EECinsights - Quantifying the Value of Information in Energy Systems

The behaviour of future energy systems is influenced by a broad and diverse range of uncertain factors, such as the characteristics of generation and storage technologies, the nature of energy demands, and weather and climatic patterns that drive power generation from renewables.

When designing and operating decarbonised future energy systems, decisions must be taken whose outcomes are impacted by the true state of the uncertainties within the energy system. A key example of this is the field of ‘capacity expansion planning problems’, where the task is to identify the minimal cost design of a low-carbon energy system for future implementation. Understanding the propagation of uncertainties through decision problems, and their impact on the outcome or quality of the candidate decisions, is critical to enabling the development of decarbonised systems that perform well and are robust to the uncertainties that influence their behaviour.

In my MRes project I investigated how a Bayesian Decision Analysis technique called Value of Information Analysis (VoIA) could be applied to energy systems problems to quantify the impact that uncertainties have on the decisions made on energy systems. This methodology can be used to determine which uncertainties are most critical to a given decision task, as well as develop decision strategies that are more robust to the underlying uncertainties of the system.

Through this work I also considered a potential extension to the VoIA framework which allows for the creation of more general Value of Information metrics that can be used to perform similar uncertainty impact quantification analyses on far more complex decisions problems, which are intractable under the classical Bayesian Decision Analysis framework.

I will be continuing to develop this work throughout the 1st year of my PhD, and will be particularly targeting the application of the extended VoIA framework to Linear Programming and Reinforcement Learning based decision agents. I will also be exploring the range of energy system decision problems to which this analysis methodology can be applied, to demonstrate its ability to improve our understanding of how decarbonised energy systems should be designed and operated so that they perform well in an uncertain world.

You can find out more about my research into VoIA for energy systems in this article.