(624n) Automatic Generation of Optimal System Configurations and Maintenance Policies for Protective Systems

Generally speaking, a protective system is used to perform two basic functions, namely alarm and shutdown. The former is facilitated by independent sensors. Based on online measurements, a predetermined logic can be applied to make alarm decision. The latter function is usually performed with actuators, e.g., solenoid valves. Since sensor and valve failures are basically random events, the availability of a protective system is highly dependent upon its configuration and the corresponding maintenance policies.

Any hardware item can fail either safely (FS) or dangerously (FD). The FS malfunctions are recoverable because they are mostly caused by noisy signals, whereas the FD failures often require repair or replacement. To achieve a desired availability level, a common practice in the process industries is to introduce hardware redundancy in design. Specifically, each critical process condition may be monitored with more than one independent sensor, and the measurements signals are sent to a voting gate to generate alarm decision. These sensors are assumed to be maintained with spare-supported maintenance program in this study. On the other hand, it should be noted that the FD failure of a shutdown unit cannot usually be detected online. A preventive strategy must be applied to enhance its availability, that is, every unit is required to be inspected regularly at constant time intervals. The broken equipment must be replaced or repaired when detected, whereas the normal ones are kept online until the next inspection. Therefore, the length of the inspection interval should be considered as a design variable.

There have been a few attempts in the past to generate the optimal configurations and/or the corresponding maintenance policies with a mathematical programming model, e.g., Liang and Chang (2008), and Liao and Chang (2010), in which the failure rates of all compoenets were assumed to be constant. The objective of the present study is to develop a generalized mathematical programming model to produce the optimal design specifications and maintenance policies of a comprehensive multichannel and/or multilayered protective system by considering the time-variant nature of the failure mechnisms. By solving the corresponding mathematical program, the optimal configurations of sensors and shutdown units, the best corrective and preventive maintenance policies and alarm/shutdown logics can all be identified automatically.

References:

Liang K.H. and Chang C.T., 2008, A simultaneous optimization approach to generate design specifications and maintenance policies for the multilayer protective systems in chemical processes, Ind. Eng. Chem. Res. 47, 5543.

Liao Y.C. and Chang C.T., 2010, Design and maintenance of multi-channel protective systems,  Ind.  Eng.  Chem.  Res. 49, 11421.