Constructing aggregated time series data for energy system model analyses

With application of energy system models one of the challenges is to achieve an appropriate balance between the number of time segments (for example, hours) in the model, the effort in solving the model and the quality of the model results. This challenge is getting more intense with increased amounts and sources of variable renewable energy, because time series for such sources display characteristics that are distinctly different from one another and from those of electricity and heat demand. This paper presents a new method for aggregating time series for energy system analyses. The method is applied to a large scale energy system model to illustrate and validate the developed method. The paper discusses criteria that may be relevant for the evaluation of the quality of a given time aggregation. This depends on the composition of the energy system in question as well as the focus of any given project. Relevant project foci include investments in renewable energy, support mechanisms, electricity prices, the role of flexibility on demand vs supply sides, transmission and bottlenecks, and emissions. The proposed method was applied to the large-scale Balmorel energy system model with data representative of the Nordic energy system. A selection of countries was chosen to feature different challenges in power market modelling. The original input time series are given on hourly basis for a full year, while aggregated time series with various aggregation resolutions down to almost 1% of the original are constructed and applied. The changes in dynamics in both the seasonal and short-term perspective are observed for various output variables. The key finding of the paper is that although the chosen aggregation technique was generally successful in regards of reduction of solution time and also for the accuracy of some of the results, attention should be given to choosing the aggregation strategy according to the investigated task at hand. Results indicate that electricity prices and fuel use (and thus also emissions) are fairly robust to aggregation in the levels tested, while investments are more sensitive.