SIMULATION GAMING AND NATURAL LANGUAGE PROCESSING FOR MODELLING STAKEHOLDER BEHAVIOR IN ENERGY INVESTMENTS

摘要

Investment decisions in energy infrastructure are multi-stakeholder decisions. Stakeholders are not only Transportation Systems Operators (TSOs) but also banks, energy traders, producers, consumers, governments and regulators. Classical approaches for modeling investment decisions are often inadequate to support decision makers in energy infrastructure investments. In the classical approaches, a single party's decision problem is modeled and the optimal course of action is selected. In a multi-stakeholder situation, however, each party has to acknowledge other stakeholders' autonomous decisions. This introduces uncertainty which cannot be modeled mathematically, but has to be handled by negotiations or in scenarios. Accordingly, this paper describes an approach to study such decision making situations by simulation gaming with (representatives of) real stakeholders. The games aim to capture the considerations taken into account by the players in order to learn about their behavior. The paper describes an innovative way of capturing these considerations empirically by using natural language processing technology to analyze written conversations and negotiations. The game is currently designed for decisions on expansion or renewal of the gas transportation infrastructure, but the technology could be applied to study multi-stakeholder decision making in more general settings.