Peak-hour metro rail traffic congestion alleviation

Abstract

Dubai Metro is the city railway transport under the Road and Transport Authority (RTA) of Dubai. It is operated as a cyclic schedule on fixed O-D pairs and two lines. The two lines are (i) the Red line which runs between Rashidiya and Jebel Ali and (ii) the Green line which operates between Al Qusias and Creek. Union and Khalid bin Waleed are the two cross-over stations between the two lines. Operational times are based on week days and weekends and the headways are based on peak hour and off-peak hours of the day. Light rails are used in Dubai Metro and each train has five cabins. On both lines, headways are designed at 90 seconds, whereas currently operated headway during peak hours is 3.5 minutes. A high ridership is observed during the last working day and particularly during peak hours, leading to severe over-crowding. To reduce over-crowding and alleviate congestions, three types of strategic solutions are available. (i) Reduce peak hour headways (ii) Increase number of cabins in trains operated during peak hours. (iii) Increase number of services between modified O-D pairs with denser traffic. The problem domain is modeled as a directed transit network and further as an LP formulation with a focus on the third solution for the Red line, under an assumption of known demand. Nodes represent arrival/departure times at each station and two types of arcs represent storage and movement of train cabins, respectively. Constraints on (i) consistency of flow, (ii) non-negativity of flow, (iii) indivisibility, (iv) demand satisfiability and (v) bounds are proposed. The objective has two components: (i) To minimize the number of train cabins in the system and (ii) To minimize the total car miles run by all trains in a schedule cycle. The LP Model developed for the third proposed solution is tested with actual data extracted in September 2012 using an open package LP/MILP IDE for Win32, namely GUSEC. Partial results are obtained and reported. Several interesting questions have arisen out of our model and we are trying to analyze some of them. Attempts continue (i) to acquire, test and evaluate validated data from RTA, (ii) apply the proposed model to green line and study congestion alleviation of red and green lines in a combined manner and (iii) evaluate the model against other possible solutions.