View Q&A

Is there any later guidance about an apparent discrepancy between the design procedure in the 2002 RDG and results from crash testing? I've seen that a version of the RDG is available with a 2006 update to Chapter 6, but do not understand that Chapter 8, including the sand barrel design issue is changed.

Table 8.3 in the 2002 RDG gives a transfer of momentum method. It gives lower predicted deceleration than observed in crash tests. The example attached uses data from Test 3-41 in FHWA acceptance memo CC-29 (6/28/1995). The calculation predicts a maximum deceleration of 8.0 G's, while the crash testing produced 14.2 G's.
I've also included an example based on test 3-41 from acceptance memo CC-52 of 7/10/1998. The latter calculation predicts 7 G's, while testing documented 9.5 G's.

By changing the distance over which the impulse acts to about 1.7', the first (CC-29) calculation can be calibrated to match the crash test result. I tried a similar approach for tests 3-41 in acceptance memo's CC-28 and CC-52 and found calibration of this parameter ranged from 1.3 to 2.2 feet.
The lower value of 1.3 feet for test 3-41 of memo CC-28 involved a heavier pickup truck (4934#) and higher speed (68 mph). In general, it appears that the more energy in the impacting truck, the higher the calibration factor is needed.

Because the transfer of momentum is elastic, some energy is going to other actions "¢â‚¬Å“ breaking the drums, deforming the vehicle, and bulldozing the sand.
The sand and the drums have some variations from test to test.
If the systems were inelastic and frictionless, the transfer of momentum equations would correctly predict the change in velocities, but accelerations would be excessive.
This appears to be a more complex problem than presented in the RDG methods.

think there are a couple of items here that are leading to your comparison issues.

First, the simple inertial procedure in the RDG generates a change in velocity based on and impulse momentum calculation. From this, you can estimate an average deceleration due to the impact of the vehicle with each row of barrels. This is an average deceleration and not an instantaneous acceleration like we measure in testing. Thus, the estimation procedure will always yield a lower acceleration value than the full-scale test. To deal with this, we have typically limited the average deceleration values for our sand barrel array calculations to 12 g's or less. This is conservative, but it is done to account for the difference between the actual and average calculated accelerations. The other mechanisms you mention have some effect on the acceleration as well, but the issue you are seeing is due mostly to the calculation of average acceleration rather than instantaneous.

I noted that you are attempting to vary the deceleration distance to yield better results. We would not recommend this. As noted above, you are only calculating average decelerations, so we cannot expect the values to match. Instead, design the arrays with the 12 g limit noted above. We typically use 36" or 3' for the deceleration distance which is the typical diameter of the sand barrels.

I have attached a spreadsheet that we have developed to analyze sand barrel arrays. Feel free to use it to develop you MASH compliant versions. The sheet allows you to vary the array configuration for different speeds and vehicle masses and notifies you with colored formatting to indicate when the array has slowed the car sufficiently or exceeded the deceleration limits.

One thing I should note is that when designing these barrel arrays you should consider more than the head on impacts. In addition, you should address a coffin corner type impact similar to test designation 3-36 in MASH. This should be done to ensure that vehicles impacting downstream of the first barrel of the system do not experience excessive deceleration but are still brought to a controlled stop.  In addition, if you the barrel array is used in installations where it can be impacted from the reverse direction, you should consider additional smaller barrels on the downstream end of the system to prevent excessive deceleration. Examples of these additional barrels can be found on page 97 and 98 of TRP-03-209-09.