Chemical Process Optimal Design and Control with Account of Separate Probabilistic Constraints

The paper proposes a new approach to solve optimal chemical process design problems, taking into account the uncertainty of initial information. Solving the problem will allow us to design a new chemical process and get the optimal values of control parameters for the entire period of operation. The initial problem has a form of one-stage optimization problem with separate probabilistic constraints. The proposed approach is based on the use of a piecewise-linear approximation of constraint functions which have to be satisfied at the predetermined probability level, as well as the approximation of the region where constraints are met by a set of multi-dimensional parallelepipeds. This provides the possibility to transform individual probabilistic constraints into deterministic form and significantly reduce the problem computational time by eliminating the multidimensional integration operation. To exclude the multidimensional integration when calculating the mathematical expectation of the function which estimates the efficiency of the chemical process, an approximation of the problem criterion proposed by the authors earlier is used. The efficiency of the method is shown on a model example in comparison with other methods. The obtained solution was also compared with the problem solution in the formulation of a one-stage optimization problem with some probabilistic constraints when the probabilistic boundary of the function for estimating the system performance was used as a criterion.