An Extended RSQP Algorithm for Process Systems

In process systems there are many large-scale problems with large numbers of equality constraints, but their degrees of freedom are not small enough. To solve this kind of problems, an extended RSQP (reduced space sequential quadratic programming) algorithm was presented. In the extended RSQP, reduced space Hessian and the cross item were expressed and computed by limited memory method, so the memory needed to store these matrixes were reduced largely, Moreover, to avoid the storage of matrix C^{- 1}N, zero space coordinate was expressed implicitly, and rules for basis selection were replaced by a heuristic basis selection strategy. Performance of the new algorithm was test by several variable large-scale problems and two real cases. Computational results demonstrate that the new algorithm can largely reduce computing time and storage required