Piotr Bobra, ul. Akacjowa 13, 46-6 Prószków, POLAND, e-mail: p.bobra@po.opole.pl EXPERIENTAL TESTING OF A WOODEN FRAE UNDER SEISIC EXCITATION Abstract Seismic experiments of wooden, three story frame are described in detail. Preliminary results of dynamic tests are shown. Differences between numerical FE (Finite Element ethod) model and experimental natural frequencies are indicated. 1. INTRODUCTION Wooden structures are widely used in civil engineering. Their low weight and high strength properties make them particularly resistant against dynamic loads. On the other hand their static and dynamic properties are often scattered and random. That is why field and laboratory tests are particularly needed for these structures (see e.g. [1], [2]). In December 29 a European project High performance composite-reinforced earthquake resistant buildings with self-aligning capabilities was initiated under the FP7-INFRASTRUCTURES-28-1 program (SP4-Capacities). This project is within soul called SERIES action (seismic engineering research) coordinated by University of Patras, Greece. Three institutions are involved in these particular project: 1) Institute of Theoretical and Applied echanics, Czech Republic, 2) Technical University Dresden, Germany, 3) Faculty of Civil Engineering, Opole University of Technology, Poland. The project lasts two years and its main purpose is to test two models of wooden structures on shaking table. The models are designed to withstand seismic actions by the application of specially designed joints and composite timber. The purpose of these paper is to present preliminary results of dynamic analysis just after tests of the second model. It covers comparison of dynamic properties of the tested model as computed by FE code and measured on the intact structure, before the main series of damaging tests. 2. DESCRIPTION OF ANALYZED ODEL OF WOODEN FRAE The investigated frame is made of four symmetric spaced columns, connected to each other by beams. It has a three stories, each stiffened by a 5mm thick plywoods slab (fig. 1). oreover four cantilever beams (two on the first and two on the second floor) are attached to column no. 1 (see detail E in fig. 1). The columns and the beams are made of glulam - class GL24, spruce. The slab is made of spruce, double plywood LVL (thickness 2x25mm=5mm). To obtain the appropriate dynamic behavior of the structure under seismic excitations, the frame was loaded by additional masses. In the vicinity of column no. 1 on the first floor there were five lead masses fasten (total weight of 125kg). In vicinity of the column no. 3, at each floor, four additional masses of total weight 1kg, were fasten. Third additional mass (25kg) was placed on two brackets of the first floor (see detail in fig. 1). The purpose of these brackets was to generate additional bending moment, simulating existence of further frame bays. In fig. 1 the tested model placed on shaking table is presented in detail. Strana 56
D3 F G2 G1 E E D2 C2 C1 G2 G1 D1 F B E B C3 C4 z z y x A y x A Fig. 1 View of the tested, three story frame. A shaking table, B supporting structure for linear position sensors, C1-C4 numbers of columns, D1-D3 slabs, E bracket, F target for high-speed camera, G1 linear position sensor, G2 end of linear position sensor, additional mass. 3. INSTRUENTATION The frame was subjected to a series of dynamic excitations on the shaking table. Its dynamic response was analyzed using the following measurement systems. The linear position sensors were attached to the support structure (see detail B in fig. 1) and used to measure the relative displacements of two, perpendicular beams of the first floor. These sensors were also used to measure change of an angle between column 1, shaking table plane and each adjoining beam. The main, structural displacements were measured along the diagonals of faces of the 1 st and 2 nd floor (see details G1 & G2 in fig 1). In addition 18 linear position sensors were used to measure remaining structural displacements. Accelerometers were attached to two perpendicular beams, on each floor. Their purpose was to measure horizontal accelerations. In the middle of each slab, additional systems of three sensors measuring acceleration along three, orthogonal directions were installed. Additional accelerometers were controlling the motion of shaking table. Total number of 24 accelerometers was applied. To analyze strains and stresses during vibrations column no. 1 was chosen, where 2 strain gauges were glued. Similar strain gauges were placed to the adjoining beams. Two high-speed cameras were applied to trace motion of 6 targets (see detail F in fig. 1) attached to three joints of column no. 2. Strana 57
The independent three-axial accelerometer was attached to the highest joint on column no. 3. This sensor was used to make a quick identification of modal properties of the frame, right after its erection (SEQUOIA FastTracer system). 4. DESCRIPTION OF DYNAIC EXPERIENTS After mounting and instrumenting the frame, a series of free vibration tests were performed. These included an impact test and snapback pullover experiment. During these tests the shaking table was locked so the frame could not interact with the mechanical structure of the shaking table. It allowed preliminary evaluation of dynamic properties of model, correcting position of additional masses and further planning of the experiment. Additional low level tests included a sweep sine and white noise excitations. After the initial tests, the actual damaging series of tests started. They were carried out using artificially generated seismic signal with increasing, intensities during each phase of the experiment. After each damaging phase low level, white noise tests were carried out to capture the changes to dynamic parameters induced by the damages. Detailed information on the sequence of the experiment is shown in table 1. Table 1. Sequence of the shaking table experiment. No. excitation type excitation direction of level excitation 1 impact - X 2 impact - Y 3 snapback pullover 25kg (force) Y 4 white noise.1g X 5 white noise.1g Y 6 white noise.1g X, Y 7 sweep sine.3g X 8 sweep sine.3g Y 9 seismic.1g /.2g* X, Y, Z* 1 white noise.1g X, Y 11 seismic.3g /.6g* X, Y, Z* 12 white noise.1g X, Y 13 seismic.5g /.1g* X, Y, Z* 14 white noise.1g X, Y 15 seismic 1.g /.2g* X, Y, Z* 16 white noise.1g X, Y 17 seismic 2.g X, Y Strana 58
5. RESULTS OF FE CALCULATION AND EXPERIENTAL TESTS. Initial dynamic analysis was done using SAP2 v.14.4.2. FE code. Two types of finite elements were applied: FRAE elements (to model beams and columns) and SHELL elements (to model the slabs). The number of dynamic degrees of freedom equaled 8362. The finite element mesh of the frame is shown in fig. 2. Fig. 2 FE mesh of the wooden frame. First the eigenproblem was solved. The resulting, first three vibration modes (along Y, along X and torsion) are listed and described in table 2 as well as in figs. 3-5. At the early stages of the experiment impact tests and low level, white noise seismic tests were carried out. In figs. 6-7 decaying accelerations of the third floor of the frame are shown together with respective power spectral densities (Fourier transforms). Clearly first natural frequency (4.75 Hz along Y axis) and second one (5.75 Hz along X axis) can be observed. The torsional, third natural frequency (16.5 Hz) can hardly be noticed. In fig. 8 the translational modes are difficult to observe, however the torsional mode with some coupled disturbances are displayed. Table 2. Description of selected natural modes of vibrations. Description mode number FE model The first cantilever mode of frame The second cantilever mode of frame The torsional mode of vibration natural frequency f natural period T measured natural frequency f 1 4.35 Hz.23 s 4.75 Hz 2 4.94 Hz.2 s 5.75 Hz 3 6.19 Hz.16 s 16.5 Hz Strana 59
Fig. 3 The first mode of vibration (un-deformed shape shown in dark). f=4.35 Hz, T=.23 s Fig. 4 The second mode of vibration (un-deformed shape shown in dark). f=4.94 Hz, T=.2 s Fig. 5 The third mode of vibration (un-deformed shape shown in dark). f=6.19 Hz, T=.16 s Strana 6
[(m/s 2 ) 2 /Hz].6 5.75 Hz.5.4.3 1s 4m/s 2.2 4.75 Hz.1 16.5 Hz 2 4 6 8 1 12 Fig. 6 Power spectral density of the accelerations of the third floor under impact load (time history in the in-set picture). Both, impact and acceleration sensor along X axis. 14 16 18 2 22 [Hz] [(m/s 2 ) 2 /Hz].14.12 4.75 Hz.1.8 1s 2m/s 2.6.4 5.75 Hz.2 16.5 Hz 2 4 6 8 1 12 Fig. 7 Power spectral density of the accelerations of the third floor under impact load (time history in the in-set picture). Both, impact and acceleration sensor along Y axis. 14 16 18 2 22 [Hz] Strana 61
[(m/s 2 ) 2 /Hz].7.6 16.5 Hz.5.4 1s 2m/s 2.3.2.1 2 Fig. 8 Power spectral density of the accelerations of the third floor under white noise, seismic excitations (time history in the in-set picture). Seismic excitations and acceleration sensor along X axis. 6. CONCLUSIONS 4 6 8 1 12 Early results of experimental seismic tests of three story wooden frame are presented. At this stage only simple isotropic FE model was applied to model dynamic behavior of the frame. The results of low level dynamic tests show substantial differences between numerical evaluations and experimental natural frequencies (see table 2). This is due to actual complicated structure of the wood and specially constructed joints which shall be modeled by nonlinear constitutive material law with orthogonal properties. Further analyses are planned by applying various nonlinear models and appropriate finite element codes (e.g. Abaqus, ANSYS). 14 16 18 2 22 [Hz] LITERATURE [1] Kasal, B., Pospíšil, S., Jirovsky, I., Drdacky,., Heiduschke, A., and Haller, P.: Seismic performance of laminated timber frames with fiber-reinforced joints. Earthquake Engineering and Structural Dynamics. John Wiley & Sons Ltd. London, K, Vol. 33. (5): 633-646, 24. [2] Heiduschke A, Kasal B, Haller P. Performance and Drift Levels of Tall Timber Frame Buildings under Seismic and Wind Loads. Structural Engineering International, Vol. 18, Nr. 2, pp. 186-191, 28. Strana 62