Incidents account for a large portion of all congestion and a need clearly exists for tools to predict and estimate incident effects. This study examined (1) congestion back propagation to estimate the length of the queue and travel time from upstream locations to the incident location and (2) queue dissipation. Shockwave analysis, queuing theory, and cellular automata were initially considered. Literature indicated that shockwave analysis and queuing theory underestimate freeway travel time under some conditions. A cellular automata simulation model for I-66 eastbound between US 29 and I-495 was developed. This model requires inputs of incident location, day, time, and estimates of duration, lane closures and timing, and driver re-routing by ramp. The model provides estimates of travel times every 0.2 mile upstream of the incident at every minute after the start of the incident and allows for the determination of queue length over time. It was designed to be used from the beginning of the incident and performed well for normal conditions and incidents, but additional calibration was required for rerouting behavior. We recommend that the Virginia Department of Transportation (1) further pursue cellular automata approaches for near-real time applications along freeways; and (2) consider adopting an approach to address detector failures and errors. Adopting these recommendations should improve VDOT's freeway real-time travel time estimation and other applications based on detector data.