Back to search

ENERGIX-Stort program energi

A New Model For Power Markets Under Uncertainty

Alternative title: En ny modell til kraftmarkeder under usikkerhet

Awarded: NOK 2.5 mill.

Project Number:

245448

Project Period:

2015 - 2017

Funding received from:

Location:

Subject Fields:

The future of any market is subject to large levels of uncertainty, and this is true also for power markets in the Nordics and elsewhere. Demand levels, fuel and emissions costs, generation technologies and costs, as well as weather-driven aspects such as hydro inflows, wind production, and short term demand changes, all drive this uncertainty. Such uncertainties can be classified into overall trends (e.g. demand trends, trends in overall levels of system inflow) and dynamic uncertainties (when exactly could hydro inflows arrive and how can this vary). To address this challenge, many actors use computational models of the power market within their medium-to-long-term analysis activities. In the Nordic market in particular such an approach is widespread. From a modeling perspective, taking account of uncertainty is hard. Models that do so are complex and take a long time to execute, and they focus on very few (generally one) source of uncertainty. This narrow focus and inability to capture a fuller range of uncertainty limits the effectiveness and accuracy of these models. The complexity and long execution times mean that the models are often difficult to use. A more accurate and useable model is an essential tool decision making in power markets. Our project has focused on developing just such a model. Our work consisted of 5 main research and development areas: analysing how each uncertainty driver affected market outcomes; surveying model users about how they use market models and which features and behaviours are critical for them; analysing alternative modelling methodologies and choosing one for our model; extending the methodology to improve solution accuracy and solve times, as well as account for the results of the user analysis; and implement the model within standard data science analysis software. The resulting model, called PUMA, models uncertainties in power demand, inflow, renewable production, infrastructure availability, and upstream fuel prices. PUMA is designed to model the markets as they are operated, capturing explicitly both market forward curves and day-ahead market outcomes, over medium-long term time frames ("now to many years ahead"). Thus PUMA can be readily extended in future research to model investments, related markets (such as capacity markets), and short term intra-day and balancing markets within a consistent framework. It is implemented as a set of python libraries, and thus works within standard data science analysis workflows.

The future of any market is subject to large levels of uncertainty, and this is true also for power markets in the Nordics and elsewhere. Demand levels, fuel and emissions costs, generation technologies and costs, as well as weather-driven aspects such as hydro inflows, wind production, and short term demand changes, all drive this uncertainty. Such uncertainties can be classified into overall trends (e.g. demand trends, trends in overall levels of system inflow) and dynamic uncertainties (when exactly could hydro inflows arrive and how can this vary). To address this challenge, many actors use computational models of the power market within their medium-to-long-term analysis activities. In the Nordic market in particular such an approach is widespread. From a modeling perspective, taking account of uncertainty is hard. Models that do so are complex and take a long time to execute, and they focus on very few (generally one) source of uncertainty. This narrow focus and inability to capture a fuller range of uncertainty limits the effectiveness and accuracy of these models. The complexity and long execution times mean that the models are often difficult to use. Actors need to thus dedicate significant resources (human and IT) to using these models; something that is expensive for many market participants. The question that our Innovation Project is designed to address is thus: How can we take advantage of existing and develop new modeling methods to build a power market model that accounts for many sources of uncertainty, and the fact that we may not know too much about these uncertainties, whilst being as easy as possible to use? Our project will develop new and improve on existing modelling methodologies and use these to build a commercial power market model for use in medium and long term market analysis work. The resulting model will be designed to take account of a wider range of uncertainty than existing models.

Funding scheme:

ENERGIX-Stort program energi