(248b) Multi-Objective Optimal Sensor Deployment Under Uncertainty for Advanced Power Systems

Authors: 
Tyagi, S., University of Illinois, Chicago and VRI-CUSTOM
Diwekar, U., Vishwamitra Research Institute, Center for Uncertain Systems: Tools for Optimization and Management
Bhattacharyya, D., West Virginia University

Advanced power plants using an integrated gasification combined cycle (IGCC) offer a competitive and economical means to produce electricity and thermal energy with significantly reduced emission levels. An efficient, safe, and reliable operation of an IGCC plant requires effective strategies for monitoring and control. Sensors play a key role in monitoring and control of the IGCC plant. The sensors in an IGCC plant operate under harsh environments leading to failures. Sensor failure  can not only create operational difficulties but can also lead to poor monitoring of the process equipment and inability to meet environmental emission limits. Sensors for IGCC application vary widely in cost, accuracy, and reliability. The objectives of the proposed research is to develop a fundamental understanding of the relationships among the sensor placement, interaction with the process, and hierarchal interactions of the sensor intelligence that can help in identifying the type, number, and location of sensors for maximum effectiveness and efficiency of the measurement technology and the process.  The optimization of the location, number, and type of sensors will contribute to enhanced control of a process, and to this end, new fundamental algorithms and hybrid hardware- virtual sensor architectures are proposed to be developed.   In order to synthesize a sensor network that can maximize the  observability, and efficiency, a multi-objective optimization framework is necessary.  Uncertainties are inherent in such a problem.  Therefore, the problem of sensor placement is formulated as a multi-objective stochastic programming problem, the solution of which is  the optimal sensor network for an IGCC plant in the face of uncertainties in performance, reliability, efficiency, and power demand. In order to solve this large scale multi-objective optimization under uncertainty, a new algorithmic framework is developed. The framework have components such as (1) an efficient sampling technique for uncertainty analysis, (2) the decomposition strategies used in L-shaped method for stochastic programming, (3) a better optimization of nonlinear system (BONUS) algorithm to circumvent multiple model runs for sampling at each optimization iteration and (4) an efficient multi-objective optimization algorithm that evaluates trade-offs in sensor network designs for various objectives such as efficiency, observability, and cost.