by Joe O » Thu Feb 17, 2011 2:19 pm
Dave,
I think you point out one of the issues that is concerning to me related to CMF's. Doesn't the benefit from an improvement vary depending on the conditions and the crash history? It seems that CMF's need to be a function that varies depending on conditions versus one number that is applied universally.
For example, assume two rural 2-way stopped control intersections. Say that one has 20 rear end crashes and no right angle crashes in a year and the other has 20 right angle crashes and no rear end crashes in a year. If one were to look at those intersections one might see that the intersection with 20 rear end crashes has a high main street to minor street left turning volume and no left turn lanes. One might conclude that installation of a left turn lane probably makes sense. At the other intersection one might find that there is a high side street volume and high speeds on the main street. At that location they might surmise that a traffic signal makes sense.
If one used that approach I suspect that the CMF's for a left turn lane at the first intersection or a traffic signal at the second would probably be appropriate to use for estimating the benefit from the improvement. It is concerning though that someone might now look at both locations and decide that a traffic signal is the right solution at both because the HSM says a traffic signal will reduce total crashes by 44%. I don't believe that is how the manual is intended to be applied but it is not clear that it is not how it is to be applied.
There is some discussion in the HSM about the need for engineering judgment but the methods for actually estimating crash reduction are very quantitative and seem fraught with opportunities for misapplication. I'd appreciate some comments from someone who was involved in the development of the manual to weigh in on the need or lack thereof for engineering judgment in the application of the CMF's.