Inventory and Flow Control in Complex Networks

Professor Erik Ydstie
Carnegie Mellon University

Date: Thursday, May 11, 2006
Time: 4:00 p.m.
Place: Engineering II, Room 3361

ABSTRACT

Accumulation and flow of inventory in supply chains, chemical plants and information systems (natural and artificial) are modeled using networks of distributed devices analogous to complex electrical circuits. We call networks of this kind process networks. The conservation laws play the role of Kirchoff’s current law. Concavity of the entropy function provides the basis for an analog to the voltage law. Nonlinear systems theory furthermore says that strictly passive feedback when applied to a passive circuit automatically gives a current distribution which stabilizes so that the dissipation of heat is minimized. By making analogies amongst process control, network thermodynamics and business decision making we show that distributed inventory and flow control leads to feedback systems that are agile and able to adapt as technologies develop and markets change. The supply chain for solar grade silicon production provides the benchmark illustration in this presentation. Other applications we have worked on include glass manufacture, automotive windshield production, silicon and aluminum smelting.