Experimental and analytical study on dynamic performance of timber floor modules (timber beams)

Construction and Building Materials 122 (2016) 391–399

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Experimental and analytical study on dynamic performance of timber floor modules (timber beams) Rajendra Rijal ⇑, Bijan Samali, Rijun Shrestha, Keith Crews Centre for Built Infrastructure Research (CBIR), School of Civil and Environmental Engineering, University of Technology Sydney (UTS), Australia

h i g h l i g h t s
Experimental and analytical dynamic investigation of six full-scale timber beams.
Performed impact hammer tests on the timber beams with two different spans.
The fundamental frequency of all tested beams was above 8 Hz, which is acceptable.
Analytical models predicted fundamental frequency of all beams within acceptable range of ±5%.
Mean damping ratio of 6 m and 8 m span beams was 0.65% and 0.58%, respectively.

a r t i c l e

i n f o

Article history: Received 19 November 2013 Received in revised form 17 May 2016 Accepted 11 June 2016 Available online 9 July 2016 Keywords: Timber floor module Vibration Instrumented hammer Natural frequency Damping ratio Mode shape Frequency prediction model

a b s t r a c t Timber floors are more susceptible to vibrations and have low impact insulation due to low stiffness and poor damping properties. Recent trends towards long-span and light-weight construction make floor vibration even more critical in satisfying serviceability requirements of floor constructions. This paper presents the results of dynamic tests conducted on timber floor modules (beams) with 6 and 8 m clear spans using an instrumented hammer for floor excitation. Dynamic parameters such as natural frequencies, damping ratios and mode shapes from the tests were evaluated to assess dynamic performance of the beams. The fundamental frequency of the beams was predicted using simple analytical models and good correlation between the test results and predicted values could be obtained. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Timber is the only truly renewable and environmentally friendly natural building material. Additionally, high strength to weight ratio, ease in workmanship and handling, good fire resistance are some of the properties that make timber attractive for use in construction. However, due to the light-weight nature of timber, timber floors have poor vibration and low impact sound insulation. Use of timber in residential buildings is widely accepted in Australia. However, its use in non-residential buildings has been limited. The availability of engineered wood products (EWPs) such as laminated veneer lumber (LVL) and glue laminated timber (Glulam) and new generation of adhesives products makes it

⇑ Corresponding author at: School of Civil and Environmental Engineering, University of Technology Sydney, PO Box 123, Broadway, NSW 2007, Australia. E-mail address: [email protected] (R. Rijal). http://dx.doi.org/10.1016/j.conbuildmat.2016.06.027 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.

possible to fabricate composite timber section to meet both strength and serviceability requirements for long-span floors. Design of long-span and light-weight floor construction may be governed by serviceability requirements rather than strength and dynamic performance is one of these requirements. Therefore, there is a growing need for measurement of dynamic characteristics such as natural frequencies, damping ratios, and mode shapes of floor systems to investigate their behaviour. The serviceability design of flooring systems requires an assessment of the fundamental frequency (or first natural frequency) in order to check the vibration behaviour of the floor and occupant comfort. The following frequency ranges must be avoided for any of the vibration modes:
Frequency below 3 Hz to prevent walking resonance [1].
Frequency range of 5–8 Hz to prevent human discomfort [1]. For residential/office floors, a natural frequency greater than 10 Hz shall be targeted [1]. A special investigation is needed if

392

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

the predicted fundamental frequency of the floors is less than 8 Hz. Hence, the prediction of natural frequency, especially fundamental frequency, becomes important for the dynamic assessment of the flooring systems. The dynamic assessment of the flooring systems is essential in recent times owing to trends towards long-span and light-weight construction. Therefore, there is a growing need for measurement of dynamic characteristics such as natural frequencies, damping ratios and mode shapes of the systems to assess their performance. Jarnerö et al. [2] assessed vibration behaviour of a prefabricated timber floor element in laboratory with different boundary condition and in field at different stages of construction and concluded that on-site conditions have significant influence on the floor damping. Hamm et al. [3], on the other hand, found that timber floors with natural frequencies less than 8 Hz could still have acceptable vibration performance based on in-situ heel drop tests on 50 buildings and 100 floors. Also, a number of other studies have investigated the effect of parameters such as joist spacing, boundary conditions, floor configuration, etc on the modal frequency, damping and mode shape of both timber only and timber-concrete composite floors [4–6]. 2. Experimental investigation Six timber beams made of laminated veneer lumber (LVL) were tested to assess their dynamic performance under the application of impact action initiated with the use of a modal impact hammer. Three of the beams had a clear span of 6 m with identical geometry (L6-01, L6-02 and L6-03) and the remaining three had a clear span of 8 m (U8-01, U8-02 and U8-03). The overall length of the 6 and 8 m span beams was 6.3 and 8.4 m, respectively. 2.1. Geometry of the beams The cross sectional dimensions for the 6 and 8 m span beams are shown in Fig. 1(a) and (b), respectively. The only differences between the two sections were the depth of the webs and width

of the bottom flanges. The top and bottom flanges of all beams were glued and screwed to the web. Further details of the specimens can be found in [7]. The design of both 6 and 8 m span beams were governed by serviceability limit state criteria [7], which is not uncommon for longspan timber beams. The cross-sectional shape of the beams was chosen based on the following advantages:
The space between the webs can be used for the installation of services and acoustic insulation.
The section has versatile application in the floor construction as it can also be used as upside-down in the modular construction and an additional layer such as concrete topping can be applied on top of it to increase the stiffness of the flooring systems which in turn reduces vibration and static issues.
It is more stable compared to the ‘‘T” and ‘‘I” sections. 2.2. Material properties 2.2.1. LVL timber Two types of LVL, hySPAN Cross-banded LVL for the top flange and hySPAN PROJECT LVL for the webs and bottom flanges, were used in the fabrication of the beams. Flat-wise properties for the top and bottom flanges and edge-wise properties for the webs, which replicate the orientation of the flanges and webs in the beams, were tested. A summary of the results (mean values) for individual components of the beams is presented in Table 1. The dynamic performance of the systems is highly sensitive to the modulus of elasticity (MOE) and the density of the timber beams. Minimum of eleven samples were tested for each component to find MOE of the LVL and the flanges were found to have higher variability in the MOE values compared to the webs. Bottom flanges had maximum coefficient of variation (CoV) of 12.3%. The density of the components showed no significant variation within individual components and among the components [7].

Fig. 1. Geometry for 6 and 8 m span timber beams (mm).

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399 Table 1 Material properties of LVL timber. Component name Top flange CoV (%) Webs CoV (%) Bottom flanges CoV (%)

MOE Ex (GPa)

q (kg/m3)

Density

9.6 10.7 13.3 4.0 13.1 12.3

607 1.0 604 1.3 601 1.8

The timber beams were stored under normal laboratory conditions. Moisture content (MC) of the tested beams was measured on the day of testing in accordance with [8] using oven dry method. Three small LVL timber blocks (referred to as moisture content samples), measuring 100  100  50 mm, were put on top of each beam one week before testing and were used to measure the moisture content of the tested beams. The moisture content values were calculated using mass of the test pieces before (initial weight) and after (dry weight) putting in the oven and the percentage of moisture content of a test piece was determined using Eq. (1). No significant difference in the moisture content values among the tested beams could be found. The moisture content in the beams varied from 8.2 to 10.0% over a period of 20 weeks. As such, the moisture content is not expected to have any significant effect on the dynamic parameters of the tested beams.

MC ¼ ðMi  Mo Þ=M o  100

ð1Þ

where Mi is the initial mass of test samples, Mo is the oven-dried mass of test samples. 2.2.2. Shear connector The top and bottom flanges of the beams were connected to the web using glue (Purbond) and normal Type 17 wood screw (Fig. 2). The screw had a measured total length of 90 mm and thread length was measured as 45 mm. The measured shank diameter was 5 mm. The screws were installed at equal spacing of 375 and 385 mm for 6 and 8 m span beams, respectively. The interface between flanges and web was assumed to be fully composite as no slip was observed during the four point bending tests under serviceability loads [7] and the bond between the flanges and the web was not expected to be broken under hammer impact load. 2.3. Test setup All specimens were tested under the same boundary conditions using pin-roller supports. The pin and roller arrangements were

393

supported on top of concrete blocks. The pin arrangement had a metal shaft between two metal plates having grooves on both plates, which constrained any horizontal movement, while there was no groove on top plate for roller arrangement which allowed horizontal movement in order to replicate the roller supports (Fig. 3). There was no mechanism to prevent vertical movement due to large complex cross-sectional geometry of the beams; nevertheless, no significant vertical movement was expected due to self-weight of the beam and hammer impact load was also applied gently during the test to minimize bouncing at the supports. The schematic diagram of boundary conditions for 6 and 8 m span beams is depicted in Fig. 4(a) and (b), respectively. 2.4. Instrumentation and testing procedure Fifteen magnetic uniaxial accelerometers were mounted at different locations along the length of the beams as depicted in Fig. 5 and their position on the cross-section at each location is shown in Fig. 6. The purpose of using seven accelerometers on the top flange along the beam centreline was to capture the first three ideal flexural modes as shown in Fig. 7, where numbers 1–5 are the nodal points of the three modes and one accelerometer was placed at each node. Two accelerometers at the supports were used to monitor any bouncing at the supports. Five accelerometers on the top flange on one side of the beam were used to capture torsional modes and the other five accelerometers on the web and bottom flanges were used to study the degree of composite action between flange and web, which is not covered in this paper. The instruments used in the test are shown in Fig. 8. The accelerometers were ICP type with PCB model 352C34 and 337A26 having a sensitivity of 100 mV/g as shown in Fig. 8 (a) and (b), respectively. All beams were subjected to free vibration initiated by impact from an instrumented modal hammer (PCB model 086D20) as shown in Fig. 8(c). Trial tests were conducted with different hammer tips with different stiffness to select an appropriate tip which results in clear impact response of the beams. Impact force was applied on the top of top flange at midspan (node 3 as shown in Fig. 7) and 1/3rd span (node 2 as shown in Fig. 7) of the beam from pin support along its centreline to excite at least the first three flexural modes. The multi-channel signal conditioner (PCB model 483B03) as shown in Fig. 8(d) was used to amplify and condition the time history signals from the impact hammer and the accelerometers. The input and output channels from signal conditioners were connected to the data acquisition system through terminal block. Terminal block (model SCB – 68) merely conveyed the time history signals acquired from signal conditioners to the acquisition system without any modification (see Fig. 8e). The acquisition system had two internally synchronised acquisition cards (model NI PCI – 6133) with resolution of 14 bit and sampling rate of 2.5 MS/s, simultaneously (see Fig. 8f). The full instrumentation for 6 m span beams is shown in Fig. 9. The instrumentation for 8 m span beams was identical to 6 m span beams. For all tests, the sampling rate was set to 1000 Hz with 16,384 time domain data points recorded while this corresponds to a frequency band width ranging from 0 to 500 Hz with 8192 Frequency Response Function (FRF) data points giving a frequency resolution of 0.061 Hz in the frequency domain. 2.5. Data processing and analysis

Fig. 2. The dimensions of the screw used in the timber beams (mm).

Modal testing (MT) and experimental modal analysis (EMA) was performed to identify the modal parameters (natural frequencies, damping ratios and mode shapes) of the timber beams. The MT and EMA configuration is shown in Fig. 10. Fourier transformation was used to convert time history signals obtained from impact hammer and accelerometers into frequency spectra. Frequency

394

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

Fig. 3. Beam supports (a) pin (b) roller.

(a) 6 m span timber beam

(b) 8 m span timber beam Fig. 4. Schematic diagram of setup for 6 and 8 m span timber beams (mm).

Fig. 5. The arrangement of the accelerometers along the span.

FRFs were obtained from the ratio of the Fourier transformed output and input signals. The modal parameters of the beams were extracted from the FRFs with the use of post-processing module in software package [9]. The dynamic behaviour of the beams was assessed by evaluating their modal parameters. 3. Experimental results 3.1. Modal parameters The captured first three flexural modes of beam U8-01 are depicted in Fig. 11. For all modes, an uplifting motion is observed as expected for all tested beams as there was no restrain in the vertical direction at the supports owing to complexity of pin-roller arrangements. The bouncing at supports was observed to be higher for the higher modes. For mode three, the bouncing at the left support is higher compared to the bouncing at the right support. A summary of the captured three flexural natural frequencies and the corresponding damping ratios for the timber beams is summarised in Table 2. For the 6 m beams, small coefficient of variation (CoV) (up to 2%) was observed for frequency results of all three modes of vibration of the beams. For the 8 m beams, the CoV for the frequency of first mode of vibration was only 2%, but it was 5 and

9% for second and third modes, respectively. Higher discrepancies in frequency results for higher modes of vibration is attributed to the higher bouncing observed at the supports and possible variation in material properties along the beam length due to the presence of localised growth characteristics, the effect of which become more prominent in higher modes of vibration. The damping ratio results were less consistent compared to the frequency results. The smallest CoV of 6% was found for the first mode of 6 m span beams while a CoV of 17% was found for the third mode of 8 m span beams. No general trend was observed between the damping ratios and the natural frequencies. Hu et al. (2001) [10] mentioned that timber only flooring systems are more susceptible to vibrations than timber-concrete composite (TCC) flooring systems as timber only flooring systems have damping ratios of about 1% while the TCC flooring systems have damping ratios of about 2%. Higher damping reduces the annoying springiness effect on the floors caused by human activities. The damping ratios of all the tested timber beams were well below 1%.

4. Analytical investigation A number of analytical prediction models can be found in the design standards and literature to determine the fundamental frequency, particularly with flooring systems. In this section, each of

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

(a) Cross-section showing location of one accelerometer at supports

395

(b) Cross-section showing location of two accelerometers at 1/6, 2/6, 4/6 and 5/6 span

(c) Cross-section showing location of five accelerometers at mid span Fig. 6. Location of the accelerometers at different sections of the beams.

Fig. 7. First three ideal flexural mode shapes and their node points.

these prediction models are first summarised and then compared against the test results to investigate reliability of these models. 4.1. Frequency prediction models The prediction models are essentially based on the natural frequency prediction relationship for a simple spring-mass (single degree of freedom) system as expressed in Eq. (2) and is also expressed in terms of elastic deflection of the system due to its self-weight and imposed action as given in Eq. (3).

qffiffiffiffiffiffiffiffiffiffi f 1 ¼ 1=2p k=m

ð2Þ

where k is the stiffness of the system and m is the mass of the system.

pffiffiffiffiffiffiffiffiffi f 1 ¼ 1=2p g=D

ð3Þ

where g is the acceleration due to gravity (9.81 m/s2) and D is the elastic deflection of the system. 4.1.1. Wyatt (1989) methods Wyatt (1989) [11] has presented four methods to evaluate natural frequencies of the beams and floors. In this section, only the first two analytical methods will be discussed as the other two methods need software packages such as iterative application of static analysis using common static analysis software at the

desk-top and dynamic analysis software packages including FE method of the structure to predict the natural frequencies. The first method predicts the fundamental frequency of the systems based on the self-weight deflection approach using Eq. (4).

pffiffiffiffiffiffiffiffiffiffiffi f 1 ¼ 1=2p g=yw

ð4Þ

where yw is the weighted average value of the static deflection (mm) due to self-weight of a slab or floor beam system and g is the acceleration due to gravity (m/s2). The weighted average value of the deflection, yw, is expressed in terms of maximum deflection, y0, due to the self-weight of the systems as yw = (3/4)y0 and Eq. (4) is then rewritten in the simpler form as given in Eq. (5).

pffiffiffiffiffi f 1 ¼ 18= y0

ð5Þ

This method can be used to estimate the fundamental frequency of a slab and floor beam system on main beams. This method does not consider long-term deflections due to shrinkage and creep effects, and the appropriate assumptions should be made on boundary conditions, material properties and the contribution of imposed loads for continuous beams. Fundamental frequency of a slab and floor beam system on stiff main beams can be estimated quite accurately using this method. The second method is based on the equivalent beam method (EBM) that can be used to predict the fundamental frequency of single span beams with different types of boundary conditions and continuous beams having maximum of three spans. The ana-

396

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

(a) Accelerometer model PCB 352C34

(b) Accelerometer model PCB 337A26

(c) Modal hammer

(d) Multi-channel signal conditioner

(e) Terminal block

(f) Data acquisition system Fig. 8. Test instruments.

lytical solution is given in Eq. (6). The frequency coefficient, CB, is the same for both pin-pin and pin-roller boundary conditions whereas a slightly higher frequency is expected for pin-pin compared to pin-roller boundary condition.

Allen (1990) [12] modified Eq. (4) to predict the fundamental frequency, f1, considering elastic deflection of the flooring system to take into account all these factors as,

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi f 1 ¼ C B EI=mL4

f 1 ¼ 1=2p

ð6Þ

where m is the mass per unit length (t/m if EI is in kNm2, or kg/m if EI is in Nm2), L is the span in meters (for continuous beams, the longest span should be used) and CB is the frequency coefficient which depends upon the number of spans and boundary conditions (Table 3 and Fig. 12). 4.1.2. Allen (1990) method Generally the deflection based prediction models estimate the natural frequency of the beams and flooring system from the flexural deflection of the beam, however, the deflection due to shear deformation and flexibility of supports influence the flexibility of the flooring system significantly and ultimately reduce the natural frequency obtained solely from beam flexure. The shear deformation is important for deep beams, girders and trusses.

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi g DB þDG þ DS 1:3

ð7Þ

where DB is the mid-span deflection of the floor beam due to flexure and shear, DG is the deflection of the girder at the beam support due to flexure and shear, DS is the deflection of the supports such as columns or walls due to axial strain. All deflections are consequence of self-weight of the member and other imposed actions. The factor 1.3 in Eq. (7) can be used for most beam-floor systems while the factor should be increased to 1.5 in the case of fixed cantilevers and two-way slabs. The deflection DB and DG should be determined based on the complete mode shape of the systems if beams or girders are continuous over supports. The deflection of adjacent spans is in the opposite directions without any change in slope over the supports, and the weight carried by individual spans acts in the direction of deflection.

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

397

Fig. 9. Instrumentation on a 6 m span timber beam.

Fig. 10. MT and EMA configuration.

4.1.3. Murray et al. (2003) method Murray et al. (2003) [13] provided a similar approach to second method of Wyatt (1989) [8] prediction model to predict the fundamental frequency of the simply supported beams or joists and girder panels under uniformly distributed permanent and imposed actions as given in Eq. (8).

f1 ¼

p 2

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi gEI=wL4

ð8Þ

4.1.4. Eurocode 5 (2008) method Eurocode 5 (2008) [14] also presents a similar approach to second method of [11] prediction model to predict the fundamental frequency of the rectangular floors simply supported along all four edges using Eq. (10). It also presents a ‘‘Gamma method” to determine the equivalent bending stiffness of the systems having partial composite action such as TCC flooring systems, but transformed section method can be used for fully composite systems.

f 1 ¼ p=2l

2

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðEIÞl =m

ð10Þ

where E is the modulus of elasticity of the system, I is the transformed moment of inertia of the system, w is the uniformly distributed load per unit length and L is the span length. This equation was further simplified in terms of mid-span deflection of a simply supported beam as;

where (EI)l is the equivalent bending stiffness of the floor perpendicular to the beam direction (Nm2/m), l is the floor span (m) and m is the mass per unit area (kg/m2). The mass includes selfweight of the floor and other permanent actions.

pffiffiffiffiffiffiffiffiffi f 1 ¼ 0:18 g=D

4.2. Frequency prediction for tested beams

ð9Þ

where D is the mid-span deflection of a beam due to the uniformly distributed permanent action, including self-weight and can be determined from 5wl4/384EI.

The fundamental frequencies of the timber beams were predicted using 1st and 2nd methods of [11] models using Eqs. (5) and (6). In addition, the frequency was also estimated using Eqs.

398

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399

(a) Mode 1

(b) Mode 2

(c) Mode 3 Fig. 11. First three flexural mode shapes of the timber beam U8-01.

Table 2 Summary of test results of the 6 and 8 m span timber beams. Beam

Natural frequencies (Hz)

Damping ratios (%)

Mode 1 Mode 2 Mode 3 Mode 1 L6-01 16.47 L6-02 16.16 L6-03 16.40 Mean 16.34 Standard deviation 0.17 CoV (%) 1.00 U8-01 13.25 U8-02 12.89 U8-03 13.27 Mean 13.14 Standard deviation 0.21 CoV (%) 2.00

56.60 54.61 55.23 55.48 1.02 2.00 43.32 42.52 46.39 44.08 2.05 5.00

100.63 97.05 97.05 98.25 2.07 2.00 80.68 71.47 84.53 78.89 6.71 9.00

0.60 0.68 0.66 0.65 0.04 6.00 0.64 0.58 0.52 0.58 0.06 10.00

Mode 2

Mode 3

0.36 0.28 0.37 0.33 0.05 14.00 0.35 0.46 0.41 0.40 0.06 14.00

0.30 0.34 0.33 0.32 0.02 7.00 0.48 0.43 0.34 0.42 0.07 17.00

Fig. 12. Frequency factor, CB for continuous beams [11].

Table 4 Summary of the parameters used to predict the fundamental frequency.

Table 3 Values of CB for a single span [11]. No. of span

End conditions

Values of CB

Single

Pinned/pinned (simply supported) Fixed/pinned Fixed both ends Fixed/free (cantilever)

1.57 2.45 3.56 0.56

(7), (9) and (10) from [12]; [13] and [14] prediction models, respectively. The deflections DG and DS were discarded in [12] prediction model for the tested beams. A summary of the parameters that were used as input to the prediction models to predict the fundamental frequency of the timber beams is summarized in Table 4, where mass corresponds to the self-weight of the beams. Bending stiffness of the beams, EI, was calculated using the transformed section method, considering fully composite behaviour between the webs and the flanges, and the material properties (as relevant) from the experimental tests were also used. The material properties were assumed to be isotropic for individual components as it was deemed sufficient to represent the flexural stiffness in longitudinal direction along the span length, which is the major contributor to the overall stiffness of the beams. The material properties for the individual components were determined by testing the component in either flat-wise or edge-wise direction, depending on their orientation

Span L (m)

Bending stiffness EI (MNm2)

Mass m (kg/m)

Mid-span deflection yo (mm)

6 8

4.03 9.47

28.0 34.4

1.15 1.90

in the tested beams. The mid-span deflection, yo, of the beams due to their self-weight with simply supported boundary conditions was also calculated. A summary of the predicted values from each prediction model for the beams and their correlation with experimental results is given in Table 5. The results show that the fundamental frequencies predicted from all models are fairly accurate as values are within ±5% of those measured experimentally. The predicted values from the 1st method of [11]; [12] and [13] models are similar as all models are based on the mid-span deflection of the systems. The prediction model [12] is more advanced compared to the 1st method of model [11] and [13] as it incorporates the deflections DG and DS of the systems (if applicable). The results from the 2nd method of [11] and [14] are identical as the latter model was originally derived from the 2nd method. A sensitivity analysis was performed on variation of LVL material properties such as mean MOE (E) and density (q) for 6 and 8 m timber beams using the 2nd method of [11] prediction model as

399

R. Rijal et al. / Construction and Building Materials 122 (2016) 391–399 Table 5 Summary and comparison of the predicted values against the test results. Span

6m f1(model)/f1(exp) 8m f1(model)/f1(exp)

Prediction, f1(model), Hz

Experiment

Wyatt (1989) 1st method 2nd method [6]

Allen (1990)

Murray et al. (2003)

Eurocode5 (2008)

[7]

[8]

[9]

16.79 1.03 13.06 0.99

16.76 1.03 13.04 0.99

16.63 1.02 12.94 0.98

16.56 1.01 12.88 0.98

16.55 1.01 12.88 0.98

expressed in Eq. (6). The properties were varied by ±5% for 6 and 8 m span beams. The analyses showed that the experimental fundamental frequencies were within the range of ±5% variation of MOE and density. Hence, it can be concluded that the discrepancies between experimental and predicted natural frequencies of 6 and 8 m timber beams were mainly due to the material properties, especially MOE, variation for LVL was up to 12.3% for CoV.

f1(exp) (Hz)

16.34 13.14

Wyatt (1989) [11] models in different forms. Wyatt’s prediction method [11] was modified by Allen (1990) [12] considering the effect of girder and support deflections.
Results from all the prediction models were within 5% of the fundamental frequencies of the timber beams.
All prediction models can be utilised to estimate the fundamental frequencies of timber beams accurately.

5. Conclusions Acknowledgement Dynamic tests were conducted on six timber beams with 6 and 8 m clear spans, to assess their dynamic performance based on the modal parameters. Fundamental frequency of the beams was predicted using five analytical prediction models and the predicted values were correlated with the test results to determine their reliability. The following conclusions can be drawn based on the experimental and analytical investigations:
The mean fundamental frequency of the 6 and 8 m span beams was 16.3 and 13.1 Hz, respectively, which are well above the minimum frequency recommended by most codes of practice (8 Hz). This shows that timber only floors with long spans can be designed to fulfil the vibration limits and ensure occupant comfort.
Low variation (up to 2%) in the lower mode frequencies in the tested beams were observed while a variation of up to 9% was observed for higher modes which could be attributed to localised growth characteristics and bouncing at the supports.
The mean damping ratio of the 6 and 8 m span beams was 0.65% and 0.58%, respectively. The results showed no general trend between the natural frequencies and the damping ratios. Timber beams have lower damping ratios compared to TCC and concrete beams, mainly due to their lighter weight.
The results showed discrepancies within 6 and 8 m span beams due to a number of factors such as material properties, shear connectors (screws and glue), minor discrepancies in the geometry, moisture content, bouncing at supports and minor discrepancies in the boundary conditions. For higher modes, the bouncing at supports was observed to be higher.
Five analytical prediction methods were summarized in this paper. The prediction models given in Murray et al. (2003) [13] and Eurocode 5 (2008) [14] were basically derived from

This project has been funded by Structural Timber Innovation Company (STIC). The financial assistance of the STIC is gratefully acknowledged. References [1] R.M. Hanes, Human Sensitivity to Whole-Body Vibration in Urban Transportation Systems: A Literature Review, Applied Physics Laboratory, The John Hopkins University, Silver Springs, MD, 1970. [2] K. Jarnerö, A. Brandt, A. Olsson, Vibration properties of a timber floor assessed in laboratory and during construction, Eng. Struct. 82 (2015) 44–54. [3] P. Hamm, A. Richter, S. Winter, Floor vibrations – New results, 11th World Conference on Timber Engineering, Trentino, Italy, 2010. [4] B. Zhang, A. Kermani, T. Fillingham, Vibrational performance of timber floors constructed with metal web joists, Eng. Struct. 56 (2013) 1321–1334. [5] N.H. Abd Ghafar, B. Deam, M. Fragiacomo, Vibration susceptibility of multispan LVL-Concrete composite floors, 11th World Conference on Timber Engineering, Trentino, Italy, 2010. [6] N.H. Abd Ghafar, B. Deam, M. Fragiacomo, A. Buchanan, Susceptibility to vibrations of LVL-Concrete composite floors, VII Workshop Italiano Sulle Strutture Composte, Benevento, Italy, 2008. [7] Z. Zabihi, B. Samali, R. Shrestha, C. Gerber, K. Crews, Serviceability and ultimate performance of long span timber floor modules, 12th World Conference on Timber Engineering, Auckland, New Zealand, 2012. [8] Australian/New Zealand Standard, Timber – Methods of test, Moisture content, AS/NZS 1080.1, 1997. [9] CADA-X, Modal Analysis Manual, LMS International, Belgium, 1996. [10] L.J. Hu, Y.H. Chui, D.M. Onysko, Vibration serviceability of timber floors in residential construction, Prog. Struct. Mater. Eng. 3 (3) (2001) 228–237. [11] T.A. Wyatt, Design Guide on the Vibration of Floors, SCI Publication 076: Steel Construction Institute, Construction Industry Research and Information Association, UK, 1989. [12] D.E. Allen, Building Vibrations from Human Activities, American Concrete Institute, Concrete International: Design and Construction, 1990, pp. 66–73. 12 (6). [13] T.M. Murray, D.E. Allen, E.E. Ungar, Steel Design Guide Series 11: Floor Vibrations Due to Human Activity, American Institute of Steel Construction (AISC), Chicago, USA, 2003. [14] Eurocode 5, Design of timber structures, Serviceability Limit States, 2008.