| Description |
For more than twenty years, the introduction of reliability-based analysis into roadway geometric design has been investigated. This type of probabilistic geometric design analysis is well suited to explicitly address the level of variability and randomness associated with design inputs when compared to a more deterministic design approach. In this study, reliability analysis was used to estimate the probability distribution of operational performance that might result from basic number of lanes decisions made to achieve a design level of service on a freeway. The concept is demonstrated using data from Interstate 15 and Interstate 80 in Utah. The basic traffic count data used for analysis were obtained from Utah Department of Transportation (UDOT). To account for the uncertainty in the design inputs, statistical distributions were developed and reliability analysis was carried out using Monte Carlo simulation. A statistical software Minitab was used to develop statistical distributions of design inputs involving variability from the traffic count data. Minitab was also used to run Monte Carlo simulation by generating random samples of the design inputs. The outcome of this probabilistic analysis is a distribution of vehicle density for a given number of lanes during the design hour. The main benefit of reliability analysis is that it enables designers to explicitly consider uncertainties in their decision-making and to illustrate specific values of the distributions that correspond their target level of service (e.g., the 65th through 85th percentile density corresponds to the design level of service). The results demonstrate how uncertainty in estimates of K (i.e., the percent of daily traffic in the design hour), directional distribution, percent heavy-vehicles, and free-flow speed significantly contribute to the variation in the vehicle density on a freeway. |
