1 Optimizing Material Sourcing and Delivery Operations Anantaram Balakrishnan and Xiajun Amy Pan McCombs School of Business, University of Texas at Austin, Austin, TX, USA Email: anantb@mail.utexas.edu, xiajun.pan@phd.mccombs.utexas.edu Keywords: supplier selection, vehicle routing, time-space network, integer programming 1. Introduction We address the following Material Sourcing and Delivery Planning problem with time windows (MSDP). Given the material requirements and delivery due dates for a set of geographically-dispersed customers, and the available sources of supply, determine which supplier(s) to use for each customer, and how to sequence the material pickup and delivery operations, using a limited number of available vehicles, in order to supply the required amount of material to each customer before the corresponding due date. This problem is motivated by the material supply planning task for scheduled track maintenance projects at a railway company. In this context, customers correspond to track maintenance jobs at different locations on the nationwide railway network. We are given the scheduled start date for each job, and the total material (e.g., ballast) needed at each job site. Each job requires one or more full-loads (e.g., train loads) of material from alternative suppliers (e.g., quarries). The current planning process to decide the quarterly or annual procurement and delivery plan is manual, and can result in delayed deliveries and high procurement cost. The goal of this work is to develop a decision support model and effective solution approach to minimize the total material procurement, transportation, and delivery cost for the MSDP problem. The MSDP problem entails three sets of interrelated decisions: sourcing, vehicle assignment, and routing. The sourcing decision requires selecting one or more suppliers from among the candidate suppliers for each customer. Some suppliers are open all the year round, while others are open only in certain time intervals. The prices offered by suppliers vary, as do the supplier-to-customer distances. Moreover, suppliers differ in the amount of material they can supply. These restrictions stem from limited weekly production and loading capacities, as well as limitations on the total amount of material that a supplier can provide over the planning horizon. Transporting the material requires specialized equipment (e.g., trains that can carry ballast, in the railway maintenance context). We consider multiple vehicle types that differ in the time they require to pickup and deliver each load of material. The vehicle assignment decision entails selecting the vehicle type(s) that will service each customer. Finally, the routing and scheduling decision requires sequencing the pickup and delivery activities for each vehicle so as to deliver on time to each customer while meeting the supply capacity restrictions. The goal of the planning exercise is to minimize the total cost of supplying all customers during the planning horizon. This cost includes: (i) material procurement cost, which depends on the amount purchased from each supplier; (ii) transportation cost, which depends on the distance from the supplier to each customer, the return distance from each customer to the next supplier, and the per-mile equipment and consumables cost for the loaded and empty vehicle; and, (iii) delivery cost, which varies by vehicle type and customer. In the rail maintenance context, this cost includes the cost of disrupting regular traffic through the job site when the vehicles deliver the materials. Thus, the MSDP problem seeks the supplier selection, vehicle assignment, and delivery sequencing plan that minimizes the total