Research Interests

To develop methods for stable and robust control and optimization of chemical processes that combine physical models and on-line learning.

Design and control of particulate processes with application to Solar Cell production
In this project we develop models and control strategies for solar grade Silicon production. Solar grade silicon is in high demand due to the very rapidly growing interest in using solar cells to generate electricity for domestic uses, telecommunications and distributed power generation in the third world. The photovoltaic industry is facing a critical shortage of solar-grade silicon, and a large number of manufacturers worldwide are trying to develop new production technology to meet the demand in a market that continues to grow at a rate of 30-35% annually. Our investigation is aimed towards design and control of fluidized bed reactors for decomposition of Silane. The project is carried out in cooperation with SGS LLC in Moses Lake WA where the pilot plant experiments are carried out.

Dynamics and Control of Complex Process Networks
We have introduced a framework for studying dynamics, distributed control, and optimization of complex networks. These networks represent self-organizing structures so that stability and optimality follow as a consequence of how the networks are put together and how they are connected with other networks. The class is sufficiently broad to cover process networks, bio-chemical networks, reaction networks, and supply chains. The study has led to a decomposition of the business decision making processes, optimal behaviors and decentralized decision making. We use the formalism of irreversible thermodynamics and the passivity theory of nonlinear control as a basis for this theory. We are currently investigating the application of the theory to bio-chemical networks, specifically in the optimization and control of metabolic pathways.

Real Time Adaptive Control and Optimization
We are developing online optimization techniques for constrained and unconstrained optimization using input output data. The aim is to develop stand alone optimization modules that gather information by simulation and/or experimentation and adaptively controls the process so that over time the optimizer converges to the optimal decision maker. The optimizer we developed is based a method referred to as Q learning which was developed in the area of computer science for optimal control of discrete and continuous Markov decision problems. We have adapted the method for real time process control and we have developed a theoretical foundation for on-line decision making using adaptive control theory. The advantages of the adaptive methods over other traditional optimization approaches are that it uses current process models to develop policies and that they self-learning in the sense that they adapt as process conditions change.

Design and Control of Multi-Phase Reactor Systems
The focus of this research is on passivity-based control of multiphase reaction systems. To study this type of system, a metallurgic process has been chosen involving the carbothermic reduction of aluminum. This process takes carbon and aluminum oxide feed to produce aluminum in a complex sequence of reaction steps. The fluid flow dynamics and the heat transfer properties must be managed carefully since the reactions occur at elevated temperatures (above 2000 degrees C). Pilot plant studies of the reaction are carried out by ALCOA, the worlds leading producer of Aluminum. AT CMU we develop dynamic models for conceptual design, process optimization and process control.

Temperature distribution in carbothermic aluminum reactor

Related Experience
Director of R&D ELKEM ASA, Oslo Norway 1999 - 2000
Board member and Chairman of Board SolarSilicon , Oslo , Norway 1999 - 2000
Professor II, Materials Technology NUST, Trondheim 2001 - 2004

Representative Publications

Alonso, A. A. and Ydstie, B. E. ``Stabilization of Distributed Systems using Irreversible Thermodynamics", Automatica ., Vol. 37, (2001) pp. 1739-1755.

Radenkovic and B.E. Ydstie (1999) Optimality of a Direct Adaptive Controller in the Presence of Model Mismatch and Bounded Disturbances, Circuits Systems and Signals, Vol. 18, No 3

Perea, E.; Grossmann, I. ; Ydstie, E.; Tahmassebi, T.; "Dynamic modeling and classical control theory for supply chain management", Computers & Chemical Engineering , Vol. 24 (2000), p.1143-1149

Coffey, Duncan P., Chad A. Farschman and B. Erik Ydstie (2000).Distillation Stability Using Passivity and Thermodynamics. Computers Chem Engng . 24: 317-322.

Ydstie, B.Erik, (2002), New Vistas for Process Control: Integrating Physics and Communication Networks, AIChE J. Vol. 48, No 3, pp. 422-426. Also in CEP , March 2002, P18.

Garcia-Osorio and Ydstie (2004) "Vapor recovery reactor in Carbothermic Aluminum Production", Che. E. Sci , Vol. 59, N0 10, pp. 2053-1064.

Ydstie, B. Erik Integrating Business, Information Technology and Process Systems Engineering, World Market Series, Business Briefing, CPI Technology , Jan. (2003), pp 71-72

Perea Lopez, E., B. Erik Ydstie and Ignacio Grossmann, A model predictive control strategy for supply chain management, Comp. & Chem. Eng . Vol. 27, (2003) pp 1201-1218.

Garcia Osorio, V. and B. Erik Ydstie, "Distributed, asynchronous and hybrid simulation of process networks using recording controllers," Int. Journal of Robust and Nonlinear Control , Vol. 14, (2004) pp 227-248

Hill H. Jennifer and B. Erik Ydstie, "Adaptive Control with Selective Memory," Int. Journal of Adaptive Control and Signal Processing , 2004, Vol 18, pp571-587.

Ydstie, B.Erik, "Distributed Decision Making in Complex Organizations: The Adaptive Enterprise," Comp. Chem Eng. 2004, Vol 29, pp11-27..

B. ERIK YDSTIE and YU JIAO "Passivity Based Inventory and Flow Control in Flat Glass, Manufacture", IEEE-CDC, Nassau , Bahamas , Dec 14-17, 2005 .

JILLSON, KENDELL and B. ERIK YDSTIE, "Complex Process Networks: Passivity and Optimality" To be published in Proc. of the IFAC World Congress, Praha, July, 2005.

RUSZKOWSKI, MARTIN, MARK READ, ROBERT KAISER, PHILIP RICHARDSON, TOM KERN, B. ERIK YDSTIE "Passivity based control and real time optimization of a Silicon process", to be published in the proceedings of the IFAC World Congress, Prague, Czech Rep. July, 2005