UC Berkeley

Diverging diamond interchange (DDI) has received increased attention from the traffic community for its efficiency in reducing delays for vehicles at on- and off-ramps. The conflicting through movements between a DDI’s two crossover intersections mean that its signal plan must concurrently consider the progression for all critical paths to ensure overall efficiency. In light of DDI’s unique geometric features, the design of its signal plans typically starts with the cycle length and green splits for its two crossover intersections, and then employs available progression models to produce the optimal offsets for those critical paths. Such a two-stage design methodology, however, often cannot yield system-wide optimal results because the optimal progression bandwidth and signal settings are interdependent. Moreover, inefficient coordination between its two crossover intersections may cause excessive queues on the DDI’s bridge segment. This paper therefore presents a mixed-integer linear programming (MILP) model that can concurrently optimize the cycle length, green splits, and offsets for a DDI’s two crossover intersections under the given traffic patterns and geometric constraints such as the link length. The results of extensive numerical analyses with a real-world DDI have confirmed the effectiveness of the proposed model and its robustness in response to demand fluctuation.