Title: Decision Making under Uncertainty:A Systems Analysis Perspective Speaker: Urmila Diwekar, Vishwamitra Research Institute Date/Time: March 7, 2011, CA: 9:00-10:00am PT, NM: 10:00-11:00am MT Location: 915/W133 (CA), CSRI Building/Room279 (NM) Brief Abstract: Uncertainties are inherent in real world systems. Decision making in the face of uncertainties is an iterative process and involves computationally expensive algorithms. This talk presents a new and efficient algorithmic framework for decision making under uncertainty based on optimization and sampling algorithms. The new sampling algorithms are based on quasi-random numbers and Latin Hypercube Sampling Technique. Properties of random numbers are used to develop new optimization algorithms. Random numbers also play a key role in convergence properties of these algorithms. These algorithms are used to solve real world case studies in the area of energy and environment. The first case study in this talk is related to the nuclear waste problem at Hanford where discrete and continuous decisions are involved in the face of uncertainties. Efficient algorithms for discrete optimization under uncertainty are developed to solve this Herculean problem. The next case study comes from green energy where water is minimized in an existence power plant by nonlinear optimization in the face of seasonal variation. To solve this problem, a new algorithm based on Bay’s theorem is used. This algorithm is called Better Optimization of Nonlinear Uncertain Systems. Green process design and green energy systems involve multi-objective problems with uncertainties. These problems will be described next with real world systems from green energy. New algorithms for multi-objective optimization based on the sampling techniques described earlier will be presented. The concept of overall sustainability goes beyond green energy, green design and brings in time dependent nature of the ecosystem and multi-disciplinary decision making. Optimal control methods and theories from financial literature can be useful in handling the time dependent uncertainties in this problem. Decision making at various stages starting from green design, green energy, to industrial ecology, and sustainability is illustrated for the mercury cycling. Power plant sector is a major source of mercury pollution. In order to circumvent the persistent, bioaccumulative effect of mercury, one has to take decisions at various levels of the cycle starting with greener power systems, industrial symbiosis through trading, and controlling the toxic methyl mercury formation in water bodies and accumulation in aquatic biota.CSRI POC: Patty Hough, (925) 294-1518 |