Robust Optimization of Sustainable Resource Planning Under Stochastic Market Conditions

Abstract

In this ever-changing and unpredictable market, sustainable resource planning requires decision-makers to strike a balance between the conflicting goals of cost-effectiveness, risk avoidance, environmental concerns, and optimum operation.Traditional decision-making models with hard rules and empirical or statistical predetermined probabilities will be less effective when the system actually operates differently than originally modelled. Similarly, overprovisioned solutions, which are robust and static, do not efficiently allocate the resources when operating conditions change significantly. The objective of this research is to provide a new framework called Scenario Feedback-Adaptive Optimization, which consists of techniques such as multi scenario analysis, multi-objective decision-making, planning for short horizons and adaptive updating of models with new information.The approach draws on actual market experiences to enhance our understanding of uncertainty and support decision-making; the planning process thus becomes adaptable when moving through time. The ultimate objective is a system capable of handling market uncertainty; it handles multiple alternatives and a range of different values at any one time and attempts to balance mean cost, damage, and risk with the environment. Actual market conditions are then compared with the created forecasts at the end of the planning period.When there is uncertainty, these fluctuations influence the likelihood of each event occurring and set new bounds of feasibility. These revised numbers are used in the planning stage. This methodology allows the planning process to adapt to changes in the underlying market circumstances, and the feedback system can tweak the resource allocation in light of new developments; it does not depend on a static view of uncertainty.The comparison of these models (deterministic, stochastic, static, robust rolling-horizon, robust, and SFAO) in terms of cost CVaR, environment, service level, and recovery constitutes an experiment to evaluate the proposed approach. This evaluation provides a generic framework for the assessment of resilient resource planning and shows how feedback from observed scenarios can further improve decision-making for real-time uncertainty management in dynamic markets.