357 Ant-based distributed optimization for supply chain management 1 Carlos. A. Silva 1 , J. M. Sousa 1 , T. Runkler 2 , J.M.G Sa da Costa 1 *Dep. Mechanical Engineering, Institute* Superior Tecnico, Technical University of Lisbon 2 Information and Communications, Siemens AG - Corporate Technology E-mail: {csilvaj ,sousa,sadacosta}@dem.ist.utl.pt Abstract Multi-agent systems are the best approach for an ef- ficient supply chain management. However, the con- trol of each sub-system in a supply-chain is a complex optimization problem and therefore the agents have to include powerful optimization resources along with the communication capacities. This paper presents a new methodology for supply-chain management, the dis- tributed optimization based on ant colony optimization, where the concepts of multi-agent systems and meta- heuristics are merged. A simulation example, with the logistic and the distribution sub-systems of a supply- chain, shows how the distributed optimization outper- forms a centralized approach. 1 Introduction In order to improve competitiveness and profitability, most of the companies today are organized as supply- chains: a world-wide network of external partners (sup- pliers, warehouses and distribution centers) through which raw materials are acquired, transformed into prod- ucts and delivered to costumers [1]. The company's job is no longer to produce the goods, but to manage all the different partners in a coordinated manner such that in the end the costumer receives a quality product on a cer- tain desired date. The different partners in a supply-chain operate un- der different sets of constraints and objectives. However, the systems are highly interdependent and the optimiza- tion of objectives such as on-time deliveries or costs of one partner will influence the performance of the remain- ing partners. The supply-chain is a pure distributed sys- tem with several parallel and independent optimization problem and the coherence between the different deci- sion making centers can be accomplished by a multi- 1 This work is supported by the German Ministry of Education and Research (BMBF) under Contract no.l3N7906 (project Nivelli) and by the Portuguese Foundation for Science and Technology (FCT) un- der Grant no. SFRH/BD/6366/2001 and "Programa de Financiamento Plurianual de Unidades de I&D (POCTI), do Quadro Comunitario de Apoio III". agent based framework, based on explicit communica- tion between constituent agents to control multiple sys- tems [1,5]. This paper introduces an innovative management methodology based on the description of the supply chain as a distributed optimization problem. The opti- mization problems are solved by the ant colonies meta- heuristic that can also be used as a multi-agent frame- work. 2 Description of a supply-chain A typical supply chain has at least two partners: the logistic system, that collects the orders from the cus- tomers, purchases the components from external sup- pliers and schedules the components gathered in cross- docking centers, e.g. airports, see [5]); and the distribu- tion system, an external company that collects the com- ponents at the cross-docking centers and delivers them to the clients as orders. The task of each system can be modeled as an optimization problem. 2.1 Logistic process The logistic system receives every day new orders re- quested by different clients, where an order Oj is a set of different types of components in certain quantities, with a certain due date dj. The different components and their quantities are purchased from external suppliers, that de- liver the components to the cross-docking centers after a certain period of time. The logistic process task is a scheduling problem that consists of observing the list of n orders and the list of components, and decides which orders are released at date rj. The difference between the release date and the due date is called the lateness Lj — rj — dj. The objective is to match the release date with the due date, i.e. to have for all orders Lj = 0. This decision step is done once per day. Two disturbances may influence the system: the fact that suppliers service may not be respected; and the fact that some clients ask for desired delivery dates not compatible to supplier services. The optimization