Methods for solving the integrated procurement and lot-sizing problem with perishable inventory and uncertain demand

• Grant number: 21/12200-0
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: September 06, 2022
• End date: September 05, 2023
• Field of knowledge: Engineering - Production Engineering - Operational Research
• Principal Investigator: Maristela Oliveira dos Santos
• Grantee: Caio Paziani Tomazella
• Supervisor: Raf Jans
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Institution abroad: École des Hautes Études Commerciales (HEC Montréal), Canada
• Associated to the scholarship: 19/10824-7 - Solving the integrated raw material procurement and lot-sizing problem with the aid of big data analytics tools, BP.DR

Abstract

This project addresses the integration of two problems from the Supply Chain Management area, raw material procurement and lot-sizing. When integrating these two problems, it is possible to take advantage of the inter-dependency of the decisions, leading to more efficient solutions. Two variants of the problem, that enrich its applicability to real-life cases, are addressed: inventory aging and demand fulfillment. Inventory aging is used in cases that items are perishable or their age affects other factors such as holding costs. Demand fulfillment decisions are made when production resources are scarce and is not possible to fulfill the entire demand in time, therefore, product deliveries must be chosen in order to ensure an adequate service level. This type of problem is addressed in both deterministic and stochastic cases, the latter in which variations of the demand can cause issues in demand fulfillment after some operational decisions have already been made. The integrated problem is approached in a modeling point of view, which allows us to analyze how the aspects of the problem impact the solutions, and efficient solving methods for solving them, such as MIP-based heursitics, will also be proposed. Lastly, addressing the stochastic variant of the problem will require the use of statistical and Big Data Analytics tools to obtain more precise input demand data for these models. (AU)