(370j) A Stochastic Programming for Optimal Fence Line Monitoring and Detection of Toxic and Flammable Gases Using Fixed and Mobile Sensors

Recently, as the number of incident at the plants using chemical materials increases, Plants make efforts to improve level of process safety management. The most important thing among them is to detect chemicals leak as quickly as possible and prevent it to lead to a large disaster. Especially scale of damage of leak incident depends on how fast we take safety measure. It's ideal to manage all area using a lot of sensors but there are some limits such as number of sensors, budget, etc. For this matter, we need to research optimal sensor placement for each facilities. It's more efficient to place sensors on/at fence than inside of plant in respect of detecting time, coverage, cost for all leak scenarios. In this research, we introduce optimal fence monitoring system using fixed and mobile sensors for toxic and flammable gas based on CFD simulation. There are some factors considered significantly: process conditions, weather condition, type of sensors, geometries, etc. First getting the leak scenarios using CFD tools. Main parameters are selected based on CFD simulation for some simple cases that are chlorine gas and toluene gas leak from the real plant. Realistic leak scenarios are generated. Mixed-integer linear programming formulation is developed for optimal sensor placement by minimizing detecting time, the number of sensors and maximizing number of scenarios detected over all scenarios. The use of mobile sensor can reduce a lot of fixed sensors by covering some area with few mobile sensors. After these works, we compare 3 types of sensor placements (fixed only, mobile only, fixed and mobile) in respect of cost efficiency, detecting time, number of scenarios detected and propose optimal sensor placement among them. The result of this study can be helpful