Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
65 6.1 Study Approach Data was collected under three different studies: the FHWA Rollover Study, NCHRP 17-11, and NCHRP 17-22. Each of these studies involved a retrospective data collection and analy- sis of historical NASS CDS cases. Supplemental site informa- tion was collected to identify characteristics of the roadway, roadside, and objects struck during the crash. This supple- mental information was then utilized to reconstruct each crash in order to determine vehicle departure and impact conditions. The data was then compiled into a relational database that can be used to analyze the data. 6.2 Findings A relational database of ran-off-road crashes has been devel- oped. The database includes detailed characteristics of the vehi- cle, trajectory, roadway, roadside, objects struck, and crash result for 877 crashes. The data are strongly biased toward seri- ous crashes with 15% fatal and 72% A+K crashes. The database can be used for many different purposes, including identifi- cation of roadway departure and roadside impact conditions, and ran-off-road trajectories. The database can also be used to develop a relationship between impact conditions and crash severity for some common obstacles, such as trees and poles. Although prior studies showed departure velocity to be most closely associated with highway functional class, this roadway classification was not available in the current data- base. In the absence of highway functional class, speed limit was found to provide the best discriminator for departure velocity and angle. Departure velocities were found to be accurately modeled with a normal distribution while no sin- gle common distribution provided a good fit to departure angles for all speed limit classes. However, the gamma distri- bution was found to fit the square root of the departure angle for all speed limit classes. The dependency between departure angle and velocity was found to be relatively insignificant for all individual speed limit classes. Chi-square tests for inde- pendence showed that departure velocity and angle could be considered independent for all speed limit classes. Further, combined velocity and angle distributions developed based on the assumption of independence were subjected to chi- square tests for goodness-of-fit. These tests showed that the differences between predicted and measured distributions of departure velocities and angles were not statistically signifi- cant at the p = 0.05 level for any speed limit class. Thus, the models of departure velocity and square root of departure angle can be reliably used to develop distributions for a vari- ety of speed limit classes included in the study. Further, the database provides definitive support for reduc- ing the length of guardrail used in advance of roadside obstacles. The distributions of longitudinal departure lengths included in the data set correlated surprisingly well with rec- ommended guardrail runout lengths generated from Cooperâs encroachment data. The only significant difference between the longitudinal departure length distributions and the mod- ified runout length guidelines was associated with the use of a 60 mph design speed for a full access-controlled freeway. In this situation, modified runout length guidelines were found to be shorter than longitudinal travel distances found in the data set. Therefore, it is recommended that states either use a design speed of 70 mph for all controlled access roadways, or an additional category should be added to the guardrail runout length table to accommodate 60 mph design speeds with full access control. 6.3 Long-Term Data Collection A detailed work plan for a long-term data collection system was developed and pilot tested. The plan involves implement- ing a continuous sampling subsystem and possibly a special study subsystem within the NASS CDS. The continuous sam- pling subsystem would provide a steady stream of new cases that would be very similar to the existing database while a C H A P T E R 6 Summary of Findings
66 special study would focus on one particular type of crash such as W-beam guardrail impacts. If implemented, the long-term data collection plan could provide information that would allow development of the relationships between impact severity and crash conditions for a wide variety of roadside features. Further, such a study would provide greater information regarding the causation of injuries and fatalities during crashes involving roadside safety hardware. This information will provide the foundation for the next generation roadside safety features designed to dra- matically reduce injuries and fatalities associated with ran- off-road crashes.