In order to establish cost-effective traffic management strategies for alleviating the delay caused by crashes, transportation professionals need to understand better the relationship between crash, traffic, and site characteristics and associated traffic delay. An accurate estimate of delay caused by crashes is a necessary first step in a traffic delay prediction model. This study considers an empirical approach for obtaining accurate estimates of delay caused by crashes on freeways.; The focus of the discussion in this thesis is on estimating traffic delay from freeway crashes using empirical data. The study provided a statistical evidence for the effect of weather on delay estimation. In addition, the results from this study revealed that the proposed formula modified from the current empirical method is more reliable.; The discussion highlights a number of important implementation issues for obtaining an accurate estimate of traffic delay using the empirical approach, such as, missing data manipulation, data smoothing, determination of the temporal and spatial extents of individual crash delay, and estimation of incident-free (reference) speeds. Particularly, the study introduced a nonlinear smoothing technique, namely "4253H twice", and verified the smoothing performances based on statistical analysis. The proposed formulas for estimating delays are applied for the 201 crash samples obtained from an instrumented urban freeway in Toronto, Ontario. |