IMPROVING CRASHWORTHINESS IN RAILCAR AGAINST ROLLOVER Sarmad Riazi, Mohammad Mahdi Feizi, Parisa Hosseini-Tehrani Center of Excellence in Railway Engineering, School of Railway Engineering, Iran University of Science and Technology, Tehran, Iran E-mail: hosseini_t@iust.ac.ir Received April 2012, Accepted October 2012 No. 12-CSME-47, E.I.C. Accession 3367 ABSTRACT Roof strength is critical to survival in rollover accidents in railcars. In this task using LS-DYNA software different scenarios are examined in order to strengthen railcar roof and to improve the crashworthiness features against rollover. The performance of each scenario is investigated through presenting the results of crushing behavior and energy absorption versus displacement response. Eventually, the best solution providing the highest ratio of Energy absorption per Mass among the studied model is presented. Keywords: rollovers; roof strength; railcar; crashworthiness. AMÉLIORATION DE LA RÉSISTANCE À L IMPACT DES WAGONS DE TRAIN LORS D UN RENVERSEMENT RÉSUMÉ La résistance du toit des wagons de train est essentielle à la survie des passagers lors d un renversement. Pour cette recherche, différents scénarios sont examinés à l aide du logiciel LS-DYNA dans le but de renforcer le toit des wagons et d améliorer la résistance à l impact lors de renversement. La performance de chaque scénario est étudiée à travers les résultats du comportement de résistance à l écrasement et l absorption d énergie versus la réponse de déplacement. Nous présentons la solution la meilleure, celle qui donnait le plus haut coefficient «d absorption d énergie par masse» parmi les modèles étudiés. Mots-clés : renversement ; résistance du toit ; wagon de train ; résistance à l impact. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 383
1. INTRODUCTION Roof collapse and buckling are the major cause of rollover head and neck injuries to both ejected and nonejected occupants. In rollover occupant compartment integrity and good resistance against deformation are essential. In railway industry testing is extremely expensive, new models of railway vehicles circulate and new passenger cars are put into service without being subjected to experimental testing. However, virtual testing and numerical modeling may be acceptable in the design cycle. On the other hand, various performance standards concerning crashworthiness of motor vehicles in case of rollover accidents have been introduced. Standards like FMVSS 216 (for cars), FMVSS 220 and ECE r66 (for buses) are accompanied by numerous publications on the subject. However, currently, there are few standards and published works addressing the desired behavior of passenger railcars in rollover. Due to similarity between structure of transit buses and railcars, and in absence of a comprehensive rollover test standard for the latter, in a study by Cuartero, J. et al. [1], the European standard ECE r66 has been implemented to study the crashworthiness of a metro car. Although the above-mentioned study proves such railcars to be exceedingly safe when tested by this standard, an investigation on the Waterfall train crash poses concern on safety of the existing fleet, since in their design rollover is not generally envisaged or catered to (Rechnitzer G. et al. [2]). Inspired by the work of Cuartero et al., a recent paper by the authors of the present work implemented FE software and modified ECE r66 to study the crashworthiness of a section of a passenger railcar. That study also investigated the effectiveness of usuall methods for strengthening the structure, i.e. using thicker components and upgraded steel (Hosseini Tehrani,P. and Riazi, S. [3]). Moreover, in the mentioned study, without any major change in the design of the structure, a certain reinforcing component was installed and proved to be effective. In this work, the objective is to study a wider range of possibilities for improving crashworthiness of passenger railcars. Initially, a variety of reinforcing components are studied without changing the original design of the wagon. Afterward, some changes are made in the profile of main pillars to examine the suitability of alternative designs. Eventually, the best solution providing the highest ratio of Energy absorption per Mass among the studied model is achieved. 2. NUMERICAL MODEL 2.1. Body Parts Based on a number of published works on buses, it is clear that it is computationally cost effective to perform the numerical analysis on a segment of a bus. This method is approved by ECE r66 (Addendum 65: Regulation No. 66, 2006). Therefore, According to the definition given in the standard for a segment, in this paper a section of the railcar is chosen and shown in Fig. 1. The shown sections are later referred to as Base Model and consist of the following parts: 1. The underframe, including longitudinal and lateral beams; 2. The side walls, composed of vertical pillars and horizontal rails; 3. The top flange, connecting the pillars to the roof structure; 4. The roof structure, including arches, rails and bars. 2.2. Initial and Boundary Conditions As illustrated in Fig. 2, a moving rigid wall hits the wagon structure at an angle of 19.5 degrees. The rigid wall has a mass of 2580.75 kg and its initial velocity is considered 5 m/s. 384 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
Fig. 1. Main dimensions of the chosen section (in meters), on which analyses are performed. The wagon is assumed to be stationary and the shell elements at the lowest part of the underframe have no degrees of freedom. This is to avoid the modeling difficulties of the tilting bench method (Addendum 65: Regulation No. 66 [4]) and to provide pure crash event resulting in shorter simulation time. Fig. 2. The rigid wall hitting the wagon section. 2.3. Elements The CAD model is based on a structure used by Wagon Pars Company for passenger railcar. FE model of this section is made by 16107 four-node shell elements with Belytschko-Tsay formulation. 3. SIMULATION SCENARIOS 3.1. Reinforcing Components Table 1 shows different phases of the simulations. Analysis 0 represents base model, i.e. original railcar as designed by the manufacturer. Analyses 1 through 3 are designed to investigate the effect of thickness change of the crossbar, a reinforcing component that is added to the base model as shown in Fig. 3. In these analyses the thickness is increased by the increment of 1 mm. the thickness of crossbar used in the first model is considered 2 mm. Analysis 4 introduces another component the bottom stiffeners, which strengthen the joints between the underframe and the pillars as illustrated in Fig. 4. It is then followed by analysis 5 that studies another type of stiffeners with the same area but different dimensions as shown in Fig. 5. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 385
Fig. 3. The crossbar mounted on the structure. Fig. 4. The stiffeners type 1. Fig. 5. The stiffeners type 2. Analysis 6 addresses the use of top stiffeners along with the crossbar type 1, to stiffen the pillar-crossbar joints (Fig. 6). Further, analysis 7 deals with the use of the top stiffeners and crossbar type 3. Also, analysis 8 studies the combined effect of the crossbar, bottom and top stiffeners as indicated in Fig 7. Fig. 6. The top stiffeners and crossbar. 386 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
! Analysis No. Crossbar thickness Bottom stiffeners Top stiffeners Reinforcing box Component Reinforcing plates Pillar type Material Number of roof arches 0 (base model)! S- Box- St- Al 2 3 4 1 2 profile profile 52 4 5 6!!! 1!!!! 2!!!! 3!!!! 4!!!! 5!!!! 6!!!!! 7!!!!! 8!!!!!! 9!!!!! 10!!!!!! 11!!!!! 12!!!!! 13!!!!!! 14!!!! 15!!!! 16!!!! 17!!!! Table!1!! Table 1. Analysis Scenarios. The concept of a reinforcing box (Fig. 8) is introduced in analysis 9 where it is used along with bottom stiffeners. This configuration is combined with the crossbars in analysis 10. In analysis 11, the combined effect of bottom stiffeners and crossbars are investigated (Fig. 9). Analysis 12 addresses the use of Reinforcing Plates and the bottom stiffeners as shown in Fig. 10, while analysis 13 supplements this design by the use of crossbars. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 387
Fig. 7. The combined use of top and bottom stiffeners and crossbar. Fig. 8. The box-shaped reinforcements mounted on the bottom stiffeners. Fig. 9. Bottom stiffeners and crossbar. Fig. 10. The reinforcing plates strengthen the s-profile of the pillars. 388 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
3.2. Alternative Designs In contrast to simulations presented in the former section, in alternative solutions, there will be considerably less fidelity to the original design i.e. pillar profile undergoes some modifications. The last row of Table 1 shows the simulation scenarios for the alternative designs. Analysis 0 is the original model with s-profile used in the pillars. In Analysis 14 the s-profiles of the main pillars are replaced with box type counterparts. In Analysis 15 aluminum box-profile pillars are used. It is also important to determine how number of roof arches affects the crashworthiness. If it is of little importance, then it will be possible to decrease the overall weight of the structure by decreasing number of these members, without compromising the crashworthiness capability of the structure. This is especially useful when the extra weight added by the reinforcing components has to be removed in some way. Therefore, analysis 16 and 17 investigate the effect of decreasing and increasing number of roof arches respectively (Figs. 11 and 12). Fig. 11. Roof structure with 4 arches. Fig. 12. Roof structure with 6 arches. 4. MODEL VALIDATION Due to similarity between a passenger railcar structure and that of a transit bus, the actual rollover test results of an M3 bus was chosen as the means of verification. The experimental test was conducted as part of the ECBOS project in accordance to the ECE R66 (G. Belingardi et al. [5]). To gain confidence on the reliability of numerical models of the present study, an FE model based on the above-mentioned test was constructed. In Fig. 13 time histories of relative displacements of two certain points, which is shown in Fig. 14, are indicated. As it can be seen, the results of the FE simulation are in close accordance with the experimental tests. As Fig. 14 shows the deformed shape of the bus section in actual and FE tests are very similar. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 389
Fig. 13. Time histories of relative displacements. Fig. 14. Comparison between the experimental [5] and FE models. 5. SIMULATION RESULTS Analysis 0 of Table 1 refers to the rollover test of the main model. As it is shown in Fig. 15, while the kinematic energy of the rigid wall decreases, the internal energy increases and the total energy remains almost constant. Moreover, hourglass energy is little. According to ECE r66, the permissible amount of non-physical energy components like hourglass, is maximum 5 % of total energy [4]. In analysis 0, satisfaction of this condition ensures the right energy balance. In addition, for the rest of the simulations in this research, hourglass energy is little and lies within the mentioned criteria. First order reduced-integration elements (which happen to be the very type of elements generally used in explicit solvers) suffer from hourglassing. These elements only have one integration point; because of this the elements can shear without introducing any energy. FEA codes which rely on 1st order reduced integration elements counter this by introducing hourglass energy. In situations where these elements would otherwise shear, this augmented energy keeps this from happening. This is reflected in the reported hourglass energy values. High hourglassing energy is often a sign that mesh issues may need to be addressed. Figure 16 depicts the results of analyses 0 through 3. As seen, the use of crossbar (analysis 1) has significantly increased the absorbed energy. In the absence of crossbars, for any 390 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
Fig. 15. Energy absorption versus time. given distance that rigid wall travels, the right sidewall undergoes more deformation than the left wall, which means less participation of the left wall in energy absorption. But, when the crossbars are present, for the same displacement of the rigid wall, the left sidewall absorbs more energy and the deformation of the structure is more symmetrical which leads in higher overall energy absorption (Fig. 17a,b). It can also be seen that thicker crossbars result in slightly higher energy absorption. It is now appropriate to define Fig. 16. Absorbed energy for analyses 0 to 3. a dimensionless value of β. According to Eq. (1), a ratio of absorbed energy of each model per absorbed energy of base model (EAR) is defined, and then in order to compute the value of β (Eq. 3) this ratio is divided by the mass ratio (MR) which is defined by Eq. (2). Use of dimensionless parameter β facilitates the comparison between suitability of each improvement method. For example as Table 2 indicates, for analysis 1 through 3, even when thicker and heavier crossbars are used, β is increased. internal energy o f each simulation EAR =, (1) internal energy o f base model MR = mass o f each simulation mass o f base model. (2) Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 391
(a) (b) Fig. 17. (a) More involvement from the left wall causes more energy absorption; (b) Early intrusion of the right wall into the residual space causes meager involvement of the left wall. β = EAR MR. (3) By observing the deformation of the base model at pillar-under frame joints, the need to use stiffeners is made clear in Fig. 17(a). In analyses 4 and 5 of Table 1, bottom stiffeners (Figs. 4 and 5) are used and the results could be seen in Fig. 18. It is interesting that stiffener type 2 has remarkably increased the energy absorption in comparison with the type 1, which has the same mass. As Table 2 indicates, these stiffeners have improved parameter β by 22.7 % and 42 % respectively. Figure 19 belongs to analyses 6 through 8 Fig. 18. Energy absorption vs time for analyse 4 and 5. which implement another strengthening component, the top stiffener. Top stiffeners improve the strength of pillar-crossbar joints and are most effective when used in conjunction with the bottom stiffeners (Fig. 7). In such case, β increases up to 71 %. As it is seen in Fig. 20, when bottom stiffeners are used, plastic hinges move upward and at their new location, pillars undergo severe deformation and sudden collapse. Therefore the design of the bottom stiffeners needs to be improved. This leads into the use of Box reinforcements that are used alongside the bottom stiffeners as illustrated in Fig. 8. The Box reinforcements are strengthening components added to the base model at the location of plastic hinges and improve the energy absorption when used with crossbars (Fig. 21). It is seen that β will increase up to 2.13 times the original value. As mentioned earlier, when bottom brackets are used, it is imperative to strengthen the new location of the plastic hinges. Plate 392 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
Model β EAR MR Base model (No. 0) 1.000 1.000 1.000 1 1.283 1.315 1.026 2 1.308 1.356 1.038 3 1.321 1.390 1.052 4 1.227 1.310 1.065 5 1.422 1.510 1.065 6 1.327 1.410 1.058 7 1.387 1.485 1.071 8 1.714 1.827 1.137 9 1.840 2.000 1.080 10 2.134 2.400 1.125 11 1.563 1.720 1.104 12 1.698 1.910 1.123 13 2.227 2.590 1.160 14 2.534 3.026 1.216 15 1.459 1.500 1.028 16 0.964 0.950 0.983 17 1.028 1.048 1.015! Table 2. β value of simulations. Table!2!!! Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 393
Fig. 19. Absorbed energy vs. time for analyses 6 through 8. (a) (b) Fig. 20. Deformed shapes (a) of model 9 and (b) of model 10. Fig. 21. Absorbed energy for analyses 9 and 10. reinforcements in models 12 and 13 (Fig. 10) are another design to eliminate this shortcoming. As Fig. 22 indicates, this design results in energy absorption of 16 KJ or about 2.5 times of the original design when used with crossbars in model 13. 394 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
Fig. 22. Absorbed energy vs. time for analyses 12 and 13. It is useful to know how the original pillar profiles could affect the energy absorption during rollover. According to Table 2, in analyses 1 and 2, the s-shaped profiles have been replaced with box profiles, and instead of the original steel in analysis 2, aluminum has been used. Table 3 states specifications of steel and aluminum used. As Fig. 23 shows, the use of steel box profile has significantly increased the absorbed! Material Density Young s Poisson s Yield stress (Kg/m 3 ) Modulus (Pa) ratio (Pa) Steel 7850 2.1E11 0.3 6.21E8 Aluminium 2700 6.9E10 0.33 2.75E7 Table!3! Table 3. Steel and aluminum specifications. Fig. 23. Absorbed energy vs. time for analyses 14 and 15. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 395
energy, compared with the base model. Also, aluminum box-profiles have improved this feature too. It is interesting that changing the pillar profiles increased β for 2.53 and 1.49 times the original design for steel and aluminum pillars respectively. As mentioned earlier, knowing how the number of roof arches affects the crashworthiness is useful. In analyses 16 and 17 of Table 2, the number has been altered to 4 and 6 arches respectively. Figure 24 shows the absorbed energy for these simulations. β values for analysis 16 and 17 suggest that omitting or adding one roof arch has no significant effect on the energy absorption of the structure. Therefore, it is a good strategy to omit one roof arch to decrease the overall mass of the structure. Fig. 24. Absorbed energy vs. time for analyses 16 and 17. 6. CONCLUSIONS In this work the crushing behavior of a section of the railcar subjected to a moving rigid wall that hits the wagon structure at an angle of 19.5 degrees was studied. Using a number of numerical models the following important conclusions are drawn: The use of crossbar has significantly increased the absorbed energy. The cross bars transmit the forces to the other sidewall and as a result, the deformation of the structure is more symmetrical, which leads in higher overall energy absorption. It can also be seen that thicker crossbars result in slightly higher energy absorption. Bottom stiffeners could remarkably increase the energy absorption. Bottom stiffener type two in comparison with the type one, which has the same mass has improved the parameter β by 42 % against 22.7 %. Top stiffeners improve the strength of pillar-crossbar joints and are most effective when used in conjunction with the bottom stiffeners. In such case, β increases by as much as 71 %. When bottom stiffeners are used, plastic hinges move upward and at their new location, pillars undergo severe deformation and sudden collapse. Therefore the design of the bottom stiffeners needs to be improved. This leads into the use of box reinforcements that are used alongside bottom stiffeners. The Box reinforcements are strengthening components added to the base model at the location of plastic hinges and improve the energy absorption when used with crossbars. In this way β will increase up to 2.13 times its original value. Bottom brackets are essential to strengthen the railcar and to move the plastic hinges to new locations which results in more energy absorption and less deformation. This design results in energy absorption of 16 KJ or about 2.5 times of the original design when used with crossbars in model 13. 396 Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012
The use of steel box profile has significantly increased the absorbed energy, compared with the base model. Also, aluminum box-profiles have improved this feature too. It is interesting that changing the pillar profiles increased β for 2.53 and 1.49 times the original design for steel and aluminum pillars respectively. It is shown that omitting or adding one roof arch has no significant effect on the energy absorption of the structure. Therefore, it is a good strategy to omit one roof arch to decrease the overall mass of the structure. REFERENCES 1. Cuartero, J., Lizaranzu, M., Castejón, L., Carrera, M. and Dieste, M., Evaluation of passenger railroad car rollover crashworthiness, International Journal of Crashworthiness, Vol. 11, No. 5, pp. 419 424, 2006. 2. Rechnitzer, G., McIntosh, A., Richardson, S., Grzebieta, R. and Jaraie, J., Crashworthiness improvements for light and heavy rail lessons learnt from crash investigations, Center for auto safety report, document ID: NHTSA 2008 0015 0039, 2008. 3. Hosseini Tehrani, P. and Riazi, S., Crashworthiness improvement of a passenger railcar against rollover International Conference on Mechanical Automotive and Aerospace Engineering (ICMAAE 11), Kualalumpur, Malaysia, May 2011. 4. Addendum 65: Regulation No. 66, Uniform technical prescriptions concerning the approval of large passenger vehicles with regard to the strength of their superstructure, pp. 61 95, 2006. 5. Belingardi, G., Gastaldin, D., Martella, P. and Peroni, L., Multibody analysis of M3 bus rollover: structural behavior and passenger injury risk, Proceedings of 18th International Technical Conference on the Enhanced Safety of Vehicles, Nagoya, Japan, 2003. Transactions of the Canadian Society for Mechanical Engineering, Vol. 36, No. 4, 2012 397