SAE Paper 97-0960: APPENDIX 3: Clarifications

  1. The existence of significant errors related to restitution that are present in both the original CRASH (EDCRASH) and original SMAC (EDSMAC) computer programs must be acknowledged. It is common knowledge that the original CRASH (EDCRASH) program underestimates the V in barrier crashes by approximately 10 to 20% at 30 MPH and by a greater amount at lower speeds as a result of the fact that restitution is completely ignored. The original SMAC (EDSMAC) form of simulation of restitution is crude, with rarely changed inputs, and it cannot rationally be expected to produce reliable and accurate combinations of dimensional recovery and partial return of absorbed energy for all vehicles under all collision conditions. In fact, its under-prediction of structural recovery produces a similar range of underestimates of V. Thus, the status quo regarding restitution effects in existing computer programs is very difficult to defend on a logical basis.
  2. The ranges of errors (underestimates) indicated in the paper for damage-based V values from the original CRASH and SMAC programs, that are produced either by a total neglect of restitution (CRASH) or by under-prediction of structural recovery (SMAC) are intended to provide the reader with approximate measures of the practical significance of the effects of restitution.
  3. On the basis of Figures 2, 6A, and A2 as well as the closely related SAE 861894 it should be clear that is set to 1.000 whenever the calculated value exceeds 1.000. Therefore any concern about an "infinite limit" for is unfounded.
  4. The modeled value of , at zero residual crush, is not necessarily equal to 1.000. Rather, it is defined by equation (16):

    Results of applications of equation (16) to test data where the limiting value of was substantially less than 1.00 are shown in Figures 13 and 14. In Figure 13, the restitution coefficient for the Escort never exceeds 0.314. In calculations related to Figure 13 performed by one reviewer the value of was incorrectly set equal to 1.00.

  1. Energy is absorbed whether or not a structure is elastic. It is the extent of return of the absorbed energy that distinguishes elastic from inelastic behavior.
  2. The term "damage" in this paper is used to refer to the generally accepted residual crush, as opposed to any cosmetic disfigurement of "fragile body parts".
  3. The general form of the A,B crush coefficients used in the CRASH(EDCRASH) program implies an effective elastic deformation range, in terms of full dimensional recovery, equal to A/B.
  4. The lack of exact repeatability of individual measurements in crash tests that are performed under identical test conditions acts to produce ranges of measured responses rather than single values. For this reason, rigorous measures of reconstruction accuracy must be based on the mean experimental measurements for a given set of test conditions.
  5. Available experimental measurements of restitution behavior over ranges of impact speed (e.g., Figures 12, 13, and 14) include only single measured values at the individual test conditions. Clearly, progress toward a rigorous and complete validation study is data-limited at the present time.

We welcome any questions, comments, corrections or requests for further information. We can be reached by e-mail at

| ARreview | News! | McHenry Software Homepage|