It has been argued recently (Bagajewicz, Chmielewski and Rengaswamy, Proceedings of Sensors Topical, AIChE Annual Conference, 2004) that the sensor network design problem can be best viewed as one where the difference between value of information and cost is maximized. This new approach, the value-paradigm sensor network design, was conceptually discussed by the three authors from three different perspectives: process monitoring, process control and process fault detection. The new idea will also be presented in area 10e "Smart Plants" session. In a nutshell, sensors need to be selected and positioned in such a way that the value of the information obtained minus its cost is maximized. The value of the information is in turn, obtained from the relationship between the accuracy of the data and the resulting economics of the plant.

In this paper, we will show new techniques to solve this problem using tree search enumeration with pruning criteria. We will also show how some suboptimal solutions can be obtained using some approximate methods.

Finally, we will extend the search to evaluate the economic value of a sensor network when used in conjunction with Real Time Optimization. Specifically, a sensor network reduces uncertainty on constraints of the optimization model and therefore the set points obtained render an operating point closer to the true objective.

Examples will be provided.