Optimal Design and Retrofit of Sensor Networks

University Of Oklahoma Norman Campus, Norman OK

Investigators

Abstract

Research: This project will address the development of new methodologies for the design of instrumentation networks in chemical process plants. The PI has been working on developing methods to obtain cost-optimal sensor networks and he has addressed new concepts including estimability, residual precision, resilience and gross error detectability of sensor networks. He has also looked into the consideration of maintenance (corrective and preventive) and has proposed several methods for the cost efficient upgrade of sensor networks. The project will continue the work in this area. Existing solution procedures do not perform well for medium industrial size problems, the ultimate user of this technology. Therefore, the present project will develop new and more efficient numerical procedures to solve instrumentation design problems. Specifically, the project will develop: (a) new network performance measures: residual reliability and accuracy of estimators; (b) an unconstrained optimization model (such models relate all sensor network properties to economic measures, allowing the problem to be solved as a net present value maximization problem); (c) inclusion of financial risk management; and (d) new and more efficient numerical procedures. The use of groups of variables (members of cutsets), instead of one variable at a time, in branch and bound procedures, will be explored. Broader Impact: This project addresses critical issues that will facilitate industry's task of retrofitting more sophisticated instrumentation as it becomes available. The project has industrial support (OSI software) and therefore technology transfer will be facilitated. Findings of this research will help produce a generation of engineers capable of looking at plant operations issues with a different perspective. The project therefore has the potential to run chemical plants more efficiently and thus positively impact the US economy. The PI himself is a member of an underrepresented group and is planning to involve students from underrepresented groups in the project.