Exploring the performance and ethicality of smart charging systems

Abstract

Increased Feed-in of renewable energy sources and an increase of sales of electric vehicles (EVs) complicate balancing of demand and supply. EVs are considered core enablers for dealing with intermittency due to their storage potential. This does however require that these EVs are charged according to smart charging protocols, which is a vehicle-to-grid (V2G) technology. Options exist for designing such a system. However, due to the gathering of personal (transactional) data, and the involvement of monetary assets, ethical concerns such as privacy and trust issues are assumed to arise at or after the introduction of such a technology. This has been the case with for instance smart meters. The effects of such ethical concerns on the usage of smart charging systems are unknown.

The main research question for this thesis is: “How can a smart charging system be designed which is both used on the short- and long-term and fulfils ethical values of EV owners?”

The purpose of this thesis was to explore and assess different smart EV charging system designs concerning factors contributing to system performance and possible ethical concerns. The research was conducted by creating an integrated framework of both the capability approach and complex adaptive systems. Agent-based modeling was used as the main research method in order to model the behaviour of EV owners within an smart charging environment. The model aims at providing valuable insights concerning which system design performs best with respect to system performance and ethical value fulfilment. Several architectural design decisions are elaborated on with respect to a decentralised or a centralised system. The research outcomes indicate possible short- and long-term ethical concerns of users with respect to the designed system. The effects of these concerns are at this point unknown, but are considered to have an ongoing effect on the performance of the system.

The research objectives involve the identification of four key architectural design decisions which consist of both decentralised and centralised alternatives. The conceptual framework built upon the capability approach, the unified theory of acceptance and use of technology (UTAUT), and complex adaptive systems is used as conceptual basis for the agent-based model. The experimental design revolves around comparing three experimental design alternatives: a (1) public centralised system, (2) public decentralised system, and (3) a private decentralised system. The public centralised system describes a system which is controlled by a single authority, and in which data is stored within an external database. Furthermore, participants are free to participate. The public decentralised system describes a system which is not controlled by a single authority, and in which transactions are validated through shared consensus. Within the private decentralised system, power is exerted towards a single facilitator, which is authorised to whitelist participants. Aside from whitelisting privileges, the platform is not controlled by a single authority. The combined theoretical framework induces a wide variety of uncertain parameters. Therefore, the experimental designs are explored through an EMA study. The main purpose of this EMA study was to assess the three designs with respect to system usage, system performance, and ethical value fulfilment across a wide range of possible scenarios.

The model outcomes provide knowledge regarding promising design directions as well as directions for further research. The developed model describes the interactions between EV owners and the smart charging platform. The performance of the system is based on three constructs originating from UTAUT. These constructs are: performance-expectancy, effort-expectancy, and social influence. The total performance of the system combined with the personality traits of the EV owners determine whether an EV owner uses the system. Effort-expectancy describes the degree of effort needed to participate on the platform. Social influence is rooted within the capability approach and incorporates a combination of effects resulting from direct and indirect interaction between EV owners. Electricity is traded through the use of the platform which creates demand and supply. For model experimentation the KPIs for assessment are: the number of system users, the number of transactions, the number of traded kilometers, and the selected ethical values.

Regarding ethical values, the experimental outcomes provide strong indications that a decentralised system scores best on privacy, security, anonymity, and confidentiality. The scores indicate that problems regarding these ethical values are less expected on the short- and long-term as compared to a centralised system. When aiming to design a decentralised EV charging platform, special focus should be placed on achieving trust, as indications are present that trust issues could arise for decentralised systems. Concerning the KPIs related to system usage and performance, the results indicate that a centralised system is highly preferred. For each of the KPIs (number of users, enabled users, number of transactions, and number of traded kilometers), a centralised system scores higher. However, due to the large number of users, centralised systems have high oscillations in demand and supply.

Concerning the main research question, the research outcomes indicate that when designing a smart EV charging system, one should consider both decentralised and centralised design elements. Both elements are needed as solely focusing on either centralised or decentralised systems has implications for value fulfilment as well as system performance. The optimal combination of design elements is at this point uncertain. The research outcomes clearly indicate that extended research in this field is justified. It is important to further identify which system components in both centralised and decentralised systems positively contribute to the chosen KPIs. These components can then be combined in order to work towards an optimal system design. The operationalisation of the capability approach opens new possibilities to assess these combined design alternatives with respect to system usage and ethical values. In that sense, this thesis provides a tool for assessing the designs of smart charging platforms by providing an integrated framework of the capability approach and agent-based modeling.

The research described in this thesis has several implications for designing smart EV charging systems. The capability approach was found to be a proper method for assessing ethical values concerned with the usage of technology. Furthermore, it was found useful to extent the capability approach with other theories and methods such as the unified theory of acceptance and use of technology, complex adaptive systems, and agent-based modeling. Using this approach for assessing smart EV charging systems helps to keep focus on the actual well-being of system users, rather than solely focusing on technological performance. Furthermore, the approach is long-term oriented. Directions for further research are presented. Essentially, these directions stimulate the further exploration of feasible smart charging system designs by extending the scope of design alternatives, further explore the integrated approach of CA and CAS, and further quantify the agent-based model.