ITR: Distributed Resource Planning of Process Networks

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Research: The goal of this Information Technology (IT) Research - Small project is to develop a framework for distributed modeling, simulation and agent based optimization. Such a system will permit the integration of classical process systems engineering tools, like process modeling, numerical integration, optimization and predictive evaluation with the distributed data base capabilities of ERP systems. The focus will be on solution strategies that take advantage of distributed computing and information exchange across a network of connected computational units. The approach achieves speed, quality flexibility, and robustness that scales favorably with problem size. Two specific problems will be addressed: 1. How to develop methods for dynamic simulation of distributed process networks, and 2. How to use agent based methods for solving large scale, stochastic control problems that have dynamics distributed in time and space. A key issue in the research will be how to interface the modeling and agent-based optimization frameworks in such a way that the performance of the process network is optimized. Broader Impact: Enterprise Resource Planning (ERP) systems and other IT tools allow companies to access, analyze and disseminate vast amounts of distributed information almost instantaneously over the web. This project could help interface business IT with classical tools for process modeling and optimization which in turn could be of great commercial benefit to US industry.