Development of Inexact Optimization Methods for Planning Environmental Management Systems Under Multiple Uncertainties

Abstract

Environmental planning is becoming a key component in environmental management due to rapid socio-economic development and steadily increasing population. In real-world applications, uncertainties and complexities are frequently involved in environmental management systems, where decision makers need sound management approaches in order to allocate their limited resources to different competing end users. Thus, more innovative approaches are required to deal with multiple uncertainties. As well more in-depth information in regard to the individual and interactive effects among multiple input parameters on the system performance is needed. In this research, a dual-interval two-stage mixed-integer inexact-chance-constrained linear programming (DITMIC) method was developed for the planning of flood-diversion management systems. DITMIC did not only reflect uncertainties expressed as intervals, dual intervals, interval-valued probability distributions but also incorporated pre-regulated diversion policies into its optimization process. Various policy scenarios were generated and they were associated with varied economic penalties if the promised targets were violated. Furthermore, DITMIC assisted in analyzing dynamic features of capacity-expansion schemes when it came to decision-making. Various flood-diversion patterns and capacity-expansion schemes were obtained under different risk levels. The reliability of satisfying the system constraints under uncertainty was also examined. This enabled decision makers to choose the most desired outcome relative to their preference and perception of future conditions. To better enable insights on the detailed effects from uncertain parameters as well as their interactions on the system objective, a factorial dual-interval programming (FDIP) method was developed for planning municipal waste management systems. Through the integration of factorial analysis and dual-interval linear programming into a general framework, the FDIP method could handle uncertainties existing in the left- and right-hand sides of the objective function, as well as in the associated constraints. Moreover, it had the advantages of identifying influential parameters along with their joint effects on the system output. Impact factors as well as their interactive effects have been identified and analyzed for the lower and upper bounds of the system output, which could further provide valuable information on their effects on the system solutions when it came to decision-making.