Building Social License for Automated Demand-Side Management—Case Study Research in the Swiss Residential Sector

Abstract

Demand-side management (DSM) is increasingly needed for answering electricity flexibility needs in the upcoming transformation of energy systems. Use of automation leads to better efficiency, but its acceptance is problematic since it is linked with several issues, such as privacy or loss of control. Different approaches investigate what should be done for building community support for automation for the purpose of DSM, but it is only recently that literature has shown interest in the application of social license as a concept merging several issues traditionally treated separately. The social license concept emerged in the mining sector before being adopted for other problematic resources. It serves to identify different levels of community support for a project/company as well as various factors that influence it, such as economic and socio-political legitimacy and interactional trust. This paper investigates, through empirical evidence from eight case studies, what has been done in different contexts to build trust and legitimacy for an automated DSM project. Our findings suggest that patterns exist in respect of benefits, risks and rationale presented, the retention of control, information gathered, and inclusion and that these factors differ according to appliances/devices automated, operators of automation, and end-users targeted.

1. Introduction

1.1. Background

The changes needed to meet the targets of the Paris Climate Accord require a major energy transition [1]. The energy landscape is transforming towards decentralised low-carbon energy systems with distributed energy resources [2]. These changes are seeing policymakers, energy users, and companies increasingly use language centred around the user. This discourse incorporates a more active role than merely addressing consumers, whereby users become prosumers investing in renewable energy technologies (e.g., rooftop photovoltaics) and storage capacities (e.g., batteries), therefore producing, storing, and exchanging energy services [3,4]. Transition towards renewable energy resources and electrification of energy demand-side leads to several electricity system issues [5], such as grid congestion due to additional total amount of electricity itself as well as additional power during peak times and network losses due to mismatch between stochastic and intermittent supply and demand, especially at the distribution level [6,7,8]. In order to avoid such issues, demand flexibility is emerging as increasingly important resource for electricity system operators to balance power systems specifically in the residential sector, mainly since that is where the potential of demand flexibility is high [9] as well as being highly needed [10]. Demand-side management (DSM), which refers to the measures that aim to influence magnitude and time patterns of energy demand, could constitute a major opportunity to provide demand flexibility in order to balance power systems [11]. These measures may require end-user manual response, as in the case of time-of-use tariffs, or real-time pricing without a firm commitment, where households respond to price signals [12]. Alternatively, DSM can be automated via (i) home energy management systems (HEMS) that households set up and (ii) other means such as direct load control (DLC), where a utility can switch on and off certain devices depending on the grid needs [12]. Considering the low response rate and/or reduced response in the long term to dynamic tariffs [13,14,15,16], automated DSM is increasingly considered as a more effective option to activate flexibility compared to pricing tariffs [17]. Indeed, response to price signals is limited since electricity consumption is inelastic and dependent upon timing of human activities rather than prices [15,18], and its end-users are also subject to inattention and decision fatigue [13]. On the utility side, automated DSM is recognised as a more “reliable and economic operation of power systems and electricity markets” [19] and a way to integrate renewable energies [20,21]. On the user side, it ensures a more economic and ecological usage of energy, avoiding efforts for end-users [22,23,24]. Nevertheless, despite the benefits for both consumers and utilities, the engagement in automated DSM programmes is low [25,26,27].

1.2. Previous Studies on Behavioural Barriers for Automated DSM Acceptance

In this section, we highlight several barriers or enablers for the exploitation of demand flexibility as discussed by diverse disciplinary origins, which can be summarised under six topics. Several researchers, specifically those with social practice theory perspective, indicate that (i) several researchers, specifically those with social practice theory perspective, indicate that providing demand flexibility (e.g., shift of energy use) is complex, as energy demand is directly linked to practices embedded in both social and temporal rhythms, such as working hours or meal times [28,29,30]. Therefore, configuration of how elements (e.g., norms, objects) of practices link together in and through performances across multiple temporal scales is a basis and a useful starting point for promoting demand flexibility [28].

Several researchers generally draw on a range of human-centred psychology to discuss various barriers and enablers for DSM. (ii) Researchers found that influencing energy demand may entail inconvenience, discomfort, or dangerous malfunctions, such as drop down of indoor temperature [31], bacterial proliferation [32], or potential fires [33]. These differ in their nature and their perceived intensity depending on practices where flexibility is provided [34] but also among the end-users targeted [35]. The socio-technical potential is derived from the technical potential focusing on devices/appliances with high amounts of energy consumption that can be technically postponed (e.g., heat pumps, EVs, wet appliances, and freezers compared to small appliances such as TV or iron) with consideration regarding acceptance likeliness (see [36]). Against this background, the literature suggests that appliances/devices with a fine balance of socio-technical potential for automated DSM are mainly washing machines [37], heating systems [38,39] or batteries (which include EV charging) [40,41,42]. It seems that the less that devices/appliances are embedded in social rhythms, the more likely they are to be accepted for DLC [43]. For example, washing machines are used in households when laundering activity occurs [44]; thus, altering washing machines’ usage requires the end-user’s willingness to change the timing of their activity. On the contrary, heating systems can be switched off in peak hours without occupants noticing the difference [28] thanks to building inertia or water tanks, which avoid any change in the household’s activities. (iii) Literature identifies fear related to data security as another barrier for acceptance since energy data taken from smart meters (which are crucial for automated DSM) can be misused for multiple purposes [45,46,47].

(iv) These concerns are magnified even more when the control of these appliances that are closely linked to daily life and bodily comfort is to be delegated to third parties by offering to automate these appliances for households. Literature suggests that perception of the loss of control is the most influent element for acceptance [48]. This perception is shaped by the end-users’ desire for ultimate control over devices they have paid for, especially thermostat controls for heating systems [49]. It differs according to personal sensitivities and socio-demographic characteristics; for example, households living in houses may need more control than those living in apartments, which are used to sharing control for heating systems or washing machines [43,50]. In fact, many studies have indicated that end-users are only ready to accept automated DSM if they feel they still retain control over automaton [31,51,52,53].

(v) Literature identifies trust (in both the technology itself and the actors operating for it) as vital for acceptance of automated DSM [25,52], especially since it impacts the perception of the loss of control, of data security, and of the benefits [31,53,54,55]. Trust in the technology related to smart grids (such as HEMS) is influenced by several elements such as household habits or ease of use [56,57], communication of the positive and negative outcomes [56,57,58], and salience or “observability” of pilots to increase social awareness and reduce concerns [58] but also trust towards its operator [57]. Nevertheless, this is critical since “people recognise a misalignment between their own ends (e.g., energy services) and those of energy companies (e.g., profit)” [49]. Trust towards utilities to roll out automated DSM is built progressively through the history of their interactions with end-users [59], which means utilities must be transparent and responsive to end-users’ needs [60]. Moreover, trust requires horizontal relationships with free flow of information and familiarity, but large energy companies are mainly seen by people as inaccessible [61]. Therefore, trust involves matching the social norms and the vision of a desirable future between end-users and utilities [22,61].

Other researchers draw on behavioural economic theories to investigate the barriers and enablers. (vi) Studies focus on the motivations and incentives regarding the adoption of automated DSM programmes. They mainly identify those as monetary savings and the feeling of contributing to environmental issues [33,35,62,63] but also as other elements such as a better understanding of their energy consumption [63]. Behavioural economists investigated the effect of incentives/disincentives on the acceptance of automated DSM and concluded that acceptance increased as long as they were justified [64], minimal [65], and fair [31].

1.3. Aim and Objectives

It is evident that earlier research has focused on different elements separately, such as economic incentives, acceptance of socio-technical objects and people’s perceptions and concerns related to them, household practices, and other topics such as trust, etc., depending on the different disciplinary origins.

In this paper, we consider automated DSM on a project basis, and we argue that there is more to getting end-users as either individuals or collectives on board than the questions of perceived inconveniences, practices, and social acceptability of artefacts when deploying automated DSM programmes. This is mainly because, given its nature, with the automated DSM, the complexity of asking energy users to cede the control of their appliances, which are closely linked to their daily lives, to a third party is amplified even more when the energy users do not rationalise why and for what reason the third parties operate such programmes in their homes, communities, or in the region. Therefore, arguments over social and political trust and legitimacy are likely to occur when actors are attempting to deploy automated DSM projects at the scale of an individual project or activity, the scale of the entire sector with its actors, and the overall concept of automated DSM.

Against this background, we conducted a case study analysis and explored the role that social license to operate plays in automated DSM projects in Switzerland. Our main aim in this paper is to investigate how social license for automated DSM projects has been granted and maintained in Switzerland, in particular, the perceptions of who grants an SLO and how actors work to obtain or maintain trust and legitimacy. For this, we first look at the existing conceptual frameworks of the SLO approach to explore which elements can be used for the automated DSM programmes. Then, we apply the adapted framework to eight Swiss case studies related to deployment of automated DSM programmes for household energy resources, such as EV charging, batteries, heating, or wet appliances (e.g., washing machines and dishwashers). By using the SLO concept as a bridging framework, we draw attention to the (mis)alignments and the negotiations between households with their practices, agency (examining the six main topics mentioned in Section 1.2 if relevant), their perceptions (examining legitimacy, trust), and the expectations of stakeholders as part of the energy system management using demand flexibility as a resource. We believe that this qualitative case study analysis through the social license to operate concept may be useful to energy policymakers and specifically actors that implement automated DSM programmes, such as grid operators and aggregators, for understanding public perceptions and considering them in deploying automated DSM projects and programmes.

2. Social License to Operate (SLO)

The concept of SLO emerged in the late 1990s in the mining industry as an answer to growing community opposition [66]. Since then, it has been widely used and spread among many other industries, including pulp and paper mills, forestry, wind power, or geothermal plants [67,68,69,70,71,72]. SLO is generally defined as “the ongoing acceptance and approval of an operation by those local communities affected by it and those stakeholder who can affect its profitability” [73,74,75,76]. It investigates community support by the observation of the perception of relationship quality between impacted communities and the company/project, which gives a rapid indication as to which elements are important or can be ignored in a specific context [68,77].

Originally a metaphor, it has been theorised and defined by different models, with the most adopted being the pyramidal model [78], which recognizes four factors that define the stakeholders’ perception of legitimacy (economic and socio-political) and trust (interactional and institutionalised) as well as different levels of community support that are cumulative to a certain extent [73]. The pyramidal model is presented in Figure 1 and can be read as follows: on the left different colours are associated to a level of social license, and on the right the social license level (colours) is presented to which the four factors lead, with the boundary of each factor defined by blue and purple lines. With that, Boutilier and Thomson (2011) theorised that when a project represents nothing more than a financial transaction (economic legitimacy), it is more likely to be accepted than rejected but is subject to variation between these two states. When a company is seen as either listening, responsive, and keeping its promise (interactional trust) or as respecting the local way of life and contributing to the well-being of the region (socio-political legitimacy), acceptance is granted, but community support is likely to fluctuate between approval and withdrawal. If the project is seen as benefiting from all the three factors exposed above, the community support granted is mainly likely to fluctuate between approval and acceptance. The highest level of community support (identification to the project) is only reached when all the previous factors are present, and trust towards the company is institutionalised.

Even though a social license is considered as an intangible concept [79], consultants and social-license practitioners take the four factors of trust and legitimacy to measure levels of community support, as exposed above [77,80]. These are sensitised through ethnographic and historical analysis of projects to produce an integrated notion of a project stakeholder’s network and gives importance to stakeholder identification to address appropriately their fear and their expectations and to empower them [73,81,82,83]. Therefore, it can be used to provide guidelines about the ways to appropriately manage relationships as the project progresses [81], going from only maintaining a good reputation to communication and integration in the decision-making process [79,84]. Social license, social acceptance, and corporate social responsibility (CSR) are concepts that are interconnected and hard to differentiate [78,81,82]. Nevertheless, we argue that SLO is a much broader concept with deeper political ties compared to social acceptance although it as a feature of SLO [78,85]. CSR is even a broader concept that can be defined as “a movement to establish responsibility to corporations that goes beyond legal requirement”, while SLO is more defined by “going beyond legal requirement” [78,86].

We argue that using the instrumental aspects of the SLO concept, which are largely concerned with the components of SLO, as an analytic lens to read public support of automated DSM might be useful for understanding how actors achieve and maintain the social support for automated DSM projects. We argue that the four factors of trust and legitimacy established by Boutilier and Thomson (2011) are particularly helpful in providing guidelines, as they merge all the identified barriers related to community support for automated DSM (see Section 1.2).

3. Methods

We conducted research on case studies covering “in-depth examination of particular subjects or phenomena (e.g., individuals, firms, cities, policies, adjustment to a new technology) as well as related contextual conditions, often using multiple sources of evidence” [87]. Literature reviews for each case study, covering reports, newspaper articles, public institution publications, and scientific articles, constituted empirical data for each case study. We did not directly interview participants in the case studies, but instead, we considered interview summaries as part of some case study reports. All the data collected were used to fill an analytical framework containing elements relevant to social license (it had been built in the frame of the SLA sub-task of the IEA and also used in [88] and presented in Appendix A). After the extraction of all information possible from the documentation, missing elements were obtained by presenting the analytical framework to project managers. We asked the project managers to fill in if there are elements missing and also to confirm whether the information filled in by the co-authors is correct. Here, it is important to note that every project manager received the same analytical framework questions for consistency. Eight case studies were investigated, with regard to providing similar and differentiated contexts (such as appliances automation, end-users targeted, and actors operating automation) to allow at the same time generation and strengthening of hypotheses.

In this paper, we focus our investigation on the factors established by Boutilier and Thomson (2011) given the specific issues related to automated DSM (

Table 1. Adaptation of the SLO original concept to automated DSM (social license to automate—SLA).

SLO’s Factors	Definition in the Original Concept (Based on [73])	Adaptation to SLA
Economic legitimacy	Refers to the perception that the project/company is beneficial to the perceiver.	Which benefits and costs have been communicated to the end-users as well as which indirect and direct monetary incentives have been used to convince the end-users.
Socio-political legitimacy	Refers to the perception that the project/company contributes to the well-being of the region, respects local way of life, meets expectations about its role in society, and acts according to stakeholder’s view of fairness.	Which actors control automation, and how does the rationale of the project match with the end-users’ interest and operator’s expected role?
Interactional trust	Refers to the perception that the company and its management listens, responds, keeps promises, engages in mutual dialogue, and exhibits reciprocity in its interactions.	Which means were available for mutual dialogue (for listening, response, and inclusion)? Which information was shared through which channels to ensure promises are kept? How was end-users’ sense of control dealt with (e.g., overriding)?

). We observe how economic legitimacy, socio-political legitimacy, and interactional trust have been built depending on the context in case studies. In the mining sector, social license to operate discussions often centre around spectacular failures of trust. However, our case studies are more modest and do not provide evidence about shifts between different levels of community support (withdrawal, acceptance, approval, identification). Additionally, we do not directly investigate institutionalised trust, as it would require an in-depth assessment of the impact of several years of trust building and field practices between stakeholders and project managers (mainly utilities in the context of Switzerland and DSM [73].

4. Case Studies

The selection of the case studies is limited strictly to those that used automated DSM (at least in part), were conducted within households, and benefitted from empirical results and open science. Different project managers were also consulted about their knowledge of pilot studies related to the criterion established above, who further on provided us a list of contacts in and potential projects related to automated DSM that occurred in Switzerland. We observed that the number of case studies is limited to eight Swiss case studies for the criterion established above. However, case studies are embedded in similar and differentiated contexts that allow us to confront the elements suggested in Section 1.2.

Table 2. Case studies reviewed.

Project Name	Date	Documentation	Appliances Automated	Size/Scale	Core Stakeholders and Location	Actor Leading the Case Studies
Quartierstrom	2017–2020	[89,90]	Community battery (4/5)	37 households; 1 battery sized for 4 household	Virtual self-consumption community (SCC) in a village and including multi-family housing	Small DSO (WEW, supplying fewer than 10,000 consumers), University (ETHZ, Bits to Energy Lab)
WarmUp	2016–2018	[91,92,93]	Shared heat pumps for SH and DHW (5/5)	15 buildings with 22 hot water boilers fed by 9 HPs	Individual consumers living in a big city (Zürich) and in multi-family housing	Big DSO (EWZ, supplying more than 100,000 consumers)
Innovative self-consumption optimization for multi-family area development with local electricity exchange (ISCO-LEE)	2017–2022	[94,95,96]	Shared heat pumps for SH (3/5) and DHW (4/5); individual EVs, washing machines, and dishwashers (1/5).	35 households with 4 HPs, one EV charging station, and 70 combinations of washing machines and dishwashers.	Self-consumption community (SCC) in a village with only households living in multi-family housing	Small DSO (RTB Möriken-Wildegg supplying fewer than 10,000 consumers), University (Fachochschule Nordwestschwiez)
GoFlex	2016–2020	[97,98,99,100]	Individual heat pumps, electric boilers for DHW, and electric heaters for SH (4/5) and individual EVs (3/5, 1/5 de facto).	195 households (with control of HPs, resistive heating, and/or electric boilers); 6 EV charging stations and 9 HEMS	Individual consumers and prosumers living in a medium city (Sion) and in single-family housing	Big DSO (OIKEN, supplying more than 100,000 consumers), University (HES-SO Valais/Wallis)
Decentralised flexibility	2020–2022	[101,102]	Individual heat pumps and electric boilers for DHW (4/5)	45 HPs and electric boilers	Individual consumers living either in a medium city (Fribourg) or in villages and in single/multi-family houses	Big DSO (Group E, supplying more than 100,000 consumers)
Luggagia Innovation Community (LIC)	2018–2022	[103]	Individual electric boilers for DHW (2/5 in the first phase and 3/5 in the second phase) and centralised battery (5/5)	17 single-family households (incl. 3 prosumers); 1 kindergarten with rooftop PV	Self-consumption community (SCC) in a village with households living in single-family housing	Small DSO (AEM, supplying fewer than 10,000 consumers)
Tiko	2014—commercial	[104]	Individual electric boilers for DHW, heat pumps, electric heaters, and night storage heaters (3/5)	6000 devices (50% HPs)	Individual consumers in single-family housing across the country	Private aggregators
OKEE	2019–2021	No publication yet	Rented EVs (3/5)	2 smart charging stations with 2 EVs for sharing with V2G capabilities	EV renters in a big city (Basel)	Private aggregators

presents the case studies investigated and their characteristics, such as project name, date, documentation, and context, including which appliances were automated and at which level (on a 5-point scale) and which were the stakeholders targeted and on which scale.

We distinguished levels of automation on a 5-point scale, considering the nature of automation (semi or full) and the possibility to override automations and to opt-out, inspired from [56]. We define semi automation as automation that can be influenced by end-users (i.e., with temperature or battery-level thresholds or any other factors), full automation as automation that cannot be influenced by end-users, opt-out as the possibility to leave the project and recover the previous settings (i.e., before opt-in), and the possibility to override automation as the end-user’s possibility to choose to activate/deactivate automation (for full automation as well as semi automation). Therefore, we define that:

• A level of automation of 1/5 refers to semi-automation with a possibility to override automation at any time.
• A level of automation of 2/5 refers to full-automation with a possibility to override automation at any time.
• A level of automation of 3/5 refers to either full automation with a restricted possibility to override automation or semi-automation with a restricted or without any possibility to override automation.
• A level of automation of 4/5 refers to full automation without any possibility to override automation but the possibility to opt-out from the project.

A level of automation of 5/5 refers to full automation without any possibility to override automation or even to opt-out from the project.

5. Results

In this section, we present how some sense of social license has been acquired in the case studies reviewed depending on the context in which they were embedded. First, we present how economic legitimacy was built by reviewing which risks and benefits were presented and how. Then, we present how case studies constructed the socio-political legitimacy of automated DSM, observing the rationale presented and the actor responsible for automation. Finally, we present how interactional trust was built by observing the level of control left to the end-users, information communicated to ensure promises are kept, and how end-users were included.

5.1. Economic Legitimacy

5.1.1. Economic Legitimacy—Benefits Presented to End-Users

Case studies show that the benefits end-users will receive are communicated under three categories: (i) monetary savings, (ii) increase in self-consumption and self-sufficiency, and (iii) improving the understanding and transparency of energy consumption. Additionally, the wider benefits of automated DSM to the region are also communicated, namely by (iv) contributing to environmental issues, (v) contributing to an innovative project and energy transition, and (vi) contributing to solving grid problems (such as blackouts, voltage, and frequency issues). Wider benefits are clearly associated with collective benefits and tend to pull end-users to the grid by making it cleaner and safer. Tangible benefits are associated with personal benefits for the end-users and are focused either on pushing them away from the grid (i.e., increased self-sufficiency) or encouraging them to stay (i.e., monetary savings, better understanding).

Table 3. Benefits and incentives presented to the end-users.

Case Study	Context *	Tangible Benefits	Wider Benefits	Incentives
Quartierstrom	(a) Centralised battery (b) SCC (village) (c) Small DSO	Reduce the bill and increase local energy consumed (PV). Better visualization.	Participate in an innovative project. Contribute to environment. Increase the community feeling.	RTP (Indirect). Free installation.
WarmUp	(a) Shared heat pumps (b) Individual consumers (big city) (c) Big DSO	More economical heating.	More ecological heating.	Free installation.
ISCO-LEE	(a) Shared heat pumps (b) SCC (village) (c) Small DSO	Reduce the bill and increase local energy consumed (PV).	Contribute to energy transition and an innovative project.	RTP (Indirect).
GoFlex	(a) Individual heat pumps, electrical boilers, electric heaters, EVs, HEMS (b) Individual consumers, prosumers (medium city) (c) Big DSO	Better visualization (resulting in energy savings and monetary savings).	Participate in an innovative project. Participate in the Swiss Energy Transition.	Free installation.
Decentralised flexibility	(a) Individual heat pumps (b) Individual consumers (village/medium city) (c) Big DSO	None.	Solving grid and network problems.	Free installation and 3 cts/kWh discount for each device automated.
LIC	(a) Individual heat pumps, electrical boilers, centralised battery (b) SCC (village) (c) Small DSO	Increase self-consumption.	Contribute to the energy transition.	Free installation.
Tiko	(a) Individual heat pumps, electrical boilers, electric heaters (b) Individual consumers (all Switzerland) (c) Private aggregators	Monetary savings. Reduce CO2 emissions.	Help to better manage network issues of the TSO.	Free installation.
OKEE	(a) Rented EVs (b) Individual consumers (Big city) (c) Private aggregators	Reduce the bill and increase local energy consumed (PV). Better visualization.	Participate in an innovative project. Contribute to environment. Increase the community feeling.	RTP (Indirect). Free installation.

* Appliances/devices automated (a), targeted end-users (b), automation’s operator (c).

shows the different identified patterns of benefits presented in the case studies as well as the context of each project with the appliances automated, the targeted end-users, and the actor responsible for automation, respectively.

When end-users were part of a self-consumption community (SCC) (e.g., Quartierstrom, LIC, ISCO-LEE), increasing self-consumption was the main benefit presented to end-users, whereas for individual prosumers, this argument was not seen as convincing enough if proposed by the DSO [99]. Contributions to the environment or the Swiss Energy Transition were communicated in all case studies reviewed (except Decentralised Flexibility, which presented instead its contribution to help to solve grid issues). End-users seemed mainly satisfied by it being presented as a broad statement, as the GoFlex survey illustrates [100].

Projects led by private actors (Tiko and OKEE) are the only ones that presented environmental benefits as tangible (i.e., avoiding CO₂ emissions). Monetary savings are presented as one of the main benefits in all projects except two (Decentralised Flexibility and GoFlex), but none presented a target such as the savings expected. Decentralised Flexibility proposed monetary incentives instead and is the only project that did so. GoFlex, given that project managers feared a decrease of goodwill and negligible savings, presented energy and monetary savings as an outcome of a better understanding of energy consumption [99]. Incentives such as free installation are widely used, and dynamic pricing was presented in some SCCs as an incentive per se. Only GoFlex and Quartierstrom presented an improved visualization of energy consumption as a main benefit, which might be explained by the fact that it is the only tangible outcome for end-users and another way to increase self-consumption, respectively.

Building community, in the sense of enhancing social bonds between residents, was only emphasised in one SCC (Quartierstrom). Being part of an innovative project was presented in the project that occurred in a rural context or medium towns but not for end-users living in larger towns. Mainly, technological innovations were associated with different aspects, such as a local energy market (in Quartierstrom), smart meters (in GoFlex), or RTP (in ISCO-LEE).

5.1.2. Economic Legitimacy—Risks and Costs Communicated to End-Users

Case studies that automated HPs as electric devices for both SH or DHW presented discomfort as something that would never or be unlikely to occur and promised that comfort had priority over other objectives (e.g., cost savings). Data security was not observed as a high concern for the participants, as also observed for Swiss households in the literature [53]. In the case studies, it was addressed mainly by a consent form explaining how data will be managed safely (i.e., with a blockchain, by storing on local servers, ensuring that user data will not be used for a commercial purpose or shared with third parties). It seemed sufficient as illustrated in GoFlex, where most end-users disagreed to associate the project with an issue such as invading privacy [100]. Similarly, in the other case studies, no further complaint emerged related to data security.

5.2. Socio-Political Legitimacy and Interactional Trust

Table 4. Benefits and incentives presented to the end-users.

Case Study	Context *	Rationale Presented	Remaining Control	Information Channels (Information Shared)
Quartierstrom	(a) Centralised battery (b) SCC (village) (c) Small DSO	Decrease imports from the grid and exports of PV produced in the boundary of the microgrid.	Full automation. No overriding options. Opt-out possible.	Web interface, monthly mail (to inform about energy costs and consumption and PV production and self-consumption at the household and the community scale). Detailed billing (to inform about monetary savings). Information sessions (to introduce project and stakeholders). Phone line (to inform about to the interface in the beginning, then to express complaints). Interviews (to express future expectations, collect perceptions, and know if households did manual DSM).
WarmUp	(a) Shared heat pumps (b) Individual consumers (big city) (c) Big DSO	Increase efficiency; decrease CO2 emissions and costs.	Full automation. No overriding options. Opt-out impossible.	Web interface (to inform about energy and power costs). Billing (to inform about monetary savings). Phone line (for any complaints). No interviews.
ISCO-LEE	(a) Shared heat pumps (b) SCC (village) (c) Small DSO	Increase self-consumption of PV in the eco-district.	HPs: Semi automation (choice of temperature thresholds). No overriding options. Opt-out possible. EVs: Semi automation (choice of departure time and journey planned). Possibility of overriding automation at any-time. Opt-out possible. Wet appliances: Semi automation (schedule). Possibility of overriding automation at any-time. Opt-out possible.	In-home interface (to inform about temperature, automation state, and electricity mix). Web interface and smartphone interface (to inform about market prices, temperature, energy consumption, PV production, and self-consumption/self-sufficiency at the device, the household, and the community scale and self-consumption/self-sufficiency at the household and the community scale). Detailed billing (to inform about monetary savings). Phone line (for any complaints). Surveys (to collect satisfaction related to comfort and prices and know if households used automation or did manual DSM).
GoFlex	(a) Individual heat pumps, electrical boilers, electric heaters, EVs, and HEMS (b) Individual consumers, prosumers, (medium city) (c) Big DSO	Better management of the grid (but a bit unclear for end-users); increase self-consumption (for prosumers).	Heating systems: Full automation. No overriding options. Opt-out possible. EVs: Semi automation (choice of departure time and journey planned, and schedule). Possibility of overriding automation at any-time (de facto, by not using the app where semi automation is made). Opt-out possible.	Smartphone interface for EVs (to inform about flexibility activation). Web interface for heating systems (to inform about flexibility activation, temperature, and energy consumption at the device scale). Billing (to inform about monetary savings). Phone line (for any complaint). Interviews (to identify preference related to benefits/rationale before the project and ask about perception related to discomfort and control as well as fear).
Decentralised Flexibility	(a) Individual heat pumps (b) Individual consumers (village/medium city) (c) Big DSO	Grid stabilization	Full automation. No overriding options. Opt-out possible.	Web interface (to inform about energy consumption at the device scale). Billing (to inform about monetary savings). Phone line (for any complaint). Interviews (to identify preference related to benefits/rationale before the project).
LIC	(a) Individual heat pumps, electrical boilers, centralised battery (b) SCC (village) (c) Small DSO	Increase self-consumption of PV and self-sufficiency of the community.	Battery: Full automation. No overriding options. Opt-out possible. Heating systems: Full automation. Possibility to override automation at any-time (phase 1)/restricted possibility to overriding automation (phase 2, confirmation needed from the DSO). No overriding options. Opt-out possible.	Web interface and biannual mail (to inform about energy consumption, PV production, and flexibility activation for the battery). Billing (to inform about monetary savings). Phone line (for any complaint). Interviews planned.
Tiko	(a) Individual heat pumps, electrical boilers, electric heaters (b) Individual consumers (all Switzerland) (c) Private aggregators	Helping the TSO by providing ancillary services	Full automation. Restricted possibility to override automation (for a day). Opt-out possible.	Web interface and smartphone interface (to inform about temperature, CO2 avoided, and energy consumption at the device scale). Billing (to inform about monetary savings). Phone line (for any complaint). No interviews.
OKEE	(a) Rented EVs (b) Individual consumers (Big city) (c) Private aggregators	Help renewables integration and reduce power peak.	Semi automation (choice of departure time and journey planned). No overriding options. Opt-out possible.	Web interface (to inform about solar consumption). Billing (to inform about monetary savings). Phone line (for any complaint).

* Appliances/devices automated (a), targeted end-users (b), automation’s operator (c).

indicates which rationales were presented, which information was communicated through which channels, and how the issue of control was dealt with. The rationales presented are linked to socio-political legitimacy and remaining control, while the information communicated and channels used are linked to interactional trust.

5.2.1. Socio Political Legitimacy—Legitimacy of the Rationale and the Operator

Case studies show that rationales presented to the end-users are communicated under four categories: (i) increase in self-consumption and self-sufficiency, (ii) solving grid issues, (iii) solving environmental issues, and (iv) increase in monetary savings.

Rationales such as “to maximise self-consumption and self-sufficiency” were used in each case study that targeted SCCs (i.e., LIC, Quartierstrom, and ISCO-LEE). Additionally, operating agents for automation were always small DSOs (supplying fewer than 10,000 consumers).

When case studies targeted individual end-users, solving grid issues was presented as the/one of the rationales of the project (except for WarmUp). While DSOs presented it directly (i.e., Decentralised Flexibility, GoFlex), Tiko, which is a private aggregator, presented it as a service sold to the TSO for grid balancing. Differences in communication might be explained by the fact that solving grid issues is the expected role of the DSO but not of a private aggregator. Some case studies operating with individual consumers presented a similar rationale, but additionally, they communicated the integration of renewables, especially PV (OKEE, GoFlex).

WarmUp emphasised the rationale for automation as providing ecological and economic heating. Nevertheless, investigation showed that the distinction between benefits and rationale was sometimes not clear for end-users, which led to misunderstandings. It is particularly evident in GoFlex, where the interviews revealed that most end-users were thinking that the purpose of the project was to inform about energy use and to help use less energy with better visualization [100].

5.2.2. Interactional Trust—Management of End-Users’ Sense of Control

Analysis of the case studies shows that the ways to deal with end-users’ sense of control is highly dependent on the type of device automated. Solutions offering to increase the sense of control can be summarised under three categories: (i) opt-out, (ii) overriding options or semi-automation (including schedules, pre-request, temperature thresholds, etc.), and (iii) informing about flexibility activation (i.e., when the third party performs the automation such as turn on/off the targeted device).

In all case studies except WarmUp (i.e., too complex to provide individual opt-out from HPs, as the heating system was centralised), end-users had the possibility to opt out from automation to recover previous settings. Community batteries (CB), since they were not owned by end-users, were fully automated without overriding options.

For heating systems and electrical boilers, case studies that were led by private aggregators or that targeted SCCs all proposed either overriding options to the end-users (LIC, which were unlimited in the first phase, and Tiko, which were limited to veto day) or semi-automation with thresholds of temperature (ISCO-LEE and Tiko). In LIC, when end-users had the possibility to override automation at any time, it resulted in almost no flexibility activation. In ISCO-LEE, complaints related to discomfort still occurred, partly since some end-users did not see the effect of their actions but also partly due to misinformation. On the contrary, in WarmUp, GoFlex, and Decentralised Flexibility, no remaining control was left to the end-users (full-automation and no overriding options). In GoFlex, end-users were asked directly about their perception of control, while in every project, end-users were able to complain in the phone line, but it did not provide any evidence that end-users perceived more loss of control when they were proposed low remaining control solutions (full automation, no overriding options). This suggests that fear of loss of control is lower when a public utility operates automation, and end-users are not part of a SCC. Moreover, GoFlex informed about flexibility activation and communicated, as a central aspect of the project, a better understanding of energy consumption with visualization. This strategy was appropriate to manage loss of control, as most end-users felt that the project increased their control over their electrical consumption [100].

For EVs, semi-automation with the possibility to set departure time and journey planned was present in all case studies. The only case study that automated rented EVs (OKEE) did not provide overriding options for end-users since it was a car-sharing activity. ISCO-LEE proposed to schedule journeys in a calendar and to override automation at any time activation, but only two EVs were operating in the project, and therefore, it is difficult to reach certain conclusions. GoFlex did not directly propose overriding options, but by not using the smartphone interface linked to the charging station (where end-users must indicate type of EV, departure time, etc.), end-users were able to bypass the automated charging of their EVs. This led to almost no flexibility activation (only 1% of the charging sessions).

ISCO-LEE implemented the lowest level of control (semi-automation with the possibility to override automation at any time) for dishwashers and washing machines. These were activated automatically when there was an overproduction of PV but should be manually loaded by the end-users and set with a pre-request for overriding automation. Additionally, the end-users could indicate what time the washing machines and dishwashers should be finished by, and these appliances were never interrupted when they were on. With that proposal, 45% of the end-users made the choice to activate automation for dishwashers and 27% for washing machines [95].

5.2.3. Interaction Trust—Information Communicated and Interface Provided

Each case study that targeted SCCs (i.e., Quartierstrom, LIC, and ISCO-LEE) informed about self-consumption, production, and consumption (through load curves) of the community as a whole as well as the concerned individual households of the community. In each case study, end-users were able to know their actual savings through billing. However, few case studies (Quartierstrom and ISCO-LEE) provided detailed billings to simplify the understanding. Some case studies informed about monetary savings but also about current costs through an interface (i.e., Quartierstom, ISCO-LEE, and WarmUp). In the project that communicated it, actual savings remained mainly limited. In OKEE, the cost for energy usage decreased by 2.5%, and in ISCO-LEE, these decreased about 8%, while in WarmUp, energy costs (here, it is matter of block of energy traded) decreased on average about 2.7% even if power peak costs decreased on average about 10% for the DSO. End-users seemed mainly satisfied with monetary savings since almost no complaints were registered. However, one-third of ISCO-LEE’s end-users claimed that their energy costs did not decrease [95]. Some case studies that included individual consumers presented load curves (which also inform about flexibility activation) through an interface (i.e., GoFlex, Tiko, or OKEE), while others did not (WarmUp, Decentralised Flexibility). For GoFlex, better visualization was one of the main aspects of the project, while for Tiko and OKEE, the fact that they were led by private aggregators might explain why a higher level of information with load curves was provided.

For community batteries (CB), LIC informed about flexibility activation, while Quartierstrom did not. Case studies led by private aggregators are the only ones that informed end-users about effective contribution to environment issues through CO₂ avoided (Tiko) or share of local PV consumed (OKEE). When a DSO operated automation, effective contribution to the environment was mainly not seen as necessary information to be communicated. This was explicitly stated in Go Flex surveys, where this information requirement was only in seventh place (out of ten) [100]. It is also suggested in WarmUp since, although environmental consumption and ecological consumption have been presented as benefits on an equal footing, end-users were only informed about money saved.

Discomfort related to temperature dropdowns is indeed subjective, and it was presented in many different ways in the case studies. For example, Decentralised Flexibility indicated “the automation of heat pump will not impact your comfort”; GoFlex, “the automation will guarantee maximum comfort”; and WarmUp, “we will improve your comfort” without any precise indications or metrics. To prove that discomfort did not occur, ISCO-LEE, GoFlex, and Tiko directly informed about discomfort through temperature in an interface. GoFlex and ISCO-LEE also asked end-users through interviews or surveys about perception of discomfort. They observed that most end-users did not experience discomfort such as temperature drop down but also that complaints related to discomfort blamed automation even when discomfort could be explained by other elements such as early phase of regulation, bad habits (e.g., not using blinds), or bad location (e.g., flat located under the roof). This highlights that not every end-user understood discomfort caused by automated DSM in the same way. Tiko also provided an alarm system informing when temperature was below threshold points according to the end-user’s choice mode. WarmUp, Decentralised Flexibility, and LIC did not provide any information related to discomfort such as temperature.

5.2.4. Interactional Trust—End-User Reciprocity and Inclusion

The inclusion of end-users goes from listening to end-users’ expectations and empowering and encouraging them to act towards pursuing DSM objectives. Several case studies provided an interface that informed about load curves (e.g., GoFlex, Quartierstrom, and ISCO-LEE) or current state of electricity mix (ISCO-LEE) and varying tariffs (ISCO-LEE and Quartierstrom) to encourage manual load shifting. In ISCO-LEE, more than 90% of the end-users claimed to do manual shifting [95]. Quartierstrom was particularly inclusive, as manual DSM was central, and end-users even had the right to influence market price. In that configuration, many end-users actively doing manual DSM also talked about the project in a positive way to their friends [89].

Only two case studies asked end-users their expectations before starting. They did it to frame the most appropriate proposal related to the benefits and the rationale presented (GoFlex and Decentralised Flexibility) or related to incentives (Decentralised Flexibility). Quartierstrom set up a session to inform future end-users about the project and to answer their questions, which is one explanation for the high opt-in rate (90%) [89]. For automation of heating systems for SH and DHW, each case study provided the possibility to complain to the automation operator. It was crucial to evaluate discomfort and adjust regulation. Each case study that targeted self-consumption communities (ISCO-LEE, LIC, and Quartierstrom) went or planned to go further in listening to the end-users by interviewing them. This helped to understand fears and what end-users would like to see implemented or improved in the next phases. The case studies that targeted individual consumers mainly did not consider this option (only Decentralised Flexibility and GoFlex out of five). The latter includes all the case studies led by private aggregators.

6. Discussion

This paper provides insights from case studies in Switzerland regarding the way to build a social license in the field of automated DSM. In this section, we present first a summary of these findings, followed by the implications for policymakers and limitations as well as future work that can be associated with this paper.

6.1. Findings and Implications

The analyses of the case studies show that building the economic legitimacy of a project goes through the end-users’ perception that their personal costs will be compensated by benefits whether they are personal or for the community. Review of the case studies showed that risks communicated were mainly focused on discomfort and as an element unlikely to occur, not impacting daily life, and that would be minimised. Focus on discomfort is not surprising since it is recognised as one of the main concerns for Swiss households [31,34]. The topic of data security was managed only by providing a consent form, which seemed enough for end-users. This suggests data security is a minor issue for Swiss households, as already suggested in the literature [53], and highlights a high level of trust in Swiss energy and wider institutions. The literature established that monetary savings is the first motivation to accept automated DSM [33,35,52,63] although literature investigating the questions in the minds of Swiss households observed that it is only one important element among others [34] but also that households can associate automated DSM to monetary losses [53]. Additionally, monetary savings, understood as a pure personal benefits, is an element on which importance should not be put on since it might abate end-users feeling that they are doing something positive by opting-in [105]. This last evidence agrees with the findings in the case studies in which monetary savings were presented in different contexts but not as a precise amount of savings. Indeed, end-users were communicated the potential monetary savings through various measures, including incentives or revenue from flexibility activation. The literature recognizes different sensitivity to benefits [35], which was also observed in the case studies. When targeting SCCs, increasing self-consumption was presented as the main benefit. When targeting individual consumers, contributions to solving grid issues and/or environmental issues were presented as the main benefits. When targeting end-users living in low-density areas rather than big cities, case studies emphasised benefits such as being part of an innovative project. Visualisation was emphasised either when it was the only tangible outcome or when increased self-consumption was central (i.e., for SCCs). Broman Toft et al. (2014) observed that both collective and personal benefits influence acceptance of automated DSM. Analysis of the case studies suggests first that, besides collective benefits (i.e., contribution to environment), individual consumers must feel they will receive personal and tangible benefits (i.e., monetary savings, better visualisation). The necessity of provide tangible benefits was also raised in [56,58], especially to build trust. Second, findings suggest that SCCs are mainly interested in benefits allowing them to be more independent from the grid, while individual consumers are mainly interested in using the grid in a more sustainable way (i.e., benefit from cheaper and cleaner electricity).

The analysis revealed that socio-political legitimacy is built by presenting a rationale for automation that must match end-users’ expectations concerning both the role of the technology and the expected role of the actor operating it. For case studies targeting individual consumers, the main rationales presented were solving grid issues, which was often presented with integration of renewables. Surveys that investigated the acceptance of automated DSM in Swiss households mainly considered these two but rather with a focus on environmental issues [53,105,106]. When targeting SCCs, the rationale presented was always that of increasing self-consumption. This highlights the fact that SCC expectations are to increase their independency from the grid. Misunderstanding of the rationale occurred in some case studies, which is a concern also raised in the literature [107], and it might lead to a withdrawal and have a long-term effect on the social license if outcomes of automation do not lead to the rationale conceived by end-users. DSOs presented rationales such as solving the grid or environmental issues. Since few concerns emerged, we can conclude that end-users perceived those rationales as matching the roles expected from their DSO. Private aggregators only proposed rationales such as trading flexibility (suggesting they are only trusted for this), and only small DSOs proposed to increase self-consumption. However, since increasing self-consumption competes with one of the main expected roles of the DSO (selling energy), as pointed out in [49,60], it might suggest a particularly high degree of trust in DSOs, especially when compared to private aggregators. In fact, small DSOs are small entities, are well-established (thus having a certain history of relationships with end-users), and are public actors, which are all essential characteristics to build trust, while private aggregators are private actors and are emergent [34,59,61]. Building interactional trust requires ensuring that promises are kept, maintaining the end-users’ perception of inclusion and being in control. However, efforts needed for this are highly dependent on the context.

The case studies relating to automated devices for which control was already shared (i.e., centralised batteries or centralised HPs) are the only ones that proposed full automation with no opt-out possibility. This might suggest that end-users used to sharing control over their appliances have a lower aversion to losing control, as highlighted in [43]. For devices providing SH or DHW, semi-automation with overriding options was proposed when the case study targeted SCCs or when a private aggregator was operating automation. Conversely, in the case studies including individual consumers and DSOs, informing about flexibility activation (with full automation) was enough to manage the end-users’ perception of control and even more if combined with a communication of a higher control in another aspect. This supports that experience of control can be dealt with by only maintaining optimism and by diversion, and the statement exposed above could be summarised by this: “In conditions in which no objective control exists, a person’s conviction that control is available is enough…” [48]. When case studies proposed to automate EVs and wet appliances, such as dishwashers and washing machines, only semi-automation with deliberate activation of automation was proposed. These observations suggest first that there is a hierarchy of aversion to losing control depending on devices/appliances automated, with wet appliances and EVs at the top and battery and heating systems at the bottom. This supports what was observed in [43]. Secondly, private aggregators are less likely to be trusted, and SCCs need more control than individual consumers.

No case studies provided information to show the contribution of automation to the grid issues. Only private aggregators provided information to prove the environmental contributions, which might suggest that DSOs do need to prove environmental contributions because, first. It is a role that they are expected to meet and, second, because as already suggested above, they are more likely to be trusted compared to private aggregators since ensuring promises is linked to trust, as exposed in Section 2. Many case studies informed about the temperature, but it did not prevent end-users from blaming automation for experiencing discomfort regardless of whether this really occurred or if automation was the effective cause of it. Moreover, ISCO-LEE showed that end-users might be disappointed about monetary savings. Those last observations indicate the need to provide transparency and better inform the end-users to increase consumer support, as stated in [60]. Quartierstrom, before starting, provided an information session for potential end-users to introduce the project as well as the stakeholders’ part of it, which could be one explanation for a particularly high opt-in rate and is actually a recommendation for implementing a microgrid [108]. We argue that an information session before even starting the project might be a capital mechanism not only to build trust towards the automation operator but also to manage end-users’ expectations by improving transparency.

Encouragements to pursue manual DSM were present in many case studies (especially when those targeted SCCs), but mainly, SCCs were considered for interviews and asked about elements they would like to include in the project. This might be an additional element that suggests that SCCs put more importance into being in control and, more widely, into independency compared to individual consumers.

6.2. Limitations and Future Work

Automated DSM is quite emergent in Switzerland. However, the number of surveys and pilot projects is still limited, and participants are primarily highly enthusiastic and technically literate early adopters of technology. The data analysed in this study were primarily secondary data sourced with the assistance of project proponents. This introduces potential biases to our findings. Furthermore, in the case studies, the level of acceptance reached was not measured, at least in the sense of Boutilier and Thomson (2011). Most field studies reviewed considered acceptance elements such as perception of benefits or influence of control options as secondary elements in their investigation.

As further improvements, we can suggest first to take part in the case studies at the earlier stages in order to be able to define metrics easily comparable among case studies (e.g., percentage of people asked that were ready to opt-in or monetary savings for households). This includes introducing surveys and/or interviews that investigate end-users’ perceptions to remove a bias layer (project managers interpretation). This will additionally help to provide equal data access and quality. Finally, another alternative approach could be to include a greater number of cases worldwide that carefully consider the above-mentioned aspects and measured with similar metrics so as to perform a more comparative analysis to provide stronger evidence. Furthermore, we argue that our findings could be strengthened by building a case study centred on the social license concept, which incorporates measures (at different stage of the project) of control, discomfort, trust, benefit perception, and identification with the goals of the project.

7. Conclusions

This study provides an empirical investigation of the emerging concept of the social license to automate (SLA), which is an application of the social license to operate (SLO) in the field of smart grids and demand-side management (DSM). The concept of an SLO has become increasingly important to creating new supply resources for energy. This study has shown the value of the concept, and SLO literature has highlighted that the identification of stakeholders as well as economic legitimacy, socio-political legitimacy, and interactional trust are key elements in building community support. Indeed, these points bridge several elements traditionally considered separately in the literature that focused on acceptance of automated DSM.

We investigated empirical data from eight case studies that trialled automated DSM in Swiss households. Findings showed that recommendations for building consumer support are highly context-dependent, which is defined by which appliances are automated, which actor is operating the automation, and which end-users are targeted. The context has a considerable influence on the appropriate way to communicate the risks, the benefits, and the rationale but also on the appropriate way to manage the end-users’ sense of control and the information communicated. Considering these elements, our findings suggest first that individual consumers’ understanding of the rationale and hence the benefits of automated DSM will contribute to a safer, cleaner, and cheaper grid as well as that benefiting them is a key factor to engage them in automated DSM projects. For self-consumption communities (SCC), on the other hand, it is a fine balance of accepting that automation will increase their independence despite more control from third parties. Second, our findings suggest that devices producing space heating and domestic hot water as well as those where ownership is shared, distribution system operators (especially when those are small) and low-density areas together form a context in which efforts to acquire a social license are minimised. Therefore, we argue that this is the configuration that policies should concentrate on to build acceptance of automated DSM. We argue that our findings demonstrate the relevance of investigating consumer support of automated DSM through the lens of the social license. We think that this work can help to avoid withdrawal and build acceptance for new and current projects of automated DSM in the residential sector.