From queueing performance evaluation to policy guidance: A new modeling framework for escalator systems

Abstract

Escalators play a critical role in vertical transportation in urban transit systems such as metro and subway stations, yet their efficiency is often constrained by informal usage norms such as the “walk left, stand right” rule. This study presents a new framework to characterize passenger boarding behaviors and evaluate the resulting delays on escalators. In particular, we focus on the average passenger queueing delay before boarding the escalator and define it as our performance metric designated as Average Delay Before Escalator (ADBE). Two distinct arrival processes, Poisson and Single Batch, are considered to estimate lower and upper bounds for the ADBE. A probabilistic output process is used to simulate boarding decisions under varying traffic and behavior scenarios. We analyze three boarding policies: (1) all-standing, (2) walk-left-stand-right with two variants, and (3) a hybrid strategy reflecting real-world behavior. Four passenger types are defined to emulate behavioral diversity, reflecting differing urgency and space preferences. Simulation results reveal that the all-standing policy significantly reduces the ADBE and increases throughput, especially under congestion. Our model offers actionable insights for transit planners and metro system operators, providing a quantitative basis for optimizing escalator usage policies and enhancing both throughput and safety.