Constraint-based Optimization: Judo for the Supply Chain
(Supply Chain Alert, June 2003)

By Chris Beck and Gene Freuder,
Cork Constraint Computation Centre

Judo is a bit counter-intuitive. Judo means the "gentle way:" when faced with an attacking opponent, the response is not to meet the opponent head-on, applying force against force. Rather the response is to give way or to be compliant. In this way, the opponents force may be turned and used against him.

This same philosophy is at the heart of constraint-based techniques that are increasingly applied to manage global supply chains. Instead of viewing the constraints on decisions as an opponent to struggle against, they are used to restrict and evaluate the impact of supply chain management choices.

Imagine a simple supply chain where factories can ship products directly to customers. When a new order arrives, a choice must be made to allocate the order to a particular factory. Many constraints bear on this decision: Is the factory technically able to manufacture the order? Does the factory have sufficient production capacity? Does the third-party logistics company contracted for a factory have sufficient capacity to make the delivery? What is the cost of production and delivery? These constraints can be used to make the decision easier. Initially, we might entertain the possibility of assigning the order to any factory and we may be at a loss for deciding amongst them. By applying the constraints, we can narrow the decision-making.

Obviously factories that are technically unable to manufacture the order can be ignored for this decision. Next we can look at the available capacity of the factories and logistics companies to reduce our options to those that can feasibly produce and deliver by the required due date. Finally, we can reason about the costs: of the feasible factories, which can deliver with minimum production and delivery costs? The use of the constraints limits the final options that must be investigated in depth.

Some constraints are "hard," e.g. the capacity of a shipping container, while others are "soft," e.g. a preferred delivery date. Researchers are even beginning to reason with "stochastic" or uncertain constraints, like customer demand.

Many of the basic constraint-based techniques can be found within existing supply chain management software from such companies as SAP and Oracle. Research is focusing on some of the deeper issues that are not yet addressed by commercial software. The Cork Constraint Computation Centre (4C), a 9 million euro research center established by Science Foundation Ireland, is at the forefront of this research. Using learning and knowledge-based techniques, 4C is working to reduce the deep technical knowledge required by people who implement constraint-based supply chain solutions. Currently, highly skilled consultants are required to implement supply chain optimization models. The goal of the research is to provide an environment where experts in the supply chain of a company can create, evaluate, and use constraint technology to aid in decision making.

The need for a company to hire a black belt in optimization technology to implement a supply chain solution may soon be a thing of the past. Through the work of 4C and others, an internal company expert will be able to make use of the distilled and automated experience of previous supply chains to develop customized solutions: the "gentle way" indeed.

Editors Note:
For more information about the work of the Cork Constraint Computation Center, see www.4c.ucc.ie or contact j.little@4c.ucc.ie.

This article was published by Retail Systems AlertSM Advisory Service. If you would like to have access to this type of information every day, become a subscriber.

Retail Systems Alert Advisory Service has provided objective technical information and analysis for the retail industry for more than 15 years. Find out more about the benefits of being a subscriber and review recent headlines.