Blog Post | Staging a Car Crash | RICSAC 2

Last Updated: August 9, 2017

In Chapter 12 of the User's Guide, you are walked through the workflow of a typical accident reconstruction analysis involving a t-bone style impact based on a staged collision test from the Research Input for Computer Simulation of Automobile Collisions (RICSAC) series. In that case, using knowledge of the pre-impact orientations and post-impact rest positions and orientations, as well as the post-impact trajectories, we were able to iteratively converge on a reasonable solution for the collision, obtaining estimates of the pre-impact speeds. In this post, we will repeat this same process for the second RICSAC collision ("RICSAC 2"). Below, we will follow the same procedure and style as in Chapter 12 of the User's Guide. Here we show the final result:

In the RICSAC 2 collision, a 1974 Chevrolet Chevelle impacted a 1974 Ford Pinto. The figure below depicts the impact configuration, as well as pre and post-impact trajectories. The figure is obtained from Figure 6 of Reference [1]. We will start by importing and scaling this crash diagram [2].


As in the User's Guide, what we would like to do is use our knowledge of the post-impact trajectory, orientation, and rest position, to help us obtain reasonable estimates of the pre-impact speeds of both vehicles. This is a common problem that every accident reconstructionist faces, and so this is a worthwhile exercise. 

On page 2-11 of Reference [3], we can find the "Summary Form" for RICSAC 2. This is shown as Table 1 below.


Table 1


The "Crash Test Summary" can be found on page 8-4 of Reference [4]. This is shown below as Table 2.


Table 2


Finally, we show the "Vehicle Test Weights" form from page 8-21 of Reference [4]. This is shown as Table 3 below.


Table 3


We will create an axis object plot the impact and rest positions noted in the table. We will also place markers at the impact and rest positions as we did in the Chapter 12 exercise. Due to the known scaling error of the Smith and Noga diagrams, the scale tool was aligned with the 3 meter imbedded scale, but the scale tool length was set to 10.2 feet. The image was then reasonably aligned with the impact and rest position markers shown above.

With our crash diagram, and Tables 1, 2, and 3, we are now ready to proceed with our simulation. 

In our simulation, we will reuse the same vehicles from our RICSAC 1 simulation shown in the User's Guide. Note, the wheel location, and geometrical size data remain unchanged; however, the vehicle weights are different for RICSAC 2.

The front axle to cg data also needs to be updated to account for the front/rear weight distribution shown in Table 3. For vehicle 1 (Chevy), the cog to front axle value was set to: (2130 lbs / 4710 lbs) x (116 in) = 4.37 ft. 

For vehicle 2 (Ford), the cog to front axle value was set to: (1630 lbs / 3260 lbs) x (93.5 in) = 3.895 ft.

For both vehicles, set the adhesion (locked-wheel drag factor) to 0.87, as indicated on Table 1.

Next, in the sequences menu, set the post-impact rolling resistance values, also given by Table 1. For vehicle 1, we have: 

For vehicle 2, we have:

Recall, the lock wheels option allows you to specify the percentage of full wheel lock for each wheel individually. 

Search for reasonable pre-impact velocities

Typically, one may have some initial indication of pre-impact velocity ranges by using momentum conservation with knowledge of the post-impact trajectories as well as pre- and post-impact orientations, as we have in this case. With some knowledge of the post-impact trajectories, one could find reasonable limits on the post-impact velocities given the particulars of the trajectories, and a simple spin-out analysis. One can conduct a post-impact spin-out analysis in Virtual CRASH in minutes; this is demonstrated in the following video: 

For our RICSAC 2 study, we will conduct a similar analysis to obtain some reasonable first estimates for the pre-impact speeds of our two vehicles. Once we have some first reasonable estimates, we can move on to our simulation. Note, performing a post-impact spin-out analysis isn't necessary, but it can help speed things up as we will have a natural starting point to start tuning our simulation. One major benefit to performing a post-impact spin-out analysis interactively within Virtual CRASH is that all effects of tire slip angles and simultaneous rolling resistance effects are automatically accounted for, whereas on paper, one would need to break up the trajectories into regions of common effective drag factors. In Virtual CRASH, simply place the vehicles at the location where impact begins, then iteratively modify the initial speeds, velocity vector orientations (vni), and yaw rates (omega-z) until both vehicles converge to a reasonable degree onto their post-impact trajectories. 

With the final velocity data, we can use momentum conservation to estimate the pre-impact velocity. Here we simply use a spreadsheet tool to perform the calculations (Reference [5]). Without loss of generality, we can simplify our problem by first rotating our x-axis by 30 degrees (our "Adjustment Angle" in the spreadsheet below) such vehicle 1 is traveling along the x-axis. This implies that vehicle 2 is traveling along the pre-impact heading given by yaw = -120 degrees. Our estimates for pre-impact speeds are given by:

These are shown in yellow below:

Therefore, from momentum conservation, we know vehicle 1 and vehicle 2 should have initial velocities in the range of 28 to 35 mph. With this information, we can proceed with our full simulation.

Note, here PDOF is defined with 0 degrees directed forwards and 90 degrees to the left. This convention means that a frontal collision with a rearward directed force has a PDOF of 180 degrees.

Full simulation

We start by placing the vehicles near their documented points of impacts. Next, as was shown in Chapter 12 of the User's Guide, we iteratively adjust (1) impact speed, (2) initial vehicle positions, (3) coefficient-of-restitution, (4) impulse centroid location (set depth of penetration = 0), until we obtain reasonable convergence. Note, by adjusting the initial positions such that the vehicles' volumes overlap at time=0 seconds, you are effectively changing the depth of penetration time, which is set to 0.03 seconds by default. For this simulation, we recommend setting the depth of penetration to 0 seconds. Recall, the depth of penetration controls the time from first overlap of vehicle bounding boxes until the moment impulses are instantaneously exchanged. By starting with the vehicles overlapping, you are effectively delaying the moment when the depth of penetration time starts, thereby causing the impulse centroid to be located deeper within the collision volume of each vehicle. 

The DeltaV values can be found using the ees object, or by using the report.

For our analysis, we would like to know by what distance our final simulated vehicle positions differ from experimental data. We would also like to calculate the difference in final orientations, initial velocities, and DeltaVs. We will calculate the differences in final positions between simulated and experimental values, using the initial position of vehicle 2 as our reference point. In the simulation, we define vehicle 2's initial point as being the cg location at the moment when the vehicles begin to overlap. Our solution came with Vehicle 1's initial (x, y, yaw) = (-7.238 ft, -6.271 ft, 30 deg), and Vehicle 2's initial (x, y, yaw) = (-0.064 ft, -2.404 ft, -90 deg). The impulse centroid for the first impact was positioned at (x, y, z) = (-2.029 ft, -3.932 ft, 1.641 ft), with the coefficient of restitution = 0.08 and coefficient of friction = 1.0. The final results are shown in Table 4:


Table 4


Here we see excellent agreement between our simulation and experimental data. Indeed, vehicle rest positions are simulated to within 1 feet, while final orientations agree to within 4 degrees. There is less than 8.4% difference between experimental and simulated initial speeds. DeltaV matches to within 13% for the Chevelle and 23% for the Pinto. We note using Newton's 3rd law to estimate the Pinto's Delta-V from the Chevelle's data yields a value of 28.17 mph, which is within 13% of the simulated value.


[1] “Examples of Staged Collisions in Accident Reconstruction,” R. Smith and J. Noga, NHTSA, US DOT.

[2] Chapter 9 | Scaling Images

[3] “Research Input for Computer Simulation of Automobile Collisions, Volume IV. Staged Collision Reconstructions,” NHTSA, US DOT, DOT HS 805 040.

[4] Research Input for Computer Simulation of Automobile Collisions, Volume II. Staged Collision Reconstructions,” NHTSA, US DOT, DOT HS 805 040.

[5] Equations (19.7) and (19.8), "Fundamentals of Traffic Crash Reconstruction, Volume II," Daily, Shigemura, and Daily, IPTM.