ABSTRACT
In offshore industries, the gas detector system with emergency shutdown employs a 2-out-of-N (2ooN) voting scheme to determine appropriate output action because of preventing process shutdown caused by spurious alarms. Arrangement of gas detectors in offshore plants is currently designed using trial and error method to detect flammable or toxic gases although the trial and error technique is time-consuming and has a difficulty to design gas detector layouts effectively. This paper proposes a predictive method to calculate detection coverage for 2ooN using response surface method. In order to establish a regression model for detection coverage, an initial layout for gas detectors is divided into several groups and a second-order prediction for a divided group is proposed. A regression model for a whole layout for gas detectors is determined using superposition of the second-order polynomial. An optimal layout for gas detectors is designed using the regression model and detection coverage of the optimal layout is increased in comparison with an initial layout.
