State of the Art Report COMBINED USE OF NDT

Transkript

State of the Art Report
COMBINED USE OF NDT/SDT
METHODS FOR THE
ASSESSMENT OF STRUCTURAL
TIMBER MEMBERS
Editors:
José Saporiti Machado and Mariapaola Riggio & Thierry Descamps
State of the Art Report
COMBINED USE OF NDT/SDT
METHODS FOR THE ASSESSMENT OF
STRUCTURAL TIMBER MEMBERS
Edited by
José S. Machado
Mariapaolla Riggio
Thierry Descamps
Year of publication: 2015
ISBN 978-87325-094-2
This publication is supported by COST
Published by: UMONS - Université de Mons
No permission to reproduce or utilise the contents of this book by any means is necessary, other
than in the case of images, diagrammes or other material from other copyrights holders. In such
cases, permission of the copyright holders is required.
Disclaimer
The views and propositions expressed herein are those of the respective authors and, unless otherwise
stated, not represent the official view of COST Association or any organisation mentioned in this
report. Also they are not responsible for the use which migth be made of the information and for
the external websites referred in this publication.
PREFACE
COST Action FP1101 ‘‘Assessment, Reinforcement and Monitoring of Timber
Structures” is a research network established under the support of the European
framework supporting trans-national cooperation in different scientific domains
across Europe (COST). The main objective of COST Action FP1101 was to increase
the acceptance of timber in the design of new structures and supporting the safety
and serviceability of existing structures. For these purposes this Action supported
a series of activities (short-term scientific missions, training courses, workshops
and conferences) aiming to develop and disseminating methods to assess, reinforce
and monitor timber structures.
This state of the art report has been prepared by the Working Group 1/Task
Group 2: “Combination of NDT/SDT for the assessment of timber structures on
site”. The present report is divided in four broad themes that describe some of
the achievements and reflections among research groups in the last decades in
several topics related to the scope of TG2 :
• Part I includes general matters related to the use of NDT/SDT methods such as
calibration, reliability of field measurements, statistical approaches to derive
properties, as well as to update and combine information from different sources;
•
Part II includes the application of NDT/SDT methods for predicting the socalled reference properties and also specific methods aimed at determining
other mechanical/physical properties whose estimation is important for the
assessment of particular problems associated to timber members;
•
Part III deals with issues fundamental to the evaluation of the nature, extent and
causes of local features that can induce failures, damage and vulnerabilities (e.g.
aging, mechanical damage, delamination, fissures and biological deterioration);
•
Part IV includes a group of chapters highlighting the application of NDT/SDT
methods in specific contexts.
Editors are indebted to the COST Action FP1101, and to all authors, reviewers
and COST members who contributed to the preparation of this report.
The Editors
José S Machado
Mariapaola Riggio
Thierry Descamps
Leader of Task Group 2
Leader of Working Group 1
“Assessment of Timber
Structures”
Short Term Scientific Missions
officer
CONTENTS
PART I – ENHANCEMENT OF NDT/SDT INFORMATION
THROUGH STATISTICAL/PROBABILISTIC MODELS
Reliability of prediction by combining direct and indirect
measurements
José S. Machado
1
Standardization of non-destructive methods in the assessment of
existing timber structures
Guillermo Íñiguez-González, Francisco Arriaga, Miguel Esteban, Daniel F.
Llana
15
Multi sensor approach combined with multivariate analysis for
assessment of timber structures
Jakub Sandak, Anna Sandak, Mariapaola Riggio
23
Hierarchical modelling of timber reference properties using
probabilistic methods: Maximum Likelihood Method, Bayesian
Methods and Probability Networks
Hélder S. Sousa, Jorge M. Branco, Paulo B. Lourenço
33
Quantitative assessment of the load-bearing capacity of structural
components using NDT, SDT and DT inspection methods
Gerhard Fink, Jochen Köhler
45
PART II – ASSESSMENT OF REFERENCE PROPERTIES
Methodology and protocols for routine assessment of wooden
members with spectroscopy
Anna Sandak, Jakub Sandak, Mariapaola Riggio
53
Structural analysis of porous media by means of thermal
methods: Theory and monitoring measurements
Olivier Carpentier, Thierry Chartier, Emmanuel Antczak, Thierry
Descamps, Laurent Van Parys
61
Prediction of mechanical properties by means of semi-destructive
methods
Michal Kloiber, Miloš Drdácký
69
Practical procedure for estimating the density of timber with
portable X-ray equipment
Thomas Lechner, Roger Kliger
93
A methodology for the determination of the timber density
through the statistical assessment of ND measurements aimed at
in situ mechanical identification
Beatrice Faggiano, Maria Rosaria Grippa, Anna Marzo
107
Evaluation of the influence of defects on the mechanical
properties of timber through the analysis of multiscale
specimens, based on NDT and DT
111
Screw withdrawal resistances for reliability-based evaluation
of timber in existing structures
Nobuyoshi Yamaguchi
115
Combined method for the in situ mechanical identification of
ancient timber based on NDTs
131
Combine information from visual and NDT/SDT methods
Artur Feio, José Saporiti Machado
137
Assessment of timber floors by means of non-destructive testing
methods
Tiago Ilharco, Thomas Lechner, Tomasz Nowak
145
PART III – ASSESSMENT OF THE INTEGRITY OF
Application of imaging techniques for detection of defects,
damage and decay in timber structures on-site
Mariapaola Riggio, Jakub Sandak, Steffen Franke
163
Assessing the integrity and strength of gluelines
Philipp Dietsch, Thomas Tannert
175
Mapping of cracks in glulam beams and assessing the effect of
environmental conditions
Philipp Dietsch
189
Estimation on-site of decay in timber structures by means of pene
tration methods
Alessandra Gubana, Ezio Giuriani
203
PART IV – CASE STUDIES
Assessment through NDT of the state of timber structures of the
historic buildings of Catalonia
Marcel Vilches Casals, Carles Labèrnia Badia, Vladimir Rodríguez
209
SDT methods as part of a conservation process
Dulce Franco Henriques, André Santiago Neves
217
Advanced model based assessment of existing structures
Kiril Gramatikov, Toni Arangjelovski, Marija Docevska
223
PART I
ENHANCEMENT OF NDT/SDT INFORMATION
THROUGH STATISTICAL/PROBABILISTIC
MODELS
PART I- ENHANCEMENT OF NDT/SDT INFORMATION THROUGH STATISTICAL/PROBABILISTIC MODELS
Reliability of prediction by combining direct
and indirect measurements
José S. Machado
Department of Structures, LNEC, Portugal, [email protected]
Abstract
Several non and semi-destructive testing (NDT and SDT) methods have already
been developed for in situ assessment of structural timber member’s properties or
deterioration level. The use of these methods as auxiliary tools to the traditional
visual strength grading method can only be current if a fully comprehension of
their limitations and reliable exploitation of their possibilities is achieved.
This chapter discusses the possibility of obtaining more reliable predictions by
combining information from NDT and SDT methods taking into account possible
sources of uncertainty. SDT methods can provide direct measurements of a
desired wood’s property (e.g. density or mechanical properties) and then be used
to cross-validate the information obtained from indirect measurements (e.g.
drilling resistance, stress wave velocity). The discussion is based on an example
of prediction of bending modulus of elasticity through the combined information
obtained from a stress wave NDT and a tension SDT method.
Introduction
Structural timber members present a high variability of properties (between and
within members) being influenced by various variables. Among these variables
wood species, density, defects and moisture content are the ones mostly taken into
account during the survey of timber structures. Considering the large spectrum of
variables involved and the difficulties in assessing their influence on the global
mechanical behaviour of a structural timber member, several non and semidestructive testing (NDT and SDT, respectively) methods have been developed.
The description and limitations of these methods were already analyzed in recent
reports [1, 2].
Visual strength grading (VSG) was the first NDT to be developed and is still
the basic tool used in the assessment of timber structural members. VSG provides
reliable results (meaning over conservative) and a long-term experience of its
application in situ. However these over conservative values lead often to the demoli-
1
tion of structures showing no signs of damage or deformation after several years
(in some cases centuries) in service. To support expert’s decision other NDT and
SDT methods were developed as auxiliary tools to VSG for the allocation of more
reliable (meaning closest to the real value) mechanical characteristics to structural
timber members in situ. However several shortcomings explains why their applications is limited and the final decision is still based solely on expert’s opinion
and application of a simplified set of visual rules.
NDT and SDT methods can be differentiated based on the type of information
provided:
• Local (limited to a small volume of the element) or global (all volume of the
element).
• Direct measurement or an indirect measurement of the desired property (density, strength, stiffness).
• Qualitative or quantitative measurements.
In the present chapter direct methods are the ones that provide a direct evidence
of the mechanical or physical property of a wood member under examination.
This definition includes methods that although providing a direct reading will require further information in order to extrapolate local to global behaviour. Other
meaning is found in literature considering direct methods those not involving the
need for empirical models (e.g. tension of micro-specimens) [3]. Direct methods
can include the removal or not of wooden material. Proof-loading is an example of
a NDT method capable of deliver a direct measurement of the global modulus of
elasticity of timber beams in service [4]. However this method can only be applied
in certain situations [5].
In some cases the removal and testing of structural members in the laboratory
provides a direct information that can increase the reliability of NDT/SDT predictions. However this destructive procedure is dependent upon the possibility of removal of timber members (e.g. not suitable in historic structures) and results have
again to take into account the gross wood’s variability between members.
Direct assessment of physical and mechanical properties can be obtained by
some SDT [6]. Indirect methods (NDT or SDT) are frequently applied in situ
based on uni or multivariate empirical models (e.g. regression analysis) linking
indirect measurements (e.g. ultrasonic modulus of elasticity, drilling resistance) to
desired properties (e.g. static modulus of elasticity, density). Unfortunately on
most cases the existent regression models are characterized only by its coefficient
of determination and nothing is mentioned about the uncertainty of the different
models.
Wood complexity and general difficulties associated to performing measurements in situ require a careful planning of the inspection works. This should begin
always by settling on the property or properties to predict. Given a certain property the second step includes the choice regarding: NDT&SDT methods to apply;
locations of testing; number of data to obtain from each method; and, suitable data
analysis procedures.
2
Regarding data analysis it should be considered the restrictions regarding extraction of samples which only allows that only a few samples can be obtained and
analyzed (small sample size). Also the majority of NDT and SDT methods provide local properties and only concern clear wood properties. For the prediction of
the global behaviour of a timber member these information needs to be combined
with information from defects (mostly knots and slope of grain).
The present chapter discuss the possible improvement of the reliability of predictions based on the combination of NDT and SDT methods.
NDT/SDT methods Indirect estimators Direct estimators Material properties
Core Density
Core drilling
Edyn
Em
Stress waves
Time-of-flight
Tension test
Etension
Fig. 1. NDT&SDT methods used for predicting the static modulus of elasticity. Edyn – Dynamic
modulus of elasticity; Em– Bending modulus of elasticity
Core drilling (SDT), stress waves (NDT) and tension tests (SDT) are used as
example for the discussion, Figure 1. The use of these methods for the prediction
of two reference properties (density and modulus of elasticity) is analysed taking
into account possible sources of errors, the effect of small sample size, the high
spatial variation of properties inside a timber member and possible error propagation.
Joint use of direct and indirect measurements
As for other material a more precise and accurate assessment of timber
members in situ can be achieved by joining information from different sources [7].
This combination of information can be done by:
3
• considering both as independent variables in a common empirical model (e.g.
multiple regression analysis);
• using direct readings to calibrate the indirect readings made in situ. This
calibration includes validation of prior regression models as well as to quantify
adjustment factors (e.g. wood moisture);
• using both data as independent predictions of the property – possibility of
cross-validation of the prediction.
• using first a fast technique as a preliminary screening of the structure followed
by a second time-consuming technique in areas selected from the results of the
first technique (e.g. thermography followed by drilling resistance for assessing
presence and extension of decay).
Although considered above as autonomous paths they can in fact act together.
Uncertainties involving in situ evaluation
Whatever the path followed it must be bear in mind that predictions are always
affected by errors, including aleatory (high spatial variability inside a timber member
– within the cross section and length) and epistemic errors (lack of knowledge on
the material or models, associated to the test method and human errors). These
errors lead to a certain degree of uncertainty of the prediction made using NDT and
SDT “…the estimate of even single parameters using established methods can be contaminated
by significant errors and caution must be exercised in interpreting experimental data.” [8]. The
awareness of the type of error is only important as a mean to recognize what are
the possibilities to diminish the amount of error. In the assessment of existing
elements the uncertainty can be considered as epistemic (the errors are only due to
our incapacity to get the necessary information, to deal with human errors or to
apply the correct test methods) [9].
Wood’s variability (aleatory error) can as a rule of thumb be considered known
using the values provided in Table 1. These values should be regarded as start up
values possible to be adapted to any particular situation (type or quality of wooden
members, amount of information possible to be collected on site).
Table 1 Coefficients of variation for clear wood and structural timber
Property
Clear wood [10]
Structural timber [11]1)
Density
10
10
Tension strength parallel to grain
25
30
Bending strength
16
25
Compression strength parallel to grain
18
20
Modulus of elasticity in bending
22
13
1)
Values for European softwood and corresponding to a number of tests equal to 10
4
This uncertainty is always present since the number of samples or measurements possible to be made are limited and the extrapolation process (local to global assessment) is always complex given the nature of a timber member – heterogeneous, anisotropic and hygroscopic.
The limitations associated to the different NDT and SDT methods usually applied are described in a recent report [2].
Uncertainties associated to NDT/SDT methods
Density prediction using core drilling
Density is an important property given his positive direct impact in the strength
and stiffness of wood. Some of issues involved in the determination of density in
situ are discussed in [8]. It is also used to predict the modulus of elasticity through
the determination of the dynamic modulus of elasticity. Density can be predicted
using core drilling (SDT), drill resistance (SDT), penetration resistance (SDT) and
pull-out resistance (SDT) [12]. The accuracy and precision of density’s prediction
model is strongly dependent on the variability showed by each individual timber
member. Density’s variation occurs along its length and within the cross-section
(width and depth). The error of prediction can be partially dealt if the NDT or
SDT method is applied taking into consideration important characteristics of the
member (namely wood species, growth ring pattern and spatial variation inside the
member) and if a sufficient number of readings are collected. Statistical models allow having an estimative of error as function of the number of readings collected,
Figure 2. However this error is underestimated since it does not consider a possible spatial variation which usually occurs in wood. Consequently the sample size
effect on error given by models as the one illustrated in Figure 2 only provides
guidance for maintaining a certain level of precision and does not ensure the accuracy of our prediction.
Core drilling is a multifunction SDT method capable of providing information
about wood species, moisture content, strength and density. The determination of
density of core samples can be done according with standard procedures. Since the
number of readings is limited (due namely to level of destruction made to the timber member) and considering the level of variability that can be found in a sole
timber member the reliability of density’s prediction is highly dependent on: the
number of readings; and, the way they are carried out in order to ensure a proper
representativeness of the material under observation. Therefore the accuracy of
this SDT method depends strongly on expert decision about where to extract and
5
the length of the cores whereas precision are more related with the number of
cores extracted.
Fig. 2. Error as function of sample size and mean density (coefficient of
variation of 10%) [13]
To illustrate consider a timber member with a cross-section 304 mm x 148 mm
and two possible annual ring patterns (quarter-sawn and boxed heart). A quartersawn pattern (A) can be considered closest to a homogeneous section. In this
example the density of each layer of latewood and earlywood is randomly generated
assuming a normal density probability function using data taken from [14]. No
lengthwise variation is assumed. A collection of four cores taken perpendicular to
the grain and with different depths was simulated.
A boxed heart pattern (B) shows a cross-section variation of density was
considered from the pith to the surface. Also no lengthwise variation was
considered. In old timber structures large cross-section usually contains the pith
inside showing an annual growth ring pattern similar to case B. For this case a
comparison was made between taking four wood cores from the edges or taking
two from the edges and two from the faces.
For all cases a bootstrap method was applied running 10000 iterations. Figure 3
shows the result of the different simulated cases. Figure 3 shows the importance to
take into account the variation of properties inside a timber member in the prediction
of density by a core drilling method. Depending on the type of growth ring
pattern, cross-section variability and type of core samples (depth and extraction
procedure) density prediction’s error can vary in average from 0% to 15%. An
estimation of the lengthwise and cross-section variability can be obtained by using
another SDT method – drilling resistance. The variation of growth ring width is
another source of variability not taken into account in the present simulation.
Density values are affected by moisture content and thus wooden cores they
6
can be dried and moisture content determined for carried out the necessary
corrections. The need to apply correction factors are always a difficult issue
being address in another chapter of the present publication [15].
Fig. 3. Estimated error associated with density prediction through core drilling. ∆ - cross-section
variability; s – type of sampling (1 – all samples taken at the edge; 2 – two samples taken at the
edge and other two on the faces)
7
The accuracy in density’s prediction is important since this variable will be
used for predicting the dynamic modulus of elasticity or directly the mechanical
properties.
Other NDT/SDT methods for predicting density are available being some
described in other chapters [16].
Modulus of elasticity prediction using stress waves
Static modulus of elasticity can be predicted by tension and/or compression tests
(SDT) [17, 18], dynamic response (NDT) [19], stress waves (NDT) and load test
(NDT) [12].
The prediction of the static bending modulus of elasticity (Em) is usually
obtained in situ using the correlation with the dynamic modulus of elasticity
(Edyn), regression equation below.
E
m
= a + bEdyn + ε
The dynamic modulus is obtained by applying stress waves NDT methods in
situ. Information about the use of stress waves for in situ assessment of structural
timber can be found in [12]. Wood is an orthotropic material being the equations
of motion for bulk waves given by Christoffel’s equation [13]. However considering the complexity of wood material and the experimental conditions on site a
simplification is made and it is applied the equation for isotropic solids and obtaining a prediction of the dynamic modulus of elasticity (Edyn) as showed above.
(
))(
2  1 + υ 1 − 2υ
Edyn = V ρ 
 (1 − υ )
)


Considering the complexity of determining in situ the Poisson’s ratio the above
simplified equation is generally applied.
E
dyn
2
=V ρ K
 (1 + υ )(1 − 2υ ) 
 =1
 (1 − υ ) 
Assuming K = 
The component of the equation regarding the coefficient of Poisson is considered a determinist value. The use of this equation is supported in different studies.
Considering the values given in Table 2, for a stress applied along the grain a
mean k value could probably be found between 0.39 and 0.57. Nevertheless since
the uncertainties surrounding k factor (random) are merged with other uncertain8
ties (density and TOF measurements) and taken into account by a final regression
curve the real value of k is not significant.
Table 2 Average Poisson’s ratio for Softwoods and Hardwoods [22]
Coefficient
Softwood
Hardwood
vLR
0.37 (0.06)
0.37
vLT
0.42 (0.07)
0.50
vij – ration for deformation along the j axis caused by stress along the i axis
TOF determination can be affected by: the lack of proper coupling between
transducers and wood (wood surface roughness, lack of coupling agent); slope of
grain or presence of other defects; uncertainty about wave path length [23, 24]. To
minimize the error several readings (at least 5 [20]) should be carried out along the
length of the beam.
Regarding density small samples can be taken from timber members for
determination in the laboratory. This procedure can be done through core drilling
as seen in the previous section and possible errors associated were already
mentioned. Finally TOF should be corrected accordingly with the temperature
and moisture content of the beam.
The combine information from stress waves and core drilling provide a prediction
of the dynamic modulus of elasticity (Edyn). Once obtained the dynamic modulus
of elasticity a prediction of the static modulus of elasticity is generally done by
applying empirical models (regression curves). The correlation between this
two variables can varied from 0.58 to 0.96 depending upon the dimension (clear
wood or structural wood), treatment and age of timber specimens [24].
Modulus of elasticity prediction using tension strength tests
Modulus of elasticity parallel to grain shows moderate dependence to type of
loading being a common value used for design of timber structures. However
the modulus of elasticity in compression is lower that in bending which in turn is
lower than in tension [25] for structural timber elements. Nevertheless it can be
considered independent of the load involved for clear wood specimens [25].
Information about the modulus of elasticity in bending of clear wood can then
be obtained from tension tests carried out on small samples removed from timber
members in situ [6]. This information although limited in terms of possible number
of tests when compared with the possible readings obtained from stress waves can
nevertheless provide us the possibility to: validate the values obtained from the
indirect method; and, increase the reliability of our prediction.
9
Combining information from indirect and direct methods
To illustrate different options for combining information obtained from direct and
indirect methods data from previous works will be used [26]. In these studies
density and modulus of elasticity were obtained using the SDTs and NDT methods
aforementioned.
However before trying to combine this information they should be careful
analysed. A forest plot (Figure 6) can be used to evaluate if the results of both
tests are consistent (values ranging in a similar interval).
Fig. 6. Forest plot showing the mean and 95% confidence intervals for the results obtained from
Edyn and Et (values in N/mm2)
The result obtained allows us to consider that the data obtained from the two
testing methods although not coincident can be accepted as being coherent.
Unfortunately the values obtained from these two methods showed a high
correlation coefficient (r2 = 0.85) representing a possible conflict with one of the
assumptions of regression analysis (independency of independent variables). The
questions related with excessive multicollinearity does not have any standard
metric and so it is a decision of the expert to make about using or not a multiple
regression model in this circumstances.
In the present case a Meta-analyse technique is used and the combining of
information is made trough an inverse variance method, see equation bellow.
−2
Ecomb = ω 1 × E1 + ω 2 × E 2 + ... + ω n × E n and ωi = σ i
n n
å σk
k=1
Where n estimators of the variable E are combined as a weighted average
according with their variance ( σ).
The combined predicted value can then be used as explanatory variable (Ecomb)
in a simple regression model. Two models are then available for predicting Em:
one as function of Edyn (r2 = 0.72); and, the other as function of a combined value
Ecomb (r2 = 0.77). The two models are very close in terms of capacity of explained
the Em variability. These regression models assumed that independent variables
are measured without errors and the error is only associated to the dependent
variable [27]. However for the present analysis it will be taken into account also
the uncertainty related with the independent variables (Edyn and Ecomb), scenario
10
more close to reality. For that purpose a Bayesian analysis applying a Markov
Chain Monte Carlo (MCMC) method is used to drawn inferences about the
models and parameters. MCMC algorithms generate a Markov chain sequence
of parameter values (the values at a given state of the chain depends only
upon the previous state) being these parameters generated randomly (Monte
Carlo). MCMC was carried out by running Winbugs inside R software trough
R2WinBUGS package (1 chain, 1000 burn-in iterations, 9000 used iterations).
Figure 7 shows the prediction for beam P20 and P21.
a)
b)
Fig. 7. Predicted Em distribution for: a) beam P20 and b) beam P21, using model Em(Edyn) – red –
and model Em (Ecomb). Real stiffness showed in the graphs by a dashed vertical line
For both beams considering the uncertainties of the variables it is clear that a
closer approximation to the real stiffness is obtained using the combined information.
Final remarks
Assessment of timber structures is always done by crossing information from
different sources. However this process is generally based on a series of individual
results obtained from different NDT/SDT methods and that the expert uses to take
an informative decision about the structural health of the structure. The possibility
of combining in one single model the information provided by two independent
methods, as for other materials (e.g. concrete), can lead to a more reliable prediction
of that property. The need to consider the uncertainties associated to each test
method, the need to understand the assumption behind a particular statistical model
and finally the need to asses the robustness of the final model are also matters
that should be considered.
The complexity of wood makes that any final decision relies heavily on the expert
capability of extracting valuable information from the test methods applied, his/
her experience and particular conditions regarding the structure under inspection.
This heterogeneous information (qualitative and quantitative) makes bayesian
methods a suitable data analysis method to be applied.
11
References
[1]
Dietsch P, Köhler J (2010) Assessment of timber structures. Shaker Verlag, Aacher.
[2]
Kasal B, Tannert T (2010) In situ assessment of structural timber. Springer.
[3]
Kasal, B (2013) Structural health assessment of in-situ timber – An interface between
service life planning and timber engineering. In 9th Meeting of Northern uropean Network
for Wood Science and Engineering (WSE). Leibniz University of Hannover. Brischke, C.
& Meyer, L. (eds). p. 157-164.
[4]
Riggio M, Tomasi R, Piazza M (2014) Refurbishment of a traditional timber floor with a
reversible technique: Importance of the investigation campaign for design and control of
the intervention. International Journal of Architectural Heritage 8:74-93.
[5]
Srpčič J (2010) Load tests (Proof Loading). In: Dietsch P and Köhler J (Eds) Assessment
of timber structures, Shaker Verlag GmbH, Aachen, p 96-99.
[6]
Kloiber M, Drdácký M (2014) Prediction of mechanical properties by means of semidestructive methods. In J S Machado, Mariapaola Riggio and Thierry Descamps (Eds) Combine
use of ndt/sdt methods for assessment of structural timber members, this issue.
[7]
Verma S K, Bhadauria S S, Akhtar S (2013) Review of nondestructive testing methods
for condition monitoring of concrete structures. Journal of Construction Engineering
article ID 834572.
[8]
Kasal B (2012) State-of-the-art in situ evaluation of structural timber – some critical
observations. Revista Portuguesa de Estruturas 11:79-85.
[9]
Kiureghian A D, Ditlevsen O (2009) Aleatory or epistemic? Does it matter?. Structural
safety 31: 105-112.
[10] Forest Products Laboratory (2010). Wood Handbook – wood as a engineering material.
General Technical Report FPL-GTR-190. U.S. Department of Agriculture, Forest
Service, Forest Products Laboratory, Madison, 508 pp.
[11] JCSS (2006). Probabilistic model code. Part 3.5 – Timber. ISBN: 978-3-909386-79-6,
Joint Committee on Structural Safety.
[12] Höls M, Dietsch P (2010) Assessment of timber structures. In: Dietsch P and Köhler J
(Eds) Assessment of timber structures, Shaker Verlag GmbH, Aachen, p 14-21.
[13] Zar J H (1996) Biostatistical analysis. 3rd ed. Prentice-Hall International, Inc.
[14] Fries A, Ericsson T (2009) Genetic parameters for earlywood and latewood densities
and development with increasing age in Scots pine. Ann. For. Sci. 66:404
[15] Iñiguez-González G, Arriaga F, Esteban M, Llana D F (2014) Standardization of
non-destructive methods in the assessment of exsting timber structures, In J S Machado,
Mariapaola Riggio and Thierry Descamps (Eds) Combined use of NDT/SDT methods
for the assessment of structural timber members, this issue.
[16] Faggiano B, Grippa M R, Marzo A (2014) A methodology for the determination of the
timber density through the statistical assessment of ND measurements aimed at in situ
mechanical identification. In J S Machado, Mariapaola Riggio and Thierry Descamps
(Eds) Combined use of NDT/SDT methods for the assessment of structural timber
members, this issue.
[17] Drdáchy M, Kloiber M (2013) In-situ compression stress-deformation measurements along
the timber depth profile. Advanced Materials Research 778:209-216.
[18] Kasal B, Anthony R (2004) Advances in in situ evaluation of timber structures. Progress
in Structural Engineering and Materials 6:94–103.
12
[19] Ilharco T, Lechner T (2014) Assessment of timber floors by means of non-destructive testing
methods. In J S Machado, Mariapaola Riggio and Thierry Descamps (Eds) Combine
use of ndt/sdt methods for assessment of structural timber members, this issue.
[20] Dackermann U, Crews , Kasal B, Li J, Riggio M, Rinn F, Tannert T (2014) In situ assessment
of structural timber using stress-wave measurements. Materials and Structures 47:787-803.
[21] Bucur V (1995) Acoustics of wood. RC Press, New York.
[22] Bodig J, Goodman R (1973) Prediction of elastic parameters for wood. Wood Science
5:249-264.
[23]
Chapman M J, Norton B, Taylor J McA, Lavery D J (2006) The reduction in errors
associated with ultrasonic non-destructive testing of timber arising from diferential pressure
on and movement of transducers. Construction and Building Materials 20:841-848.
[24] Kasal B, Lear G, Tannert T (2010) Stress waves. In: B Kasal and T Tannert (Eds) In situ
assessment of structural timber.
[25]
Gehri E (1997) Timber as a natural composite: explanation of some pecularities in the
mechanical behaviour. Case: Assessment of the modulus of elasticity of timber parallel to
grain. In: Görlacher R (Ed) International Council for Building Research Studies and
Documentation – Working Commission W18 – Timber Structures, Vancouver, Paper 30-6-3.
[26] Machado J S, Lourenço P B, Palma P (2011). Assessment of the structural properties
of timber members in situ – a probabilistic approach, in: SHATIS’11 International
Conference on Structural Health Assessment of Timber Structures, Lisboa.
[27] Gillard J (2010) An overview of linear structural models I errors in variables regression.
REVSTAT – Statistical Journal 8:57-80.
13
14
Standardization of nondestructive methods in
the asssessment of existing timber structures
Guillermo Íñiguez-González, Francisco Arriaga, Miguel Esteban, Daniel F.
Llana
ETSI de Montes, Forestal y del Medio Natural, Universidad Politécnica de Madrid, Spain.
[email protected]
Abstract
The aim of this paper is to define the reference conditions and to propose
modification factors to standardize the nondestructive variables recorded by
different NDT, in order to gain uniform practical data and to develop a procedure
for the standardization of timber assessment.
Introduction
The standardization process consists of the development of a common NDT
procedure for the evaluation of structural timber properties. This process could
be based on previous works (Íñiguez-González et al. 2013) and should include
the following features:
- Compilation of nondestructive test results from different research groups
and studies, taking into account the species studied and devices used;
-
The creation of a standardized data sheet to compare results, based on the
adjustment factors proposed;
-
Standardized equations.
This paper therefore focuses on the second and third features.
Background
2.1 Nondestructive variables
The following variables are usually measured for the in situ assessment of timber
15
properties by means of nondestructive techniques:
Time of flight (or equivalent velocity)
The propagation of stress waves through material can be used to estimate its
mechanical properties, mainly stiffness. Time of flight (TOF) or the equivalent
velocity, is the main parameter measured. Attenuation can also be recorded and is
related to strength properties. Two types of waves are used: sonic stress
waves for frequencies within the audible range, and ultrasonic stress waves at
frequencies above 20 kHz.
Natural frequency
In this method timber pieces are made to vibrate longitudinally or transversely
by means of an impact in the corresponding direction. The vibration of the
piece occurs primarily in the system Eigen frequencies. These frequencies are
related to the stiffness properties of the piece and its dimensions/geometry. In
the case of longitudinal vibration, it is also possible to obtain the equivalent
velocity of stress wave transmission.
Pullout resistance
The pullout resistance method consists of the measurement of the withdrawal
force of a screw with a known diameter inserted into a timber piece to a certain
depth. This force is related to the density of the timber.
Penetration depth
This method is based on a similar principle to that of material hardness
measurement, and it consists of measuring resistance to the penetration of a
hard solid piece. Penetration depth is related to the timber density in the outer
part of the piece of wood.
Drill resistance
This method mainly focuses on gathering data on the internal condition of timber
members and trees. It uses a small diameter drill (1.5-3.0 mm) to bore into timber
members while measuring resistance to penetration (energy consumed at constant
velocity). Resistance to drilling is proportional to density.
16
2.2 Factors affecting nondestructive variables
Moisture content
The moisture content (MC) of timber depends on the hydrothermal conditions of
the surrounding air. In a normal dry condition inside a building the MC of timber
is from 8 to 12 %; this is slightly higher, 10 to 16 %, if the building is close to the
coast. MC can be measured with electrical resistance equipment, according to
the EN 13183-2:2002 standard. A reference value of 12% MC is usually adopted.
Under usual conditions the range of MC variation is about ±4% (12±4%).
Temperature
In general, the mechanical properties of wood decrease when it is heated and
increase when it is cooled. However, in the practical range from -10 to 50 ºC the
effect is negligible; for example, in structural mechanical timber design it is
assumed that the properties of timber do not change and are not affected by
temperature below 50 ºC. Consequently, the effect on measured nondestructive
variables is very small.
Size/length
The size effect may have an influence on some local nondestructive variables such
as pullout resistance and penetration depth, as a consequence of the sawn pattern
and cross-section size of the piece. In the case of large cross-sections, the outer
wood of the cross-section usually has narrow rings, and the direction of probe
penetration is more radial than it is tangential. On the other hand, small or narrow
cross-section pieces may have juvenile wood in the outer part of cross-section, so
that tangential penetration is possible.
Wave velocity propagation is independent of frequency. But different commercial
devices work at different wave frequencies, and the means used to detect signal
start and stop may differ. The result is influenced by signal attenuation and
therefore depends on the length and size of the piece. It is more difficult to
establish a reference value for the size and length factors.
Positioning of sensors/angle to the grain
This factor affects the measurement of time of flight (or the equivalent wave
propagation velocity). The best positioning of sensors for measurements is at each
end of the piece, obtaining the velocity parallel to the grain.
This is not possible in practice in existing structures because the ends of pieces
are not accessible. It is therefore common practice to take measurements at an angle
17
to the grain, positioning the sensors on opposing faces at the maximum possible
distance or, in the central portion of a span, obtaining the velocity for a certain
angle. For frequent dimensions of timber pieces and cross-section slenderness
(width/thickness = 1-2) and length (length/depth of the beam = 17-22) the angle
with respect to the grain varies from approx. 2 to 6 º. In some situations only the
lower face of the piece is accessible, and in this case sensors are placed on
the same face, obtaining velocity parallel to the grain but under special conditions.
3. Standardization proposal
3.1 Reference conditions
Moisture content
12% moisture content of timber is proposed as reference value to correct the
nondestructive measurements (stress wave, ultrasound wave velocity and penetration
depth). This value corresponds to the target MC for coniferous timber in service
class 1 according to Eurocode 5 (EN 1995-1-1:2004). Service class 1 is characterized by a MC in the materials corresponding to a temperature of 20ºC and a relative
humidity of the surrounding air only exceeding 65% for a few weeks per year.
Temperature
20ºC is the reference value proposed in general, and according to the definition
of mechanical properties in Eurocode 5. But considering its low effect
on nondestructive variables this correction may be neglected in frequent
variable conditions (-10 to 50ºC).
Size/length
The effect of size is considered here in terms of length, and 2.7 m is the proposed
reference length to correct nondestructive measurements (stress wave and
ultrasound wave velocity). This value is based on standard bending test
slenderness (span/depth = 18) and the reference depth for bending strength of
solid timber in Eurocode 5 (150 mm, 18•150 = 2700 mm). In practice length
may usually be in the range from 3 to 6 m, and considering the reference value
of 2.7 m the maximum effect would be a 7% variation in velocity.
18
Positioning of sensors/angle to the grain
The end to end positioning of sensors, measuring the velocity parallel to the grain,
is the proposed reference position of sensors and reference angle. In practice the
angle may usually be in the range from 2 to 6º, so the maximum effect would be
a 6% variation in velocity.
3.2 Modification factors
Moisture content
The reference velocity of stress wave propagation, v12 (referred to 12% MC) may
be obtained by equation 1, from velocity at H % MC, vH,
vH
v12 =
1 − ( H − 12) k H
(1)
Where kH is the adjustment factor for MC obtained as the ratio between the
linear variation of velocity relative to MC (∆velocity/∆MC) related to velocity
at 12% MC. A preliminary value of 0.01 (1% velocity decrease for every 1% MC
increase) is proposed for this, as it is a common result in several research works.
The reference depth penetration of the Pilodyn 6J Forest, P12 (referred to 12%
MC) may be obtained by equation 2 from depth penetration at H %, PH,
P12 =
PH
1 + ( H − 12) k P
(2)
Where kP is the adjustment factor for MC obtained as the ratio between the linear
variation (depth/MC) related to depth penetration at 12% MC. A preliminary
value of 0.02 is proposed (approx. 2% depth penetration increase for every 1%
MC increase) for this factor. There are other experiences suggesting that its effect
be neglected for practical purposes.
Temperature
The reference velocity of stress wave propagation, v20 (referred to 20 ºC temperature)
is obtained by equation 3 from velocity at T ºC temperature, vT,
v 20 =
vT
1 − (T − 20 ) k T
19
(3)
Where kT is the adjustment factor for temperature obtained as the ratio between
the linear variation of velocity relative to temperature (∆velocity/∆T) related to
velocity at 20 ºC. This factor has been obtained in several research works with a
value close to 0.00075 (a 0.075% velocity decrease for every 1 ºC temperature
increase). In practice the temperature may usually be in the range from 5 to 35 ºC,
and considering the reference value of 20 ºC the maximum effect would be 1.12 %
of variation in velocity.
Size/length
The reference velocity of stress wave propagation, v2.7 (referred to a reference
length of 2.7 m) is obtained by equation 4 from velocity at L length in m, vL,
v 2.7 =
vL
1 − ( L − 2.7 ) k L
(4)
Where kL is the adjustment factor for length obtained as the ratio between the
linear variation of velocity relative to length (∆velocity/∆L) related to velocity at
reference length 2.7 m.
Positioning of sensors/angle of the grain
The reference velocity of stress wave propagation, v0 (parallel to the grain and end
to end) is obtained by equation 5, from velocity angle α in sexagesimal degrees
(α ≤ 10º), v,α
v0 =
vα
1 − α kα
(5)
Where kαis the adjustment factor for angle obtained as the ratio between the linear
variation of velocity relative to angle (∆velocity/∆α) related to velocity parallel to
the grain and end to end. This factor was obtained in several studies with a value
close to 0.01 (a 1% velocity decrease for every additional grade increase in angle
deviation) for ultrasound waves (at 22 kHz).
Finally, if velocity is measured only using sensors in the lower face of the
piece, these values should be divided by a factor of 0.972 to obtain the velocity v0.
20
4. Conclusions and future work
Given the lack detected in the state of the art, research work compiling
nondestructive testing values measured in coniferous and deciduous species is being
undertaken in Spain. This has the purpose of collecting existing test results measured
in raw material and existing structures with different devices and procedures. A
testing protocol has also been designed and proposed in order to make the results
of future nondestructive tests by different researchers comparable. Although this
study does not yet include sufficient material to propose a grading method of
timber based on nondestructive testing results, it does mark the route to this goal.
Acknowledgments
Ministerio de Ciencia e Innovación: Proy.: BIA 2010-18858. Plan Nacional I+D+i 2008-2011.
Proy.: BIA 2006-14272. Plan Nacional I+D+i 2004-2007. Proy.: AGL 2002-00813. Plan
Nacional I+D+i 2000-2003.
References
EN 13183-2. (2002). Moisture content of a piece of sawn timber. Part 2: Estimation by electrical
resistance method. European Committee for Standardization. Brussels/Belgium.
EN 1995-1-1. (2004). Eurocode 5. Design of timber structures. Part 1-1: General. Common rules
and rules for building. European Committee for Standardization. Brussels/Belgium.
Íñiguez-González, G., Llana, D.F., Montero, M.J., Hermoso, E., Esteban, M., García de Ceca,
J.L., Bobadilla, I., Mateo, R., Arriaga, F. (2013). Preliminary results of a structural
timber grading procedure in Spain based on non-destructive techniques. Proceedings of 18th
International Nondestructive Testing and Evaluation of Wood Symposium. Madison,
Wisconsin, USA. Pp. 386-395.
21
22
Multi sensor approach combined with
multivariate analysis for assessment of timber
structures
Jakub Sandak1, Anna Sandak1, Mariapaola Riggio2
1 CNR Ivalsa, S.Michele all’Adige (TN), Italy,
[email protected], [email protected]
2 Wood Science & Engineering, Oregon State University, USA,
[email protected]
Abstract
Numerous methods are used currently for existing timber structure assessment,
including continuous monitoring of the structure performance on-site. However,
its comprehensive characterization is often problematic issue due to wood
anisotropy and heterogeneity. Trends for using multiple sensors simultaneously
become more popular due to realistic representation of the real-world cases and
advancement in hardware development. Multi-sensor approach even if possessing
plenty of advantages require integration of non-destructive testing methodologies
and data handling techniques to currently assess, monitor and predict wooden
members properties. It is crucial to assure proper pre-processing of the signals
from sensors, appropriate data fusion and optimal data analysis. The paper
presents sucessful applications of the different data analysis techniques used for
the assessment and monitoring timber structures. It is assumed that, after
additional developments, such methodologies can provide supplementary data to
be considered when inspector decision is made in order to support selection of
optimal conservation process.
Introduction
Currently blooming engineering research provides us with numerous methods to be
used for improving (reengineering) existing structure assessment routines, including
also continuous monitoring of the structure performance. The availability of novel
statistical tools to handle many variables simultaneously is another stimulus for
rapid changes within the field of measurement technology and the sensors domain.
Current trend for using multiple sensors simultaneously is more favorable than a
single sensor approach due to far better representation of the real-world cases: the
world is multivariate.
23
Multi-sensor monitoring generates new issues and challenges, where the fusion
of different sources of information is fundamental. Data collected from different types
of sensors are often based on diverse physical phenomena, therefore interpretations
of results is complicated. A general flowchart of the multi-sensor approach in
timber structures assessment is summarized on Figure 1. It consists of several
layers, including sample/object/case, sensors measuring member properties (through
generating various types of data), and numerical models/tools to deal with such
data in order to support expert in decision making.
Sample
Samples, cases or objects are the physical units on which the evaluation/measurements
are performed. It can be a single wooden member or the whole structure, depending
on the scope of evaluation and/or the goal of inspection. There are numerous
sample(s) characteristics of interest when assessing timber structures, including:
material properties
degradation stage of wooden members due to biotic and a-biotic agents
presence, position and incidence of strength-affecting defects in wooden
members
presence of damp areas and not uniform moisture distribution in wooden
members
mechanical damage in wooden members and connections
geometrical alterations in the wooden members and assemblies
overall performance of the structure
and others.
Sensors
The visual assessment of timber structures on can be complemented by a series of
instrumental techniques, giving information on unreachable object, about not-visible
features, and on measurable/quantifiable parameters. The range of sensing techniques
suitable for characterization of wood within structures is very wide and includes:
vision systems in various spectral ranges; visible, infrared, thermovision, hyperspectral cameras
measurement of electromagnetic radiations penetrating structure of wood;
γ-rays, X-rays, microwaves, radar detectors
analysis of mechanical/stress waves propagation; vibration; ultrasound
sensors, accelerometers, microphones, laser vibration-meters
semi-destructive methods; drilling/penetration/cutting equipment as well
as screw withdrawal portable testers
portable spectrophotometers in visible, near-infrared, mid-infrared, XRF
spectral bands
24
wood moisture meters of various types
Environment condition monitoring systems including measurements of
temperature, relative humidity, solar radiation, rain events and intensity,
among others
Fig. 1. Multi sensor approach combined with multivariate analysis for assessment of timber
structures
Pre-processing
The pre-processing of raw signals is a routine task usually performed before any
further data evaluation. Several treatments are available, including:
electronic signal manipulation, amplification, filtering, compensation, etc.
numeric signal manipulation; normalization, filtering, correction, derivative,
25
integration, noise reduction, smoothing, interpolation, averaging, convolution,
etc.
numerical processing; compression, filtering, wavelet analysis, Fourier
transform, etc.
The optimal selection of signal pre-processing is crucial for the overall performance
of the multi-sensor system, as well as the presentation of the results/parameters/ data.
Even if all the existing sensors help in characterization, the human contribution to
the structure assessment is indispensable.
Data types
The final output of the sensor (following the pre-processing procedures) is very
diverse and depends on the sensor itself, the moment of data analysis, and nature of
investigated object/structure, among others. Various sorts of data may be accessible,
including:
single/multiple variables such as scalars or vectors; value at a given time
of measurement, change of this value (in the function of time, temperature,
pressure, frequency, etc) measured with constant time laps or randomly
waves, in the form of series of measurements, with defined starting point
and distance between measurement points; such as stress-waves, vibrations,
radar signals, microwaves
images, or spatially resolved data, in the form of matrices (a rectangular
array of numbers/variables arranged in rows and columns); gray, color, xray absorption, thermal images
spectra, where series of data are representing frequency,
wavelength, wavenumber resolved signals; UV-, Vis-, NIR-, IR-, XRFspectra
hyperspectral/multispectral cubes, what are hybrids of images and spectra
where each pixel represents the full spectra in a given range
Data fusion
The data fusion strategies are different when combining data in real-time or when
analysis can be performed after measurement on the archived data. In the first
case, dedicated interfaces are indispensable and such data fusion systems are rather
complex/case-dependent.
Whilst data evaluation can be performed in post-process mode, the data fusion is
rather straightforward. It can be carried out with the help of different software tools
(suitable for dealing with various signals/sensors) and accessing diverse databases.
The most common result of the data fusion process is a spreadsheet containing
26
series of parameters extracted from various sensors corresponding to single sample/
case, collected at a given time or period of time. The size of the spreadsheet may
vary depending on the system complexity, number of sensors utilized, quantity of
samples/cases and/or duration of the measurement/monitoring. In general, the more
complex (in terms of the number of variables and samples) collection of fused data
the more reliable/generalized numerical models may be created.
It has to be stated that having huge data sets is not equal to possessing
“information”. In fact, high number of data may be cause of disturbance,
misinterpretations or confusion, especially when “conventional” data analysis
techniques are applied. Multivariate analysis (MVA) techniques, which allow more
variables to be analyzed at once, are thus an alternative. The MVA can be divided
into three groups:
exploratory data analysis (data mining) – attempts to find the hidden
structures in large and complex data sets
classification models – are useful when identification of unknown sample/
object within one of previously established classes is required
regression analysis and predictive models – are used for developing
statistical models on the base of available reference data.
Number of software packages suitable for MVA and for non-linear systems is
already available on the market including, among others, Unscrambler X, OPUS,
SIMCA. There is also a high number of dedicated modules for various software
development environments such as C++, R, Matab or LabView, in some cases
offered as an open source code. A list of MVA techniques suitable for applications
toward timber structure assessment is presented below.
Cluster analysis
Cluster analysis (CA) is a statistical method used for matching multivariate data
into particular groups according to their similarities. CA divides similar samples
into groups called classes or clusters. Clustering methods belongs to unsupervised
statistical algorithms; therefore do not require previous information about the objects’
memberships, which are obtained according to the data’s intrinsic characteristics,
or dissimilarities. The clustering can be displayed in the form of a dendro-gram
where the heterogeneity explains the similarity between the samples. The higher
the heterogeneity, the higher is the difference between samples.
Principal Components Analysis
Principal Components Analysis (PCA) is a powerful statistical method for decorrelation of highly correlated data and to reduce the high dimensional data set to
27
lower dimensions. PCA decomposes a linear combination of original variables into
few PC (principal components or factors). Each PC explains part of the data set
variability. The number of significant factors is case dependant, but as a rule it
should be as low as possible. In analogy to cluster analysis, PCA searches for
unique properties of samples and separates set of input data into groups of peculiar
similarities allowing visualization of natural clustering of the data.
Identity test
Identity test is a method extending use of Principal Components Analysis
for differentiation of sample/cases and classification of unknown samples
within previously defined groups. The identification of the unknown sample/case
is based on computation of principal components by using loadings corresponding to
the model. The resulting components are then compared to each group within the
model. The result of such comparison is the sample distance called hit quality. The
better sample match with the model group, the smaller is the distance. The hit quality
of each comparison is weighted against threshold corresponding to each modeled
group/class. Three possibilities of unknown sample identification are possible as a
result of the identity test:
the sample is identified as one of the modeled classes (hit quality
< threshold in case of only one class)
the sample is identified as probably belonging to more than one modeled
classes, therefore not unique identification is possible (hit quality < threshold
in case of more than one class)
the sample is identified as none of the modeled classes (hit quality >
threshold in case of all classes)
SIMCA
Soft Independent Modeling of Class Analogy (SIMCA) is another classification/
identification algorithm using Principal Components Analysis for differentiation of
sample classes. In analogy to identity test, set of meaningful principal components is
derived from the data set. The difference lays in modeling of classes, as in SIMCA
each class is modeled separately (local models) and number of principal components
may vary between classes. The prediction of a probable class membership for new
samples/observations is performed by determination of best fitting to the respective
class (local model).
28
Partial Least Squares
Partial Least Squares (PLS) is a statistical method considered as an expansion of
Principal Components Analysis toward quantitative analyses. Basically, PLS finds
a linear model describing some predicted variables in terms of other observable
variables. The development of PLS model starts with computation of principal
components on the base of calibration dataset. The obtained principal components
are regressed in the next step against reference variables to be predicted. The PLS
model has to be validated after calibration. The coefficient of determination, the rootmean square error of prediction and ratio of standard error of prediction to sample
standard deviation are commonly accepted indicators of the PLS model quality.
Multiple linear regression
Multiple linear regression (MLR) is a multivariate analysis tool for modeling the
relationship between two (or more) explanatory variables by fitting a linear equation
to the reference data. MLR can also be used to estimate the linear association
between the predictors and responses, in analogy to Partial Least Squares. Another
use of multiple regression is to understand the functional relationships between the
dependent and independent variables, by discovering the cause of the variation.
The relationship between all predictors and a given response is summarized by the
regression coefficients.
Expert systems
The most common perceptive of the expert systems is a rule-based programming.
In this programming paradigm, rules are used to represent heuristics, which specify
a set of actions to be performed (or decisions to be taken) for a given situation. A
rule is composed of an if portion and a then portion. The if portion of a rule is a
series of patterns which specify the facts (or data) which cause the rule to be
applicable. The then portion of a rule is the set of actions to be executed when the
rule is applicable.
Fuzzy logic
The easiest method to emphasize “knowledge” is to use not exact expressions,
avoiding precise quantifications and classifying variables into rough values/sets.
The scientific usage of such semantics is implemented within fuzzy logic expert
systems. The value of each variable (obtained from one or more sensors) is
29
“fuzzyfied” according to pre-defined classes. As a result, detailed numerical value
of variable is replaced by the fuzzy value such as “low”, “medium” or “high”. The
set of fuzzyfied variables is then propagated to the module where if-then rules are
tested. The output of the if-then module is a fuzzy number. It has to be “defuzzyfied” in order to make it a numerical/predicted value and become compatible with
the following steps of evaluation and/or decision making.
Neural networks
Neural networks (NN) are widely used for processing of very complex and multivariable data sets. NN are very functional and have a number of great advantages;
they have a parallel computation nature, can be applied in various applications, can
continuously learn and adopt themselves for changing circumstances. NN is alike a
“black box” as the knowledge acquired by the NN is hidden in the neuron weights.
The set of variables is propagated to the input of the NN. The properly trained NN
will process the input vector and will generate an output – predicted value to be
later used for decision making or other actions. Back propagation algorithm is the
most popular method for NN training, even if other procedures (including nonsupervised learning) are also available.
Decision
The overall purpose of characterizing wooden members with different sensors, as
well as developing MVA solutions to deal with multivariate data, is to assist the
inspector in making the correct decision regarding the structure assessment, safety
and optimal maintenance. It is impossible to generalize the final reasoning process,
and to even think to reduce the importance of the inspector in the final decision
making, as each wooden structure is a unique case. It is clear, however, that the result
of multivariate analysis has to be handy, reliable and intuitive for interpretation.
Conclusions
The proper timber structure assessment is of great importance to assure safe service
of buildings as well as to preserve cultural heritage objects for future generations.
A multi-sensor approach may be a very attractive alternative to conventional
nondestructive method assessment and can provide supplementary data to be
considered when inspector decision is measured. The problematic issue is, however,
the high number of data/signals to be dealt as a result of measurement with several
sensors. It is important, therefore, to assure proper pre-processing of the signals
30
from sensors, appropriate data fusion and optimal data analysis.
The multivariate analysis systems can be implemented for various applications.
Exploratory analysis are the most basic, but also most useful for preliminary data
screening and identification of trends within data.
Very important task is to continuously upgrade the models. In a perfect case, the
software system should acquire new knowledge automatically along the service.
It is assumed that, after additional developments, such methodologies can serve
as assisting tools for non-destructive assessment of the wooden structures, service
life prediction of structural elements and to support selection of optimal conservation
process.
31
32
Hierarchical modelling of timber reference
properties using probabilistic methods:
Maximum Likelihood Method, Bayesian
Methods and Probability Networks
Hélder S. Sousa ([email protected])a*, Jorge M. Branco
a
a
([email protected]) , Paulo B. Lourenço ([email protected])
a
ISISE, Department of Civil Engineering, University of Minho, Portugal
*
corresponding author: ISISE, University of Minho, Department of Civil Engineering,
Azurém, 4800-058 Guimarães, Portugal
Abstract
In recent decades, an increased interest has been evidenced in the research on multiscale hierarchical modelling in the field of mechanics, and also in the field of wood
products and timber engineering. One of the main motivations for hierarchical
modelling is to understand how properties, composition and structure at lower scale
levels may influence and be used to predict the material properties on a macroscopic
and structural engineering scale.
This chapter presents the applicability of statistic and probabilistic methods, such
as the Maximum Likelihood method and Bayesian methods, in the representation
of timber’s mechanical properties and its inference accounting to prior information
obtained in different importance scales. These methods allow to analyse distinct
timber’s reference properties, such as density, bending stiffness and strength, and
hierarchically consider information obtained through different non, semi or
destructive tests. The basis and fundaments of the methods are described and also
recommendations and limitations are discussed. The methods may be used in several
contexts, however require an expert’s knowledge to assess the correct statistic fitting
and define the correlation arrangement between properties.
1 Application
Wood is a natural material that by itself has its own hierarchical structure, which
is defined as the number of levels of scale with recognized structure. Hierarchical
modelling requires the distinction and differentiation between different scales,
33
such that a homogenization step may be taken to each of those scales in order to
define similar properties within a given scale. In [1], the hierarchical levels of
timber were defined regarding the structural member as the main unit of analysis.
In that case, three levels were defined: micro (material), meso (local) and macro
(global). At the material level, several attempts have been made to hierarchically
model the stiffness and strength of timber elements, by considering the presence
of weak sections separated by segments of clear wood [2-5].
Early, in [6], a stochastic model of hierarchical series system was used to
represent the bending strength of spruce regarding the anticipation of failure
in a weak section with defects. The model parameters were defined
regarding Maximum Likelihood estimates, and assuming that the estimated
parameters are applicable in the series system model for the full uncut beams, a
theoretical bending strength distribution function was obtained in dependence
of number of defect clusters within the span of constant bending moment
loading. A strong test of the prediction power of the model was established by
experiments with 54 long beams from the same population of beams from which
the small test pieces were cut. Also in [4], a hierarchical model was built for the
multi-scale variability of modulus of elasticity (MOE) that included an explicit
representation of the stiffness variability between timber boards and the
stiffness variability within boards. All parameters of the hierarchical stiffness
model were estimated based on a sample of 30 randomly selected timber
boards within the strength class L25 of Norway spruce grown in southern
Germany. The elements were differentiated along its length in weak sections
and in clear wood sections and a model was proposed by definition of the mean
modulus of elasticity within an element and the differences between that
mean and the results within sections of the same element and between other
elements. These values were modelled by probabilistic distributions with the
parameters obtained using the Maximum Likelihood method.
In [5], a hierarchical model for inferring on the reference properties of timber was
proposed by considering the distinction between clear and knot wood zones. This
work, however, presented a framework for timber members in service and thus
differentiated from the previously mentioned. The model procedure was based in
three main steps: i) visual identification of clear and knot wood zones; ii)
nondestructive prediction of the properties of clear wood zones; iii) prediction of
the reference materials using clear wood properties and applying a knot factor for
predicting the strength reduction effect of knots on clear wood properties. The
application of this procedure to maritime pine beams evidenced a good relationship
between experimental and predicted global modulus of elasticity (r2 between 0.76
and 0.55). Nevertheless, for bending strength weaker results were obtained, evidencing
the need for improvement in the method for determining the strength reduction
effect of weak zones.
In [7], Bayesian methods were used to update the mechanical properties of timber
and reliability assessment was performed using First Order Reliability Methods
34
(FORM). The results showed that different degrees of belief in the updating data
may significantly influence the reliability level. The used updating data were derived
from non-destructive test (NDT) results obtained with ultrasound, resistance drilling
and pin penetration equipments. The tests were conducted on chestnut wood
specimens, and were combined with results from compressive strength parallel to
the grain tests. The uncertainty of the different NDT results was also modelled by
Maximum Likelihood estimates.
Hierarchical modelling has also been carried out by use of Bayesian Probabilistic
Networks (BPN) for the analysis of variability of timber mechanical properties [810]. BPNs are used to represent knowledge based on Bayesian regression analysis
describing the causal interrelationships and the logical arrangement of the network
variables. In [8], a hierarchical model was used to determine the influence of the
origins (different tree growth locations) and cross-sectional dimensions of timber
elements on the probability distribution of its material properties. On that work,
BPNs were used to describe and inference on the dependence of different origins
and dimensions of sawn structural timber on the relevant timber material properties
conditional on indicator values assessed by machine grading indicators (Fig. 1).
Following, the parameters of the prior probability distribution functions as well as
the regression parameters were estimated as random variables with mean values,
standard deviations and correlations using the Maximum Likelihood method.
cross-section
dimension
influence
variables
material
properties
NDT
indicators
origin
minimum tension
strength
Modulus of
elasticity
mean density
strength
indicator
stiffness
indicator
density
indicator
Fig. 1 Example of a proposed BPN for inference on timber material properties, adapted from [8].
35
The work in [10] considered the findings and results of a multi-campaign
experimental campaign [11] as input data for the construction of the hierarchical
levels and its interrelationships within a BPN inferring on the bending
stiffness and strength of different size timber elements. In that case, the
probabilities within the BPN were updated through Bayes' theorem regarding
the belief propagation within the arrangements of nodes of different considered
BPNs. The results were modelled by posterior distribution probabilistic
parameters, accounting the results of Maximum Likelihood estimates and χ2
tests initially performed to the global sample. In that work, BPNs were
developed in order to infer the influence of local data (smaller size specimens),
with respect to both visual inspection and bending tests, on the results of
stiffness and strength of structural size elements. An example of a studied BPN
is presented in Fig. 2, with the results of different prior information in visual
inspection grading given in Fig. 3 by cumulative frequency functions. Visual
grading was considered as parent node in the analysis, as it provides a link
between scales and, also as it is commonly an available parameter in the
assessment of existing timber structures.
Visual grading VI
Structural size
S
VI Beams
VIB
MOE bending
Em,l
VI Boards
VIb
Fig. 2 Example of a proposed hierarchical model for inference on bending modulus of elasticity
(MOE) regarding different scales on visual grading, adapted from [10].
36
VIboard = I
VIbeam =
75
no
evid.
frequency (%)
100
I
50
II
III
25
NC
0
0
10000
20000
2
MOE (N/mm )
Fig. 3 Cumulative frequency distributions results obtained through a hierarchical BPN inferring
on bending stiffness regarding different evidences on visual grading, adapted from [10].
When assessing a timber element by visual inspection several works have analyzed the accuracy obtained by automatic systems and manual graders, and also
comparing the effectiveness and subjectivity of different inspectors [9,12-13]. In
[9], BPNs were applied to infer on different scales of visual inspection accounting the
information provided by inspectors with different levels of expertise. In that case, the
use of BPNs permitted to individually assess the accuracy in stiffness prediction of
different level of inspectors, and also by combination of their information. By use
of a parallel combination for prior information, it was evidenced the significance in
combining information by several inspectors even with similar or inferior individual
accuracy. Moreover in that study, due to the large variability in the visual grading
process, a second opinion improved the global efficiency, even if provided by a
less experienced inspector.
2 Methodology
2.1 Maximum Likelihood method
A possible way of defining the Maximum Likelihood method may be taken by the
following premises [1]. Considering that the parameters θ = (θ 1, .
T
, θ n) of the distribution of X are known, the joint probability of a random
sample X1, X2, ..., Xn can be written as:
37
f X ( x | θ ) = f X1 , f X 2 ,..., f X n (x1 , x2 ,..., xn | θ ) = f X1 (x1 ) f X 2 (x2 )... f X n (xn ) =
= ∏ f X (xi | θ )
(1)
i=1
However, it is often the contrary situation that is present in engineering
applications, such that a sample xˆ1 , xˆ2 ,..., xˆ n is observed and the distribution
parameters are unknown. In that sense, Eq. 1 can be understood as a relative
measure for the likelihood that the distribution determined by the parameters θ
is appropriate in the statistical definition of the sample x̂ . Along the full domain
of all possible parameters θ the likelihood L(.) that the parameters belong to the
sample is:
n
L(θ | xˆ1 , xˆ 2 ,..., xˆ n ) = ∏ f X (xˆ i | θ )
(2)
i =1
The Maximum Likelihood estimates can be defined by the parameters θ which
maximize the likelihood function L(.) over the domain of θ, thus being assumed as
the most likely to represent the data sample, as:
θ = max L(θ | xˆ1 , xˆ 2 ,..., xˆ n )
θ
(3)
The Maximum Likelihood method besides being used to fit the statistical parameters
in distribution functions can also be used to fit the parameters in linear and
non-linear regression analysis [14]. Also when considering a sample of results
taken from tests a linear regression may be estimated including an uncertainty
parameter or also called lack-of-fit parameter.
For parameter estimation for linear regression lines, the following linear regression
model in x1, ..., xm – space is considered:
y = α 0 + α 1 x1 + ... + α m x m + ε
(4)
where α0, α1, ..., αm are the regression parameters and ε models the lack-of-fit. ε is
assumed to be Normal distributed with expected value 0 and standard deviation
σ ε . It is assumed that n sets of observations or test results of (x,y) are available
and denoted as: (x1, y1), ..., (xn, yn). The regression parameters are determined
using a Maximum Likelihood method. The likelihood function is written with xij
being the j th coordinate of the i th observation:
m
L(α 0 , α1 ,...,α m ) = ∏ P( yi = α 0 + α1 xi1 + ... + α m xim + ε )
i =1
38
(5)
or, as in this case if it is used that ε is Normal distributed and σ ε is included as a
parameter to be estimated, then it follows:
m
1
i =1
2πσ ε
L (α 0 , α 1 ,..., α m , σ ε ) = ∏
 1  y − (α + α x + ... + α x )  2 
0
1 i1
m im   (6)
exp  −  i
 
 2
σ
ε

 

The log-likelihood function becomes:
ln L (α 0 , α 1 ,..., α m , σ ε ) = − n ln
(
m 1  y − (α + α x + ... + α x ) 
0
1 i1
m im 
2πσ ε − ∑  i

2
σ
i =1 
ε

)
2
(7)
Finally, the optimal parameters are determined from the optimization problem:
max
α 0 ,α 1 ,..., α m ,σ ε
ln L (α 0 , α 1 ,..., α m , σ ε
)
(8)
2.2 Bayesian Probabilistic Networks
BPNs are used to represent knowledge upon a system based on Bayesian
regression analysis describing the causal interrelationships and the logical
arrangement of the network variables. BPNs are represented by directed acyclic
graphs (DAG), composed by a set of nodes, representing each system variable,
connected by a set of directed edges, linking the variables according to their
dependency or cause-effect relationship. The causal relationship structure of a
BPN is often described by family relations that differentiates child node variables
with ingoing edges (effects), from parent node variables with outgoing edges
(causes) [15]. A (parent) node without any ingoing edges, thus without any parent
node converging to it, is often called a root node. The direction-dependent
criterion of connectivity evidences the induced dependency relationship between
variables and is classified as converging, diverging or serial (or cascade),
according to its arrangement [16]. Each variable node represents a random
variable, either defined as a continuous random variable or as a finite set of
mutually exclusive discrete intervals. In a BPN it is possible to coexist different
nodes with either continuous or discrete variables, in so called hybrid BPNs.
The main objective of a BPN is to calculate the distribution probabilities
regarding a certain target variable, by considering the factorization of the variables'
joint distribution based on the conditional relations within the developed generic
algorithm. In this light, the DAG is the qualitative part of a BPN, whereas the
conditional probability functions serve as the quantitative part. Therefore, the
algorithms themselves are indifferent to the scope for which the BPN is
employed, and thus have been employed in several different real-world problems,
39
besides the hierarchical modelling of reference properties, such as in diagnosis,
forecasting, automated vision, sensor fusion, manufacturing control, and
information retrieval [17]. A review of application of BPNs in environmental
modelling is found in [18], while a review in BPNs applications on dependability,
risk analysis and maintenance is provided by [19]. The applicability and
framework for construction of BPNs in the field of reliability analysis has been
addressed in e.g. [20-21]. In the case discrete states are used, each random variable
is defined by conditional probability tables, with the exception of nodes without
parents which, in that case, are defined by their marginal probabilities. Taking as
example a converging BPN with two parent nodes (A and B) with corresponding
marginal probabilities P(Ai) and P(Bi) for each given state i (i = 1, 2, 3,... n with n
= number of states), the conditional probability of the child node (C) given states
of A and B is calculated as:
P(C A, B)
(9)
The joint probability of all nodes is then calculated by the multiplication of
the conditional probabilities of the individual nodes, as:
P(C, A, B) = P(C A, B) ⋅ P( A) ⋅ P(B)
(10)
The marginal probabilities of the child node C are obtained by the sum of
the individual joint probabilities in every state, as:
P (C ) =
∑ ∑ P (C , A , B )
A
(11)
B
One of the main advantages of BPNs is that information may be easily
implemented to the network allowing for an update of the target variable. By
instance, if information about the state of a parent node is known with certainty,
then is referred that an evidence, e, is given in that state. Back to the example
of the converging BPN, considering that information is given to the state of
parent node A by evidence eA as it belongs to state 1 (i = 1), therefore A = A1 and
the probability P(A1) = 1, the probability distribution of the remaining variables
of the network can be updated following Bayes theorem, as:
P(C , B A1 ) =
P (A1 C , B ) ⋅ P(C , B )
P( A1 )
(12)
By application of Eq. 12 the posterior joint probabilities are obtained regarding
the prior given evidence. The previous methodology can also be extended to
converging BPNs with more than two parent nodes, or even to diverging or serial
BPNs, being most often found that complex engineered systems are composed
by the combination of smaller BPNs with these different arrangements.
40
3 Recommendations
In the Bayesian probability methods, probabilities are considered as the best
possible expression of the degree of belief in the occurrence of a certain event.
The Bayesian probabilistic approach does not consider that probabilities are direct
and unbiased predictors of occurrence frequencies that can be observed in
practice. The only consideration is that, if the analysis is carried out carefully, the
probabilities will be correct if averaged over a large number of decision situations
[22]. To fulfil that consideration it is necessary that the subjective and purely
intuitive part is neither systematically over conservative nor over confident.
Therefore, calibration to common practice on the average and to empirical data
may be considered as an adequate path to that aim.
In the case of BPNs, it is recommended that the parents nodes are composed
by indicators with strong correlations with the child node (reference property in
analysis) given, as instance, by high coefficients of determination. After determining
the indicators with higher predictive power, the dependencies within the DAG are
created with different levels of hierarchy according to expert decision. The
levels of hierarchy should attend to the source of the data, its relevance and both
its size and material scale.
When considering decay in timber elements, dynamic BPNs should be
implemented as to incorporate a time dimension, mainly by adding a direct
mechanism for representing temporal dependencies among the variables, see e.g.
[23-24]. Dynamic BPNs have also been extended to the modelling of deterioration
as reported in [25], while aspects of optimization inspection and maintenance
decision regarding deterioration have also been addressed by BPN analysis in e.g.
[26-28].
In a probabilistic analysis, as the inference on characteristic values is of special
interest in the field of structural safety assessment, it is also recommended that
special focus is given to the extreme values of the distributions. Therefore a
scheme for estimating the parameters of probability distributions focusing on the
tail behaviour should also be addressed, as considered in [29] where a censored
Maximum Likelihood estimation technique was used.
4 Limitations
Bayesian methods allow quantifying an approximation about the statistical
uncertainty related to the estimated parameters, regarding both the physical
uncertainty of the considered variable, as well as the statistical uncertainty related
to the model parameters. Therefore, they offer a suitable method for parameter
41
estimation and model updating. However, for making this possible, it is necessary
to take into account the measurement and the model uncertainties in the
probabilistic model formulation. Since Bayesian methods grant the opportunity to
incorporate different considerations about the uncertainty of models in the
upgraded stochastic model, the comparison between different experts’ results may
be regarded as a limitation, such that a consensus about a comparison basis has not
yet been established. Nevertheless, in the case of the construction of a BPN,
experts often promptly assert the causal relationships among variables in a domain,
without pre-ordering the variables in different levels. In almost all cases, by doing
so, results in a BPN which conditional-independence implications are accurate
[30]. Another limitation of Bayesian methods is the overall requirement of a
sufficient large sample for a reliable analysis. In a parallel BPN, it should also be
noted that a small sample may lead to the situation that the joint probability
factorization may not be possible due to the non occurrence of a given intersection
of evidences. In this case, when using discrete variables the construction of a BPN
is highly dependent of the choice of the intervals’ range.
Both the Maximum Likelihood method and Bayesian methods are statistic tools
that may consider data from different sources, such as different NDTs, SDTs or
destructive tests, or even their combination. However, these data must be classified
and arranged with respect to its relevance and dependability, in order to obtain an
adequate hierarchical modelling and inference on different reference properties of
timber.
References
1. Köhler J (2007) Reliability of timber structures. PhD thesis, Institute of Structural Engineering
Swiss Federal Institute of Technology, Zurich, Switzerland
2. Riberholt H, Madsen PH (1979) Strength of timber structures, measured variation of the cross
sectional strength of structural lumber. Report R 114, Structural Research Lab., Technical
University of Denmark
3. Isaksson T (1999) Modelling the variability of bending strength in structural timber. Report
TVBK-1015, Dept. of Structural Engineering, Lund University, Sweden
4. Fink G, Köhler J (2011) Multiscale variability of stiffness properties of timber boards. In: Faber, Köhler, Nishijima (ed) Applications of Statistics and Probability in Civil Engineering.
Taylor & Francis Group, pp 1369-1376
5. Machado JS, Palma P (2011) Non-destructive evaluation of the bending behaviour of inservice pine timber structural elements. Mater Struct 44(5):901-910
6. Ditlevsen OD, Källsner B (1998) System effects influencing the bending strength of timber
beams. In: Proceedings of 8th IFIP WG 7.5 Working Conference, Krakow, Poland, pp
129-136
7. Sousa HS, Sørensen JD, Kirkegaard PH, Branco JM, Lourenço PB (2013) On the use of NDT
data for reliability-based assessment of existing timber structures. Eng Struct 56:298-311
42
8. Deublein M, Schlosser M, Faber MH (2011) Hierarchical modelling of structural timber material
properties by means of Bayesian Probabilistic Networks. In: Faber, Köhler, Nishijima (ed)
Applications of Statistics and Probability in Civil Engineering. Taylor & Francis Group, pp.
1377-1385
9. Sousa HS, Branco JM, Lourenço PB (2013) Effectiveness and subjectivity of visual inspection
as a method to assess bending stiffness and strength of chestnut elements. Advanced Materials
Research, vol 778. Trans Tech Publications, pp 175-182
10. Sousa HS (2013) Methodology for safety evaluation of existing timber elements. PhD thesis,
Civil Engineering Department, University of Minho, Guimarães, Portugal
11. Sousa HS, Branco JM, Lourenço PB (2014) Use of bending tests and visual inspection for
multi-scale experimental evaluation of chestnut timber beams stiffness. J Civ Eng Manag (in
press)
12. Lycken A (2006) Comparison between automatic and manual quality grading of sawn
softwood. Forest Prod J 56(4):13-18
13. Grönlund U (1995) Quality improvements in forest products industry: classification of
biological materials with inherent variations. PhD thesis, Luleå University of Technology,
Swe-den
14. Sørensen JD (2003) Statistical analysis using the Maximum-Likelihood Method, Aalborg
University, Aalborg, Denmark
15. Bayraktarli YY, Ulfkjaer J, Yazgan U, Faber MH (2005) On the application of Bayesian
probabilistic networks for earthquake risk management. In: Proceedings of 9th international
conference on structural safety and reliability, Rome, Italy
16. Pearl J (1988) Probabilistic Reasoning in Intelligent Systems: Networks of Plausible
Inference. Morgan Kaufmann Pub. 552pp
17. Heckerman D, Mamdani A, Wellman MP (1995) Real-world applications of Bayesian
networks. In: Communications of the ACM 38(3):24-26
18. Aguilera PA, Fernández A, Fernández R, Rumí R, Salmerón A (2011) Bayesian networks in
environmental modelling. Environ Modell Softw, 26(12): 1376-1388
19. Weber P, Medina-Oliva G, Simon C, Iung B (2012) Overview on Bayesian networks applications
for dependability, risk analysis and maintenance areas. Eng Appl Artif Intel 25(4):671-682
20. Langseth H, Portinale L (2007) Bayesian networks in reliability. Reliab Eng Syst Saf
92(1):92-108
21. Marquez D, Neil M, Fenton N (2010) Improved reliability modelling using Bayesian
networks and dynamic discretization. Reliab Eng Syst Saf 95(4):412-425
22. Vrouwenvelder ACWM (2002) Developments towards full probabilistic design codes. Struct
Saf 24(2):417-432
23. Allen JF (1981) An interval-based representation of temporal knowledge. In: Proceedings
of 7th International Joint Conference on Artificial Intelligence, Vancouver, Canada, pp
221-226
24. Ghahramani Z (1998) Learning dynamic Bayesian networks. In: Adaptive processing of
sequences and data structures. Springer Berlin Heidelberg. pp 168-197
25. Straub D (2009) Stochastic modelling of deterioration processes through dynamic Bayesian
networks. J Eng Mech 135(10):1089-1099
26. Friis-Hansen A (2000) Bayesian networks as a decision support tool in marine applications.
Department of Naval Architecture and Offshore Engineering, Technical University of Denmark 27. Attoh-Okine NO, Bowers S (2006) A Bayesian belief network model of bridge
deterioration. In: Proceedings of the ICE-Bridge Engineering, 159(2):69-76
43
28. Montes-Iturrizaga R, Heredia-Zavoni E, Vargas-Rodríguez F, Faber MH, Straub D (2009)
Risk Based Structural Integrity Management of Marine Platforms Using Bayesian Probabilistic
Nets. J Offshore Mech Arct Eng 131
29. Faber MH, Köhler J, Sørensen JD (2004) Probabilistic modelling of graded timber material
properties. Struct Saf 26(3):295-309
30. Heckerman D, Breese JS (1996) Causal independence for probability assessment and
inference using Bayesian networks. In: IEEE Transactions on Systems, Man and Cybernetics,
Part A: Systems and Humans 26:826-831
44
Quantitative assessment of the load-bearing
capacity of structural components using NDT,
SDT and DT inspection methods
Gerhard Finka and Jochen Kohlerb
1
a
ETH Zurich, Institute of Structural Engineering, Zurich, Switzerland
b
NTNU, Department of Structural Engineering, Trondheim, Norway
Introduction
In timber constructions a significant number of failures and damages have been
detected within the last decades, well-known documentations are e.g. Frühwald et
al. (2007), Blaß & Frese (2010) and Kohler et al. (2011). In many cases, only a
partly damage of the construction accord (e.g. a single structural component, a
single connection, a group of components, etc.), whereas the remaining (nonfailed) structural members are apparently unaffected - they might be also damaged
due to additional loading caused by load redistribution.
In such cases the corresponding engineer has to make a decision for the repair
alternatives. This might be an exchange of the failed member(s), a reinforcement
of the non-failed members or in the worst case a complete renovation of the entire
construction. However, to make the optimal decision, it is essential to estimate the
load-bearing capacities of the remaining structural members; thereby it has to be
considered that the load-bearing capacity is composed of the load-bearing capacity
at the time of construction and the deterioration during utilization (Fink & Kohler
2014b).
In many situations the estimation of the remaining load-bearing capacity is
rather complicated and connected to large uncertainties. However, under the
consideration of information available, such as the target material properties, the
age of the building, the history of load (e.g. amount and duration of load) or the
amount of the damage, a first estimation can be made. Often this first estimation is
not sufficient to make a final decision. In such cases different non-destructive,
semi-destructive and destructive inspection methods (referred to as NDT, SDT and
DT) can be performed to enhance the estimation.
In this chapter a summary about the quantitative assessment of the load-bearing
capacity of structural components based on information available and the results
of different NDT, SDT and DT inspection methods using Bayes updating is
presented (for a more detailed description see Fink & Kohler 2014a). At first
information is classified according their characteristics in respect to a qualitative
45
assessment of the load-bearing capacity of structural components. Afterwards, the
corresponding updating-procedure is introduced. The application of the updating
procedure is illustrated on two selected examples. In the last part of this chapter,
the application of Bayes updating for the decision support is introduced.
2
Updating
For the estimation of the actual load and the resistance of a structural system, the
way of treating available information is of particular importance. To do this, the
structural model (static system), the applied load, the geometrical properties and
especially the material properties of the structural components have to be considered.
One handsome approach to combine different types of information is the socalled Bayes updating. Using Bayes updating, given information (so-called prior
information) will be updated with additional information; e.g. the results of a NDT
inspection. The prior information can be the planed conditions (if available), such
as the target material properties or (if nothing better available) the assessment of
an expert.
The Bayes approach is related to the quality of the prior information and the
characteristics of the information used for updating. For this purpose, the
information is classified according the characteristics necessary for Bayes
updating. Thereby it can be distinguished between so-called equality type
information and inequality type information. Equality type information are
measured variables, whereas inequality type information denotes information
that some variable is greater than or less than some predefined limit. Furthermore,
it can be differentiated between direct information (direct measurements of the
quantity of interest) and indirect information (measurement of some indicator of
the quantity).
For the different types of information the corresponding updating procedure is
introduced (according to Rackwitz 1983, Faber 2012, Faber et al. 2000, Fink &
Kohler 2014b). The general scheme for updating the parameters having equality
type information is given in Eq. (1). The inspected parameter, here the loadbearing capacity of the structural members, is represented by the variable X with
the probability distribution function FX(x). The parameters θ = ( 1, 2, ..., n)T of
the distribution function are not precisely known; they are product of engineering
knowledge, physical understanding or earlier observations of the quantity. In general
the parameters θ are expressed as random variables specified by the so-called prior
density function f Q' ( θ ) . The uncertain parameters θ can be updated with new
T
information (new observations of realizations of the variable X,x^ = ( x^1, x^2 ,..., x^n ) ).
^
f Q' ' ( |x)
^ denote the posterior distribution function of the parameters θ, L(θ x) de-
46
note the likelihood function (representing the knowledge gained by the
new information), and n is the number of observations.
f ' '' ( x ) =
∫ f X ( x θ) f '' (θ x^)d θ
Q
f Q'' (θ x^ ) =
f Q' ( θ)L(θ x^ )
∫
f Q' (θ )L(θ x^ ) θ
(1)
The load-bearing capacity of the structural members can also be updated having
inequality type information. Assuming a structural member will be proof loaded up
to specific stresses effect σ l without failure. Thus the load-bearing capacity of the
structural members can be represented by the variable X following a truncated
distribution function. In Faber et al. (2000) the following approach is proposed to
calculate the load-bearing capacity FR'' (r); here FR' (r) is the prior distribution function of the resistance:
F ' (r ) − FR' (σ l )
FR'' (r ) = R
1 − FR' (σ l )
r ≥ σl
(2)
The principle of updating using equality type information and updating using
inequality type information is illustrated in Fig. 1 on two examples. In both
examples, the bending strength of GLT beams (strength class GL24h) is
updated. One time with equality type information and one time with inequality
type information. The characteristic value of the bending strength fm of strength
class GL24 is fm,k = 24 MPa. The bending strength fm is assumed to be lognormal distributed with COV = 0.15, in accordance to JCSS (2006). Thus, the
logarithm of the bending strength is normal distributed: ln(fm) ~ N(µ’, σz), with
σz ≈ COV = 0.15. In Fig. 1 (left), the bending strength fm of GLT beams is
updated with the results of three bending tests fm,i = 22, 30, 35 MPa (equality
type information). All three test results are within the expected range, but slightly
below the expected value. As a result the predictive bending strength of the not
tested GLT beams is slightly reduced, in particular within the upper tale of the
distribution function. In the second example (Fig. 1, right), the bending strength
f m of the GLT beams is updated after proof loading. The load is applied
constantly over the entire construction and corresponds to σl = 22 MPa. In this
example no GLT beam failed under this load and thus it is obvious that the loadbearing capacity of all GLT beams is at least equal to the specific load effect:
fm > σl = 22 MPa. As a result, in Fig. 1 (right), the lower tail of the predictive
distribution function is truncated.
47
Fig. 1 Schematic illustration of the principle of (left) updating using equality type information,
and (right) updating using inequality type information
3
Decision support
The presented summary about the quantitative assessment of the load-bearing
capacity of structural components using NDT, SDT and DT inspection methods
concludes with a discussion about the application of Bayes updating for the
decision support. Two fields of application are discussed:
- Support the corresponding engineer to choose the optimal
inspection methods: Using the Bayes updating combined with the
assumption of possible outcomes the optimal inspection method as
well as the optimal number essential
test can be estimated. Such pre-investigations can be very
efficient for the choice (type and amount) of inspection methods.
As a result useless inspections can be avoided and the total cost of
the renovation (inspection costs and repair costs) can be minimised.
- Support the corresponding engineer to find the optimal decision:
Under consideration of the information available and all test results
(NDT, SDT, and DT) the load-bearing capacity of the non-failed
structural members can be estimated using Bayes updating. This can
support the corresponding engineer to make a final decision.
References
Blaβ H.J. & Frese M. (2010). Schadensanalyse von Hallentragwerken aus Holz. Karlsruher Berichte zum Ingenieurholzbau 16, KIT Scientific Publishing, Karlsruhe, Germany
48
Faber M.H. (2012). Statistics and Probability Theory: In Pursuit of Engineering Decision
Support, vol 18. Springer Verlag
Faber M.H., Val D.V. & Stewart M.G. (2000). Proof load testing for bridge assessment and
upgrading. Engineering Structures 22(12):1677-1689
Fink G. & Kohler J. (2014a). Quantification of different NDT/SDT methods in respect to
estimate the load-bearing capacity. under preparation
Fink G. & Kohler J. (2014b). Risk based investigations of partly failed or damaged timber
constructions. In: Materials and Joints in Timber Structures, Springer, pp 67-75
Frühwald E., Serrano E., Toratti T., Emilsson A. & Thelandersson S. (2007). Design of safe
timber structures - How can we learn from structural failures in concrete, steel and timber?
Tech. rep., Lund University, Sweden
JCSS (2006). Probabilistic Model Code Part III - Resistance Models (3.05 Timber). http://
www.jcss.byg.dtu.dk/Publications/Probabilistic Model code
Kohler J., Fink G. & Toratti T. (eds) (2011). Assessment of Failures and Malfunctions. Shaker
Publishing Company, Aachen, Germany
Rackwitz R. (1983). Predictive distribution of strength under control. Materiaux et Construction
16(4):259-267
49
50
PART II
ASSESSMENT OF REFERENCE PROPERTIES
51
52
PART II- ASSESSMENT OF REFERENCE PROPERTIES
Methodology and protocols for routine
assessment of wooden members with
spectroscopy
Anna Sandak1, Jakub Sandak1, Mariapaola Riggio2
1 CNR Ivalsa, S.Michele all’Adige (TN), Italy,
[email protected], [email protected]
2 Wood Science & Engineering, Oregon State University, USA,
[email protected]
Abstract
Current procedures for characterization of wood members on-site are often limited
to few characterizations (visual inspection supported by the localized resistance
analysis and moisture content estimation). The development of electromagnetic
wave-based methods, such as spectrometric techniques, electric and optical
methods, as well as the increasing availability of portable instruments, has opened
up new perspectives for on-site characterization and monitoring of building
materials. The objective was to highlight the potential of the infrared spectroscopy
as a tool capable of providing complementary information for the expert inspection
assessing the timber structures. Advantages and drawbacks of the techniques are
illustrated together with recommendations regarding samples presentation, correct
measurement as well as data evaluation. Finally examples of sucessful application
of infrared spectroscopy for assessment of selected wood properties are presented.
Introduction
Nondestructive testing (NDT) methods find a particular place among analytical
methods used for structure assessment due to the limited amount of required sample,
or in case of portable instruments possibility of performing measurement directly
on-site. Unfortunately many of instrumental methods, even possessing plenty of
advantages, are rarely applied on-site due to the lack of standardized procedures.
Infrared spectroscopy, in both near and mid ranges, is a well known technique
with a great potential for chemical characterization of materials. It is useful for
identification of various organic compounds, on the base on their selective absorption
of radiation in the infrared region. As an effect of this phenomenon, the infrared
radiation reflected from the surface can be used for estimation of the physicchemical structure of the surface. Several researchers are focused on the evaluation
of physical and mechanical properties of wood or estimation its degradation level.
The objectives of this research are to highlight potentials and limitations of
53
proposed techniques and to provide list of requirements for correct implementation of
spectroscopy in routine assessment of wooden members.
Strength and limits of IR spectroscopy
Traditional infrared (IR) technique is used to analyze solid, liquid or gases by means
of transmitting the infrared radiation through the samples. The Attenuated Total
Reflectance (ATR) allows measurement without necessity of complicated sample
preparation and furthermore provides good spectral reproducibility. The advantages
of this technique are:
• little or no sample preparation;
• possibility to measure samples that are too thickor too opaque for traditional
transmission IR;
• relatively fast measurement (minutes, depends on resolution);
• possibility of measurements solids, powders or liquids;
• possibility for determination of many components simultaneously;
• high degree of precision and accuracy;
• information related to chemical fingerprint;
• direct measurement with very low cost.
The restrictions of this technique are related to limited dimension of samples,
pH constrains of used crystals and necessity of good contact between the sample
and the crystal.
The FT-NIR technique is relatively simpler and possesses some very important
advantages (in comparison to other analytical methods):
• simple sample preparation (to assure controlled MC and surface finish)
• non-destructive or semi-destructive testing (fast screening method applicable on site
and more accurate estimation method with controlled samples in the laboratory)
• relatively fast measurement (seconds, depends on resolution);
• no residues/solvents to waste
• possibility for determination of many properties simultaneously
• high degree of precision and accuracy
• direct measurement with very low cost
The most important limitation of the FT-NIR is that the spectra are rather
complicated and includes a complex overlapping of different overtones corresponding
to vibrating functional groups. Consequently, data evaluation is necessary for proper
interpretation and understanding of results. The resolution of the spectrometer is also
limited, thus complicating the spectra interpretation even more. Moreover sensitivity
to moisture variations, surface preparation, aging of surface affects measurement.
It must be also highlighted that in case of assessment of timber structural members,
with a certain thickness and length, sampling criteria must be defined so that global
54
characteristics can be reliably inferred from a number of local (and superficial)
measurements.
Methodology and protocols for routine assessment of
wooden members with spectroscopy
Sampling and sample preparation
Representative reference sampling is essential especially when heterogeneous
material such as wood is investigated. Precise sampling criteria must be defined in
case of assessment of timber structural members. It has to be optimized depending
on the structure topology, member size, state of preservation and due to the
purpose of measurement. It has to be also highlighted that any measurement of mid
or near infrared spectra may provide information related only to the local wood
characteristics and is limited to the subsurface of the member. Sample preparation
and presentation significantly affect reliability of presented techniques. Milling
procedure, particle size, and quality of the solid wood surface influence the
performance of the models to predict chemical properties. Recommendations
regarding milled wood samples were presented by Schwanninger et al. [1]. Also
the effects of sample presentation (solid or milled) affect reliability and quality of
the models [2]. It is also important to minimize influence of temperature and
relative humidity of the environment during measurement. Therefore, if it is
feasible to take small amount of material from the member on site, it is highly
recommended to condition it in climatic chamber prior spectroscopic
measurements. Recently it is also possible to perform measurements directly onsite, especially with modern, portable instruments. Particular attention should be
focused however on moisture content of measured wooden members and further
correction, for proper data mining and interpretation.
Measurement procedure/sample presentation
Until recently, the preparation of samples to be measured with infrared
spectrometers has been rather complicated; for example through solubilization of
the sample or preparation of potassium-bromide (KBr) pellets for sample analysis
(10 to 20mg of material). Fourier Transform Mid Infrared Attenuated Total
Reflectance (FT-MIR-ATR) is a relatively new advancement of traditional mid
infrared spectroscopy. It uses a phenomenon of absorbing infrared energy during
reflection from the measured surface. This technique allows measurement of the
wood powder or even wooden block surface without time consuming KBr sample
preparation. Some instruments equipped with external reflectance module allow
measurement of large objects without contact. This technique might be
particularlyuseful during on-site analysis by placing the instrument in front of the
analyzed object. Fourier Transform Near Infrared (FT-NIR) spectroscopy offers
55
even simpler measurement and its applicability for in-field measurement is good.
Due to low absorption coefficients, bulk or thick samples of intact cellular
structure can be measured [3]. Spectrometers equipped with fiber optic allow
direct measurement of samples at a certain distance from the instrument
(depending on fiber optic length). Use of fiber optic is most convenient approach
for acquiring near infrared spectra. The reference light is emitted from the probe
and reflected part of it is transmitted to the detector. The area of the detector varies
and may cover from 1 to 20mm2. As a result, the spectrum acquired is an
“average” from the surface area corresponding to the probe having direct contact
with measured object. No any extra pressure is required, even if the probe positing
(deviation from the perpendicular placement) may influence the spectra outline.
FT-NIR does not require sample preparation or hazardous chemicals, making it
quick and reliable for quantitative and qualitative analysis. It is ideal for rapid
material identification and is also a powerful analysis tool capable of accurate
multi-component quantitative analysis. Recently hyperspectral system able to
measure and characterize whole surface with high spatial resolution become more
used. Hyperspectral images add a new dimension (spatial resolution) to the field of
spectroscopy. They provide a means of accurately quantifying and locating
constituent variation within the field of view of the camera, in addition to the
identification and quantification of bulk constituents provided by integrating
spectrometers. The measurement distance vary from millimeters up to meter and
the spatial information is acquired directly by the spectrometer optics, by means of
controlled positioning of the sample. The spectral band of hyperspectral camera
may include various spectral ranges (UV, VIS, NIR, IR) as well as Raman scatter.
It provides the great possibility to characterize various physical-chemical
properties of the surface with high spatial resolution.
Collection of representative spectra
The routine testing procedure for the measurements should be determined through
a series of preliminary tests, in order to optimize the scanning procedure and
improve the quality of results obtained from wood samples. Due to anisotropy and
heterogeneity of wood, it is important to repeat measurements and average spectra.
It is recommended to measure different samples on the corresponding/ analogous
points (for example on the radial plane). According to Tsuchikawa and Schwanninger
[4] spectra collected from transverse and radial surfaces provide better prediction
than those from tangential surfaces. The measurement location can be selected
randomly; however, any visible abnormalities of wood surface (such as resin canal,
knot or discoloration) should be intentionally omitted, unless the measurement of
defect rate is intended.
56
Data evaluations
NIR absorption spectra are often complex and normally possess broad overlapping
absorption bands that require special mathematical procedures for data analysis. In
contrary MIR absorption bands are well-resolved, assignable to specific chemical
components. Moreover the signal-to-noise ratio of NIR is poorer than that of FTMIR-ATR and interpretation of spectra more problematical. However NIR spectra
contain a lot of information related to hydroxyl groups linked to several chemical
components, which, in case of wood, are very relevant also for the estimation of
physical and mechanical properties of the material. Interpretation of spectra is very
important and is highly recommended to include this step in routine analysis.
Recently published works provides valuable information in regards of bands
assignment [5-8]. Visual observation of spectra is also the easiest method to detect
outliers caused by errors during measurement (not parallel position of fiber optic
(in case of NIR) or incorrect contact between sample and crystal (in case of ATR).
Spectra identification and qualification (quantitative analysis) can be done by
comparing a sample spectrum to reference spectra of known materials (or in case
of decayed samples with reference samples infested with various fungi).
Quantification is done by using mathematical models and so-called multivariate
data analysis (MVA): these approaches are generally referred as chemometrics.
MVA techniques are statistical design tool for dealing with very large datasets,
which allow more than two variables to be analyzed at once. Multivariate data
analyses are usually divided into three groups: exploratory data analysis, regression
analysis and classification models. Exploratory analysis (data mining), attempts to
find the hidden structure in large complex data sets, examples are Cluster Analysis
or Principal Component Analysis. Regression analysis and Predictive Models, such
as Partial Least Squares Regression or Multiplicative Linear Regression, are used
for developing the models from available data and predict desired response. Even
if both spectral noise and reference method noise affect the accuracy and the
precision NIR predicted values sometimes model based on the noisy reference data
led to good results [9]. Classification Models (Cluster Analysis Test, Identity Test
or SIMCA) allow separation of group of objects into one or more classes, on the
base on distinguished characteristics. Figure 1 summarizes recommended protocol
for routine assessment of wooden members.
57
correct sampling
optimal sample
preparation
proper sample
presentation
collection of
representative
spectra
elimination of
outliers
spectra
interpretation
qualitative
analysis
quantitative
analysis
Fig. 1. Schema of proposed stages for wooden members characterization. Source: [10] modified.
Application of spectroscopy for assessment of wooden members
Spectroscopy allows understanding of chemical changes of the wooden material
during various degradation processes. It allowed classification of decay type and
prediction of modification of physical properties, as a consequence of the decay
process. It has been also successfully applied for monitoring the weathering
process of different wood species, understanding the weathering dynamic and
estimating both exposure time and service life of wooden structures [10]. Reports
related to characterization and evaluation of waterlogged wooden samples, both
from archaeological site and recent wood during short term waterlogging are
recently published [11-12]. Several researchers already proved applicability of
infrared spectroscopy for species recognition, prediction of moisture content,
density, tensile strength, mechanical stresses, bending MOE and MOR [4]. In
58
general, spectroscopy works with all wood species. Most of up-to-data prediction
models reported in the literature are valid for a single wood species only. However,
the development of generalized chemometric model suitable for several wood
species is of great interest [9]. The important concern is related to the timber
surface in use, including oxidation, ageing and/or weathering [13]. As a
consequence, chemometric models developed for one sample lot may not be
functional for same wood samples, but processed with different preparation
procedure. In such case, calibration transfer formula is necessary. It has to be
mentioned, that spectroscopic measurements are related to the surface, therefore
can not be straightly correlated with internal properties of the members. The
solution is to combine sampling with other methods (e. g. radial coring for
dendrocronological analysis or semi-destructive analysis) and acquire the spectra
along the members depth (e.g. radial profile) [14].
The development of electromagnetic (EM) wave-based methods, such as
spectrometric techniques, imaging techniques, electric and optical methods, as well
as the increasing availability of portable instruments, has opened up new perspectives
for on-site characterization and monitoring of building materials. Moreover the trend
for material characterization by using multiple sensors simultaneously has become
well accepted. It is more favorable than a single sensor approach due to far better
representation of the real-world cases. The speed of measurement is comparable
with other NDT for decay detection, such as dynamic indentation (Pilodyn®) and
infra-red thermovision, among others. The other challenge is however the correct
interpretation of measurement data assuring integration/fusion of all indicators as
the multisensory data is usually correlated with each other.
Conclusions
Practical application of spectroscopy for timber structure assessment provides very
essential supplement to the typical information collected traditionally with
standard procedures. Special attention is focused here to highlight potentials and
limitations of proposed techniques and to provide list of requirements for correct
implementation of spectroscopy in routine assessment. Up to now several
researchers confirmed advantages of spectroscopy for evaluation of wood
properties. However application of this technique for on-site wooden members
inspection requires a proper definition of sampling criteria, according to the
properties investigated, as well as prior preparation of dedicated databases of high
precision reference values. Those are essential to build reliable, flexible and
sufficiently generalized models. The method then might be a tool assisting experts
in the estimation of reference material properties, and other relevant mechanical/
physical characteristics, in a fast and repeatable way.
59
Acknoledgments
P art of the work was conducted within project BIO4ever (RBSI14Y7Y4) within call SIR funded
by MIUR.
References
[1]
Schwanninger M, Rodrigues JC, Pereira H, Hinterstoisser B. Effects of short-time
vibratory ball milling on the shape of FT-IR spectra of wood and cellulose. Vibrational
spectroscopy 2004;36(1):23-40
[2]
Hein PRG, Lima JT, Chaix G. Effects of sample preparation on NIR spectroscopic
estimation of chemical properties of Eucalyptus urophylla S.T. Blake wood.
Holzforschung 2010;64(1):45-54.
[3]
Fackler K, Schwanninger M. How spectroscopy and microspectroscopy of degraded wood
contribute to understand fungal wood decay. Appl Microbiol Biotechnol. 2012;96(3):
587– 599.
[4]
Tsuchikawa S, Schwanninger M A review of recent near infrared research for wood
and paper (Part 2). Applied Spectroscopy Reviews 2013;48:560-587
[5]
Rowell RM. Handbook of Wood Chemistry and Wood Composites, CRC Press. 2005
[6]
Workman J, Weyer L. Practical Guide to Interpret Near-Infrared Spectroscopy. CRC
Press, 2007.
[7]
Burns DA, Ciurczak EW. Handbook of Near-Infrared Analysis. CRC Press 2008
[8]
Schwanninger M, Rodrigues JC, Fackler K. A review of band assignments in near infrared
spectra of wood and wood components. J. Near Infrared Spectrosc. 2011;19(5): 287-308.
[9]
Rodrigues J, Alves A, Pereira H, da Silva Perez D, Chantre G, Schwanninger M. NIR
PLSR results obtained by calibration with noisy, low-precision reference values: Are
the results acceptable? Holzforschung 2006;60(4),402-408
[10] Sandak A, Riggio M, Sandak J. Non destructive characterization of wooden members
using near infrared spectroscopy. Advanced Materials Research, 2013;778:328-334
[11] Sandak A, Sandak J, Babiński L, Pauliny D, Riggio M (2014) Spectral analysis of
changes to pine and oak wood natural polymers after short-term waterlogging. Polymer
Degradation and Stability, 99:68-79
[12] Riggio M, Sandak J, Sandak A, Pauliny D, Babiński L (2014) Analysis and prediction
of selected mechanical/dynamic properties of wood after short and long-term
waterlogging. Construction & Building Materials 68:444-454
60
Structural Analysis Of Porous Media By Means
Of Thermal Methods : Theory And Monitoring
Equipment
Olivier CARPENTIER*, Thierry CHARTIER*, Emmanuel ANTCZAK*,
Thierry DESCAMPSº, Laurent VAN PARYSº
* Civil Engineering Department, LGCgE, University of Artois (Bethune, France)
° CESM Department, Univerisity of Mons, Mons (Belgium)
Abstract
Thermal methods are based on the knowledge of both theory of heat and mass
transfers and thermal sensors. To be effective, thermal analysis must be able to
make the logical connection between the theoretical thermal behaviour of a
medium under solicitations and the ability of sensors to measure thermal physical
quantities. Concerning porous media like wood, thermal analysis is very complex,
mainly because of the difficulty to accurately describe micro-structure geometry
and its influence in thermal behaviour of whole wooden structure. A description of
current thermal methods (non-destructive and semi-destructive testing methods)
are given for both laboratory and in situ experiments. Latest advances in thermal
analysis are also presented. In conclusion, examples of dedicated applications and
possible connections with some non-destructive testing methods give the necessary
information to identify perspective of development.
Keywords: Monitoring, Diagnosis, Numerical analysis, Thermal sensors
1. Introduction
Thermal metrology primarily responds to three needs [1]. The first is mastery of
manufacturing process in which a monitoring or an automated control solution is
needed. The second is to highlight thermal effects in a system for which there is
no obvious numerical solution (local heat transfers in huge system). The third
responds to the evaluation of thermo-physical properties of materials and interfaces
needs in order to know and improve materials. This evaluation can add new fields
of knowledge to numerical modeling of complex systems. The quantitative thermal
analysis falls into this last category.
61
2. Thermal metrology
Thermal metrology is based on the exploitation of measurable physical quantities,
i.e. temperature and heat flux, respectively expressed as Kelvin (K) and Watt (W)
in the international system of units. In addition to temperature and heat flux, two
other groups of physical quantities are defined : physical quantities linked to the
material itself and physical quantities linked to its interface with environment or
surrounding objects. Only temperature and heat flux can be directly measured via
thermal sensors. Evaluation of other physical quantities needs to master the resolution
of inverse problems.
Table 1. Example of several physical parameters that can be determined from thermal analysis
Physical parameters
Determined by...
Direct measure- The resolution of an inment
verse problem
Temperature
x
Heat flux
x
Thermal conductivity
x
Thermal diffusivity
x
Thermal effusivity
x
Evaluation through others physical parameters
Specific heat
x
density
x
Water content
x
Convection coefficient
x
Contact thermal resistance
x
Radiative surface
characteristics
x
Enthalpy
x
Viscosity
x
Porosity
x
62
Thermal analysis methods of thermo-physical properties of materials have the
same basic protocol. First, temperatures and/or heat flux are measured with
thermal sensors. Then, one or several thermo-physical properties (e.g. thermal
conductivity)associated with studied material are evaluated via the use of mathematical
formalisms [2]. Finally, several thermo-physical properties may be seen from
previous ones (e.g. water content given by thermal effusivity [3]). Solid, granular,
liquid, gaseous or other phase change materials (PCM) [4] can be studied.
Table 1. is given for a single material. But thermal methods can be extended to
composite materials. It may be possible to have global information of a system while
generally the purpose is to highlight the presence of thermal resistance inside the
medium (e.g. poor adherence [5] or cracks). There are a lot of thermal methods
because they cover a wide field of analyses:
- destructive (probes [6]) or non-destructive methods (infrared thermography,
IRT [7]),
- local (studies on micro and nano-components [8]) or huge areas can be
investigated (thermal Doppler for atmosphere analysis [9]),
- monitoring over time [10],
- methods can be used in laboratory or on site [11].
With regard to positive points, it should be pointed out that thermal methods
are very widely used, adapted to many materials and have applications on site.
The significant presence of these methods also involves the existence of a wide
variety of sensors to meet the most varied individual needs and budgets.
However, although thermal methods are widely used, they have some limits.
Disadvantages are related to the use of specific thermal method but there are
common denominators :
- Thermal methods are usually slow compared to other mechanical methods
(Thermal diffusion effects),
- Absolute accurate temperatures (accuracy of about 50mK) are difficult to
obtain,
- Global analysis. It is difficult to link information to a space discretization
(anisotropy),
- Quantitative analysis usually needs thermal solicitations that can modify
thermo-physical properties of the medium.
- While thermal sensors are steadily improving, heat flux measurement is
still difficult.
To solve these problems, the use of coupled mechanical or optical methods
with thermal method is usually the prescribed solution in order to
determine in particular thermo-physical properties.
63
3. Thermal methods and wooden structures
As stated above, there are a lot of thermal methods and it is not possible to provide
an exhaustive list. So, we focus on thermal methods suitable for wooden structures.
Wood is a porous material with a very complex microscopic architecture. Thermal
methods usually employed consider wood as a homogeneous medium for a specific
direction [12]. If on site applications were considered, non-destructive thermal
methods (NDTM) should be retained and could be active or passive. If wooden
structure could receive solar radiations (random solicitations), passive methods
would be considered [13]. Out of this specific context, NDTM have to use artificial
thermal solicitations in which a deterministic signal is used. Two kinds of methods
could be employed in laboratory or on site : thermal methods with contact and
contactless thermal methods.
3.1 Thermal methods with contact
Guarded hot plate test (laboratory) [14], hot wire or hot plane (laboratory) [15] and
the combined use of thermocouples and flux meters (in laboratory and on site) [16]
are widely used.
The principle of the procedure is based on the installation of sensors on the medium
boundaries. Then, thermal solicitation (deterministic signal) creates a disturbance.
Thermal sensors measure thermal response of the medium (temperature variation
and/or heat flux). Global thermo-physical properties access (thermal conductivity,
diffusivity and effusivity) is given by mathematical formalism (resolution of inverse
problem) and depends on injected signal (step, sinus,).
These methods can be very accurate and suitable for small samples with low
rugosity on sample surface in order to limit thermal contact resistance between
sensors and samples [17].
It is recommended to work with thermal sensors able to measure temperature
and heat flux with a minimal accuracy of respectively 100 mK and 1 W/m2. Wood
samples have to be adapted to sensors size. Measurement area of thermocouple is
about one square millimetre but flux meters (plane sensors) have a measurement
area from 25 cm2 to 400 cm2. Wood samples will usually have a volume greater
than 100 cm3.
Monitoring on site is often more difficult than carry out laboratory tests. So,
another interesting recommendation is to create a database of thermo-physical
properties of their own wood samples in order to easily specify limits of thermal
methods carried out on site.
3.2 Contactless methods
It is without question that IRT is the predominant NDTM without contact [18].
64
It is a well known method and applications are numerous [19-22], even if wood is
often less studied than other building materials because of its complex thermal
behaviour.
IRT protocol is the same as active thermal methods with contact but instead of
the determination of thermo-physical properties, the first objective is mainly to
detect thermal contrast (directly or after post-processing image [23]). In wooden
structures, defects usually detected are void, cracks or moisture [24].
Wood is a material with a low thermal conductivity, so it is recommended that
thermal solicitations last long enough to have a good thermal diffusion in the medium.
For a study on the first millimetres of a wooden structure, tests should last about
ten minutes. For a deeper investigation (5 to 6cm), tests can last more than two
hours. Thermal solicitations have to be low in order to not disturb thermo-hydro
behaviour of porous medium [25].
4. Data analysis
Inverse methods are used for data processing. Numerical method is chosen
according to the nature of injected signal and thermo-physical properties or thermal
effect that need to be determined. There are temporal methods based on simplified
description of thermal problem to solve (function series decomposition) or spatial
and temporal interpolation (Finite Differences [26] and Finite Elements Methods
[27]). Other methods are based on frequential description of thermal problem like
thermal quadrupoles methods [28].
Signals like steps are usually used with temporal description to determine
conductivity. To better characterize thermal behaviour of wooden structure, more
complex signals (sinus, sweep [29], pseudo-random binary sequence PRBS [30],...)
are the most used. Thermal diffusivity and thermal effusivity can be so determined.
There is a correlation between thermo-physical properties that can be directly
identified (conductivity, diffusivity and effusivity) and some physical quantities such
as water content [3] or porosity [31,32]. Correlation functions can be established
to link measurements on site to laboratory data.
5. Conclusions
In summary, thermal methods can be used to determine thermo-physical parameters
of wooden structures, other physical quantities (water content, porosity) and defect
inside the medium (cracks, voids). On site global thermal analysis (IRT) works fine
but makes anisotropic properties of wood difficult to evaluate i.e. thermo-physical
properties are known for a specific volume and not necessarily for a specific
direction. Thermal methods are still attractive even in terms of preliminary-structural
65
analysis. However, a precise diagnosis requires an association with otherNDT
methods (e.g. ultrasound [33]). Data fusion is in constant development andNDT
methods associations have to be followed with attention.
Acknowledgements
We would like to thank members of FPS COST Action FP1101 Assessment, Reinforcement and
Monitoring of timber Structure for their informed advice and the quality of the discussions.
References
[1]
Degiovanni A., Rémy B., (2005) Métrologie thermique : une histoire, un appareil, une
application, In: JITH 2005, Tanger, Maroc
[2]
Brown M.E., Gallagher P.K. (2008) Vol.5, Chap1. Recent advances, techniques and
applications of thermal analysis and calorimetry. Brown M.E (ed) Handbook of Thermal
Analysis and Calorimetry, Elsevier Science B.V., Grahamstown
[3]
Antczak E., Chauchois A., Defer D., Duthoit B. (2003) Characterisation of the thermal
effusivity of partial saturated soil by the inverse method in frequency domain, Applied
Thermal Engineering, Vol. 23, Issue 12
[4]
Zalewski L., Joulin A., Lassue S., Duthil Y., Rousse D. (2012) Experimental study of smallscale solar wall integrating phase change material, Solar Energy, Vol. 86, Issue 1
[5]
Berglind H., Dillenz A. (2003) Detecting glue deficiency in laminated wood – a thermography
method comparison, NDT & E International, Vol. 36, Issue 6
[6]
Carpentier O., Defer D., Antczak E., Chartier T. (2012) Frequency methods applied to the
characterization of the thermophysical properties of a granular material with a cylindrical
probe, International Journal of Thermophysics, Vol. 33, Issue 1
[7]
Carpentier O., Defer D., Antczak E., Duthoit B. (2005) The use of infrared thermography
and GPS topographic surveys to monitor spontaneous combustion of coal tips, Applied
Thermal Engineering, Vol. 25, Issue 17-18
[8]
Hosokawa M., Nogi K., Naito M., and Yokoyama T. (2008) Chapter 6 – Evaluation methods
for properties of nanostructured body. In Nanoparticle Technology Handbook, Elsevier
Science B.V., New York
[9]
Clemesha B.R., Martins Jorge M.P.P., Simonich D.M., Batista (1997) A new method for
mesuring the Doppler temperature of the atmospheric sodium layer, Advances in Space
Research, Vol. 19, Issue 4
[10] Keo S.A., Brachelet F., Breaban F., Defer D. (2014) Steel detection in reinforced concrete
wall by microwave infrared thermography, NDT & E International, Vol. 62
[11] Carpentier O., Brachelet F., Defer D., Aubagnac C., Cannard H., (2009) Characterization
of defect under waterproofing layer by IR thermography and thermal impedance, In:
NDTCE’09, Nantes
[12] Maldague X., Wyckhuyse A. (2001) A study of wood inspection by infrared thermography,
Part I and II, Res. Nondestr. Eval, Vol. 13
66
[13] Carpentier O., Defer D., Antczak E., Chauchois A., Duthoit B. (2008) In situ thermal
properties characterization using frequential methods, Energy and Buildings, Vol.
40,Issue 3
[14] ISO 8302:1991 Thermal insulation -- Determination of steady-state thermal resistance
and related properties -- Guarded hot plate apparatus
[15] ISO 8894-1:2010 Refractory materials -- Determination of thermal conductivity -- Part 1:
Hot-wire methods (cross-array and resistance thermometer)
[16] THERY, P. (1979) Fluxmètre calorifique, Brevet ANVAR
[17] Guo W., Lim J., Bi X., Sokhansanj S., Melin S. (2013) Determination of effective
thermal conductivity and specific heat capacity of wood pellets, Fuel, Vol. 103
[18] Maldague X.P.V. (2001) Theory and practise of infrared technology for nondestructive
testing, John Wiley & Sons
[19] Büyüköztürk O., Taşdemir M.A., Oğuz G., Akkaya Y. (2011) Non-destructive Testing
Materials and Structures, Springer, In: Proceedings of NDTMS, Istanbul, Turkey
[20] Arndt R.W. (2010) Square pulse thermography in frequency domain as adaptation
of pulsed phase thermography for qualitative and quantitative applications in cultural
heritage and civil engineering, Infrared Physics and Technology, Vol. 53
[21] Ludwig N., Redaelli V., Rosina E., Augelli F. (2004) Moisture detection in wood and
plaster by IR thermography, Infrared Physics and Technology, Vol. 46
[22] Kortados, E.Z., Exarchos D.A., Stavrakos C., Moropoulou A., Matika T.E. (2013)
Infrared thermographic inspection of murals and characterization of degradation in historic
monuments, Construction and Building Materials, Vol. 48
[23] Du T., Brachelet F., Defer D., Antczak E. (2013) Quantitative evaluation of thermal
diffusivity and thickness of mortar cover using induction thermography, In Nondestructive
Testing of Materials and Structures RILEM Bookseries, Volume 6
[24] Taoukil D., El bouardi A., Sick F., Mimet A., Ezbakhe H., Ajzoul T. (2013) Moisture
content influence on the thermal conductivity and diffusivity of wood– concrete composite,
Construction and Building Materials, Vol. 48
[25] Krishnaiah S., Singh D.N. (2003) Determination of influence of various parameters on
thermal properties of soil, In: International Communication of Heat Mass and Transfer 30
(6)
[26] Ozisik N. (1994) Finite Difference Methods in Heat Transfer, CRC Press
[27] Lewis R.W., Morgan K., Thomas H.R., Seetharamu K. (1996) The Finite Element Method
in Heat Transfer Analysis, Jhon Wiley & Sons
[28] Maillet D., André S., Batsale J.C., Degiovanni A., Moyne C. (2000) Thermal Quadrupoles
: Solving the Heat Equation through Integral Transforms, JohnWiley & Sons
[29] Von Seggern D. (1993) CRC Standard Curves and Surfaces, Boca Raton, FL: CRC Press
[30] Mahmoudi Y. (2014) Effect of thermal radiation on temperature differential in a porous
medium under local thermal non-equilibrium condition, International Journal of Heat and
Mass Transfer, Vol. 76
[31] Lopez G., Basterra L.A., Acuna L. (2013) Estimation of wood density using infrared
thermography, Construction and building materials, Vol. 42
[32] Carpentier O., Antczak E., Brachelet F., Defer D., Descamps T., Van Parys L (2014)
Characteristization of density variations of historic timber structure by thermal methods,
In: The 12th International Conference on Quantitative Infrared Thermography, Bordeaux,
France
67
[33] Kandemir-Yucel A., Tavukcuoglu A., E.N. Caner-Saltik (2007) In situ assessment of
structural timber elements of a historic building by infrared thermography and ultrasonic
velocity, Infrared Physics & Technology, Vol, 49, Issue 3
68
Prediction of mechanical properties by means of
semi-destructive methods
Michal Kloiber1, Miloš Drdácký1
1
Czech Academy of Sciences, Institute of Theoretical and Applied Mechanics
Abstract
Methods and devices for in-situ establishment of mechanical properties of wood
have been recently developed in cooperation of Czech and US researchers. The
development of new methods was motivated by requirements from engineers who
need most accurate data of mechanical properties of specific elements when
planning renovations of historic buildings. The established data replace mean
standard values of strengths in their static calculations, which results in the
demanded possible retaining of a larger amount of original material as the
behaviour of the specific elements is safely assessed. The newly developed
methods, which are described in this chapter, are tensile s tr e n gth o f s ma ll
samples, compression strength of cores, compression strength in a drilled hole, and
mechanical resistance to pin pushing. Although the devices for the testing of
mechanical properties by the mentioned methods provide more accurate results
than the methods used so far, they are not mass produced, which should change
soon.
Key words: semi-destructive testing, wood, in-situ assessment, strength, pin,
resistance.
1 Tensile strength of small samples
1.1 Sampling and testing methodology
The assessment of bending strength is important for in-situ assessment of timber
elements as it is the prevailing manner of loading in e.g. ceiling constructions.
Bending strength of integrated timber cannot be established without damage done
to the construction. However, it is close to tensile strength and according to some
authors it can be considered almost the same (Kasal and Anthony, 2004).
Therefore, a new method to establish strength of integrated timber using small
samples taken from its surface was devised.
69
Samples for the establishment of tensile strength are extracted in a simple way
using an adjustable circular saw. Sampling is carried out by two cuts inclined in an
angle of 45° in relation to the element surface (Fig. 1) parallel to the grain. The cut
depth is adjusted so that a triangular bar with rectangular sides of about 5–8 mm is
gained. The saw runs in guides, which are fixed to the surface of a tested element
by screws. The damage done to the surface is remedied by an insertion of
triangular bar with the same dimensions; it can also be totally mended by
restoration. During the production of a sample, the area of the bar section is
reduced to about 8–12 mm2 in the central part, which corresponds to production of
tensile samples in compliance with (ASTM Annual Book). The samples should not
contain any natural defects (knots, cracks or other damage). Rectangular wooden
blocks are glued to both ends of the samples (Fig. 2) in order to fix the small
samples in coaxial articulated grips of the loading device during the tensile
strength testing (Kasal et al. 2003).
Fig. 1 Circular saw with guides, modified
Fig. 2 Tensile test of a triangular bar
The tensile sample is inserted in simple grips designed for this purpose and
loaded in a common testing device (Fig. 2). The test is not standard but its concept
is very close to the standard test in compliance with (ASTM Annual Book) as it
uses the same simple layout eliminating parasite movement and a cross-section
with a small number of annual rings. The test measures the tensile strength and the
modulus of deformability to calculate the modulus of elasticity. The maximum
tensile loading for each sample is the ultimate load and the tensile strength is determined by the formula:
70
fc =
where
Fmax
[MPa]
0,5 ⋅ bh
ft – tensile strength [MPa],
Fmax – ultimate load [N],
b – triangular bar hypotenuse [mm],
h – triangular bar height [mm].
The results gained by this test need not be correlated and can be declared comparable with the standard test. For the purpose of the construction safety assessment and dimensioning, mechanical properties established by the described test
must be converted to technical properties of timber which take account of the locally measured strength of clean timber reduced by defects that commonly appear
in large elements (knots, cracks, and others). The disadvantage of this method is
the damage done to the surface of the assessed element, which is undesirable in
the case of historic construction timber assessment (Drdácký et al. 2005).
1.2 Limitations
The method for the establishment of tensile strength and modulus of elasticity
uses small triangular samples that are taken from a relatively shallow surface part
of a timber element, where the historic timber is often damaged by biotic factors.
The method is very sensitive to fibre deflections in the sample and requires a
careful choice of a sampling place and careful sampling. The cross-sections of
the sample is small, which increases the effect of a higher earlywood proportion.
This effect is negligible in larger cross-sections. As a result, the values of strength
and modulus of elasticity will be highly variable. Therefore, it is necessary to
think well about the places where samples are to be taken and also take account
of potential damage.
1.3 Application examples
The tests of tensile samples have been successfully used e.g. when investigating
the quality of ceiling joists in the St. Mary’s Tower of the Karlstejn castle or
strength values of the storage hall in the trade fair facilities in Brno (South Moravia – ČR).
The method can be also used to determine the level of damage to the surface
layer of the timber. As an example we can name the study into the effect of fire
protection treatment that has been repeatedly applied to timber constructions of
historic buildings. The application of agents with fire retardants on the basis of
71
sulphate and ammonium phosphate has caused damage to timber surface referred
to as “fibrillated surface” which gives the look of timber elements a “fuzzy”
character.
The in-situ surveys of damaged timber construction elements proved that the
fibrillated timber layer manifests a considerable loss of cohesion and deterioration
of mechanical properties. There was the question to what depth chemical corrosion
reaches and how much mechanical properties are affected. Mechanical properties
in particular levels of damaged timber were established by special tests of small
tensile samples. The samples were taken from the truss construction of a
former brewery malthouse in Děčín (North Bohemia – ČR) (Fig. 3). The surface of
the timber investigated manifested an advanced level of fibrillation (Fig. 4).
Fig. 3 Truss construction in the former
brewery in Děčín
Fig. 4 A detail of an analyzed beam surface
The tensile strength parallel to the grain was tested using small triangular samples (5×5×7.5 mm) 200 mm long. The specially made small samples allowed for a
more accurate establishment of a property investigated at various depths under the
surface of the damaged timber. The samples were made from the superficial layer
of the timber (0 – 5 mm – damaged layer) and the inner part of the timber (25 mm
deep – undamaged layer, reference samples). The considerable deterioration of
mechanical properties of timber in the damaged surface layer was manifested by a
50 % decrease in strength compared to the values ascertained in undamaged timber (Tab. 1). The surface was damaged to a depth of 5 mm, which was confirmed
using small tensile samples (Kloiber et al. 2010).
72
Table 1 Mean values of strength and modulus of elasticity from the tests of small tensile samples in
the surface (damaged) and inner (undamaged) layers of timber
Tensile tests parallel to the grain
Beam side
Strength S c (MPa)
Modulus of elasticity M OE (MPa)
surface
inner
surface
inner
Top
18.85
47.49
14801.66
12541.10
Bottom
17.52
55.16
16116.45
13957.38
Lateral
20.15
40.86
12444.37
13261.16
1.4 Summary
The testing of small tensile samples is a direct and partially destructive method
which can be used for the measurement of modulus of elasticity and tensile
strength parallel to the grain. The method is not compromised by uncertain correlations between the measured and the estimated parameters. The tensile properties
of small clean wood samples are measured directly but their information capacity
is reduced by the high variability of results dependent on the effect of earlywood/latewood proportion. The method only gives information about timber in
the close vicinity of the element surface, similarly to method Hardness test
presented in Riggio and Piazza, 2011.
1.5 Recommendations
It is recommended to choose a sufficiently large part of timber surface without
defects – typically, a band of 20 mm x 300 mm is needed for the method
application. Sampling demands that the guides are fixed either directly on
the surface or on an auxiliary construction so that the sampled band area
merges with the plane of the guides. The band axis needs to be in the same
direction as the axis of the guides and the direction of fibres. It is recommended
to determine the fiber direction using light scratching the surface between the
annual rings. If, for the reasons of the element geometry, the sampling
orientation cannot be maintained, it is necessary to correct the strength
values measured based on table dependences (e.g Bodig, 1993). The samples
in the test need to be loaded coaxially so that their bending and thus effect
on measured quantities are prevented. Moisture content should be measured
or controlled during sampling and laboratory testing as the values of
mechanical properties decrease with increasing moisture content. Moisture
content of 12% is recommended for the tests. Accurate measuring of sample
dimensions is a prerequisite for the test interpretation and establishment of
mech. properties.
73
2 Compression strength of cores
2.1 Sampling and testing methodology
Testing of radial cores is a semi-destructive method. Samples are of a cylindrical
shape (Fig. 5) and they are used to establish strength and modulus of elasticity in
compression parallel to the grain using a special loading device (Kasal et al.
2003). The holes that remain after sampling are smaller than most knots that
appear in timber elements and they do not reduce the element strength considerably
meeting thus the requirements of conservation institutes regarding low invasiveness
(Kasal and Antony, 2004). The sampling holes can be plugged to prevent
moisture penetration, insect attacks, probability of decay or if aesthetic qualities
are to be preserved (Kasal, 2003). Radial cores are 4.8 mm in diameter and the
holes in the element are 10 mm in diameter. The length of the cores should be at
least 20 mm to ensure reliability of results and elimination of result variability in
consequence of early- and latewood alterations.
Fig. 5 Taking a cylindrical sample
Fig. 6 Equipment for sampling of radial cores
Radial cores are taken using an electric drill with a special bit (Fig. 6), which
was developed in ITAM, ASCR. Occasionally, soap or wax is applied to facilitate
drilling. The drilling speed must be constant and the drilling is usually performed
in steps to prevent damage to samples. The bit tip must be sharp and clean. Blunt
and dirty bits cause damage to samples. The samples are transported to the laboratory in containers that prevent their damage and moisture content changes. The
containers are marked with a number, place and date of sampling and other important information (Drdácký et al. 2005).
74
Fig. 7 A detail of the loading device
Fig. 8 An example of the stress-strain diagram for
the compression test of a radial core
The samples should be extracted from healthy and undamaged material in the
radial direction because the tree-ring orientation is an important basis for correct
testing. The shear forces that can be very high during drilling make the sampling
conditions unfavourable. Therefore, the drill construction eliminates them. Due to
the threat of drill sideways motion, the drill is fitted in a special device ensuring
fixation and constant progress of the bit into the material. Radial cores can be used
for examining other properties of wood such as density, moisture content, modulus of elasticity and compression strength parallel to the grain. It can also be used
for the determination of the tree species, dendrochronological dating, microscopic
analysis of decay, and visual evaluation of the element condition and the penetration of protective agents (Schwab et al. 1982).
The actual testing of radial cores uses testing grips with grooves that facilitate
loading in the direction perpendicular to the core axis, i.e. parallel to the grain
(Fig. 7). Two linear-variable displacement transducers (LVDT) are used for monitoring the distance between the grips and thus the radial core deformation. A correct insertion of the radial core in the testing apparatus is critical for the correct
determination of the compression strength and the modulus of elasticity. Wood
has the highest strength in the grain direction, and uncentered radial cores in the
grips lead to higher variability of results (Lear, 2005). The compression force and
the core deformation are recorded in a stress-strain diagram, see Fig. 8. The compression strength is calculated as follows:
fc =
Fmax
l ⋅ dc
[MPa]
where
fc – compression strength [MPa],
Fmax – load [N], load Fmax is taken from the diagram, see Fig. 8,
l – radial core length [mm],
dc – radial core diameter [mm].
Kloiber and Kotlínová (2006) found the correlation between the strength of radial
cores and the strength of standard specimens in the longitudinal direction.
75
Schwab et al. (1982) established the coefficients of determination in the interval
R2 = 0.77–0.96 for the same dependence, in dependence on the wood species.
Kasal (2003) found a strong dependence between moduli of elasticity in compression parallel to the grain for radial cores and (ASTM 143-94) samples (American
Technical Standard), coefficient of determination R2 = 0.89. The variability of
measuring is comparable for both methods (Kasal, 2003). One of the problems
when establishing the regression between the properties of radial cores and standard specimens is the destructive character of both methods due to which both of
the methods cannot be used for absolutely the same samples.
2.2 Limitations
Due to the dimensions of the radial cores, this method is of local
character. Therefore, it may not provide relevant information about the
condition of the integrated timber because of wood variability. This inefficiency
can be eliminated and the reliability increased by a higher number of
samples taken from one element. However, this step would increase the
damage done to the element, the time consumed, and the expenses of field
measuring and the strength of the element would decrease (Kasal, 2003).
The testing sample is extracted using a special bit which is fixed in an
electric or manual drill (Fig. 6). The bit outer diameter is 9.5 mm. Speed is
controlled during drilling so that the samples are not damaged. For the same
reason, the bit tip needs to be kept sharp and clean. Blunt or dirty bits can
cause that the samples look damaged or decayed, or they can be pushed
outside the bit, which creates distortion of results. The samples should be
extracted from healthy and undamaged material in the radial direction because
the tree-ring orientation is vital for the correct test. The shear forces that
can be very high during drilling make the sampling conditions
unfavourable. Therefore, the bit inner diameter decreases towards the tip. To
eliminate a possible sideways motion the bit is fitted in a special device which
ensures fixation and constant speed towards the material.
Radial cores can be used to establish physical, mechanical and strength
properties of timber. When investigating moduli of eleasticity, it is necessary to
release the load partially from the radial core during testing (Fig. 9) and then
measure the elastic response of the core to the change of external loading only;
otherwise, considerable inaccuracy occurs (Micka et al. 2006) as the total sample
deformation is affected by the plastic deformation in the places of contact with
the grips. Radial cores can also be used to determine density. This is especially
important for valuable timber elements, in which every piece of material matters.
The variability of the data gained is comparable with standard tests. However,
radial cores need to be taken from undamaged places.
76
Fig. 9 Stress-strain diagram force –
compression with partial release
Fig. 10 Correlation of pressure deformation
characteristics of the radial cores with the
number of annual rings.
(Micka et al. 2006)
2.3 Application
Radial cores have been successfully used for non-destructive surveys of timber
constructions e.g. with the aim to find out mechanical properties of a timber
construction of storage halls in the trade fair facilities in Brno (South Moravia –
ČR) or the quality of the ceiling timber in the St. Mary’s Tower of the Karlstejn
castle (Central Bohemia – ČR). In the case of the St. Mary’s Tower, we had a
sample of an authentic beam from a damaged part of the building (emperor palace
stables). We used this piece for calibration tests in compression parallel to the
grain in compliance with (ASTM 143-94). We tested short columns cut from the
beam with exactly oriented fibres. Further, radial cores were sampled so that each
standard sample by ASTM was matched with two cores taken from each end of the
specimen. In total, 38 radial cores were available. Typical deformation properties
(modulus of elasticity in compression parallel to the grain) were investigated using
both the radial core samples and the standard samples. The values correlated very
well. The mean values for the modulus of elasticity and compression strength
ranged around 7600 N/mm2 and 42 N/mm2, respectively. The characteristic values
with 5% quantile followed, reduced based on the appearance of defects in the
element. For the reduction, visual assessment was used (Drdácký et al. 2003).
Further, non-standard tests were conducted using 12 radial cores from the joist of
the treasure room ceiling. The found values of deformation characteristics correlate
with the tests of cores from the reference beam from the emperor palace stables.
The comparison shows that the mechanical properties of the treasure room ceiling
timber elements were of the same quality as the reference beam whose strength
and modulus were established by ASTM standard tests (Fig. 10). The values of
mechanical properties showed a very high quality of the historic timber.
77
These values can be used to determine design characteristics for the purpose of
construction safety assessment in a common way (Drdácký et al. 2003).
2.4 Summary
Radial cores are to be used for a direct establishment of physical (specific density),
morphological (tree-ring width) and mechanical properties (compression
strength parallel to the grain and modulus of elasticity) with a relatively high accuracy
in defectless timber. The cores can also be used for microscopy, dendrochronology,
visual assessment, and measuring of protective agent penetration.
2.5 Recommendations
The cores need to be taken with a special hollow bit in the radial direction, i.e. the
direction of a normal to annual rings so that the test of mechanical properties can
be performed parallel to the grain. An area with a diameter of about 70 mm is
needed for the sampling as the drill fittings need to be fixed. The fittings ensure
the bit movement in the radial direction without deviations and allow for a gradual
drilling into the demanded depth with constant speed. After the demanded core
depth is achieved (usually 40 mm), the bit is taken out and a thin-walled tube is
inserted. The tube is used to take the core out of the timber. The core is then
pushed out of the tube. Broken or damaged cores are excluded from the test.
Moreover, the core needs to be placed into the testing grips with high accuracy.
Moisture content of 12% is recommended for the tests. Accurate measuring of
sample dimensions is a prerequisite for the test interpretation and establishment of
mechanical properties.
3 Compression strength in a pre-drilled hole
3.1 Testing methodology
The range of the existing methods and devices lacked a solution enabling the
measurement of mechanical properties of wood using gently destructive investigation
of its behaviour when being loaded by a miniature loading jack inserted in a predrilled hole. Determine exact mechanical properties and strength grades important
for structural design (Machado, 2013). During the application the dependence of
deformation on the voltage is measured while symmetrically arranged grips
("stones") are being pushed apart in a pre-drilled radial hole with a diameter of
12 mm (Drdácký and Kloiber, 2013). The semi-destructive procedure of making
a hole into the tested material allows the investigator to assess other aspects of the
material condition (e.g. based on the core, sawdust, videoscopy, etc.).
78
Fig. 12 A detail of a drawbar with a push-apart wedge
and rounded grips
Fig. 11 A view of the device
The device (Fig. 11) is designed to measure mechanical properties of wood
using gently destructive investigation of its behaviour when loaded by the
miniature loading jack inserted in the prebu-drilled hole. The device can be used
both in the laboratory and in the field to assess the condition and the quality of
timber. The advantage of the device is the possible gradual recording of the
force and shift of grips (loading jack) at different depths corresponding to the
required dimensions of commonly investigated constructions. The device is laid
on the tested unit (usually a constructional element of a rectangular profile) by
means of a cylindrical shell, which allows for measuring in four positions of
the pre-drilled hole. The shell arresting is provided by two grooved screws, for
positions (core depths) 5– 25, 35–55, 65–85, and 95–115 mm. When the
measuring part of the device is inserted in the drilled hole and the device is laid
on the tested element, the rounded grips are pushed apart by the drawbar with a
push-apart wedge (Fig. 12) into the walls of the hole. The maximum depth of
possible loading on both sides is 1.5 mm. The rounded grips are 5 mm wide
and 20 mm long. The grips also include flexible arms whose movement during
pushing is provided by a push-apart bronze wedge fitted to the lower end of the
drawbar by means of a pin and screw. The apex angle of the wedge is 15°.
This angle is not self-locking and to release the grips it is sufficient to release
the push-apart force (Drdácký and Kloiber, 2013).
The force of the drawbar when pulling out is continually scanned and recorded.
It is calibrated to the real force of the loading jack and simultaneously related to
the measured distance of movement of the grips (Fig. 13). The signals are wirelessly transmitted to a portable computer, where they are processed.
79
Fig. 13 Example of the device output: record of the force of grip pushing apart
related to the measured distance of movement of the grips
Kloiber et al. 2013 introduced the construction and usage of this new device for
in-situ assessment of integrated timber. The application of the new device was
verified. It was found out that the device is sufficiently sensitive to the natural differences between individual elements of healthy timber. Strong correlations were
mainly found for the measured CSC (L) strength in compression parallel to the grain
and SC (L) strength of standard samples assessed in compliance with ČSN 49 0111
(correlation coefficient 0.92). The relations were described by practically usable
linear regression models. The measured compression strength parallel to the grain
correlates with the other investigated timber parameters, e.g. density (correlation
coefficient 0.87). Another parameter for the assessment of mechanical properties
using the new device was MOD (L) modulus of deformability, which correlates
well with MOE (L) modulus of elasticity parallel to the grain (correlation coefficient 0.87). The construction of the device is lightweight and due to its independence from the electrical grid, it is easy to use in the field. In contrast to other
methods, the new device enables a highly accurate establishment of mechanical
properties in the entire depth profile of the assessed element.
3.2 Limitations
The prerequisites of an appropriate use of the method are drilling a hole through
the wood fibres purely in the radial direction, where there is a regular alternation
of earlywood and latewood within annual rings, and the orientation of the measuring
probe parallel to the grain; in structural elements it is generally parallel to
the axis of the element. Measuring is affected by a higher proportion of
earlywood or latewood within a tree ring in the tangential direction, which leads to
distorted results. The hole needed for the test is created by a bit which is fixed in
an accumulator drill. To prevent sideways motion of the bit, the drill is fitted to a
special stand which fixes it to the element. The outer diameter of the bit is 12
80
mm. Speed is controlled during drilling so that the hole is not damaged. For the
same reason, the bit must be maintained sharp and clean. Blunt or dirty bits can
cause fibres being torn out of the hole walls, which distorts the results. The hole
should be made to undamaged places of the element without visible defects and
damage.
An essential feature of the in-situ testing is the fact that the measuring of a
loaded element is conducted with unknown internal forces present. It was proved
by measuring the deformation around the drilled hole using image digital correlations
that the state of tension recedes after drilling into a distance of about 2 mm from
the hole edge and the measuring is thus not affected by the inner tension of the
constructional element unless the element was damaged by the elasticity limit
being exceeded. The above mentioned assertion has been verified by tests of a
bended timber console (Maddox at al. 2014).
3.3 Application
The method for the establishment of strength in a pre-drilled hole has been
successfully used for the investigation of mechanical properties of the timber truss
of the St. Mary’s Church in Vranov nad Dyjí from the 17th century (South Moravia
- ČR) or when investigating the quality of a larch ceiling from the 14th century in
Spišské podhradí (UNESCO site - SK).
In the case of the Vranov nad Dyjí truss, four samples of tie beam ends were
available. They were taken away because timber scarf joints were to be used instead
of the damaged ends. The aim was to verify the application of the device using the
wood of Silver Fir (Abies alba Mill.) in the common variability of properties of timber
integrated in a historic building. The measuring by the new device was conducted
in 135 positions. Decayed beam ends were also measured but due to the low values
gained, the results were not taken into account.
Mechanical properties were determined from the record of measured data in the
form of a stress-strain diagram showing the force of drawbar pulling out, which
was calibrated to the real force of grip pushing apart and simultaneously related to
the measured distance of grip movement. Measured strength CSC was determined
from the ratio of the ultimate load and the area of the grips. Modulus of elasticity
cannot be calculated from the stress-strain diagram directly; however, the modulus
of deformability was determined based on the force slope and deformation. The
new device was verified by experiments based on the comparison of values
measured during grips being pushed apart in a drilled hole and the results of
standard sample testing by destructive tests in compliance with ČSN 49 0111 using
a universal testing device. Two standard samples with dimensions 20x20x30 mm,
complying to ČSN 49 0111 were cut from places adjacent to the positions
measured by the new device. Compression strength parallel to the grain established
based on the standard samples was correlated with the results of the new device.
81
Strong correlations were mainly found for CSC (L) measured strength in compression
parallel to the grain and SC (L) strength of standard samples (correlation coefficient
0.75). The relations were described by practically usable linear regression models.
Similar dependences are found when the pin penetration device is used (Kloiber et
al.). The initial tests conducted using timber from the historic truss construction
showed that the new method is sufficiently sensitive to natural changes in properties
(distribution along the element width). It should be noted that the natural material
variability was increased by the presence of defects (knots and cracks).
3.4 Summary
The advantages of this method include the high accuracy of the establishment of
mechanical properties (measured strength and modulus of deformability in
compression parallel to the grain) of timber tested and assessed in the field. In contrast
to other methods, the new device is able to establish mechanical properties in the
entire depth profile of the assessed element. The measuring is accurate if the drilling
is oriented perpendicular to the grain in which direction early- and latewood
alternate regularly within annual rings and if the grips are pushed apart parallel to
the grain or parallel to the element axis in the case of constructional elements. The
effect of larger proportions of earlywood or latewood in the tangential direction
leads to a distortion of results. The results of measuring depend on the quality of
the drilled hole production, i.e. it is necessary to check the bit constantly and
replace blunt bits immediately.
3.5 Recommendations
To guarantee the planeness of the drilled hole and to eliminate sideways motion of
the bit, the drill needs to be fixed to the assessed element by means of a special
stand during drilling. The stand can be fixed to the element directly or via an
auxiliary construction. The fixing demands an area of 150 x 150 mm. To ensure a
good quality of drilling, it is recommended to control the drilling speed, especially
the bit progress into the hole. The hole should be drilled in undamaged places of
the element without natural defects and visible damage. A higher number of holes
improperly placed can affect the mechanical resistance of the element assessed.
Like other in-situ methods used for the diagnostics of integrated timber, this
method for the measurement of strength and modulus of deformability manifests a
considerable dependence on the moisture content in the investigated material.
Therefore, the measuring of moisture content in the tested place is an essential part
of the test.
82
4 Mechanical resistance to pin pushing
4.1 Testing methodology
A device for in-situ establishment of mechanical resistance to gradual pin pushing
was developed in cooperation of the Institute of Theoretical and Applied Mechanics
of the Academy of Science of the Czech Republic and the Department of
Wood Science, Mendel University in Brno. The device measures the demanded
values to the depth corresponding to typical dimensions of timber constructional
elements and is applicable for an indirect establishment of density and mechanical
properties of wood. Similar penetration test is based on repeated pin hammering
into the wood by means of a hammer with a constant energy (Ronca and Gubana,
1998).
Fig. 15 A detail of pin penetration through the
device base during pin pushing
Fig. 14 A view of the device for in-situ
The device body can be fixed to the tested element in various ways, most often
with a fabric strap (Fig. 14) or a roller chain. The device body can also be fixed to
the tested element by mounting screws. After the device has been fixed to the object, a pin is gradually pushed into the timber perpendicularly to the device base
(Fig. 15) by a toothed rack and pinion gear driven by two opposite manual cranks
for both hands. The force of pin pushing is continuously recorded in relation to the
distance measured (Kloiber et al. 2011). The measuring application processes the
data, shows them in the real time and saves them. The progress of force measured
83
in the real time is shown either in dependence on time x-t or in x-y mode together
with pin displacement. During the measurement, basic characteristics are calculated by a PC. These are work [N.mm] as the area under the force curve related to
the displacement, penetration length [mm], time of pin displacement [s] and the
maximum and minimum force [N]. The mean force [N] necessary for pin pushing
is calculated by dividing the area under the curve by the penetration depth. This
parameter is of key importance for the assessment of the timber mechanical resistance.
The continuous record of the force related to pin displacement is able to indicate a change in properties within the entire depth of pin penetration caused by either a natural distribution of properties or biodegradation. The curve of forces in
the case of undamaged spruce (Fig. 16) corresponds to earlywood and latewood
alternations within annual rings (latewood with a higher mechanical resistance and
early wood with a lower mechanical resistance). The curve also describes the different tree-ring widths (increments) in the element cross-section. The general progress reflects the equal distribution of mechanical resistance in the cross-section,
i.e. a balanced quality of sound spruce timber.
When measuring the resistance of pine wood, there is again a visible difference
between latewood and earlywood as well as tree-ring widths (Fig. 17). The record
with increasing forces reveals that there is heartwood typical of pine wood with a
higher density and mechanical resistance. The absolute values of forces correspond to the mechanical resistance to pin pushing in sound pine wood. Fig. 18
shows a record of measuring a spruce element containing biodegradation. The relative decrease in the zone with decay compared to sound wood and the absolute
force values indicate a decrease in mechanical resistance.
Fig. 16 Record of the force related to pin displacement - spruce wood
Fig. 17 Record of the force related to pin displacement - pine wood
84
To sum up, the device is applicable for a wide range of properties of sound as
well as damaged wood. The test results manifest very good correlations of the
mean force needed for pin pushing with wood density and strength determined using standard samples in compression perpendicular and parallel to the grain (Kloiber et al. 2009; Tipner et al. 2011). The measured parameter can be changed by a
simple replacement of the indentation pin with a hook for withdrawal of screws or
other fixings, as for example presented in Yamaguchi, 2011.
Fig. 18 Record of the force progress and displacement of pin pushing into spruce wood with decay
and feeding of wood-damaging insects (Tannert et al. 2014)
4.2 Limitations
The device was designed and tested for the assessment of wood integrated in
buildings, both sound or with various levels of degradation. The device records a
relatively wide range of timber mechanical resistance to pin pushing caused by
natural properties of different tree species as well as different levels of
degradation. The resistance is affected by the species, wood quality and density
but also moisture content (Tannert et al. 2014). These parameters need to be taken
into account when interpreting the results. Potential wood defects, such as cracks,
knots, foreign objects, etc. considerably distort the results. It is advisable to avoid
measuring in places with wood defects, or interpret results of such measuring very
cautiously. It is necessary to push the pin into wood in the direction perpendicular
to fibres - only in the radial direction - where earlywood and latewood alternate
regularly.
4.3 Application
Visual inspection of the flat roof truss construction of the Čechy pod Kosířem
Castle Orangery (North Moravia – ČR) conducted in 2011 proved that the truss
and ceiling timber is locally damaged by wood rot and insects. Damaged places
were found at the ends of tie (ceiling) beams, where rainwater had leaked and
had provided favourable conditions for brown rot. The resulting rot changed
physical properties of the timber (colour, decrease in density, increase in
absorption, etc.). Wood mass was considerably disintegrated at some places.
85
The wood-decaying insects identified as the cause of the general damage of
constructional elements were Cerambycidae or Anobiidae. The attacked tie beam
had to be replaced or fitted with scarf joints. The renovation needed to be
approached with utmost caution and maintain the largest possible proportion of
historic material. The construction of the Orangery truss was divided into 25
cross sections, each containing a tie (ceiling) beam and a rafter. The
construction renovation of the truss was designed based on visual inspection and
mainly results of mechanical resistance measuring by the diagnostic device with a
pin 2.5 mm in diameter (Fig. 19). A third of the total volume of the elements was
deemed for replacement. The accurate establishment of damage helped save a
large part of the material, which despite the surface damage (Fig. 20) met the
necessary mechanical properties to the same or higher extent than new timber.
Fig. 19 Investigation of a tie beam
condition using pin pushing
Fig. 20 A detail of damage to a tie beam,
which was left in the original position
Fig. 21 shows the record of measuring of a tie beam with surface rot and damage by wood-damaging insect feeding. The measuring was conducted using the
device with pin pushing method. The device was fixed to the element by a fabric
strap and the force needed for pin pushing was developed continuously by two
cranks (Fig. 21). The distance and force were recorded during pushing. The relative decrease in the zone with rot and feeding compared to sound wood and the
absolute force values indicate a decrease in mechanical resistance caused by degradation to a depth of 15 mm only.
86
Fig. 21 The record of measuring tie beam 8 northern end,
where only 15 mm layer of surface was damaged
4.4 Summary
The method of measuring the wood mechanical resistance to pin pushing can be
used to estimate parameters of wood density and strength, up to a depth of 110
mm. By contrast to the commonly used Pilodyn, this method provides data about a
larger cross-section part of the tested element and allows the researcher to identify
internal defects hidden deep below the timber surface. The slow progress of
pushing enables the researcher to quantify damage in any depth. The size of the
resistance recorded is affected by the wood species, wood quality and density,
as well as moisture content. These parameters need to be considered for a correct
interpretation of results. Potential wood defects, such as knots, cracks, foreign
objects, etc. distort the results considerably.
4.5 Recommendations
The test demands a free area of 150x150 mm so that the device can be fixed. It is
recommended to avoid measuring in places with defects as the interpretation of results
is then difficult. Pin pushing is only accurate if the device is fixed perpendicular
to the grain and if the pin penetrates the timber in the radial direction. The
acceptable deviation is about up to 10° from purely radial direction. When the pin
is pushed in the tangential direction, the results may be distorted as the pin then
often penetrates the weaker earlywood only and does not enter denser latewood
increments in heterogeneous wood types. The measuring of moisture content in
the tested place is again an important part of the test.
87
5 Conclusion
Testing of small tensile samples is a direct, partially destructive method that can be
used for strength and modulus of elasticity measuring in tension parallel to the
grain. The method is not hindered by uncertain correlations between measured and
estimated parameters. However, the method is sensitive to deflections of wood
grain from the sample longitudinal axis. The sample cross-section is small, which
strengthens the effect of early- and latewood alternation in annual rings. This
effect cannot be avoided, which means that the results (strength and modulus of
elasticity) are highly variable and the selection of suitable places for sampling is
vital. In-situ sampling and laboratory testing are time consuming and costly. The
method only informs about superficial properties of an element without defects;
moreover, it represents damage to the surface of the assessed element. The method
of radial cores represents a smaller invasion in the material based on taking
cylindrical samples that are used to test strength and modulus of elasticity in
compression parallel to the grain using a special device. The holes that remain
after the core is removed are smaller than most knots and are located not to
deteriorate the element strength. The unfavourable effect of earlywood and latewood alternations within annual rings is sufficiently diminished if the core length
is at least 20 mm. The cores can be used also for microscopy, dendrochronology,
visual assessment and measuring of protective agent penetration. The core
diameter is 5 mm only, i.e. the method is of local character. Due to the high
variability of wood, the core may not provide a good idea about the mechanical
properties of the element investigated. This deficiency can be countervailed by a
larger number of cores taken from an element. In-situ sampling and laboratory
testing are time consuming and costly, just as in the case of small tensile samples.
To reduce the costs and the time needed, we devised a method for in-situ direct
establishment of strength and modulus of deformability in compression parallel to
the grain in a pre-drilled hole with 12 mm in diameter. With this method no sample
needs to be taken for laboratory tests. The method measures mechanical properties
very accurately in any depth of the timber element. The unfavourable effect of
earlywood and latewood alternations within annual rings is counteracted by grips
of 20 mm in length. Again, wood variability needs to be taken into account as it
affects the mechanical properties determined locally.
The method of pin pushing is the least invasive out of the presented methods.
The hole created has 3 mm in diameter only. By contrast to dynamic indentor
(Pilodyn), the length of the pin and thus the depth of the penetration into the
investigated element allows for the identification of internal defects. Thanks to the
slow progress of pushing, the values of mechanical resistance are gainedgradually.
Mechanical resistance correlates well with mechanical properties.
88
The disadvantage of this method is the necessity to fix the device to the element
investigated. Just like the other invasive methods for timber diagnostics, this method
depends heavily on internal factors, moisture content, and tree species.
Table 2 A synoptic table describing the success rate of prediction of mechanical properties, the time
consumed and the extent of invasive damage caused by the presented methods in comparison with
commonly used methods
Methods described in the chapter
Tensile
strength of
m icrosam ples
Com pression
strength of
cores
Com pression
strength in a
drilled hole
Mechanical
resistance to
pin pushing
Mechanical properties in
the elem ent surface
60-90 %
60-90 %
60-90 %
40-70 %
Mechanical properties to
a depth of m in. 100 m m
30-60 %
60-90 %
60-90 %
50-80 %
Test tim e consum ption
high
high
low
low
Extent of invasion
high
low
low
negligible
Methods comm only used
Mechanical
resistance to
pin penetration
(Pilodyn)
Mechanical
resistance to
screw
w ithdraw al
Hardness test
(Piazza)
Microdrilling
energy
Mechanical properties in
the elem ent surface
50-80 %
50-80 %
50-80 %
30-60 %
Mechanical properties to
a depth of m in. 100 m m
30-60 %
30-60 %
30-60 %
40-70 %
Test time consum ption
negligible
low
low
low
Extent of invasion
negligible
low
negligible
negligible
Acknowledgments
This paper was created with a financial support from grant project DF11P01OVV001 “Diagnostics
of damage and life span of Cultural Heritage buildings”, NAKI program, provided by the
Ministry of Culture and research supported by the project CZ 1.05/1.1.00/02.0060 from the European Regional Development Fund and the Czech Ministry for Education, Youth and Sports.
References
1.
ASTM: Annual Book of ASTM Standards, Section 4. Construction. 04.10 Wood. ASTM,
Philadelphia, PA. 2002.
89
2.
ASTM: D 143-94 Standard Test Methods for Small Clear Specimens of Timber. ASTM
International. For referenced ASTM standards, 2000.
3.
Bodig, J.: Mechanics of wood and wood composites. Malabar: Krieger Publish. Comp. 1993,
712 pp.
4.
ČSN 49 0110: The breaking strength in compression along fibres, 1980.
5.
ČSN 49 0111: Tests of the properties of solid wood. Method of detection of the module of
elasticity in compression along fibres, 1992.
6.
Drdácký, M., Jirovský, I., Lesák, J., 2003, Non-destructive survey of masonry and timber
structures of the Maria Tower on the Karlštejn Castle, Research Report ITAM-ARCCHIP,
Praha, 2003, 62 p.
7.
Drdácký, M., Jirovský, I., Slížková, Z.: On Structural Health and Technological Survey of
Historical Timber Structures. Proceedings of the International Conference The Conservation
of Historic Wooden Structures, Florence, Vol. I, 2005. pp. 278-284.
8.
Drdácký, M., Kloiber, M.: In-situ compression stress-deformation measurements along the
timber depth profile. In: Advanced Materials Research 778: Trans Tech Publications, 2013,
pp. 209-216.
9.
Kasal, B., Anthony, R.: Advances in in situ evaluation of timber structures. Progress in
Structural Engineering and Materials. John Villey & Sons Ltd. London. UK. Vol. 6 No 2
April-June 2004, pp. 94–103.
10. Kasal, B., Drdácký, M. Jirovsky. I.: Semi-destructive methods for evaluation of timber
structures. Structural Studies, Repairs and Maintenance of Heritage Architecture VIII. C.A.
Brebia, Editor. Advances in Architecture. WIT Press. Southampton, 2003, pp. 835–842.
11. Kasal, B.: Semi-Destructive Method for In-situ Evaluation of Compressive Strength of Wood
Structural Members. Forest Products Journal, 53 (11/12), 2003, pp. 55–58.
12. Kloiber, M., Drdácký, M., Tippner, J., Sebera, V.: New construction NDT device for in situ
evaluation of wood by using compression stress-deformation measurements parallel to grain.
In: 18th International Nondestructive Testing and Evaluation of Wood Symposium, 2013,
September 24-27, Madison, Wis-consin, USA. FPL-GTR-226: pp. 585-592.
13. Kloiber, M., Frankl, J., Drdácký, M., Kučerová, I., Tippner, J., Bryscejn, J.: Change of
mechanical properties of Norway Spruce wood due to degradation caused by fire retardants.
In: Wood Research, Vol. 55(4): 2010, Slovakia. ISSN: 1336-4561, pp. 23-38.
14. Kloiber, M., Kotlínová, M., Tippner, J.: Estimation of wood properties using pin pushing in
method with various shapes of the penetration pin. In: Acta Universitatis Agriculture et
Silviculturae Mendelianae Brunensis, ISSN: 1211-8516, LVII, 2/2009, Brno, pp. 53-60.
15. Kloiber, M., Kotlínová, M.: Prediction of mechanical properties by means of radial cores in
situ. In: In-situ evaluation & non-destructive testing of historic wood and masonry structures,
RILEM Workshop, 10-14 July 2006, Prague, ISBN: 978-80-86246-36-9, pp. 56-65.
16. Kloiber, M., Tippner, J., Drdácký, M.: Semi-destructive Tool for "In-situ" Measurement of
Mechanical Resistance of Wood. In: SHATIS´ International Conference on Structural Health
Assessment of Timber Structures, June 2011, Lisbon, Portugal. 3 pp.
17. Kloiber, M., Tippner, J., Hrivnák, J., Praus, L.: Experimental verification of a new tool
for wood mechanical resistance measurement. In: Wood Research 57(3): 2012, Slovakia.
ISSN: 1336-4561, pp. 383-398.
18. Lear, G.Ch.: Improving the Assessment of In Situ Timber Members with the Use of
Nondestructive and Semi-Destructive Testing Techniques. M.S. thesis – North Carolina
State University, Raleigh, 2005, 137 pp.
90
19. Maddox, J., Drdácký, M., Kloiber, M, Kunecký, J.: In-situ assessment of strength of historic
th
wood In: 9 International Conference on Structural Analysis Historical Constructions, 2014.
20. Machado, J.S.: In situ Evaluation of the Reference Properties of Structural Timber Members.
Use of Available Tools and Information. In: Advanced Materials Research, 2013, Vol. 778:
pp. 137-144.
21. Micka, M., Minster, J., Václavík, P.: Compression test of a timber core – An-sys model
and Moiré interferometry experiment. In Engineering mechanics 2006, Svratka, 2006,
pp. 219/1-219/12.
22. Riggio, M., Piazza, M.: Hardness test. In: In Situ Assessment of Structural Timber: Discussion
of Classical and Modern Non-Destructive and Semi-Destructive Methods for the Evaluation
of Wood Structures; Series: RILEM State of the Art Reports, Kasal B.; Tannert T. (Eds.),
Springer, 2011, Vol.7: pp. 85-95.
23. Ronca, P., Gubana, A.: Mechanical characterisation of wooden structures by means of an in
situ penetration test. Elsevier Publishing Co., Oxford, England. Construction and Building
Materials, 1998, Vol. 12(4): pp. 233-243.
24. Schwab, E., Wasshau, A., Willetneier, H.: Bohrkerne. Zur Beurteilung der Festigkeit holzerner
Rammpfahle. Bauen mit Holz (9), 1982, pp. 566–570.
25. Tannert, T, Anthony, R., Kasal, B., Kloiber, M., Piazza, M., Riggio, M., Rinn, F., Widmann,
R., Yamaguchi, N.: Recommendation of RILEM TC 215-AST: In-situ assessment of structural
timber using semi-destructive techniques. In: Materials and Structures: 2014, Netherlands.
ISSN: 1359-5997. DOI: 10.1617/s11527-013-0094-5
26. Tippner, J., Kloiber, M., Hrivnák, J.: Derivation of Mechanical Properties by Pushing of a Pin
into Wood. In: 17th International Nondestructive Testing and Evaluation of Wood Symposium,
9/2011, Sopron, Hungary. ISBN: 978-963-9883-83-3, Vol. 2: pp. 575-582.
27. Yamaguchi, N.: Screw Resistance. In: In Situ Assessment of Structural Timber: Discussion
of Classical and Modern Non-Destructive and Semi-Destructive Methods for the Evaluation
of Wood Structures; Series: RILEM State of the Art Reports, Kasal B.; Tannert T. (Eds.),
Springer, 2011, Vol.7: pp. 81-86.
91
92
Practical procedure for estimating the density of
timber with portable X-ray equipment
Thomas lechner, PhDa), Robert Kliger, Professorb)
a)
NCC Construction Sverige AB and Chalmers University of Technology,
Gothenburg, Sweden
b)
Chalmers University of Technology, Gothenburg,
Sweden Abstract
Wood density has a strong relationship with several mechanical properties, such as
strength and stiffness. An X-ray image calibration procedure, which enables the
determination of density for the in-situ assessment of timber structures, has been
developed. This non-destructive method is useful for evaluating the internal
condition for global assessments of the structure. Finally, a calibration wedge was
set up to verify the procedure. The density calibration procedure from X-ray
images was verified on a timber beam specimen, resulting in good agreement and
an average accuracy of 97%. The values obtained from the image calibration
presented a very good linear correlation between the measured density and the
greyscale from X-ray images, with coefficients of determination (R2) ranging from
0.90 to 0.98. The main advantage compared with conventional techniques is the
detection and quantification of internal damage, defects, disturbances and
deterioration that may reduce the mechanical properties of the structure. This study
shows good potential when it comes to the development of a viable tool for in-situ
assessments of timber structures. In a subsequent step, the above-mentioned
procedure was also implemented in two case studies of historical wooden
structures in Sweden. This technique could be used indirectly in analyses of
structural behaviour.
Introduction
Timber structures are generally complicated when it comes to condition
assessment. This is particularly true in the case of structures and buildings of historical
value, when ageing can be suspected of having a diminishing effect on their
strength and stiffness or abnormal structural behaviour has been discovered. It is
therefore of great importance to adopt structural health-monitoring techniques to
assess the remaining load-bearing capacity of timber structures. Through the
reliable and appropriate assessment and monitoring of timber structures, it is
possible to detect any weaknesses at an early stage and appropriate action can then
93
be taken to extend the service life of the structures. This kind of assessment
often requires appropriate non-destructive testing (NDT) and quasi-nondestructive testing techniques. Improved, new methods based on scientific
knowledge and guidelines are needed for their application. The application of
digital imaging processing and increasing resolution has made it possible to
quantify and assess components inside timber elements, such as the internal
deformation of fasteners and the dimensions of hidden elements and strains [1].
However, the aim of this paper is to propose a practical methodology for
assessing the density of timber in situ using X-ray measurements. These
measurements are based on X-ray attenuation, as it is impossible to measure
wood density directly with X-rays. Density is an important parameter when
defining wood materials. Wood density is strongly related to stiffness and
strength properties and can therefore be used in the evaluation of timber
structures [2]. The definition is, however, problematic, as both mass and
volume are dependent on moisture content. In this case, the estimated density
relates to structural timber in the in-situ condition and the moisture content is
related to the relevant service class. The basic principles of X-ray attenuation
were used in this study to establish an in-situ calibration procedure to predict
density on site. The method which will be described produces local property
values of density measurements using an X-ray calibration procedure.
Furthermore, this procedure can also be used to evaluate density distribution in
all timber structures. The variation in density in a piece of timber can be seen
using photographic imaging [3].
General specifications of the X-ray equipment
Radiographic investigations can be conducted using X-ray densitometry,
diffraction analysis, computed tomography (CT) and micro-tomography and they
are mainly used for the quantitative assessment of wood properties of different
species, the detection of deterioration and to acquire knowledge of whether the
element is sapwood or heartwood by researching the density profile. These
differences can be detected through the attenuation of X-rays [4-5].
X-rays are short-wavelength electromagnetic radiation travelling at the speed of
light. These rays are not affected by electromagnetic fields and can be diffracted
but not deflected. Emitted X-rays lose intensity and this appears as lighter/darker
in terms of greyscale values (RGB) on the imager [6].
The penetration capability and intensity of the radiation are controlled by the
electric potential (kV), the current (intensity, mA) of the X-ray tube and the expo-
94
sure time. The penetration is the intensity projection on the image plate and it is
governed by:
The type of material and the material characteristics
The material composition
The density of the material
The porosity of the material and its moisture inclusion
The attenuation factor (µ)
The penetration thickness of the X-rayed object.
In digital form, the image can be expressed as a matrix and processed in an
image-processing toolbox, such as the one in Matlab®, to quantify and compare
relative positions in real-time radiography, for example.
The equipment used in the studies was a battery-powered portable X-ray
source, Inspector XR200® from Golden Engineering Inc. However, other X-ray
equipment can be used in situ. The digital image plate system, DIMAP® from
Logos Imaging Inc., was used to scan the photographic X-ray images using
reusable phosphor-layered imaging plates that capture the intensity levels from
the X-ray exposed objects. An example of an X-ray set-up, which includes the
principle for the X-ray procedure, X-ray source and the digital image plate
system, is shown in Fig. 1. The X-rays leave the tube at a certain exit angle,
which is decisive for the minimum distance to the specific object to ensure the
maximum utilisation of the imaging plate.
The laser scanner releases the accumulated energy from the image plate and
stores the image at selectable resolution on the laptop imaging software, where
every single image can be post-processed in terms of specific details. The safety
instructions from the manufacturer regarding the danger of X-rays must be
followed carefully in order to avoid any kind of harm.
Fig. 1 Example of X-ray system and recording process. The upper part illustrates the principle for the X-ray procedure, whereas the lower part shows the X-ray source and the digital image
plate system, DIMAP® from Logos Imaging Inc.
95
Applications of radiographic measurements
Until recently, the opportunities for X-ray investigation have been used for the
qualitative assessment of timber structures, but the opportunities to carry out
quantitative evaluations are of great importance. A number of applications for
using X-ray equipment on site, which can be useful for the evaluation of structural
behaviour, are revised in this paper.
Depending on the material properties, such as the chemical properties, density
and thickness of the inspected object, the energy absorption will vary and this is
reflected by the photographic image [3]. Anthony also investigated termite activity
using infrared thermography and acoustic non-destructive methods but without
satisfactory success when it came to quantifying the loss of material [3]. By
comparing the measured intensities on a radiograph, the extent of deterioration
in wood members could be quantified using image processing techniques [3, 6].
Image enhancement techniques are a common technique when adjusting digital
images [7] to facilitate the interpretation of deteriorated wood. The primary benefit
when using X-rays is the opportunity to determine the condition of structures on
site. Further advantages are the ability to acquire precise dimensions by measuring
the distances between the X-ray source, the imager and the object of interest, the
ability to identify the physical condition of wood such as rot, for example, to
estimate the structural capacity, identify the types of connection, such as nails,
bolts and so on, to detect corroded areas and to identify the construction details for
historical dating [6].
All these advantages involve some difficulties, especially in crack identification,
which requires an adequate size of at least 2% of the member thickness and must
be oriented parallel to the radiation in order to be detected [8]. Limitations to the
intensity or energy level can also limit the investigation [8].
Real-time radiography (radioscopy) allows the study of component behaviour
under moderate loads and it is particularly suitable for timber structures due to the
density differences.
There are also several limitations to using X-ray for detecting defects (including
cracks). They can be discovered with visual inspections and may be related to
chemical contamination, variations in the attenuation coefficients, variations in the
dimensions and the distribution of moisture.
Detection of corroded areas
As corrosion in metal fasteners might cause severe failure, radiographic equipment
can be used as a tool to detect corrosion inside the structure and, by taking
appropriate action, the collapse of the structure could be prevented [3]. Using
commercial image-editing programs, distances can be measured fairly accurately
in relation to a reference unit and the actual capacity of the fastener can be
96
recalculated. Fig. 2 shows a nail that has corroded as a result of a shrinkage crack
in timber.
Fig. 2 Deterioration of the metal fastener due to corrosion in the shrinkage crack of the beam.
Reduction of cross-section
Old timber may have lost its full capacity due to deterioration either as a result of
insect attacks or due to shrinking cracks [9]. Moreover, fungal degradation may
reduce a cross-section. When accessibility with an X-ray camera along the fibre
direction is guaranteed, a more reliable cross-section can be obtained and
predictions of the maximum permissible stresses at a specific point could be made
with a reduced cross-section, before any strengthening or remedial work is carried
out.
“Timber-to-timber” hidden geometry
Hidden metal details in a timber structure can be assessed using X-ray. However,
this does not necessarily mean that hidden timber details can be visualised with
sufficient accuracy. As part of the current investigation, a preliminary study has
been carried out and it shows promising results in this field, cf. Fig. 3.
Fig. 3 A hidden dowel with approximately the same density as the surrounding wood can be
detected using X-ray. Original X-ray image (top left corner) vs. edited image. The numbers correspond to the mean density through the thickness of the beam at different positions.
97
Density distribution in components
The development of the equipment and methods for digital image analysis has
made it possible to determine variations in apparent density values and distribution
in timber and wood composites. These differences can be detected through the
attenuation of X-rays passing through the material [5,10].
Determination of material properties through image calibration
X-rays are already in use as a means of determining material properties and
strengthening grade timber. The current methods are not suitable for in-situ
assessment and are outside the scope of this article.
In-situ methods for determining material properties are most probably available
for materials with great homogeneity in their properties, such as steel [11]. As
timber is a material with large-scale variations in all its physical and mechanical
properties, these methods cannot be applied without further reflection and awareness
of the scatter in the obtained results.
Failure modes in metal fasteners
In-situ X-ray imaging also provides an opportunity to determine the actual
behaviour of dowels in joints, see Fig. 4 [3]. Obtaining the dimensions of non-visible
fasteners or cross-section reductions, as well as the connection of joints that are
decisive for the judgement of boundary conditions, offers an excellent opportunity
for further interpretation in the structural analysis [4]. Moreover, the exact
position of the plastic hinges (bent nails in Fig 4) can be easily determined.
Fig. 4 The X-ray images show the behaviour of a nailed joint (left) and a bolted connection
(right) [4].
98
Calibration procedure
A practical methodology for assessing the density of timber in situ using X-ray
measurements is proposed. First, the in-situ procedure, the preparation of the
equipment and the need for a calibration wedge are described. The thickness of the
member and the moisture of the X-rayed component have to be measured for
calibration purposes.
The calibration wedge, see Fig. 5, is placed by the component that is going to
be imaged, wherever there is room (above, below or next to). In order to minimise
the effects of the exit angle from the X-rays in the tube (larger dosage in the centre
and less dosage at the edges), the targeted distance from the generator to the Xrayed object is between 2.5 and 3 metres [12]. In practical applications, this
distance is seldom possible to achieve, so the effects on the image have to be
adjudged in the evaluation process when the density variation is obtained. The
parameters that may affect the measurement results include the distance to
the object, the exposure time, the energy level and the irradiation geometry
(such as source size and fan angle).
Subsequently, the density evaluation process of the X-rayed component starts
using image processing software. The first step here is to establish a density
relationship from the calibration wedge by determining the mean greyscale
values from the specimens with different densities. The density variation between
specimens can be represented by a trend-line equation, see Fig. 5.
The coefficient of determination (R2) in the equation should be 0.90 or higher
and the number of specimens from the calibration wedge included on the image
should be at least eight. The number of specimens included in the equation is
dependent on the image disturbances and the distance to the X-rayed object. So,
the number of specimens has to be evaluated from image to image. As a result,
the calibration wedge should consist of 10 to 15 specimens with dimensions that
can be captured on the image plate, cf. Fig. 1.
99
Fig. 5 Example of X-rayed beam (on top): calibration wedge (below), consisting of eight
clearwood specimens with varying densities, and its established trend-line equation (bottom).
The mean greyscale of the inspected area (µmean) has to be determined and corrected in relation to reference values for moisture (±µMC). This mean value is then
inserted in the trend-line equation, where a preliminary density value is obtained.
Finally, this value is corrected in relation to reference values from the calibration
wedge with a correction factor (∆RGB) for the thickness, according to Eq. (1). The
∆RGB value is tabulated in Tab. 1 and the µMC value is tabulated in Tab. 2.
µmean_corr = µmean + (∆µMC) + ∆RGB
Eq. (1)
Image evaluation procedure
Due to the cone-beam effect of the portable X-ray equipment, where the X-ray
dosage is not evenly spread in the raw X-ray image, image corrections using imaging
software, e.g. ImageJ®, are a great advantage in digital images and are needed in
any visual application. The relevant attenuation ratio (I/I0) can be measured using
imaging software, where I is the intensity of the X-ray beam after penetration of the
sample and I0 is the initial intensity. The ratio can then be calculated as an average
value over the energy spectrum [13].
In practical applications, this principle can be used as follows: to correct the
defaults/noise level of the raw image, several small steps must be applied to evaluate
the noise level and subtract it from the complementary background image without
illumination. A further image with X-ray illumination accounting for the non-
100
uniformity of the cone-beam effect to reach the final pixel grey value is needed.
This principle of image correction due to background noise level was applied in the
in-situ determination of the material density in the different projects. The principle
for the image correction procedure is illustrated in Fig.6.
(a)
(b)
(c)
Fig. 6 The procedure in principle for image background correction due to the cone-beam effect according to [10] at micro-level: (a) raw image, (b) noise level due to cone-beam effect, (c)
final corrected image.
Calibration wedge
The X-ray images of beams and joints can be calibrated in a further step
towards obtaining their density by a calibration specimen comprising at least
eight to 15 different wood specimens with different density characteristics, see
the example in Fig. 7. The thickness (depth) of the specimens is restricted to
30 mm, which is the reference thickness for the thickness calibration.
The specimens need to be conditioned at a relative humidity (RH) level of
about 66%, which corresponds to about 12% (air-dried density reference level) in
moisture content (MC). The conditioning process can be performed in a climate
chamber with a constant temperature level of 20ºC, or, preferably, the specimens
can be stored in a climate-controlled box where a saline solution (NaNO2; sodium
nitrite) maintains the RH at a level of about 66% [14].
Fig. 7 Calibration wedge composed of wood specimens (in this case 13) with increasing density.
Correction for thickness
This section mainly provides the instructions for using the tabulated ∆RGB values, see
Tab. 1. The ∆RGB values are based on the reference thickness of the calibration wedge,
101
which is 30 mm. The values were taken from the study carried out at Chalmers
University of Technology, and are the result of the principle of radiation energy
absorption, when X-rays pass through the object [9].
First, a qualified expectation of the density range is formulated, i.e. typical
softwood, for example, ranges in density between 300 and 500 kg/m3, as the ∆RGB
value is a function of both the density of wood and the thickness of the object or
specimen. Once the first estimated prediction of density is made, this first calculated
density value can be further corrected for ∆RGB using linear interpolation between
points representing 400, 550 and 800 kg/m3 respectively. However, only close to
the range limits, i.e. 521 and 579 kg/m3 (second column in Tab. 1, for example), it
may change very slightly. Due to many other uncertainties in the timber material,
making this further correction is not thought to be necessary. The obtained
value from Tab. 1 can then be inserted in Eq. (1) and the corrected mean value
(µcorr) can be calculated.
Tab. 1 Change in ∆RGB [RGB] on the image due to the influence of the thickness of the
specimen – tabulated and graphical.
102
Correction for moisture
In this case, the appropriate factor for the measured MC should be inserted in Eq.
(1). The µMC value between two moisture ranges has to be interpolated from Tab.
2.
This procedure is optional, at least for the range between 8% and 16% MC, as
the MC only has a small influence on the attenuation of X-rays. The influence of
moisture that was captured in this study could reveal a larger scatter between
different images.
The µMC values are based on the reference MC for air-dried density conditions
(12%) and decrease or increase linearly by approximately 1.25-1.75 [RGB] for
every 1% change in MC.
Tab. 2
Linear change in ∆µMC [RGB] on the image due to the influence of different moisture
conditions in timber.
~MC
[%]
6%
8%
10 %
12%
14 %
16%
20%
22%
~RHequ
[%]
35%
43%
50%
66%
79%
85%
90%
93%
∆µMC
[RGB]
7.5- 10.5
2.5-3.5
0 .0
2.5-3.5
5.0-7.0
5.0-7.0
103
10.0-14.0 12.5-17.5
Limitations of the procedure
The calibration of the images is made from a subjective point of view and
has to be performed for each individual image.
The uneven distribution of radiation dosage is restricted by the distance
between the generator and the X-rayed object, so the light that appears
from the power initiation of the X-ray tube has to be considered in
the evaluation process, but the shadowed areas can easily be detected and
excluded from the evaluation following an image evaluation procedure.
The size of the image plate restricts the overall evaluation of the structural
member. Imaging plates of several sizes are available.
The method cannot be directly applied to composite structures and
therefore requires further investigation and improvement.
Discussion and conclusions
Based on the results of a study carried out by [15-16], accurate estimates of the
in-situ density of timber have been obtained using the procedure described in this
paper. When using this procedure, the linear correlation is valid for timber densities
within a range of 250 kg/m3 to 1000 kg/m3. Furthermore, adjustment for the
thickness of the evaluated specimen has a significant influence on density and
cannot be ignored when the density of timber is evaluated.
The influence of the moisture content can, however, be ignored, as it has only a
minor impact on the attenuation of the X-rays; at least for the range between 8%
and 16% MC, which is typical for the structural timber in-situ in Service Classes 1
and 2 [17]. The influence of moisture could have produced a larger scatter
between different images. Uncertainty relating to the accuracy of the method when it
comes to obtaining the density values in situ can be attributed to various factors,
such as the individual image and the conditions governing access to the object that
is going to be X-rayed. The distance also impacts the individual images as a result
of the cone-beam effect, which necessitates a background correction. It should,
however, be remembered that this is a direct measurement method which is used
to obtain the density of an object, with a calibration wedge in the actual image.
Further development of the density calibration procedure is needed to apply the
procedure to composite materials or wall elements with several different layers.
As a general conclusion, digital radioscopy provides real potential for the
development of a successful future tool for the in-situ assessment of timber
structures and it is as easy or difficult to use as any other non-destructive method.
It also contributes to the detection of failures and deterioration in the material in
the early stages, which may provide a better basis for making correct decisions
104
when it comes to repairing or replacing the timber in question and, by doing so,
increasing the service life and durability of the structure.
As commonly stated for timber engineering purposes, the density governs the
stiffness and to some extent also the strength properties. The procedure can therefore
be used in the analysis of structural behaviour. The density of timber components
is also relevant for the examination and evaluation of mechanical connections.
References
[1]
Kasal B., Adams A. and Drdacky M. (2008): Application of Digital Radiography
in evaluation of Components of Existing Structures. RILEM Symposium on On
Site Assessment of Concrete, Masonry and Timber Structures - SACoMaTiS 2008.
Varenna - Lake Como, Italy.
[2]
Dinwoodie J. (2000): Timber: Its nature and behaviour. London, Taylor & Francis.
[3]
Anthony R. (2003): Examination of Connections and Deterioration in Timber Structures
Using Digital Radioscopy. American Society of Civil Engineers, Vol., pp. 320-328.
[4]
Tomazello M., et al. (2008): Application of X-ray Technique in Nondestructive Evaluation
of Eucalypt Wood. Maderas. Ciencia y tecnología, Vol. 10 (2), pp. 139-149.
[5]
Rinn F., Schweingruber F. and Schär E. (1996): Resistograph and X-ray density charts
of wood comparative evaluation of drill resistance profiles and X-ray density charts of
different wood species. Holzforschung, Vol. 50 (4), pp. 303-311.
[6]
N.C.P.T.T. (2005): Advances in Digital Radioscopy for Use in Historic Preservation.
National Center for Preservation Technology and Training, pp. 125.
[7]
Maini, R and Aggarwal H (2010): A Comprehensive Review of Image Enhancement
Techniques. Journal of Computing, Vol. 2 (3), pp. 8-13.
[8]
Lear G. C. (2005): Improving the Assessment of In Situ Timber Members with the Use
of Nondestructive and Semi-Destructive Testing Techniques. Master of Science Master's
Thesis, Civil Engineering, North Carolina State University.
[9]
Brozovsky J., Brozovsky J., Jr. and Zach J. (2008): An Assessment of the Condition
of Timber Structures. 9th International Conference on NDT of Art.
[10] Chen S., et al. (2010): Digital X-ray analysis of density distribution characteristics of woodbased panels. Wood Science and Technology, Vol. 44 (1), pp. 85-93.
[11] Bateni A., Ahmadi M. and Parvin N. (2008): Prediction of density in porous materials by
x-ray techniques. 17th World Conference on Nondestructive Testing. Shanghai, China.
[12] Hughes J. F. and De Albuquerque Sardinha R. M. (1975): The application of optical
densitometry in the study of wood structure and properties. Journal of Microscopy, Vol.
104, pp. 91-103.
[13] Badel É. and Perré P. (2002): Predicting oak wood properties using X-ray inspection:
representation, homogenisation and localisation. Part I: Digital X-ray imaging
and representation by finite elements. Ann. For. Sci., Vol. 59 (7), pp. 767-776.
[14] Nevander L. and Elmarsson B. (1994): Fukthandbok: Praktik och teori. Stockholm,
AB Svensk byggtjänst.
[15] Lechner T., Sandin Y. and Kliger I. R. (2013): Assessment of Density in Timber Using Xray Equipment. International Journal of Architectural Heritage: Conservation, Analysis
105
and Restoration, Vol. 7 (4), pp. 416-433.
[16] Lechner T. (2013): In-situ assessment of timber structures - Assessment methods and case
studies. Doctoral Thesis, Civil and Environmental Engineering, Chalmers University
of Technology, pp. 62.
[17] EN 1995 (2004). Eurocode 5: Design of Timber Structures. European Standard EN
1995 :2004, Comité Européen de Normalisation CEN.
106
A methodology for the determination of the
timber density through the statistical assessment
of ND measurements aimed at in situ
mechanical identification
University of Naples Federico II, Dept. Structures for Engineering and Architecture, Italy
[email protected]; [email protected]; [email protected]
Abstract
The research context [1-5]. A wide experimental activity has been developed in
the framework ofthe Italian project DPC-RELUIS 2010 – 2013, Line 1, Task 1 and
it is ongoing on the current DPCRELUIS 2014, with the aim to
provide a procedure for in situ mechanical identification of ancient timber
members made of old chestnut wood (Castanea sativa Mill.) by means of
non-destructive techniques (ND) and to define standardized guidelines to be
used for practical applications [STAR Paper 1]. The following specimens with
standard dimensions according to UNI EN 480 (2004) UNI ISO and 3789
and 3132 Italian codes (1985) were obtained: structural elements in actual
dimensions (SA), squared elements in small dimensions (SS) and defectfree specimens (DF). The following ND techniques are employed [4]:
Hygrometric tests (H), for the evaluation of the moisture content of
wood, Ultrasonic (U) tests, for the determination of the elastic properties
of wood, Sclerometric (S) tests, for the assessment of the quality and
surface hardness, Resistographic (R) tests, for the detection of density
variations and internal defects of wood. Destructive tests (DT) in compression
(C) and in bending (B) were performed in order to assess stiffness and strength
properties, post-elastic behaviour and collapse mechanisms of chestnut timber.
By means of a statistical approach, linear regression of the following relations
are examined: correlations between NDT parameters, relating the L and T
measures by U, S and R tests; correlations between DT parameters, defining the
mechanical behaviour in C and B; NDT-DT correlations, using both simple and
multiple models for ND estimation of density, modulus of elasticity and
strength of the material.
107
The definition of the methodology
Based on the results obtained, the most reliable correlation for achieving the
mechanical properties of timber is the one defining a relationships among the density
and the mechanical properties. Therefore the correct determination of density is
compulsory. To this purpose, a method that allows the in situ identification of
density of old chestnut timber members, through the use of ND techniques, has
been developed. Starting point of the procedure consists in the individuation of the
most reliable correlation between the examined parameters. The best linear regression
equation found is the one between the density (ρ) and the sclerometric (PT) and
resistographic (Am,T) parameters in transverse (T) direction, for the small
dimensions specimens.
This could be used considering that SA and SS specimens show similar behaviour
in terms of strength, stiffness and collapse modes.
Tab. 1. Correlation between density [ρ (kg/m3)] and sclerometric [PT (mm)] and resistographic
[Am,T (%)] measures
The correlation allow for the determination of a so called theoretical value of
the density (ρt), it being affected by some approximation. For enhancing the
accuracy of the estimation of the actual density (ρs), which is measured in
laboratory, a correction coefficient Cadj, ρ is introduced, obtaining a design density
value (ρd = ρt/Cadj, ρ). Such a coefficient is defined as the maximum value (for the
sake of safety) of the ratio ρt/ρs evaluated for every examined specimen, it being
quite always larger than 1. The assumed value is Cadj, ρ=1.15.
The proposed method allows the in situ density identification (ρd) of
timber element by only ND tests with an error quantified at most as 13% of
the actual value ρs (in reduction).
Once estimated the density, the strength and modulus of elasticity both
in compression and in bending are evaluated through the correlations with
density obtained for SA specimens.
Indications for executions
In order to comply to the above procedure, indications for the executions are
provided. After the preliminary phases consisting in traditional visual inspections
for evidencing any possible significant defect or degradation, lack of members,
108
together with the geometrical survey, the cross sections for testing should be
identified. This is a very important task because the selected cross sections should
be representative of the timber consistency. The number differs if the members is
subjected to compression or bending. Then instructions for the application of
sclerometric and resistographic tests are given, concerning either the preparation
of the testing surface, or the mapping and the number of the shots (the last differing
if the member is subjected to compression or bending), the data reading and
elaboration.
Conclusions
The proposed method allows the in situ density identification ( ρd) of timber element
by only ND tests with an error quantified at more as 13% of the actual value (ρs).
The methodology also allows to perform NDT in the direction transverse to the
axis of the tested member. In this way, it is possible to solve the problem of
the access at the end sections of members, which often involves the inconvenience
of dismantling the entire structures. Further experimental activities would allow to
reach more robust correlations for in situ mechanical characterization of existing
timber members.
Acknowledgments
The research activity was developed within the Italian projects PRIN 2006 and DPC-RELUIS
2010-2013 and RELUIS 2014 (B. Faggiano responsible of research unit). Authors acknowledge
Letizia Esposito and Francesco Grasso for the contribution to the set-up of the method, given
within the research activity.
Main authors references
[1]
Faggiano, B., Grippa, M.R., Marzo, A., Mazzolani, F.M.: Experimental study for
nondestructive mechanical evaluation of ancient chestnut timber. Journal of Civil
Structural Health Monitoring: Volume 1, Issue 3, page 103-112, 2011.
[2]
Faggiano, B., Grippa, M.R., Marzo, A., Mazzolani, F.M.: Experimental analysis on
old chestnut timber by means of non-destructive techniques. 11th World Conference on
Timber Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June 2010.
[3]
Faggiano, B., Grippa, M.R., Marzo, A., Mazzolani, F.M.: Structural grading of old
chestnut elements by compression and bending tests. 11th World Conference on Timber
Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June 2010.
[4]
Grippa, M.R., Faggiano, B., Marzo, A., Mazzolani, F.M.: Combined methods for in situ
mechanical identification of ancient timber structures based on non-destructive tests. 11th
World Conference on Timber Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June
2010.
109
[5]
B. Faggiano, M.R. Grippa, B. Calderoni, 2013. Non-destructive tests and bending tests
on chestnut structural timber. Advanced Materials Research Vol. 778 (2013) pp
167-174© (2013) Trans TechPublications, Switzerland, doi:10.4028/www.scientific.net/
AMR.778.167
110
Evaluation of the influence of defects on the
mechanical properties of timber through the
analysis of multiscale specimens, based on NDT
and DT
Abstract
The research context: The work illustrated is part of the extensive experimental
activity, developed within the Italian projects PRIN 2006 and DPC-RELUIS
2010-2013 and 2014 (B. Faggiano responsible of research unit) [1, 2, 3, STAR 1].
According to a defined investigation procedure, non-destructive (ND:
hygrometric, sclerometric and resistographic) and destructive tests (D: compression
and bending) on samples made of old chestnut wood were performed with the aim
of obtaining reasonable ND-D correlations based on statistical elaborations for
the prediction of wood density, stiffness and strength of the tested material.
The paper focus
The experimental activity was developed on multiscale specimens, such as
structural timber and defect free clear elements, with the aim to evaluate the influence
of natural defect patterns of material on both non-destructive parameters by
sclerometric and resistographic tests and mechanical properties by compression
and bending tests of timber.
Multiscale specimens and tests methods
The experimental campaign was developed on timber elements made of
old chestnut wood (Castanea sativa Mill.), provided from roofing trusses of a
masonry building of Naples, built up at the beginning of the 19th century. The
111
investigations were carried out on both structural elements in actual and small
dimensions (types SA and SS respectively) and defect-free (types DF) specimens
with standard sizes, according to UNI EN 480 (2004) and UNI ISO 3789 and
3132 Italian codes (1985). Before the experimental tests, the conservation state of
the selected samples was detected by checking wood defects and damage, such as
longitudinal cracks due to shrinkage ring shakes, large isolated knots or knots
groups. Ultrasonic (U), sclerometric (S) and resistographic (R) ND methods were
used for SA and SS specimens, whereas compression (C) and bending (B) tests
were performed on both S and DF elements. Fig. 1 summarises the multiscale
specimens and tests types [3].
Fig. 1. Multiscale specimens tested in the experimental campaign.
Experimental evidences
The results of compression tests parallel to grain show that the presence of
extended natural defects on SA reduces of about three times the compression
strength of the base clear material, being the stiffness properties nearly similar
each other (Fig. 2a). Furthermore, the bending strength of DF samples is about
twice than the same one of SA and SS elements. It is also apparent that the
bending modulus of elasticity in clear wood is slightly larger than in structural
timber (Fig. 2b) [1, 2, 3].
Concerning the influence of macroscopic defects on NDT parameters, it is
observed that the presence of superficial layers of wood with low consistence
increases of about 18% the penetration depth values by sclerometric test in transversal direction. Moreover, extended knots and internal more resistant parts produce
an increment of the resistographic amplitude of about 40%.
112
Fig. 2. Average curves: a) Stress ( σc,0)-strain (εc,0) for C test; b) Strength (fm)-middle rotation
(ϕ) for B tests.
Acknowledgments
The research activity was developed within the Italian projects PRIN 2006 and DPC-RELUIS
2010-2013 and 2014..
References
[1]
Faggiano B., Grippa M.R., Marzo A., Mazzolani F.M., 2011. “Experimental study for
nondestructive mechanical evaluation of ancient chestnut timber”. Journal of Civil
Structural Health Monitoring: Volume 1, Issue 3, page 103-112.
[2]
Faggiano B., Grippa M.R., Marzo A., Mazzolani F.M., 2010. “Structural grading of
old chestnut elements by compression and bending tests”. 11th World Conference on
Timber Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June.
[3]
Faggiano B., Grippa M.R., Calderoni B., 2013, “Non-destructive tests and bending tests on
chestnut structural timber”. 2nd International Conference on Structural Health Assessment
of Timber Structures (SHATIS’13), Trento, Italy 4-6 September.
113
114
Screw withdrawal resistances for reliabilitybased evaluation of timber in existing structures
Nobuyoshi Yamaguchi
Dpt. of Building Materials and Components, Building Research Institute, Tsukuba, Japan
Abstract
Screw withdrawal measurement is a semi-destructive testing method, which are
used in detailed inspections of existing timber structures. Screw withdrawal
measurements had used wood-screws for probes; the method using metric-screw
type probes with short-threads has been developed. The new screw withdrawal
measurements are able to estimate physical/mechanical properties of timber such
as their densities and shear strengths. Distribution of properties along the timber
depths is also obtained from coaxial multiple withdrawal resistance measurements.
Estimated properties from these screw withdrawals are applied for the evaluation
of degradation of timber structures. Benchmark method, nominal value method
and structural calculation method are proposed for the evaluation. The benchmark
method and nominal value method use integrity indexes of components and joints
of timber structures. Properties estimated from these screw withdrawals are also
used in structural calculations of existing timber structures. These indexes and
results of structural calculations are applied for the evaluation of existing timber
structures. Physical/mechanical properties estimated from these screw withdrawals
are able to apply for reliability-based evaluation of existing timber structures.
Introduction
Scales of Inspection Objects
Timber structures use many structural timbers and other materials. The components
have joints connecting them to the others. Integrity of structures is dominated by
the integrity of their components and joints. Most of joints use fasteners such
Structure
Component & Joint
Material
Fig. 1 Scale of Inspection Objects
115
as pins, nails, bolts and metal plates, etc. Interface of these timbers and metal hardware
tend to accumulate water or humidity between them, these wet/humid interfaces
between them accelerate degradation of timber and fasteners. The performance of
these components and joints are dominated by the integrity of their materials. Fig.1
illustrates scale of inspection objects from structures to their materials. Integrity of
structures is dominated by the integrity of their materials consequently.
Homogeneous and Inhomogeneous Degradation of Components
Homogeneous degradation of timber causes global degradation of components by
weathering, aging and fungus, etc. Inhomogeneous degradation of timber causes
partial degradation of components by fungus, insects, etc. Dense degradation of
timber caused defects on the surface and inside of the components, and make effective
cross-sectional areas of them reduced consequently. Strengths of the components
are reduced by both of decrement of the cross-sectional areas and degraded stress
capacity(strength) of the residual cross-sectional area of them.
Global Inspection for Screening
A flowchart of inspection and evaluation from phase 1 to 5 is proposed in Fig.2. In
Phase 1, document and drawings on object structures, information on climate and
environment and fungus/insects information around the structure are required to be
accumulated as possible. Those are generally provided from owners and users of
the object structures. Probable degradation areas of timber structures might be
known by the people in surrounding communities empirically. Document and
drawings of the structures are the bases of global inspection in Phase 2. Timber
structures are composed of thousands of components; detailed inspections for all of
components are not cost effective. Positions of detailed inspection are to be
minimized. Screening of structures which identify the detailed inspection areas is
effective to minimize the number of detailed inspections. Basic nondestructive
testing methods as visual, sounding, knife test, small core sampling and others are
applied for the structures. These basic inspections provide clues of probable
degradation areas of the structures. Modern nondestructive testing are available
such as X-ray, bore-scopes, etc [1,2]. These tests provide visual information of
timbers behind finishing. Contact type nondestructive testing is able to provide
visual information of components such as GPR, ultra-sound reflection tools, etc.
Some semi-destructive testing is also used such as needle penetration, pin pushing,
etc. Combined usage of information from owners, empirical knowledge in
communities, non/ semi-destructive testing will realize cost effective inspections.
Global inspection will result tentative map of detailed inspection positions of the
structures. Information obtained from Phase 1 and Phase 2 would be the bases for
cost effective inspections.
116
Start
Phase 1
Phase 2
Documents
Information from Owner, Users
Drawings
Environment ,Climate
Global Inspection
Screening
Non-destructive
Non-contact testing
(Remote)
Contact testing
Semi-destructive
Fungus, Insects, etc.
In situ
Visual
X-ray
Knife test
M.C
AE
Bore scope
Needle Penetration
Drilling resistance
etc.
Sounding
Stress Wave
GPR
etc.
Pin pushing, etc.
etc.
Tentative mapping of position of detailed inspection
Phase 3
Detailed Inspection
Non-destructive
Contact testing
Semi-destructive
Dent
Holes
In situ
Knife test
M.C
AE
Bore scope
Sounding
Stress Wave
GPR
etc.
In situ
Hardness test
Pin pushing, etc.
Needle Penetration
Drilling resistance
Compression
Laboratory
Screw Withdrawals, etc
Cores
Compression
Core sampling, etc.
Micro-specimens
Shear test of Glue lines
Compression, etc.
Tension
Positions of additional detailed inspection
Phase 4
a) Benchmark method b) Nominal value method
Comparison with
benchmark timber
Comparison with nominal
values of timber
Integrity Index
c) Structural calculation
Residual properties of
components and joints
Structural calculation
(Global or Partial)
Analysis of serviceability
Estimation of risk of failure
Phase 5
Global or partial mapping of integrity index, lack of serviceability
and risk of failure on structures
Judgment of use, usage restrictions, repairs, reinforcement, demolish
(Global and Partial)
Fig. 2 Inspection and Evaluation of Timber in Existing Timber Structure
117
Detailed Inspections
In phase 3, detailed inspections are applied for the components and joints of
structures. Physical/mechanical properties of components and joints are obtained
from non/semi-destructive testing in situ and laboratories [1,3]. Non-destructive
testing such as stress wave, acoustic emission (AE) tools will be available to obtain
physical/mechanical properties. Semi-destructive testing such as pin pushing, needle
penetration and drilling resistances will assist to estimate properties of timber.
New semi-destructive testing measuring force-deformation relationship along the
depths of timber is also proposed [4]. Results of detailed inspections may require
additional inspection positions other than the tentative inspection positions derived
from the global inspection. Physical/mechanical properties obtained from detailed
inspections will be used in Phase 4.
Screw withdrawal measurements for detailed inspections
Screw withdrawal measurement is one of semi-destructive testing methods. Screw
withdrawals were examined from 1980’s for inspection method of timber [5,6]. Screw
withdrawal measurements are one of methods which leave scars in timber as other
testing methods of needle penetration, pin pushing, drilling resistance and small core
sampling, etc. Screw withdrawal measurements had used wood-screws, lag-screws
and nails as probes. New screw withdrawal measurements method using metric-screw
type probes with short-threads have been proposed [7]. Screw withdrawals are able
to provide specific densities and shear strengths of timber. Screw withdrawals also
provide distribution of them along the depths in timber [7].
Judgement
After the detailed inspection in Phase 3, three evaluation methods of benchmark
method, nominal value method and structural calculation method are prepared in
Phase 4. Physical/mechanical properties obtained from detailed inspections are
compared to the reference values in benchmark method and nominal value method.
The benchmark method and nominal value method use relative comparison between
residual properties and initial/original properties of timber. Estimated residual
mechanical properties are also used in structural calculations.
Benchmark method
Benchmark timber is need in benchmark method. Benchmark timber is the reference
selected from the same species as the object timber. The benchmark timber is
required to have average properties of the object timber species. The same testing
method should be applied for the measurement of properties of the object timber
and the benchmark timber. Integrity index IBS of benchmark method is given from
Equation (1). Index IB is ratio of measured properties of the object timber and the
benchmark timber, which is defined in Equation (2). Pr and PB are measured
properties of the object timber and the benchmark timber. Index IS correspond to
118
the decrement of cross-sectional areas by inhomogeneous degradation is defined in
Equation (3), where Ar is residual cross-sectional areas of the object timbers, and
Ao is original/initial cross-sectional areas of them. Integrity of the object timber is
evaluated by the integrity index IBS. In case of screw withdrawal resistance
measurements, index properties are withdrawal resistances both of the object timber
and the benchmark timber for the same directions to the grain. Benchmark method
is available for non/semi-destructive testing methods, especially for the case that
nominal values of physical/mechanical properties are not listed in wood handbooks,
etc.
I BS = I B × I S
(1)
Pr
PB
A
IS = r
Ao
IB =
(2)
(3)
Nominal value method
In case benchmark timber is difficult to prepare, nominal value method will be
alternative of benchmark method. Reference for the comparison is nominal
properties of the object timber species. These nominal properties are generally the
average or the lower limit (5 percentile) values listed in wood handbooks, etc [8,9].
Integrity index INS of nominal value method is given from Equation (4). Index IN
is ratio of the properties estimated from the object timber and nominal properties
of them, which is defined in Equation (5). Pr is estimated properties of the object
timber. P N is nominal properties of the object timber species. Nominal
value method requires less measurement works than those by benchmark
method, however, need nominal properties of the object timber species. In case
of screw withdrawal measurements, densities or shear strengths parallel to the
grain are used for the index properties, because both of nominal values are
listed in wood handbooks. Although shear strengths perpendicular to the grain
is obtained from measured withdrawal resistances, the shear strengths parallel
to the grain is not directly obtained from withdrawal resistances. Estimation of
shear strength parallel to the grain is required in screw withdrawal measurements.
I NS = I N × I S
P
IN = r
PN
(4)
(5)
Structural calculation method
The other method is structural calculation. Structural calculations are applied for
119
object structures or part of them. Structures are required to resist actual and design
loads in principle. One of the methods to evaluate integrity level of existing
structures is to calculate structural safety of existing structures with their residual
properties against their actual and design loads. Structural calculations result stress
level of components, safety level of joints, deformation of components and
displacement of structures. These deformation and displacement are used for
serviceability analysis of the structures. Stress level of the components and joists
are used for risk analysis of failure of the structures. Structural calculations need
principal mechanical property of components and joints of the object structure.
Screw withdrawal measurements provide only densities and shear strengths of
timber. MOR and MOE of timber are estimated from screw withdrawal resistances
with velocities of stress waves. Compression of timber is also correlated with
screw withdrawal resistances. Expedient inspection methods should be used for
mechanical property estimation of structural components and joints. In order to
estimate required mechanical properties, screw withdrawal measurements would
be used with other inspection methods.
Mapping and judgement
Detailed inspections provide physical/mechanical properties of timber. These
properties are used for integrity indexes of benchmark and nominal value methods.
Integrity indexes, serviceability rank and risk of failure of the structures are mapped on
the drawings and included in the inspection reports. Judgment for the future of the
object structure is made such as use, usage restrictions, repairs, reinforcements and
demolish of all or part of the structure. Value of the structures and cost of repairs
and reinforcements of them are also considered for the judgements.
Methodology
Probes
Screw withdrawal measurements had used nails, wood-screws and lagscrews. Wood-screw based probes with short-threads were developed,
and probes manufactured from threaded rod with metric-screws were
developed[7,10]. Shape of wood-screws is not standardized internationally;
however, shape of metric-screws is standardized by ISO 261 and 724. Metricscrew probes are longer and have smaller diameters of their threads than typical
wood-screws; the probes are suitable for measuring withdrawal resistances in
deep of timber. The short-threads clarify measuring positions in deep timber,
and reduce withdrawal resistances. Coaxial multiple withdrawal resistance
measurements (CMWR) using long metric-screw type probe provide distributions
of withdrawals along pre-drilled holes. The metric-screw probes have outer
cylindrical shear planes around the threads. In case of timber, these shear planes
have shear strengths correspond to shear directions. Fig.3 and Fig.4 show the
120
probes for the screw withdrawal measurements which have short-threads and
are manufactured from threaded rods of ISO standard metric-screws.
Diameter, pitch and length of the probe thread in Fig.3 was 3.87mm,
0.7mm and 12.85mm respectively. The probe has double heads of conical
heads on top for withdrawing and hexagonal heads below for screwing.
Depth
Indicator
15
5
Thread Length
15
20
Thread Diameter
Fig 3. Short-thread of the Metric Probe
for Screwing and Withdrawing
Fig.4 Metric-screw Probe with Heads
Equipment
Tools for withdrawing and measuring
Two cramp types of withdrawing tools are shown. Photo1 shows typical withdrawal
tools developed by Prof. F.Divos [11]. Photo 2 shows withdrawal tool developed for
the coaxial multiple withdrawal resistant (CMWR) measurements [12]. Load-cells
are installed in these tools. Peak loads are indicated on the indicators connected to the
load cells. Rate of withdrawing should be constant as possible. Electric screwdrivers
are able to apply constant rate of withdrawing better than that by hands.
Photo 1 Typical Withdrawal Tool
Photo 2 Withdrawal Tool for CMWR
121
Procedures
Pre-drilled holes
Wood-screws do not need pre-drilled holes, but lag-screws and metric-screw type
probes need pre-drilled holes to screw them into timber. Pre-drilled holes are
applied by drilling tools. These pre-drilled holes are required to be orthogonal to
the surfaces of timber generally. Diameter of the pre-drilled holes used with
the probes shown in Fig.3 was 3mm.
Single Withdrawal Resistances
The probes are screwed into the pre-drilled holes by hands or electric screwdrivers,
etc. Single withdrawal resistance (SWR) measurements are used for typical withdrawal
measurements. The probes are screwed into timber as Photo 3. The probes are able
to be screwed into timbers through finishing such as gypsum boards and plaster, etc.
Fig.5 shows withdrawal resistance measurements of timber behind the finishing.
Multiple measurements of SWR are required to obtain reliable SWR values.
Photo 3 Single withdrawals Fig. 5 Measurement of Timber behind Finishing
Coaxial Multiple Withdrawal Resistances
SWR is a method to measure withdrawal resistance at one depth position. In order
to obtain distribution of withdrawal resistances along depth direction of timber,
coaxial multiple withdrawal resistance (CMWR) measurements were developed
[7,10,12]. Fig.6 illustrates procedures of CMWR using the same pre-drilled hole.
The probe in Fig.6 equips attachment adjusting depths of the probe thread in
timber. Typical procedures of CMWR are as follows. The probe is screwed into
timber 20mm deep into the wood. The probe is pulled out and withdrawalresistances are measured simultaneously by the withdrawal tools. Then the probe is
removed from the hole to get rid of sawdust from the hole. The probe is then
screwed in 15mm deeper than before. The probe is also pulled out and
withdrawal-resistance is measured simultaneously. The probe is removed from the
hole to get rid of sawdust from the hole. These procedures are repeated. Tip
positions of the probes in timber will be 20, 35, 50mm depths in this case. Multiple
measurements of CMWR are required to obtain reliable CMWR values.
122
1. Pre-drill
2. Screw Probe
into Pre-drilled
Hole
3. Pull out Probe
& Measure
4. Screw Probe
deeper than
Before
5. Pull out Probe
& Measure again
Fig. 6 Procedure of CMWR Measurement
Normalized Withdrawal Resistance (NWR)
Measured withdrawal resistances are affected by the area of outer cylindrical shear
plane around the probe threads. Removing the effect of dimensions of the probe
threads, measured withdrawal resistances were normalized by the outer cylindrical
area of the thread. These normalized withdrawal resistance (NWR) is obtained by
Equation (6)[12]. NWR indicates estimated shear strength of wood on the outer
cylindrical shear plane shown in Fig.7. When the probe is screwed into the timbers
from their longitudinal surfaces, direction of the estimated shear strength (NWR)
will be RT-direction (radius or tangential direction = perpendicular to the grain) of
timber.
(6)
τ
P
Rt
Lt
π
: Estimated shear strength (NWR)
: Withdrawal resistance
: Diameter of the thread (peak to peak)
: Length of the thread of probes
: Circular Constant
123
[N/mm2]
[N]
mm]
[mm]
Fig.7 Cylindrical Shear Plane around Probe Thread
Application
Density
Withdrawal strengths of wood-screws are correlated with densities of timber,
withdrawal resistances of wood-screws are calculated using equations with density
terms [8]. Fig.8 and Fig.9 are examples of relationship between measured densities
and withdrawal strengths. Fig.8 shows relationship between timber densities and
withdrawal strengths of typical wood-screws [13]. Fig.9 shows relationship between
densities and NWR of metric-screw probes [14]. The tests in Fig.9 used three
coniferous species. NWR(–RT) in Fig.9 means NWR for radius(R) or tangential(T)
directions of timber, those are for perpendicular to the grain directions. Both figures
show clear correlations between densities and withdrawals. A regression equation
between densities and withdrawal strengths is proposed in Fig.9. A regression
equation in Fig.9 is obtained using three coniferous species totally. Withdrawal
forces include the effect of thread dimensions; however, NWR exclude effects of
thread dimensions.
Fig. 8 Densities-Withdrawals Relationships by Wood-screws [13]
124
Fig.9 Densities-NWR Relationships by Metric-screws[14]
Shear strength
Perpendicular to grain
NWR is shear strengths on outer cylindrical planes around the threads of probes.
Withdrawal measurements are applied for perpendicular to grain in general; NWRs
obtained from these withdrawal measurements are for perpendicular to grain direction.
Measured NWRs by withdrawal measurements are shear strength perpendicular to
the grain, but it is difficult to measure shear strengths perpendicular to the grain by
typical shear strength tests using chair type block specimens.
Parallel to grain
Typical shear strengths of timber are those parallel to the grain. Nominal shear
strengths listed in typical wood handbooks are those for parallel to the grain
direction [8,9]. Nominal shear strengths are obtained from shear strength tests
based on testing standards(ASTM D143, JIS Z2101). Shear testing standards uses
chair type specimens which have flat shear planes, use offset distance between two
loading shear planes parallel to the grain. NWR is shear strengths on cylindrical
shear plan e s, have no offset, for perpendicular to the grain. Fig.10
shows relationship between densities of timber and measured shear
strengths(L) Sh L parallel to the grain [14]. A regression equation between
strengths(L) ShL and densities is proposed in Fig.10. A regression equation in
Fig.10 is obtained using three coniferous species totally. The specimens used for
the tests of Fig.9 and Fig. 10 were end-matched. Shear strengths(L) ShL are able
to be calculated from NWRs of withdrawal resistances using these two regression
125
equations shown in Fig.9 and Fig.10. Regression equations in Fig.9 and Fig.10
are examples respectively.
Fig. 10 Densities-Shear Strength( L) ShL Relationship [14]
CMWR
Depth (mm)
Fig.11 shows comparison of averaged NWRs distributed in depth direction measured
by SWR and CMWR measurements. The depth direction was perpendicular to the
grain. CMWR used a 105x105x105mm specimen of Douglas-fir with nine predrilled holes and six depth positions for CMWR measuring respectively. SWR
used twenty seven 30x30x30mm specimens with a pre-drilled hole and a depth
position respectively. These specimens for CMWR and SWR were cut off from the
same timber. Those were also end-matched and depth-matched. NWR distribution
100
90
80
70
60
50
40
30
20
10
0
CMWR
(105mm)
SWR
(30mm)
0
10
NWR (N/mm2)
20
Fig.11 Distribution of NWRs by CMWR and SWR Measurements
126
by CMWR measurements was just about close to that by SWR measurements;
distributions of NWRs for depth direction are measured by CMWR measurements
[12].
Compression
Relationship between compression and screw withdrawal/pull-out force was studied
[15]. Fig.12 shows one of these results. The withdrawal/pull-out force values are
correlated with compression strengths of timber.
Fig.12 Compression-Pull out Force Relationship[15]
M OR
Relationship between MOR and screw withdrawal resistances is known from
1990’s. F.Divos et al. proposed Equation (7) and Equation (8) which predict
MOR of coniferous and hardwood species respectively. Screw withdrawal forces
and velocities of stress wave are used in these equations. The applied units in
these equations are: MOR[MPa], F[kN], ν[km/s]. Screw withdrawal forces F were
measured using wood-screws. Diameter and length of their threads was 4mm and
18mm. 2.7mm diameter pre-drilled hole was made in timber to accommodate
screw. The results are shown in Fig.13. In sound wood, longitudinal transmission
velocities generally fall within the range of 3.5 - 5 km/s. Screw withdrawal forces
using wood-screws for transverse direction of timber generally fall with the range
of 1 – 3.5 kN. These withdrawal forces are considered to be equivalent to 4.5 –
15.5 N/mm2 of NWR. The correlation coefficient between measured and predicted
MORs by Equation (7) and Equation (8) was 0.74 [1,11,16].
2
MOR = 0.809 F × ν + 26.8
2
MOR = 1.258 F × ν + 36.9
(7)
(8)
127
Fig.13 Measured and Predicted MOR Relationship [11]
M OE
Relationship between static MOE and screw withdrawals of timber were
examined. Static MOE were correlated with dynamic MOE. Static MOE(E)
was predicted by Equation (9) with regression coefficients a and b, density and
stress wave speed v. The densities were estimated from measured screw withdrawals
with their regression coefficients. Predicted MOE (E) was correlated wiht the actual
static MOE [3, 13].
E = a + bρν 2
(9)
Reliability-based Evaluation
In case that the structural performance and design loads are probabilistic, structural
safety is evaluated stochastically. Reliability-based integrity analysis of existing
timber structures will be realized by structural calculations using residual mechanical
properties of timber obtained from detailed inspections such as screw withdrawals
and others.
Further Application/Calibration of Output from
Nondestructive Testing
Non-destructive testing method is expected to provide physical/mechanical properties
of materials, but information obtained from non-destructive testing is limited. For
example, applying screw withdrawal tests in conjunction with nondestructive testing
for the same location of the same timber, relationship between non-destructive testing
output and screw withdrawal resistances are obtained. Output from non-destructive
testing could be calibrated by the physical/mechanical properties estimated from
128
screw withdrawals. Calibrated output provided from non-destructive testing is more
informative and useful than those without calibrations.
Limitations
Screw withdrawal is one of cost effective inspection methods. But the followings
are limitations of screw withdrawals. Screw withdrawals leave holes on timber.
Repair of the holes may be required. Knots are often hidden in object timber.
Screw withdrawals provide very large withdrawal resistances when the probe is in
and around knots hidden in object timber. These withdrawals should be eliminated
preventing the false estimation of timber properties. Screw withdrawal measurements
are time consuming. CMWR measurements are more time consuming than SWR.
In order to obtain reliable results from screw withdrawal measurements, multiple
measurements of screw withdrawals are required. Screw withdrawal measurements
need skills for pre-drilling and withdrawing. Lengths of drills for the pre-drilling
and metric-screw probes are limited.
References
[1]
Bohumil Kasal, Thomas Tannert, In Situ Assessment of Structural Timber, Springer,
RILEM State of the Art Reports, Volume 7, 2010
[2]
Mariapaola Riggio, Ronald W.Anthony, Francesco Augelli, Bohumil Kasal, Thomas
Lechner, Wayne Muller, Thomas Tannert, In situ Assessment of Structural Timber using
Nondestructive Techniques, Materials and Structures, DOI 10.1617/s11527-013-0093-6,
2014
[3]
Thomas Tannert, Ronald W.Anthony, Bohumil Kasal, Michal Kloiber, Mau-rizio Piazza,
Mariapaola Riggio, Frank Rinn, Robert Widmann, Nobuyoshi Ya-maguchi, In situ Assessment
of Structural Timber using Semi-destructive Techniques, Materials and Structures, DOI
10.1617/s11527-013-0094-5, 2014
[4]
Kloiber, M., Drdácký, M., Tippner, J., Sebera, V.: New construction NDT device for in
situ evaluation of wood by using compression stress-deformation measurements parallel to
grain. In: 18th International Nondestructive Testing and Evaluation of Wood Symposium,
2013, September 24-27, Madison, Wisconsin, USA. FPL-GTR-226: pp. 585-592.
[5]
Talbot, J.W. unpublished research, Pullman, WA: Washington State University, 1982
[6]
Robert J. Ross, Roy F. Pellerin, Nondestructive Testing for Assessing Wood Members
in Structures, General Technical Report FPL-GTR-70, Forest Products Laboratory, US
Department of Agriculture (1994)
129
[7]
Nobuyoshi Yamaguchi, Shiro Nakajima, Hirofumi Sakuma, Inspection Method of Wood
Integrity Using Distribution of Resistances to Axial Withdrawals of Wood-Screw Probes,
Proceedings of 15th international symposium on nondestructive testing and evaluation of
wood, Duluth (2007), pp.233-241
[8]
Wood Handbook, 2010 Edition, Forest Products Society, Madison, WI, USA (2011)
[9]
Wood Industry Handbook, ISBN 4-621-07411-3 C3550, Forest and Forest Products Research
Institute, Maruzen, Japan (2004) in Japanese
[10] Nobuyoshi Yamaguchi, Withdrawal resistances by screw-based probes for in-situ assessment
of wood, Proceedings of international conference on structural health assessment of timber
structures (SHATIS’11), Lisbon (2011)
[11] Ferenc Divos, Laszlo Nemeth, Laszlo Bejo, Evaluation of the Wooden Structure of a Baroque
Place in Papa, Hungary, Proceedings of 11th international symposium on non-destructive
testing and evaluation of wood, Washington State University, USA (1998) , pp.153-160
[12] Nobuyoshi Yamaguchi, “In situ Assessment Method of Wood using Normal-ized Withdrawal
Resistances of Metric-screw Type Probes”, Advanced Materials Research Vol.778 (2013),
pp.217-224
[13] Cai, Z., Hunt, M.O. R.J. Ross, and L.A. Soltis, Screw Withdraw-al - A Means to Evaluate
Densities of In-situ Wood Members, Proceedings of the 13th International Symposium on
Non-destructive Testing of Wood, Forest Products Society, Madison, USA, pp.277-281,
2002
[14] Nobuyoshi Yamaguchi, Inspection Method of Integrity of Wood Components in Existing
Timber Construction (Part 4) In situ Evaluation using Screw Probe Withdrawals, (in
Japanese), Summaries of Technical Papers of Annual Meeting, Structure III, Architectural
Institute of Japan, 2014, in Japanese, Print Pending
[15] Gilfillan, J.R., Christie, D. and S.G. Gilbert, The Residual Strength of Timber Degraded by
Woodworm Infestation, Proceedings of Durability of Building Materials and Components
8(Volume One), NRC Research Press, Ottawa, Canada, pp.714-722, 1999
[16] Ferenc Divos, Evaluation of Single Members in Historical Structures, RILEM meeting,
Prague, May 30. 2005
130
COMBINED METHOD FOR THE IN SITU
MECHANICAL IDENTIFICATION OF
ANCIENT TIMBER BASED ON NDTS
Abstract
The research context. A wide research activity based on experimental tests and
statistical analysis, is ongoing at the Department of Structures for Engineering and
Architecture of the University of Naples Federico II, by the research team leaded
by B. Faggiano, composed by M.R. Grippa, A. Marzo, L.Esposito. The research
aims at providing a methodology for in situ mechanical identification of ancient
timber members by non-destructive techniques [1, 2]. In particular the main goals
are:
database
collection
through
laboratory
NDT
and
DT investigations; mechanical characterization of old chestnut defect-free
and structural elements through compression and bending tests;
elaboration of experimental results for evaluating the influence of typical
defect patterns on timber performances; identification of reliable NDT-DT
correlations to be used in situ for the estimation of wood density,
strength and stiffness properties; definition of standardized guidelines for
practical applications. The starting point is the research carried out within the
national project PRIN2006 “Consolidation of timber structures” (Prof.M.
Piazza coordinator, B. Faggiano responsible of research unit) and the international
project PROHITECH “Earthquake Protection of Historical Buildings by
Reversible Mixed Technology” (2004-2008, prof. F.M. Mazzolani
coordinator). The activity is actually developed in the framework of the
Italian projects DPC-RELUIS 2010 – 2013 and 2014 (B. Faggiano
responsible of research unit). The study is articulated in three main phases
correlated each other, such as experimental investigations, data
processing, structural identification by combined NDT-DT relationships.
Supplying of the test material
The experimental campaign is developed on timber elements made of old chestnut
wood (Castanea sativa Mill.), provided from roofing trusses of masonry buildings
131
in Naples, dated beginning of 19th century. Tests are performed on both
structuralelements and defect free specimens with standard dimensions, according
to UNI EN 480 (2004) and UNI ISO 3789 and 3132 Italian codes (1985), in order
to evaluate the influence of defect patterns on both non-destructive quantities and
mechanical properties measured through destructive tests [3, 4, 5]. Specimens have
the following features (Fig. 1):
Fig. 1. Specimens tested in the experimental activity.
Structural elements in actual dimensions (SA): from trusses king posts, 14
elements (SA-C) for compression tests parallel to grain, D 14.5-16 cm
mean equivalent diameter, 6D length; from the trusses struts, 10 elements
(SA-B) for bending tests, D 15-16.5 cm mean diameter, 19D length;
Squared elements in small dimensions (SS): from 11 S A-C specimens,
after destructive tests: 20 samples 5x5x30 cm for ND and D tests in
compression parallel to grain (SS-C); 16 elements 5x5x15 cm for ND
tests (SS-NDT); from 6 SA-B specimens, after destructive tests: 24
elements (undamaged parts, 4 from each specimen) 4x4x76 cm for ND
and D tests in bending (SS-B);
Defect-free specimens (DF): from S A-C samples, 2x2x4 cm: 33
specimens for longitudinal tests (DF-CL); 22+22 specimens for both
radial (DF-CR) and tangential (DF-CT) tests; from SA-B samples,
2x2x40 cm: 35 specimens for bending tests (DF-B).
Table 1 specifies for each group of specimens the ND and D tests performed.
Tab 1. Non-Destructive (H: hygrometric; U: ultrasonic; S: sclerometric; R: resistographic) and
Destructive (C: compression; B: bending) Tests.
132
Visual inspection
Preliminary, the conservation state of the selected elements is examined by checking
timber features and defects, signs of damage and deterioration. In particular, on
the lateral faces of SA-C specimens macroscopic longitudinal cracks due to
shrinkage are detected; whereas, ring shakes, large isolated knots or knots groups
are surveyed on the SA-B specimens. Therefore, as a result of the visual structural
grading, according to UNI 11119 (2004) standard, all specimens are assigned to
the third class.
Non-destructive tests (NDT).
The following ND techniques are employed [3]:
Hygrometric tests, for the evaluation of the moisture content of wood;
Ultrasonic (U) tests, for the determination of the elastic properties of
wood;
Sclerometric (S) tests, for the assessment of the quality and surface
hardness;
Resistographic (R) tests, for the detection of density variations and internal
defects of wood.
For each specimen, the average values of NDT measures, such as ultrasonic
stress wave speed (SWS), sclerometric penetration depth (PD), resistographic
mean amplitude (Am), are calculated.
Since in situ surveys allow only perpendicular to grain measures on the
accessible external surfaces of timber members, laboratory tests are performed in
both longitudinal (L) and transversal (T) directions as respect to grain orientation
for evaluating the relations between L and T ND variables.
Destructive tests (DT)
Compression (C) and bending (B) tests were carried out for determining the
mechanical behaviour of timber elements in terms of stiffness, load bearing
capacity and collapse mechanisms [4].
Statistical analysis of laboratory data
The experimental results are critically examined and compared, with aim to obtain
statistical ND and D parameters, such as average values, standard deviations and
coefficients of variation.
133
Determination of NDT-DT correlations
By means of a linear regression approach the following relations are examined
[1, 7]:
Correlations between NDT parameters, relating the L and T measures by
U, S and R tests;
Correlations between DT parameters, defining the mechanical behaviour
in C and B;
NDT-DT correlations, using both simple and multiple models for ND
estimation of density, modulus of elasticity and strength of the material.
In order to identify the best combination of NDT parameters for the prediction
of wood density, simple and multiple linear regression analyses have been carried
out. The goodness of regression fit, such as the fitting of the linear model to a given
body of data, is formally assessed by the coefficient of determination (R2).
Note that 0 ≤ R2 ≤ 1, high values indicate a strong linear relationship between the
variables involved in the model.
As an example simple and multiple correlations are presented in Fig. 2.
Fig. 2. Correlations: a) Density vs compression strength; b) Density vs sclerometric+resistographic parameters
Combined method for timber mechanical characterization
Based on the results achieved, the method consists in the determination of the density
of timber through the combination of in situ sclerometric and resistographic parameters
in transverse direction. Once estimated the density, the strength and modulus of
elasticity both in compression and in bending are evaluated through the correlation
with density. Indication for executions are also provided.
134
Further development
The extension of the database would allow to reach more and more robust correlations
for in situ mechanical characterization of existing timber members.
Acknowledgments
The research activity is developed within the projects PRIN 2006, PROHITECH 2004-2009,
DPC-RELUIS 2010-2013 and 2014.
References
[1]
Faggiano B., Grippa M.R., Marzo A., Mazzolani F.M., 2009. “Experimental evaluation of
the mechanical properties of wood by means of non-destructive compared techniques for
the characterization of existing wooden structures”. In Consolidation of timber structures,
ed. M. Piazza, Hevelius Publisher, pp. 25-78, Italy, 2009 (in Italian).
[2]
Faggiano B., Grippa M.R., Marzo A., Mazzolani F.M., 2011. “Experimental study for
nondestructive mechanical evaluation of ancient chestnut timber”. Journal of Civil
Structural Health Monitoring: Volume 1, Issue 3, page 103-112.
[3]
Faggiano B., Grippa M.R., Marzo A., Mazzolani F.M., 2010. “Experimental analysis on
old chestnut timber by means of non-destructive techniques”. 11th World Conference on
Timber Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June.
[4]
Grippa M.R., Faggiano B., Marzo A., Mazzolani F.M., 2010. “Combined methods for in
situ mechanical identification of ancient timber structures based on non-destructibe tests”.
11th World Conference on Timber Engineering (WCTE2010), Riva del Garda, Italy, 20-24
June.
[5]
B. Faggiano, M.R. Grippa, A. Marzo, F.M. Mazzolani, 2010. Structural grading of old
chestnut elements by compression and bending tests. 11th World Conference on Timber
Engineering (WCTE2010), Riva del Garda, Italy, 20-24 June 2010.
[6]
B. Faggiano, M.R. Grippa, 2012. Mechanical characterization of old chestnut clear wood
by non-destructive and destructive tests. In: World Conference on Timber Engineering.
vol. III, p.359-364, Auckland (New Zealand), 16-19 July 2012.
[7]
B. Faggiano, M.R. Grippa, B. Calderoni, 2013. Non-destructive tests and bending tests
on chestnut structural timber. Advanced Materials Research Vol. 778 (2013) pp
167-174© (2013) Trans Tech Publications, Switzerland, doi:10.4028/www.scientific.net/
AMR.778.167
135
136
Combine information from visual and
NDT/SDT methods
Artur Feio1, José Saporiti Machado2
1)
2)
Department of Architecture, University Lusíada, Portugal, [email protected]
Department of Structures, LNEC, Portugal, [email protected]
Abstract
Visual strength grading (VSG) standards are generally the most common nondestructive method used for assessing the mechanical properties of timber member
in situ. The limitations of the full application of visual grading rules to timber
members in situ lead to the necessity to provide a set of simplified rules for this
task. However given the subjectivity and the limitations related to the application
of visual grading the results can be supported/validated by complementary
information provided by non and semi-destructive testing (NDT/SDT) methods.
The present paper discusses how VSG information can be optimized facing a
particular timber member and how its information can be combined with other
type of NDT/SDT information.
Background
Structural evaluation of ancient or recent timber structures presents particular
problems (related to inherent wood material properties) and difficulties. In situ
evaluation (without damaging) of timber structural elements represents an initial
and crucial step for the success of the rehabilitation process. Support for
nondestructive inspection works includes nowadays a variety of tools offering
valuable information about the quality and biodeterioration status of timber elements.
Appraisal and repair of ancient timber structures has become a major topic of
interest in the last decades. This renewed interest considerably increased the number
of technical interventions and design developments. Conservation or rehabilitation
of existing timber structures imply extensive knowledge about the properties of
materials from which the structure is made. This knowledge constitutes the
support for short-term structural behaviour assessment as well as to foresee the
continuous adaptation and capacity of response of the material under long-term
actions.
137
Due to the high variations intra and inter species, a large volume of wooden
material is needed to be tested to characterize its mechanical properties with a
minimum level of confidence [1]. Quality control and preservation of artistic value
were considered important issues leading to the development of some nondestructive and semi-destructive test methods (NDT/SDT) for wood, which sometimes were used in the evaluation of the mechanical and physical properties of
other materials [2]. Combined with a visual grading survey, these evaluation
methods are an excellent complement to achieve a good level of reliability in the
structural analysis, diagnosis and inspection of existing constructions.
Visual Strength Grading
The evaluation of a timber structure begins by conducting a preliminary survey
that includes an appraisal of the general quality of the timber members. This
evaluation consist in examining directly, and preferably at close distance, checking
and registering wood features, signs of damage or deterioration, sometimes with
the help of simple instruments (knife, chisel, hammer, etc.), providing a rapid
means of identifying areas that may need further investigation. This is an essential
part of diagnosis but the results depend severely on the experience of the person
carrying out the task.
Since modern design standards require quantitative more than qualitative
parameters, visual inspections, as the basis of any visual strength grading (VSG)
appraisal, are crucial in the inference of mechanical properties. Visual grading
is done mainly based on national standards trying to optimize the grading results
for the timber resources of the country, taking into account growth conditions,
local preferences for certain cross sections, silviculture differences, and historical
developments concerning structural applications [3]. All the VSG standards
are based on indirect visual methods focus in understand the effect of important
macroscopic strength-reduction reducing characteristics, such as knots and slope
of grain, upon the basic quality of a timber member, defined by the properties of
its clear wood zones. Therefore mechanical properties of wood elements are
highly dependent on the inherent wood properties of each species as well as on
the presence and type of defects [4].
Appraisal in situ includes the identification of the wood species, selection of a
proper VSG standard and consequent allocation of suitable reference mechanical
properties in terms of allowable stresses, characteristic values or strength class.
VSG was the first non-destructive testing method to be applied for the sorting of
sawn timber according with their mechanical characteristics. The simplicity of its
application, not requiring any special equipment, made this method also the one
first adopted for the in situ assessment of structural timber members. However in
this preliminary survey other NST/SDT methods are already used as support to
VSG results (e.g. moisture meters).
138
Such properties are then used for a first structural analysis. If results showed
that structural failed complied with the ultimate state requirements then a more
detailed evaluation – detail survey – of the timber members should be carried out or
the structure demolished [5]. The focus of this second evaluation is on critical
members of the structure (more stressed or involved in failure has shown by the
first structural analysis results).
Recent publications address the way VSG should be adapted to on site conditions
[5, 6, 7, 8]. These publications try to optimize the grading process (optimal visual
grade) by considering the most important defects present in a timber member and
the type of stress applied. The use of NDT/SDT methods to ensure a more reliable
prediction of timber member’s properties is vaguely mentioned. For instance,
[7] establishes criteria for the diagnosis of old timber elements and strength
grading can be performed using both on site inspections and NDT techniques.
Prepared in the scope of COST Action IE0601, [5] proposed a first harmonized
method for determining the structural grade based on visual grading appraisal
supported in the current European procedures and standards. Although NDT for
strength prediction is presented as an option in some cases, the level of reliability
was thought to provide always an over conservative prediction (safe side).
This document is also the reference document for the WG 10 of the CEN/TC
346/WG 10 activity (Conservation of cultural heritage - Historic Timber
Structures) aiming the final preparation of a harmonised visual strength grading
methodology that could be approved by CEN.
However, there are several restrictions to the full application of visual grading
standards to timber members in situ. VSG standards were developed having in
mind the grading of sawn timber at sawmill yards. The full application on site of
the rules applied at the sawmill yard is not possible or logical and leads to gross
underestimation of the real mechanical performance of timber members. Considering
the biased and subjectivity of the grading process, dependent upon human
judgment, and variability associated to each visual grade, it is expected that a large
proportion of the pieces real resistance would be largely underestimated [9, 10].
However the estimation of the serviceability properties of timber constructions, by
means of the VSG standards, is not entirely reliable due to the many factors:
• timber is a highly anisotropic material – heterogenetic material;
• specific to a particular wood species and provenance, the possibility of
extrapolation to other wood species and origins is not certain – correlation
inaccuracy;
• susceptible to human error, given the limitation of its full application in situ –
measurement error.
VSG principles can be differentiated between those following the concept of
characteristic values [11, 12] and those following the concept of allowable working
stress [7, 13, 15].
Since VSG is a group classification based on a heuristic approach with limited
139
support from models (e.g. statistical), limited options remains for a possible upgrade of a timber member. In fact, in the absence of the data used for defining the
grades of a certain VSG standard, the only option is to shift a timber member from
one grade to another (lower or upper), not being possible to derive intermediate
grades.
For making these changes significant, the data used for the derivation of the
visual strength grades or at least a probabilistic model adjustable to the timber
species under analyse, needs to be available [16].
When no grading standard is available to a certain species, it is selected one
appropriate to a species showing similar wood anatomy and wood density. For
example, while [12] establishes different grading criteria for different timber groups,
[7] applies the same grading criteria to all species.
In-situ diagnosis of ancient timber structures has been described by several au
thors [5, 17, 18, 19, 20]. All the authors state that an initial visual inspection of the
entire structure and of the singular elements is required in order to determine the
original timber characteristics and the changes suffered due to service conditions.
This survey follow several steps, beginning with the purpose of a general prediction
of mechanical properties and ending in a thorough examination using
NDT/SDT. But an important characteristic of several ancient timber structures is
that they can effectively bear higher loads than expected [21], which stresses the
need of adequate procedures for diagnostic and assessment of the real bearing
capacity, which cannot be obtained with a simple visual inspection.
Combine visual and NDT/SDT methods
Recent guidelines propose the upgrading of preliminary VSG application by
conditioning it upon the stress condition and position of defects in relation to applied
stress. As referred, and based on this, recent standards and studies establish that
strength grading can and should be performed using both on site inspections and
NDT techniques [1, 5, 22, 23].
Thus, NDT/SDT can be used to increase the precision of the prediction of the
mechanical properties provided by visual grading. Generally these methods are
used as validation of visual information and not combined in the sense of a fusion
process between the data obtained from visual observation and other type of
information (NDT/SDT methods). Usual applications of these methods are related
with the prediction of:
• the element residual section by analyzing abnormal density variations in the
element, generally associated with mass loss;
• density;
• important mechanical properties (bending, shear, tension and compression
140
strength, local and global stiffness) by measuring one or more indicative
parameters that can be correlated with it.
There are several NDT/SDT that can be applied to wood and wood composites
namely: thermography [24, 25], sonic stress waves [26], X-Ray [27, 28], isotope
method [29, 30], hardness tests [31], drilling resistance [20, 32, 33], moisture
content meter and core drilling [34, 35, 36].
Among these, core drilling, the moisture content meter and drilling resistance
currently play key roles in the preliminary survey. The large majority of research
and in-situ application relies on the use of NDT/SDT results to support or confirm
visual grading measurements. Drilling resistance is the most often used method to
get information about the presence of hidden features and their extent (e.g. knots,
decay and fissures). Other NDT/SDT methods (e.g. dynamic modulus of elasticity
or penetration resistance) are used as independent variables in regression models.
The upgrading of VSG information (reference properties) using other types of
information (NDT/SDT methods) is limited to a few studies.
The development of these methods is in fast progress; however, owing to safety
concerns, high costs involved, technical issues, etc., their use has been quite
limited in structural timber evaluation. One of the main constrains continues to be
the lack of generalized procedures or standards concerning the application
of NDT/SDT methods, although recent publication attempt to harmonize the
way these methods are used [37, 38, 39]. Although some publication shows
the advantages of combining (joining) information from VSG with those
from NDT/SDT methods [40] very few present methods that can be used for
a true combination of information [41, 42].
Conclusions
VSG will remains as the basic method for assessing the mechanical performance
of timber members in situ. However several studies agrees on the fact that the sole
use of these method can lead to the demolishing of safe structures and that other
NDT/SDT methods should be used to ensure a proper assessment procedure. The
question about how to combine this information is still under study but will probably
depend upon many variables (e.g. type of structure, load stresses involved, type
of evaluation) making that the expert will have to make the final decision upon
which method to use and how to use all the methods and data for taking a final
decision.
141
References
[1]
Feio A O, Lourenço P, J S. (2007) Non-Destructive Evaluation of the Mechanical Behavior
of Chestnut Wood in Tension and Compression Parallel to Grain. International Journal of
Architectural Heritage 1:272 – 292.
[2]
Riggio M, Tomasi R, Piazza M (2014) Refurbishment of a traditional timber floor with a
reversible technique: the importance of the investigation campaign for the design and the
control of the intervention. International Journal of Architectural Heritage 8:74-93.
[3]
Stapel P, Kuilen J (2014) Efficiency of visual strength grading of timber with respect
to origin, species, cross section, and grading rules: a critical evaluation of the common
standards. Holzforschung 68:203–216.
[4]
Sousa H, Branco J, Lourenço P (2014) Prediction of global bending stiffness of timber
beams by local sampling data and visual inspection. Eur. J. Wood Prod. 72:453–461.
[5]
Cruz H, D, Tsakanika E, Macchioni N, Jorissen A, Touza M, Mannucci M, Lourenço P
B (2013) Guidelines for the on-site assessment of historic timber structures. International
Journal of Architectural Heritage DOI: 0.1080/15583058.2013.774070.
[6]
Anthony R W, Dugan K D, Anthony D J (2009) A Grading Protocol for Structural
Lumber and Timber in Historic Structures. Apt Bulletin: Journal of Preservation Technology
40:3-9.
[7]
UNI 11119 (2004) Cultural Heritage. Wooden artefacts. Load-bearing structures – On site
inspections for the diagnosis of timber members. Italian Organization for Standardization
(UNI).
[8]
UNI 11035-1 (2003) Structural timber — Visual strength grading: terminology and
measurement of features. Italian Organization for Standardization (UNI).
[9]
Bonamini G (1995) Inspection and evaluation. D3/1-9. Timber Engineering STEP 2
– Design and structural systems. Centrum Hout. The Netherlands.
[10] Yeomans D (1999) The problems of assessing historic timber strengths using modern building
codes. The use of and Needs for Preservation Standards in Architectural Conservation,
ASTM STP 1355, L. B. Sickels-Taves Ed. American Society for Testing and Materials.
[11] EN 1912 (2012) Structural Timber – Strength Classes – Assignment of Visual Grades
and Species. CEN European Committee for Standardization, Brussels.
[12] UNI 11035-2 (2003) Visual strength grading rules and characteristic values for Italian
Structural Timber Population. Italian Organization for Standardization (UNI).
[13] ASTM D245 (2012) American Society for Testing and Materials, Annual Book of Standards,
Standard Practice for Establishing Structural Grades and Related Allowable Properties
for Visually Graded Lumber. ASTM, West Conshohocken, PA.
[14] ASTM D1990 (2012) American Society for Testing and Materials, Annual Book of Standards,
Standard Practice for Establishing Allowable Properties for Visually-Graded Dimension
Lumber from In-grade Lumber. ASTM, West Conshohocken, PA.
[15] BS 4978 (2007) British Standards Institution. Specification for the Visual Strength
Grading of Softwoods. BSI, London.
[16] Faber M H, Köhler J, Sorensen J D (2004) Probabilistic Modeling of Graded Timber
Material Properties. Journal of Structural Safety 26:295-309.
[17] Uzielli L (1992) Valutazione della capacità portante degli elementi strutturali lignei.
L’Edilizia 12:753-762.
142
[18] Ross R, Brashaw B, Pellerin R (1998) Non-destructive evaluation of wood. Forest Products
Journal 48:101-105.
[19] Tampone G, Mannucci S, Macchioni N (2002) Strutture di Legno. De Lettera editore,
Mila-no.
[20] Feio A O (2006) Inspection and diagnosis of historical timber structures: NDT correlations
and structural behaviour, PhD Thesis, University of Minho, Portugal. Available
from www.civil.uminho.pt/masonry.
[21] Togni M (1995) Elasticità e resistenza di travi lignee antiche di grande sezione: stima con
metodologie non distruttive applicabili in opera. Tesi presentata per il conseguimento del
tito-lo di dottore di ricerca in Scienze del Legno. Università degli Studi di Firenze, Firenze,
Italia.
[22] Piazza M, Riggio M (2008) Visual strength-grading and NDT of timber in traditional
structures. Journal of Building Appraisal 3:267–296.
[23] Feio A O, Lourenço P, J S. (2014) Testing, NDT and modeling of a traditional timber
mortise and tenon joint. Materials and Structures 47:213-225.
[24] Tanaka T (2000) Wood inspection by thermography. NDT.net – March 2001 Vol. 6 no. 03.
[25] Berglind H, Dillenz A (2003) Detection of glue deficiency in laminated wood with
pulse thermography. Journal of Wood and Science 49:216–220.
[26] Ross R, Pellerin R, Volny N, Salsig W, Falk R (1999) Inspection of timber bridges using
stress wave timing non-destructive evaluation tools – A guide for use and interpretation.
Gen. Tech. Rep. FPL-GTR-114. Madison, WI: U.S. Department of Agriculture, Forest
Service, Forest Products Laboratory: 15 pp.
[27]
Bucur V, Garros S, Navarette A, De Troya M, Guyonnet (1997) Kinetics of wood
degradation by fungi with x-ray microdensitometric technique. Wood Science and
Technology 31:383-389.
[28]
Bergsten U, Lindeberg J, Rindby A, Evans R (2001) Batch measurements of wood
density on intact or prepared drill cores using x-ray microdensiometry. Wood Science and
Technology 35:435-452.
[29] Madsen B (1994) Radiological density scanning portable gamma camera based on backscatter
tomography. In Proceedings of the 9th Non-destructive Testing of Wood Symposium.
Washington State University: pp. 131-139.
[30] Feinberg J (2005) Frontier Architecture Found in the Wild Wild West – Conservation
of U.S. Western Heritage Wooden Structures, 3 Studies. In Proceedings of the
International Conference The Conservation of Historic Wooden Structures, Florence, Vol.
I: pp. 383-390.
[31] Turrini G, Piazza M (1983) Il recupero dei solai in legno. Esperienze e realizzazioni, In
Re-cuperare, Vol. 7.
[32]
Acuna L, Basterra L A, Casado M, Lopez G, Ramon-Cueto G, Relea E, Martinez C,
Gonza-lez A (2011) Application of resistograph to obtain the density and to differentiate
wood species. Materiales de Construccion 61:451-464.
[33] Rinn F (2012) Basics of micro-resistance drilling for timber inspection.
Holztechnologie 53:24 – 29.
[34] Kasal B, Drdácký M, Jirovský I (2003) Semi-destructive methods for evaluation of timber
structures. In Proceedings STREMAH VIII, Series: Advances in Architecture, Vol. 15:
pp. 835-842.
143
[35] Kasal B (2005) Estimate of the Design Values of the In-Situ Wood Structural Members
Based on Semi-Destructive Experiments. In Proceedings of the International Conference
The Conservation of Historic Wooden Structures, Florence, Vol. I: pp. 304-306.
[36] Romagnoli M, Nocetti M, Sarlatto M (2005) Datazione Dendrocronologica di Strutture
Lignee nei Tetti in Italia Centro-Meridionale. Limiti e Prospettive. Proceedings of the
International Conference The Conservation of Historic Wooden Structures, Florence,
Vol. I: pp. 19-23.
[37] Riggio M, Anthony R W, Augelli F, Kasal B, Lechner T, Muller W, Tannert T (2014)
In situ assessment of structural timber using non-destructive techniques. Materials and
Structures, 47:749-766.
[38] Tannert T, Anthony R W, Kasal B, Kloiber M, Piazza M, Riggio M, Rinn F, Widmann
R, Yamaguchi N (2014) In situ assessment of structural timber using semi-destructive
techniques. Materials and Structures 47:767-785.
[39] Dackermann U, Crews K, Kasal B, Li Jianchun, Riggio M, Rinn F, Tannert T (2014) In
situ assessment of structural timber using stress-wave measurements. Materials and
Structures 47:787-803.
[40]
Cavalli A, Togni M (2013) How to improve the on-site MOE assessment of old
timber beams combining NDT and visual strength grading. Nondestructive Testing and
Evaluation 28:252-262.
[41] Machado J S, Lourenço P B, Palma P (2011) Assessment of the structural properties of timber
members in situ – a probabilistic approach. In SHATIS’11 International Conference on
Structural Health Assessment of Timber Structures, Lisboa, 2011.
[42] Sousa H (2013) Methodology for safety evaluation of existing timber elements. PhD Thesis,
University of Minho, Portugal. Available from www.civil.uminho.pt/masonry
144
Assessment of timber floors by means of nondestructive testing methods
Tiago Ilharco1, Thomas Lechner2 & Tomasz Nowak 3
1 NCREP- Consultancy on Rehabilitation of Built Hertage, Ltd., Science and Technology
Park of
University of Porto (UPTEC), Praça Coronel Pacheco no. 2, 4050-453 Porto, Portugal.
2 NCC Construction Sverige AB, Division of NCC Technology & Sustainable
Development, SE-405 14 Gothenburg, Sweden; Chalmers University of Technology,
Dept. of Civil and Environmental Engineering, SE-412 96 Gothenburg, Sweden
3 Faculty of Civil Engineering, Wroclaw University of Technology, Wybrzeze Wyspianskiego
27, Wroclaw 50-370, Poland
Abstract
In the process of rehabilitation of built heritage, the preservation of timber floors is
an essential issue. These structures have characteristics that are not entirely known,
namely the connections between elements, the load distribution between beams,
the importance of secondary elements, such as struts and floorboard, for the
attenuation of vibrations and reduction of deformations of the floor, etc. If properly
analysed and considered, these aspects can contribute to upcoming well-succeeded
interventions, improving the global behaviour of the floors and, consequently, of
the buildings. One of the focuses of the present paper is the assessment of the
global behaviour of timber floors by means of dynamic analysis, which is one of
the Non Destructive Tests (NDT) used to evaluate the reference properties of the
wood. In particular, this technique allows estimating the timber floors’ stiffness
and, consequently, assessing their efficiency and integrity. Furthermore, the paper
focuses on the use of other NDT, namely involving stress-wave timing, X-ray and
resistance drilling, which can provide very useful information about these
characteristics. The information obtained with the combined NDT allows a better
understanding of the timber floors behaviour and the implementatio of more
efficient rehabilitation and (or) strengthening techniques.
Assessment and assessment strategy of timber floors
Historical structures represent a part of the cultural heritage of every nation and
societies pay considerable attention to their preservation and maintenance. The insitu assessment of timber elements and their properties is essential in the
continuous maintenance and preservation of historical timber structures. Much of
the damage observed in historical timber structures can be attributed to
145
biodegradation. The deterioration of structural members results in changes in
geometry and load-bearing capacity. The replacement of members that have
deteriorated may not be an acceptable option for structures of historical
significance and redesign may be necessary to sustain the functionality of the
structure. The structural strength assessment of timber structures, which uses
various procedures and evaluation tools, is based on a multidisciplinary approach
aimed at providing information about the mechanical properties and actual
condition of timber members and the mechanical behaviour of joints. Abnormal
structural behaviour can be suspected when the strength and stiffness of a structure
is diminished due to deterioration, creep and the natural ageing of old timber [1, 2]
which implies changes in load-bearing capacity. Strategies for the analysis of
structures of significant cultural value must therefore be established.
A structural investigation procedure should be based on adapting a general
assessment methodology [3-6] to evaluate the structural condition and the mechanical
performance of the floor structures in an efficient manner.
The methodology comprises the following steps:
1) Diagnosis of the structure from previous repair work and action during
service life
2) Preliminary assessment and visual inspection
3) Detailed assessment and investigation including material testing with
nondestructive and quasi-non-destructive testing methods at critical sections
4) Evaluation of the results of the material tests
5) Structural analysis and evaluation of results
This methodology shall include global and local NDT, namely with seismographs,
resistance drilling machines, pilodyn, stress-wave timing, X-rays, etc. The thorough
interpretation of the tests’ results and the estimation of timber floors’ properties
can only be achieved with an analysis of the state of conservation, along with a
constructive/structural characterization. The characterization of the wood species
and of its density is also an important step to build a preliminary model of the
mechanical behaviour of the floor.
Global assessment through dynamic response
The discussion about the global assessment of timber floors through dynamic response
will be systematized in four main topics: 1) the dynamic behaviour of timber floors;
2) the techniques and instruments used to assess this behaviour; 3) the prediction
of the wood reference properties; 4) the identification of the damaged areas based
on the dynamic analysis. These topics will be analysed through an overview of this
subject and making use of results from several NDT performed in timber floors of
old buildings in Portugal included in structural survey campaigns.
146
The dynamic behaviour of timber floors
In residential buildings, the design of timber floors taking into account the vibration
limit state has in consideration the excitation caused by the movement of people,
which produces vibration frequencies of about 2Hz and 3,5Hz for walking and
running steps, respectively. The dynamic response of a floor is determined by
several factors, such as its mass, stiffness, damping and geometrical and structural
characteristics, namely the existence of struts, the thickness of the floorboard, the
type of connection between beams and walls, etc. In most cases, the floor stiffness
ensures a satisfactory dynamic behaviour. However, the traditional deflection criterion
does not always guarantee satisfactory vibration behaviour [7].
The issue of vibration induced by people walking on timber floors is more
complex than the static behaviour due to the resonance phenomena. Resonance
occurs when the frequency of the impacts that forces the vibration coincides with
the natural frequency of the floor, resulting in an increase in the magnitude of
vibration, leading to an eventual structural failure [8]. In an occupied building, with
high permanent loads, the increased mass may decrease the floor natural
frequencies to "dangerous" levels, since timber floors themselves have low mass
(50-100kg/m2). Therefore, it is fundamental that the timber floors’ design respect
the vibration limit states to fulfil comfort and safety requirements. [9] concluded
that two criteria for lightweight floors with fundamental frequencies above 8 Hz
should be considered: one related to the deformation due to a concentrated load and
other to the speed of the vertical vibration. These criteria were adopted in [10] in
the design of timber floors to the vibration limit state, stating that the vibration
levels should be estimated by tests or calculations, taking into account the
parameters that determine floors’ dynamic behaviour, namely mass, stiffness and
damping coefficient. The knowledge of all these characteristics allows the
assessment of the natural vibration frequencies and vibration modes associated to
each frequency, i.e. of the response of timber floors when subjected to known
dynamic actions.
Description of the method and instruments
Dynamic tests
Dynamic tests using ambient vibration are one of the most effective nondestructive
in situ testing techniques to identify the mechanical characteristics of structures.
The existence of highly sensitive sensors allows testing without imposing a forced
excitement on the structure and considering only environmental dynamic actions,
such as wind, traffic, movement of persons etc. [11]. Still, some authors consider
that, in the case of timber floors, the forced vibration allows a stronger response
and may provide more consistent results [12].
147
Instruments used and precautions to have during data acquisition
For measurements of the ambient vibration of timber floors, seismographs that
include tri-axial accelerometers (GeoSIG GSR-18bit), with an acquisition frequency
of 250Hz, can be used, resulting in temporal registries of the accelerations to which
the structure is subjected, Fig. 1. The seismographs allow the transference of the
information to a computer to be analysed. Nowadays, there are simple electronic
devices, such as smart phones, which are equipped with accelerometers and provide
reliable results, Fig. 2.
Fig. 1. Frequency measurement on a single
timber beam.
Fig. 2. Calibration of smart phones accelerometers with seismographs.
The registration of the dynamic response of a structure is a fundamental phase
of the tests. If the acquisition is carried out with errors, it will be very difficult to
correct them during the post-processing phase. It is, therefore, essential to perform
a careful planning of the tests, defining the equipment to use, its location and the
duration of the test. The positioning of the devices should be chosen so as to avoid
areas of zero modal displacements and the data acquisition, in particular using ambient
vibration, should be made by recording the dynamic response of the structure over
a pre-defined time interval. Some other specific precautions should be considered
during the tests, such as not disturbing the floor with the introduction of additional
masses, such as those given by the test operators, and taking into account the
presence of superimposed dead loads and its position.
Data processing and results achievement
After processing the collected data, the dynamic identification is done through the
determination of the natural frequencies and the corresponding modes of vibration,
which can consist on vertical, horizontal or coupled modes, depending on the main
direction of vibration. One of the used methods is the Advanced Method of Frequency
Domain Decomposition [13], currently implemented in the software ARTeMIS
[14]. The fundamental frequencies of vibration of the floors are identified using the
peaks observed in the records in the frequency domain (obtained via Fast Fourier
Transforms, FFT). Fig. 3 show an example of the data obtained in the frequency
148
domain, identifying the direction of higher vibrations associated. The value of the
1st frequency is 9.1Hz (z) and the 2nd is 10.0Hz (y).
Fig. 3. Identification of the main frequencies of a timber floor (y and z).
Prediction of reference properties
The influence of the constructive elements of the floors
Timber floors are simple structures with a complex behaviour that depends on the
performance of the whole system: the beams, the struts and the floorboard. In fact,
the load distribution factor conferred by struts and floorboard, designated ksys in
[10], which accounts their stiffening effect, is essential to the estimation of the MOE
when using the natural frequencies obtained in the dynamic tests. [15] determined
that, in a common timber floor, the load distribution factor is 1,15, close to the one
defined by [10] (1,1). Some in situ tests performed by [16] indicated that this factor
can be even higher. The connections between different structural elements have
also a strong influence in the vibrational behaviour of the floor.
Analysis of the results and prediction of the modulus of elasticity (MOE)
The dynamic tests performed with accelerometers positioned in different locations
of a timber floor allow estimating some of the reference properties of the floor,
namely the MOE. This approach can include simple calculations or more complex
numerical models, which reproduce the in situ tests through numerical modal analysis.
In this case, the numerical results are fitted to the results obtained experimentally
by adjusting the mechanical properties through an iterative process. One must note
that it’s a very complex task to properly simulate the timber floors, namely due to
their geometrical irregularities, types of connections between structural elements,
etc. All these characteristics should be carefully integrated in the numerical
structural analysis and, therefore, the calibration process should also be based on the
data resulting from the visual inspection (geometrical, structural and conservation
assessment) and other in situ tests.
If the estimation of MOE is developed through more simple calculations, rather
than with numerical models, the need for understanding thoroughly the geometrical/
structural characteristics of the timber floors is the same. For a simply supported
beam, the fundamental frequency f1 can be calculated using Eq. (1). Although the
equation is defined for simply supported beams, [10] suggests its use for timber
149
floors simply supported on the four sides. In this case, “(EI)long” is the stiffness of
the plate equivalent to the floor in the direction of the beams; “m” is the value of
the mass per unit area and “L” is the span of the floor.
f1 =
π
(EI )long
2 L2
m
(1)
E=
4 × f1 × L4 × m
Ilong × π 2
2
(2)
[8] states that the frequencies obtained in situ are typically up to 50% higher than the
frequencies estimated using Eq. (1), thus suggesting its multiplication by a factor up
to 1,5. This is due to the mentioned stiffness conferred, particularly, by the nailed
floorboard and to the support conditions of the beams in the walls, which, in fact,
don’t correspond to simple supports. “(EI)long” must account the increase of stiffness
of the timber floor due to these conditions and should be obtained multiplying the
stiffness of a simply supported beam (EI) by a stiffness factor (Sf) that, according
to the frequency results obtained by [8], can go up to 2,25.
The MOE of the floor can be estimated following Eq. (2). Some experimental
campaigns in a set of chestnut beams [16], including dynamic and bending tests
[17] indicated a good approximation between the results of MOE obtained in both
tests when multiplying the Eq. (2) by a value between 1,2 and 1,5 (equivalent to
increase the stiffness of a simply supported beam with a factor (Sf) of 1,5 to 2,25).
This methodology is very useful to estimate the behaviour of timber floors in their
present conditions and, in particular, in their future use, regarding, for instance, an
increase of the live loads. The results can indicate the need to strengthen the timber
floor in order to increase its stiffness and improve its dynamic behaviour.
Identification of damaged areas
Dynamic tests have the advantage, compared to other NDT methods, of allowing a
global assessment of floors by measuring the frequencies and modes of vibration.
However, they don’t allow the separate analysis of the structural elements and may
even lead to the occultation of some local damages. Therefore, the use of other NDT
methods, such as X-ray, resistance drilling, stress-wave timing, etc. is fundamental
to analyse thoroughly the state of conservation of timber floors. Still, since the
dynamic response of the floor depends on the type of structural elements and
connections, as well as on their level of degradation, these tests allow evaluating
the global condition of a floor, helping in the detection of damaged areas.
The dynamic identification of other structural elements of a building, particularly
walls, may also help to assess the behaviour of timber floors. In the particular case
of a building in the city centre of Lisbon, whose in situ tests were conducted in
collaboration with the National Laboratory of Civil Engineering (LNEC), the
observation of two opposite facades responding in phase for lower modes indicated
that the timber floors were effectively linking both walls [18]. This result, together
with the observations made during the survey, confirmed the good condition of the
150
timber floors. On the other hand, in a specific area of the main facade, an anomalous behaviour was detected, with the identification of modes associated with a
“free” vibration of this area. This result indicated that the connection between the
timber floor and that area of the wall was deficient, probably due to the degradation
of some timber beams, situation confirmed afterwards with a detailed survey.
Local assessment
As previously stated, the use of NDT, such as X-ray, resistance drilling, stress-wave
timing, is fundamental in the local assessment to analyse thoroughly the state of
conservation of timber floors and thereby the structural health of timber floor structures
and their performance regarding the strength and stiffness values. Once the critical
sections are identified from a preliminary investigation, the corresponding actions
on the floor structure can be quantified and therefore serve as a valuable input in
the global analysis of the structure to achieve an as accurate response as possible.
Appropriate properties, such as density (ρ), the modulus of elasticity (MOE) and
the cross-sectional properties, relating to the quality and health of single members
need therefore to be determined using NDT.
The sequence of use for different assessment devices to detect and localise internal
deterioration and damage, for example, is of great importance in an effective
assessment procedure. It is therefore preferable initially to identify members requiring
further investigation using global measurements, before applying methods such as
resistance drilling and X-ray that require more effort and time. The discussion about
the local assessment of timber floors treats the techniques and instruments used to
assess the structural soundness and performance. This also implies the prediction
of local and semi-global material properties as well as the identification of damage
and deterioration of structural timber members. Those aspects are integrated in the
single method. These steps are presented in an overview and results from NDT
investigations performed from a timber floor investigation of a historical structure
in Sweden are roughly included in this survey.
NDT techniques for local assessment purposes
Among a number of different NDT tools to investigate timber structures, stress wave
timing, X-ray measurements and Resistance drilling were studied to efficiently assess
material properties and locally assess the performance of timber floor structures
Stress-wave timing
The transmission time is highly correlated with the modulus of elasticity (MOE),
Eq. (3), which is of great importance for evaluation of the structural soundness of
151
beams, purlins and columns in large timber structures [19]. Several commercial
instruments, such as FAKOPP®, are available to measure and assess transmissiontime in materials.
MOEdynamic = ρ " v 2
(3)
where ρ is the density and v the transmission time of the stress wave.
Stress wave measurements are a simple and efficient measurement technique to
identify the internal soundness and condition of structural elements, but also to
determine stiffness parameters for structural analysis. This technique requires an
appropriate measurement strategy and approach to efficiently determine the structural
performance of in-situ elements and to successfully detect internal damage, but also
the extent of both external and internal damage. Such a stress-wave-based condition
assessment strategy is simply illustrated in Fig. 4, where critical areas from the
visual inspection were measured stepwise in different directions to identify decay
and its extent on the structural element at different locations along the beam.
Fig. 4. (Right) Illustration of a stepwise (1-5) stress-wave-based assessment approach along a
structural beam and (left) measurements in different directions to detect the extent of the
damage/deterioration (A-B, C-D and E-F), adopted from [20].
Decayed and degraded wood show clear increases in stress wave transmission times,
which also leads to a significant loss of strength [20]. An increase of the velocity sound
by about 30% results in a loss of strength by about 50% [21, 22]. The longitudinal
propagation of the stress waves vary from 4000 m/s to 5500 m/s depending on the
wood species. Transverse propagation of the stress waves are about 25% of the
value in the longitudinal direction and is mainly used as a qualitative parameter to
assess the condition of structural elements and is the most efficient way to detect
decay and its extent [23].
An appropriate measurement strategy to efficiently determine the structural
performance as aforementioned is illustrated in Fig. 4. Comprehensive longitudinal
measurements on single element properties and the structural soundness were
explored on a historical floor structure in Sweden [24].
152
The average velocity from the measurements throughout both of the timber floors was
4969 m/s (std. dev. 335 m/s), which is in the range of sound and good performing
timber.
The principle to detect eventual deterioration as illustrated in Fig. 4 was performed
on a historical floor structure in Sweden [24] in the upper and the lower floor,
where the measurements for the upper floor showed rather constant velocities for all
measurements with a slight decrease near the column, so no signs of degradation/
deterioration were captured. In comparison to the upper floor, the lower floor,
implied some signs of degradation/deterioration within the support region, where
the velocity (3916 m/s) was about 20% lower than the velocity measured for sound
members.
On-site X-ray investigations
The application of digital imaging processing and increasing resolution has made it
possible to use quantitative assessments of components, such as the internal
deformation of fasteners, the dimensions of hidden elements and strains [25]. Until
recently, the opportunities for X-ray investigation have been used for the qualitative
assessment of timber structures, but the opportunities to carry out quantitative
evaluation are of great importance to evaluate the on-site structural behaviour. The
portability of available X-ray devices was a great step forward which especially
facilitates the in-situ operation of e.g. timber structures. Portable units have shown
to be promising, both with regard to quality and feasibility [26]. The ability to
penetrate wood with lower-level energy X-rays and to record images with adequate
quality, was evaluated in 1996 and further evaluations identified technical and
logistical issues [27].
It is a well-known fact that timber density correlates well with other significant
parameters such as MOE and bending strength (modulus of rupture, MOR), which
makes it possible to provide indirect information about these properties by X-ray
imaging, since real-time radiography (radioscopy) allows the study of component
behaviour under moderate loads and is particularly suitable for timber structures due to
the density differences. In order to obtain correct density data the X-ray equipment
must be calibrated and an example of this is presented and thoroughly described by
[26].
Mapping damage and deterioration of timber and mechanical connections is another
powerful application for implementation of X-ray equipment on-site. As most of
the portable X-ray equipment delivers images in a two-dimensional perspective,
additional help using resistance drilling, for example, may be needed for the
volumetric mapping of deterioration as a result of insect attacks. In many cases, a
two-dimensional image is sufficient for determining the severity and progress of
the invisible damage [28, 29], as decay due to rot and high moisture content can be
seen and determined by measuring the area of the void.
In general, the qualitative radiographs from the investigation of a historical timber
floor in Sweden [24] showed that the grain structure of the members was intact at
153
critical sections. Radiographs of connections showed that the connection details
were generally in good condition at the inspected locations except at one location
where signs of deterioration close to the support structure was found, see Fig. 5,
as verified by using resistance drilling.
The effect of local interior deterioration in structural elements needs to be taken
into account in the general assessment and should be remedied.
Fig. 5. A qualitative radiograph (A) & (B) of a beam in a floor structure indicates hidden
deterioration (in the centre), as drilling resistance results verified.
Resistance drilling
Resistance drilling can be used to detect and quantify the internal condition and
decomposition of the wood in timber structural elements. Although the drilling
resistance causes tiny holes, it can be considered as a negligible influence on the
structure, but should be preferably planned properly in the assessment in order to
minimize the surgical intervention.
The use of that small diameter needle-like drill (1.5-3.0 mm at the tip) was introduced
by Rinn [30]. Nowadays, there are some different commercial instruments available,
e.g. IML RESIF400-S® (Fig. 13). The rate at which the wood is penetrated is
constant. The torque needed to maintain a constant penetration rate corresponds to
the drilling resistance and is graphically recorded versus drilling depth. Zones of
lower drilling resistance can be identified as the ones with lower density. As a
consequence those zones usually have lower strength and elasticity. Moreover, lower
drilling resistance may indicate decayed zone, cavities, cracks and crevices. Peaks
in the graph correspond to high resistance and high density. They also indicate the
presence of knots in the cross section. Declines and low points correspond to low
resistance and low density, including decayed zones, cavities or cracks (Fig. 6).
Totally decayed wood shows no drilling resistance.
The drilling resistance is proportional to the relative variations in density, i.e. that
decreasing drilling resistance is followed by decreased torque of the drill. A Resistance
Measure (RM) parameter was implemented that allowed the comparison between
the density of the drilling resistance and mechanical and physical properties of the
timber. The RM parameter is though defined as the integral of the area of the drilling
diagram divided by the length l of the drilled perforation [31], see Eq. (4).
=
(4)
154
Fig. 6. (Left) IML RESIF400-S® and (right) density profile of drilling measurement. Drill
shape and dimensions (mm) are shown in lower left corner.
This method is used to a great extent in the quantification of deteriorated timber.
Resistance drilling enables to locate defects and structural discontinuities in timber
members without affecting the performance, which is particularly important in the
case of heritage structures [32, 33].
An investigation of a historical timber floor structure in Sweden showed that the
coefficient of variation (CoV) of the obtained RM values is high [24], which
increases the uncertainty when it comes to correlating the RM values with the wood
parameters. In investigations of structural timber members in different historical
structures carried out by the authors, partly reported by [34], the coefficient of
variation reached values up to 37%.
Uncertainty about the potential for evaluating wood strength parameters using the
drilling-resistance method was raised. It was found that many parameters, such as
wood moisture content, drill-bit sharpness and drilling angle and direction, affect
the drilling-resistance diagram [35]. The main purpose of the drilling-resistance
method was therefore not the assessment of the mechanical properties of wood but
simply the qualitative investigation of wood based on internal material defects.
Due to its local measurement character, the parameter estimation requires several
measurements.
Drilling resistance measurements can also serve as an input for determining
effective cross-sectional areas of timber beams that affect the second moment of
area and the load-carrying capacity of structural members in general.
Strength and Stiffness predictions from NDT
The densities from the radiographic measurements and the MOE calculations from
the stress-wave measurements as mentioned previously might serve as an input in
the analyses of the material resistances of the individual structural elements.
155
For this evaluation, the characteristic bending strength,
, (MOR), and shear
strength, , , for example can be calculated according to [36] and [37] respectively,
using the following expressions:
(5)
[
] = 0.002065 ∙
,
,
!
[
] = 0.2 ∙
,"
,
(6)
The static MOE (MOEstatic) is usually acquired from the dynamic MOE by a
linear relationship equation according to Eq. 7 [38], but they can also be obtained
directly from the density measurements, Eq. 8 according to [36].
!
= 407.2 + 0.796 ∙
!
= 25.186 ∙ G
.-./.
'()
!
(7)
(8)
The quantitative evaluation of the mechanical properties and the density
using stress-wave timing and radiographic measurements provided both good
agreement and reasonable input for the structural analysis.
It is, however, important to remember that there may be some uncertainty about
the correlations for timber between the output from the measured properties and
the strength parameters using assessment techniques that might weaken the
estimation of the actual capacity.
Conclusions
The focus of the present paper is the assessment of the behaviour of timber floors
with the use of global and local NDT, namely involving dynamic behaviour analysis
with seismographs, stress-wave timing, X-ray and resistance drilling.
The use of instruments to assess the dynamic behaviour of the floors, such as
seismographs, has the advantage of allowing a comprehensive assessment of the
timber floors’ global behaviour, by measuring the characteristics associated with
their dynamic performance, which allow estimating MOE. It can also evaluate the
global state of conservation of the floors and detect damaged areas, which are
usually associated to lower natural frequencies and, consequently, to lower MOE.
However, the use of seismographs does not allow the separate analysis of the
structural elements that compose the floors and it may even lead to the occultation
of damaged elements. For that reason, it should be used together with other NDT
that can lead to the prediction of local properties, such as stress wave timing,
resistance drilling and X-rays. To assess the general quality of the timber, it is
sufficient to apply stress-wave measurements, in combination with resistance
drilling and X-ray measurements. Reliable results can be obtained, thereby
increasing the ability to minimise interventions and prolong the service life as a
part of sustainable development. The extent of measurements should be adjusted to
match the structural condition and existing information relating to the structure
[24].
156
The combined use of these NDT allows a better understanding of the timber
floors behaviour and the implementation of more efficient rehabilitation
and (or) strengthening techniques.
References
[1]
Mohager S. Studier av krypning hos trä: med särskild hänsyn till inverkan av konstanta och
cyklist varierande fukttillstånd. Stockholm: KTH - Institutionen för Byggnadsmateriallära;
1987.
[2]
Rug W, Seemann A. Strength of old timber: Test cores from 50-400 year old timber
indicates strength of old timbers corresponds to that of new. Building Research and
Information. 1991;19(1):31-7.
[3]
ICOMOS. Recommendations for the analysis, conservation and structural restoration of
architectural heritage. ICOMOS International Committee for Analysis and Restoration
of Structures of Architectural Heritage; 2005.
[4]
Macchioni N, Piazza M. Italian standardisation activity in the field of diagnosis and
restoration of ancient timber structures'. Structural Analysis of Historical Constructions,
Macmillan India Ltd, New Delhi. 2006.
[5]
ISCARSAH. Recommendations for the analysis, conservation and structural restoration
of architectural heritage. ICOMOS; 2003.
[6]
Cruz H, Yeomans D, Tsakanika E, Macchioni N, Jorissen A, Touza M, et al. Guidelines
for the on-site assessment of historic timber structures. International Journal
of Architectural Heritage. 2013(just-accepted).
[7]
Hu, L. J., Chui, Y. H., Onysko, D. M. (2001). Vibration serviceability of timber floors in
residential construction, John Wiley & Sons, Ltd., Quebec, Canada.
[8]
TRADA. (1998). Vibration in timber floors, TRADA Technology Ltd, Hughenden
Valley, High Wycombe, Bucks.
[9]
Ohlsson, S. (1988). "Ten years of floor vibration research - A review of aspects and some
results." Symposium/Workshop on Serviceability of Buildings (Movements, Deformations,
Vibrations), Ottawa, Canada, 435-450.
[10] European Committee for Standardization CEN, EN 1995-1-1: Eurocode 5: Design of
timber structures - Part 1-1: General - Common rules and rules for buildings, 2004, CEN,
Belgium.
[11] Caetano, E. (1992) Identificação Experimental de Parâmetros Dinâmicos em Sistemas
Es-truturais. Master thesis on Civil Engineering, University of Porto.
[12] Soltis, L. A., Wang, X., Ross, R. J., Hunt, M. O. (2002). "Vibration Testing of
Timber Floors Systems." Forest Products Journal, 52, 75-82.
[13] Brincker, R., Ventura, C. (2001) Damping Estimation by Frequency Domain. Proceedings
of the 19th International Modal Analysis Conference (IMAC), Kissimee, Florida, USA.
[14] SVS (2006) ARTeMIS Extractor Pro, Release 4.0, Structural Vibration Solutions, Aalborg,
Denmark.
[15] Blass, H. J. (1995). Load Sharing. Timber Engineering -Step 1.Lecture B16, Almere
Cen-trum Hout, Holanda.
[16] Ilharco, T. (2008) Pavimentos de madeira em edifícios antigos. Diagnóstico e intervenção
estrutural. Master Thesis, University of Porto. [17] European Committee for Standardization
157
CEN, EN 408: Timber structures – Structural timber and glued laminated timber –
Determination of physical and mechanical properties, 2003, CEN, Belgium.
[18] Lopes, V. et al (2010). Conference Património 2010. Porto, Portugal.
[19] Wang X, Divos F, Pilon C, Brashaw BK, Ross RJ, Pellerin RF. Assessment of decay
in standing timber using stress wave timing nondestructive evaluation tools. USDA Forest
Service Forest Products Laboratory General Technical Report FPL-GTR-147, 12pp. 2004.
[20] Dackermann U, Crews K, Kasal B, Li J, Riggio M, Rinn F, et al. In situ assessment of
structural timber using stress-wave measurements. Materials and Structures. 2013:1-17.
[21] Pellerin RF, Ross RJ. Nondestructive evaluation of wood: Forest Products Society; 2002.
[22] Ross RJ, White RH, Pellerin RF, Wang X, Brashaw BK. Wood and timber condition
assessment manual: Forest Products Society; 2004.
[23] Ross RJ, Hunt MO. Stress wave timing nondestructive evaluation tools for inspecting
historic structures: a guide for use and interpretation. General Technical Report Forest
Products Laboratory, FPL-GTR-119, USDA Forest Service2000.
[24] Lechner T, Nowak T, Kliger IR. In situ assessment of the timber floor structure of the
Skansen Lejonet fortification, Sweden. Construction and Building Materials. 2014;58:85-93.
[25] Kasal B., Adams A. and Drdacky M. (2008): Application of Digital Radiography in
evaluation of Components of Existing Structures. RILEM Symposium on On Site
Assessment of Concrete, Masonry and Timber Structures - SACoMaTiS 2008. Varenna Lake Como, Italy.
[26] Lechner T, Sandin Y, Kliger R. Assessment of Density in Timber Using X-Ray Equipment.
International Journal of Architectural Heritage. 2013;7(4):416-33.
[27] Anthony RW. Examination of Connections and Deterioration in Timber Structures Using
Digital Radioscopy. In: Bosela PA, Delatte NJ, Rens KL, editors. Third Forensic Engineering
Congress. San Diego, CA: American Society of Civil Engineers; 2003. p. 320-8.
[28] Rinn F., Schweingruber F. and Schär E. (1996): Resistograph and X-ray density charts
of wood comparative evaluation of drill resistance profiles and X-ray density charts of
different wood species. Holzforschung, Vol. 50 (4), pp. 303-311.
[29] Lear G. C. (2005): Improving the Assessment of In Situ Timber Members with the Use
of Nondestructive and Semi-Destructive Testing Techniques. Master of Science Master's
Thesis, Civil Engineering, North Carolina State University.
[30] Kasal B, Anthony RW. Advances in in-situ evaluation of timber structures. Progress
in Structural Engineering and Materials. 2004;6(2):94-103.
[31] Lourenço PB, Feio AO, Machado JS. Chestnut wood in compression perpendicular to the
grain: Non-destructive correlations for test results in new and old wood. Construction
and Building Materials. 2007;21(8):1617-27.
[32] Jasieńko J, Nowak T, Hamrol K. Selected methods of diagnosis of historic timber structures principles and possibilities of assessment. Advanced Materials Research.
2013;778(2013):225-32.
[33] Branco JM, Piazza M, Cruz PJS. Structural analysis of two King-post timber trusses:
Nondestructive evaluation and load-carrying tests. Construction and Building
Materials. 2010;24(3):371-83.
[34] Jasieńko J, Nowak T, Hamrol K. Selected methods of diagnosis of historic timber structures
principles and possibilities of assessment. Advanced Materials Research.
158
2013;778(2013):225-32.
[35] Branco JM, Piazza M, Cruz PJS. Structural analysis of two King-post timber trusses:
Nondestructive evaluation and load-carrying tests. Construction and Building
Materials. 2010;24(3):371-83.
[36] Dinwoodie JM. Timber: Its nature and behaviour. 2nd edition ed. London: E & FN Spon;
2000.
[37] Glos P. Solid timber - Strength classes. In: Blass HJ, Aune P, Choo BS, Görlacher R,
Grif-fiths DR, Hilson BO, et al., editors. Timber Engineering - STEP 1: Centrum Hout;
1995. p. 4-5.
[38] Íñiguez G. Clasificación mediante técnicas no destructivas y evaluación de las propiedades
mecánicas de la madera aserrada de coníferas de gran escuadría para uso estructural.
Doctoral Theses Universidad Politécnica de Madrid, ETS de Ingenieros de Montes. 2007.
159
160
PART III
ASSESSMENT OF THE INTEGRITY OF
161
PART III - ASSESSMENT OF THE INTEGRITY OF STRUCTURAL TIMBER MEMBERS
162
NDT imaging techniques for the inspection of
timber structures
Mariapaola Riggio1, Jakub Sandak2, Steffen Franke3
1
2
3
Wood Science & Engineering, Oregon State University, USA
[email protected]
CNR Timber and Trees Institute IVALSA, S.Michele all’Adige (TN),
Italy, [email protected]
Bern University of Applied Sciences, Biel/ Bienne, Switzerland,
[email protected]
Abstract
This chapter deals with the application of NDT imaging techniques as
complementary methods to be used during visual inspection. NDT imaging
can be used to detect inhomogeneity and to identify the areas at the highest risk
for damage in timber structures. The potential of some imaging techniques
accepted and practiced for the assessment of timber structures is here discussed.
1. Introduction
Visual inspection is an essential phase for the assessment of timber structures,
during which vulnerability factors and visible damage can be identified.
Nondestructive imaging techniques can complement visual information. According
to the penetration depth of the wave field, imaging techniques make it possible
to analyse either surface/subsurface features or internal heterogeneities of the
wood material.
Imaging methods use wave-based techniques: they analyse the response of the
material to wave fields of different nature.
Two different kinds of waves can be utilized: they are elastic or mechanical
waves (or called stress waves) and electromagnetic waves (i.e. X-rays, gamma
rays, ultraviolet rays, visible light, infrared, microwaves).
Non-destructive images map a measured physical parameter, over a defined plane
perpendicular to the direction of the radiation/wave transmission (e.g. radiographs,
thermograms). Tomographic techniques, instead, map parameter values on a plane
across the object and parallel to the direction of the radiation/wave transmission,
allow cross sectional imaging.
163
Reflectometric techniques, such as the ultrasound echo technique and the GPR,
allow different kind of scans, in particular:
• A-scan: a trace that show the wave transition time and the intensity of the
pulse;
• B-scan: a profile composing different A- scans recorded at a given distance, resulting in 2D cross sectional map on a plane parallel to the direction of the
wave transmission;
• C-scan: 2D cross sectional map on a plane perpendicular to the direction of the
wave transmission, obtained by interpolating a horizontal layer from several Bscans.
Depending on the feature of interest and the related properties, a specific imaging technique can be selected [1].
2. Non-destructive imaging methods for inspection of
timber members.
In the following sections some imaging techniques applicable for the inspection
of timber members on site are described.
2.1. Optical methods: photogrammetry
2.1.1. Principle of the method
Photogrammetry (wave field in the visible range) can be used to extract threedimensional metric models from photographic images. It also allows reconstructing
a textural database for the selected surfaces of an object. It can be adopted for
analysing the extent and position of material features, visible on the inspected
element surface [2]. Close range photogrammetry basically involves the use of a
network of photographs of an object taken from different angles. In monoscopic
photogrammetry, convergent shots at different scale can be used, thus allowing the
adoption of much more flexible geometric acquisition schemes than those of the
stereoscopic method. The monoscopic method allows the use of non-metric camera,
after a preliminary calibration. The accuracy of the restitution depends on several
factors: primarily on the scale of the photogram, secondarily on the geometry of
the acquisition scheme and on the accuracy of the interior and exterior orientation
[3].
Different types of light sources could be utilized for enlightenment. The
illumination angle and light source position in relation to the specific heterogeneity
and camera are crucial to obtain good detecting capabilities.
164
2.1.2. Applications
Photogrammetric techniques permit to obtain reliable metric information of
defects, damage and distortions, from the analysis of orthophotos.
From image data, geometrical data of the detected features can be extracted, for
further analysis (e.g., measurement of strength affecting characteristics, 3D
visualization and modelling of the macroscopic material features). Therefore,
photogram-metry can be used for supporting advanced visual strength grading of
timber elements in service and for the acquisition of metric/geometric data for the
numerical analysis of wooden elements [4].
2.1.3. General remarks
Site conditions and accessibility can strongly affect the applicability of the method.
Metric accuracy of orthophotos depends on the acquisition methods and the
camera used.
It can be advisable to use natural points (e.g. features of the wooden texture)
for buddle adjustment.
The photogrammetric survey should be coupled with a topographic survey of
control points. In case of elements with very rough/non-planar faces, the use of orthophotos can be insufficient to completely describe the material characteristics.
Radiometric quality of photographs should be ensured using constant uniform
lighting.
2.2. Thermography
Thermography provides information related to the thermal properties of the sample,
through a map of the temperature distribution on the surface of the object.
Because of the low thermal conductivity of wood, thermography allows the
detection of defects near the surface (approx. 1 mm).
There are two main variations of thermography the active method (i.e. inspected
element is illuminated by a heat source) or the passive one (wood surface is
subjected to natural heating).
165
The most popular device for thermal imaging is the infrared camera, which is a
thermal wave detector in the infrared domain, detecting infrared radiation emitted
from the object.
The thermal gradient (and not the temperature itself) makes it possible to obtain
information about the integrity of the structure. The thermal gradient is calculated
from the temperature distribution.
2.2.2. Applications
Thermography can be used as a preliminary non-contact screening procedure, to
select areas for more detailed analysis.
For defects detection, active thermography is generally advisable while, in
many circumstances, passive thermography can be apt to map areas with higher
moisture content.
Thermography is also an advisable technique, for the geometrical survey of
elements hidden/covered by plaster, or by other similar thin layers [4]. It can be
also used to assist visual strength grading of coated timber elements (e.g. painted
historical wooden members).
Analysis of temperature gradients generally allows detection of most superficial
and sub-superficial abnormalities. However, the thermal image does not
provide any information about the depth of discontinuities. As regards information
about moisture, only qualitative indication about location of moist areas can be
gathered, while the technique is not appropriate for quantitative estimation of
moisture content.
2.3. X-ray radiography
X-ray radiography uses electromagnetic waves with wavelengths in vacuum
between 10-8 m and 10-12 m. The X-rays are absorbed depending on the material
density.
For wood, the X-ray absorption coefficient is defined as:
µ = µ'• ρ
166
(1)
With µ’ as the mass absorption coefficient in [m2/kg] and ρ the density of the
material in [kg/m3].
The detectable X-ray waves on the film plate depends on the pulse intensity of
the X-rays, the distance of the test object to the transmitter as well as to the film
plate and also the thickness of the material. In the X-ray acquisition process, the
test object is located between the X-ray transmitter and the film plate. The X-rays
transmitted travel through the test object and will be absorbed with different
intensities before they hit the film plate. The material specific absorption of the
X-rays leads to the so called radiogram which will finally be transferred into a
grayscale picture. The whole volume of the three dimensional test object in front
of the film plate will be reproduced as a two dimensional picture.
2.3.2. Applications
The X-ray radiography is mainly used for the detection of internal heterogeneities.
In typical timber elements the differentiation of various densities like soft- or
hardwood, sap wood or heart wood, early or late wood. Furthermore the assumptions
and growing direction of knots can be distinguished.
Fungal or insect decay can be observed within the X-ray radiography as well.
Normally the visual inspection is used for fungal or insect decay, but in some
cases structural elements are covered or only accessible from one side, so that
the mobile X-ray system can be used for detailed analyses or specification of
assumptions.
Further assessments of timber structures using the X-ray radiography are
shown e.g. in [6-7-8].
The general principle of the X-ray radiography is similar to taking a picture with a
photo camera. But here for the quality of the radiogram the pulse intensity, the distance
of the test specimen between the transmitter and the film plate and the thickness
respectively the density of the object influence the resolution and accuracy of the
method. The limitations described are observed using the X-ray unit XR 200
167
with a maximum photon energy of 150 KVP and X-ray dose per pulse of 0.026
0.040mSv and a test object made of European Spruce with a density of 480kg/m3
and a moisture content of about 15%. The X-ray unit can be used within to the
following regulations [6]:
• Increasing the distance between the transmitter and the test object results in the
projection of a smaller area where the object is represented enlarged but with
less sharpness and more noise.
• A minimum distance between the transmitter and the film plate of about 1
meter is necessary using a film plate of 30 by 40 cm. A further reduction of
this distance leads to a clear “burned” spot and unusable radiograms.
• Typical structural timber elements with thicknesses up to 300 mm can reliable
be assessed with the used system. For greater thicknesses the contrast vanishes
and only objects with distinctly different densities, e.g. parts of steel embedded
in wood are visible in the radiogram.
The safety requirements for the use of the mobile X-ray system do not restrict
the practical use on existing timber structures. In practical use, the safety zone for
this unit is specified as follows: 3 meters around the transmitter, 30 meters in
measuring direction and 11 meters perpendicular to it.
In general the users carry a personal dosimeter to register any irradiation.
2.4. Microwave
2.4.1 Principle of the method
Microwaves are electromagnetic waves in the frequency range between about 300
MHz and 300 GHz. The microwave technique in connection with range or flight
of time measurement can be termed as radar technique (radar: radio detection and
ranging) or GPR (ground penetrating radar).
There are two basic microwave techniques, the transmission and the reflection.
In a transmission scanner the transmitting antenna, illuminates the piece of wood,
with a uniform microwave field.
The probes used to evaluate the signal of microwaves can be either scatterometers/
radars, which measure the scattering properties, or reflectometers, which
measure the reflectivity of the workpiece due to inhomogeneous and defects. In
the first mode, the material is located between two antennas (one for transmission
the other for reception), in the second mode, both antennas are located in front of
the material. Usually, the following equipment is required for microwave imaging:
a) two dipole or horn antennas (transmitting and receiving); b) generator of
electromagnetic wave at relatively high frequency and low power; c) data acquisition
system[9].
168
GPR system generates a series of short pulses that travel through the material
and back-scatter.
GPR measurements consist in recording a profile by moving the antennas on
the tested structure along a linear direction. A trace (A-scan) can be recorded every
centimetre of the profile. From series of A-scans, B-scan can be generated.
2.4.2. Applications
Depending on the microwave frequency and measurement hardware it is possible
to detect the defect presence directly, or its presence can be noticed only by the
introduction of “noise” into attenuation, phase and polarization signals.
In [10] a new system for the in situ evaluation of timber structures based on
microwave reflectometry is presented. The obtained results show that the reflectometric methodology can clearly highlight the presence of discontinuities inside
the wood, but nature and dimension of the heterogeneities cannot be characterised.
Due to its high permittivity, free water in materials like wood strongly influences
the reflection and transmission behaviour of microwaves. The changes of the
microwave properties due to moisture can be used to treat the moisture as a material
property and to detect, image and quantify it and its distribution in the object [11].
As a rule of thumb, the approximate minimum detectable defect size is around half
of the wavelength of the microwave frequency used. By decreasing the microwave
frequency generally the penetration depth will be enhanced, but the lateral
resolution at imaging will be lowered. By taking into account the parameters of
the test object (dimensions, shape, moisture content, fibre orientation) and of the
measurement equipment (frequency, type of antenna, microwave power, distance
of object, etc.) some optimization will be necessary to get a good compromise
between the resolution and penetration depth.
2.5. Stress-wave tomography
Acoustic tomography is a technique used to reconstruct the properties of the materials under inspection from stress-wave propagation data. In acoustic tomography,
the most typical parameter to measure is the time of flight (TOF), i.e. the time that
169
it takes for the wave to travel a distance through a medium. The transit times,
recorded for each pair of transmitting-receiving points, as well as the coordinates of
these points, are the input data for the tomographic analysis. Velocity maps are the
output data. Local apparent velocities are computed over a Cartesian grid of square
pixels, according to the geometric arrangement of the sensors on the element surface.
Tomographic images are then generated as the velocity distribution throughout the
inspected section.
The basic equipment for stress-wave TOF data acquisition is composed by:
•
•
•
•
•
•
an oscilloscope, for visualization and analysis of the signal;
a function generator, with a given pulse repetition frequency;
a timer, which controls both the trigger of the generator and the counter;
a signal amplifier;
a signal filter;
an instrumented hammer for emitting low frequency signals (< 10 kHz);
alternatively, transducers can be tapped with a steel hammer to generate sound
waves;
• piezoelectric transducers, which are used for emitting high frequency signals
(typically, 20÷100 kHz);
• piezoelectric transducers for receiving the signal (or micro-accelerometers, in
case of low frequency signal);
• preamplifiers, which are required in most applications on wood, because of the
high attenuation of the transmitted waves in the material, especially in case of
thick elements.
For tomographic data acquisition, it is desirable to use a multi-channel device to
speed up the measures. Alternatively, a set of probes, operating both as signal emitter
and receiver, can be used.
The definition of the parameters for data acquisition depends on the experimental
conditions and the scale of the investigated characteristics. The minimum size of
the detectable defects is predetermined, depending on the frequencies used and the
geometric resolution of the tomography.
Frequency range of the emitting source has to be chosen in order to optimize
resolution of the analysis and attenuation of the signal. The acquisition scheme for
each specific test should be carefully designed, considering the characteristics of
the investigated section and the accessibility of the sensed surface [11].
2.5.2. Applications
Acoustic tomography can be applied to timber structural elements to detect strength
affecting heterogeneities (i.e., knots) and damage (i.e., decay, cracks).
In particular, longitudinal tomograms are useful for screening the element along
the entire length, identifying problematic areas, where low velocity values are
mapped. Imaging of selected transverse sections is aimed at gross estimation of the
170
heterogeneity extent. Very high velocity values in acoustic tomograms of wood are
generally associated with knots. Low velocity areas are associated with low material
density, often caused by decay. In particular, decay due to rot fungi and diffuse
insect attacks can be detected by means of the acoustic tomography.
The main methodological aspects, which influence acoustic tomography, are the
applied frequency, the number of independent measurements, the adopted acquisition
scheme and the applied inversion technique.
The technique permits only qualitative, large-scale analysis (e.g., maps of entire
elements),whereas complementary non-destructive/semi-destructive methods should
be used to obtain local quantitative information.
In general, it is recommended to couple acoustic tomography with local mechanical
tests, such as resistance drilling tests, for the detection of internal zones of lower
densities, and with visual/photographic analysis of the element faces for correlation
of internal features and external indicators [12].
3. Conclusions
For inspection of timber structures it is important to reliably detect defects,
damages and material heterogeneities in wooden members. Normal praxis
based on visual inspection and point measurements of drill resistance can be
improved through multi-sensor, multi-scale, multiresolution analysis [13].
Most of the techniques described in the paper allow qualitative analysis of
the elements and detection of the presence and location of main gross defects
and damage.
High-resolution photogrammetry and IR thermography have great potential as
preliminary noncontact screening procedures, to select areas for introspective
analysis.
Stress-wave ToF tomography and microwave scanning, thanks to their
completely non-destructive nature and the possibility to map large timber sections, can
be adopted as large-scale global evaluation method, for decay and defect detection
in the interior of the wood material, to be followed by further investigation on
anomalous velocity areas.
X-ray radiography can be used for more detailed investigation of specific
points in the structure.
171
References
[1]
Bucur V. (2003) Nondestructive characterization and imaging of wood. Springer,
Berlin/Heidelberg.
[2]
Riggio M., Prandi F., De Amicis R., Piazza M. (2011). Geometrical characterization of timber structural elements in the roof of S. Lorenzo church in Tenno (TN,
Italy) using close range photogrammetry. CULTURAL HERITAGE Istanbul 2011. 5th
International Congress "Science and Technology for the Safeguard of Cultural
Heritage in the Mediterranean Basin". Istanbul, Turkey, 22-25 November 2011.
[3]
Kraus, K. (2007). Photogrammetry. Geometry From Images and Laser Scans. Walter
de Gruyter, Berlin.
[4] Riggio M, Prandi F, De Amicis R, Piazza M (2012). Use of high resolution digital
images and NDT imaging techniques for the characterization of timber structural
elements. In: Nondestructive Testing of Materials and Structures. O. Büyüköztürk
et al. (eds), RILEM Bookseries 6:365-371.
[5]
Cantini L, Tedeschi C, Tiraboschi C, Binda L. Use of thermovision for the survey of
a timber vault in Torino. In: Nondestructive Testing of Materials and Structures,
O. Büyüköztürk et al.(eds.), RILEM Bookseries 6, Pages 1203- 1208. RILEM 2012.
[6]
Franke S., Franke B., Scharmacher F. (2013). Assessment of timber structures using
the X-Ray technology, Advanced Materials Research, Vol. 778, pp. 321-327
[7]
Brashaw B.K., Bucur V., Divos F., Goncales R., Lu J., Meder R., Pellerin R.F., Potter
S., Ross R.J., Wang X., Yin Y. (2009). Nondestrucitve testing and evaluation of wood: A
worldwide research update, Forest Products Journal 2009/59, pp. 7-14.
[8]
Wei Q., Leblon B., La Rocque A. (2011). On the use of X-ray computed tomography
for determining wood properties: a review, Can. Journal for Res. 2011/41, pp.
2120-2140.
[9]
M. Sbartaï. Ground penetrating radar. In: In Situ Assessment of Structural Timber:
Discussion of Classical and Modern Non-Destructive and Semi-Destructive
Methods for the Evaluation of Wood Structures; Series: RILEM State of the Art
Reports, Kasal, Bohumil; Tannert, Thomas (Eds.), Springer, 2011, Vol.7, Pages
25-36
[10] Macchioni N., Mannucci M., Olmi R., Sabrina Palanti1,d, Cristiano Riminesi. (2013)
Microwave reflectometric tool for non-destructive assessment of decay on timber
structures. In Proceeding of the 2nd international Conference on Structural Health
Assessment of Timber Structures. Advanced Materials Research Vol. 778. Pages 281-288
Trans Tech Publications, Switzerland
172
[11]
Schajer G.S., Orhan F.B. (2006). Measurement of wood grain angle, moisture content
and density using microwaves, Holz als Roh- und Werkstoff. 64, 483– 490
[12] Dackermann U., Crews K., Kasal B., Li J., Riggio M., Rinn F., Tannert
T. (2014). In situ assessment of structural timber using stress-wave measurements.
Materials and Structures . 47 (5): 787-803.
[13] Riggio M, Sandak J, Franke S (2015). Application of imaging techniques for detection
of defects, damage and decay in timber structures on-site, SI: Appraisal of Wooden
Members, Construction and Building Materials, 2015 -Elsevier (in press).
173
174
Assessing the integrity and strength of gluelines
Philipp Dietsch
Thomas Tannert
Dr.-Ing., Team Leader Timber Structures
Chair of Timber Structures and Building
Construction
Technische Universität München, Germany
Assistant Professor, PhD, PEng
Associate Chair Wood Building Design and
Construction
The University of British Columbia, Canada
1 Introduction
Historically, timber structures were characterized by wood elements which were
limited both in their cross-sectional dimensions and length by the dimensions of
the existing tree population in the surrounding area. The step from solid timber
elements to glued(-laminated) timber elements, started by Hetzer [1], represented
a significant technological progress, widening the range of application of timber
structures in the building sector. Joining single boards to continuous lamellae and
subsequently gluing and stacking them enabled to disengage from the size of the
stem cross section and to reduce the effect of defects (e.g. knots) in the material.
The results were larger and more variable (e.g. curved) geometries and timber
being increasingly used, e.g. for large-span structures. Consequently, the importance
of assessing large timber structures grew, naturally resulting in an increased
interest of the professional community in assessment methods for existing timber
structures. The need for an assessment of an existing structure can be based upon
a multitude of reasons, e.g. (taken from [2]):
 change of the requirements to the use or to the structure (increased loading,
increased service life …)
 doubts whether the assumptions applied during the design are fulfilled (no
inspection for longer period or negative results of an investigation, unexpected
degradation, accidental loads, new knowledge …)
The performance of glulam beams depends on the quality of the individual
lami-nations, the quality of the finger joints and the quality of the glue-lines. This
chapter focuses on methods to assess the gluelines of glued timber elements. The
assessment of the laminations is comparable to the assessment of solid sawn timber
(moisture content, rot …) and finger joints have shown less probability of causing
structural damages or failures. There are multiple methods available to assess the
properties of interest.
175
Nevertheless each method only allows assessing a certain type
of property or damage. Therefore the application of just one method might not be
suitable to enable confident decisions, making it necessary to combine different
methods to derive a full picture about the residual performance of the gluel-ines.
The objective of this chapter is therefore to present feasible methods to assess
gluelines in structural timber elements and to evaluate each of them with regard to
the following objectives, see also [3]:
 Which properties can be determined / which properties cannot be determined?
 How accurate and precise are the results (e.g. degree and size of damage; local/
global results)?
 How complex and time consuming is its application (on-site, (non-) destructive)?
 Which combinations of methods are useful to derive a clear picture of the
structural integrity of the assessed structure?
2 Approach to the assessment of gluelines
The objective of any assessment of a structure is to detect, localise and document
damage, to determine the actual degree of damage, to identify the reasons for damage
and to define means to repair the damage. Typically the assessment is divided into
different phases, structured by the detailedness of the investigations. The obligatory
visual inspection (Preliminary Evaluation - Phase 1) can be followed by more precise
methods to detect and localise damage, e.g. non-destructive or semi-destructive
techniques (General Investigation - Phase 2, Detailed Investigation - Phase 3).
Sometimes the on-site assessment is supported by investigations in the laboratory.
The sequence of a structural assessment can be summarized in a flowchart, see
Fig. 1.
The number of phases necessary depends on the level of doubt, the feasibility and
simplicity of repair or strengthening, in combination with economic considerations.
Depending on the type of structural property to be assessed and the phase of
assessment, there are a number of applicable assessment methods. With respect to
the assessment of gluelines, the three major properties of interest are:
 type of glue used, e.g. to assess its durability in given conditions
 integrity, i.e. potential loss of bond of the glueline due to cracking or delamination
 strength of the glueline, e.g. shear strength
176
Fig.1 Illustration of the phases approach (from [4]).
Methods that are applicable for the assessment of the integrity and strength of
glu-elines are described in the following. Hereby the described methods are listed in
the order 1) property of interest and 2) frequency of application. Each description
is concluded by a short evaluation of the method in tabular form.
177
3 Methods to determine the type of glue used
3.1 Review of documentation (and visual inspection)
Before starting any assessment of a structure, adequate attention should be given
to any existing documentation. A thorough review of documentation can give an
expeditious overview of the structure which would require more time if all necessary
information had to be acquired on site. Some information can only be drawn from
existing documentation, since it is simply not obtainable anymore by other means
(e.g. producer, conditions during gluing, previous inspections …). In the case of
historical structures, original planning documents are in most cases not available
anymore, but chances are that they can be found in historical archives.
For buildings constructed after the turn to the 20th century, copies of original
blueprints or sketches may be available which contain useful dimensions. If a building
has already undergone renovation or reconstruction, newly established documentation,
based on measurements of dimensions in the existing building is sometimes available.
For such cases or for recently constructed buildings, documentation is – in many
countries - kept in administrative offices of the building authorities (municipalities
or states). In other countries, such documentation is kept by the building owner.
Companies involved in the design and construction are also a good source to
receive planning documentation (original designs). If these sources do not reveal
the necessary information, openly available literature can be an eligible source
since quite a few buildings are covered in books on architecture or architectural
magazines as well as magazines for structural engineers or craftsmen.
In the case of glued timber elements, existing documentation can yield
information on e.g. wood species and strength class; type of glue, producer and
gluing process; transportation, gluing on site and coatings. Such information can
oftentimes eliminate costly and time-expensive assessment methods.
If the type of glue cannot be identified by available documentation, it can - in
the case of resorcinol glue - be determined by its apparent characteristics, i.e. its
dark colour and its typical odour when scoring the surface. A white glue line is
oftentimes an indicator for urea-formaldehyde or melamine formaldehyde glue (at
least for buildings constructed before the millennium). A first indication which of
the two types of glue was used is given by the age of the glued timber element:
the first melamine glues were used from 1975 onwards.
3.2 Laboratory tests
The most common objective potentially necessitating laboratory tests for glued
timber specimen is the determination of type of glue. A first indication can be
derived from a boiling test on the basis of EN 314-1 [5] or, more recent, EN
302[6]. The sample, mostly a drill core (see chapter 5.1) is exposed to boiling
178
water for six hours and subsequently to cold water for two hours. Specimens
featuring a glue line from urea-formaldehyde will in most cases fall apart during
this treatment, since this type of glue is not permanently moisture proof.
Specimens featuring a glue line from melamine-formaldehyde should still be
intact and should still deliver acceptable strength when tested. It is the authors
experience even specimens with glue lines from melamine-formaldehyde
sometimes fail during this treatment. A more accurate option is to use spectral
analysis to identify the type of glue used. This method is based on a
differentiation by means of the different absorption spectra of substances. A
common tool for this analysis is the X-ray diffractometer.
Table 1 Laboratory tests - Evaluation.
Type
Optical; thermo-mechanical; Application
non-stationary; semidestructive
Phase III (Rare)
Extent
Local / Global
Time / Cost
High / Medium
Validity of
results
Good (exact)
Constraints
Rather complex and time-consuming;
only in laboratory
4 Methods to assess the integrity of gluelines
4.1 Visual inspection
A visual inspection is part of every structural assessment and is a mandatory basis for
further examinations. It should be carried out in combination with a review of existing
documentation and planning documents (see preceding paragraph). The latter can aid
to receive a clearer picture of the structure before the actual assessment on-site. It
helps to determine, if the structure was carried out according to plans and to carry
out comparative calculations in order to check which members are highly utilized.
There is no universally valid approach to a visual inspection. Every engineer has to
find an individual approach according to his/her experiences.There is a multitude
of reference books and check-lists which can support the assessment of failures,
e.g. [7], [8], [9]. These instructions oftentimes only apply to certain building and
construction types and are only meaningful if applied by experienced engineers.
The visual inspection yields an impression of the overall condition of the structure
and the glued timber elements, including a first overview and registration of the
visible degree of damages. All accessible structural elements are examined handson, e.g. in terms of cracks in the glueline or delamination. A magnifying glass can
aid in determining the age of a crack or delamination (dust, discolouration within
crack) and potential adhesion or cohesion problems. Magnifying glasses often come
179
with an integrated lamp to lighten the crack, enabling to take pictures magnifying
the interior of the crack, see Figs. 2-5.
It is self-explanatory that the visual inspection has to be well documented. Thereby,
the benefits of photography as part of an easily accessible and comprehensive
documentation are obvious. The captions of the pictures picked for the written
documentation should include a clear indication of the location of the detail within
the structure. Areas that are designated for further examination by more specific
methods should be marked on site and in the documentation. A building book has
proven to be a good tool to facilitate future inspections and to guarantee a consistent
documentation, even with the change of authorized personnel [10].
Fig. 2 Open glueline in glulam beam.
Fig. 3 Magnifying glass and camera
Fig. 4 Magnified detail of open glueline, inadequate
gluing during production
Fig. 5 Magnified detail of open glueline,
failure due to changes in timber moisture
content.
Table 2 Visual Inspection - Evaluation.
Type
Visual; on-site;
non-destructive
Application
Phase I (always)
Extent
Local/ Global
Time / Cost
Low / Low
Constraints
Limited to surface; not quantifiable
information; dependent on experience;
accessibility can be limited
Validity of Limited (qualitative; on
results
surface; experience)
180
4.2 Mapping of delaminated areas and/or cracks
For a description of this method, the interested reader is referred to the chapter
“Mapping of cracks and effect of environmental conditions” in this publication.
Apart from the crack dimensions (length, width, and depth), the position of cracks
within the structural element is essential for the evaluation of their consequence
on the structural integrity of the timber element. It is also relevant if the cracks
appear predominantly or with a certain frequency in the timber or in the glueline.
A crack in the direct vicinity of the glueline is not in all cases an indication of
adhesion or cohesion problems. The wood material in the vicinity of the gluelines
is the weakest part of the lamella due to the cutting of fibres during sawing. In
addition, there is a discontinuity between the density of the boards and
also between the shrinkage movements of both sides. Oftentimes the cracks
occur in the boundary layer next to the glue line but within the wood. The
crack should therefore be closely investigated before conclusions on the failure
mechanisms are drawn [11].
Table 3 Mapping of delaminated areas and/or cracks - Evaluation.
Type
Visual ; on-site ; Application Phase I (Frequent)
Extent
non-destructive
Time / Cost Medium / Low
Validity of
results
Local
Constraints Local and temporary assessment; investigation of
larger areas time-consuming
4.3 Stress waves - Ultrasonic Echo Technique
The use of stress waves is based on the propagation of sound waves
through material. Amongst the different possibilities to apply stress
waves in the assessment of timber members (see e.g. [12], [13], [14]), the
ultrasonic echo technique could prove promising with respect to assessing
the integrity of gluelines, since it can provide information on the distribution
of interior defects. The ultrasonic echo technique is based on the reflection of
waves on material inhomogenities. The sonic wave passes through the element
and is reflected. The so-called back-wall echo is received by the sensors.
Structural irregularities produce a change in signal structure of the back wall
echo. The results are given as A- and B-scans, see Fig. 6. The A-scan shows the
transmission time and intensity of the pulse, while the B-scan is a
composition of various A-scans that are recorded with a defined distance
between them, see Fig 7. The B-scan is a 2D cross section through the
specimen and enables to identify a change in signal structure along the
measured axis. By combining several B-scans, it is possible to interpolate a
horizontal layer (C-scan) that gives 3D information about the internal structure.
181
The experimental set-up consists of a signal generator, a preamplifier, an amplifier,
a transducer and a data acquisition system. For the measurement, the sensor
is placed on a surface of the element. To limit signal retardation, a specific
coupling pressure or a coupler (e.g. water, oil, paste, rubber) has to be applied to
create complete contact between the transducers and the material surface. In some
cases, the surface of the member may need preparation as smoothing the surface
though planning or sanding. In [16] it was shown that the application of shear
waves and reflection measurements is mandatory if thin defects (like small cracks
or delami-nations) are to be assessed. Some of the abovementioned challenges
are overcome by using air-coupled ultrasonic technique [17] but its application is
still limited to laboratory conditions or production control and smaller samples.
Fig. 6 Illustration of A (left), B (middle) and C (right) scans, from [15]
Fig. 7 Result of B-scan (transversal waves) of glulam member with crack, from [14]
The technique has shown to have sensitivity to delamination [18], it is lowcost and portable. However, its application over larger areas is time-demanding.
Several factors can affect the transmission of stress waves in timber, equipment
application and the interpretation of results. Access to in-situ members may also limit
182
the use of some stress wave techniques. The ultrasonic echo technique allows the
direct localization of a reflector like a back wall or any inhomogenity like damage
in the structure. It is difficult, however, to locate the exact position of damage or
to distinguish between one large irregularity and many small ones. Nevertheless,
it can be assumed that a clear back wall echo shows that the specimen is free of
defects. Systematic measurements on sample specimens with artificial imperfections
are required to make qualitative conclusions from measuring results. An exact
classification of damage by analysing the signal structure is not yet possible.
Table 4 Ultrasonic Echo Technique - Evaluation.
Type
Stress Waves
Application
Rare (Phase III)
Extent
Local
Time / Cost
Medium / High
Validity of
results
limited (experience, Constraints
type of damage)
Accessibility; sensitivity to defects; interpretation experience-dependent; time-consuming
5 Methods to assess the strength of gluelines
5.1 Shear tests on core samples
Shear tests on circular core samples are used to derive the shear strength of glue
lines in glued-laminated timber elements. The method was developed for production
control [19], [20], [21] but is being applied to determine the shear strength of builtin glued timber members as well, although this approach is reason for discussion
[22], [23]. When extracting core samples in-situ, proper drilling and orientation is
important. The glue line needs to be in the centre of the specimen and perpendicular
to the drill axis. The necessary equipment includes an electrical drill for sample
extraction which is mounted on a supporting fixture. The equipment is temporarily
fixed onto the glued timber member (e.g. by screws) to guarantee a correct drilling
position and prevent lateral motion of the drill during extraction, see Fig. 9.
The core samples taken from a member give information about the specific
location from which they were taken. Since the quality of the glue line can vary
significantly within and between members, multiple samples must to be taken to be
representative for the structural member which they are taken from and to receive
global estimations of a members mechanical properties. Samples should be drilled
in zones featuring low stresses and/or in compression zones of critical locations,
e.g. close to bearings. Access to these locations of in situ members can be very
difficult. In some locations, the necessary attachment of the supporting fixture onto
the member is inhibited.
The drill bits need to be sharp to extract the shear core samples of
approximately 35 mm in diameter and up to 120 mm in length, see Fig. 8 and Fig.
10. Voids left by drilling should be fully plugged by glued-in wood cores to
restore stress-transfer, to prevent moisture and insect penetration and to restore
the appearance of the member. The samples should be labeled with member, date,
183
location, and other pertinent information and be stored and transported in
containers that provide adequate protection. The samples are usually tested in
shear, but density and moisture content can also be determined. Furthermore
samples can also be used for species and adhesive type identification by
performing microscopic and chemical investigation. For shear testing, the
specimens need to be flattened on two sides and the exact dimensions of the shear
plane need to be measured before testing. The shear cores may be divided into
sub-samples, featuring lengths of at least 50 mm, see Fig. 8 and Fig. 11. After
reaching constant moisture content, the specimens are placed into the space
between the jaws of a test machine with the glue line oriented parallel to the
loading direction, see Fig. 12. Proper alignment of the specimen in the fixtures is
crucial for accurate strength estimates (max. 1 mm deviation). However, the shear
stresses in typical shear test fixtures are not evenly distributed and no pure
condition of shear stress scan be created, rather a combination of shear and
normal compressive stresses. The shear strength is determined from the load upon
failure through the following equation
0.78
0.044 ∙
∙
Fig. 8 Geometry of drill core
Fig. 9 Extraction of core from glulam beam.
Fig. 10 Drill core in laboratory
184
Fig. 11 Specimens prepared from drill core Fig. 12 Shear testing of specimen
In addition to the shear strength, the percentage wood failure (PWF) has to be
determined after testing. This is a critical index to determine the quality of a bond
and is usually measured by trained personnel by visual examination of the failure
surface. In specimens with colorless glue, the surfaces are treated with chemicals
(first hydrochloric acid, then by phloroglucinol) to better distinguish regions
of wood failure from regions of glue-line failure [24]. The areas that turn red
indicate fiber failure, as the lignin changes color. To determine PWF, usually
two people independently evaluate the specimens. If their results deviate by more
than a certain percentage, they consult a third person. Nevertheless, the determined
PWF remains subjective, making it difficult to compare test results of different
studies. Image analysis technique could be an alternative to calculate the PWF
[25]. Depending on PWF, different requirements on the strength of glue lines
exist [21] [26]. A higher percentage of glue failure leads to higher
requirements on the reference strength to which the determined shear
strength is compared. If the determined shear strength reaches the reference
strength given in e.g. [21], the applicable design shear strength (listed in the
respective product or design standard) may be used. In no case should the
determined shear strength be used as reference shear strength for structural
calculations. If the reference values are not achieved, further investigations are
necessary. These include the maximum utilization of the member in shear and the
size of the area under high shear stresses.
Table 5 Shear tests on core samples - Evaluation.
Type
Mechanical; nonstationary, semi-destr.
Application
Phase II (meduim)
Extent
Local
Time / Cost
High / Medium
Constraints
Accessibility; local assessment; variation of
glueline quality; percentage wood failure;
semi-destructive (surfaces)
Validity of High (on a local level)
results
185
6 Conclusions
As for other timber properties to be determined through assessment methods it can
be stated that most assessment methods utilized today can give qualitative information
but only a few non- or semi-destructive methods can give quantifiable information.
Applicable methods to determine global strength parameters of built-in timber
elements are very scarse. An approach would be to correlate strength parameters
from measured stiffness parameters which are easier to derive. But the correlation
between stiffness parameters and strength parameters for common britle failure
types like tension perpendicular to grain and shear failure are too low, leaing a low
level of confidence. As a consequence, all data received requires a very careful
evaluation by an experienced engineer. Expert’s reports treating the structural
safety of a structure featuring gued timber elemnts will oftentimes be set up from a
standpoint which could be summarized as “safe on the best knwledge we have”.
References
[1]
Hetzer O. Patentschrift Nr. 197773 vom 22. Juni 1906: Gebogener Holz-Bauteil für vereinigte Dach-Pfosten und-Sparren. Kaiserliches Patentamt, Berlin. 1906
[2]
Steiger R, Köhler J. Development of new swiss standards for the assessment of existing
load-bearing structures. Paper 41-102-2. In: Proceedings of the 41st meeting of CIB W-18.
St. An-drews, Canada. 2008.
[3]
Dietsch P, Köhler J (eds). Assessment of timber structures. Shaker Publishing Company,
Aa-chen (Germany). ISBN 978-3-8322-9513-4. 2010.
[4]
Diamantidis D (ed). Probabilistic assessment of existing structures - a publication of the
joint committee on structural safety (JCSS). RILEM Publications S.A.R.L The publishing
Company of RILEM. 2001.
[5]
EN 314-1:1993. Plywood, Bonding Quality, Part 1 – Test methods. CEN, Brussels. 1993
[6]
EN 302-2:2013. Adhesives for load-bearing timber structures – Test methods – Part 1:
Determination of longitudinal tensile shear strength. CEN, Brussels. 2013
[7]
Görlacher R. Hölzerne Tragwerke Untersuchen und Beurteilen. Sonderforschungsbereich
315 „Erhalten historisch bedeutsamer Bauwerke“, Baugefüge, Konstruktionen, Werkstoffe.
Uni-versität Karlsruhe (TH). Verlag Ernst & Sohn, Berlin. 1996
[8]
Blaß H-J. Brüninghoff H. Kreuzinger H. Radovic B. Winter S. Guideline for a First Evaluation
of large-span Timber Structures. Council for Timber Technology, Wuppertal. 2006
[9]
SEI/ACSE 11-99. Guideline for Structural Condition Assessment of Existing Buildings.
ASCE Publications. ISBN 0-7844-0432-1. 2000
[10] Dietsch P. Winter S. Assessment of the Structural Reliability of all wide span
Timber Structures under the Responsibility of the City of Munich. 33rd IABSE Symposium
Proceedings. Bangkok, Thailand. September 9-11 2009
[11] Brüninghoff H. Effects of (seasonal) changes in Environmental Conditions on structural
Timber Elements. in: Dietsch P, Köhler J (eds.). Assessment of timber structures. Shaker
Publishing Company, Aachen (Germany). ISBN 978-3-8322-9513-4. 2010
[12] Bucur V. Ultrasonic Imaging. In: Nondestructive Characterization and Imaging of Wood.
Springer, Berlin. 2003. pp. 181–213.
186
[13] Kasal B. Lear G. Tannert T. Stress Waves. in: Kasal B, Tannert T. (eds): In Situ Assessment
of Structural Timber. Springer. ISBN 978-94-007-0559-3. 2011
[14] Hasenstab A. Integritätsprüfung von Holz mit dem zerstörungsfreien Ultraschallechoverfahren. Bundesanstalt für Materialforschung und –prüfung BAM. Dissertationsreihe Band
16. Berlin. 2006.
[15] Krautkraemer J, Krautkraemer H. Werkstoffprüfung mit Ultraschall. Springer, Berlin. 5th
edition. 1985
[16] Dill-Langer G. Aicher S. Non-destructive detection of glue line defects in glued laminated
timber. WCTE 2008 Conference Proceedings. Miyazaki, Japan. 2008
[17] Sanabria Martín S. J. Air-coupled ultrasound propagation and novel non-destructive bonding
quality assessment of timber composites. Dissertation. ETH Zurich. 2012 [18]
[18] Maeva E. Severina I. Bondarenko S. Chapman G. O’Neill B. Severin F. Maev R. G. Acoustical
methods for the investigation of adhesively bonded structures: A review. Canadian Journal
of Physics, Vol 82, No 12. pp. 981–1025. 2004.
[19] EN 392. Glued laminated timber - Shear test of glue lines. CEN, Brussels. 1995.
[20] Steiger R. Risi W. Gehri E. Quality control of glulam: shear tests of glue lines. Paper
40-12-7. Proceedings of the 40th meeting of CIB-W18. Bled, Slovenia. 2007.
[21] EN 14080. Timber structures – Glued laminated timber and glued solid timber
- Requirements. CEN, Brussels. 2013.
[22] Brüninghoff H. Shear tests on core samples - critical discussion on the evaluation
of results, in: Dietsch P, Köhler J (eds). Assessment of timber structures. Shaker
Publishing Company, Aachen (Germany). ISBN 978-3-8322-9513-4. 2010.
[23] Tannert T. Vallée T. Müller A. Critical review on the assessment of glulam structures
using shear core samples. Journal of Civil Structural Health Monitoring, Vol 2, No 1.
pp 65-72. Springer. 2012
[24] Künniger, T. Automatische Bestimmung des prozentualen Faserbruchanteils bei
der indust-riellen Klebfestigkeitsprüfung. FFWH Project report 2006.05. EMPA,
Dübendorf, Switzer-land. 2007.
[25] Yang Y, Gong M, Chui Y.H. A new image analysis algorithm for calculating percentage
wood failure. Holzforschung, Vol 62. 2008. pp. 248–251
[26] ASTM D 5266. Standard Practice for Estimating the Percentage of Wood Failure in Adhesive
Bonded Joints. ASTM International, Conshohocken PA, USA. 1999.
187
188
Mapping of cracks in glulam beams and
assessing the effect of environmental conditions
Philipp Dietsch
Dr.-Ing., Team Leader Timber Structures
Chair of Timber Structures and Building
Construction
Technische Universität München, Germany
1 Introduction
For the timber community, the step from solid timber elements elements to glued(laminated) timber
elements, started by Hetzer [1], represented a
t
significant technological progress, widening the range of application of timber
structures in the building sector. The results were larger and more variable
(e.g. curved) geometries and timber being increasingly used, (e.g. for largespan structures. Consequently, the importanec of assessing large timber
structures grew. naturally resulting in an increased interest of the professional
community in assessment methods for existing glulam structures. The
performance of glulam beams depends on the quality of the individual
laminations, the quality of the finger joints, the quality of the glue-lines and
the integrity of the cross-section which is made up of single lamellae which are
stacked and glued to cross-sections. The most frequent damage affecting the
integrity of the cross-section is cracking. Cracks are a form of stress relief. The
reason for the frequent occurrence of cracks in grain direction of timber elements
is the very low tensile strength perpendicular to the grain of wood. The reason
for tensile stresses perpendicular to the grain in glulam elements can be
manifold: external loading, internal stresses due to deviation forces in e.g.
curved members, uneven shrinkage of the cross-section due to changes in
moisture content (moisture gradients).
The objective of this chapter is to present feasible methods to map cracks,
to assess environmental conditions and to evaluate each of them with regard to
the following objectives, see also [2]:
 What can be determined / what cannot be determined?
 How accurate and precise are the results (e.g. degree and size of damage; local/
global results)?
 How complex and time consuming is its application (on-site, (non-) destructive)?
189
2 Approach to the mapping and assessment of cracks
The objective of any assessment of a structure is to detect, localise and document
damage, to determine the actual degree of damage, to identify the reasons
for damage and to define means to repair the damage. Typically the
assessment is divided into different phases, structured by the detailedness of the
investigations, see e.g. [2]. The number of phases necessary depends on the
level of doubt, the feasibility and simplicity of repair or strengthening, in
combination with economic considerations. The detection and assessment of
cracks is part of almost every assessment of timber structures, especially largespan glulam structures. While a partial detection of cracks is performed
during the first site-visit, a detailed investigation, including a complete
mapping of all cracks may be carried out during the detailed inspection. With
respect to the assessment of cracks, the major properties of interest are:
 Location, length and depth of crack
 Reasons for the crack (inevitable, remedial measures possible)
 Effect of crack on structural safety (stress transfer) of timber element
2.1 Manual/visual mapping of cracks
Fig. 1 Common equipment for the mapping
of cracks: magnifying glass; thickness gauge;
laser distance measuring device; folding rule
Fig. 2 Measurement of crack depth with
thickness gauge (0.1 mm)
Fig. 3 Indication of crack depth with folding
rule
Fig. 4 Magnifying glass and camera
190
The most common approach to detection and documentation of crack distribution
as well as measurement of crack dimensions is to use a thickness gauge (0,1 mm)
and a tape measure/folding rule, see Fig. 1 - Fig. 3. A magnifying glass can aid in
determining the age of a crack (dust, discoloration within crack) and potential
adhesion or cohesion problems if cracks appear mainly in the glueline, see Fig. 4
[3]. It is self-explanatory that the mapping and assessment of cracks has to be well
documented on the timber element and in the written documentation, including
clear indication of the location of the crack within the structure, see Fig. 5 and Fig.
6. The marking/mapping of crack width, depth and crack tips is important to analyse
possible changes in crack dimensions over time. This is essential for structures
which are subjected to seasonal environmental changes, leading to changing timber
moisture content which can have an influence on crack dimensions.
I
t = 60 mm
II
t = 45 mm
Numbering
Largest depth incl. location
Marking of crack ends
Fig. 5 Example of marking of cracks on structural timber element
Element / Axis
Cracks:
Pos:
i
di
View from
xi
di
ℓi
wi
N E S W
1
Moisture measurement: yes
no
2
Crack width
Crack length
Crack depth
Distance from e.g. bottom flange
Distance from e.g. support
Comments
Fig. 6 Example of written documentation of cracks
The determination of the crack dimensions aids to assess the remaining residual
cross section of structural elements. In this context the crack depth is of particular
importance since it indicates the residual cross-section to transfer stresses, see
Fig. 7. The crack depth should be measured at multiple locations along the crack. [4]
indicates possible space intervals for the measurement of crack depths. For longer
cracks, these intervals (measurement at each ¼ of the length) should be reduced.
Diverging cracks are problematic in that aspect since their real depth cannot be
accurately measured with a thickness gauge, see Fig. 8. A core sample can give
clearer information, see Fig. 9 and Fig. 10, nevertheless this remains a local and
destructive measure.
191
Thickness
Gauge
di
Remaining
cross-section
Remaining cross-section to transfer
shear stresses or tensile stresses
perpendicular to the grain
Remaining cross-section
from measurement
Fig. 7 Schematic of crack depths and
remaining cross-section to transfer stresses
Fig. 8 Difficulty to accurately measure depth
of diverging cracks with thickness gauge
Fig. 9 Indication of crack depth on a core sample,
here: crack in glueline
Fig. 10 Magnification of crack in a core sample, here: crack in glueline
Table 1 Manual mapping of cracks - Evaluation.
Type
Visual; on-site;
non-destructive
Application
Phase I (always)
Extent
Local
Time / Cost
Medium / Low
Constraints
Local and temporary assessment; depth of
diverging cracks cannot be examined
accurately; time-consuming
Validity of Limited (incertitude
results
about crack depth)
2.2 Alternative methods – potential for application
Potential options to manual assessment and mapping of cracks are optical devices
or laser scanning. Preliminary investigations have shown, however, that laser
scanning incorporates the same drawbacks as manual measurement with a
thickness gauge since it requires a straight crack of sufficient width [5]. The
light-section method constitutes another option. However, this method requires
adequate crack width which is not always given.
192
Another method to assess the location and dimension of cracks is the ultrasoundecho-technique whereby the air-coupled ultrasonic technique seems to be the most
promising [6] since it is low-cost and portable. Currently its application is still limited
to laboratory conditions or production control and smaller samples. Its application
over larger areas is time-demanding. The ultrasonic echo technique allows the
direct localization of a reflector like a back wall or any inhomogenity like damage
in the structure. It is difficult, however, to locate the exact position of damage or to
distinguish between one large irregularity and many small ones. For a description
of the ultrasound-echo-technique, the interested reader is referred to [3] (this
publication) and to [6]-[11].
Radiography (X-Ray) represents another option since it can deliver information
on the internal structure of members. Detection of cracks using radiography requires
that the crack be of adequate size, at least 2% of the member thickness [12], and
the crack being oriented parallel to the radiation beam. In combination with the
requirement to access the opposing sides of a member, these can constitute exclusion
criteria when assessing glulam elements. Safety issues and high initial costs of the
equipment constitute additional drawbacks of radiography. For a description of this
method, the interested reader is referred to [13]-[15].
2.3 Evaluation of the consequence of cracks
To evaluate the consequence of a crack on the structural integrity of the timber
element, information on the crack dimensions (length, width, and depth) and
especially the position within the structural element is essential, Fig. 11 and
Fig. 12.
h =hap
Tension perp.
ℓ
Shear
Shear


max


t,90,max

m

max
max
t,90
Fig. 11 Consequence of cracks - Schematic illustration of the distribution of bending stresses,
shear stresses and tensile stresses perpendicular to the grain in straight beam and curved beam,
from [16]
193
The measured crack dimensions and location are to be evaluated individually
for each structure. This includes the structural system, the relevance of the structural
element, building use and environmental boundary conditions to determine the
causes and possible consequences of cracks. Within the scope of structural boundary
conditions it should to be differentiated between cracks in areas of high shear and
areas of high tension perpendicular to the grain stresses, Fig. 11 and Fig. 12. Some
literature indicates permissible crack depths for such areas, e.g. [17]-[20].
ℓ
0
10
20
Bending stresses/
Bending capacity
σm,d [MN/m2]
fm,d [MN/m2]
0
1
2
Shear stress/
Shear capacity
τv,d [MN/m2]
fv,d [MN/m2]
1 t = 70
2 t = 85
3 t = 45
5
4 t = 50
t = 85
6 t = 65
7 t = 90
Illustration of
crack distribution
ℓ
0 20 40 60 80 100%
Explanation:
0 20 40 60 80 100%
0 20 40 60 80 100%
Percentage of member width (b = 160 mm)
Required width according to standards (max. permissible crack depth)
Required width considering distribution of shear stresses over member depth
Required width considering distribution of shear forces over member length
Remaining width respectively remaining crack depth
Fig. 12 Schematic of remaining member width in dependence of location and depth of cracks in
comparison to required width to transfer existing stresses, here shear stresses, from [21]/[22]
194
It is also relevant if the cracks appear predominantly or with a certain frequency
in the wood or in the glueline. A crack in the direct vicinity of the glueline is not
in all cases an indication of adhesion or cohesion problems. The wood material in
the vicinity of the gluelines is the weakest part of the lamella due to the cutting of
fibres during sawing. In addition, there is a discontinuity between the density of
the boards and also between the shrinkage movements of both sides. Often times
the cracks occur in the boundary layer next to the glue line but within the wood.
The crack should therefore be closely investigated before conclusions on the failure
mechanisms are drawn [23].
3 Assessing the effect of environmental conditions
3.1 Measurement of temperature and relative humidity
The measurement of temperature and relative humidity is part of every assessment
of a structure. Information on the macroclimate is – among other things - needed
for a comparison with the timber moisture content. The measurement of air
temperature and humidity should be realized in the vicinity of the structural
component by means of thermometer and hygrometers. The relative humidity is
commonly measured by capacitative hygrometers which measure the change in
the dielectric constant of a material between two condensator plates during water
adsorption. The drawback of measurements carried out during an assessment of a
structure is that these represent temporary measurements and do not allow for
extrapolation with respect to seasonal changes of macro climate. Information on
the latter is e.g. necessary to explain moisture gradients in the timber elements. In
these cases, the monitoring of environmental conditions, which can be coupled
with the monitoring of timber moisture content, represents a feasible solution [24].
3.2 Measurement of wood moisture content
The measurement of wood moisture content can be realized by several methods.
The traditional method of measuring moisture content is the gravimetric method –
i.e. oven drying of wood specimens. It necessitates the extraction of specimen
from the timber element. These are weighed before kiln-drying them at a
temperature of 103 °C േ 2 K until constant mass is achived. The moisture content
u is determined from the ration between the mass of water in the moist specimen
(mu-mdr) and the mass of the kiln-dried specimen (mdr). Investigations on the
moisture gradient in timber elements are possible but difficult since this
necessitates the extraction and segmentation of specimen without influencing the
moisture content. The method is standardized, see [25]. Since this method is a
destructive method and rather time-consuming, it is less suitable for in-situ
investigations. One common method to determine timber moisture content in situ
is the dielectric / capacitative method. Wood changes in proportion to its moisture
content, i.e. its dielectric constant increases with increasing moisture content. For
195
measurements, a condensator is placed on the wood surface, the wood acts as
dielectric, see Fig. 13. The meters do not penetrate the wood material, i.e. only
measurements of the average wood moisture content near the surface (< 35 mm)
are possible, measurement of a moisture gradient is impeded, see Fig. 15. The
method is standardized, see [26] and delivers an acceptable accuracy for wood
moisture contents from 2 % up to the fibre-saturation point [27], [28], although
temperature considerably affects the dielectric properties of wood [30].
Fig. 13 Commercially available device using the
capacitative method, from [24]
Fig. 14 Commercially available device using
the resistance method, from [24]
Fig. 15 Schematic of areas measured by capacitative method, from [24]
Fig. 16 Schematic of areas measured by resistance method using insulated and noninsulated electrodes, from [24]
The most commonly applied method to determine timber moisture content in situ
is the resistance method with hand-held moisture meters, see Fig. 14. Since water
has a much higher conductivity than wood, the electrical resistance decreases with
196
increasing moisture content [29]. In practice, two electrodes are rammed into the
wood or attached to the surface with a defined distance of usually 30 mm between
them and the resistance to electrical current is measured between them.
The electrodes should be teflon-insulated to measure moisture content in
clearly defined depths of the cross-section, see Fig. 16. This allows for
collecting moisture content readings at multiple depths (e.g. 10, 30 and 50 mm)
to evaluate moisture gradients. The method is standardized, see [31] and
delivers an acceptable accuracy of ± 1,0 % for wood moisture contents
between 6% and the fibre-saturation point (28 % - 30 %). Above the fibresaturation point, the accuracy decreases considerably. Below 6 %, the electrical
resistance reaches very high values which are challenging to measure. Another
factor influencing the electrical resistance of wood is the material
temperature, see Fig. 17. The resistance decreases with increasing
temperature, see e.g. [27], [28]. Moisture readings can also be affected by
species and direction of measurement (parallel or perpendicular to grain). If the
wood contains salts, terpens, oils or preservatives, the resistance-measurementmethod is not adequate and should be replaced by the distillation method [27].
Fig. 17 Influence of wood MC (x-axis) and temperature on electrical resistance of wood (y-axis) ,
from [28]
Fig. 18 Influence of wood moisture content on
select wood properties, from [27]
2.3 Evaluation of the effect of environmental conditions
Changes in wood moisture content lead to changes of virtually all physical and
mechanical properties (e.g. strength and stiffness properties) of wood, see Fig. 18.
The existence of high moisture content can initiate decay or growth of
fungi. Another effect of changes of the wood moisture content is the
associated shrinkage or swelling of the material. Shrinkage and swelling are
significantly more pronounced in radial and tangential direction than in
197
longitudinal direction. Since the outermost sections of the wood cross-sections
will adapt to the climatic conditions at first, the resulting moisture gradient and
the associated shrinkage or swelling will lead to equilibrium of internal
compression and tension perpendicular to the grain stresses in the crosssection. Although these stresses are partly reduced over time by relaxation
processes, an excess of the tension perpendicular to the grain strength results
in instant stress relief in form of cracks, see Fig. 19. In case of shrinkage of the
cross-section, the cracks will appear on the surface of the cross-section, in case of
swelling of the cross-section, the cracks can develop inside the cross-section.
The effect is more pronounced in larger cross-sections, for example large or
block-glued glulam members. When the free shrinkage or swelling of the
member is restrained, due to e.g. fasteners or reinforcing elements, the
equilibrium of internal stresses is suspended, resulting in larger stresses and
correspondingly larger, deeper cracks, see Fig. 20.
Moisture gradient
∆uoutside
∆uinside
Stresses perp. to grain
Initial MC t = 0
Moisture gradient 0 < t < ∞
σt,90
σc,90
Equilibrium MC t = ∞
ft,90
Fig. 19 Schematic of moisture gradient (left) due to changed environmental conditions and resulting stresses perpendicular to the grain due to shrinkage (right)
Moisture gradient
∆uoutside
∆uinside
Stresses perp. to grain
Initial MC t = 0
Moisture gradient 0 < t < ∞
Equilibrium MC t = ∞
σt,90
ft,90
Fig. 20 Schematic of moisture gradient (left) and resulting stresses perpendicular to the grain in
case of restrained free shrinkage (due to e.g. connections) of the timber beam (right)
198
Low or high moisture contents or severe changes of the same can sometimes
be attributed to local conditions (e.g. roof leakage) but in the majority of cases,
they can be explained by the climatic conditions, depending on the construction
type and use of the building, and seasonal variations of the building climate
[32]. Hereby it has to be differentiated between (see Fig. 21)
 changes of moisture content between the various phases of processing and shape
until the timber element is finally subjected to the environmental conditions
during operation of the building
 seasonal changes of moisture content due to seasonal changes of environmental
conditions during operation of the building
 changes in climatic conditions due to temporary interventions (e.g. renovations)
or changes of use (temporary or permanent) of the building
Fig. 21 Sketch of a possible „moisture chain“, i.e. exposure to moisture from the tree to gluedlaminated timber elements in the building
In closed, insulated and heated buildings (e.g. living spaces, office spaces,
gymnasiums, production and sales facilities), featuring constant but dry climate,
the most severe change of timber moisture content will mostly occur during the
first winter of operation, after assembly and closure of the building. Here, surface
treatment e.g. in the form of products which damp the moisture absorption and
release in the first years of operation of the building could be a means to counter
fast drying of newly installed elements. Strong but periodic changes of moisture
content can occur in buildings with seasonal change of use (e.g. ice-skating rinks)
or in buildings with a considerable influence of the outdoor climate on the indoor
climate (e.g. unheated and non-insulated buildings like riding rinks, stables,
warehouses). For this group, the application of insulation on the roof could help to
199
dampen the strong changes of indoor climate and correspondingly the timber
moisture gradients. In case of temporary interventions or changes of use, care
should be taken during such interventions to realize a dampened and controlled
change of moisture content (e.g. in form of evaporation basins or surface
treatment). For a more detailed description of these effects, the interested reader is
referred to [32].
References
[1]
Hetzer O., Patentschrift Nr. 197773 vom 22. Juni 1906: Gebogener Holz-Bauteil für
vereinigte Dach-Pfosten und-Sparren, Kaiserliches Patentamt, Berlin, 1906.
[2]
Dietsch P., Köhler J. (eds), Assessment of timber structures, Shaker Publishing
Company, Aachen (Germany), ISBN 978-3-8322-9513-4, 2010.
Dietsch P., Tannert T., “Assessing the integrity and strength of gluelines”. This
publication.
DIN 4074-1, Strength grading of wood – Part 1: Coniferous sawn timber, DIN, Berlin,
2008
[3]
[4]
[5]
Talke D., Capturing cracks in timber constructions with the David laserscanner,
Bachelor Thesis, Chair of Timber Structures and Building Construction, Technische
Universität München, 2011.
[6]
Sanabria Martín S. J., Air-coupled ultrasound propagation and novel non-destructive
bonding quality assessment of timber composites, Dissertation, ETH Zurich, 2012.
[7]
Bucur V., ”Ultrasonic Imaging”. In: Nondestructive Characterization and Imaging of
Wood. Springer, Berlin. 2003, pp. 181–213.
[8]
Kasal B., Lear G., Tannert T., „Stress Waves“, in: Kasal B., Tannert T. (eds): In Situ
Assessment of Structural Timber, Springer, ISBN 978-94-007-0559-3, 2011
[9] Hasenstab A., Integritätsprüfung von Holz mit dem zerstörungsfreien
Ultraschallechoverfahren, Bundesanstalt für Materialforschung und –prüfung BAM,
Dissertationsreihe Band 16, Berlin, 2006.
[10] Krautkraemer J., Krautkraemer H., Werkstoffprüfung mit Ultraschall, Springer, Berlin,
5th edition, 1985.
[11] Dill-Langer G., Aicher S., “Non-destructive detection of glue line defects in glued
laminated timber”, WCTE 2008 Conference Proceedings, Miyazaki, Japan, 2008.
[12] Raj B., Jayakumar T., Thavasimuthu M., Practical Non-Destructive Testing, 2nd ed.
Woodhead Publishing, Abington, UK, 2002.
[13] Kasal B., Lear G., Anthony R., “Radiography” in: Kasal B, Tannert T. (eds): In Situ
Assessment of Structural Timber, Springer, ISBN 978-94-007-0559-3, 2011.
[14] Bucar V., Nondestructive Characterization and Imaging of Wood, Springer, Berlin,
2003.
[15] Kasal B., Anthony R., “Advances in in-situ evaluation of timber structures”, Progress in
Structural Engineering and Materials, Vol. 6, No. 2, 2004, pp. 94-103.
[16] Dietsch P., Kreuzinger H., Winter S., “Design of shear reinforcement for timber beams”,
CIB-W18/ 46-7-9, Proceedings of the international council for research and innovation
in building and construction, Working commission W18 – timber structures, Meeting 46,
Vancouver, Canada, 2013.
200
[17] Frech P., „Beurteilungskriterien für Rißbildungen bei Bauholz im konstruktiven
Holzbau“, Bauen mit Holz, Vol. 89, No. 9, 1987, pp. 582-585.
[18] Erler K., Alte Holzbauwerke Beurteilen und Sanieren, 3.Auflage, Verlag Bauwesen,
Berlin, 2004.
[19] Radovic B., Wiegand T., „Oberflächenqualität von Brettschichtholz“, Bauen mit Holz,
Vol. 107, No. 7, 2005, pp. 33-38.
[20] Blaß H.-J., Brüninghoff H., Kreuzinger H., Radovic B., Winter S., Guideline for a First
Evaluation of large-span Timber Structures, Council for Timber Technology,
Wuppertal, 2006.
[21] Dietsch
P.,
Einsatz
und
Berechnung
von
Schubverstärkungen
für
Brettschichtholzbauteile, Dissertation, Technische Universität München, 2012.
[22] Dietsch P., & Schänzlin J., personal communication, 17.8.2010.
[23] Brüninghoff H., “Effects of (seasonal) changes in Environmental Conditions on
structural Timber Elements”, in: Dietsch P., Köhler J. (eds.): Assessment of timber
structures, Shaker Publishing Company, Aachen (Germany), ISBN 978-3-8322-9513-4,
2010.
[24] Dietsch P., Franke S., Franke B., Gamper A., “Methods to determine wood moisture
content and their applicability in monitoring concepts”, Journal of Civil Structural
Health Monitoring, Vol. 4, No. 3, 2014
[25] EN 13183-1:2002, Moisture content of a piece of sawn timber - Part 1: Determination
by oven dry method, European Committee for Standardization CEN, Brussels, Belgium,
2002.
[26] EN 13183-3:2005, Moisture content of a piece of sawn timber - Part 3: Estimation by
capacitance method, European Committee for Standardization CEN, Brussels, Belgium,
2005.
[27] Kollmann F., Coté W. A., Principles of Wood Science and Technology I: Solid Wood,
Springer, Berlin, 1968.
[28] Niemz P., Physik des Holzes und der Holzwerkstoffe, DRW-Verlag, LeinfeldenEchterdingen, 2003.
[29] Villari E., Annalen der Physik und Chemie - Untersuchungen über einige Eigenschaften
des mit seinen Fasern parallel oder transversal durchschnittenen Holzes, Bd. CXXXIII,
Leipzig, 1886.
[30] James W.L., “Fundamentals of hand held moisture meters: An outline”, Proceedings of
ASTM Hand-held Moisture Meter Workshop, Forest Products Society, Madison, WI,
1994, pp. 13–16.
[31] EN 13183-2:2002, Moisture content of a piece of sawn timber - Part 2: Estimation by
electrical resistance method, European Committee for Standardization CEN, Brussels,
Belgium, 2002.
[32] Dietsch P, Gamper A, Merk M, Winter S, “Monitoring building climate and timber
moisture gradient in large-span timber structures”, Journal of Civil Structural Health
Monitoring, Vol. 4, No. 3, 2014.
201
202
Estimation on-site of decay in timber structures
by means of penetration methods
Alessandra Gubana1), Ezio Giuriani2)
1)
DICA, University of Udine, Italy, [email protected]
2)
DICATA, University of Brescia, Italy, [email protected]
Summary
The problem of the evaluation on-site of decay in timber structures can be faced
from different points of view and reliable information are necessary to decide on
the structural adequacy, the possibility of preservation, the kind of the eventual
strengthening intervention or the need of substitution.
The assessment of in situ timber members is always firstly based on visual
inspection, the information so obtained can be deepen by several techniques
proposed and applied to locate and quantify deterioration, such as sonic and ultrasonic
techniques [1], application of electromagnetic waves [2], resistance drilling [3], screw
resistance [4], hardness test [5], pylodin [6] and penetration test [7]. Correlations
to predict mechanical properties are also disposable in literature, but not always
strong or reliable [8]. An organic and extensive state of the art is reported in [9].
Among all these different tests, those based on penetration into the cross
section of structural elements seem to be particular interesting for the structural
assessment, as the information obtained can be useful and sufficient to check
and verify the load bearing capacity of timber elements.
Penetration tests in general can give information about the layers decay, as the
response of wood to the penetration is different in relation to the density of the
material. The presence of decay, due to organic attacks or moisture variations or
any other causes, in generally alters this material property.
Moreover, when it is necessary to check the deformation or the bearing
capacity of a timber beam, it is important to know the cross section stiffness,
which is determined by the dimensions of the base and the depth of the beam. So
if a structural element is affected by external decay, and if its deepness is known,
it is possible to rely only on the dimensions of the no decayed internal part of
the cross section. As it is well known, the inertia dependence on the beam depth
size varies with the third power and an error in computing this dimension has a
relevant consequence on the final evaluation.
203
On the other hand, if the decay is not superficial, but only internal, the penetration up to the core or across the section can reveal inside decay which do not appear during the visual inspection.
All the penetration techniques tries to give this information, even if some are
more suitable for the structural assessment. The results of every test herein described is affected by the influence of the different angles of inclination of the pin
penetration.
The pylodin technique consists in the measure of the penetration of a thin rod
after only one constant energy blow: this means that if the external layer density is
high, the penetration length is small and the other way round. In this case the response is just a medium of the consistency of the different layers crossed by the
rod.
Screw-withdrawal tests resistance consists in the measure of the required force
to pull out a screw of 4 mm diameter and 18 mm length and it can be similarly put
in relation with the decay of the layers crossed by the screw itself. Different length
can be used to explore deeper inside. As in the pylodin test, the results correspond
to a medium values of the characteristics of the crossed layers.
Resistance drills use small diameter needle-like drills to bore into timber members and it measures the resistance the drill bit encounters as a function of the penetration depth. The results are given in terms of graphs, where peaks correspond to
higher resistance or density, while lowest points are associated with lower response and are index of possible decay. Experience is necessary to correctly interpreter the result data.
The pushing pin device [10] is based on the same principle, as a pin is gradually pushed into the timber perpendicular to the surface by rack and pinion gear,
driven by two opposite manual cracks moved by hands. The device is equipped
with a linear magnetic position sensor and a force sensor, which allow to measure
the progress of the force and the displacement of the pin.
The so called Wood Penetration test [7,8] can be regarded as an extension to
timber structures of the dynamic soil penetration test, as it is based on the insertion
into timber structural members of a steel graduated rod, which advances by means
of repeated constant energy blows transmitted by a rebound hammer. The test
makes possible to distinguish between different degrees of decay as a function of
the number of blows necessary for each centimetre layer penetration. This technique proved to be effective and reliable for investigating the extent and depth of
wood decay. The test was frequently proposed and adopted in Italy and several
ancient building timber structures were checked by means of this methodology.
The test is easy to perform and the output data, shown in terms of penetration histograms, which vary along the penetration depth (Fig.1), can be interpreted very
clearly by structural engineers: it is so possible to determine the depth of the decayed layers and consequently the reliable dimensions of the resisting timber sections.
204
(a)
(b)
Fig. 1. (a) State of decay of a timber truss arch; (b) Penetration test results (from [7]).
In conclusion all the penetration tests are useful to investigate the resistance
opposed by wood to the penetration of thin rods, and a simple and clear
information about the depth and the position of the decayed layers can be taken
from the output data or graphs in a less easy or easier way, depending on the
instruments involved.
205
In most cases for timber beam and timber elements structural assessment this
information is more important and decisive than the precise knowing of timber
strength resistance or elastic modulus.
Acknowledgments
The present paper is a result of the activities of COST FP1101 Action “Assessment,
Reinforcement and Monitoring of Timber Structures”.
References
[1]
R.J. Ross, R.F. Pellerin, Nondestructive testing for assessing wood members in structures:
a review. Gen.Tech.Rep. FPL-GTR-70, US Department of Agriculture, Madison, WI,
Forest Product Laboratory, 1991.
[2]
M. Lualdi, L.Zanzi, and L.Binda, Acquisition and processing requirements for high
quality 3D reconstruction from GPR investigations, Proceedings NDT-CE International
Conference, 2003.
[3]
F. Rinn, Resistographic inspection of building timber, Proceedings Pacific Timber
Engineering Conference, Gold Coast, Australia,1994.
[4]
Z. Cai, M.O. Hunt, R.J. Ross, L.A. Soltis, Screw withdrawal – A means to evaluate densities
of in-situ wood members, International Symposium on NDT of Wood, Richmond,
USA, (2002), pp.277-281.
[5]
J.G. Sunley, A comparison of the Janka and Monnin methods of testing the hardness of
timber and wood products, J.Inst. Wood Science, 14 (1965), pp.40-46.
[6]
P. Hoffmeyer, The Pilodyn Instrument as a Non-Destructive Tester of the Shock
Resistance of Wood, Proceedings of the 4th Symposium on the non-destructive testing of
wood, Wash-ington State University, 1978 , pp.47-66.
[7]
E. Giuriani, A. Gubana, A penetration test to evaluate wood decay and its application to
the Loggia monument, RILEM Materials and Structures, 1993, 26, 8-14.
[8]
E. Giuriani, A. Marini, S. Cominelli, A. Gubana, The Penetration Test to Evaluate Wood
Decay after 20 Years Timber Structure Assessment Experience, Proceedings of SHATIS
2013, Trento, 2013.
[9]
B. Kasal, T. Tannert, In Situ Assessment of Structural Timber, State of the Art Report of
the RILEM Technical Committee 215-AST, 2010.
[10] M.Kloiber, J.Tippner, V.Hermankova, J.Stainbruch, Comparison of results of measuring
by current NDT methods with results obtained through a new device for wood
mechanical resistance measuring, Proceedings of SAHC 2012, Wroclaw, pp.2035-2043.
206
PART IV
CASE STUDIES
207
PART IV- CASE STUDIES
208
Assessment through NDT of the state of timber
structures of the historic buildings of Catalonia
Marcel Vilches Casalsa, Carles Labèrnia Badiaa and Vladimir
Rodríguez Trujillob
a Catalan Institute of Wood (INCAFUST). Lab. and Technical unit of Lleida. Parc
Científic i Tecnològic. Edifici H2 • E-25003 Lleida, Spain, [email protected],
[email protected]
b Timber Building Consultant. C/Fontrodona 1, AT 2a E-08004 - Barcelona, Spain,
[email protected]
Abstract
The wood was the material most commonly used for the construction in cities
and towns of Catalonia in the past, for this reason, nowadays a lot of
diagnosis in timber structures are performed. The NDT, in the diagnosis of
old buildings, help to complete the inspections, therefore in some diagnosis we
have used NDT, like Microsecond timer or drilling resistance device, to improve
the assessment conditions of the structure. In order to NDT are reliable previous
studies of laboratory should be made, for getting warranty correlations. These
correlations presented by different authors have been used in some inspections of
old buildings in Catalonia, in which timber roofs and the floors of the buildings
are analyzed with NDT. These methods have been a complement for deciding the
strength class of the timber putted in the old buildings.
Keywords Nondestructive techniques, timber structures, diagnosis, old
buildings, Catalonia
Introduction
Historically, the timber has been the material most used for construction of the
resistant structures (roof, floors, …). For this reason, a lot of rehabilitations of catalan historic buildings are performed in timber structures, that can be degraded by
biotic pathologies. Nowadays, the development of the NDT for the evaluation of
properties of timber also are used in the inspection of the structures as a
complementary tool for deciding the strength class of timber.
The main advantage of the NDT is that they don't produce damage in the
examined sample and we can continue using the timber elements. The acoustic
methods are most used in the diagnosis of timber, permitting to predict the modulus
of elasticity with the transfer speed of the wave. The densitometers also are
209
employed as NDT for determining the density of timber, so this physic property
is a good indicator of the mechanic properties. In this case the estimations are in a
point and his information is complemented by other methods, as ultrasonic or
vibration techniques for calculating dynamic modulus of elasticity, which is a
good estimator of the real elasticity of the material.
Technicians and researchers, before using the NDT in the inspections, first
have performed several studies in laboratory for obtaining good correlations
between the values of the nondestructive methods and the properties of timber
(MOR, MOE and density). Normally, for the acoustic methods are correlated the
dynamic modulus of elasticity1 and the MOE2 and MOR3, but the Pearson correlation
coefficients do not exceed the 80% [2, 4, 5, 6]. Although it is difficult to find
higher correlations that give better warranties, it should be taken into account that
these techniques must be a help to the inspections of timber structures. The case of
the densitometers is similar, inasmuch the correlations between density and the
values obtained of the tool used can vary between the 50% and 70%.
Methodology
Species identification
The timber specie identification of the inspections is performed through a
macroscopic identification of a sample extracted in situ. The macroscopic description
includes the observation of several characteristics of the timber to the naked eye
or with help of a magnifier of 10x. The structure, size and form of the tissues are
different in almost all timber species, so that every timber specie can be avowed.
Xy-lotheque samples are used for the macroscopic description, with the three
cutting planes: transversal, radial and tangential.
Moisture content
The moisture content in the inspections is determinate with a xylohygrometer of
electrical resistance (FMW), that permits obtaining the moisture content of the
timber with a microprocessor and we can know the moisture quickly. This tool has
a sensor on top and measures the moisture content holding the sensor on the timber
element. The measuring range of the humidity is 2 to 30%, with a resolution of
0.1% and an accuracy of 0.5%. The measuring depth is 10 to 20 mm, adjustable to
intervals of 1 mm.
1
Edyn =ρ(v2 ) ρ=Density,v=wave speed
2
MOE=A+B·Edyn
3
MOR=A+B·Edyn
210
Determination of density
For the determination of density we used a drilling resistance device, model IML
Resi F300S. This tool is an electrical drill that is introduced, with a constant
speed, in the timber and it measures the degree of resistance offered. It is a good
technique for the inspection of service structures. The drill does holes of 3 mm of
diameter. Normally, it is used to verify the status of certain critical points, as the
inspection of hidden timber parts, which building are sensitive to the material loss,
thus decreasing the strength capacity.
Determination strength properties
Another tool used is the Microsecond Timer (FAKKOP), which is based on the
generation of the impact waves, with a frequency of resonance of 23 kHz. The
function of this tool is based on the higher absorption capacity of the impact
waves of the wood degraded and the wood healthy. The impact waves are generated
with the mechanical hammer impact on an electrical transmitter. The mean of
the wave speed is performed with a timer and an electronic receptor that detects
the transit time.
Case of study
Old Building at the Pyrenees
This building, where teaching activities were performed, has an old construction
and is distributed in ground floor, first floor, second floor and penthouse floor.
Structurally is built with walls of load and horizontal floors with timber beams
(Figure 1).
Fig. 1 Exterior building front view and beams of the second floor
211
In the inspection was observed that all timber of the structure was performed
with the same specie, for this reason a wood piece of a beam was extracted for the
specie identification. The identification showed that the genus of the specie was
Pinus sp., probably Pinus sylvestris. The moisture content was determined with a
contact hygrometer, obtaining humidity around 10% of the timber elements. So
that, the characteristic values of elasticity will be modified for the 12% of humidity.
The density, used for determining the dynamic modulus of elasticity, was
determined in laboratory with beams that were not to the inspection. We made
drilling resistance curves in areas free of defects, in eight beams of Scot pine,
with similar dimensions and characteristics to the beams of the inspected building.
We had these eight elements available in laboratory for an own characterization
project and we used them for this analysis because we did not make the drilling
resistance curves on the elements of the inspection in situ by unavailability of the
drilling resistance device. The moisture content of the eight beams in the moment
of drilling was approximately of 12% and the mean of coefficient of growth was
of 6 mm by ring in each beam. With mean values of the drilling resistance, of the
beams of laboratory, and the correlation obtained by Acuña et al. [1], for Pinus
sylvestris, a density of 411 kg/m3 was consider for the inspected timber (Table 1).
Table 1 Density values obtained with the mean values of the drilling resistance of different
timber elements of Pinus sylvestris
Timber elements
Mean values of Drilling resistance (%)
Density valuesa (kg/m3)
Pinus sylvestris 13
23,34
412,53
Pinus sylvestris 21
38,06
423,71
Pinus sylvestris 22
18,52
408,87
Pinus sylvestris 23
25,51
414,18
Pinus sylvestris 81
15,42
406,51
Pinus sylvestris 81b
12,24
404,10
Pinus sylvestris 82
25,75
414,36
Pinus sylvestris 83
14,25
405,62
a
Mean
411,24
ρ=394,797+0,7598·Drilling resistance Mean Value (r2 = 0,82) [1]
With the density value can be calculated the MOEdyn, but this density will not
be considered for the assignation of the strength class, as this property is not a
limiting factor because it has higher values than 400 kg/m3. For this reason, in
order to estimate the strength class of the timber inspected we decided to consider
only the MOE and the MOR obtained with impact waves.
The analyzed timber structure of Pinus sylvestris showed a good conservation,
which has given an excellent structural functionality. We only found an excessive
212
deformation in some beams of the first floor. This was originated by the prolonged
action of humidity of the water tubing of the top floor. The other beams near the
water tubings didn’t observe any degradation signs of the timber. In the penthouse
we didn’t find any kind of alteration.
Once the condition of the timber structure was valued positively we performed
the assignation of the strength class with the help of the time-of-flight of ultrasonic
waves values. Six pieces of timber were analyzed, two bottom chords, which
formed the trusses of roof, two beams of penthouse and two beams of the second
floor, and in each of them were measured three readings wave speed (Figure
2). The measures were performed on the central part of the timber pieces, with a
distance between sensors of approximately two and three meters (Table 2).
Fig. 2 Evaluation of the bottom chords of the trusses of roof with the el Microsecond timer
Table 2 Values of time transmission, the mean value of the time and the wave velocity in each
timber element analyzed
Element
Time 2
(µs)
Time 3
(µs)
Mean value Distance WaveVelocity
(µs)
(cm)
(m/s)
Bottom chord
Penthouse 480
DEF
482
480
480,60
215
4.473,57
Bottom chord
Penthouse 434
LLLM
435
435
434,60
217
4.993,10
Beam 4
Penthouse 315
314
315
314,60
171
5.435,47
Beam 5
Penthouse 331
332
331
331,30
177
5.342,59
Beam 11
Second
634
630
632
632,00
290
4.588,61
Beam 13
Second
664
663
662
663,00
283
4.268,48
Floor
Time 1
(µs)
213
These values of time-of-flight of ultrasonic waves were for determining the dynamic modulus of elasticity and with this value we can allocate the strength class
using the equations of bending strength (MOR) and modulus of elasticity (MOE),
published by Íñiguez (2007) for Scots pine. The bottom chord obtained a characteristic value of the bending of 23,49 N/mm2, of 35,14 N/mm2 for the beams of
penthouse and of 20,83 N/mm2 for the beams of the second floor, assigning them
respectively a class C22, C35 and C20. In return, the elasticity values limited the
strength class of the inspected elements, obtaining characteristic values of 7.404
N/mm2 for the bottom chords, 9.403 N/mm2 for the beams of penthouse and 6.541
N/mm2 for the beams of the second floor, assigning them respectively a class C14,
C18 and C14 (Table 3).
Table 3 Wave speed, mean modulus and bending values of each element. Characteristics values
of Mean modulus and bending. Strength class assigned
MOEdyna
(N/mm2)
MOEb
(N/mm2)
MORc
(N/mm2)
Bottom chord
Penthouse 8.230,09
DEF
6.655,74
23,14
Bottom chord
Penthouse 10.252,63 8.151,82
LLLM
30,02
Beam 4
Penthouse 12.149,83 9.555,19
36,47
Beam 5
Penthouse 11.738,13 9.250,65
35,07
Beam 11
Second
8.658,79
6.972,85
24,60
Beam 13
Second
7.492,75
6.110,33
20,64
Element
a
b
Floor
MOEd
(N/mm2)
MORe
(N/mm2)
Strength
class
7.403,78
23,49
C14
9.402,92
35,14
C18
6.541,59
20,83
C14
Edyn =ρ v2 ρ=411kg/m3 ,v=wave speed
MOE=A+B·Edyn =579,5+0,7548·Edyn (r2 = 0,74) [4, 5]
c
MOR=A+B·Edyn =-4,84+0,0034·Edyn (r2 = 0,60) [4, 5]
Mean values
e
5th percentile values
d
We didn't use the visual grading for allocating the strength class because the
timber elements of the penthouse were hidden by the ceiling and we couldn't see
the singularities of timber. The beams of the first and second floor were painted
and we neither could see the defects of timber.
214
Conclusions
Nowadays the speed of the wave transmission is, together with visual classification,
the NDT technique more used. The disadvantage of these methods are the low
correlations, because they don’t fulfill with the expectative. The values of the
elasticity and the resistance prediction with the nondestructive variables are
between 40% and 70% of the mechanical properties; however these techniques
are a good complement for the rehabilitation of buildings with timber structure,
to estimate a strength class of the timber elements.
The drilling resistance device is a good tool for predicting biotic degradations
in hidden areas and the difficult access areas, as the timber beams ends embedded
in masonry walls. This tool also serves like estimator of the timber density, using
the mean value of the drilling resistance curve as indicator [3, 6]. The results
obtained by the authors reinforce the utility of this tool in the inspections, but it
is necessary to use different NDT techniques for getting an adequate diagnostic.
Bibliography
[1]
Acuña, L.; Basterra, L.A.; Casado, Mª. M.; López, G.; Ramón-Cueto, G.; Relea, E.; Martínez,
C.; González, A. (2011). Aplicación del resistógrafo a la obtención de la densidad y la
dife-renciación de especies de madera. Materiales de construcción. Vol. 61, 303, 451-464
[2]
Arriaga, F; Íñiguez, G; Esteban, M; Bobadilla, I. (2009) Proposal of a Methodology for
the Assessment of Existing Timber Structures in Spain. Proceedings of the 16th
International Symposium on Nondestructive Testing of Wood (pp. 145-151)
[3]
Casado, M; Pinazo, O; Martínez, C; Vegas, F; Pando, V; Acuña, L; Relea, E. (2005)
Deter-minación de la capacidad resistente mediante métodos no destructivos. Aplicación
en vigue-tas de forjado de un edificio singular. Actas IV congreso forestal español. 4CFE05[523]
[4]
Esteban Herrero, M., Arriaga Martitegui, F., Bobadilla Maldonado, I., Íñiguez González,
G. y García Lantarón, H. (2009) Análisis y consolidación de la estructura de madera del
edificio antiguo del aserradero de Valsaín, Segovia. Actas V congreso forestal español.
5CFE01-[614]
[5]
Iñiguez, G. (2007). Clasificación mediante técnicas no destructivas y evaluación de las
pro-piedades mecánicas de la madera aserrada de coníferas de gran escuadría para uso
estructural. Tesis Doctoral. UPM. Escuela técnica superior de ingenieros de montes
[6]
Ramón, G; Basterra, A; Casado, M; Acuña, L. (2005) Analysis of the structural timber
diagnosis techniques at the cultural patrimony with an architectural project orientation. I
Jornadas de investigación en construcción. Tomo 1
215
216
SDT methods as part of a conservation process
Dulce Franco Henriques1*, André Santiago Neves1
1
Civil Engineering Departament, ISEL - Instituto Superior de Engenharia de Lisboa, Portugal.
* corresponding author: [email protected]
Abstract
This paper refers to the assessment on site by SDT (semi-destructive testing)
methods of the consolidation efficiency of a conservation method developed by
[1]. This is a solution for improve the physic and mechanical characteristics of
wood moderately degraded by fungi, avoiding its substitution and contributing to
the conservation of building heritage. The decay level and the evaluation of the
proposed solution efficiency were checked on site by SDT methods of
drill resistance and of penetration resistance. The objective was to
assess the consolidation efficiency on site by SDT. The technique
involves traditional methodologies used in the conservation of wooden cultural
heritage area applied to wooden elements in buildings. This study set out
to ascertain on site the mechanical performance of scots pine (Pinus
Sylvestris L.) wood degraded by fungi after treatment with a biocide product
followed by consolidation through impregnation with a polymeric product.
The SDT methods used showed good sensitivity to the presence of the products
and could evaluate their effectiveness.
Introduction
This conservation process was developed in order to help keeping the timber in
buildings that lies slightly deteriorated but yet has strength capacity determined on
site. The process of consolidating degraded timber by impregnation consists of forcing
a specific fluid material into it, which when hardened will restore its integrity and
improve the physical and mechanical characteristics [2-4].
In addition to strengthening the wood structure, the materials used may also
provide some protection against biological pests [5,6]. However it was found that
synthetic consolidants, including epoxies, do not significantly increase the resistance
of wood against fungi [5-7]. So the application of biocides before or with the
consolidant became necessary and the use of boron was a possibility because of its
good fungicide and insecticide properties [8,9].
The pair of products which showed the best mechanical results by laboratory
tests was an epoxy consolidant and a boron-based biocide [10-12].
217
After that it was necessary to evaluate the efficiency of the method in wooden elements in buildings [10]. This evaluation was done using the SDT methods of drill
resistance and of penetration resistance due to the superficiality of the consolidation (up to 15 mm) and to the necessity of some tools sensitive to that.
The drill resistance device has been seen as a reasonable tool to evaluate
mechanical characteristics of timber, even though that was not its original objective
[14], nor its most usual application field. Due the sensibility of the tool several
authors have recently been evaluating wood properties like the density of some
species with the drill resistance equipment in laboratory conditions to estimate this
characteristics in timber applied on site [10,14-17]. The mechanical strength and
modulus of elasticity have also been correlated with the drill resistance results
[14-16]. The penetration resistance technique is also applied to evaluate the surface
physic-mechanical characteristics of timber as well as the level of damage of the
timber, which depends on its surface hardness and density [10,15,17].
Case study
This case study presents the in situ experimental conservation process performed
on six moderately degraded sections of structural timber elements from a XIX
century palace: three floor beams, a staircase, a wall and a roof beam. In every
case the degradation was located in a small part of the element with an extent
generally lower than 80 cm. The laboratorial development of the process considered
it applicable to wood degraded by fungi, with mass losses lower than 20%. This
value is regarded as a limit for the intervention success. For higher values of
mass loss, laboratory tests indicated that the resistance was lower than the
minimum structural class of Scots Pine [1,18].
The evaluation of the local timber elements condition was made with a drill
resistance device and with an penetration resistance device before and after
the treatment and consolidation application. The main goals of the use of
nondestructive techniques in situ were the evaluation of the local degradation
condition [19,20] as well as a physical / mechanical efficiency evaluation of the
applied treatment and consolidation products.
Materials and Methodology
Treatment and Consolidation
The process consists on an initial application of a boron-based aqueous biocide
(Bora-care® - Nisus Corporation). On a second stage, and after the stabilization of
218
the water content, the two component consolidation product is applied (EPO 155 ®
+ K 156® - C.T.S. Srl.). This pair of products was selected among others with a
similar individual efficiency, because they proved to have the best joint mechanical efficiency [1,11,12].
Both products were applied on the timber elements by brush (Fig.1). It is also
possible to resort to injections whenever it is justifiable [10].
In each of six timber element to be analyzed, three analysis zones were determined: in sound wood (zone C), to control, moderately degraded by fungi (Zone
A) and heavily degraded by fungi (Zone B), for comparison
Fig. 1. Application of consolidant a) Wall; b) Staircase; c) Pavement beam 1 [18]
Evaluation tools
Aiming to identify local timber elements condition, drill resistance and penetration
resistance equipments were used. To assess the increase of mechanical strength
after applying the proposed method, the devices were used before and after
the treatment and consolidation application. As verified by [1], the penetration
depth of consolidation product is in the order of 10-15 mm. Therefore,
the drill resistance measure was determined in an extension of 10 mm for
nonstructural elements and 15mm for structural elements, always excluding the
initial 2 mm, considered as a perturbation zone.
It was verified that drill resistance equipment allowed the identification of
strength increase through the density profiles analysis [18].
By a comparative analysis penetration resistance equipment also allows the
identification of strength increases, identified by a reduction of the penetration
depth.
Results
The results are presented through medium values comparing the situation before
and after treatment and consolidation in each timber element moderately degraded
by fungi.
219
Comparing C and A zones is possible to get a rough idea of timber decay
levels through both drill resistance and penetration resistance methods, but was no
possible to quantify it because of the great variability of results obtained [10,18].
Also with C and B zones, used as references, it is possible to see the significance
of the resistance gain after the application of the process. In fact, as shown in table
1, the values of zone A after the treatment and consolidation, in a great number of
the cases, approached the registered value for sound wood (zone C).
Table 1 - Summary results of tests performed
Timber Properties – mean values
Element
Zone A
moderately degraded
Initial
Wall
Staircase
Timber
beam 1
Timber
beam 2
Timber
beam 3
Roof
beam
Moisture content (%)
Drill resistance
Penetration resistance (mm)
Drill resistance
Drill resistance
drill resistance
penetration resistance (mm)
Drill resistance
Drill resistance
5
8,5
10
6,4
14,1
7
11,8
14,8
9
10,8
19,7
9,5
9,1
24,4
9,5
9,4
19,3
Treated and
consolidated
5
9,2
10
8,7
11,4
8
15,6
13,9
9
11,1
18,2
9
12,3
19,1
9
11,1
18,4
Property
variation
0%
8%
0%
36%
-19%
14%
32%
-6%
0%
3%
-8%
-5%
35%
-22%
-5%
18%
-5%
Zone B
(heavy
degr.)
Initial
Zone C
(sound)
2
6,3
7
1,0
17,8
3
7,0
21,3
8
8,2
37,3
8
5,98
38,1
7
5,8
34,3
12
13,1
12,3
9
11,6
11,4
7
12,8
16,2
8,5
11,7
16,6
9
10,3
18,0
Initial
The penetration resistance and the drill resistance measurements are affected by
moisture content and to obtain correlations of those values with wood properties
one must adjust the measurement to a common wood moisture content, such as
12% [21]. For the purpose of this study the goal was to equilibrate de MC for the
initial conditions to be possible to do the value comparison in almost the same
conditions.
The rise in the drill resistance measurements and the decrease in penetration
depth, point to the increase in mechanical strength after the application of the
conservation process proposed.
Unable to perform the penetration resistance test on the wall because this element
was not supported, which influences the results. For other elements a tendency
220
to decrease of the penetration resistance penetration was generally verified,
which indicates a gain of mechanical resistance.
Final comments
The SDT methods of drill resistance and penetration resistance were very
important auxiliary instruments to assess and confirm on site the results obtained
in laboratory for the treatment and consolidation process [18].
Conclusions of the case study:
• Efficiency of consolidation by impregnation as a technique of local mechanical
strength increase of wood moderately degraded by fungi, applied after the
biocide treatment;
• SDT drill resistance and penetration resistance as methods suitable for assessment
of consolidation efficiency levels;
• Good sensitivity of the methods to the superficiality of the conservation
method (up to 15 mm);
• Suitability application of that technique on site.
Acknowledgments
The authors would like to thank the Foundation for Science and Technology (FCT) for
the financial support in the CONSERV-TIMBER project (ref. EXPL/ECM-COM/
0664/2012). Also would like to thank East-Banc, owner of the building for their willingness.
References
[1]
Henriques, M. D., 2011. Treatment and consolidation of pine wood degraded by fungi
in structural elements of ancient buildings. (In portuguese). PhD Thesis. Lisboa: Instituto
Superior Técnico.
[2]
E. Schaffer, 1074, Consolidation of painted wooden artefacts, Stud. in Conserv. 19, 212-220.
[3]
J.R. Loferski, 2001, Technologies for wood preservation in historic preservation, Archives
and Museum Informatics 13, 273-290.
[4]
S.M. Nakhla, 1986, comparative study of resins for the consolidation of wooden
objects, Studies in Conservation, 31, 38-44.
[5]
A. Unger, W. Unger, 1994,Conservation of wooden cultural property, Proceedings of The
International Research Group on Wood Preservation, IRG/WP 94-30038, Bali, Indonesia.
[6]
A. Unger, A.P. Schniewind, W. Unger, 2001, Conservation of wood artefacts - A handbook,
Springer-Verlag, Germany.
[7]
J.-D. Gu, 2003, Microbiological deterioration and degradation of synthetic polymeric
materials: recent research advances, International Biodeterioration & Biodegradation, 52,
221
69-91.
[8]
F.C. Jorge, L. Nunes, C. Botelho, 2004, Boron in wood preservation: problems, challenges
and proposed solutions. An overview on recent research. Journal of the Faculty
Science Technology, University Fernando Pessoa 1, 3-15.
[9]
S.N. Kartal, W.J. Hwang, K. Shinoda, Y. Imamura, 2006, Laboratory evaluation of boroncontaining quaternary ammonia compound, didecyl dimethyl ammonium tetrafluoroborate
(DBF) for control of decay and termite attack and fungal staining of wood. Holz Roh
– Werkstoff, 64, 62–67.
[10] Henriques D.F., Nunes L., Brito J. de, 2011 “An experimental approach to the treatment
and consolidation of degraded timber elements from a XIX century building”. SHATIS –
International Conference on Structural Health Assessment of Timber Structures. LNEC,
Lisbon, Portugal, pp. 47-48;
[11] Henriques, D., Brito, J., Duarte, S. & Nunes, L., 2013a. Consolidating preservativetreated wood: Combined mechanical performance of boron and polymeric products in
wood degraded by Coniophora puteana. Journal of Cultural Heritage, 15, 10-17.
[12] Henriques, D., Nunes, L. & Brito, J., 2013b. Mechanical evaluation of timber conservation
processes by bending tests. Advanced Materials Research, 778, 612-619.
[13] Rinn, F. 1994, ‘One minute pole inspection with RESISTOGRAPH micro drillings’, in
Intern. Conf. on wood poles and piles, March 21-23, Fort Collins, Colorado, USA.
[14] Feio, A., 2006. Inspection and Diagnosis of Historical Timber Structures: NDT Correlations
and Structural Behavior. PhD Thesis. Braga: Universidade do Minho.
[15] Palaia, L., Monfort, J., Sánchez, R., Gil, L., Álvarez, A., López, V., Tormo, S., Pérez,
C., Navarro, P., 2008, ‘Assessment of timber structures in service by using combined
methods of non-destruct testing together with traditional ones’, 9th Inter. Conf. on NDT of
art, Israel.
[16] Branco, J.M., Piazza, M., Cruz, P.J.S. 2010, ‘Structural analysis of two King-post timber
trusses: Non-destructive evaluation and load-carrying tests’, Con. Build. Mat, 24, 371–383.
[17] Iniguez, G., Arriaga, F., Esteban, M., Bobadilla, I.González, C., Martinez, R., 2010. Insitu non-destructive density estimation for the assessment of existing timber structures.
WCTE - Word Conference on Timber Engineering.
[18] Neves, A. S., 2013. Assessment, treatment and consolidation of timber in ancient buildings
(In portuguese). Master Thesis. Lisboa: Instituto Superior de Engenharia de Lisboa.
[19] Tannert, T., Kasal, B. & Anthony, R., 2010. RILEM TC 215 In-situ assessment of structural
timber: Report on activities and application of assessment methods. World Conference
on Timber Engineering. Trentino, Itália: Curran Associates, Inc.
[20] Machado, J., 2013. In situ Evaluation of the Reference Properties of Structural
Timber Menbers. Use of Available Tools and Information. Structural Health Assessment
of Timber Structures. Trento, Itália: Trans Tech Publications Ltd, Switzerland, 137- 144.
[21] Kasal, B., Anthony, R., 2004, Advances in situ evaluation of timber structures, Prog. Struct.
Eng. Mat.6, 94–103.
222
Advanced model based assessment of existing
timber structures
Kiril Gramatikov1), Toni Arangjelovski2) and Marija Docevska3)
1)
Prof. Ph.D., University "St.Cyril and Methodius"-Skopje
2)
Assistant Prof. Ph.D., University "St.Cyril and Methodius"-Skopje
3)
Ms.C. student, grad.civ.eng., University "St. Cyril and Methodius"-Skopje
Introduction
Assessment of the timber structures in R.Macedonia follows the procedure where
proof load test was used to verify the available stress state and serviceability state
of deflections. This procedure is given in the national standard MKS U.C9.300/1984
[1]. The knowledge and experience in the field of assessment of new and existing
or damaged structures, gained from the past years gives us enough data to
harmonize behaviour of existing structures according to the standard EN1995-1 as
a special case of assessment known as design code assessment.
For example the standard ISO13822 [2] gives instructions that structures
designed and constructed based on earlier codes may be considered safe to resist
loads if there is no evidence of significant damage, distress or deterioration,
planned maintenance ensures sufficient durability and structure has demonstrated
satisfactory performance for sufficiently long period of time for displacements.
Despite this fact, for two case studies we have performed structural reliability
verification, on the basis of the proof load test results. For this purpose we have
used Level 3: Advanced Model Based Assessment given in Annex A of SAMCO
F08a Guideline for the Assessment of Existing Structures [3]. Structural reliability
analysis was performed for:
• Glued laminated timbers Structure, Cambered beam span of L=24.00m BrewerySkopje
• Glued Laminated Pedestrian Timber Bridge in Struga, Two-hinged arch, span
of L=26.00m plus two simple supported beams span of L=12.5m each.
For these structures design documents and data from serviceability test after
observed damage and after construction by proof load test were available [4, 5 and
6].
The semi-probabilistic approach was used based on limit state principle to
define partial safety factors. The targeted reliability level for the ultimate limit state
223
and serviceability limit state was defined using the table C.1: Target reliability indices for assessment of existing structures [3].
For ultimate limit state reliability index β was set to β=4.27, for design working
life (e.g.50 years) and for high consequences and probability of failure P f=10-5.
For the irreversible serviceability limit state reliability index β=1.5 for the remaining working life and probability of failure Pf=10-1. These values are the same as
given in the Eurocode EN1990.
Reliability verification was done using the Reliability Based Code Calibration
program "Code Call" developed by Joint Committee on Structural Safety [7, 8].
Case study: Cambered beam L=24.90m, Brewery Skopje
First structure analyzed in this study is a roof glued laminated timber structure
constructed for the Coca-Cola section of Brewery-Skopje, shown in figure 1. The
structure is consist of cambered glued laminated timber beams, span of 24.00m,
with curved intrados in the middle of the span, 6° slope on the extrados and 4.5°
slope on the intrados.
Fig. 1. Industrial building of Coca Cola section, Brewery Skopje
One year after the construction, cracks parallel to the grain direction were appeared at the middle of the height where cambered beam is joined with the steel
bracing near the supports. To obtain the accurate state of the stresses and deflec224
tions, of damaged cambered beam, proof loading test was performed. The structure was loaded in ten different phases, but in this paper we have analyzed only
the ninth phase, in which the test load simulates approximately 90% of the full
value of snow load.
The reliability index β was calculated for different values of partial safety factor for the material γM (1.0; 1.10; 1.20; 1.30). The results are given in Figure 2. Reliability index β for γM=1.20 and ratio between variable and total stresses α=0.74
(for this case) is 4.71. It means that probability of failure for β=4.71 is Pf=10-6.
Fig. 2. Reliability index β for different values of γM as a function of stresses ratio α
Case study: Pedestrian Bridge over Crn Drim River, Struga
As a second case study in this paper, assessment of existing glued laminated
pedestrian bridge over Crn Drim River in Struga was analyzed (Figure 3).
The bridge is consist of two parallel main beams which are composed as twohinged arch beams with span of 25.70m and two single supported beams with
span of 12.50m on both sides of the arch as access ramps.
Proof load test was performed for variable load of 5KN/m2 which corresponds
to design variable load. The test load was set in three unsymmetrical and four
symmetrical phases of loading. We were considering only the two phases for
which we obtaied maximum measured deflections.
The reliability index β was determined for the coefficient α=0.84 according to
the measured deflections from the permanent and variable load. For this value of
α, reliability index β=1.50 was obtained as shown on the figure 4. For this value
of β, the probability of failure Pf is 10-1.
225
Fig. 3. Two hinged arch pedestrian bridge over Crn Drim river in Struga
Fig. 4. Reliability index β for irreversible serviceability limit state of deflection for different values of α
Conclusion
From the results of reliability verification analysis, for the two cases of structures,
following issues can be concluded:
226
• For the two type of structure, the special case of assessment new code assessment
EN1995 can be applied using Level 3: Advanced Model Based Assessment.
• Analysis of ultimate limit state for the cambered beam L=24.00m, for the ratio
of stresses α=0.74, reliability index β was calculated β=4.27 for value of partial
factor for the materials γm=1.2 and for the loads γG=1.35 and γQ=1.5.
• Proving irreversible serviceability state of deflections in the case of the
pedestrian bridge, two-hinged arch, span of L=26.00m, reliability index was
obtained β=1.5, for the given value of de-flections according to serviceability
test after construction.
• Structures designed and constructed by the old codes, which during the
serviceability period have satisfactory behavior also fulfill the design criteria
according to Eurocode EN1995.
References
[1]
MKS U.C9.300/1984-Design of glued laminated timber structures, Technical requirements
[2]
SAMCO Final Report 2006, F08a Guideline for the Assessment of Existing Structures,
Dr.W.Rucker, F.Hille, R.Rohrmann, Federal Institute of Materials Research and Testing,
Division VII.2 Buildings and Structures, Unter den Eichen 87, 12205 Berlin, Germany.
[3]
ISO 13822 "Bases for design of structures - Assessment of existing structures", 2001
[4]
K. Gramatikov, "Report on initial analysis for the state of stresses and deflections in the
roof structure for the Coca-Cola section", Brewery Skopje, 1996
[5]
Kiril Gramatikov, Experimental Testing and Assessment of the Bearing Capacity of GluLam Timber Structure at Skopje Brewery”, Report Faculty of Civil Engineering, 2002
[6]
S.Atanasovski, K.Gramatikov "Assessment of the bearing capacity of glulam timber bridge
in Struga", Report Faculty of Civil Engineering Faculty 2003
[7]
Joint Committee on Structural Safety, Probabilistic Model Code, 12th draft version 2000,
in-ternet version http://www.jcss.ethz.ch, 2014
[8]
Joint committee on Structural Safety, CodeCal, Reliability Based Code Calibration
program updated 12 June 2013, internet version: http://www.jcss.ethz.ch, 2014
227

State of the Art Report COMBINED USE OF NDT

Transkript

Podobné dokumenty

Prospěch ETP pro průmysl a vlastníky lesů – případ Německa

Wood Anatomy of Akaniaceae and