Modelling the influence of building types on the building response to railway vibration M. Villot 1, P.Ropars 1, P.Jean 1, E.Bongini 2, F.Poisson 2 1 CSTB, 24 Rue Joseph Fourier, 38400 Saint Martin d Hères, France { michel.villot@cstb.fr, pierre.ropars@cstb.fr, philippe.jean@cstb.fr } 2 SNCF, I&R/PFC, 45 Rue de Londres, 75379 Paris cedex 08, France { estelle.bongini@sncf.fr, franck.poisson@sncf.fr } Abstract A 2D BEM / FEM ground structure vibration interaction model is used to estimate the influence of building types on the building response to railway vibration. The results are expressed in terms of transfer function between free field ground vibration and building floor vibration. The ground / structure model is first experimentally validated from field measurements of ground and building vibration performed near railway and then used in a parametric study to evaluate the influence of floor thickness and span, as well as the thickness of buried walls and façades. Real railway ground signals are finally used to compare the results in terms of currently used vibration exposure indicators. Keywords: ground borne vibration from railways, building vibrational response. 1 Introduction A vibration measurement campaign has recently been carried out in France in order to compare freight trains and passenger trains (Vibsolfret project). Ground vibration measurements have been performed in free field at different distances from the tracks as well as vibration measurements in a building close to the train tracks. A dozen of passages for each train type were measured so that statistical data could be calculated. Having all these data for one ground-building configuration, a decision was made to try to extend the measured results to other building types by calculation, using a 2D ground structure vibration interaction model (MEFISSTO software) developed at CSTB. In the first part (section 2) of this paper, the model is described, the main parameters given and the model experimentally 1
validated. A parametric study is then performed in section 3 in order to extend the measured results to other building types by calculation and evaluate the influence of floor thickness and span, as well as the thickness of buried walls and façades. In the last part (section 4), real railway ground signals are used to express the results of the parametric study in terms of currently used vibration exposure indicators and to show the influence of building structures on people s perception and annoyance. 2 Model used; experimental validation 2.1 Model used A BEM FEM ground structure interaction model either 2D or 2D ½ has been developed at CSTB (MEFISSTO software, see [1,2]) in order to estimate the effect of environmental ground vibrations on buildings. The usual configuration tested consists of a half space ground (BEM approach) and a simplified building (FEM) with building elements either underground or above ground (see example in section 2.2). Three typical ground types are often used at CSTB, with the parameters shown in table 1 below. Table 1 Typical ground parameters Young modulus MPa Density kg/m3 Shear velocity m/s Softer ground 50 1400 100 Standard ground 200 1600 200 Harder ground 800 1800 400 2.2 Experimental validation 2.2.1 Ground propagation MEFISSTO has been used here in a 2D ½ configuration composed of a free half space ground (no structures) and an infinite uncorrelated line source (1N/m force). The vibration acceleration levels at different distances from the line source were calculated; then the free field ground vibration level (narrow band spectrum) measured at 6m from the track during the measuring campaign were used to calibrate the calculated results. The 1/3 octave band acceleration spectra measured and calculated at different distances from the tracks (3, 6, 9, 12, 17 et 20m) are given in figure 1, showing that the attenuation with distance is well modeled with MEFISSTO using the standard ground parameters (see table 1) over a 1/3 octave 12 200 Hz frequency range. 2.2.2 Transfer function between free field ground and building MEFISSTO has been used here in a 2D configuration composed of a half space ground and the building structure (given in figure 2), simplified compared to the real building but with elements having the right length and thickness: all the elements are made of 20cm thick reinforced concrete, except the façade (embedded part and part above ground), which was particularly thick (40 cm) in the (old) building measured. The building is located at 6m from the tracks (excitation force shown in figure 2). The transfer function between free field ground vibration level at the same distance as the building (configuration with no structures) and first floor vibration level in the building (location close to the center of the floor) was calculated 2
and compared to the transfer function obtained from measured spectra (free field ground at 6m and first floor in the building). Figure 1 Attenuation with distance calculated using MEFISSTO (continuous line) and measured (dotted line). Figure 2 Simplified 2D geometry of the measurement site The results given in figure 3 show that (i) the orders of magnitude are acceptable, (ii) the best calculated results are obtained using the standard ground parameters (which confirms the results of section 2.2.1) and (iii) the calculated transfer function frequency spectra are much rougher than the measured one. These rather acceptable results justify the use of a 2D model in the parametric study presented in the next section. 3
Figure 3 Transfer function between free field ground vibration (at the same distance from the tracks as the building) and first floor vibration in the building 3 Parametric study 3.1 Method In this section, a parametric study is performed in order to evaluate the influence on the floor vibration (at a location close to the center of the floor), of floor thickness and span, thickness of buried walls and façades as well as distance of the building from the tracks. The results are expressed here in terms of floor vibration velocity levels estimated from the calculated transfer functions obtained with MEFISSTO and real (measured) ground vibration spectra. 3.2 Parametric study 3.2.1 Distance from the tracks The building originally at 6m from the track is virtually moved away at 20m. Assuming the transfer function between free field ground at the same distance and building is the same at 20m, the floor velocity level at 20m is estimated from the transfer function measured at 6m and the ground vibration measured at 20m. Ground and floor vibration levels are given in figures 4 and 5 respectively. The results show that there is an attenuation of more 20 db at 20m from the tracks. The frequency shift (of about 20 Hz toward the low frequencies) observed in Figure 4 between the 2 ground vibration spectra disappears in figure 5, showing a spectrum shape imposed by the floor modal response. 4
Figure 4 Free field ground levels measured at 6 and 20m from the tracks Figure 5 floor levels in the building measured at 6m and estimated at 20m from the tracks 3.2.2 Floor thickness The floor thickness originally of 20cm is virtually either decreased (15cm) or increased (25cm). The floor vibration levels obtained are given in figure 6, showing that (i) the calculated floor velocity levels are close to the measured ones and (ii) the floor modal response (20 cm) is frequency shifted compared to the measured one (result expected with the 2D model used) and this shift varies with the floor thickness. 3.2.3 Façade thickness The façade (embedded part and part above ground) was particularly thick (40 cm) in the (old) building measured. Its thickness is now virtually decreased to a more current one (20cm). The floor vibration levels obtained are given in figure 7, showing that (i) the floor response is 5
frequency shifted (because of the change in its boundary conditions) and (ii) the vibration levels are more than 10 db higher. Figure 6 Floor velocity levels in the building either measured or estimated for different floor thicknesses Figure 7 Floor velocity levels in the building either measured or estimated for different façade thicknesses 3.2.4 Floor span The floor span, originally of 3.4m is virtually doubled (6.8m), with the same floor thickness. The floor vibration levels obtained are given in figure 8, showing that the floor response is 6
frequency shifted (toward the low frequencies) and its vibration levels increased (the floor becomes more supple); these two results were expected. Figure 8 Floor velocity levels in the building either measured or estimated for different floor spans 4 Results expressed in terms of vibration exposure indicators 4.1 Method Without any proper standards / regulation on vibration from railways and its effects on people in France, the Norwegian standard NS 8176.E [3], in accordance with ISO 2631-1 and very well documented, was used to express the results of the parametric study in terms of vibration exposure indicators and corresponding annoyance. The procedure for calculating the Norwegian vibration exposure indicators is the following: (i) either the acceleration time signal or the velocity time signal is used (ii) the signal is filtered in 1/3 octave band in order to get N filtered time signals s i corresponding to the N 1/3 octave band considered (iii) the time dependant rms value (with time window τ =1s) of each signal s i is calculated according to 1 / 2 t 1 2 si,rms( t ) = si ( t ) dt (1) τ t τ (iv) at each instant t, the N 1/3 octave signals s i,rms (t) are then weighted using the w m weighting (expressed in 1/3 octave values w i ) defined in standard ISO 2631-2; the weighted signal s w,rms (t) is calculated according to s w,rms 1 / 2 2 ( t ) = si,rms( t ). wi (2) i 7
An example of weighted signal s w,rms (t) is given in figure 9, calculated from a measured ground acceleration signal generated by a passenger train at 6m from the tracks. Figure 9 Weighted signal s w,rms (t) calculated from a measured ground acceleration signal generated by a passenger train at 6m from the tracks For each train passage (index i), the above procedure was first applied to the floor velocity signal measured in the real building; then, the variations in the 1/3 octave floor vibration spectra calculated using MEFFISTO in the parametric study were added to the 1/3 octave w m weighting (step (iv) in the procedure) in order to get the modified weighted signal v w,rms,i (t) (which would be obtained in the modified building configuration).finally and according to standard NS 8176, the maximum value v w,max,i during the train passage was calculated. In a second step, the same procedure was applied to all the floor signals measured for all the freight trains (a dozen) and the mean value as well as the standard deviation σ calculated in order to get the 95% confidence value used in the Norwegian standard. 4.2 Results From the real building measured (located at 6m from the track, with a façade 40cm thick and a first floor 20cm thick and 3.40m long), the 4 configurations presented in the parametric study were considered: (i) same building at 20m from the tracks, (ii) building at 6m with modified façade thickness (20cm), (iii) building at 6m with modified floor thickness (15 and 25cm), (iv) building at 6m with modified floor span (6.80m). The exposure indicator obtained for all the building configurations are given in table 2. Table 2 Indicator obtained for all the building configurations calculated Real building measured Same building at 20m Façade thickness of 20cm Floor thickness of 15cm Floor thickness of 25cm Floor span of 6.80m mm/s 0.11 0.04 0.89 0.26 0.11 0.56 8
Following the Norwegian regulation, the results given in Table 2 show that: (i) the building at 20m from the tracks corresponds to a very good vibration environment (Class A of the Norwegian standard) where residents should normally not notice the presence of vibration, (ii) the real building and the building with floor thickness increased correspond to a relatively good vibration environment (Class B) where a low percentage (less than 10%) of the residents are expected to be annoyed, (iii) the building with reduced floor thickness corresponds to recommended values for a new building (Class C) where already about 15% of the residents are expected to be annoyed, (iv) the building with long floor span corresponds to recommended values for an existing building (Class D) when the cost for improvement is prohibitive; about 25% of the residents are expected to be annoyed, and (v) the building at 6m with smaller (but common) façade (20cm) corresponds to values higher than the recommended ones; the building must therefore be treated to at least reach Class D because more that ¼ of the residents are expected to be annoyed. 5 Conclusions From a whole set of railway vibration data, measured for one ground / building configuration, the results were extended to other ground /building configurations by calculation using a 2D BEM/FEM ground structure interaction model. The modified floor vibration was estimated for each configuration and expressed first in terms of 1/3 octave vibration velocity spectra and then in terms of vibration exposure indicator and corresponding annoyance; the example of the Norwegian standard NS 8176, in accordance with ISO 2631-1 and well documented, was used but other standards such as the similar German standard DIN 4150-2 or the British standard BS 6472-1 based on vibration dose could have also been used. The data presented were obtained from freight train passages; passenger trains usually faster than freight trains would have given very similar results. Acknowledgments The authors gratefully acknowledge the support of this work by the French Agency ADEME (Agency for Environment and Energy Management) References [1] Jean P., Boundary and finite elements for 2D soil-structure interaction problems, Acta Acustica 87, 56-66, (2001) [2] Jean P., Guigou-Carter C., Villot M. A 2.5D BEM model for ground structure interaction. Building Acoustics 11(3), 157-163. (2004) [3] Norwegian Standard NS 8176.E: Vibration and shock Measurement of vibration in buildings from land based transport and guidance to evaluation of its effects on human beings, Norway, 2005. 9