Abstract: Operating disruptions result in enormous costs across spatial-temporal networks. For instance, air traffic disruptions result in flight delays, cancellations, passenger misconnections, creating high costs to aviation stakeholders. Businesses often design recovery plans in response to past disruptions while preparing for future disruptions. However, future disruptions often can only be characterized partially and probabilistically. We propose a joint stochastic reactive and proactive approach to disruption management (SRPDM), which optimizes airline disruption recovery given partial and probabilistic forecasts of future congestion at hub airports. disruptions. Ultimately, it aims to mitigate operating costs in spatio-temporal networks through more flexible and robust recovery. We develop an online solution procedure based on look-ahead approximation and sample average approximation, which enables the model’s implementation in short computational times. Leveraging partial and probabilistic estimates of future disruptions can reduce expected recovery costs by 1% to 2%, as compared to a baseline myopic approach based on observed disruptions alone. To evaluate the performance of this model, we introduce a novel lower bound based on penalized information-relaxations, generalizable for time-space networks. We formulate this as a tractable integer program, avoiding the need to enumerate prohibitively large state spaces. The performance bound demonstrates that when improvement is possible, the SDRPM bridges significantly higher gap than could have been estimated using a regret bound. This indicates that networked operations can strongly benefit from even partial and probabilistic forecasts of future disruptions, based on available demand and capacity information.

Bio: Lavanya Marla is an Assistant Professor in Industrial and Enterprise Systems Engineering at the University of Illinois at Urbana-Champaign. Prior to her current position, she was a Systems Scientist with the Heinz College at Carnegie Mellon University; and earned her PhD in Transportation Systems from MIT and Bachelors degree from IIT Madras. Her research interests are in robust and dynamic decision-making under uncertainty and game theoretic analysis for large-scale transportation and logistics systems; combining tools from data-driven optimization, statistics, simulation and machine learning. Her research is funded by an integrative National Science Foundation grant, a Department of Homeland Security cyber-security grant, the Department of Transportation, the US-India 21st Century Knowledge Initiative, the INFORMS Transportation and Logistics Society and aviation companies. Her work has received an Honorable mention for the Anna Valicek award from AGIFORS, a best presentation award from AGIFORS, a KDD Startup Research award, and a Top-10 cited paper recognition from Transportation Research – Part A.