Fluid Structure Interaction V 49 Fluid-tructure interaction between train and noie-reduction barrier: numerical and experimental analyi M. Belloli, B. Pizzigoni, F. Ripamonti & D. Rocchi Mechanical Engineering Department, Politecnico di Milano, Italy Abtract Thi paper deal with the fluid-tructure interaction related to the tranit of railway vehicle in proximity to noie-reduction barrier. During train tranit a preure wave i generated and conequently, due to the barrier large urface area, non-negligible load on the upport element arie. Thi problem applie in particular to high-peed railway line and become critical when the train-barrier gap i mall. Moreover, the preence of lateral wind can increae the load acting on the tructure, influencing the deign of the upport upright. The problem ha been numerically and experimentally evaluated. A preliminary tudy conited of Computational Fluid Dynamic (CFD) imulation, reproducing the train and barrier geometrie and the relative motion with the liding meh technique. The condition of both the abence and preence of lateral wind were invetigated. The preure ditribution along the barrier wa analyed for different condition. Then, experimental tet were performed in the wind tunnel at the Politecnico di Milano. Different wind peed and expoure angle were conidered in order to define lateral wind contribution. The preure on the barrier wa calculated at everal point by mean of preure tap. Finally, the reult obtained were compared with the data collected during an experimental campaign performed on the Rome-Naple high-peed railway line. Keyword: train tranit, noie reduction barrier, CFD, wind tunnel, preure ditribution. 1 CFD code tudy A Computational Fluid Dynamic (CFD) code wa ued to perform imulation of the flow field induced by the train tranit in the proximity of a noie reduction doi:1.2495/fsi951
5 Fluid Structure Interaction V barrier [1 3], in till air or with lateral wind. Simulation were performed by mean of a commercial code on a three dimenional domain, uing an unteady approach and the liding meh technique. The computational domain wa plit into two part, one containing the noie reduction barrier and the urrounding, which i conidered at ret, and one containing a mall part of air around the train, moving together with the train at the train velocity. The moving domain lide on the till domain along interface plane where the grid point belonging to the different ub-domain do not coincide and their relative poition change with time. Figure 1 how a ketch of the computational domain at ret and only the train wall of the moving domain, at the intant of time when the train head reache the midpoint of the noie reduction barrier. The boundary condition adopted in the imulation are alo illutrated in the ame figure. Two different et of boundary condition were conidered to perform imulation with and without lateral wind. For the imulation without lateral wind, atmopheric preure condition were aumed at the preure inlet and outlet boundarie of the domain and the computed flow field wa due to the relative motion between the train and the till barrier. For the imulation with lateral wind, a tatic preure value wa precribed at preure inlet and at preure outlet, o that the preure difference originated a flow in the computational domain, directed tranverally with repect to the train direction of motion. No lip condition were ued for the ground wall, while ymmetry condition were aumed for all the other boundarie in both imulation i.e. with and without lateral wind. The reference frame ha the x-axi aligned with train moving direction, the y- axi aligned with the lateral wind, if any, and the vertical z-axi directed upward. Symmetry Preure outlet Symmetry Wall Preure inlet Symmetry Figure 1: Computational domain, boundary condition and reference frame.
Fluid Structure Interaction V 51 The train geometry i reproduced by a imple prim repreenting the volume of the whole convoy, neglecting all the particular in the under floor region (bogie and auxiliary part) and on the roof, keeping account of the variation of the train head ection only. The length of the train i choen in order to prevent the train tail from entering the domain, ince only the effect induced by the train head are conidered. 1.1 Simulation without lateral wind Figure 2 how the preure time hitorie computed during a train tranit at 3 km/h on different point belonging to a vertical ection of the noie reduction barrier poitioned at x=8 m from the train entrance into the till domain. The conidered point are poitioned at z=.5; 1; 1.5; 2; 2.5; 3 and 4 m from the top of rail (TOR) level. The overpreure recorded at different height from the ground i compared, howing how the peak value become lower the higher i the quote from the ground. Thi reduction i due to the effect of the contraint acted by the barrier on the air trapped between the lateral wall of the train and the barrier itelf and moving toward the upper edge of the barrier. An overpreure peak of more than 5 and a uction peak of almot -9 are computed when the train head reache the invetigated ection poition. Before the overpreure peak a poitive preure i already preent, due to the ma of air dragged by train. 6 4 2-2 -4 p-x8m-z5m p-x8m-z1e5m -6 p-x8m-z1m p-x8m-z2e5m p-x8m-z2m -8 p-x8m-z3m p-x8m-z4m -1.2.4.6.8 1 Figure 2: Time hitorie of the preure computed by the CFD imulation, at different height from the ground.
52 Fluid Structure Interaction V Figure 3: Contour plot of the tatic preure on the noie reduction barrier. Figure 4: Static preure contour plot. Figure 3 how the contour plot of the tatic preure field on the noie reduction barrier. The red zone i characteried by overpreure and i poitioned jut in front of the train head; the blue one i characteried by uction and i poitioned aide the train head.
Fluid Structure Interaction V 53 The reduction of preure moving upward from the ground level can alo be appreciated a well a the extenion of the poitive preure zone in front of the train. In Figure 4, a contour plot of the tatic preure on an horizontal plane poitioned at a height of 2.5 m from the TOR i illutrated, decribing the lack of ymmetry in the preure field around the train a induced by the barrier. The air i compreed by the train head on the barrier, giving rie to the overpreure zone on the barrier and to a flow in the channel between the lateral wall of the train and the barrier itelf. The preure time hitorie computed along the barrier in different ection along the x-axi at 1 m height from the TOR are compared in Figure 5. The conidered ection are equally paced along the x axi (train direction) every 1 m, tarting from x=7 m. The time hitory trend are very imilar and a time lag eparate the different reult, according to the time required by the train to reach the different poition along the line. An increae both in the poitive and the negative peak value i viible, baically due to the unteadine of the imulation, which continue to change the boundary condition while the train i moving. 6 4 2 p-x1m-z1m p-x7m-z1m p-x8m-z1m p-x9m-z1m -2-4 -6-8 -1.2.4.6.8 1 Figure 5: Preure time hitorie on different ection along the barrier at 1 m from the TOR. 1.2 Simulation with lateral wind Figure 6 how the net preure time hitorie, computed on the ame point of the x=8 m ection conidered in Figure 2, for a train peed of 3 km/h and a
54 Fluid Structure Interaction V 15 1 5 p-x8m-z5m -5 p-x8m-z1e5m p-x8m-z1m p-x8m-z2e5m -1 p-x8m-z2m p-x8m-z3m p-x8m-z4m -15.1.2.3.4.5.6 Figure 6: Net preure time hitorie at different height on the barrier ection at x=8 m. 15 1 5 p-x1m-z1m p-x7m-z1m p-x8m-z1m p-x9m-z1m -5-1 -15.1.2.3.4.5.6.7 Figure 7: Net preure time hitorie on different point along the noie reduction barrier at 1 m height from TOR.
Fluid Structure Interaction V 55 lateral wind of 3 m/. In thi cae, the barrier, which i ubjected to the wind load and how a poitive mean value of the net preure between upwind and downwind urface, experience a reduction of preure, taking advantage of the wind hielding effect during the train tranit. The peak value of the negative preure are lightly higher than thoe reached during the imulation without lateral wind. An almot null preure condition i reached after the negative peak at the conidered barrier ection, due to the wind hielding effect of the train body. An analyi of the preure ditribution along the barrier i illutrated in Figure 7, where the net preure time hitorie at the ame poition conidered without lateral wind in Figure 5 are compared. The phenomenon repeat itelf in all poition, conidering the time delay, and an ocillating negative preure trend appear after the negative peak i reached. 2 Wind tunnel tet Wind tunnel tet [4] were performed in a 1:1 cale to determine the preure ditribution on the noie reduction barrier during the train tranit. Figure 8 how the complete wind tunnel et-up intalled in the boundary layer tet ection of the Politecnico di Milano wind tunnel. A 1:1 cale model of the EMUV25 train i intalled on a flat ground cenario, with a plitter plate at the height of.3 m from the wind tunnel floor, o a to conider the incoming wind to have a block vertical profile of the mean velocity. Figure 8: Tet et-up intalled on the wind tunnel turntable. To meaure the preure ditribution due to the train tranit, the noie reduction barrier wa intrumented with 32 preure tap, dipoed according to the ketch reported in Figure 9. A can be noted, the preure tap ditribution i more refined cloe to the train head, poitioned at 1.68 m from the barrier upwind vertical edge (a vertical line i drawn at the train head poition in
56 Fluid Structure Interaction V Figure 9: Preure tap poition on the barrier and train head poition. Table 1: Yaw angle and lateral wind peed at full cale. Angle Train velocity Lateral wind peed 83 m/ m/ 1 83 m/ 14.6 m/ 2 83 m/ 3 m/ 3 83 m/ 48 m/ Figure 9). The preure tap have been intalled at three different height level, correponding to 1m, 2 m and 3 m at full-cale ize. Each preure tap i connected to a high frequency preure canner. The preure ignal have been acquired at a ampling rate of 1 Hz. To imulate lateral wind condition the angle between the wind tunnel incoming wind and the train direction wa changed. Four different yaw angle have been analyed in particular, repectively equal to, 1, 2 and 3. The deg tet correpond to the abence of lateral wind. Auming a train peed of 3 km/h, Table 1 how the correponding lateral wind peed determining, at full cale, the ame angle of the relative velocity of the wind conidered in the wind tunnel tet. The wind tunnel tet and the full cale operating condition are different ince, in the wind tunnel tet, both the barrier and the train are contrained to the ground, i.e. there i not relative motion. Therefore the preure data, meaured on the wall at different location along the direction of the train axi, are, at full cale, repreentative of thoe meaured at fixed point at different intant of time during the train tranit. 2.1 Experimental reult In order to compare the reult obtained for different incoming wind velocitie the preure have been expreed in term of preure coefficient. The free tream velocity i adopted to compute the preure coefficient C P related to each preure tap:
Fluid Structure Interaction V 57 5 4 3 2 Reale 1 m/ 12 m/ 15 m/ 1-1 -2-3 -4-5 6.6 6.7 6.8 6.9 7 7.1 7.2 7.3 7.4 Figure 1: Comparion between wind tunnel tet and full cale meaurement (1 m). 2 p C P = (1) ρ 2 V where p i the meaured preure, ρ i the air denity and V i the wind free tream velocity. From the Cp value it i poible to define the preure at the barrier a a function of the train peed. Uing the following equation: 1 2 p = C ρv (2) 2 P T and by conidering the train velocity V T equal to 83 m/ (3 km/h), a time ditribution can be given to the wind tunnel data by mean of the following equation: x t = (3) VT being t i the time delay between two preure tap and x i the ditance between the ame two preure tap. So it i poible to compare the reult obtained during wind tunnel tet with the reult obtained at full cale on the high-peed line cloe to Anagni-Fiuggi, a hown in Figure 1, where a very good agreement can be een. The comparion between wind tunnel reult and CFD one highlight a good agreement on the value of the poitive peak, while the numerical approach overetimate the negative peak value, poibly due to the very implified train geometry ued for the imulation. Figure 1 how the value obtained at the height of 1 m, full cale, while Figure 11 and Figure 13 are concerning, repectively, to 2 m and 3 m.
58 Fluid Structure Interaction V 5 4 3 Reale 1 m/ 12 m/ 15 m/ 2 1 Figure 11: -1-2 -3-4 -5 6.6 6.7 6.8 6.9 7 7.1 7.2 7.3 7.4 Comparion between wind tunnel tet and full-cale meaurement (2 m). 5 4 3 Reale 1 m/ 12 m/ 15 m/ 2 1 Figure 12: -1-2 -3-4 -5 6.6 6.7 6.8 6.9 7 7.1 7.2 7.3 7.4 Comparion among wind tunnel tet and full-cale meaurement (3 m). The peak-to-peak amplitude of the preure decreae with the height, both in full cale and in wind tunnel tet, a already explained by mean of the CFD code. It i noticeable that there i a time hift between the laboratory condition and the real condition data. Thi could be due to the different hape of the real train and the 1:1 cale model, particularly to the difference in the noe of the real train compared to the model one.
Fluid Structure Interaction V 59 2.2 Lateral wind Lateral wind effect ha been tudied by changing the yaw angle of the whole cenario by mean of the turntable. There are no reference data with lateral wind at full cale, o the laboratory finding have been compared to no wind condition, ee Figure 13. The data are reported a a function of time and, in order to extend the reult to the full-cale value, the reference velocity ha been calculated a: V = V + V (4) 2 2 a t lat where V a i the reference velocity, V t i the train peed and V lat i lateral wind velocity. The lateral wind ha the effect to increae the preure peak, a already pointed out by the analyi of the CFD reult. 14 12 1 8 6 4 2-2 -4-6 -8 Reale v lat = m/ v lat = m/ v lat =14 m/ v lat =3 m/ v lat =46 m/ -1 6.6 6.8 7 7.2 7.4 Figure 13: Comparion between wind tunnel and full-cale meaurement, lateral wind effect at h=1 m. 3 Concluion The preure ditribution due to a train tranit on a noie reduction barrier ha been invetigated through CFD imulation, wind tunnel tet and full-cale meaurement. Full-cale meaurement are the more reliable data but, with thi approach, it i difficult to take into account the lateral wind effect. CFD analyi i able to conider both lateral wind and no lateral wind condition and allow one to have a good inight in the phenomena. The trong implification in the train geometry may be the caue of the overetimated negative preure peak
6 Fluid Structure Interaction V obtained by the CFD reult. Wind tunnel tet on a till train model proved to be a valuable tool to invetigate the preure ditribution not only in abence of lateral wind. Even though in the wind tunnel experiment there i not relative motion between the train and the cenario and the wind-train velocity compoition i not reproduced, the wind tunnel tet data highlight the effect of the lateral wind on the value of the negative preure peak. Reference [1] Copley, J.M., 1987. The three-dimenional flow around railway train. Journal of Wind Engineering and Indutrial Aerodynamic, 26(1), pp. 21-52. [2] Barcala, M.A. and Meeguer, J., 27. An experimental tudy of the influence of parapet on the aerodynamic load under cro wind on a twodimenional model of a railway vehicle on a bridge. Proceeding of the Intitution of Mechanical Engineer, rt F: Journal of Rail and Rapid Tranit, 221(4), pp. 487-494. [3] pech, A.J.G., 1972. Model Study of Windbreak on Railway Bridge. New Zealand Engineering, 27(4), pp. 132-139. [4] Tanemoto, K., 24. Wind tunnel tet on aerodynamic characteritic of train/vehicle in cro wind and effect of wind-protection fence. Railway Technical Reearch Intitute Report, 18(9), pp. 17-22.