European Journal of Wood and Wood Products (2021) 79:335–347

https://doi.org/10.1007/s00107-020-01644-4

ORIGINAL ARTICLE

Nondestructive assessment of cross‑laminated timber using

non‑contact transverse vibration and ultrasonic testing
Lu Zhang1 · Alan Tiemann2 · Tonghao Zhang2 · Tom Gauthier3 · Kevin Hsu3 · Mustafa Mahamid2 ·
P. K. Moniruzzaman3 · Didem Ozevin2

Received: 11 April 2020 / Accepted: 26 November 2020 / Published online: 1 January 2021
© The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature 2021

Abstract
Cross-laminated timber (CLT) is a new generation of engineered wood product developed initially in Europe and has been
used in North America for various applications. CLT is composed of at least three layers of lumber boards stacked crosswise
in orthogonally alternating orientation to the adjacent layers. CLT has been gaining popularity and increasing application
areas in the construction industry, such as building systems, mats for construction and temporary road applications as well
as temporary bridges. It is especially identified as the prevailing material in modular construction for the future building
industry. However, CLT as a relatively new structural material has limited studies related to their durability and long-term
performance. Moreover, nondestructive evaluation methods to understand their mechanical behavior after exposure to harsh
environment are lacking. This paper contributes to the development of a rapid nondestructive evaluation method to assess
the mechanical properties using transverse vibration method and ultrasonic testing. The free vibration response triggered by
the impulse excitation is captured using contact accelerometer and non-contact laser Doppler vibrometer (LDV). The LDV
measurement is proven to respond similar to conventional accelerometers but more convenient as a rapid assessment method.
While the longitudinal wave speed is measured by ultrasonic testing (UT) in a through-transmission mode, the modulus of
elasticity was calculated in terms of measured wave speed. By investigating the different structural states and dimensions of
CLT samples, the laboratory and the manufacturing facility, the influences of aging on effective dynamic modulus of elasticity
(MOE) and damping are shown. The numerical and experimental results indicate the increase in moisture content decreases
the fundamental frequency. Therefore, moisture content should be measured in order to examine the coupled effects of mois-
ture content and aging on the fundamental frequency. The damping ratio extracted using half-power bandwidth method is
shown as an additional information related to the moisture content. The non-contact transverse vibration method and UT are
shown as an efficient nondestructive evaluation method for assessing the CLT structural state right after the manufacturing
as a quality assurance method or after their service in field as a diagnostic tool.

1 Introduction high-rise building with wood is still challenging, some

prominent examples of wood structures can be found in the
Wood is one of the common structural materials used by the world. For example, T3 office building in Minneapolis has
construction industry in North America. Nowadays, in the seven stories, Mjøstårnet in Norway is an 18-story wood
US, about 87% of the country’s multi-family construction building and also the largest wood structure in the world,
is constructed using wood (Slowey 2016). Though building and the Brock Common dormitory at the University of Brit-
ish Columbia is 18 stories with 17 stories of CLT over a
* Lu Zhang first story of concrete. Tacoma Dome in Washington state
[email protected] is the largest wooden dome in the world. Moreover, many
wood high-rise buildings are under construction (Parajuli
1
College of Civil Engineering and Architecture, Guilin et al. 2018). Comparing with the other common construction
University of Technology, Guilin, China
materials such as concrete and steel, CLT has advantages
2
Department of Civil and Materials Engineering, University such as high dimension stability, strength, and stiffness, cost
of Illinois at Chicago, Chicago, IL, USA
and energy efficiency, renewability, high strength to weight
3
Research and Development Department, Sterling, Phoenix, ratio, sustainability, good thermal and sound insulating
IL, USA

13
Vol.:(0123456789)
336 European Journal of Wood and Wood Products (2021) 79:335–347

properties and aesthetics. However, wood has other chal- to capture the response. The experimental results showed
lenges including vulnerability to water, fire and termites and that the damping was not sensitive to the boundary condi-
biodegradability. Wood is a porous material and can absorb tion, but the frequency change was highly correlated to the
or loose a significant amount of water. The moisture content junction quality. Labonnote and Malo (2010) identified the
can directly affect the density, which may lead to the degra- relationship between vibration properties and stiffness of
dation of mechanical strength. To improve the mechanical connections between the CLT panel by measuring the vibra-
performance of wood, new engineered wood products have tion properties numerically and experimentally. Ross and
been developed, such as plywood, densified wood, fiber- his team (Ross 2008; França et al. 2018) took advantages
board, particle board and cross-laminated timber (CLT). of transverse vibration to develop a method of assessing the
Herein, CLT has been widely used in Europe and gaining lumber properties. The stiffness and strength were estimated
in popularity in the US due to its high strength, dimensional according to the vibration behavior. Existing literature shows
stability and rigidity relative to other wood products. the potential of vibration method for diagnosing the residual
The mechanical properties of CLT have been studied strength of wood. The current studies are based on using
by destructive tests. For instance, Song and Hong (2016) contact accelerometers to monitor the acceleration response
evaluated the bonding strength of CLT by combining the of the sample upon receiving an applied load (Chui 1991;
soaking and boiling delamination test with block shear test. Haines et al. 1996; Labonnote et al. 2013). Attaching accel-
The results show that temperature and humidity can impact erometers on the surface of CLT with high moisture content
the bonding quality of CLT. Wang et al. (2018) assessed the is a challenging problem. Thus, it is more convenient to use
bond quality and duration of CLT in terms of block shear non-contact sensor for the rapid on-site testing and assess-
and delamination tests. They found that the adhesive type ment. Using a laser Doppler vibrometer (LDV) as opposed
could significantly affect the CLT performance. Jacquier and to accelerometers has been shown to have advantages in the
Girhammar (2015) investigated the performance of shear quality of the results (Rossi et al. 2002). Another distinct
connection area in the glulam-CLT cassette floor with a four- benefit of the LDV is faster setup than accelerometers as
point bending test. The bending stiffness and load resistance accelerometers need to be securely attached on the surface.
were evaluated. Cao et al. (2020) studied the strength of In this paper, CLT samples with different structural states
CLT-concrete connection by using the cyclic load test. and dimensions are tested using transverse vibration method
With those destructive tests, the mechanical properties employing LDV as the receiving sensor. Beam-like and
can be directly evaluated with high accuracy; however, the plate-like CLT samples are tested with different moisture
experimental configuration is complicated and costly. In contents, age and damage levels in the laboratory and in the
particular, the destructive tests are not suitable for on-site manufacturing site, respectively. Comparing the vibration
assessment, express examination or quality assurance in the analysis results between samples of different states shows
manufacturing process. As the mechanical properties of CLT the correlation among moisture content, the modulus of
are time-dependent and vary depending on its environmen- elasticity, fundamental frequency and damping. Ultrasonic
tal exposure, finding a reliable nondestructive evaluation testing as an alternative method can also be implemented
(NDE) method to understand the residual strength of CLT to obtain the dynamic modulus of elasticity of individual
is needed. NDE allows measuring the structural properties layer. Following this introduction, the vibration method and
rapidly without damaging the sample. In the literature, the theory used in this study, description of wood sample and
vibration analysis is utilized to assess the mechanical perfor- experimental setup, results and discussion, conclusions and
mance (Ross et al. 1991). Morandi et al. (2020) conducted an future work are presented.
experimental study on the diffuseness of the vibration field.
Through measuring the spatial distribution of the vibration
mode under impulse response, the non-diffusive frequency 2 Analytical and numerical models
was obtained. The correlation between wood species, stiff-
ness and modal density was obtained from the experimen- 2.1 Vibration equations for beam‑like and plate‑like
tal results. Quang Mai et al. (2018) evaluated the natural CLT samples
frequency of hybrid CLT-concrete floor specimen. They
found that the portion of the concrete could highly impact The planar dimensions of CLT samples vary depending on
on the fundamental frequency of the hybrid system. The their applications. Depending on the width to length ratio,
dynamic properties are good indicators of the wood qual- it can be categorized as beam-like. However, when width
ity. Morandi et al. (2018) identified a discontinuity at junc- increases relative to length, structural behavior switches
tion in the CLT wall system by characterizing the vibration to a plate-like. For the beam-like behavior, the relation-
reduction indices. The electrodynamic shaker was used for ship between dynamic modulus of elasticity, Ed , and the
generating the vibration, and the accelerometers were used

13
European Journal of Wood and Wood Products (2021) 79:335–347 337

fundamental frequency (the first mode) is given by ASTM y or m by n, which depend on the coordinate system. Equa-
D6874-12 (ASTM D6874-12 2012) as tions (2) and (3) are rewritten as Eq. (4) to obtain the rela-
tionship between dynamic MOE, Ed , and the fundamental
f1 2 Wl3 frequency, f11 as:
Ed = (1)
Kd Ig
2
48a4 𝜌f11 (1 − 𝜈 2 )
where f1 is the fundamental frequency of vibration, W is the Ed = ( )2 (4)
weight of the specimen, l is the clear span length between 𝜋 2 h3 (m − 1)4 + 6h3 ba (1 − 𝜈)(m − 1)2
supports, Kd is a constant equal to 2.47 for simply supported
case, I is the moment of inertia, and g is the acceleration Using Eqs. (1) and (4), dynamic modulus of elasticity
due to gravity, which is equal to 9.81 m/s2. The fundamen- of CLT samples depending on width to length ratios can
tal frequency of vibration is measured experimentally by be calculated by non-contact transverse vibration method.
introducing an impulse load at the center of the sample to
excite the vibration modes. The time history signal of sensor 2.2 Numerical model
is the summation of transient response and free vibration
response, which starts when the transient response ends. The correlation of frequency and dynamic MOE has been
The fundamental frequency of vibration is measured in the illustrated analytically in Sect. 2.1, which do not consider
free vibration regime. As the other variables of Eq. (1) are moisture content. However, moisture content can influence
known, dynamic modulus of elasticity can be measured dynamic MOE and density (Wang 2008; Llana et al. 2018).
experimentally. A parametric study is conducted by a numerical model to
For the plate-like geometries, the vibration modes are show the influence of moisture content on the fundamen-
plotted in two-dimensional coordinates. The Rayleigh tal frequency. The dimensions and boundary conditions of
method is used, and the relationship between natural fre- numerical CLT model are similar to the laboratory setup.
quencies and dynamic modulus of elasticity can be derived The numerical model is built using COMSOL Multiphysics
using the equation below (Leissa 1970): software, see in Fig. 2. Pinned and roller boundary condi-
{ }
𝜋4 × D
( )4 ( )2
a a (2)
𝜔2mn = (2𝜋fmn )2 = 4 G4x + G4y +2 [𝜈Hx Hy + (1 − 𝜈)Jx Jy ]
a ×𝜌 b b

Ed h3 tions are defined. Three layers with different orientations are

D= (3)
12 × (1 − 𝜈 2 )

where Gx , Gy , Hx , Hy , Jx , and Jy are the frequency coeffi-

cients, which depend on the boundary conditions; 𝜈 is the
Poisson’s ratio; a and b are the lengths of two edges; h is
the thickness; 𝜌 is the density. 𝜔mn is the angular frequency;
herein 𝜔mn = 2𝜋fmn , fmn is the natural frequency; m and n
refer to the mode numbers. In this study, two opposite sides
are simply supported; and the other sides are free, see in
Fig. 1. Therefore, G is m − 1, H and J are (m − 1)2 for the
simple supported boundary condition. For the free sides,
G and H are equal to 0, and J is 3∕𝜋 3 . The quantities of Fig. 2  Numerical model showing dimensions, layers, boundary con-
Gx , Gy , Hx , Hy , Jx , and Jy can be obtained by replacing x by ditions and excitation location

Fig. 1  Boundary conditions of

tested samples. The dashed lines
indicate the simply supported
boundaries

13
338 European Journal of Wood and Wood Products (2021) 79:335–347

Table 1  Materials properties Property Longitudinal layer Transversal layer

of different layers used in the
numerical model (Aydin and Moisture content (MC) (%) 12–18
Ozveren 2019)
Elastic modulus (MPa) − 210.22 × MC + 13,468 − 31.892 × MC + 892.6
Density (kg/m3) 3 × MC + 500
Poisson’s ratio 0.34

(a) (b) (c)

Fig. 3  Numerical results a mode shape of fundamental frequency, b the results of frequency sweep result, and c the relationship between mois-
ture content versus fundamental frequency

modeled: the transversal layer is in the middle, and the layers

on the top and bottom are defined as longitudinal layer. The
modulus of elasticity and density are defined as a function of
moisture content (Aydin and Ozveren 2019) as summarized
in Table 1. To show the correlation between moisture con-
tent and dynamic properties, eigenfrequency and frequency
response analyses are conducted with four different mois-
ture contents: 12%, 14%, 16% and 18%. The mode shape
of fundamental frequency at 46 Hz is shown in Fig. 3a. A
frequency sweep in the range of 0–100 Hz is defined at the
perturbation location of the frequency response analysis.
Figure 3b shows that the maximum response coincides with
the fundamental frequency of 46 Hz. With the increase in
moisture content, the natural frequency decreases as shown
in Fig. 3c. In general, the mechanical properties are influ-
enced by moisture content, which can also cause a shift in Fig. 4  CLT plate, beam and layer configuration
the fundamental frequency.

Fig. 4, and the plate samples were tested at the manufactur-

3 Descriptions of CLT samples ing facility. Four categories of samples with different aging
and experiment setup conditions based on the visual observation were prepared,
which are denoted as: new, used good, used moderate and
3.1 CLT samples dead poor. The CLT beams and plates were assumed to have
the same condition because of similar aging time and ser-
The CLT beams and plates were used in this study. The CLT vice condition for each category. To assure the repeatability,
samples were made by stacking three layers crosswise at 90° four beam samples and three plate samples were prepared
by a special adhesive. The species of wood is southern yel- for each category. The surface conditions are shown in
low pine (Pinus taeda). The experiments were conducted at Fig. 5. With the increase in aging, more wood-decay fun-
both the laboratory and the manufacturing site. The beam gus is observed on the surface with high moisture content.
samples were prepared for the laboratory testing see in The dimensions and weights of the beam and plate samples

13
European Journal of Wood and Wood Products (2021) 79:335–347 339

Fig. 5  Surface conditions with

different aging levels

Table 2  Summary of beam and plate samples

Samplea Beam Plate
b
Dimensions (m) Date: 2019/5 Date: 2019/10 Date: 2019/12 Dimensions (m)b Date: 2019/12
L W T Weight (kg) Weight (kg) Weight (kg) L W T Weight (kg)

N1 2.20 0.46 0.10 64.80 64.60 63.60 4.15 2.31 0.10 526.17
N2 2.21 0.42 0.10 58.80 58.20 55.80 4.15 2.31 0.10 517.09
N3 2.20 0.46 0.10 61.60 60.40 59.20 4.15 2.31 0.10 533.42
N4 2.22 0.40 0.10 55.60 55.20 54.20
UG1 2.36 0.46 0.10 69.40 65.40 63.20 4.13 2.31 0.10 687.65
UG2 2.35 0.46 0.10 72.20 66.20 62.40 4.13 2.31 0.10 714.86
UG3 2.34 0.46 0.10 68.40 63.80 62.60 4.13 2.36 0.10 670.41
UG4 2.34 0.46 0.10 69.20 61.00 58.40
UM1 2.35 0.46 0.10 71.60 59.60 53.80 4.20 2.36 0.10 659.52
UM2 2.36 0.46 0.10 81.20 62.40 57.80 4.13 2.31 0.10 653.17
UM3 2.36 0.46 0.10 71.80 59.40 53.80 4.12 2.30 0.10 649.54
UM4 2.36 0.46 0.10 75.20 61.60 57.40
DP1 2.13 0.46 0.10 77.40 58.20 51.20 4.24 2.39 0.10 850.03
DP2 2.13 0.46 0.10 83.20 57.20 52.00 4.20 2.35 0.10 673.13
DP3 2.13 0.46 0.10 79.00 58.40 56.60 4.21 2.37 0.10 782.90
DP4 2.13 0.46 0.11 85.00 62.20 56.80
a
N: new; UG: used good; UM: used moderate; DP: dead poor
b
L: length; W: width; T: thickness

are summarized in Table 2. The beam samples were placed 3.2 The experimental setup
in the laboratory environment with the relative humidity
of 50% and the temperature of 20 °C. The environmental Both CLT beams and plates were placed on simple sup-
parameters in the laboratory were kept constant. Each CLT ports, which can be kept at a fixed distance such that all
sample was weighted right before or after testing. The beams CLT samples would have the same clear span of 2.0 m. The
in the laboratory were tested at three different times to study experimental setup is illustrated in Fig. 6. In the laboratory
the effect of drying on weight and fundamental frequency. test, with the beam in place, one accelerometer was attached
using hot glue in the middle of the span. Directed at the
center of the beam, adjacent to the middle accelerometer,
the PDV-100 manufactured by Polytec was mounted on a

13
340 European Journal of Wood and Wood Products (2021) 79:335–347

Fig. 6  Experimental setups at a UIC laboratory, b manufacturing facility, and c schematic of sensor connections to oscilloscope

tripod which pointed the laser orthogonal to the surface, influenced by different surface conditions and environmental
see in Fig. 6a. To generate free vibration, an external force exposures.
excitation was generated by a hammer manufactured by Three weight measurements were taken since the beam
PCB Piezotronics Inc. with the sensitivity of 2.2 V×10−2/ samples were brought to the laboratory in May 2019. The
kgf. The hammer is equipped with force sensor, which is weight changes of CLT beam samples are measured. The
connected to an oscilloscope channel to measure the force corresponding density of each sample is calculated and illus-
versus time plot, and obtain the initiation of free vibration trated in Fig. 7a. The weight changes are mainly attributed
response. With the accelerometer associated with the signal to water loss in this case. With the increase in the aging state
conditioner, LDV, and hammer connected to the MSO2014 more water loss is observed due to porous condition usually
oscilloscope, the velocity and acceleration responses due to accompanied by the poor surface. Additionally, the density
hammer impact were recorded for further analyses. Similar does not change with time and drying rate is less than 5%
experimental setup was used in the manufacturing facility. see in Fig. 7c, which means the moisture content is stable.
However, to ensure the LDV is seated on a fixed position, To show the ability of conserving water of CLT, the drying
the tripod was placed on the stationary stand, see in Fig. 6b. rate is calculated and plotted in Fig. 7c. The drying rate of
The sampling frequency as 1302 Hz was used for the data first 120 days is higher than that from the following 45 days.
acquisition. The measurements from each sample were The water loss is faster when the moisture content is higher,
repeated three times in order to ensure repeatability. All of and the drying rate is highly correlated to the aging state.
the recorded signals were post processed using MATLAB.
4.2 The comparison of contact accelerometer
and LDV measurements
4 Experimental results and discussions
In general, contact-type accelerometer is the most common
4.1 The moisture contents of CLT beams sensor used to collect vibration data; at the same time, the LDV
as a non-contact sensor is a non-intrusive approach to measure
In this section, the beam samples were taken as examples the vibration response. The vibration data was recorded using
because they were preserved in a constant humidity and accelerometer and LDV simultaneously to show that LDV is
temperature environment. The moisture content mainly a good alternative to test CLT samples. Taking beam sample
controls the development of fungi on wood, which can be N4 as an example, the frequency spectra are calculated for
considered as one of the most dominant factors of wood the signals when the force vibration ends, shown with dashed
decay. As shown in Fig. 5, more fungus is observed with the lines on the signals in Fig. 8. The voltage values of acceler-
increase in aging. In turn, the ability of conserving water is ometer and LDV indicate acceleration response and velocity
response, respectively. The fast Fourier transform (FFT) is

13
European Journal of Wood and Wood Products (2021) 79:335–347 341

(a) (b)

Fig. 7  Influence of moisture content, a the corresponding density change, and b the drying rate of CLT beam samples

Fig. 8  Comparison of accelerometer and LDV responses, a time history signals, b the frequency spectra of free vibration response, and c the
windows of forced vibration and free vibration

13
342 European Journal of Wood and Wood Products (2021) 79:335–347

applied to the free vibration regions. The fundamental trans- where fB= frequency intersecting at a ∕ 2, highest value,
√
max
verse vibration frequency is measured through the first peak √
fA= frequency intersecting at amax ∕ 2 , lowest value, amax =
signal in the frequency spectrum of free vibration response.
The comparison shows that the LDV is able to identify the peak amplitude at fr . The variables are shown in Fig. 9.
peak frequency, which corresponds to the first vibration mode.
The results obtained from LDV are presented in the remaining 4.4 Condition evaluation of CLT samples
of the paper.
Dynamic modulus of elasticity values are calculated using
4.3 Damping calculation Eqs. (1) and (4). A good relationship between mechanical
properties and dynamic MOE has been identified in the liter-
Damping ratio is calculated using two common methods as ature (Ross 2008). It is aimed to develop a calibration curve
illustrated in Fig. 9: (i) general decrement method, and (ii) between MOE and the structural state of CLT to determine
half-power bandwidth method. For method (i), the peak ampli- the remaining service life. The results from laboratory and
tudes are measured at certain cycles. The damping ratio is manufacturing site are shown in Fig. 10. For the beam sam-
calculated in terms of decay factor as: ples, two measurements were taken in 120 days and 165 days
( ) as discussed above. In general, the dynamic MOE values
1 Ui from October and December measurements agree with
𝜉= ln (5)
2×𝜋×j Ui+j each other due to comparable moisture contents. The over-
all trends of beams and plates are similar: the fundamental
where 𝜉 is damping ratio, Ui and Ui+j are the magnitudes frequency and the corresponding dynamic MOE decrease
of cycle i and cycle i + j , j is number of cycles between with the increase in aging. Some variations in the samples
two amplitudes, shown in Fig. 9. The half-power bandwidth with similar states are observed, which can be attributed to
method is obtained from the frequency spectrum and calcu- differences in manufacturing process and exposure to dif-
lated using the following equation: ferent environment.
f B − fA The damping ratio represents the energy dissipation due to
𝜉= (6) the internal friction caused by dynamic stresses. The literature
2fr
shows that the moisture content in the wood can influence the
damping ratio, which means that the damping ratio measure-
ment can also be used as a quality indicator (Ouis 2003). The

Fig. 9  Methods to calculate the damping ratio as general decrement and half-power bandwidth methods

13
European Journal of Wood and Wood Products (2021) 79:335–347 343

(a) (b)

(c) (d)

Fig. 10  Frequency measurement (a, b), and dynamic MOE measurements (c, d) from beam samples and plate samples, respectively

(a) (b)

(c) (d)

Fig. 11  Damping measurement: a, b general decrement method; c, d half-power bandwidth method from beam samples and plate samples,
respectively

13
344 European Journal of Wood and Wood Products (2021) 79:335–347

damping ratios from two methods are compared and shown in where Ei is the modulus of elasticity of each layer,
Fig. 11. Half-power bandwidth method provides more consist- E1 = E3 = 11.03 GPa, and E2 = 1.25 GPa (Mahamid and
ent results, see in Fig. 11c, d; while more flucturations were Torra-Bilal 2019); bi and hi are the width and thickness of
observed in the results from general decrement method, see in each layer, respectively; Ai is the cross-sectional area of each
Fig. 11a, b. In general, the samples with higher moisture con- layer; zi is the distance from the centroid of each layer to the
tents (with some exceptions) have relatively higher damping neutral axis. In this case, we assume each layer has the same
ratios, see the comparison October and December in Fig. 11a, width and thickness.
c. Water trapped in wood may increase the dissipation of Based on Eq. (7) and geometric information in Table 2,
energy between particles, which causes higher damping ratio. the effective bending stiffness of each beam sample for the
If each CLT sample is periodically tested with the transverse pristine state is calculated, see in Fig. 13. The dynamic bend-
vibration method, the fundamental frequency and the damping ing stiffness is measured experimentally according to Eq. (1)
ratio can be tracked, and decay in strength with time and usage as:
can be quantified.
f1 2 Wl3
Ed I = (8)
4.5 Comparison with bending modulus of elasticity Kd g

The effective bending stiffness can be considered as the

The section and material properties should be defined as the
baseline condition after fabrication. It is based on static
first step of CLT design. The effective bending stiffness can be
modulus of elasticity. To compare the effective bending
calculated using shear analogy method (Mahamid and Torra-
stiffness, the measured Ed is converted into static modulus
Bilal 2019). The calculation diagram is shown in Fig. 12: in
of elasticity using the empirical relationship provided in the
layers 1 and 3, modulus of elasticity parallel to fiber orienta-
literature (Liu et al. 2006):
tion denoted as longitudinal Young’s modulus (­EL) is con-
siderably higher than the tangential modulus ­(ET) in layer 2. Es = 0.628 × Ed + 0.847 (9)
Therefore, the effective bending stiffness is calculated based
on the equation below: The unit in Eq. (9) is GPa. The comparison between static
and dynamic modulus of elasticity is shown in Fig. 13a. The
n n
∑ h3i ∑ comparison of effective bending stiffness shows that new
EIeff = Ei bi + Ei Ai z2i (7) and used good samples preserve their original stiffness,
i
12 i=1
whereas used moderate and dead poor samples have decay
in stiffness, see in Fig. 13b. In addition, the effective bending

Fig. 12  Diagram of calculation

of effective bending stiffness of
a 3-layer CLT

(a) (b)

Fig. 13  Comparison of a dynamic and static modulus of elasticity; and b corrected analytical effective bending stiffness and experimental results

13
European Journal of Wood and Wood Products (2021) 79:335–347 345

stiffness can be calculated by the shear analogy method. The taken on the layer 2 in which the moisture condition is
calculation is shown as Eq. (7). To consider the environ- constant as compared to layers exposed to environment.
mental factor, the wet service factor, CM should be applied The UT measurement is implemented using the function
to calculating the effective bending stiffness. If the moisture generator (WaveGen) and built-in DAQ system manufac-
content is larger than 16%, CM = 0.833; the rest condition, tured by MISTRAS. The testing setup is shown in Fig. 14.
CM = 1 (American Wood Coucil 2012). Usually, the stable Two R6 sensors are used as transmitter and receiver. A
moisture content of the CLT is about 13% (Schmidt and 7-cycle tone burst signal with 70 kHz frequency is used
Riggio 2019), and the density of new sample group does not as the excitation signal. The time of flight (TOF) is deter-
change with time, see in Fig. 7. In this case, it is assumed mined using the threshold-based method as the first wave
that the samples N1 to N4 have a stable MC, and the cor- arrival with amplitude greater than 40 dB, see Fig. 14 on
responding MC is 13%. Therefore, the corresponding MC an example waveform obtained from the sample N2. For
of each sample is calculated, accordingly: all of the samples each sample, three points are selected to ensure the repeat-
were brought to the laboratory in May, which can be treated ability of measurement. The dynamic modulus of elastic-
as initial condition with highest MC. With the water loss, ity for the second layer is calculated using the following
the MC of all samples tends to be stable. The comparison equation (Gray et al. 2008):
of effective bending stiffness is summarized in Fig. 13b.
Therefore, a reduction factor in effective bending stiffness Ed = 𝜈 2 𝜌 (10)
can be defined based on the result of non-contact transverse
Here, 𝜐 is the wave speed, m∕s ; 𝜌 is the density, kg∕m3;
vibration method.
Ed is dynamic MOE and can be converted into static MOE
using Eq. (9). The static MOE is summarized in Fig. 15. For
4.6 Comparison of modulus of elasticity using UM1, no signal could pass through the width due to the pres-
ultrasonics and vibration methods ence of through-crack. While, for the remaining samples, the
MOE values of new and used good samples are higher than
Ultrasonic testing (UT) is an alternative method to vibra-
those of used moderate and dead poor samples. The trends
tion method to obtain dynamic modulus of elasticity indi-
are similar to the static MOE obtained from LDV measure-
rectly. Instead of testing entire structure, a layer can be
ment. The results are compared in Fig. 16. LDV measures
tested to obtain mechanical properties. To identify the
the effective property as all three layers contributing to the
influence of the material decay, the UT measurement is

Fig. 14  Setup of UT measurement to measure the velocity of elastic wave. The grey line added to the output signal indicates the time of flight

Fig. 15  Static MOE from UT

measurement

13
346 European Journal of Wood and Wood Products (2021) 79:335–347

additional information related to the property of individual

layer, which can be used as a quantitative assessment per
layer other than the effectivity property obtained through
transverse vibration method.

Acknowledgements This work was performed with support in part

by Sterling Lumber. The support from the sponsoring organization is
gratefully acknowledged. The authors would like to thank Professor
Tom Royston for lending the LDV system for testing. Any opinions,
findings, and conclusions or recommendations expressed in this paper
are those of the authors and do not necessarily reflect the views of the
organizations acknowledged above.

Fig. 16  Comparison of static MOE obtained from UT and LDV References

measurements
American Wood Council (2012) NDS national design specification for
wood construction: with commentary, 2012th edn. The American
measurement. while UT measures only the property of inter- Wood Council (AWC), Leesburg
mediate layer in the direction of fiber. Therefore, MOE of ASTM D6874-12 (2012) Standard test methods for nondestructive
evaluation of wood-based flexural members using transverse
UT is higher than MOE of LDV while they indicate similar vibration. ASTM International, West Conshohocken, pp 1–8. https​
trends. The intermediate layer has lower degradation than ://doi.org/10.1520/D6874​-12.2
top and bottom layers as expected. Aydin TY, Ozveren A (2019) Effects of moisture content on elastic
constants of fir wood. Eur J Wood Prod 77:63–70. https​://doi.
org/10.1007/s0010​7-018-1363-3
Cao J, Xiong H, Wang Z, Chen J (2020) Experimental investigation
5 Conclusion and numerical analysis for concrete-CLT connections. Constr
Build Mater 236:117533. https​://doi.org/10.1016/j.conbu​ildma​
The nondestructive evaluation methods based on the trans- t.2019.11753​3
Chui YH (1991) Simultaneous evaluation of bending and shear moduli
verse vibration and ultrasonics to understand the remaining of wood and the influence of knots on these parameters. Wood Sci
bending strength of CLT samples are studied. The transverse Technol 25:125–134. https​://doi.org/10.1007/BF002​26812​
vibration method utilizes the LDV to measure the fundamen- França FJN, Seale RD, Ross RJ (2018) Using transverse vibration non-
tal frequency extracted from free vibration excited by the destructive testing techniques to estimate stiffness and strength of
southern pine lumber. Reserach paper FPL-RP-695. US Depart-
impact hammer. Dynamic modulus of elasticity is calculated ment of Agriculture, Forest Service, Forest Products Laboratory:
using the theoretical vibration equations assuming homog- 8, Madison, WI, p 695
enous structure. The experiments show that the MOEs of Gray J, Grushecky S, Armstrong J (2008) Stress wave velocity and
the CLT samples decrease with the increase in severity of dynamic modulus of elasticity of yellow-poplar ranging from 100
to 10 percent moisture content. In: Proceedings of 16th central
deterioration, which can be used as the indicator of residual hardwood forest conference, vol 24, pp 139–142
strength of CLT samples. Additionally, the moisture content Haines DW, Leban JM, Herbé C (1996) Determination of Young’s
measurements reveal a clear trend between surface condi- modulus for spruce, fir and isotropic materials by the resonance
tion and the ability to conserve water. As the moisture con- flexure method with comparisons to static flexure and other
dynamic methods. Wood Sci Technol 30:253–263. https​://doi.
tent also influences the fundamental frequency, additional org/10.1007/bf002​29348​
experimental feature is needed. It is shown that the damping Jacquier N, Girhammar UA (2015) Evaluation of bending tests on com-
ratio obtained using half-power bandwidth method can be posite glulam—CLT beams connected with double-sided punched
used to measure the moisture content. Ultrasonic method metal plates and inclined screws. Constr Build Mater 95:762–773.
https​://doi.org/10.1016/j.conbu​ildma​t.2015.07.137
can supplement the vibration measurement to understand the Labonnote N, Malo KA (2010) Vibration properties of cross lami-
decay of different layers. As the degradation level of CLT nated timber floors. Structures and architecture—Proceedings of
sample increases, weight loss and drying rate increase. The 1st international conference on structures and architecture ICSA
fundamental frequency associated with the effective MOE 2010, pp 428–435. https​://doi.org/10.1201/b1042​8-56
Labonnote N, Rønnquist A, Malo KA (2013) Experimental evaluations
decreases. As a quality assurance approach, the baseline of material damping in timber beams of structural dimensions.
measurement can be taken from each pristine sample, and Wood Sci Technol 47:1033–1050. https​://doi.org/10.1007/s0022​
the measurements can be periodically repeated to under- 6-013-0556-5
stand the decay in CLT properties with exposure to different Leissa W (1970) Vibration of plates. National Aeronautics and Space
Administration: NASA SP-100, U.S. Government Printing Office,
environmental conditions, and the residual effective bend- Washington, D.C.
ing strength can be identified. Ultrasonic method provides

13
European Journal of Wood and Wood Products (2021) 79:335–347 347

Liu Z, Liu Y, Yu H, Yuan J (2006) Measurement of the dynamic modu- nondestructive testing wood, 10–12 Sept 2007, Duluth, Minne-
lus of elasticity of wood panels. Front For China 1:425–430. https​ sota, USA, pp 13–15
://doi.org/10.1007/s1146​1-006-0048-y Ross RJ, Geske EA, Larson GH, Murphy JF (1991) Transverse vibra-
Llana DF, Íñiguez-González G, Martínez RD, Arriaga F (2018) Influ- tion nondestructive testing using a personal computer. Research
ence of timber moisture content on wave time-of-flight and lon- Paper FPL- RP-502. US Department of Agriculture, Forest Ser-
gitudinal natural frequency in coniferous species for different vice, Forest Products Laboratory, Madison, WI
instruments. Holzforschung 72:405–411 Rossi G, Marsili R, Gusella V, Gioffrè M (2002) Comparison
Mahamid M, Torra-Bilal I (2019) Analysis and design of cross-lami- between accelerometer and laser vibrometer to measure traffic
nated timber mats. Pract Period Struct Des Constr 24:1–18. https​ excited vibrations on bridges. Shock Vib 9:11–18. https​://doi.
://doi.org/10.1061/(ASCE)SC.1943-5576.00003​90 org/10.1155/2002/96850​9
Morandi F, De Cesaris S, Garai M, Barbaresi L (2018) Measurement Schmidt E, Riggio M (2019) Monitoring moisture performance of
of flanking transmission for the characterisation and classification cross-laminated timber building elements during construction.
of cross laminated timber junctions. Appl Acoust 141:213–222. Buildings 9(6):144. https:​ //doi.org/10.3390/BUILDI​ NGS90​ 60144​
https​://doi.org/10.1016/j.apaco​ust.2018.07.009 Slowey K (2016) Wood construction: how does it stack up? In: Constr.
Morandi F, Prato A, Barbaresi L, Schiavi A (2020) On the diffuseness dive, Retrieved from URL: https​://www.const​r ucti​ondiv​e.com/
of the vibrational field of a cross-laminated timber plate: compari- news/wood-const​ructi​on-how-does-it-stack​-up/42572​1/
son between theoretical and experimental methods. Appl Acoust Song YJ, Hong SI (2016) Evaluation of bonding strength of larch cross-
159:107104. https​://doi.org/10.1016/j.apaco​ust.2019.10710​4 laminated timber. J Korean Wood Sci Technol 44:607–615. https​
Ouis D (2003) Effect of structural defects on the strength and damping ://doi.org/10.5658/WOOD.2016.44.4.607
properties of a solid material. Eur J Mech A Solids 22:47–54. Wang X (2008) Effects of size and moisture on stress wave E-rating of
https​://doi.org/10.1016/S0997​-7538(02)01256​-1 structural lumber. In: Proceedings of the 10th world conference
Parajuli R, Gale C, Vlosky R, Martin RO (2018) An overview of cross- on timber engineering: Miyazaki, Japan, 2–5 June 2008, pp 1–8
laminated timber in North America. Increasing use wood global Wang JB, Wei P, Gao Z, Dai C (2018) The evaluation of panel bond
bio-economy—Proceedings of Science Paper, pp 217–225 quality and durability of hem-fir cross-laminated timber (CLT).
Quang Mai K, Park A, Nguyen KT, Lee K (2018) Full-scale static and Eur J Wood Prod 76:833–841. https​://doi.org/10.1007/s0010​
dynamic experiments of hybrid CLT–concrete composite floor. 7-017-1283-7
Constr Build Mater 170:55–65. https​://doi.org/10.1016/j.conbu​
ildma​t.2018.03.042 Publisher’s Note Springer Nature remains neutral with regard to
Ross RJ (2008) Transverse vibration techniques logs to structural jurisdictional claims in published maps and institutional affiliations.
systems. In: Proceedings of 15th international symposium on

13