Pricing Mechanisms for Energy Communities - The GridFlex Heeten Project
Victor Reijnders is a PhD student in the research group Mathematics of Operations Research (MOR). His (co)supervisor are prof. dr. J.L. Hurink and dr. ir. M.E.T. Gerards from the Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS).
One of the most challenging problems of the recent years is climate change. To decrease its effect, lowering CO2 emissions is crucial. One important step to take in this context, is the switch from using fossil fuels to using renewable energy sources. This switch also implies an electrification of appliances, e.g., electric vehicles, or heat pumps.
The resulting energy transition, however, causes problems in our electricity grid, as the electricity from renewable energy sources is often not produced at the same time (and location) as the electrical appliances need it. Moreover, this production, and the mentioned consumption, is highly synchronized. Next to a mismatch between supply and demand, this also causes huge peaks in the networks, leading to quicker degradation of assets, or even black outs.
At the low voltage level of the grid, this nowadays already causes issues. In certain neighbourhoods, solar panels need to be curtailed on sunny days, or at other times the simultaneous charging of electric vehicles may blow fuses. Fortunately, for these problems, solutions exists, amongst others, in the form of demand-side management, that aims, for example, match the consumption and production of renewable electricity to lower the stress on the grid. One way to apply demand-side management is with pricing. Pricing of electricity aims to activate people to use electricity at given moments. Furthermore, the usage of smart devices can be steered based on such prices.
As a single household does not provide enough flexibility, in this thesis, we focus on neighbourhoods, or energy communities, as a whole. This, e.g., supports households to capture the solar energy produced by neighbours. It is clear, however, that hereby not only technical aspects, but also a social component needs to be addressed.
The main motivation for the research of this thesis came from the GridFlex Heeten project, which focussed on implementing innovative pricing mechanisms in an energy community to lower the stress on the grid. The community considered in the project is located in Heeten, a village in the Netherlands, within a neighbourhood of 47 households.
The core focus of this thesis is on the design of pricing mechanisms for energy communities to help alleviate the stress on the electricity grid. Some of these pricing mechanisms were also used in the GridFlex Heeten project and corresponding field test. The specific contributions of this thesis are:
- Pricing mechanisms based on losses using grid topology: As the goal of the pricing mechanism is to reduce the stress on the grid, a logical choice is to have the pricing somehow reflect the costs that occur in the grid. Losses give a good indication of the stress on the assets in the grid and are used as basis of the pricing mechanism.
The proposed pricing mechanism is based on the Shapley value and it distributes the costs based on the average marginal contribution of a household to the neighbourhood losses. This results in a pricing mechanism with the `polluter pays' principle. Normally, calculating the Shapley value is computationally inefficient, but as losses scale quadratic with the power consumption, it can be calculated efficiently for this specific case.
The disadvantage of this pricing is that the costs assigned to the households are highly dependent on the location of the household in the electricity grid. As this is not seen as a fair criterion by consumers, we introduce the concept of an average location of households in the grid. For this, we permute the location of the households and use the Shapley value based on the caused losses to assign the costs. These permutations of location are used in two ways: either by considering all possible permutations and taking the average over the resulting cost assignments (called the average location cost), or taking the average of the cost assignments using only the permutations where the location of a pair of households are exchanged (called the approximate average location cost). The latter can be calculated much faster, but keeps more locational bias.
Calculating all the Shapley values for all these permutations (in case of the average location cost) normally would be computationally infeasible as the number of permutations scales factorial with the number of households in the general case. However, when considering radial grid structures, explicit expressions can describe the costs. For the approximate average location costs, the number of considered permutations scales only linearly with the number of households in the general case, so extensions to different grid structures can be done more easily. For radial grid structures, explicit expressions for the approximate average location cost can be found in a similar way as for the average location cost.
For the resulting two cost assignments, larger consumers still have to pay higher costs, showing the possibility to address the locational bias that arises from the `polluter pays' principle.
- A hybrid pricing mechanism for joint system optimization and social acceptance: As with the aforementioned pricing mechanism, social acceptance is crucial to take into account. Therefore, a framework for local electricity pricing mechanisms focussed on social acceptance is proposed. The goal of these mechanisms should be to flatten the overall electricity profile of the neighbourhood. In the proposed mechanisms, the price of electricity depends on the electricity load of the neighbourhood, and it is based on a linear price function, as this achieves the goal mentioned. However, these cost functions are deemed to be too complex, and consumers are generally unwilling to participate in systems offering these prices.
The problem with simpler pricing mechanisms that are accepted by consumers, is that they do not always help to achieve the intended goal, or might even worsen the situation. Therefore, the challenge is to bring together these conflicting aspects in a pricing mechanism: low complexity and flattening the neighbourhood load. For this, an electricity pricing is proposed that is a step-wise function that expresses the price per kWh for individual consumers based on the overall neighbourhood load. The resulting cost function is piecewise linear and approximates a quadratic cost function. These cost functions penalize periods with high neighbourhood peaks, and incentivize the neighbourhood to flatten their overall load.
Within the given framework the number of pieces and the prices per piece can be tailored to the neighbourhood. This way, the prices can be made fair, and the neighbourhood receives suitable incentives to lower or flatten their electricity load. A further advantage of this pricing is that the prices only change a couple of times a day, giving the consumer more certainty.
The performance of the presented pricing mechanisms is tested under various conditions and compared to other steering mechanisms. This hybrid pricing mechanism was implemented and tested in the GridFlex Heeten field test. Both the results from GridFlex Heeten as well as those from simulations are comparable to that of quadratic costs, while having low computational complexity. Furthermore, based on the feedback of participating consumers and criteria from literature, we conclude that the proposed mechanism is socially accepted. This shows the potential for these pricing mechanisms in practice.
- An overview of the GridFlex Heeten field test and research project: The aim of the GridFlex Heeten project was to test innovative pricing and steering mechanisms at a field test location with local electricity production and storage. This could result in a local energy market which should be scalable, and aim to improve the local matching of supply and demand. This thesis addresses the main project results, thereby providing a reference for future research projects.
These project results mainly comprise the pricing mechanisms created and tested in the project, namely the mentioned hybrid pricing mechanism, and a neighbourhood net-metering mechanism. In the latter, the consumers pay more for importing energy from outside the neighbourhood, and receive less for exporting it compared to using it within the neighbourhood. This stimulates self-consumption within the entire neighbourhood.
The information on the used prices was shared with the participants via an app. With this information they could adapt their electricity usage to save money. Furthermore, batteries were installed in the neighbourhood with a controller that responded to the given prices. This resulted in annual savings of € 1,403.38 and € 753.47 for the two considered pricing mechanisms, respectively.
Statistical tests showed that the participants did not structurally change their behaviour based on the pricing mechanisms. This was also confirmed by a neighbourhood team, which consisted of participants who were more involved in the project.
With the proposed pricing mechanisms, the potential for using pricing mechanisms to alleviate the stress on the grid, while taking into account the social acceptance was shown. However, before these pricing mechanisms can be applied outside of research projects, many challenges still need to be tackled. Some of such important aspects are the legal frameworks, the required infrastructure, and the necessary investments.