Decarbonization through Active Participation of the Demand Side in Relatively Isolated Power Systems

Abstract

In the context of power system decarbonization, the demand-side strategy for increasing the share of renewable energy is studied for two constrained energy systems. This strategy, which is currently widely suggested in policies on the energy transition, would impact consumer behavior. Despite the importance of studying the latter, the focus here is on decisions regarding the type, location, and timeframe of implementing the related measures. As such, solutions must be assessed in terms of cost and feasibility, technological learning, and by considering geographical and environmental constraints. Based on techno-economic optimization, in this paper we analyze the evolution of the power system and elaborate plausible long-term trajectories in the energy systems of two European islands. The case studies, Procida in Italy and Hinnøya in Norway, are both electrically connected to the mainland by submarine cables and present issues in their power systems, which are here understood as relatively isolated power systems. Renewable energy integration is encouraged by legislative measures in Italy. Although not modeled here, they serve as a backbone for the assumptions of increasing these investments. For Procida, rooftop photovoltaics (PV) coupled with energy storage are integrated in the residential, public, and tertiary sectors. A price-based strategy is also applied reflecting the Italian electricity tariff structure. At a certain price difference between peak and off-peak, the electricity supply mix changes, favoring storage technologies and hence decreasing imports by up to 10% during peak times in the year 2050. In Norway, renewable energy resources are abundant. The analysis for Hinnøya showcases possible cross-sectoral flexibilities through electrification, leading to decarbonization. By fine-tuning electric vehicle charging tactics and leveraging Norway’s electricity pricing model, excess electricity demand peaks can be averted. The conclusions of this double-prospective study provide a comparative analysis that presents the lessons learnt and makes replicability recommendations for other territories.

1. Introduction

1.1. Background of the Study

Many countries are counting on the massive development of renewable energy to meet their commitments to the CO₂ emission reduction targets in the framework of the Paris Agreement and more broadly in the fight against global warming. As part of its overall energy policy, particularly the European Green Deal, the European Union is aiming, for example, for a minimum of 42.5% renewable energy participation in its energy mix by 2030. This target has been chosen to reach the goal of reducing greenhouse gas emissions by 55% in 2030, with respect to 1990 levels [1]. Among these renewable energies, solar and wind power are leading figures, with an increase of 309 GW and 389 GW in their respective capacities planned between 2015 and 2050 [2]. However, the integration of producing electrical energy using these variable renewable sources involves rethinking the electrical system to ensure its stability and reliability. Grid stability is crucial for power system operation, with a balance needed between production and consumption. The variation between these two factors impacts the system’s inertia, which affects its ability to maintain voltage and frequency stability after a disturbance. Lower inertia leads to faster frequency changes and challenges concerning system reliability [3].

Thus, to adapt to the increasing variability of electricity production from renewable energy, that is, to guarantee supply and take advantage of the energy produced at all times, the electricity system must be more flexible. This encourages setting up strategies that necessitate the participation of the demand side to ensure robust, decarbonized energy systems.

Many European islands possess abundant renewable energy resources, such as wind, solar, geothermal, and wave power, offering them opportunities to transition toward self-sufficient, sustainable energy generation [4]. The favorable regulatory framework of the Clean Energy for all Europeans Package (CEP) [5] indicates that the European Commission is dedicating a Clean Energy for EU Islands Initiative that “provides a long-term framework to help islands generate their own sustainable, low-cost energy” [6]. Studies related to the energy or electricity systems of islands have until now focused on renewable energy integration for energy autonomy, such as the cases of Reunion Island [7] and Madeira Island [8]. Some have also focused on the reduction of greenhouse gas emissions through the inclusion of climate mitigation targets, such as Nationally Determined Contributions (NDC). The International Renewable Energy Agency (IRENA) dedicates specific studies to the energy system of Small Island Developing States (SIDS) [9]. One example is the study of Antigua and Barbuda, where pathways for renewable energy integration, electric vehicles, and green hydrogen were elaborated [10]. Importance is assigned to assessing the potential of RE on islands in [11] and analyzing the energy system through indicators related to energy resources, energy profiles, and energy management in [4]. Through technical scenarios and future planning, the authors of [12] were able to determine the alignment between national and successful policies, while focusing on five critical policy areas. As can be seen, some islands aim at achieving complete energy autonomy, while other goals include reducing emissions by having low percentages of diesel in their future energy supply mix. In other words, they seek diversification of their energy resources through harnessing the full potential on their territories. In some cases, interest is directed to policy evaluation. The literature highlights the benefits and importance of involving the demand side yet fails to provide guidance on its long-term facilitation in the energy systems of islands, especially considering their restricted interconnections with the mainland.

1.2. Motivation and Contribution

In this study, the subjects delve deeper into decarbonization and renewable energy integration in insular energy systems, focusing on strategies that would enable flexibility. First, it relies on the applied methods that would be employed in the energy systems. Hence, this research isolates and evaluates the impacts of these methods within the energy systems themselves. It provides a unique lens on the feasibility of integrating renewable energy and flexibility into the system without altering demand.

Second, the study highlights the need to assess the energy transition and action plans implemented by islands, which are often not adequately addressed in national policies. For this reason, islands are taken as case studies: Procida in Italy, and Hinnøya in Norway. As a novelty, the paper emphasizes island-specific energy transitions and underlines the importance of localized energy solutions that cater to the unique needs and potentials of island communities.

Third, utilizing the TIMES model generator from the IEA Energy Technology Systems Analysis Program (ETSAP), the study models the future evolution of power systems for these islands, aiming to accommodate higher shares of renewable energy for their decarbonization. This long-term, high-resolution modeling provides a detailed framework for understanding how renewable energy and flexibility solutions can be optimally integrated over extended timeframes, considering local resources and data [13,14], and demand participation strategies. Furthermore, it provides an assessment of different scenarios or hypotheses that are crucial for making decisions now on investments for the sustainability of decarbonization pathways.

Fourth, owing to the time structure of the two energy systems, the load curve variations are obtained across different timescales (seasons, days, and hours of the day). This method is particular to this long-term study and provides detailed analysis of the optimal use of the potential of renewable energy. It enhances the understanding of how flexibility solutions and strategies can be tailored to account for the grids’ needs (peak demands, imports, etc.) in different timeslots.

Fifth, the decarbonization of end-use sectors is also part of the study. The analysis for each island offers a comprehensive view of how different elements can synergistically contribute to decarbonization.

Sixth, the selected case studies are representative of two contrasting EU island territories. Thus, this study offers broader policy recommendations and insights for replicability in other territories. Indeed, it includes a comparative summary in the discussion (Section 5) on implementation challenges, success factors, and cross-regional lessons. It aims to enhance the generalizability and applicability of its findings.

1.3. The Different Roles and Strategies of Active Demand Participation for Increasing Renewable Energy

Power system flexibility encompasses maintaining the supply–demand balance, based on [15], the existence of storage capacity for balancing [16], the adjustment of demand (i.e., demand-side management (DSM)), and the operation in the context of a market that allows flexibility [17]. Active participation of the demand side, as explained by the authors of [18], involves activities such as load management and self-production, which are integral components of DSM. In addition, the International Energy Agency analysis of the demand response (described in the following section) notes that “this flexibility will become increasingly important as grids become progressively dominated by variable power generation, such as wind and solar PV”. These mechanisms, including load levelling, valley filling, and load shifting, enable consumers to actively engage in the energy system. To facilitate DSM, enabling technologies, such as IoT-driven automation, connect devices such as PV systems, battery storage, and smart meters to exchange information on energy consumption and consumer responsiveness to price signals [19]. This discussion also delves into the adoption of load management and self-consumption solutions in the studied territory. Three types of DSM activities in relation to the analysis are non-exhaustively described below.

(a)

Demand response

Historically, electricity production and transmission have evolved to meet rising consumption demands. However, the demand response is now shifting this dynamic by encouraging consumption to adapt to electricity production and transportation conditions [20]. The demand response involves temporarily reducing a site’s consumption in response to an external request. This reduction modifies the initially planned consumption, possibly involving a time shift [21]. The advantages of the demand response include reduced network connection investments and power system decarbonization. Flexible consumption can lower emissions from conventional peak-hour power plants. The demand response comes in two forms: explicit and implicit [22]. The first involves consumers facing time-variable electricity prices and participating by responding to price signals that change over time [23]. In Europe, time-of-use (TOU) pricing, which divides the day or year based on peak/off-peak hours or seasons, prevails for the energy component, but real-time pricing (RTP), with hourly rate adjustments based on market price adoption, is on the rise [24]. The second form is explicit, based on incentive programs, whereby consumers who shift demand receive payments through the demand response. Moreover, according to the IEA Net-Zero scenario, 500 GW of demand response will be deployed by 2030, which corresponds to 10 times the value of the year 2020 [22].

(b)

Self-consumption

The concept of self-consumption of electricity is widespread across Europe [25]. The interpretation of this concept is found in the EU Directive 2019/944 (article 2 (8), related to “active customers”) and the REDD II, which defines the “renewable self-consumer” in EU Directive 2018/2001 (article 2 (14)). However, the former definition offers more opportunities to customers, as it states that they can participate in flexibility and energy efficiency schemes.

Nowadays, the most competitive renewable energy sources are solar photovoltaics (PV) and wind [26]. According to [25], solar PV in Europe has approached “grid-parity costs for the kWh produced”. A common solution in territories constrained by surface area, such as islands, is PV electricity self-generation with solar panels integrated on building rooftops. Residents become “prosumers” by self-consuming generated electricity, selling excess power to the grid, or storing it for later use, fostering energy transition, raising awareness, and addressing environmental concerns. These systems are usually connected to storage technologies for optimal use. Nevertheless, the focus is on reducing the carbon footprint of these solutions. As the industry grows, more transparency is needed in the value chain of these technologies, while the options of second-life use and recycling enhance its sustainability. In addition, policy plays a significant role in regulating the battery market [27].

(c)

Electric vehicle charging

Creating synergies across the energy system can also identify possibilities for enhancing the flexibility and economic efficiency of the energy system [28], as well as making electricity prices more elastic for residential users [18]. This can be applicable with the electrification of the transportation sector. By controlling the charging of electric vehicles (EVs), the operator can avoid creating additional demand during peak hours and employ EVs to perform peak shaving or valley filling [18]. Two strategies exist in electric vehicle (EV) charging control: passive control, which encourages charging during low-electricity tariff periods, and active control (smart charging), which not only targets off-peak or low-tariff periods, but also involves modulation of charging power, allowing for better electricity demand management through input current adjustments. Smart charging offers two further options: unidirectional (V1G) and bidirectional (V2G), enabling power injection back into the grid [29].

1.4. Organization of the Paper

This paper is organized as follows: it begins by describing the modeling methodology in Section 2, including a presentation of the energy system modeling, followed by a focus on the tool used. Section 3 introduces the main assumptions integrated into the two models to understand the energy systems of each island and their evolution. Section 4 details the case tests implemented to form scenarios, showing renewable energy and storage investments and load curve variation. Section 5 elaborates on the impacts of the pricing structure with the integration of renewables and offers recommendations for future pathways, before concluding with final remarks in Section 6.

2. Method

To analyze the future development of territories’ power systems considering the different challenges facing them (e.g., decarbonization, integration of high amounts of renewable energy, choice of investments, etc.), it is crucial to consider the plausible long-term trajectories of their evolution. Energy system models are frequently used in this context, bearing in mind that investments in the energy sector are often expensive and their success mainly depends on planned, well-defined scenarios covering a certain time horizon [30]. In this section, a review of energy system analysis and TIMES modeling is presented. Nevertheless, short-term energy planning is also employed for scheduling the operation of electricity grids and evaluating the integration of intermittent renewable energy, storage technologies, and pricing structures. This was the case of [31], which used battery storage as a flexible demand response framework and TOU to make a financial and economic analysis.

2.1. Background on Energy System Modeling

Generally, in energy system modeling, the scientific community distinguishes two approaches to integrating and representing the energy sector: bottom-up and top-down approaches [32]. Bottom-up optimization is known to provide a detailed representation of the energy system (technologies, processes, emissions, economic and financial properties, etc.) and the energy demand is satisfied by a certain market or technology option [33]. In contrast, top-down models provide a representation of the interaction of the whole energy sector with other economic sectors, for example, by identifying relationships between the economy and energy through macroeconomic indicators, or even by adding socio-technical aspects to the energy transition [32]. However, these models are not appropriate to assess the evolution of the energy system with all available low-carbon technologies. For instance, in [34], it was shown that to model carbon capture and utilization (CCU), bottom-up models still dominate due to the lack of information on this technology’s socioeconomic impact. Top-down and bottom-up models can be linked together, where the output of the former is used as input for the latter. These models are known as hybrid models combining techno-economic details and constraints to assess economic (or socioeconomic) impacts, such as the case of the POLES-JRC model [35], which integrates the energy sector with a macro-economic evaluation and climate policy assessment [32].

The modeling tool employed in this study adopts a bottom-up optimization featuring two approaches for long-term planning: perfect foresight and myopic foresight, referring to the expectations of the economic actors [30,32]. In the first case, the development of energy prices, future improvements in certain technologies, and future decommissioning of power plants are known a priori. As for the myopic model, decisions are made based on the status quo of these factors, which requires that the optimization problem be divided into recursive subproblems, yielding suboptimal solutions. Babrowski et al. [30] switched from perfect foresight to a myopic approach in PERSEUS-NET, a bottom-up, linear optimizing energy system model used for studying the evolution of the German electricity generation system until 2030. As a consequence, the computation time was reduced ten-fold. In case of sudden events, such as the development of a new technology, perfect foresight provides an upper limit for an optimal performance of the system, but the myopic approach yields a more realistic output. However, for foreseen events, such as an expected rise in CO₂ prices, myopic approach solutions are suboptimal. This highlights the importance of exploiting the complementarity of both approaches.

A set of characteristics helps energy modelers to compare these tools, and they mainly utilize the techno-economic detail (scores can be allocated to the inclusion of flexibility options, especially when addressing future low-carbon energy systems), time resolution, geographical representation, sectoral coupling, and its representation [32,33]. Some of the well-known energy system models dealing with long-term planning, employing a bottom-up analytical approach and perfect foresight, are TIMES/MARKAL (version 4.8.1) from the IEA-ETSAP [36], LEAP (https://leap.sei.org/ (accessed on 19 January 2022) [37] from the Stockholm Energy Institute (SEI) [38], MESSAGE (https://docs.messageix.org/en/lates (accessed on 1 March 2024) [39] from the International Institute for Applied Systems Analysis (IIASA) [40], and OSeMOSYS (http://www.osemosys.org/ (accessed on 1 March 2024) [41] by the Royal Institute of Technology (KTH) [42]. The solution algorithm for LEAP is based on a heuristic algorithm, in contrast to the others, which use linear programming. This method provides a suboptimal solution in a rapid computational time. In addition, emerging energy system modeling tools are built with the objective of addressing the high integration of renewable energy in electric systems, thus handling operation and investments. This is the case of PyPSA, (v0.28.0) which addresses linking “power system analysis software and general energy system modeling tools” [43], but has other features such as sector coupling with other energy carriers, such as synthetic fuels and carbon removal with direct air capture (DAC). PyPSA is open source and it has been used to study various subjects, including the European transmission network for the ENTSO-E area and the optimization of power systems with renewable energy integration [44]. Another model aimed at exploring carbon-neutral energy systems is the Lappeenranta University of Technology (LUT) model. Its key features include representing the energy system in detail and finding the least-cost solution considering sets of constraints. It has been used for various energy transition scenarios, such as variable renewable energy integration in the Indian power system [33].

Observing that each model possesses distinct advantages over the other, the choice of tool often relies on the modelers’ expertise, experience, and primary research objectives. For instance, TIMES/MARKAL is known for its high granularity and capability of showcasing the sector coupling, which avoids losing the coupling’s impact. OSeMOSYS (http://www.osemosys.org/ (accessed on 1 March 2024), on the other hand, is open source and thus does not require upfront financial investment. Furthermore, tools such as the TIMES/MARKAL family are being constantly developed and extended, such as stochastic programming for optimization with imperfect foresight for assessing the robustness of policies under uncertainties [32]. TIMES modeling is further elaborated anddescribed in the following section.

2.2. TIMES Modeling

In this section, an overview describing the TIMES paradigm is provided, with a focus on the mathematical formulation of its objective function and characteristics relevant to this study. TIMES is a model generator developed by the IEA-ETSAP [45], whose code is available online (https://github.com/etsap-TIMES/TIMES_model (accessed on 1 February 2024)).

2.2.1. Description of a TIMES Model

A TIMES model describes the energy sector of a chosen region by sets of processes connecting primary energy sources and final energy demand through explicit and implicit input and output technologies [46] or commodity flows (commodities can be energy carriers, materials, emissions, etc.) [47]. Hence, present and future sources of primary energy supply, their potentials, and available and future technologies can be integrated using their characteristics. This allows adding new technologies that are currently too expensive but may be competitive in the future [47]. Another important capability of this energy system model is that it catches dynamic aspects of the power sector, such as flow equilibrium conditions, to follow load curves [46]. TIMES allows the user to divide the time horizon into years and subdivide it into four sub-annual segments: annual, seasonal, weekly, and daily [47].

The quantities and prices of the different commodities are in a state of equilibrium, meaning that their prices and quantities at each point in time are aligned in a way that ensures that suppliers supply precisely the amounts demanded by consumers. This state of balance possesses the characteristic of maximizing the overall economic surplus. Figure 1 conceptualizes the mathematical description of TIMES.

2.2.2. The Objective Function and Scenarios for Possible Energy Futures

The objective function is considered as a linear mathematical expression of decision variables subject to linear constraints. It represents a minimization of the total discount costs of a power system over a long time horizon, as per the equation below. The decision variables are determined by the optimization (i.e., endogenously), whereas the constraints are equations or inequalities involving these variables and must be satisfied by the optimal solution. They could represent a number of environmental, technical, and demand constraints [48]. A TIMES model is vertically and horizontally integrated. Processes or technologies can be linked through inputs and outputs, where the outputs are expressed as linear functions of the inputs.

All these properties imply that TIMES is a partial equilibrium model that uses a linear programming (LP) approach. The objective function is, therefore, expressed as [45]:

$$min\left(NPV\right)=min(\sum _{r\in R}\sum _{y\in Y}{\left(1+d_{r,y}\right)}^{T_{0-y}}·ANNCOST(r,y))$$

where:

• $NPV$ is the net-present value of the total cost for all regions, r, and years, y,
• $R$ is the set of all the regions,
• $Y$ is the set of years with costs,
• $T_0$ is the reference year,
• $d_{r,y}$ is the general discount rate in region, r, and year, y,
• $ANNCOST(r,y)$ is the total annual cost in region, r, and year, y. It includes capital costs from investing or dismantling processes, maintenance and operation, and trade costs.

Long-term simulations with TIMES make the scenario approach the best option, while short-term simulations may use econometric methods. Scenarios are based on coherent assumptions about future trajectories, organizing the system under study. Scenario builders need to test assumptions for internal coherence through a credible storyline. The demand component is essential when building a TIMES scenario. Demand drivers can be exogenous, such as using population, GDP, households, etc. National and sectoral output growth rates can be obtained from general equilibrium models, such as GEM-E3 [45,49].

2.2.3. Time Granularity

The time horizon in TIMES can be divided into several time periods, where each period contains a number of years that are possibly different. In a period, each year is considered identical, except for the cost objective function. It includes differentiation between payments in each year of a period. Input and output values, related to a period, t, are applied for each year in t. This concerns capacities, commodity flows, operating levels, etc., except investment variables, which are implemented once in a period (unless the period exceeds the technical life of the investments). There is a possibility to create time divisions using time slices (seasons, portions of the day/night, etc.). This is useful with production technologies that have different characteristics depending on the time of the year, such as wind turbines. Technologies storing commodities for later discharge can be defined and modeled. Different time slices may require different production technology deployment and investment decisions for peak reserve capacity. The initial period and its quantities of interest are fixed by the user as part of the calibration, an important step in creating a TIMES model.

3. Case Studies and Model Description

3.1. Case Studies: Italian and Norwegian Islands

The islands studied here are each situated in different geographical locations of Europe, and thus subject to different weather conditions and natural resources’ availability. Their size and population density are other parameters that constitute a contrasting setting. In addition, the challenges they face, especially from an energy system perspective, are common to electricity systems on relatively isolated islands (i.e., existence of electricity connection to mainland), but the consequent issues remain diverse for these islands. A description of each territory is presented in

Table 1. Territory characteristics [3].

	Procida, Italy	Hinnøya, Norway
Territory characteristics	Smallest island in the Gulf of Naples. Area = 4.26 km2 Inhabitants = 10,428 [50] Density = 2449.1 inhabitants/km2	Fourth largest island in Norway, Harstad City is the center of economic activities. It forms an island cluster with the smaller islands Grytøya, Bjarkøya, and Sandsøya [51]. Area = 4533 km2 Inhabitants = 66,690 Density = 16.4 population/land km2 [52]
Electric system	Dependent on electricity imports.	Situated in Nordpool region 4 (NO4) and imports its electricity from this market, which is based on hydroelectric power generation [53,54].
Challenges	Limited by space and surrounded with a protected marine area. Grid congestion and seasonality of demand.	Limited possibilities for new grid connections.

, followed by an overview of the policy setting related to renewable energy in the respective countries.

For the case of Italy, the legal framework surrounding climate change policies has been developed in the context of European recommendations, while the government is responsible for implementing various measures. Also, within the framework of the Covenant of Mayors, local administrations have implemented climate plans and measures. Italy has a goal of integrating the use of renewables at nearly 37% of the gross final energy consumption in 2030 to be in line with the Fit for 55 goals. The main types of support mechanisms to promote renewables are the feed-in tariff (FiT) and the feed-in premium (FiP). Other mechanisms include energy efficiency certificates and net billing (“scambio sul posto”) preceded by the “conto energia” feed-in tariffs and green certificates that ended in 2013. To further promote the growth of renewable electricity generation, the Italian government encourages self-consumption and energy communities, including on small islands. By the Ministerial decree RES1 [55], FiT/FiP schemes became the main support, where FiT targets small plants and FiP targets large ones. The support scheme for islands’ “RES auctions” was established by the FER1 Decree and provides a premium in addition to a base incentive paid for the PV electricity produced, whether it is fed into the grid or self-consumed [56]. Such mechanisms are related to energy and capacity markets and may impact investors and prosumers’ profitability. In [57], it was found that the demand response and FiP drive an increase in renewable energy investments (wind energy in the case of the study) and a decrease in baseload generation and, hence, emissions. FiP increases the costs for consumers; thus, regulations, such as premiums, are established to protect their participation and reduce trade-offs. Currently, Italy’s support for the IEA Digital Demand-Driven Electricity Networks initiative underscores the purpose of the demand response by modernizing power systems and establishing the appropriate regulatory framework [55].

Norway’s climate policies target a 90–95% reduction in greenhouse gas emissions compared to 1990 levels by 2050, excluding carbon sinks. Its carbon pricing mechanism (included in the modeling, see Section 3.3) and abundant renewable resources in hydro and wind provide a strong basis for achieving this goal [58]. Nevertheless, the country still faces challenges to decarbonize its industry and transport sectors. In terms of legislative measures for increasing the integration of RES, the country only implemented an electricity certificates system in 2015, following Sweden, which introduced it in 2012. All producers and some consumers are obliged to buy certificates for a certain percentage of their production (or consumption). This percentage was gradually increased until 2020 before decreasing until 2035 when the system will be withdrawn. Italy considers a demand response mechanism as a way to ensure the stability and flexibility of its grid, whereas Norway adopts the mechanism to deal with high electricity prices, especially in the southern part of the country, and to delay investments in new grid capacities.

These island case studies present a similar goal of decarbonizing the energy system with the integration of high shares of renewable energy in the power mix. The solutions to achieve the targets are analyzed in this study according to the specificity of each island. This offers decision-makers and local authorities a broader view of the types of solution needed, but also opens the discussion on the replicability of solutions on other European islands or territories.

3.2. Procida Energy Model: TIMES-Procida

We opted to model Procida’s electricity system (Figure 2), as it is the energy commodity that is mostly used in this territory. From the supply side, the island acquires 99% of its electricity through the marine cable connecting it to Ischia. The electricity demand of the island in 2018 was 19,915,668 kWh and, considering the limited data refinement, the sector distribution estimation is based on information from Procida’s 2015 sustainable energy action plan (PAES), submitted as part of the European Covenant of Mayors program [13]. The rooftop solar photovoltaics (PV) represent a small percentage (1%), making the island extremely energy-dependent and suitable for improvements by increasing the share of renewable energy with self-consumption PV investments. The residential sector stands as the largest consumer, accounting for 62.7% of demand, trailed by the tertiary (service) sector at 28.4%. The public sector constitutes 5.4% of demand, while agriculture, industry, and transport represent 2.1%, 1.3%, and 0.1%, respectively.

To increase the renewable energy share, new photovoltaics have been considered for installation on building rooftops from 2019. The aim is to install PVs on residential, public, and tertiary buildings to analyze their contributions. The PV Watts calculator by the National Renewable Energy Laboratory (NREL) calculates the maximum PV capacity for the island’s buildings. Lithium-ion batteries are considered for storing electricity daily in residential, tertiary, and public buildings to enhance flexibility and manage electricity generation. Battery characteristics vary for each sector where PVs are installed. Different batteries are chosen for each sector based on their capacity and C-rate. Constraints are set to determine the upper limit of battery capacity for each sector. The maximum allowable capacity for each sector is calculated using linear equations, multiplying the average battery capacity in a sector by the number of buildings. Only distributed storage systems that can be coupled with PVs are implemented, meaning storage does not charge from the grid but can inject electricity in case of excess production (more details are found in [13]).

Three scenarios were chosen to evaluate the results from the implementation of self-consumption on the island. These scenarios are based on the deployment of PV at low and high investment capacities, and the third one considers high deployment of PV with storage technologies (HIGH_STG), as storage integration becomes relevant with considerable amounts of solar electricity production. TIMES-Procida includes the future technologies of hydrogen storage and electric transportation, and the choice was made to avoid concentrating on these two aspects within this analysis, since hydrogen storage is costly, and it is only chosen by the techno-economic optimization when enforced, as investment and transportation demand for electricity is negligible on this island compared to the remaining sectors. Time periods were divided into 15 time slices: 3 seasons (winter, summer, and intermediate) and 5 daytime intervals. The dates separating the seasons were set by the Italian decree (DPR n. 412 del 26 Agosto 1993), which outlines heating periods for different areas based on climatic zones (Procida falls under zone C). Winter spans from 15 November to 31 March, while the summer season lasts from 15 May to 15 September, chosen due to higher average irradiation values observed in Procida from April to September (Figure 3). The intermediate season covers 1 April–14 May and 16 September–14 November. The defined time slices of TIMES-Procida for the summer and winter seasons are presented in

Table 2. Seasons and time slices of TIMES-Procida.

Season	Time	Timeslice Name	Hours
Intermediate “I”	Night	NGT	0–6:00
	Morning	MOR	6–10:00
Winter “W”	Midday	MID	10:00–15:00
Summer “S”	Afternoon	AFT	15:00–19:00
	Evening	EVE	19:00–24:00

and can be read by aggregating the season with the time slice name; for example, for the period between 6:00 am and 10:00 am in the summer, the time slice would be denoted SMOR, while for the period between 7:00 pm and 12:00 am in the intermediate season, the time slice is IEVE. The horizon of the study was 32 years (from 2018 to 2050), where the sub-periods of one year were defined until 2035 and then became wider sub-periods (seasons). To determine the daytime level, the annual load curve was compared to the global clear-sky irradiance [13].

3.3. Hinnøya Energy Model: TIMES-Hinnøya

A first approximation of the RES is shown in Figure 3. The reference year for the model was 2015. The energy system of the island relies on imports of electricity from the main grid connection of the country of Norway, which is the Nordpool electricity market [54]. Fuel imports are mainly used for the transportation sector, with only diesel used for producing electricity to power the fish farming sector [51] and included in the primary sector. The residential sector consumes electricity, which pertains to households’ use of electricity for heating, lighting, appliances, and other domestic activities. Economic activities are also present on the island, such as farming, wholesale, tourism, and kindergarten, and they mainly use electricity. Thus, the demand side covers the primary, secondary, and tertiary sectors. The electricity in Norway is hydro-based, and thus these sectors are already considered decarbonized. The remaining sector is transportation, and demand for mobility is disaggregated in the model into private, public, transport for merchandise, and marine transport, with a distinction between short and long distances.

According to Harstad municipality in their Climate Budget of 2021 [59], the transport sector makes up the majority of emissions (69% in 2018), which includes maritime transport (i.e., shipping; Figure 4). A possible measure to decrease these emissions is the deployment of low-emission transportation, hence contributing to the decarbonization of the island.

The evolution of the electricity supply with renewable energy, namely, hydroelectricity and wind, was based on the study of the Norwegian Water Resources and Energy Directorate (NVE), which shows that there is a potential for investing in new hydroelectricity power plants. The choice of technologies took into consideration on-shore and off-shore wind potential [14].

Moreover, the Norwegian government adopts several policies for the deployment of low-emission transportation. For instance, in 2025, all new purchases of passenger cars or light vans should be zero emission. Also, a taxation on fossil fuel in the transport sector is a central policy [60] and is implemented in TIMES-Hinnøya [61].

This taxation is applied in the following form:

$$\mathrm{TAX}=C_{{emission}_i}\times \sum {Emission}_c$$

where C_emissionsCO2 = 50 Euros/tCO₂ and C_emissionsNOx = 2280 Euros/tNOx [60], and c is the emitting commodity whose emission is defined in terms of kt/TJ, from [62].

This deployment also includes proper incentives and taxation rules. The new low-emission vehicles are exempted from tax on registration, traffic insurance, and road use tax and tolls ([3], found in [63]). Hence, one solution is the electrification of this sector, which is accelerated in the whole of Norway to reduce emissions and to benefit from the demand response mechanism. This is reflected on the local level, where numbers of battery electric vehicles (BEVs; in this case, private cars) are noticeably increasing in Harstad [51]. Consequently, the hypothesis of investing in battery electric vehicles for passenger car transportation (Table 2) is upheld, along with the examination of user flexibility through demand reduction or the shift to electricity. However, the model considers several modes of transportation (transport passenger cars, light and heavy duty, maritime transport, etc.). Transport passenger cars were chosen due to their potential to provide flexibility, with battery capacities of 15 kWh, 30 kWh, and 60 kWh. Based on the information provided in [51], the average commuting distance is 30 km and, therefore, the power consumed by these vehicles is sufficient for consideration. The BEVs are modeled in TIMES-Hinnøya as investment capacities in storage batteries, so their technical parameters are included in the declaration of technologies. The TIMES model invests in these technologies in a cost-efficient manner while optimizing the battery charging in such a way as to meet demand by the transport sector, which is provided as exogeneous input. This allowed us to define the place of BEVs in the energy transition of the island. In addition, BEV chargers were included in the new technologies, including fixed and O&M costs. Storage technologies in the model consist of a hydrogen bromide flow battery [64]. However, this was not included in demand participation since it is likely to be piloted by the DSO of the island.

For the tariff structure, Norway adopts the RTP pricing scheme, which is widely accepted in this country, where 71% of households and 88% of small and medium enterprises and small industries use this scheme [24]. Also, it is adequate for customers to reduce their bills, as the heating sector is highly electrified in Norway, for instance, in residential homes, and the deployment of electric vehicles is on the rise, as mentioned earlier, so this pricing structure was implemented in the model. Electricity trade was calibrated for 2015 (the base year, Figure 5). Prices were obtained from the Nordpool database. NVE forecasts were used for future electricity prices ([65], found in [3]). The temporal structure in Figure 5 shows three levels, distinguishing between the months forming a year, weekdays and weekends, and finally, hours. In total, there were 576 time-steps that could be used for studying supply and demand in the long term [3].

Table 3. Summary of the main assumptions of the models.

Assumptions	Procida	Hinnøya
Demand evolution	Italian GDP growth	Population growth for residential sector GDP growth for the remaining sectors
		Transport service demand
Price evolution	Constant (EUR 184.7/MWh) [66]	Variable on different timescales (hour, month, and year)
Horizon	2018–2050	2015–2050
Time slices	15 (3 seasons, 5 daytimes)	576 (12 months, weekday, weekend day, 24 h)
Discount rate	6%	6.50%
Currency	Euros (€)	Million euros (M€)
Regions	1 region (Procida)	3 regions (Harstad, Grytoya, and the rest of Hinnøya)
Future technologies	Solar PV	Hydro power Wind power
	Storage: Li-ion Smart Energy Hub [67] (hydrogen electrolyzer, tank and fuel cell + Li-ion battery) Long-term hydrogen storage	Storage: Li-ion and flow battery (Hydrogen Bromide Elestor [64])
	Electric vehicles and bikes	Electric transport
Import/export	Electricity imports Electricity exports not allowed	Electricity, fossil fuel (diesel and gasoline), biofuel, MGO Electricity exports allowed
Trade	N/A	Between the regions
Emissions	N/A	Emissions from fossil fuels

summarizes the main assumptions included in the two models.

4. Results

In this section, the outcomes obtained from the TIMES optimization with the relevant scenarios are examined. For the case of Procida, solar PV and storage investments as part of the “self-consumption” measures were analyzed for the three scenarios. Then, a pricing structure based on the time of use was applied. For the case of Hinnøya, the transport sector was scrutinized, highlighting its electrification with renewable energy integration and demand-side load-shifting measures.

Table 4. Different cases set in the hypothesis for the two islands and the main results.

	Procida	Hinnøya
Objective: renewable integration and decarbonization	Solar PV integration. Reduced imports from mainland.	Decarbonization of the transport sector with RE increase. Reduced grid tension.
Implementation	Rooftop PV and storage investments. Electricity tariffs (TOU).	Electric vehicle charging structure. Electricity tariffs (RTP).
Hypothesis or test cases	Test case 1: Modest to high shares of PV with constant electricity prices: $\begin{array}{ll}{CapacityPV}_{High}& \ge Capacity{PV}_t\\ & \ge {CapacityPV}_{low}\end{array}$ Test case 2: Allowing investments in batteries + high PV shares + constant prices of electricity: $C_{max,BAT,s}=\overline{C_{BAT,s}}·N_{B,s}$ Test case 3: Time-of-use structure + high PV + storage: $P_{peak}=\left(1+20\%\right)\times P_{off-peak}$ where P is the price of electricity	Test case 1: Taxation’s impact on future passenger car deployment: $\mathrm{TAX}=C_{{emission}_i}\times \sum {Emission}_c$ Test case 2: The introduction of EV and constant prices. ⮚ Load curve variation analysis: $P=constant\forall i$ where P is the price of electricity and i is the hourly time slice. Test case 3: Introduction of EV with RTP (charging adaptation V1G). ⮚ Load curve variation analysis
Results	Shift in the electricity supply with up to 10% decrease in imports at peak times in 2050. Favoring storage technologies, with TOU, optimizes the use of RE and improves system flexibility (Section 4.1).	Management of the additional electricity demand. Preventing peak loads by load shifting to low-demand hours (1:00 am to 6:00 am and 7:00 pm to 12:00 am) (Section 4.2).

shows the establishment of testing used for validating these decarbonization approaches.

4.1. Procida Island

First, three scenarios were chosen to represent the integration of solar PV on building rooftops in Procida. The trend of these investments is shown in

Table 5. PV and storage investments according to three scenarios.

LOW Scenario			HIGH and HIGH_STG Scenarios
PV investments:			PV investments:
2018–2020	50	kW/year	2018–2020	50	kW/year
2020–2025	80	kW/year	2020–2025	150	kW/year
2020–2030	80	kW/year	2020–2030	200	kW/year
2030–2040	100	kW/year	2030–2040	250	kW/year
2040–2050	120	kW/year	2040–2050	300	kW/year
Storage investments: not included as possible technology.			Storage investments: allowed only for the HIGH_STG scenario.

, where buildings’ geographical constraints on PV installation were taken into account, as per [13]. The LOW scenario represents a case of modest PV, which follows the trend of current installations. Figure 6 shows the results obtained from the optimization of the model under the HIGH scenario, which was allowed to reach 300 kW of PV installed by mid-century. In terms of the optimal solution, the investments reached the allowable capacity in all the studied sectors.

The impact of the integration of self-consumption on the island’s demand for imports (ELC Grid) is shown in Figure 7. With the HIGH scenario, the self-sufficiency of the island was enhanced, with reduced imports, especially during hours of PV generation. The benefit of coupling storage to PV was noticeable during peak hours, namely, in the afternoon and the evening.

It was also observed that the introduction of significant PV capacity, even when coupled with storage technologies, failed to mitigate imports during winter evenings. Nevertheless, as per local authorities, the highest demand was projected for the summer, a trend not mirrored in the existing data. This is mainly due to tourism activities, which cause congestion problems on the grid. Measures that can decrease imports during summer peak times are useful in this case. The increase in the shares of PVs or the decrease in imports directly relate to the decarbonization of the island, as the only two possible supply technologies are PV and imports. In the case of the HIGH scenario, during the summer midday time slice, it can be seen from Figure 7 that PV electricity covered all consumption, even in sectors where no PV was installed, such as agriculture and industry. When storage investment was allowed in the model, the electricity mix changed: PV electricity was stored for later use in the evening and afternoon when demand was high, involving long-term storage investment.

Next, the decision was made to incorporate dynamic electricity pricing, considering Italy’s pioneering implementation of time-of-use (TOU) tariffs in 2010. This pricing reform became mandatory in mid-2010 for residential customers subscribing to basic contracts and touched 42% of the population [24]. It was chosen to start with a difference of 10% (set to be increased [68]) of the price between peak and off-peak (peak price being the highest) based on [24], since this is the only difference able to incentivize around 60% of Italian households to shift their power demand. According to [69], the Autorità di Regolazione per Energia Reti e Ambiente (ARERA) sets the different time slots for the pricing structure. There are three time slots, F1, F2, and F3, associated with a specific period of the day or an entire day. In view of this tariff scheme, and according to the model structure built in five time slices, a TOU can be implemented where the prices of imported electricity in the NGT and MOR are lower than in “peak hour” time, thus approaching the F2 and F3 slots for domestic users [69]. Note that in the current modeling structure, storage did not charge from the grid. This decision was mostly made to reduce reliance on the grid, as there are congestion problems involved in acquiring electricity through the submarine cable.

In this configuration, the renewable electricity price remained competitive on average. Thus, the peak price was further increased to a 20% difference between peak and off-peak, and the scenario including storage technologies was used. This scenario would allow the assessment of price-based mechanisms and a comparison between measures for integrating high shares of renewable energy in the grid. Figure 8 shows the impact of TOU on technology investments on the island: the model invested globally in higher capacities of Li-ion batteries compared to a scenario with a constant electricity price. This amounted to almost 1.5 times more total installed capacity when reaching the end of the horizon, compared to HIGH_STG, due to the appearance of investments in tertiary-dedicated batteries. Moreover, this led to increased investments in batteries for the residential sector earlier in the time horizon. In Figure 9, a decreasing pattern is seen in the share of electricity imports from the mainland with respect to the total supply, for the scenarios with constant and changing electricity prices. The difference between them can be observed until 2040. The share of electricity imports in the electricity supply reached 73.5% and 65.2% at the end of the horizon for HIGH_STG and HIGH_STG_TOU, respectively.

Conversely, the choice of investing in batteries changed the energy supply mix for the residential sector, since it is the most energy-intensive sector on the island. Figure 10 shows the charge and discharge of residential batteries, corresponding to a total installed capacity ranging from 320 kWh to 510 kWh (Figure 8).

In comparison with Figure 7 for the HIGH_STG scenario, the TOU pricing structure provides an alternative option to directly using PV. The strategy aims to meet demand with grid electricity when prices are low on winter mornings and store the electricity from PV to release later when both prices and electricity demand are higher, during the afternoon and evening. Summer midday remained interesting in all cases due to the availability of solar power.

Due to the investments in storage, the electricity supply mix differed, especially for electricity provided by PV and batteries. In fact, for the residential and tertiary sectors, the annual production of PV directly injected into the respective sectors decreased, and more storage was used to inject electricity during high demand periods. At the system level, the TOU decreased imports at the peak hours (i.e., afternoon and evening time slices, Figure 11) in 2050 (−4% in the evening and −9.78% in the afternoon), while shifting the demand to the morning period.

A disaggregation of the time slices on a finer level would enhance this representation but is still not possible with the available data for this island. To sum up, this demonstrates one of the ways electricity storage can be used to provide flexibility and help relieve the grid during congestion periods.

4.2. Hinnøya Island

In terms of decarbonization, transport evolves toward electrification, with an increased share observed starting from 2025 (Figure 12). The country of Norway adopts taxation on fuel used by cars, which was implemented in the model by assigning a cost per kilo ton of energy used for CO₂ and nitrogen oxide (NOx) emissions related to the fuels used. In addition, diesel and gasoline are blended with a concentration of biofuel, denoted in Figure 12 as DSL blend and GSL blend, respectively. However, EVs remained the winners since no emissions are associated with this technology, considering the hydro-dominated electricity supply, with a total electrification at the end of the horizon.

With the increase in electrification of transport passenger cars, opportunities for flexibility offered by this sector would present benefits for the grid. This can be carried out through demand-side management strategies applied to the charging of EVs. This translates into an economic approach used by TIMES, which chooses to charge the lowest electricity prices while meeting demand.

In addition, the optimization of TIMES performed on the RTP pricing structure identified the periods of charging (during low-electricity tariffs) while responding to mobility demand. The period and electricity costs corresponding to this structure are shown in Figure 13.

The variation in the load curve (Figure 14) was analyzed if the prices were fixed throughout the day by taking the average prices for a day. It represents a fixed or “static” tariff structure, which lacks economic incentives for consumers to charge during off-peak hours (Figure 14). Note that the EV charging in that case will be carried out throughout the whole day, as shown by the “red” curve, responding to mobility demand but adding loads to peaks, which proved the projected increase in load due to the electrification of the transport sector without appropriate load management techniques. Nevertheless, the DSM mechanisms would benefit the grid since, for the “gray” curve, during the off-peak periods (1:00 am to 6:00 am and 7:00 pm to 12:00 am), the DSM strategy of EV charging consisted of the valley-filling phenomena. The shaded “gray” area shows that the valley-filling occurs upward in the off-peak and downward in the peak periods. At the country level, the increased electrification of the transport sector in Norway would be challenging to the grid [70]. Moreover, for the island of Hinnøya, new connection possibilities are limited (Table 1). The charging strategies, if they are controlled, provide flexibility so that electrification would be successful and lead to decarbonization of the highly emissive transport sector. No additional demand is created during peak hours, as seen in Figure 13.

5. Results Discussion

The possibilities for demand-side strategies in a context of decarbonizing the power systems of the two territories were investigated according to the issues and application environments where they are applied, which remains a major aspect in developing long-term power system analysis.

First, the results for Procida demonstrated that PV is cost-optimal, and that storage would alleviate imports of electricity at hours of high demand. The only way to harness PV potential in this case is rooftop PV, supporting “self-consumption”. This is important because the island is “relatively isolated” in terms of its power system, as it relies only on submarine cables for its electricity supply. From the scenarios with price variation according to the time-of-use structure, the model changed the choice of investment, in terms of capacities of batteries. These investments are impacted by the different load curves of each sector where PV and storage systems would be installed. The tertiary sector sees its peak load ranging from midday to evening, where the price of electricity imports is still high, making it economically convenient to rely on batteries. However, for the public sector, peak demand starts in the morning in this scenario, and the MOR time slice (low tariff) coincides partly (for two hours) with high demand; hence, fewer batteries would be needed. The option of self-consumption (rooftop PV) decreases dependency on the grid, particularly for relatively isolated islands. Furthermore, this combination enhances the integration of solar-based electricity in the grid, thus decarbonizing the island’s energy system. This structure is commonly used for managing the integration of high shares of RE, especially with high storage costs. However, it incurs costs for system operators that are not captured in this long-term planning. Short-term planning could harness the benefits of the demand variation due to varying electricity prices. Here, long-term planning was used to compare investments in new technologies that could reduce the reliance on the marine cables for the Italian case. The results demonstrate its contribution to increasing self-consumption as an active participation of the demand side.

One major challenge in decarbonizing the energy system of Hinnøya island relates to the transportation sector. It is necessary to increase the share of renewable energy for the island by harnessing emission-free electricity. However, this would increase the pressure on the grid. Then, the demand response would facilitate decarbonization. RTP, which is already installed, is one way to respond to the load. This pricing system makes it easier to manage additional demand created by the massive electrification of the transport sector. Also, it would provide a way to avoid the creation of peaks. The Norwegian case demonstrated an opportunity to decarbonize the transport sector, which is an important aspect of development for the island and the country. With the policies implemented, such as taxation on fuel, the optimal solution found relies on electric vehicles, although other options are available with biofuel. Modeling the whole energy system provides a global and precise indication of the transition of the transport sector by including the supply side with fuel imports, the blending processes with biofuel, and the distinction between traffic volumes and new market shares of petroleum, diesel, and electric vehicles.

The analysis of the decarbonization pathways with the active participation of the demand side showed that planning for the electrification of the emissive sector with RE is a common success factor. In fact, the same goal was studied for the two EU islands, although different approaches were required. To achieve this goal, one common practice is to establish the energy transition through new investments in technologies for electrification (solar PV, storage, and EV) and demand-side mechanisms. This is also supported by some policy instruments, such as fuel taxation and tax exemptions. The outcomes of the study demonstrated how each solution influences the other, for example, in the case of the TOU pricing structure and storage investments. The main characteristics outlining the outcome of the study are shown in

Table 6. Comparative summary of the outcomes.

	Procida	Hinnøya
Strategies analyzed	Self-consumption Electricity price TOU	Eletric vehicle charging Electricty price RTP
Implementation challenges	Renewable energy is limited The small size of the population-dense island Protected areas surrounding the territory	Limited possibilities of new grid connections Diesel generators to mitigate this issue
Success factors	Local production of electricity with PV Storage solutions applicable with rooftop PV	Cross-sectoral flexibility solutions with the electrification of the transport sector with policy support Load control
Policy implications and cross-regional lessons	Interconnected systems that will see further integration of renewable energy Socioeconomic impacts

. Therefore, the success of this transition is dependent on the coordination between solutions; here, renewable energy development and large-scale electrification. Further, other solutions can be adopted, such as energy efficiency, which might impact demand-side strategy outcomes. Moreover, the two territories present contrasting strucutures in terms of geography and challenges. For instance, reducing imports for Procida is in line with the energy autonomy of the island, while for Hinnøya, relying on imports has an important role in the decarbonization pathways of the island.

This study also examined building-interconnected energy systems and better integrated grid support in the framework of the EU’s clean energy transition [71]. Indeed, microgrids or weakly connected grids often encounter similar technical challenges to those found on islands [72]. This underlines the cross-regional lessons learnt from the two case studies. Another lesson pointed out when demand-side strategies were applied was the impact of these solutions on sectors such as agriculture, commerce, etc., where participation is sometimes difficult to establish, as it affects the quality of the products.

One part of this double-territory analysis was to determine recommendations for the replicability of the analysis in other territories. Replication would require producing an energy profile of the islands, the assessment of the main energy and climate challenges alongside their long-term objectives, for instance, in decarbonization, the territory’s implementation plans, and the potential roles and stakeholders of the participating sectors. It was also noted that some factors determine the employment of solutions, such as storage. This depends on the type of interconnection between the island and the mainland system, the amount of RES installed or foreseen, and the size of the island in terms of population. For the case of available interconnection, the choice of storage solution relies on information about the energy prices and the grid’s resilience, and the need to reduce the use of and dependency on conventional fuels. Another important aspect for relatively isolated islands is imports of electricity, which, in terms of decarbonization, remain the responsibility of the main country to reduce their emissions.

The demand-side strategies evoked in this paper are dependent on consumers’ engagement in three dimensions: investing in new and enabling technologies, enrolling in programs, and actively reducing/shifting loads [73]. As part of the climate mitigation solutions highlighted by the IPCC [74], demand in energy plays a key role in achieving the Paris Agreement. Demand-side strategies would incentivize consumers but require their participation for their success. For example, instead of participating by charging their vehicles, consumers might opt for public transportation. This remains in line with decarbonization strategies if proper policies are implemented. Nevertheless, behavior change studies are particularly beneficial for the transportation sector [75]. These studies would induce “behavioral realism in energy models” and capture the adoption of “novel technologies” through two main methods. The first involves endogenously representing modal choices and infrastructure availability, driving patterns or new mobility trends in the energy model. The second involves linking/coupling between models, which requires data exchanges. Representing consumers’ behavior would account for the preferences and incomes of consumers and actors and can be provided by general equilibrium models, such as GEM-E3 [76]. The monetary value of the strategies would also play an important role in their adoption. Models such as TIMES can be linked to lifestyle and macroeconomic models [77], which capture the overall energy mix and emissions output. Such linkings highlight the interdependencies between economy, technology, and lifestyle, but do not capture the impact of strategies (carbon tax, electricity pricing, etc.) on consumers’ reactions. Nevertheless, dynamic models, such as system dynamics (SD) and agent-based models (ABMs), can include feedback loops that cause endogenous changes in the system based on the decisions of the actors.

Finally, there were certain limitations evident in the results, stemming from data availability, which in turn impacted the selection of time slices in the TIMES models. Disaggregated data on the level of hours are helpful to follow the change in the load curve due to the applied strategies, but they increase the computation time. Moreover, real data that reflect the situation of the island enable a more specific judgement of the solutions (for instance, sector-specific data and seasonal demand variations). In addition, energy models often overlook non-economic dimensions, such as indirect emissions, critical raw materials, land availability issues, and impacts on land, water, and biodiversity, limiting their effectiveness in informing energy policy [78]. Implementing such factors affords more credibility to the results obtained from energy models. In fact, investment in energy projects in the EU sometimes receives social resistance, which delays their implementation. It is suggested to depict and endogenize factors such as regional density and the sizing of existing windfarms and make them drivers of the analysis. Concerning environmental factors, it is possible to manipulate life-cycle data in Python environments and thus obtain life-cycle assessment (LCA) calculations, as seen in the PREMISE model, which adapts to various future scenarios of renewable energy use and technological advancements. Other ways to improve policy evaluation include the agent-based technology adoption model (ATOM), which revealed the shortcomings of Greece’s net-metering policy, demonstrating its inability to meet future photovoltaic targets, a finding that would have remained undetected with traditional system optimization models. In [79], the design and implementation of Floating Photovoltaic (FPV) systems with storage was evaluated for three representative islands in Indonesia considering techno-economic–socio-environmental impacts. The research addressed the complex balance of benefits and challenges associated with FPV installations, highlighting the potential impacts on water conservation, algae control, and marine ecosystems. It also focused on understanding the social influence of floating photovoltaic systems, highlighting the economic activities, roles of men and women, and challenges faced due to the lack of electricity infrastructure.

6. Conclusions

Throughout this study, three types of enablers were the focal point for studying the integration of renewable energy and decarbonization: demand response, self-consumption, and control of electric vehicle charging. This strategy presents benefits for the flexibility of the energy system, an important factor for its operators when balancing supply and demand.

To address these issues and to handle the evolution of the two relatively isolated power systems studied, while minimizing the costs of investment and implementing political and environmental constraints, the bottom-up TIMES model generator approach was applied. This involved TIMES-Procida for the Italian island of Procida and TIMES-Hinnøya for the Norwegian island of Hinnøya. The prospective analysis highlighted some key points for consideration in planning the evolution of the energy systems of these two islands, in terms of investments and policies that align with their local plans. The findings can be used on a broader level, which identifies possible opportunities for replicability. For instance, the decarbonization of energy systems is facilitated by flexibility solutions and mechanisms in different end-use sectors, hence creating synergies between the sectors and finally allowing increased renewable energy integration. Electricity storage is effective in managing electricity supply, particularly with intermittent renewables, and can be applied across various power system sectors. Electric vehicles also offer advantages for renewable integration on islands, and to reduce pollution, as they have low emissions and enable flexibility from prosumers, though demand-side participation is currently limited but promising for reducing grid congestion and costly investments.

This study, based on long-term energy planning, also shed light on the local energy needs of islands, which require particular attention when it comes to strategically planning for their future development and monitoring their energy transition. The topics of energy and climate are usually handled at the national level but need to be addressed differently when it comes to islands. From the analysis, the characteristics requiring attention for such territories are the spatial and grid constraints, demographics, and policies.

Nevertheless, public willingness to participate in flexibility measures is a determining aspect for obtaining effective results. The presented conclusions would impact choices and encourage the active participation of the demand side and are relevant for all relatively isolated islands; however, practical on-the-ground methods are needed. For instance, in [80], it was proven that decision-making by end-users would rely on incentives and coordination with other stakeholders. In [81], solutions such as Virtual Power Plants would allow such coordination and valorization of the flexibility provided by the participants.