MIXED-INTEGER PROGRAMMING FOR PRESSURE SWING ADSORPTION CYCLE SCHEDULING

A model is presented that allows the cyclic operating schedule of a pressure swing adsorption system to be determined from the sequence of operations that are to occur in the beds. The problem is structured as an optimization program. The model, as it is formulated, represents a mixed-integer nonlinear program. Because of the solution difficulty of such a problem, it was desired that the model be reformulated. For almost all cases, with the proper variable transformations, the nonlinear constraints may be linearized. The model then becomes a mixed-integer linear program. Since the models use only a moderately large number of binary variables, they can be solved quite efficiently. An extension of the original model, termed the relaxed existence model, allows a more versatile operating sequence to be used since a superstructure of all possible operations may be specified. The choices given then allow the best schedule to be determined by the optimization procedure. © 1990.