Page 1 of 1

Should a CRASH type damage analysis include induced damage?

Posted: Fri Jun 19, 2009 11:26 am
by MSI
2 words of wisdom on CRASH type damage analysis (whatever flavor: EDCRASH, WinCRASH, etc):
  • Be Careful!
In 1988, Tumbus and Smith wrote SAE paper 880072:
  • In the paper they presented some ideas for a standardized protocol to measure damaged vehicles which included special cases like bowing, etc. I recommend the paper for any and all who use damage analysis techniques, whatever flavor of CRASH or damage analysis you use. The standardization of technique is so that anyone measuring a damaged vehicle should produce the same crush measurements.
    For example, in cases of bowing, some extra energy was required to bow the vehicle so the damage measurement protocol should include consideration for the extra energy (or a 'did the vehicle bow?' and 'how much' question should be added and considered in the analysis).
Since publication of the paper there have been some efforts to create a standardized protocol for damage measurement but so far there have been no results.

And note: CRASH was originally created as a preprocessor for SMAC.
  • The mathematical formulation for CRASH is that damage measurements were to be of the direct contact damage only. When NHTSA adopted CRASH damage analysis for the NASS statistical studies they found it was underestimating the DeltaV so they changed the protocol to include the induced damage.
Today, the CRASH clones use the same mathematics and simplifying assumptions as CRASH3 but also include customized crush coefficients. The cottage industry that has grown out of damage analysis has generally continued the use of induced damage in the measurement 'protocol' however remember this is an indication of the limitations of damage analysis.
Be sure to use other techniques such as momentum analysis, simulation, etc. in addition to damage analysis to check and refine your damage analysis results.

Re: Should a CRASH type damage analysis include induced damage?

Posted: Tue Sep 01, 2009 2:03 pm
by brian
Some additional information related to the inclusion of induced crush in damage analysis:
  • 1) The NASS training manual instructs to include induced damage but indicates ‘most other CRASH users measure only direct damage

    2) The CRASH3 Manual:
    • Section 3, 3-32-3-32 indicates
      a general rule has evolved that specifies the inclusion of induced and direct damage in the profile for CRASH computation
      (p 132-137 of the PDF of the CRASH3 Manual)
    • p21-23 of the PDF about CRASH
      not for individual reconstructions!
      note the underlining is by someone at NHTSA.

Re: Should a CRASH type damage analysis include induced damage?

Posted: Thu Mar 26, 2015 3:41 pm
by MSI
Received a followup questions today:
  • Q: I did a Google search of "direct or induced damage" and came across your 2009 response to a question about which measurement is appropriate with CRASH. I was hoping you would answer some questions. When the length of the crush in the damaged are is less than the original length, most of the resources I found suggest that you plug in the original length of the crushed area in CRASH, even if the sum of the crushed zones is less than the original undamaged length
    induced.jpg
    induced.jpg (52.6 KiB) Viewed 2182 times
    • I was curious what the crash program is doing with the field length and the original length?
    • Does CRASH simply increase the width of each crushed zone proportional to the difference between the original length and the crush length?
    • For example, if the original length of the front end of the car shown is 25% greater than the sum of the crush zones, does crash multiply the width of each zone by the same factor (1.2) in order to increase the sum of the effective crush zone width to the actual undamaged width?[/b]
A: In answer to your questions:
  • I was curious what the crash program is doing with the field length and the original length?
    • Nothing. For CRASH damage analysis the equations only uses the user entered ‘crush damage width’ or whatever it is called in whatever flavor of the CRASH program is being used.
  • Does CRASH simply increase the width of each crushed zone proportional to the difference between the original length and the crush length?
    • Since CRASH simply uses whatever damage width the user enters, the answer to the question is essentially YES, although CRASH does not check/care about the overall width or length of the vehicle as compared to the damage width.
      So it isn’t increasing or decreasing….it is simply calculating the segment width based on user entered width.
      But I guess that is essentially ‘increasing the width of each crushed zone’ based on crush length with no consideration of original length (the user has done that by entering the overall length rather than the actual crush length)
      Some other flavor (versions) or CRASH damage analysis may do a check of the damage width to the vehicle width or length and some even allow more than 2,4 or 6 'damage measurements'. However the basic calculation procedure is to calculate the area of damage (below i have posted the 2,4 & 6 point equations from the CRASH3 manual
  • For example, if the original length of the front end of the car shown is 25% greater than the sum of the crush zones, does crash multiply the width of each zone by the same factor (1.2) in order to increase the sum of the effective crush zone width to the actual undamaged width?
    • NO. It simply uses the user input damage width and make each damage segment 1/6, 1/4, 1/2 of the user entered damage width
One important item to consider:
  • Some commercial versions of the CRASH/damage analysis have added options to allow for more than 6 damage measurement and different sized 'segments' for damage measurements
    • (which is mainly to try to give the illusion of some aura of additional accuracy however in tests run using measurements with a simple ruler vs survey equipment to make measurements the accuracy and limitations of CRASH are not improved at all. (i will add a reference to these tests when i get a chance, i believe it was documented in an SAE paper) However i guess it makes a good sales pitch for their 'better mousetrap'!)
    Part of their input process may include input of the length between each random damage measurement. Those programs may include a check of the 'overall damage width' and the summation of the individual segment width and either alert the user to the discrepancy OR do some adjustment to the individual segment lengths to handle the difference
    You will need to check with the individual program vendors to see how they are handling that situation.
The 2,4 & 6 point equations from pages 2.29, 2.30 of the CRASH3 manual
crash 246 pt calcs.jpg
crash 246 pt calcs.jpg (43.97 KiB) Viewed 2182 times

Re: Should a CRASH type damage analysis include induced damage?

Posted: Mon Jun 15, 2020 1:42 pm
by MSI
Saw a post today on Facebook NAPARS pagedamage mesurement:
  • Measuring Crush.
    An investigator for NHTSA came to an inspection of this car which experienced a serious frontal impact. He brought a pair of folding music stands, and set them at a common distance from the undamaged rear axle to serve as a reference baseline. He tied a string at bumper height and used tape on the string to flag the locations for his 6 crush measurements (equidistant on the original face, now distorted on the car). Low tech, but also low cost, easily transportable on a plane, highly adjustable, easy to explain, and repeatable enough for the task of measuring crush. He let me take a picture of it. Thanks, NHTSA-guy!
And my response:
  • Love it! i'll take on ANY cloud scan in damage analysis with my plumb bob and ruler! There have been tests/papers on comparing different measurement techniques when applying damage analysis and it all boils down to that a cloud scan for data in crush analysis is like using an electron microscope to measure something to the nearest inch!! what's the point? Whole lotta gigabytes for a crude technique.(yea they sure make purdy looking pictures which has jury appeal!!)
    CRASH (on which all damage analyses techniques are based) was invented by Ray as a quick/fast way to get starting speeds for NHTSA to start a SMAC analysis! (and yea, they decided crude was OK for statistical studies, etc etc (see our many papers on CRASH)
    Be sure to check Smith/Noga SAE 880072 paper on measurement techniques to accommodate the crudeness of damage analysis and what about induced damage, etc.

Re: Should a CRASH type damage analysis include induced damage?

Posted: Mon Jun 15, 2020 5:01 pm
by MSI
another post responded to that post as follows:
  • In the day and age of all vehicles having EDRs I fear these types of analysis will come to an end. Even mathematical calculations that have been the foundation of all that has been done in crash reconstruction will begin to take a back seat to the new reality of crash investigation.
And my response:
  • I disagree...EDRs are NOT infallible and also only tell part of the reconstruction story...sure they are a great additional piece of the puzzle however they can and do have issues/failures and don't tell you everything you need to know about a crash...we will always need a good vehicle and site examination and measurements to fully understand what happened in many crashes, etc. I've always been wary of using simple 'damage analysis' as a primary reconstruction technique (unless all other evidence not available) since it is a crude first approximation technique and should only be used in concert with other analyses and examinations. And funny that NHTSA, for whom Ray invented CRASH and who embraced it for statistical studies, was using tape measure, strings and plumb bobs to measure which i believe is in their recognition of the crudeness of simple damage analysis: being within an inch or so of measurements is GOOD ENOUGH FOR DAMAGE ANALYSIS don't kid yourself otherwise.