(290a) Design of Multi-Actor Distributed Processing Systems:a Game-Theoretical Approach

The manufacturing of a product could be the result of a value chain involving several processing plants distributed across distinct owners who form a network that exchange materials and/or energy among the involved actors. An example of such networks is given by the petrochemical value-chain which consists of refineries, petrochemical crackers, intermediates processing facilities and manufacturers of consumer products. Another example is given by the design of biorefineries, where conversion of biomass into intermediate platform chemicals and upgrading of these into other products will most probably take place at different corporations.

In our previous work, we proposed a novel strategy for the design of distributed manufacturing networks in which, suppliers and consumers linked through different intermediate products (mass or energy) sought maximization of their own profit. Under this assumption, we formulated a multiobjective optimization problem and proved that a 2-level Lagrangian approach could be used to obtain the optimal solution for the overall network (amount of mass/energy to be exchanged as well as transfer prices) through the optimal designs of the individual actors that participate in the network. This approach, assumed the existence of a centralized coordinator that collects the information from all the actors and thus implicitly knows in detail the form and parametric values of their objective functions and internal constraints.

In this presentation we propose and develop a game-theoretical framework and specific methodologies, which, without a centralized coordinator or share of internal information, allows for the optimal design of distributed processing systems. Within this framework the conditions of optimality established through the previously mentioned 2-level Lagrangian approach are satisfied through a negotiation process between suppliers and consumers which leads to a Nash equilibrium point. As this equilibrium point generally leads to an unequal distribution of the benefits between the actors, a Nash bargaining problem is analyzed to reach to an equitable distribution. The presentation will also discuss the use of penalty-term approaches to extend the framework to problems for which the underlying convexity assumptions of the 2-level Lagrangian approach may not be possible to ascertain. A series of case studies that illustrate the application of the proposed ideas to processing networks of various structures will be presented.

This work was funded by the Cooperative Agreement between the Masdar Institute of Science and Technology (Masdar Institute), Abu Dhabi, UAE and the Massachusetts Institute of Technology (MIT), Cambridge, MA, USA - Reference 02/MI/MI/CP/11/07633/GEN/G/00 for work under the Second Five Year Agreement.