1. Supply Chain Design Problem Tuukka Puranen Postgraduate Seminar in Information Technology Wednesday, March 26, 2009

2. Contents  What is a supply chain  Supply chain optimization  Supply chain design problem  Solving SCDP

3. What is a Supply Chain  A supply chain consists of  Production and storing facilities  Transportation lanes  Commodities and raw materials  Customers  There is a cost associated to each activity  Procurement  Production  Storing  Transportation

4. Supply Chain Optimization  Application of processes and tools to ensure the optimal operation of a manufacturing and distribution supply chain  Mathematical modeling techniques  Using computer software  To support decision making

5. Components of a Supply Chain Optimization System Network design Integrated planning MRP and materials management Production planning Distribution planning SchedulingVehicle routing Demand management Order management Source: Cordeau, 2008

6. Why Supply Chain Optimization?  Increasing focus on logistics and supply chain management in large companies  Globalization and increasing complexity of operations  Increasing variety of products, on shorter lead times and of high quality; introducing new technology in processes and materials  Advances in information technology  Increased data availability from ERP systems  Better software tools for modeling and solving optimization problems  Faster computers Source: Cordeau, 2008

7. Optimization Process Reality Model Data MILP Solution Decision Support Data Gathering Optimization Processing Modeling Analysis Decision

8. Data Gathering  Aim is to describe the conditions in reality as accurately as possible  Includes, for example,  Costs and capacities for each existing and potential location  Costs, production capacities, demands, and prices for each commodity in each location  Transportation costs, transportation times, and CO 2 emissions for each transportation lane  All values must be given for each existing and potential location and lane, and over given set of periods  In practice a tedious task and result in large datasets

9. Supply Chain Design Problem  Aim: determine the structure of the network in which products will flow from their points of origin to their points of consumption  Main decisions to be made:  Number, location, capacity and technology of facilities  Supplier selection  Product range assignment  Supply channels and transportation modes  Product flows (amounts purchased from suppliers, made in plants, stored in warehouses, transported, etc.) Source: Cordeau, 2008

10. Network Formulation SupplierFactory Supplier Warehouse Customer Factory Customer Supplier Customer

11. Network Formulation SupplierFactory Supplier Warehouse Customer Factory Customer Supplier Customer

12. Mixed Integer Linear Program Formulation

13. Solving SCDP  Mixed Integer Linear Programs can be solved using commercial solvers; typically based on SIMPLEX algorithm and its variants  Branching and cutting used to find integer solutions  A variety of cutting methods  Different heuristics in branching  Local search  Relaxations  Pre- and post-processing  Parameter tuning essential  1,3 million variables and 0,7 million constraints in 5 min.

14. A Note on Multi-objective SCDP  A Supply Chain Design Problem may have multiple objectives  Minimize cost or maximize profit  Minimize lead times (maximize customer satisfaction)  Minimize CO 2 emissions  Ensure robustness  Three approaches usually used to provide a single objective  Assigning weights to different objectives  Introducing additional constraints  Interactive approach

15. Additional Extensions  Ensuring robustness by optimizing using multiple scenarios simultaneously  Stochastic demand  Stochastic costs  Pricing decisions  Nonlinear demands  Nonlinear costs  All of the above

16. Summary  A supply chain consists of production facilities, transportation lanes, commodities and customers  Optimization attempts to ensure the optimal operation of a supply chain  Increasing focus on logistics and supply chain management, increasing complexity  Modeled mathematically; data gathering also challenging  Solved using Mixed Integer Linear Programs  Used to support decision making  Additional extensions continue to provide computational challenges