**If you shifted some of your electricity usage, what**

**Figure 9.** Decision basis for demand response actions.

#### **4. Follow-On System Modelling and Commercial Implementation**

## *4.1. Agent-Based Modelling*

To investigate the wider applicability of this business and technical approach to localisation of energy use, an agent-based model (ABM) was constructed embodying both the technical features and the economic and social incentives of the real-life CEGADS trial. The model used the CASCADE framework [29] in which each agen<sup>t</sup> is a household for which their energy-using appliances, energy flows, costs, and decisions are simulated taking account of their physical and social environment. The environmental factors included weather, tariffs, and the feedback and encouragemen<sup>t</sup> provided by participation in the project. The governing attributes of each agen<sup>t</sup> (such as the number of household occupants and size of dwelling) were given values selected randomly from an appropriate range and distribution. Each run of the model (typically simulating a year's operation) therefore gave different outcomes, and as is conventional for an ABM, interpretation of results is based on the range of outcomes.

The initial correspondence between empirical and modelled results indicated that the model was useful. A series of tests were then undertaken to test how robust the positive results of the empirical trial were to changes in the scenario. Firstly, different weather files were used to investigate how dependent the overall results and benefits to individuals were on weather. This showed that the important characteristic of no participant being expected to lose out financially was preserved in differing weather conditions. Next, several communities were programmed to co-exist in the model and the smart signals sent to consumers in each model were examined after the model had evolved. It was noted that the signals exhibited similar characteristics, but that they evolved in a way that was specific to the community—indicating that the model was transferrable, but that the specific evolution in response to the demographic and technology mix within the community would differ, resulting in different levels and patterns of demand response.

## *4.2. Commercial Implementation*

The success of the CEGADS trial has led to a first fully commercial implementation of the concept for a community in the small town of Bethesda, North Wales [30]. This is based around a 100 kW micro hydro generator. 100 consumers have been recruited, who pay £0.07/kWh for matched use of local generation, and a small charge for membership of the co-operative club. Any electricity not supplied by the hydro is charged according to a time-of-day tariff similar to that in Figure 1. Because the output of this generator varies seasonally, power availability is signalled to users. Figure 10 shows the proportion of each user's power matched to low cost local generation, with an average of 65% over the year of operation. The overall shape of the graph reflects the availability of hydro generation, including the impact in May 2017 of a period of dry weather.

**Figure 10.** User demand matched to local micro-hydro.

The social enterprise promoting this model, Energy Local, is now developing a "starter pack" of processes and documentation [31] with follow-up support, allowing social enterprise clubs to be formed and the model implemented wherever appropriate generation, network configuration and community enthusiasm exist. New micro-hydro clubs are in progress elsewhere in Wales, and PV-based clubs in London and Gloucester.

## **5. Discussion**

A limitation of this study is that there is no unambiguous counterfactual against which the impact on electricity use of the full range of interventions can be assessed. The process of recruiting participants and installing equipment in their homes inevitably involved engaging them with the objectives of the study. So the comparison of electricity demand measurements from the end of the study with those at the start cannot fully demonstrate the behavioural shifts that may have occurred, but they do reliably show the effect of the heating control and battery technologies. With further development, the ABM modelling techniques tested as part of this study could provide policymakers with additional predictive insight on the impact of wide-scale adoption of this kind of energy localisation scheme.

The increased demand response from those participants who had a significant supporting technology is very evident from Table 1. The smart plugs that were given to all participants clearly did not have the same impact as the other technologies. The increased response from participants with controlled electric heating and PV is notable since both started the trial with an incentive to attend to the timing of electricity demand, in the heating case through their longstanding use of the Economy 7 tariff, and in the PV case to make use of their own generation. So it must be concluded that their initial

sensitivity to timing was further reinforced through the various inputs categorized in Figure 7, and for the heating users by the control signal alignment with the tariff. For the battery-equipped participants, the response seen fully reflects the automatic operation of the technology because the batteries were installed in January 2016. For the remaining participants it is evident from the data that only a few enthusiasts were able to sustain an increasing level of demand response that could be detected in the start-to-finish comparison. This does not of course preclude that many of this group may have taken demand response actions on an intermittent basis as suggested by the interviews. These results are very consistent with previous studies such as the review of 21 trials by Frontier Economics [32] which has a key finding "Interventions to automate responses deliver the greatest and most sustained household shifts in demand where consumers have certain flexible loads".

The financial benefit of the business model is clear from Figure 5, amounting to an average of £109 to each participant over a year, on average consumption of 4854 kWh that would cost £688 at the benchmark fixed rate used as a comparator of £0.135/kWh and standing charge of £0.09/day. These substantial savings illustrate the value that will be made accessible by the UK national smart meter rollout as long as regulation, and suitable smart meter data processing capability, facilitates this form of community energy scheme and tariff structure. The way in which the benefit of the PV generation is spread across the community makes this model particularly attractive for social housing, where otherwise the restriction of local generation to a subset of dwellings with favourable roof orientation can lead to perception of unfairness [33].

The recent UK regulatory review [22] has identified the issue that consumers who benefit from local generation, as in the present scheme, pay less towards the distribution and transmission infrastructure through per-kWh tariffs, but the burden they place on that infrastructure is determined by their peak network demand. This is ultimately likely to lead to tariff structures that are less dependent on the volume of electricity consumed and have some dependency on peak demand, such as those proposed by Simshauser [34] and Nijhuis et al. [35]. The battery and smart heating control technologies demonstrated in this project directly address this issue by substantially reducing peak demand.

This trial also provides a more positive perspective on the economics of home batteries than other trials such as Uddin et al. [36] who tested exactly the same product as used in the present project and concluded there was no saving from increased self-consumption of PV generation and a substantial loss to the user from battery depreciation. It is worth noting that the trial reported by Uddin et al. was purely techno-economic, with empirical testing in a single household as the basis for economic modelling: there was no community or social dimension. In the case of the CEGADS/SWELL trial, the battery users made savings at an average rate of about £23 per annum from tariff arbitrage, by avoiding 1 kWh per day in the evening peak rate period and drawing the corresponding charge overnight. Clearly this alone would not justify the investment, or cover the depreciation, but in combination with other grid services such as Short Term Operating Reserve, for which the UK system operator National Grid currently pays about 12 p/kWh [37] adding c. £10 per annum to the return, a path to viability, as battery costs fall, can be seen. Another role for batteries relevant to community energy is to facilitate connection of generation capacity. In [38] Idlbi et al. elaborate a case study in which 0.5 MWh of lithium battery capacity enables connection of 3 MWp PV generation to a network that would otherwise require reinforcement at much greater cost to maintain voltage compliance. However, batteries are not essential to the value of the Energy Local model, as shown by the results from the Bethesda follow-up project, which did not deploy them.

## **6. Conclusions**

This trial, combined with the subsequent commercial implementation of the business model, has demonstrated that valuable technical, economic, and social outcomes can be achieved by localisation of electricity generation and consumption within a community-of-place-based organisational framework. The generally positive user experience described in Section 3 reflects

the support given to participants by the project team allowing them to understand and engage with the novel technology and tariffs. The business model delivered useful financial savings for consumers which have been shown to be repeatable elsewhere, while the technology successfully demonstrated the use of both electrical and thermal energy storage to reshape the daily profile of electricity demand, in response to technical and financial signals, such that peak demand is reduced and local consumption of local generation is increased. This demand response was stronger for the automated mechanisms managing energy storage than from the simpler devices providing appliance scheduling which required repeated user configuration. These results show that the sustainability of the transition to renewable energy can be strengthened with a community-oriented approach that supports users through technological change and improves the return on investment by localising generation and consumption. The enterprises and institutions taking part in this trial are now seeking to build on this experience by evolving the business model and technology so that they can be widely deployed, and are also motivating adjustments to the regulatory environment that facilitate local initiatives and consumer engagement.

**Author Contributions:** Formal analysis, P.B., J.R.S., R.M. and J.H.; Investigation, P.B., J.R.S., R.M., J.H. and S.D.; Writing—original draft, P.B., J.R.S., R.M., J.H. and S.D.

**Funding:** As detailed under Acknowledgements and Conflicts of Interest below.

**Acknowledgments:** The authors would like to thank the Engineering and Physical Sciences Research Council (EPSRC) and Innovate UK for providing the financial support for this study as part of the CEGADS project (EP/M507209/1 and EP/M507210/1), including funds for covering the costs to publish in open access. The project implementation was also part funded by Energy Local (Development) Ltd., Exergy Devices Ltd., Moixa Technology Ltd., Westmill Sustainable Energy Trust, and Co-Operative Energy who provided the supermarket vouchers to participants reflecting their benefit from the simulated tariffs.

**Conflicts of Interest:** Two authors (Boait and Morris) have roles in the enterprises acknowledged above that provided match funding and technology for the project. None of the other authors has a conflict of interest which could inappropriately influence this work. The results reported were generated and reviewed by all the authors. This work (i.e., the data analysis and drafting of the manuscript) was funded by the UK Engineering and Physical Sciences Research Council and Innovate UK. They had no role in the collection of data or in its analysis and interpretation in the paper and none in the decision to submit to this journal.
