In Virginia, sections of I-77 and I-64 in mountainous parts of the state have significant recurring fog events. These

locations have also been the sites of several chain reaction crashes involving more than 50 vehicles during fog. These crashes

were typically caused by drivers traveling too fast for the visibility conditions. To improve safety on the I-77 corridor, the

Virginia Department of Transportation constructed a variable speed limit (VSL) system that posts dynamic speed limits based on

the visibility condition. As of April 2016, the system was undergoing pre-deployment testing. Before the system was activated,

it was important to understand existing driver speed choice behavior during low visibility conditions. It was possible that posting

a VSL speed based only on the stopping sight distance (SSD) could create significant speed variance and decrease safety if

drivers were driving much faster than conditions would warrant. In this study, crash, speed, and visibility data were examined at

several locations on I-64 and I-77 where there were recurring fog events.

The crash history for I-77 revealed that crashes during low visibility conditions were more likely to be severe and

involve more than two vehicles than crashes during clear conditions. Mean speed analysis found that observed mean speeds

exceeded safe speeds for all low visibility conditions and at all sites. In the worst visibility conditions, drivers often exceeded the

safe speed by more than 20 mph. Standard deviation analysis found that speed variance did not increase as visibility decreased

on I-77, but for several locations on I-64, the standard deviation was different during low visibility when compared to clear

conditions.

Models were developed to allow a better understanding of the relationship between speed and visibility. The models

showed that although motorists reduce their speeds in low visibility, there is still a significant differential between observed

speeds and the safe speed calculated using the SSD. The models showed that speeds for I-64 were much less sensitive to

changes in visibility compared to I-77. A possible explanation for this difference is the presence of illuminated in-pavement

markers on I-64. The improved delineation provided by these markers during foggy conditions may cause drivers to perceive

less of a need to reduce speed during limited visibility. It is also possible that mean speeds in low visibility conditions are higher

on I-64 because of the regular commuters who may be more comfortable driving during foggy conditions.

The observed driver behavior from this study is being used as a basis for the VSL control algorithm that is being

implemented in the field. A primary concern of the operators of the VSL system is that it will not be heeded by all motorists and

thus will result in increased speed variance in foggy conditions. The developed model was used to create a VSL control

algorithm to help bridge the gap between current driver behavior and safe speed. It is recommended that future VSL system

deployments reflect existing driver behavior in the initial algorithms as well. After VSL activation, speed and crash data for I-77

should be analyzed to determine the operational and safety effects of the system. If the system on I-77 reduces the frequency and

severity of crashes, improves speed limit compliance, and reduces speed variance, a similar system should be developed for I-64

using the current driver behavior models from this study as part of the initial algorithm.