Exploring trade-offs

A


Introduction
The energy transition has reached a new phase now that several renewable energy technologies are cost competitive.In the built environment, this leads to the adoption of rooftop PV, electric vehicles, and heat pumps.The critical challenge in this phase is integrating these bottom-up developments into the existing sociotechnical system, while simultaneously accelerating adoption.
In the built environment, this embedding is especially challenging.The heating, electricity, and mobility transitions here coincide on the same limited infrastructure.Furthermore, the built environment is characterized by bottom-up developments of individual homeowners and other stakeholders such as grid operators and energy companies.In other words, the energy transition in the built environment is a complex system shaped by multi-system and multi-scale interactions, where stakeholders and limited resources must be aligned to create a fair and desirable transition.This need for alignment calls for multi-scale governance and system orchestration [1,2].In addition, policymakers must plan local energy systems under deep uncertainty [3].The high infrastructure costs require long-term planning (often exceeding 30 years).A period in which energy costs (e.g., the war in Ukraine), technological development (e.g., PV efficiency), and social discourse (e.g., view towards nuclear energy) will be under constant and unexpected change.
Local and regional policymakers are key stakeholders within this multi-scale process [4].They know the local context [2], are responsible for local sustainability goals, and can act as a linking pin between different stakeholders [5][6][7].The importance of local policymakers in the energy transition has been highlighted by IRENA [8], and governments in Europe [7,[9][10][11] the U.K. [12], and the U.S.A. [13].In local energy system planning, municipalities often have a central role in system design [14][15][16].However, local policymakers often lack the capacity and know-how to adequately play the part [13,17].To overcome this gap Beauchampet & Walsh [11] emphasize the need for tools "because as a municipality, you don't have the money, knowledge, nor power to do it on your own." The body of knowledge on local energy system modeling is broad.From specific urban-energy planning tools [18][19][20] to flexible energy system models [21].These models provide valuable results and are technologically advanced.However, there is a gap between these scientific results and local energy planning [22].Chang et al., highlight key challenges of energy models in accessibility and perceived policy relevance and call for more "out-of-the-box" usability for a wider multiplicity of actors [23].Similarly, the UK Energy Research Center calls for more focus on how modeling outputs can be used to support local community engagement, planning and, decision-making [15].Süsser et al., state that closer collaboration between modelers and users is "imperative to truly improve models and unlock their full potential" and [24].
In this paper we address this gap by using a collaborative modeling approach [25] with local decision makers.This approach has been used before, from co-creation of scenarios [26] to high level system modeling [27].However, few studies use it throughout the entire modeling process, from model development to scenario analysis and communication of results, and with non-academic stakeholders [28].This paper adds a novel contribution to that by using collaborative modeling in the entire process, resulting in novel model analysis, focusing on policy implications, trade-offs, and no-regret decisions.This leads to the research question: What are key trade-offs in local energy system planning and how can we use decision support tools to explore these?
The aim of this paper is to investigate how these key trade-offs can be identified and communicated to policymakers using a decision support tool.This is done with a focus on heat transition plans in three neighborhoods in the Netherlands.The collaborative approach included local policymakers and grid operators in a participatory process from research design to interactive workshops with the tool.The tool includes multiple technological options, such as individual electric heat pumps and collective district heating.All potential system configurations have advantages and disadvantages regarding costs, emissions, and grid constraints.This paper investigates these trade-offs in transition pathways for renewable energy strategies in a multi-domain and multi-scale environment.Furthermore, it explores the potential of collaborative modeling to aid policymakers in dealing with the inherent system complexities that exist in the energy transition.
The following section provides insights into local Dutch energy transition programs and energy system planning literature.Section 3 highlights the potential of collaborative modeling for energy transition planning, and Section 4 describes the tool itself.The actual analysis of energy system trade-offs and the results follow this.

The role of local governments in the Dutch heat transition
The Dutch government has created a tangle of policy programs to accelerate the energy transition, each focusing on a different domain or scale level.See Table 1 for the main programs relevant to local renewable energy systems.Within this set of policy programs, the national government has given local governments a central role in the energy transition.They must bridge the gap between top-down national strategies and local implementation.This was done with the idea that local governments are more suited to embed new technologies in the local energy system, taking spatial planning, stakeholder alignment, including residents, and communal infrastructure into account [2,6,7] However, in practice most local governments lack the executive power, technical expertise, and data required to do this well [17].As a result, the transition has been slow.Local governments rely on external expertise, with little learning from other regions [29], and many heat transition projects are hampered by high costs, low feasibility, and a lack of public support [30,31].
This paper focuses on the policy program Transition Vision Heat (TVH) (see Table 1).The TVH obliges all municipalities to create visions for sustainable heating on a neighborhood level for all neighborhoods in the municipality.Key choices in these TVHs are a strategy towards home insulation and the choice between collective or individual heating systems.Collective heating is usually a district heating system in which further choices occur regarding temperature levels, storage, and flexibility.Individual heating systems are natural gas boilers and electric or hybrid heat pumps.Hybrid heat pumps result in less grid congestion but still require natural gas, compared to electric heat pumps.The last option is natural gas boilers running on green gas, but the availability of green gas is limited.
The other policy programs are included in Table 1 as we take a systems perspective in this paper.Choices in EV charge points affect the same grid as heat pumps, so charging and heating strategies affect each other.On a larger scale, infrastructure is required for heating sources like green gas, hydrogen, and electricity.Using a system perspective, integrating multiple scales and domains within heating strategies, the strategy becomes more robust and works towards improving the entire system instead of a single sub-system.
Currently, the Netherlands is still at the beginning of the heat transition.Fig. 1 shows the penetration of different residual heating systems in the Netherlands.The sector is dominated by individual natural gas boilers (82%), which has created societal barriers against collective solutions [33].Also, compared to other countries, the Netherlands has a high dependency on natural gas (see Fig. 2).As heating systems are always context-specific, the results of this paper will be somewhat different in other regions.However, the method and trade-offs explored in this paper also apply to the heating transition in other countries and

Table 1
Overview of governance programs on different scale levels in the Netherlands.Adapted from (Netbeheer Nederland, 2019).

Currently used tools in Dutch heat transition planning
Multiple tools exist to aid policymakers with energy system planning.The most used tools for heating strategies in the Netherlands include the Heat transition atlas, VESTA MAIS, CEGOIA, and the Energy Transition Model [14,35].VESTA MAIS has a central role, it was developed by PBL Netherlands Environmental Assessment Agency [36], the Dutch national institute for strategic policy analysis in the field of environment and spatial planning.VESTA MAIS is a techno-economic model developed to analyze national scenarios in the heating transition of the built environment.It has been used to calculate the effects of five different strategies for sustainability measures on a neighborhood level for all neighborhoods in the Netherlands.The analysis is seen as a starting point for policymakers but also receives critique as it does not acknowledge local complexity and does not consider an integral system perspective, including different scale levels and connected transitions such as electrification of transport, PV, and local grid congestion [37].Furthermore, it does not allow for an exploration of key trade-offs in a collaborative way.Instead, it presents societal costs per heating strategy, lacking the broader insights that policymakers require to make wellinformed decisions.In this paper, we present a model that does meet this need.The first step is a literature review of local energy system models, resulting in a list of model requirements.

Tool requirements from an energy system modeling perspective
Many researchers have classified key characteristics of local energy system planning models.Based on multiple reviews of local energy modeling tools, the following list of tool requirements has been made (see Table 2).

Include multiple energy carriers
A critical development is integrating and connecting all relevant energy carriers in an energy system.Especially Lund and others using the EnergyPLAN framework have been vocal about the importance of this [38][39][40][41].In the context of local smart energy systems, the relevant energy carriers are electricity, heat, (natural) gas, and fossil fuels for combustion engine cars.Important sector-couplings are the electrification of heat, mobility, and cooking and other smart heating solutions, such as low-temperature district heating based on a residual heat source coupled to heat pumps and heat storage.This multi-energy carrier integration covers both energy supply and demand.

Use heterogeneous energy demand and supply profiles
Urban energy models require heterogeneous demand profiles [42].Energy demand is getting more detailed and specific in literature [23].Especially when including demand side-management, or other technologies which include more usage behavior, heterogeneous demand is important [43,44].For example, to gain insight into the costs of insulation, suitable heating technologies, and grid impact of flexibility solutions, heterogeneous mobility patterns to simulate smart charging potential, and heterogeneous residential energy consumption behavior to simulate demand-side management strategies.

Define spatially explicit assets
Implementing renewables in an urban context requires detailed, spatially explicit models.Spatial context largely determines the options for local and regional energy system planning [45].The core requirement is to spatially define the heterogeneous energy demand and supply profiles and match them with electricity or heating grid calculations [23].Decision makers require even more detail as they must consider spatial planning and other decision parameters, such as home ownership [35,44].Two examples are: Issues regarding a lack of space in dense urban environments to place new energy infrastructure, and the likely transition pathways vary depending on house ownership (private, rental, or housing corporation) [35].

Define temporally explicit energy calculations
Temporally explicit models are required to calculate peak production and demand adequately.With more intermittent renewable systems and timing issues such as grid congestion and supply and demand imbalances, temporal explicitness is necessary.Most studies show hourly or even sub-hourly time scales [18,19].

Include energy distribution grids
Distribution grids are amongst the most pressing bottlenecks in the Dutch energy transition [46], creating urgency for modelers to analyze the grid impact of different renewable energy strategies.This includes electricity and heating grids and requires a connection between loads (e. g., charge points or households) and their location on the grid.

Include flexibility and storage
Flexibility and storage technologies are important methods to balance supply and demand [23,47] and alleviate grid congestion [48][49][50].Grid congestion can occur either from supply peaks with excess renewable power or demand peaks to charge an electric car or running a heat pump from the grid.Supply-side grid congestion can be dealt with by either curtailment, battery storage, or heat storage.Demand-side congestion can be dealt with by smart charging and other forms of demand-side management, battery storage, and heat storage, or choosing renewable energy strategies that have reduced grid demands at peak loads, such as hybrid heat pumps compared to electric heat pumps.Demand-side management, storage, and curtailment are usually considered in energy system models [22,47]; however, they typically focus on energy balance or national markets.More specific grid studies also focus on different strategies of how local flexibility can help reduce grid congestion [48,49].

Tool requirements from a policy perspective
Two approaches were used to identify requirements from a policy perspective.A literature study and collaborative modeling approach, as described in Section 3.This section highlights the results of both.The results align with each other and the requirements from the previous section on energy modeling.The most important requirements are multi-system interactions, multi-scale dynamics, and transition pathway dynamics.

Multi-system interactions
Multi-system interactions occur when multiple sociotechnical systems interact (Papachristos et al. [51]).Examples are the increasing interactions between the heating and the electricity system [33,52,53] and those between the mobility and the electricity system [53][54][55].Rosenbloom [56] argues that "multi-system interactions are central to sustainability transitions" and that "potential alignments and tensions amongst systems that could be leveraged to politically accelerate change." Also, in the stakeholder sessions, this requirement was clear.Although many governmental organizations and grid operators have single-domain departments (e.g., see the policy programs in Table 1), the representatives in these sessions wanted to remove this gap and work towards a complete system approach.This approach comes back in the scenario analysis, where the interaction effects of the heating, mobility, and electricity transition are evaluated.

Multi-scale dynamics
Multi-scale dynamics result from the interaction between local, regional, national, and international processes and affect local transition outcomes [57].Bottom-up diffusion results from the decision-making of individuals when they invest in renewable energy technologies.At the same time, national policies and large-scale infrastructure heavily influence this bottom-up effect.Local and regional policymakers must bridge this gap [6,58].
Furthermore, multi-scale dynamics are required to address the planning and distribution of scarce resources amongst neighboring regions, for example, in dividing energy sources such as residual heat, potential for wind and PV, and high voltage grid capacity.Regions and municipalities must align their energy planning to distribute these resources fairly.Hofbauer et al. [59] state that energy models largely overlook multi-level governance and, thus, multi-scale dynamics.When included, multi-scale models can provide insights and foster coordination across scales.
Multi-scale dynamics are eminent in electricity infrastructure.This is shaped by the interaction between distribution and transmission grids, or low, medium -and high voltage grids.In current industry practice, distribution and transmission grid planning is done independent of each

Table 2
Model requirements from review studies.
To the stakeholders in the collaborative process, this requirement was eminent.The province and municipalities saw this as a natural division of roles.The grid operator is inherently a multi-scale stakeholder, considering the grid structure with low, medium, and high voltage levels.In many current policy processes, multi-scale alignment is part of the process, yet no models provide detailed insight into this.In the results section, we included an analysis of different scale effects.

Transition pathway dynamics
Transition pathway dynamics show the non-linear development of transitions in and across sociotechnical systems.They are relevant in energy system planning as they indicate key components or bottlenecks towards a new energy system configuration [64,65].Transition studies highlight and examine an elaborate set of non-linear dynamics, such as regime inertia [66], but also, more recently, enabling conditions and positive tipping points [65,67,68].Tipping points or tipping dynamics are points where transition pathways are triggered into reinforcing feedback loops, accelerating the transition [69].On the other hand, transition dynamics can also lead to deceleration due to inertia, bottlenecks or lock-ins [69,70].The aforementioned multi-system and multi-scale dynamics can both be causes of these non-linear transition dynamics.
We translated these pathway dynamics to policy advice with noregret decisions, lock-ins, and enabling conditions.Policymakers see lockins as undesirable outcomes, while many renewable energy technologies require a long-term commitment to pay back infrastructure investments.In the scenario analysis, we identified bottlenecks in transition pathways to investigate lock-ins.These bottlenecks could be alleviated by enabling conditions or addressed at all times by no-regret decisions.

Method: A collaborative approach
The tool presented in this paper has been developed in collaboration with policymakers and representatives from the electricity grid operator (distribution system operators or DSO).It involved local policymakers from the Province of Noord-Brabant and three local municipalities (see Table 3 for an overview of the cooperative sessions).The goal of the tool was to aid policymakers in devising their local renewable energy transition pathway by increasing knowledge of crucial trade-offs and a system perspective.Two concepts have been applied to attain this goal.
1.A collaborative modeling approach to create the tool and model scenarios.And 2. A scenario analysis to highlight key trade-offs and transition dynamics.The collaborative approach is described in this section and reflected in the model and scenario development described in Sections 6 and 7.
Collaborative or participatory modeling supports stakeholder understanding, social learning, collaboration, and policy advice.[71][72][73] More specifically, in energy system planning, participatory modeling can help to create integrated knowledge and understand long-term systemic effects [27,72].In literature, participatory and collaborative modeling concepts are often used interchangeably.Basco-Carrera et al. [74] define collaborative modeling as integrating end users more in the modeling process, including co-design and co-decision-making in joint actions.According to this definition, collaborative modeling has been used in this paper.
The joint actions in co-design entailed the entire modeling process, aligning to participatory modeling frameworks in literature [73,75,76].This process started with a collaborative exploration of research questions.Then, sessions followed for feedback and revision during the modeling process and extensive feedback regarding usability in workshops (see Table 3).The original research questions were defined in collaboration with provincial policymakers and DSO experts.The research and modeling process entailed six monthly meetings with a core team consisting of two provincial policymakers and two DSO experts, two demonstration and review sessions with a broad group of experts (provincial, regional, and municipal policymakers and DSO experts), and two user tests with three different municipalities to evaluate the functionality and applicability of the tool in decision-making processes.After each session, feedback was gathered to enable and spur model development towards a more applicable tool.
The scenarios to explore trade-offs have also been developed using a collaborative approach.This analysis aimed at finding key trade-offs and increasing system understanding.The number of potential solutions, in other words, the design space in the model, is enormous.Every combination of heating supply, heating demand, and electricity solutions is possible.Optimization or pre-determined scenarios often do not co-align with existing ideas and local context.For example, policymakers described how the national modeling analysis done by the Dutch environmental assessment agency in Vesta MAIS (discussed in Section 2.2.) [37], a neighborhood was assigned suited for an electric heat pump strategy while the local policymakers where already planning a district heating grid in the area.
With this approach, the research questions, scope, and requirements are determined collaboratively in sessions with policymakers, and the resulting tool has been thoroughly tested with users.The resulting model criteria align with the criteria found in the literature review (see Section 2.4) and were further used in the model development (Section 4) and scenario analysis (Section 5).

Model description
The model is a GIS-based simulation model of the current energy system on two levels.The first is the neighborhood level, which is aimed at exploring renewable energy strategies on a neighborhood level.The second is the provincial level, which is aimed at testing whether these strategies work in a larger system context.This separation enables detailed calculations of flexibility in individual households while being able to scale up to the provincial level.Moreover, it enables a detailed simulation model and visual representation of both levels (see Fig. 3).Both models are described in more detail in the following section.Full model documentation can be read here [77], and the open-source models can be downloaded here [78].
The models are agent-based representations of all energy flows in a specific region.Agent-based modeling fits the model requirements as it is suitable for modeling detailed heterogeneous agents in multi-domain and multi-scale systems [79].The model consists of bottom-up modeled energy assets and agents, such as buildings, cars, and households.Heterogeneous instances of these objects or agents are created based on data from, for example, households, cars, people, and transformers.The neighborhood model contains data from three different neighborhoods: a rural village and urban districts from the 60s with and without an operational district heating grid.In this paper, the focus is on the urban neighborhood without district heating.The other neighborhoods were created to let the local policymakers interact with their location-specific characteristics in the participatory process.The neighborhoods are based on data from the central bureau of statistics [80], basic registration of buildings and addresses (BAG) [81], and the grid operator [82].They contain between 1201 and 2904 residence objects connected to 7 to 23 transformers.
The provincial model was used to check whether a strategy fits a larger system context.For this purpose, it required heterogeneous neighborhoods, each with its specific number of households, inhabitants, cars, and heating strategy.These neighborhoods were combined with data on larger supply and demand assets (industry, agriculture, wind turbines, and solar parks), which resulted in a model with 1939 neighborhoods, 62 municipalities, and 34 high-to-medium voltage transformer stations.
Heat demand is calculated based on energy labels, which are the Dutch implementation of the European Energy Performance of Buildings Directive required for all buildings that are sold or rented and made by independent energy advisors [83].Energy consumption depends on building type, year, size, and energy assets.The numbers used in this study originate from Vesta MAIS [84].Mobility demand is taken from the 'activity-based´Albatross model [85][86][87] and domestic hot water consumption profiles are taken from DHWcalc [88].Lastly, energy generation and demand is based on hourly weather profiles from the Royal Netherlands Meteorological Institute (KNMI) [89].
An overview of the data used can be found in Appendix A. Furthermore, all data sources can be downloaded except for the exact capacity of low-voltage transformers, which were shared under an NDA.An industry standard of 1.5 kW per connection is supplied as an alternative in the model if this data is missing.
All cost data is taken from other models used in Dutch policy making to align as much as possible to other models and results.An overview of costs data can be found in Appendix B and downloaded here.

Model structure and energy calculations
Energy demand and supply are calculated hourly at a household level.Energy demand is subdivided into room heating, domestic hot water, mobility, electricity, and cooking demand.The model is activitybased, which ensures that any change in appliances (e.g., gas stove to induction stove, conventional car to electric car) is a shift in energy carrier while maintaining behavioral characteristics.
The energy calculations are executed per agent for households, transformers, and the district heating grid on an hourly basis.See Fig. 4 Fig. 3. GUI of the models with neighborhood left and province right.for a model overview.Local demand is always matched by a combination of local supply (PV generation and renewable heat sources) and imported supply (natural gas, gasoline, and electricity).The energy sources are included in the costs.However, it is up to the user to create viable scenarios where the required resources (e.g., residual heat) do not exceed the potential.
Flexibility and storage are covered by the curtailment of PV systems, smart charging algorithm for electric vehicles and electric boilers, and home battery systems.PV systems will curtail if the aggregated feed-in power exceeds the connected low-voltage transformer limits on nominal capacity.Smart charging is implemented according to an algorithm aimed at alleviating grid congestion.If transformers exceed the maximum capacity, the charging session of connected EVs is delayed until the maximum capacity is not exceeded anymore.This system is not yet possible in current legislation but was modeled to show the potential of smart charging for grid congestion specifically.Home batteries are implemented with the scope of minimizing homeowners' electricity bills.This means charging based on excess energy and discharging to alleviate grid congestion.This is not an optimized charging schedule or aimed at maximizing homeowners' profits.Further exploring the effects of different charging strategies is up for future research.

Scenario settings in the graphical user interface
When using the model in the participatory process, the user can select a range of scenario settings.The settings are presented as the number of households with a specific application or technology.Changing this setting means an algorithm will select which household gets this technology.For example, if more electric heat pumps are selected, households with higher energy labels will get the heat pumps first.This aligns with existing examples of renewable energy technology adoption and ensures realistic costs and energy loss assumptions are considered.A full description of 'selection algorithms' behind scenario settings is described in the model documentation and a list of all possible input setting can be found in appendix C.
Next to different input settings to create scenarios, users can select different map areas to highlight different scale levels.In the neighborhood model, users can select individual households and medium-to-low transformers to see the effects of specific solutions for these individual objects.The same applies to neighborhoods and high-to-medium voltage transformers at the provincial level.

Key performance indicators
Results are shown in five key performance indicators (KPIs) on both an individual and neighborhood level (see Table 4).More detail per specific solution can be obtained from the user interface.These KPIs were chosen after extensive discussions in the collaborative sessions.Costs, emissions, and overloaded transformers clearly show key tradeoffs.Individual electrification solutions lead to low costs and emissions, but many overloaded transformers.Collective district heating solutions lead to less overloaded transformers but higher costs and more emissions.The percentages of renewable energy and self-consumption are included in the user interface of the model presented to the policymakers and grid operators, but not focused upon in the results section because they give relevant insights but are not goals themselves and can be misleading.For example, renewable heating fueled by local residual heat leads to very high shares of renewable energy generation and self-consumption, outweighing the effects of PV generation.In such a case, high shares of renewable energy and self-consumption could lead to suboptimal system designs.
The participatory process resulted in many more requested indicators.Every policymaker has their specific expertise and concerns, leading to a large and diverse set of secondary performance indicators.These indicators ranged from investment costs and payback periods per household to costs for stranded assets and above and underground spatial requirements.These, however, were deemed less important as only specific policymakers asked for them.

Scenario analysis
The scenario analysis is aimed at exploring key trade-offs and transition dynamics.These lead to an increased system understanding and the ability to make well-informed decisions in transition strategies.As described in the method section, the scenarios have been created collaboratively with policymakers.They have indicated an extensive list of relevant policy questions.These questions were then categorized according to relevant transition dynamics and ordered to add  complexity systematically.The resulting list (see Table 5) contains 1.Initial insights on end system designs.2. Adding interaction effects of coinciding transitions.3. Introducing time dynamics in transition pathways.4. Highlighting potential solutions or enabling conditions for the transition pathways' bottlenecks.And 5. Identifying no-regret options and lock-ins in these pathways.The scenarios that resulted from this structure are described in Table 5, and the numeric input is included in Appendix D. The scenarios are described based on the scenario goal (1-5 from Table 5) with sets of sub-scenarios for more specific analysis.Each scenario is run with all different heating methods, which are natural gas boilers, green gas boilers, all-electric air-source heat pumps, hybrid air-source heat pumps, a medium-temperature district heating grid (70 • C), and low-temperature district heating (30 • C).These options are selected as they align to earlier studied options discussed in Section 2.2 and [37].Scenario 1 start with end designs showing 100% renewable heating systems of the different heating methods and insulation levels included in the model.Insulation upgrades are medium to very good insulation corresponding to energy labels C, B, and A from [83,84].Scenario 2.1-2.4 show the interaction effects of all heating technologies when including 100% electric cooking, electric vehicles, PV, and combining these three.Scenarios 3.1 and 3.2 show steps in getting to these end scenarios by analyzing 25%-50%-75%-100% penetration of all technologies with (3.2) and without (3.1)electric vehicles.Scenario 4.1 show the enabling conditions, by investigating how smart charging and boiler controls can alleviate grid congestion.In scenario 4.2 home batteries are also added.
Run settings for all scenarios are based on one year of simulation in hourly time steps, with gas, heat, and electricity costs assumed at 2020 values and grid infrastructure data from 2022.The highlighted results show costs, grid impact, and emissions.Costs are in €/per household/ year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.
Apart from the scenario analysis two additional analysis have been done.First, the effects of scenarios on different geographical scales are analyzed using the provincial model.This answers policymakers' and electricity grid operators' questions on how strategies on a neighborhood level impact and interact with the larger energy system, such as high voltage grids and the availability of heating sources.Second, the model's sensitivity to and effect of varying energy prices has been analyzed, as these can be highly volatile and dependent on external shocks.In contradiction to investment costs, which can be predicted well [90,91], energy prices are sensitive to external events like COVID-19 or the war in Ukrain [92] and their effects should therefore be analyzed.

Results
The results show that orchestration and coordination by policymakers is required.An initial overview of the modeled scenarios is given in Fig. 5. Initial insights into the trade-offs that are explored throughout this chapter are already visible, natural gas boilers have high-emissions, heat pumps cause grid overloads, and district heating systems have the highest costs.
However, trade-offs between heating systems occur not just in terms of costs, emissions, and grid impact.They also lead to different transition pathways, require policy on enabling technologies, and impact different scale levels.A lack of coordination leads to an unjust transition as early adopters will invest in individual technologies, leading to congestion on communal electricity grids.Sustainable heating strategies require a balancing act between multiple energy carriers, investing in grid reinforcements where transition pathways will rapidly lead to congestion issues, and using collective heat sources and smart spatial planning to avoid or delay congestion.The results of the scenarios below highlight these policy requirements.

100% sustainable heating
The first set of scenarios shows the differences between 100% renewable heating solutions.Trade-offs between increasing insulation levels and the five different heating solutions are investigated.Each Fig. 5.An overview of all modeled scenarios.Ideal solutions are in the bottom-left corner.Initial trade-offs are clear, natural gas boilers have high-emissions, heat pumps cause grid overloads and district heating has the highest costs heating solution has its advantages and disadvantages.Table 6 highlights this in a heatmap where the KPIs costs, grid congestion, and CO2 emissions are plotted next to each other, visualizing the trade-offs.The table shows the results of 24 model runs, one for every heating solution and insulation level combination.The costs, grid impact, and emissions are displayed for each run, and the cells are colored based on their relative score (green being the cheapest and red most expensive).The different heating solutions score green in one of the indicators and red in the other, or an overall yellow, indicating that each has advantages and disadvantages.
On the supply side, natural gas boilers are cheap and do not result in grid congestion but are unsustainable.Green gas is a viable solution for this neighborhood.However, it is not available in large enough quantities to make this a viable strategy for all neighborhoods.Electric heat pumps result in high emission reductions but are moderately expensive and result in high levels of grid congestion.Hybrid heat pumps and district heating are comparable; both have moderate costs, moderate grid congestion, and moderate emission reductions.Low-temperature district heating leads to high emission reductions and moderate grid congestion.However, it is expensive in the modeled set-up due to high infrastructure and in-house costs.Furthermore, district heating is often posed as a solution to grid congestion.However, it can still lead to congestion due to the increased electricity demand for domestic hot water in auxiliary boilers.
On the demand side, refurbishing the household to medium insulation (label C) is cost reducing in most scenarios.Refurbishing until good insulation (label B) is still a sound emission reduction for moderate costs.However, the effectiveness disappears towards very good insulation (label A) with limited energy savings and much higher costs.

Interaction effects of rooftop PV, electric cooking, and EVs
The second set of scenarios shows the interaction effects of 100% renewable heating solutions with 100% rooftop PV, electric cooking, and electric vehicles.These scenarios explore how much grid congestion caused by electrified heating systems interferes with other renewable energy technologies.The technologies are compared individually in scenarios 2.1-2.3 (see Appendix D).Scenario 2.4 (Table 7) combines these interaction dynamics.
PV has a positive effect as it reduces costs in all scenarios.In this case, PV curtails if voltage issues occur and, therefore, does not increase grid congestion at a transformer level in the model.Electric cooking has limited effects; it slightly increases costs and decreases emissions.Furthermore, it is, of course, required to substitute or remove existing natural gas infrastructure.Lastly, electric vehicles have a tremendous effect on grid congestion.Emissions are reduced substantially at limited costs, but all scenarios show almost all transformers being overloaded, independently of any heating strategy.For policymakers, this means two things: They must manage electric vehicle charging (smart charging, dedicated charging plazas, etc.).And two, if these measures do not alleviate congestion enough, additional electricity demand for electric vehicles requires reinforcing low-voltage grids.The additional reinforcement for an electric heating system should be considered coincidingly, depending on timing and the pace of electrification of heating, cars, and grid reinforcements.

Transition pathways
The third set of scenarios shows transition pathways towards 100% renewable energy systems.These scenarios explore when bottlenecks like grid congestion occur.Analyzing bottlenecks helps to identify which technologies allow for higher emission reductions before reaching these bottlenecks and to see if autonomous adoption patterns (adoption without any policy interference) are heading towards the identified bottlenecks.In these scenarios, the pathway towards the fully renewable energy system from scenario two (Section 6.2) is outlined by calculating adoption percentages of 10%, 25%, 50%, 75%, and 100% for heating methods, PV, electric cooking, and electric vehicles.These pathways are calculated with (Table 8) and without (Table 9) electric vehicles.
Autonomous adoption patterns will lead to grid congestion, especially when including electric vehicles.A penetration grade of 25% already leads to the first overloaded transformers, while at 50%, the entire grid overloads.This happens regardless of any choices in heating systems; however, electric heating systems will strongly amplify the effects.This step's second set of scenarios (Table 9) shows the transition pathways while disregarding electric cars.This is done to explore scenarios in which the dominance of EV charging is mitigated by, for example, smart charging or centrally located charging plazas with a separate grid connection.In this case, the design choices of heating systems to alleviate grid congestion are more relevant.Hybrid heat pumps are a perfect example, allowing for very high penetrations (75%) of heating at minimum grid impact (1 overloaded transformer).At similar levels, electric heat pumps and medium and low district heating systems have far higher grid impacts.Policymakers could use hybrid heat pumps as a 'transition technology.' Enabling immediate action and strongly reducing emissions while providing the grid operator time to enable the far more sustainable solution of electric heat pumps.

Enabling conditions
The fourth scenario set shows enabling conditions.The key bottleneck considered is grid congestion.Dealing with grid congestion can be done by adjusting the renewable technology mix (e.g., stimulating hybrid heat pumps instead of electric ones) or by adding technologies to alleviate grid constraints caused by the newly introduced sustainability technologies.The first method is viable for heating; however, electric vehicles are becoming the dominant technology for personal transportation.In that case, a policymaker could either try to take the technology out of the system and create a dedicated system by, for example, charging plazas.Alternatively, they could use smart charging and batteries to alleviate this congestion.These two technologies are implanted Costs in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kilotonne per year.
N. Loomans and F. Alkemade in scenarios 4.1 (Table 10) and 4.2 (Table 11).Comparing scenario 4.1 (Table 10) to scenario 3.1 (Table 8) shows that demand-side management is crucial for alleviating congestion.Therefore, to enable current autonomous adoption trends towards higher levels of sustainability, policymakers should rapidly roll out policies to promote smart charging to reduce grid congestion.Note that different forms of smart charging exist.In this case, the grid operator can reduce charging power to delay charging.Other forms of smart charging could have less or even adverse effects on congestion management.
The second method to deal with grid congestion tested in the model is home batteries.Batteries are often mentioned as a promising technology in renewable energy systems to align supply and demand and alleviate congestion.In the model, batteries charge according to current energy prices.The result is no effect on congestion as there is no price   Costs are in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.Costs are in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.
N. Loomans and F. Alkemade incentive to let the batteries act upon this.It does result in more selfconsumption, leading to a slight decrease in emissions, but at very high costs.In order to make batteries more cost-efficient and also able to alleviate grid congestion, other pricing schemes or tariff structures are required.

No-regrets and lock-ins
Policymakers indicated they are most interested in no-regret solutions and lock-ins.As stated before, the most crucial no-regret solution is ensuring that electric vehicle charging can be done within grid limits.No-regret solutions are less evident for heating systems, as each has advantages and disadvantages.However, the transition dynamics of the different solutions are very different.Collective solutions like district heating require strong and rapid policy action, while individual solutions are prone to autonomous (no-policy) adoption behavior.Furthermore, collective solutions require large shares of inhabitants to participate, as they scale less well.These dynamics mean that autonomous adoption leads to lock-ins.As more and more people buy individual heat pumps, the district heating system becomes less advantageous.However, the autonomous adoption trends over the past years (Fig. 6) show a clear preference towards electric heat pumps, which is the strategy that will most rapidly lead to the congestion bottleneck.This indicates that the heating transition is moving towards a tipping point, slowing down heat pump uptake if the grid bottleneck is not addressed.
Other no-regret solutions include good insulation (label B) and demand-side management of electric boilers.The effects of insulation differ based on the chosen heating solution's investment and variable costs.This can lead to conflicts of interest and suboptimal system design [93].So, when calculating the business case of the district heating system operator, it is essential to take insulation upgrades into account.This is reflected in model results where medium insulation (label C) leads to cost declines for individual solutions and to cost increases for district heating solutions.In any case, good insulation (label B) is still cost-effective, while above, the costs rise drastically while the energy consumption reductions do not.These results are confirmed in the VestaMAIS model [36].
The second no-regret of managing demand from electric boilers applies to all large, partially synchronized new consumption.Policymakers also noted other forms of electric heating (infrared heating, electric heaters, and electric domestic hot water heaters), which all have an efficiency of 1 (compared to 2.5-4.5 of heat pumps) and thus lead to high loads.If these cannot be managed through demand-side management, they will rapidly lead to congestion.

Scale level effects
The last set of policy questions shows the effects of the transition at different scale levels.With neighborhoods at the central level, there are two important effects.The first is to the individual household, building, and street level, where heterogeneous households can have significant local effects.In one of the modeled neighborhoods, certain streets and buildings are much better insulated than others, causing all the initial heat pumps at low percentages of heat pump penetration to be concentrated within this one building or street.This leads to one rapidly overloaded transformer, while all others have no additional load yet.In transition planning, this is important, as the grid operator can often tackle specific bottlenecks but cannot reinforce the entire grid.Immediately upgrading this single transformer enables a much larger share of Costs are in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.
Fig. 6.Renewable residential heat systems in the Netherlands.The other 82% are natural gas boilers.Data retrieved from [32].
N. Loomans and F. Alkemade electric heating and mobility throughout the neighborhood and buys time for the grid operator to reinforce all other transformers later.
A similar effect occurs on the high voltage grid scale (see Fig. 7 right picture), where a high share of neighborhoods suited for heat pumps within the distribution area of a single high to medium voltage transformer determines at which percentage of renewable heating the transformer will get overloaded.Policymakers can use this tool to plan which neighborhoods to address first within grid limits and to focus on a balance in heat sources (electricity, green gas, residual heat, or other district heating sources) within the geographical area served by one transformer.

Energy prices cost assumptions
Three different cost assumptions were compared to show the effect of different energy costs: Low fuel costs from 2019; Peak fuel and electricity costs due to the Ukrainian war in 2022 which were over three times as high as 2020 and projected 2030 values; And medium long-term projected costs from 2030 [94], where electricity tariffs are a combination of still high fuel costs and increased low-cost renewables (See Appendix B for exact values).
Clear conclusions can be drawn when comparing different fuel and electricity costs (Fig. 8 or Appendix E for more detailed scenarios).The rising costs of natural gas and other fuels have amplified the electricity costs.This makes electric heat pumps more expensive than district heating with 2022 energy cost assumptions compared to 2020 or 2030 cost assumptions.The second interesting finding is that natural gas is the only fuel source for which a cost increase is expected when comparing 2030 to 2022.This is reflected in the natural gas boiler and hybrid heat pump solutions.Lastly, these results could be further evaluated when adding a CO2 price, which improves the relative costs of lowtemperature district heating and electric heat pumps.

Discussion and limitations
Quantified models are required to guide decision-makers when developing local energy strategies.The models give insight into complex problems and system interactions.However, these models are often difficult to comprehend and implement in decision-making processes.The impact of using models in decision-making largely depends on the user's level of knowledge and understanding and how they deal with assumptions and uncertainty.Collaborative modeling is an auspicious method to increase this understanding, thus improving the usage of models in decision-making processes.However, it is time-consuming and challenging to transfer the right level of knowledge and correctly align the model to the wide variety of trade-offs a policymaker must consider.Their frame of reference usually extends far beyond the modeled scope, ranging from underground space allocation of new electricity cables to socio-demographic inequality in a particular neighborhood.
Tool effectiveness was assessed based on discussions and workshops with policymakers.In these sessions, the scenarios were not just represented as in this paper, but explored in workshops where the policymakers could interactively create scenarios.All decision-makers describe this as insightful and engaging.The principal contribution is an insight into transition dynamics and trade-offs in energy transition strategies.The most important features are geographical representation, interactive interface, and structured analysis.The lessons learned from the workshop and interaction with the tool help policymakers discuss strategies with grid operators, housing cooperations, and energy communities.
Furthermore, the collaborative workshops were essential to working with the model.At the start of the project, individual usage of the tool was also foreseen.However, it proved difficult for policymakers to comprehend the tool and its results adequately without guidance in a workshop.Of course, further development of the user interface could partially solve this, for example, towards a serious game [95].On the other hand, it also proves the educational power of a collaborative approach, where the inherent complexity and uncertainty of energy strategy planning can be adequately addressed by investigating scenarios with no-regrets and lock-ins and analyzing the robustness of results when varying key parameters such as energy costs in sensitivity analysis.
Lastly, the tool requires further improvement in energy system modeling with technology adoption and human behavior.The clear next steps are including more detail in different district heating system designs, introducing more demand-side management technologies, and including different tariff schemes or other incentives individuals can use to adapt their energy consumption.Apart from increasing the number and detail of included technologies, allocating costs and benefits per stakeholder (households, grid operators, energy suppliers, energy community) and actual business cases or payback periods per actor would increase the usefulness of the model.Although this would increase the required number of simulations, no immediate bottlenecks are foreseen, on one hand researchers could more carefully address which scenarios they want to explore, on the other hand there is still a lot of room for performance optimization and increased computing capacity.The agent-based modeling method enables relatively easy integration of these factors in future research.With the potential to include socioeconomic factors, ultimately leading to accurate sociotechnical energy Fig. 7. Visualization of multi-scale impacts.In the left image of a neighborhood, the grid is impacted by a single transformer, as most heat pumps are located within a single building and street.The right picture depicts a similar phenomenon on a high voltage scale where neighborhoods suited for heat pumps lead to congested high voltage transformers while neighborhoods with district heating or gas heating do not.
transition models [43] to support decision-making.

Conclusion
The transition towards renewable local energy systems in the built environment is a balancing act for local policymakers.They must stimulate the bottom-up adoption of individual technologies like PV panels, electric vehicles, heat pumps, and insulation.However, they must do this while ensuring a fair division of collective resources.Current adoption trends of individual technologies should be managed for two reasons.Firstly, grid congestion will lead to only the early adopters being able to benefit from these technologies.Second, the adoption of individual technologies leads to suboptimal lock-ins as collective technologies become less favorable if more and more inhabitants have already opted for the individual solution.
Our modeling exercise demonstrates these trade-offs.Each of the renewable heating technologies has its advantages and disadvantages, making it hard for policymakers to pick clear winners: green gas is hardly available; electric heat pumps lead to rapid grid congestion; hybrid heat pumps alleviate this but have fewer emission reductions; medium-temperature district heating has similar characteristics, and low-temperature district heating is the most expensive solution.Clear no-regrets are insulation, at least to medium insulation (label C) but preferably good (label B), depending on housing type, and demand-side management of electric vehicles and boilers.These effects are highlighted in the neighborhood analyzed in this paper and replicated in all other neighborhoods evaluated in the tool.Label B as cost-optimum for insulation is seen across all building types and ages, and all neighborhoods show grid overloads without smart charging and boiler controls, where the severity of these overloads depends on the grid itself and the density of cars and electric heating in a grid.Furthermore, the tool allows for detailed simulation models from which policymakers can identify and address bottlenecks ranging from high-voltage grids to individual streets and buildings.
Energy system tools are essential for policymakers and energy system planners to understand these trade-offs and transition dynamics.The collaborative modeling approach and interactively using the model in workshops greatly enhances this system understanding.This helps policymakers in their balancing act to align all system stakeholders, multiscale, and multi-domain dynamics, ensuring a fast and fair transition for all their citizens.

Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Naud Loomans reports financial support was provided by Dutch Research Council.Naud Loomans reports a relationship with Zenmo simulations that includes: employment.The project was partially executed by Naud Loomans as employee of Zenmo simulations commisioned by the Province of Noord-Brabant and grid operator Enexis, who were stakeholders in the process as described in the article.If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix B. Cost assumptions
Cost assumptions are shown in Table 13 (costs per energy carrier), Table 14 (costs per heating system per household) and Table 15 (costs for a district heating system).Costs for insulation are gathered in a large pivot table depending on building type, building year and level of retrofitting from one energy label to the next.This table can be downloaded here and found in the supplementary material.All cost assumptions are based on models used in Dutch policy making, to maintain as close as possible to existing frameworks and other model results.

Table 13
Cost assumption per energy carrier for households.*ODE ('increment renewable energy") is an additional tax.

Fig. 2 .
Fig. 2. Share per energy carrier in energy consumption in the residential sector 2021.Retrieved from [34].

Fig. 4 .
Fig. 4. Flow diagram of neighborhood model.Flexible supply and demand sources are controlled by the HEMS (Household Energy Management System).Households are a heterogenous population of different households connected to their respective low-voltage transformer and if required a district heating grid.

Table 3
Overview of cooperative sessions.
* Depends on the session and municipality.N.Loomans and F. Alkemade

Table 4
List of key performance indicators as used in the model.An * indicates KPIs are included in model user interface but not used in scenario analysis.

Table 5
Scenarios 1-5 with a brief scenario description and the policy questions that can be answered with these scenarios.For a detailed numeric description of these scenarios and sub.

Table 6
Trade-offs in scenario set 1 -The results of 20 model runs for energy systems with different heating solutions and insulation levels.Results per scenario show the costs, grid impact, and emissions.Colors indicate the relative score in comparison to other model runs.

Table 7
Trade-offs in scenario set 2 -Interaction effects in renewable energy systems.Costs are in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.

Table 8
Trade-offs in scenario set 3.1 -Transition pathways.Costs in €/per household/year, the grid is the number of overloaded LV-transformers, and CO2 is the total neighborhood emissions in kt per year.

Table 9
Trade-offs in scenario set 3.2 -Transition pathways without EVs.

Table 10
Trade-offs in scenario set 4.1 -Transition pathways with demand-side management.

Table 11
Trade-offs in scenario set 4.2 -Transition pathways with demand-side management and batteries.

Table 12
Data sources and accessibility.

Table 14
Costs household heat method.

Table 15
District heating costs.Sets insulation level to a minimum requirement (e.g., setting at C means all households below C would get insulation measures until level C) Heating supply Natural gas boilers 0 -max number of households Households with a natural gas boiler Green gas boilers 0 -max number of households Households with a green gas boiler Electric (air-source) heat pump 0 -max number of households Households with an electric heat pump