Data-Driven Queueing Science Approach to Manage Secondary Queue Formation of Passenger Flow in Indian Railways: A Case of Overbooking on Vaishali Train

Abstract

Overbooking has been far and wide embraced to save any system from being idle. Rarely does the literature focuses on the unique queue formation nomenclatures as secondary queue due to overbooking. In this background, deciding the buffer capacity in the secondary queue is a major concern. The hypothetical design of a buffer often leads to wrong decision-making. In this paper, we take a real data-driven approach to explore the passenger flow through the lens of a queueing scientist. The means is to perform the statistical analysis using stationarity (Augmented Dickey-Fuller test), independence (Pearson's correlation) and distribution fitting for the data to validate the queueing system of Advanced Passenger Reservation. Further, what are the unique features of this queueing system in terms of its system primitives, key performance measures, and buffer capacity have been evaluated. What-if analysis has been explored for the different buffer capacities, and operational policy has been suggested for the passenger flow based on different scenarios. Our results show that Inter arrival and service time must be governed by a combination of N and F policies of queueing with suitable channels (servers) to quantify the overbooking.