Prof. Ramandeep S. Randhawa
University of Southern California
Talk:
Using fluid approximations for designing service systems
Abstract:
Service systems tend to be very complicated to analyze exactly. This has led to a large literature aimed at developing good approximations of performance measures of interest, and then further using these for decision making, in particular, for capacity selection, routing customers etc. These approximations typically involve using asymptotic analysis by utilizing versions of laws of large numbers, via fluid limits, and/or central limit theorems, via diffusion limits. It is typically believed that fluid limits provide a crude approximation, which is refined by diffusion limits (which also tend to be more complicated). In this talk, we will demonstrate that in many settings, fluid limits can in fact provide extremely accurate prescriptions without the need for any further refinements. Given the simplicity of computing fluid limits, this allows for much easier decision making. Focusing on a call center application, we will demonstrate how these fluid limits may be used for capacity selection and customer routing.
Biography:
Ramandeep S. Randhawa is an operations research scholar whose research interests include designing service systems, revenue management, stochastic modeling, and mechanism and incentive design. His work has been published in journals that include Management Science, Manufacturing and Service Operations Management, and Operations Research. He received his PhD in Operations, Information, and Technology from Stanford Graduate School of Business, Stanford University. Prior to joining the Marshall School, he was a faculty member at the McCombs School of Business at the University of Texas.
