Production Planning Under Uncertainty: Robust and Stochastic Approaches

Production planning problems have been studied for many decades not only due to their practical importance in various systems from integrated manufacturing to logistics, but also due to their inherent theoretical complexities (e.g., even the single-item problem with varying capacities is NP-complete.) Despite the literature being primarily focused on deterministic variants of the problem, there has been increasingly more work considering uncertainty. In this talk, we will discuss two specific examples of production planning under uncertainty. In the first part, we will look into a lot-sizing problem with the option of remanufacturing including a two-level, multi-component variant. Following deterministic formulations to explore the problem setting, we discuss the impact of imposing uncertainty on customer returns and introduce uncertainty sets to redefine customer returns. We propose robust formulations and a decomposition approach to tackle this problem, and present a wide range of computational tests with practical and computational insights. In the second part of the talk, we present a novel way of modeling uncertainty on demand in the single-item setting. More specifically, the uncertainty is not, as it most often does, related to the demand quantity, but rather to the demand "timing" (i.e., demand occurs fully in a period, but we do not know which period). Dynamic programs are proposed for the general case of multiple demands with stochastic demand timing and for several special cases, and we discuss further directions of interest including complexity and more complex problem settings such as multiple items. First part is joint work with Oyku Naz Attila, Agostinho Agra and Ashwin Arulselvan, and second part is joint work with Melek Rodoplu and Stéphane Dauzère-Pérès.

Bio

Kerem Akartunali is Professor at the Department of Management Science, Strathclyde Business School, where he leads the Optimisation & Analytics Research Group. Before joining Strathclyde in 2010, he worked as a postdoctoral researcher at University of Melbourne, in collaboration with CTI, an Australian software company, on the development of methodology and software for airline planning and scheduling problems. He gained his PhD in 2007 from University of Wisconsin-Madison. He has research expertise in integer, network and robust optimisation and their applications, including production planning, transportation scheduling/planning, and health and energy applications. He worked with many companies and organisations, including AGS Airports, Capita, Cordia, NHS, Preactor, Scottish Power Renewables, SSE, and The Drone Office, from short-term consultancy to long-term research partnerships. His research has been funded by various organisations, such as Innovate UK, EPSRC, Horizon 2020, US Air Force Office of Scientific Research, and various industry partners including Capita, SSE, SPR and Technip.