Loading...
Please wait, while we are loading the content...
Similar Documents
International LS-DYNA ® Users Conference Occupant Protection June 10-12 , 2018 1 LS-DYNA ® Belted Occupant Model
| Content Provider | Semantic Scholar |
|---|---|
| Author | Kang, Stephen Dongmin Chen, Cong Guha, Shekhar Paladugu, Manjeera Sundaram, Murugan Ramasamy Gade, Lalitha Zhu, Fuchun |
| Copyright Year | 2018 |
| Abstract | The seat belt is one of the most critical components in automotive crash safety. The three-point belt system has been around for fiftyeight years, belt pretensioners for thirty years and retractor torsion bar load limiters for eighteen years. Though the belt system has been around for so long, CAE correlation to physical test is still limited and far from having high confidence predictive capability. There are numerous CAE parameters and all their values have to be carefully determined, to represent the physics of crash testing and for the CAE models to have good predictive value. How well the belt system is modeled in CAE can directly affect occupant correlation and our predictions. There is an increased need to correlate and predict occupant results in various crash modes, from the ever changing USNCAP, FMVSS and IIHS to the other NCAP updates from outside the United States. Through this study, the belt modeling has been greatly improved leading to much better occupant CAE results. Like airbags, one of the challenging parts of modeling the seatbelt is the modeling of the fabric and other related devices (like pretensioners). To validate CAE models, different levels of component and subsystem testing are required. These test procedures, setups and fixtures have to be carefully designed to create a controlled environment, which will determine the properties of the components in focus. What we have done in this particular study is to follow certain precise steps to exclusively determine the seatbelt properties and parameters, which can later be applied with fullest confidence. Introduction As the abstract above briefly outlines, the purpose of this entire study was to improve the modeling of seat-belts in occupant CAE. In such an effort, not only is it important to determine accurately the fabric material properties of the belt but also to determine energy managing retractor (EMR) and pyrotechnic-pretensioning characteristics, together with belt friction coefficients at D-rings and buckles and as they slide over dummies. The broader aim is that once these parameters are discovered and verified through “controlled environment tests”, they can be used later in different models with the fullest confidence, instead of each individual analyst continuously tuning them to achieve a desired result in a specific situation. With this goal in mind, we have used a progressive and a systematic approach to improve occupant CAE models in frontal crash scenarios. There were essentially three steps that we took and they are: • Step 1 Determination of belt fabric properties via dynamic webbing test; an FE model built & correlated later. • Step 2 Rigid Body Belted Sled Test (RBBS); CAE correlation using belt properties from Step-1 above. • Step 3 Sled Test with Belted Hybrid III 50th on rigid seat; CAE correlation using parameters from Steps 1&2. It can be noted that the first of the three above can be termed a “component level” test, while the next two can be called a “sub-system level” test. Our purpose was to first determine all the properties and parameters possible from the tests at each level and then carry them over to the next level with practically no alteration (if possible at all), in order to check out how the model correlations automatically turned out to be at each subsequent step. With that in mind we move on to the descriptions of the tests themselves at each step, the results obtained from them and how our attempted model correlations turned out to be for each one of them. 15th International LS-DYNA® Users Conference Occupant Protection June 10-12, 2018 2 Step 1: Dynamic Seat_Belt Webbing Test (using an Impactor) and Equivalent Model Correlation The setup for this test is shown in Figure 1. The dynamic test procedure in our study simulates the average loading rate under NCAP test conditions. The fixture as shown in Figure 1 had been used at the Ford Component Lab. Two load cells on each side of the webbing were attached, close to the fixed ends. A pretension in the range 4-5 lbf was applied to the webbing before the test. The impactor displacement was calculated by using the double integration of the impactor deceleration. A switch was used to signal the contact between the impactor and the webbing. Four feet of webbing was needed for each test and four tests were conducted to establish the average properties of the webbing. The fixture deformations in the setup itself turned out to be insignificant. Figure 1: Fabric Test setup An Impactor of mass of 201.7-lbs moved initially at 7.3-mph toward the middle of the webbing. The webbing sample had 5-lbf of pretension. Webbing load cells were used on both sides of the sample. The mass and speed were chosen such that the webbing load can rise to about 2000-lbf in 40-msec. This corresponds to the average NCAP webbing loading rate without torsion bars. A finite element model of the same test configuration was created, as shown in Figure 2 below. A fabric material model based on past experience was initially used to start the simulation process. Through the process of correlation of the CAE model and the test results (using the seatbelt force and impactor displacement of each respectively), we were able to accurately determine the different parameters of the seatbelt material. Once determined, the seat-belt material properties were never altered in future steps of this study. The various material properties have been shown in Table-1 on the following page. Figure 2: Seatbelt webbing test CAE model. Note the 8-element triangular mesh in the lateral direction. 15th International LS-DYNA® Users Conference Occupant Protection June 10-12, 2018 3 Figure 3 shows the belt cross-section force as recorded during the test by the load cells at the two ends of the belt and the impactor displacement, both being with respect to time. Figure 4 shows the average belt force plotted with respect to impactor displacement. The “green curves” in all of these are equivalent model simulation results. As we can see, the fabric model showed remarkable correlation with respect to the test results, differing slightly only in the “unloading phase”. The webbing test correlation provided significant direction in finalizing the seatbelt material properties like Young’s Modulus, “force vs. strain” characteristics in loading and unloading in both the longitudinal and lateral directions, damping, liner properties, etc. Our hope was that this could in turn improve the “system level” CAE models such as those for RBBS, regular sled and barrier. Figure 3. Seatbelt Force & Impactor Displacement. Note: The “green curves” are from the model. Figure 4. Force vs Impactor Displacement. The “green curve” is from the model. In the following two pages we have displayed what the final “Section Properties” and “Fabric Material Model” we chose were, to achieve the above correlations to test. Obviously, these are in LS-DYNA format. Not only are the necessary curves shown overlaid on each other but the values themselves have been tabulated to make it easy for those who wish to use this very fabric material model directly in their own models to try out. Here we would like to add that the fabric belt we modeled had “8” triangular elements across the width, instead of the default “4” which most people normally use. We have found that the performance of an 8-element belt model is superior to that of a 4-element model. Currently LSPP gives the option of not only generating an 8-element belt when we start from scratch but also gives us the possibility of “refining” our existing 4-element belts to 8element belts. 15th International LS-DYNA® Users Conference Occupant Protection June 10-12, 2018 4 Figure 5. Belt Fabric Material Model “Stress vs. Strain” Curves (as per LS-DYNA requirement for Fabrics) Figure 5 shows the belt fabric loading and unloading curves in the longitudinal and lateral directions of the belt. It is to be noted that LS-DYNA expects “stress vs. strain” curves for the fabric material model (not “force vs. strain”). We can convert the same curves back into a “force vs. strain” curve by simply scaling the “Y” by the cross-sectional area of the belt. Since a typical belt has a width of about 47-mm and a thickness of 1.2-mm, scaling the “Y” by “47.0*1.2” (i.e., by 56.4 mm**2) would give us each of the above back as a “force vs. strain” curve. The latter can then be used for “single line segment belts” in a “mixed belt system” (requiring only the “longitudinal curves” from the above and not the “lateral ones”). The “Section Properties and Material Models” of the fabric used are given below: *SECTION_SHELL $ $# secid elform shrf nip propt qr/irid icomp setyp 11 9 0.0 2 1.0 0 1 1 $# t1 t2 t3 t4 nloc marea idof edgset 1.2 1.2 1.2 1.2 0.0 0.0 0.0 0 $# bi bi bi bi bi bi bi bi 90.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 *MAT_FABRIC $ $# mid ro ea eb ec prba prca prcb 8 1.062E-6 2.0 2.0 2.0 0.3 0.3 0.3 $# gab gbc gca cse el prl lratio damp 0.769 0.769 0.769 1.0 0.3 0.3 0.1 0.2 $# aopt flc fac ela lnrc form fvopt tsrfac 0.0 0.0 0.0 0.0 0.0 14 0.0 0.0 $# unused rgbrth a0ref a1 a2 a3 xd xl 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 $# v1 v2 v3 d1 d2 d3 beta isrefg 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 $# lca lcb lcab lcua lcub lcuab rl 1000058 1000059 |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://www.dynalook.com/conferences/15th-international-ls-dyna-conference/occupant-protection/ls-dyna-r-belted-occupant-model |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
