Resolving Conflicts among Stakeholders in Real-time Operations

Planning, scheduling, and control decisions often involve conflicting priorities from multiple stakeholders (e.g., due to different perceptions of risk). We present a new framework for multi-stakeholder optimization to compute optimal compromise solutions among stakeholders. In this setting, stakeholder opinions are interpreted as random variables, establishing a parallel between stochastic and multiobjective optimization. Risk metrics are used to shape the distribution of stakeholder satisfactions. We demonstrate the approach by considering the operation of a combined heat and power utility system that monetizes excess capacity by participating in energy markets. Stakeholders express their perceived risk by discounting (weighting) potential revenues from different market products. We find that despite strongly conflicting priorities, the proposed framework is able to identify efficient compromise solutions where each stakeholder is at least 95% satisfied. We compare compromise solutions computed using the conditional value-at-risk and entropic value-at-risk metrics and discuss how the selection of risk metric impacts the distribution of stakeholder satisfactions.