Process system manufacturers face a dilemma in designing new products.
There are economic trade-offs between the two extremes of providing a
customized design for each application, and providing only a few stan
dardized designs. To solve this problem, we must have knowledge of pro
cess design and the customers. We are integrating superstructure optim
ization through MINLPs for the optimal design with knowledge of the di
stribution of the customers in a ''requirement'' space. A master mathe
matical program is being developed to link together this information t
o solve the overall question of how to most economically provide stand
ardized designs for the customers. An example is presented with our fo
rmulation, along with the results of an enumerative solution.