A simulation-based fuzzy possibilistic programming model for coal blending management with consideration of human health risk under uncertainty

Abstract

In this research, a simulation-based fuzzy possibilistic programming (SFPP) model was advanced through integrating California puff (CALPUFF), fuzzy sets theory and inexact optimization within a general framework. It has advantages in uncertainty reflection, pollutant dispersion modeling, and the management of coal blending and the related human health risks. The developed SFPP model was solved through a direct search approach which coupled fuzzy simulation and Genetic Algorithm (GA). This approach can not only handle a coupled simulation–optimization problem considering uncertainties that can be expressed as fuzzy sets, but also provided the additional information (i.e. possibility of constraint satisfaction) indicating that how likely a decision maker can believe the decision results. It also can reduce the chances of being trapped in local optima as GA converges to global optima. Moreover, the employed direct search method can avoid the approximation error originating from surrogate simulators and enhance the confidence level of the generated optimal solutions. The developed model was applied to the planning of coal blending in Gaojing and Shijingshan power plants in the west of Beijing. The results indicated that the developed SFPP model was useful for generating a series of coal blending schemes under different acceptable possibility levels, ensuring that the risk to human health reduce to an acceptable level, identifying desired coal blending strategies for decision makers, and considering a proper balance between system costs and acceptable possibility levels.