From the Crash Forum: Q: Is there any publication(s) known dealing with the measurement of accelerations (resp. integrated delta-v) in different locations during vehicle crash tests? (E.g. several locations in the passenger compartment or on the chassis.)
It includes in the appendix the derivation and the steps to calculate DeltaV etc at locations other than the CG.
It also includes results of simulation and comparison of accel and deltaV from the msmac3D simulation with full scale crash tests measurements.
Of particular interest is RICSAC#10 which has a sideslap and depending on the acceleration location can produce very different results for DeltaV (see below)
ABSTRACT
Research performed in the 1970's revealed significant limitations in the available documentation of vehicle crush information and trajectory spinout information. As a result a series of full-scale crash tests were performed which became known as the Research Input for Computer Simulation of Automobile Collisions (RICSAC) crash tests. Previous research using the RICSAC test results, particularly in relation to the validation of accident reconstruction computer programs, has varied widely in acceptance, interpretation and presentation of the RICSAC test results.
This paper presents a detailed review and decipherment in useable form of the original 12 crash tests that were performed within the RICSAC program. A new method of analyzing accelerometer data from arbitrary sensor positions, on the basis of discrete measures of the vehicle responses rather than complete time-histories, is defined. A discussion of previous research which included reference to the RICSAC test results as a measure of the validity of reconstruction computer programs is included.
sample figures from the paper
be sure to see text in paper RICSAC97 for a detailed explanation
For additional information on RICSAC and specifically RICSAC Test#10 with better scanned photos see
Yesterday was the first day of SAE's free virtual conference on crash reconstruction. NAPARS member and all-around EDR-guru Rick Ruth gave a good discussion on the problem of measuring Delta-V in offset-collisions.
The EDR itself is usually close to the car's CG, but not quite exact. In big offset collisions, like when a crossing vehicle catches the rear bumper of a thru-traveling car, the Damage Centroid and the line of action for the PDOF are far away from the EDR - it only gets SOME of the message, more or less.
It is actually possible for the EDR to underreport or over-report. Perhaps by 50% or more, not including the inherent uncertainties of accelerometers.
The math to deal with it requires just a few measurements on the car itself. If the EDR is on the centerline, the math is easier.
The paper from NAPARS-members Andy Rich and Bob Scurlock can be found online here: https://arxiv.org/abs/2001.03165v4 - the pdf download link is on the right. Amusingly, I realized when reading the paper that I was driving the bullet car (V1?) for the 2018 test they discuss. Ha!
The takeaway is this: The Speed Change recorded by an EDR may not match that experienced by a car in an offset hit. But it can be dealt with.
Our comment:
Been there, done that in our 1997 RICSAC-97 research...
Obviously easier to do when the simulation program models forces and moments for every millisecond during a crash (like msmac3D) instead of 'instantaneous momentum exchange' programs like with virtual crash and pc-crash...
We also addressed dealing with simplified analysis in our RICSAC-97 research
from above:
RICSAC-97 includes in the appendix the derivation and the steps to calculate DeltaV etc at locations other than the CG.
RICSAC-97 also includes results of simulation and comparison of accel and deltaV from the msmac3D simulation with full scale crash tests measurements.
Of particular interest is RICSAC#10 which has a sideslap and depending on the acceleration location can produce very different results for DeltaV
That work of yours freaked me out when I realized the uncertainties (errors?) in those RICSAC test data. This newer paper reminded me of that work
Our response
Reread the RICSAC-1997 paper!...
There were NO errors or uncertainties in the RICSAC data...it was simply that the accelerometers measuring things were at locations other than the center of gravity (they were on the firewall, rear deck, etc etc).
When there are angular velocities in a crash the recordings need to be adjusted accordingly...
from the paper...
CONCLUSIONS
The RICSAC data, when interpreted with the CG-Transform procedure, are reasonably accurate and are suitable for their intended purpose of testing the validity of reconstruction techniques.
Previously reported findings of gross errors and violations of Newton’s laws in the reported RICSAC data are erroneous.
The SMAC program has been demonstrated to correlate well with properly analyzed full-scale test results
"P.S.
Brian McHenry has reminded me of the forum discussion of their 1997 SAE paper on what is basically the same issue:
Accelerometers in the RICSAC test cars reported values that some claimed were wrong, but the issue was really the offset nature of some of the crashes and the accelerometers' not being on the PDOF line of action.
His forum page and reference to that 1997 SAE paper are here: