Sensitivity Constrained Nonlinear Programming: A General Approach for Planning and Design under Parameter Uncertainty and an Application to Treatment Plant Design

摘要

One important problem with using mathematical models is that parameter values, and thus the model results, are often uncertain. A general approach, Sensitivity Constrained Nonlinear Programming (SCNLP), was developed for extending nonlinear optimization models to include functions that depend on the system sensitivity to changes in parameter values. Such sensitivity-based functions include first-order measures of variance, reliability, and robustness. Thus SCNLP can be used to generate solutions or designs that are good with respect to modeled objectives, and that also reflect concerns about uncertainty in parameter values. A solution procedure and an implementation based on an existing nonlinear programming code are presented. SCNLP was applied to a complex activated sludge wastewater treatment plant design problem.