Vehicle-To-Grid for Peak Shaving to Unlock the Integration of Distributed Heat Pumps in a Swedish Neighborhood

: The city of Stockholm is close to hitting the capacity limits of its power grid. As an additional challenge, electricity has been identiﬁed as a key resource to help the city to meet its environmental targets. This has pushed citizens to prefer power-based technologies, like heat pumps and electric vehicles, thus endangering the stability of the grid. The focus of this paper is on the district of Hammarby Sjöstad. Here, plans are set to switch from district heating to heat pumps. A previous study veriﬁed that this choice will cause overloadings on the electricity distribution grid. The present paper tackles this problem by proposing a new energy storage option. By considering the increasing share of electric vehicles, the potential of using the electricity stored in their batteries to support the grid is explored through technical performance simulations. The objective was to enable a bi-directional ﬂow and use the electric vehicles’ (EVs)’ discharging to shave the peak demand caused by the heat pumps. It was found that this solution can eliminate overloadings up to 50%, with a 100% EV penetration. To overcome the mismatch between the availability of EVs and the overloadings’ occurrence, the minimum state of charge for discharging should be lower than 70%.


Introduction
Sweden has adopted the ambitious goal of becoming a zero carbon emission society by 2045 [1]. Within this context, the energy sector has attracted a special attention because it accounts for more than 70% (including transport) of total greenhouse gas emissions [2]. When looking at the internal electricity mix of this country [3,4], electricity is promoted as a clean resource and thus a promising option to achieve the 2045 target. As a consequence, in Stockholm, citizen-driven initiatives work to promote the installation of distributed residential heat pumps (HPs) and the adoption of electric vehicles (EVs) [5]. A real example is represented by Hammarby Sjöstad, which is a residential neighborhood constituted by multi-apartment buildings. This district is currently connected to the city's district heating (DH) network, but projects are already approved for some customers to switch to their own domestic HP.
On an opposite side, this electrification trend is challenged by a structural problem that impacts the power grid system in Stockholm. In fact, the electricity transmission to the city is close to its maximum power capacity. This means that the distribution system operators (DSOs) are being forced to impose power limits on the distribution grid [6]. Thus, the connection of new loads, like HPs and EVs, could soon be impossible. If capacity expansion investments are to be avoided or delayed, alternative solutions should be quickly found.
A promising option is offered by energy storage [7]. This technology can shift loads in time by charging when surplus energy supply is available and discharging during periods of peak demand.
The remainder of the paper is as follows. Firstly, the case study is presented in Section 2. Secondly, the two scenarios and the corresponding simulation models are described in Sections 3 and 4, respectively. Finally, the results are discussed in Section 5.

Case Study
The case study presented in this paper corresponded to a specific area in Hammarby Sjöstad, as shown in Figure 1. All the buildings in this area are connected to a single medium-to-low voltage transformer substation. The conceptual illustration of the case study area in Figure 1 sketches the current type of energy infrastructure in the neighborhood. The multi-apartment buildings are connected to a DH network for space heating and domestic hot water purposes. The installed capacity is about 1.2 MW for a total of 94,555 m 2 of heated area. Electricity is provided by a low voltage distribution grid (400 V) that covers a capacity of about 1.6 MW.
Energies 2020, 13, x FOR PEER REVIEW 3 of 13 The remainder of the paper is as follows. Firstly, the case study is presented in Section 2. Secondly, the two scenarios and the corresponding simulation models are described in Sections 3 and 4, respectively. Finally, the results are discussed in Section 5.

Case Study
The case study presented in this paper corresponded to a specific area in Hammarby Sjöstad, as shown in Figure 1. All the buildings in this area are connected to a single medium-to-low voltage transformer substation. The conceptual illustration of the case study area in Figure 1 sketches the current type of energy infrastructure in the neighborhood. The multi-apartment buildings are connected to a DH network for space heating and domestic hot water purposes. The installed capacity is about 1.2 MW for a total of 94,555 m 2 of heated area. Electricity is provided by a low voltage distribution grid (400 V) that covers a capacity of about 1.6 MW. Within the scope of the scenarios later described in Section 3, the buildings are grouped in 22 building blocks (BB) according to the energy declaration documents of the corresponding housing associations ([29]). As reported in Table 1, the BBs are identified through their heated area. The choice of a limited area was motivated by the focus of the study, which was to show the potential of peak power demand management by using V2G technology. This could be easily replicated to show the impact on the whole neighborhood (or beyond).

Reference Scenario
The reference scenario for the present work was taken from a previous study presented in [12]. Given the interest of a few housing associations (Section 1), the focus was to assess the technical Within the scope of the scenarios later described in Section 3, the buildings are grouped in 22 building blocks (BB) according to the energy declaration documents of the corresponding housing associations [29]. As reported in Table 1, the BBs are identified through their heated area. The choice of a limited area was motivated by the focus of the study, which was to show the potential of peak power demand management by using V2G technology. This could be easily replicated to show the impact on the whole neighborhood (or beyond).

Reference Scenario
The reference scenario for the present work was taken from a previous study presented in [12]. Given the interest of a few housing associations (Section 1), the focus was to assess the technical feasibility of replacing a DH network with domestic-distributed HPs. Since the sole installation of the distributed HPs was shown to cause overloadings on the electricity distribution grid, a demand response solution was further assessed. The thermal mass of the buildings was used as TES to alleviate the grid from the overloadings caused by the HPs. A positive contribution was shown with a reduction of the heat demand dependence from a DH of 6%, compared to the case without the thermal mass control. Furthermore, up to 50% fewer overloadings were detected at the HP level. However, it was also highlighted that a full disconnection from the DH network was not possible given the current infrastructure capacity. These outcomes [12] represented the reference scenario for the present study.

V2G Integration Scenario
In the present paper, the reference scenario was extended by assessing the potential of V2G technology as a way to further support the electricity distribution grid. A simplified illustration of this concept is shown in Figure 2. For the sake of simplicity, a BB is represented as one building, though it can also correspond to a group of buildings. The V2G charging/discharging station is located upstream of each HP installation. The EVs' charging/discharging patterns were based on stochastic profiles. These were generated according to the assumptions described in Section 4.
Energies 2020, 13, x FOR PEER REVIEW 4 of 13 feasibility of replacing a DH network with domestic-distributed HPs. Since the sole installation of the distributed HPs was shown to cause overloadings on the electricity distribution grid, a demand response solution was further assessed. The thermal mass of the buildings was used as TES to alleviate the grid from the overloadings caused by the HPs. A positive contribution was shown with a reduction of the heat demand dependence from a DH of 6%, compared to the case without the thermal mass control. Furthermore, up to 50% fewer overloadings were detected at the HP level. However, it was also highlighted that a full disconnection from the DH network was not possible given the current infrastructure capacity. These outcomes ( [12]) represented the reference scenario for the present study.

V2G Integration Scenario
In the present paper, the reference scenario was extended by assessing the potential of V2G technology as a way to further support the electricity distribution grid. A simplified illustration of this concept is shown in Figure 2. For the sake of simplicity, a BB is represented as one building, though it can also correspond to a group of buildings. The V2G charging/discharging station is located upstream of each HP installation. The EVs' charging/discharging patterns were based on stochastic profiles. These were generated according to the assumptions described in Section 4. The EVs' parking stations are located in the BBs, in line with the "Charge at Home" initiative (Section 1). In this scenario, as illustrated in Figure 2, both charging and discharging processes-and thus V2G-were enabled in each BB. The aim was to cover the grid overloadings by performing peak shaving with the electricity stored in the EVs' batteries. This study considered a 100% penetration of the EVs. Furthermore, the EVs' charging was assumed to happen overnight with no overloadings. More specific assumptions are described in Section 4.

Models
Different combinations of models were used to represent the two scenarios described in Section 2. These models are discussed in detail in the following sub-sections. The EVs' parking stations are located in the BBs, in line with the "Charge at Home" initiative (Section 1). In this scenario, as illustrated in Figure 2, both charging and discharging processes-and thus V2G-were enabled in each BB. The aim was to cover the grid overloadings by performing peak shaving with the electricity stored in the EVs' batteries. This study considered a 100% penetration of the EVs. Furthermore, the EVs' charging was assumed to happen overnight with no overloadings. More specific assumptions are described in Section 4.

Models
Different combinations of models were used to represent the two scenarios described in Section 2. These models are discussed in detail in the following sub-sections.

Models for the Reference Scenario
This sub-section briefly highlights that, from a modeling and simulation perspective, a co-simulation method [28,30] was used to assess the interaction between the loading of the electricity distribution grid, the heat supply from HPs and a DH network, and the indoor temperature of residential buildings. Figure 3 sums up the modeling tools that were used for each system involved in the scenarios. The scheme also shows the main heat (red lines), power flows (black lines), and control signals (blue dashed lines).
Energies 2020, 13, x FOR PEER REVIEW 5 of 13 This sub-section briefly highlights that, from a modeling and simulation perspective, a cosimulation method ( [28,30]) was used to assess the interaction between the loading of the electricity distribution grid, the heat supply from HPs and a DH network, and the indoor temperature of residential buildings. Figure 3 sums up the modeling tools that were used for each system involved in the scenarios. The scheme also shows the main heat (red lines), power flows (black lines), and control signals (blue dashed lines). The co-simulation environment allowed us to capture two feedback loop signals. On one hand, the loading status of the grid was checked to decide whether the HPs could be operated. An overloading signal was generated when 100% of the capacity was reached. On the other hand, the indoor temperature of the BBs was monitored to decide if the thermostat set point could be changed to load/unload the thermal mass. In this case, the indoor temperature was to be maintained within the range of 20 ± 0.5 °C.
The following models were implemented for this reference scenario: • A power flow model for the electricity distribution grid. An open source tool called Pandapower [31] was used for this purpose.  The co-simulation environment allowed us to capture two feedback loop signals. On one hand, the loading status of the grid was checked to decide whether the HPs could be operated. An overloading signal was generated when 100% of the capacity was reached. On the other hand, the indoor temperature of the BBs was monitored to decide if the thermostat set point could be changed to load/unload the thermal mass. In this case, the indoor temperature was to be maintained within the range of 20 ± 0.5 • C.
The following models were implemented for this reference scenario: • A power flow model for the electricity distribution grid. An open source tool called Pandapower [31] was used for this purpose. • A reduced order model for the thermal demand of the BBs. This model is part of the Modelica Buildings library [32], maintained by the International Building Performance Association (IBPSA). • A Python model based on a coefficient of performance formula [33] for the HPs.
• Python models based on energy balance equations for the DH, the TES, and the electrical backup units. • A Python in-house operation logic for the control of the heating supply.
These models were used within a one year simulation with an hourly time step. A detailed explanation, parameters, variables, and assumptions for all the models are provided in [12].

Models for the V2G Integration Scenario
Concerning the V2G integration scenarios, two models were combined: • The grid overloading power profiles generated as a result of the reference scenario for the critical cables. • Stochastic driving patterns for the EVs owned by the inhabitants of Hammarby Sjöstad, assuming a 100% penetration of this technology.
Regarding the first model, a critical day was selected (10th January) from the reference scenario, together with the identified critical cables. With reference to this day, the overloading power for each critical cable was obtained by subtracting the power corresponding to a 100% loading of each critical cable from the actual loading. For the hours that did not present overloadings, the overloading power was set to zero.
In this study, the analysis was extended to all cables, including the parallel ones. Figure 4 shows the one-line diagram of the electricity distribution grid of the case study. The four red-highlighted cables are the ones that were identified as critical in the reference scenario. Two of these cables corresponded to bundled parallel cables. The corresponding labels are listed as the following: Energies 2020, 13, x FOR PEER REVIEW 6 of 13 These models were used within a one year simulation with an hourly time step. A detailed explanation, parameters, variables, and assumptions for all the models are provided in [12].

Models for the V2G integration scenario
Concerning the V2G integration scenarios, two models were combined: • The grid overloading power profiles generated as a result of the reference scenario for the critical cables.

•
Stochastic driving patterns for the EVs owned by the inhabitants of Hammarby Sjöstad, assuming a 100% penetration of this technology. Regarding the first model, a critical day was selected (January 10 th ) from the reference scenario, together with the identified critical cables. With reference to this day, the overloading power for each critical cable was obtained by subtracting the power corresponding to a 100% loading of each critical cable from the actual loading. For the hours that did not present overloadings, the overloading power was set to zero.
In this study, the analysis was extended to all cables, including the parallel ones. Figure 4 shows the one-line diagram of the electricity distribution grid of the case study. The four red-highlighted cables are the ones that were identified as critical in the reference scenario. Two of these cables corresponded to bundled parallel cables. The corresponding labels are listed as the following: The stochastic driving patterns of the EVs were generated by means of an in-house, Pythonbased model. Table 2 summarizes the main inputs to the model, with the related references for the assumptions. These values were considered valid for the selected location. The number of cars per person referred to statistical data of a typical Stockholm neighborhood. To determine the density of The stochastic driving patterns of the EVs were generated by means of an in-house, Python-based model. Table 2 summarizes the main inputs to the model, with the related references for the assumptions. These values were considered valid for the selected location. The number of cars per person referred to statistical data of a typical Stockholm neighborhood. To determine the density of inhabitants, it was reasonably assumed that two people could live in 48 m 2 . The level of penetration of EVs was related to a future case, in line with the current trend of an increasing EV market share [34]. These three inputs were used by the model to determine the number of EVs per each BB. From further statistical data, it was possible to assume average driving kilometers and speeds. The state of charge (SoC) of an EV, which refers to the charge level of its battery, was constrained by charging and discharging limits. An EV had to be charged if its SoC was below 50%, while charging was prevented if its SoC was over 90%. This is mainly due to technical performance reasons of the battery. Discharging to the grid was enabled when an EV's SoC was over 80%. Since this represented a key parameter for the present study, a sensitivity analysis was performed for values between 70% and 90% of the minimum SoC for discharging. The number of charging stations for each BB and the number of trips per day referred to the context of a residential neighborhood. This was also in line with the initiative "Charge at Home," presented in Section 1. Thus, the two trips represented a way to work and a way back to home. The potential charging/discharging of an EV could happen after these two trips were completed each day.
Assumed initial hourly profiles for the share of cars starting a trip at a certain hour were taken as a further input to the model. These profiles were randomized using the parameters' ranges presented in Table 2.
Given all these inputs (Table 2), the in-house Python model performed three main steps: • The total number of vehicles was allocated to each BB proportionally to the number of inhabitants and to the number of cars per person.

•
The starting times for the cars' trips were randomized around an initial value defined per each hour of the day [36]. The "random" Python module was used for this purpose.

•
The daily trips for each car were simulated by assigning, on a random basis ("random" module), the travel distances, the speed boundaries, and the initial SoC.
The main outputs of the model were: • The number of cars per each station (or BB).

•
The share of each cars' brand per each station (or BB).

•
The starting and ending times of each trip per each car.

•
The end-of-the-day SoC per each car.

•
The available power hourly profile per each car per each station (or BB).
Finally, the EVs' available power profiles were combined with the overloading power profiles taken from the reference scenario. The objective was to estimate the potential of V2G to alleviate the grid overloadings caused by the installation of the distributed HPs. Thus, a peak power shaving strategy was implemented by using the electricity stored in the EV's batteries. This was done by overlapping the two hourly profiles along the selected critical day. In particular, a V2G discharging service was activated when the following combination of events occurred: • A potential overloading was detected on an upstream cable.

•
One or more EVs were parked with an SoC over 80% at a corresponding charging/ discharging station.
The validity of the results was bound to the assumption that the EV's driving patterns did not present large differences from one day to the other. This was considered reasonable for a residential district, like Hammarby Sjöstad.
Since it was expected that the assumption for the minimum SoC for discharging played a relevant role, a sensitivity analysis was conducted for minimum SoC values equal to 70%, 75%, 80%, 85%, and 90%.

Results
Within the scope of the present paper, the results are presented and discussed in Figure 4 and in the remainder of this section. Figure 4 highlights the cables that were found to be still critical within the reference scenario. This means that, despite the utilization of the thermal mass of the BBs as a TES device, not all the grid overloadings, caused by the HPs, could be compensated for [12]. This was mainly due to the constraints in terms of thermal mass capacity of each BB and indoor temperature comfort. The latter was assumed to 20 ± 0.5 • C. By introducing V2G technology, according to the models discussed in Section 4.2, the objective was to explore the technical potential of EVs' batteries to provide further capacity to the grid. In this way, it could be shown how V2G can help covering these remaining overloadings generated by the HPs' demand peak.
Concerning the V2G integration scenario, Table 3 shows the number of EVs that were assigned to each BB. This outcome was based on the parameters presented in Table 2, regarding the number of cars per person, the number of people per BB's heated square meters, and the share of EVs in the studied area. In terms of electricity discharging to the grid, the number of EVs reported in Table 3 shows the total availability of EVs when no constraints about trips and SoC were considered. However, within the V2G integration scenario, the EVs were firstly required to be parked at their BBs' stations after completing two daily trips (to and from work). Secondly, their SoC had to be above 80%. Based on these constraints, Figure 5 illustrates the EVs' availability when back from work, i.e., after the second daily trip. When looking at the number of EVs parked (top graph in Figure 5), it could be verified that the cars were driven back from work along the day, and, around 6 p.m., most of them were parked at home. However, since an SoC higher than 80% was also required, the actual, much lower, EVs' availability is shown in the bottom graph of Figure 5. For example, at the charging station of BB16, at 6 p.m., only 25% of the parked EVs could discharge electricity to the grid. Energies 2020, 13, x FOR PEER REVIEW 9 of 13 Figure 5. EVs' availability at each BB station during the selected critical day. Figure 6 illustrates the impact of the EVs' availability for V2G on the grid's overloadings. In this figure, the reference scenario (without V2G integration) and the V2G integration scenario are compared over the selected critical day. The comparison is done by showing the percentage overloading performance for each critical cable. The loading performance below 100% is not shown in order to set the focus on the overloadings only.  Figure 6 illustrates the impact of the EVs' availability for V2G on the grid's overloadings. In this figure, the reference scenario (without V2G integration) and the V2G integration scenario are compared over the selected critical day. The comparison is done by showing the percentage overloading performance for each critical cable. The loading performance below 100% is not shown in order to set the focus on the overloadings only.  Figure 6 illustrates the impact of the EVs' availability for V2G on the grid's overloadings. In this figure, the reference scenario (without V2G integration) and the V2G integration scenario are compared over the selected critical day. The comparison is done by showing the percentage overloading performance for each critical cable. The loading performance below 100% is not shown in order to set the focus on the overloadings only.  As it can be noticed in Figure 6, the overloadings were concentrated during the late afternoon and during the evening. This is reasonable since the studied area is mainly a residential one, so people are at work during the day hours.
The integration of V2G helped to alleviate the overloadings by shaving the peak demand generated by the HPs. All the cables, except Llugnw6 and Llugnw7, fully benefited from this new system with no overloadings left. The other cables remained critical, especially before 6 p.m., when most of the cars were still away from the parking stations. This means that, with a 100% level of EV penetration in the neighborhood, the discharge of electricity from EVs' batteries could cover the amplitudes of the critical power peaks. The time matching between overloadings and the EVs' availability remained a challenge.
From the perspective of the main objective of the case study, Figure 7 shows the relation between the cables and each BB and, thus, each distributed HP unit. Considering that this analysis was related to a critical day (representative according to the assumption in Section 4.2), it can be concluded that the integration of V2G has the potential to fully enable the installation of HPs in BB6, BB19, BB20, and BB21. Concerning the other BBs, further measures should be implemented. For example, if a grid upgrade is to be either avoided or delayed, a different V2G operation logic should be tested. The implementation of smart control devices based on forecasting methods is a suggestion for future work. As it can be noticed in Figure 6, the overloadings were concentrated during the late afternoon and during the evening. This is reasonable since the studied area is mainly a residential one, so people are at work during the day hours.
The integration of V2G helped to alleviate the overloadings by shaving the peak demand generated by the HPs. All the cables, except Llugnw6 and Llugnw7, fully benefited from this new system with no overloadings left. The other cables remained critical, especially before 6 p.m., when most of the cars were still away from the parking stations. This means that, with a 100% level of EV penetration in the neighborhood, the discharge of electricity from EVs' batteries could cover the amplitudes of the critical power peaks. The time matching between overloadings and the EVs' availability remained a challenge.
From the perspective of the main objective of the case study, Figure 7 shows the relation between the cables and each BB and, thus, each distributed HP unit. Considering that this analysis was related to a critical day (representative according to the assumption in Section 4.2), it can be concluded that the integration of V2G has the potential to fully enable the installation of HPs in BB6, BB19, BB20, and BB21. Concerning the other BBs, further measures should be implemented. For example, if a grid upgrade is to be either avoided or delayed, a different V2G operation logic should be tested. The implementation of smart control devices based on forecasting methods is a suggestion for future work.
Energies 2020, 13, x FOR PEER REVIEW 10 of 13 Figure 6. Cables overloading performance day before and after the integration of V2G.
As it can be noticed in Figure 6, the overloadings were concentrated during the late afternoon and during the evening. This is reasonable since the studied area is mainly a residential one, so people are at work during the day hours.
The integration of V2G helped to alleviate the overloadings by shaving the peak demand generated by the HPs. All the cables, except Llugnw6 and Llugnw7, fully benefited from this new system with no overloadings left. The other cables remained critical, especially before 6 p.m., when most of the cars were still away from the parking stations. This means that, with a 100% level of EV penetration in the neighborhood, the discharge of electricity from EVs' batteries could cover the amplitudes of the critical power peaks. The time matching between overloadings and the EVs' availability remained a challenge.
From the perspective of the main objective of the case study, Figure 7 shows the relation between the cables and each BB and, thus, each distributed HP unit. Considering that this analysis was related to a critical day (representative according to the assumption in Section 4.2), it can be concluded that the integration of V2G has the potential to fully enable the installation of HPs in BB6, BB19, BB20, and BB21. Concerning the other BBs, further measures should be implemented. For example, if a grid upgrade is to be either avoided or delayed, a different V2G operation logic should be tested. The implementation of smart control devices based on forecasting methods is a suggestion for future work. Finally, in order to assess the relevance of the assumption for the minimum SoC for discharging, a sensitivity analysis is presented for values equal to 70%, 75%, 80%, 85%, and 90%. It was found that the availability of cars with a minimum SoC of 70% allowed for the covering of all the overloadings during the selected critical day. However, this is expected to increase the following charging requirements, which should be further tested. On the opposite side, no cars were available when the minimum SoC was set to 90%, so none of the overloadings could be balanced. As an example, Figure 8 shows the results of this sensitivity analysis at 5 p.m. during the selected day for all the critical cables. At this specific hour, a minimum SoC of 75% already solved the overall criticality. However, the cable Llgnw7 was still overloaded by about 8% at 1 p.m. Finally, in order to assess the relevance of the assumption for the minimum SoC for discharging, a sensitivity analysis is presented for values equal to 70%, 75%, 80%, 85%, and 90%. It was found that the availability of cars with a minimum SoC of 70% allowed for the covering of all the overloadings during the selected critical day. However, this is expected to increase the following charging requirements, which should be further tested. On the opposite side, no cars were available when the minimum SoC was set to 90%, so none of the overloadings could be balanced. As an example, Figure  8 shows the results of this sensitivity analysis at 5 p.m. during the selected day for all the critical cables. At this specific hour, a minimum SoC of 75% already solved the overall criticality. However, the cable Llgnw7 was still overloaded by about 8% at 1 p.m. As a more general conclusion, it can be stated that the approach presented in this paper can play a relevant role as a decision support tool for city planners, energy utilities, and engaged citizens. A district-level perspective linked to a sector-coupling approach (heating, electricity, and transport) can unlock the implementation of innovative technologies like active thermal mass and V2G.

Conclusions
In this paper, a district level perspective was applied on an integrated energy infrastructure problem where the electricity, heating, and transport sectors were interconnected.
Hammarby Sjöstad, a residential neighborhood in Stockholm, was selected as a relevant case. A previous study assessed that the plan of installing distributed domestic HPs will overload the local electricity distribution grid. The criticality of this situation can be improved, but not be solved, by using the thermal mass of the buildings as TES.
In the present work, V2G was presented as a new solution to further support the grid. This technology enables a bi-directional flow of electricity between the EVs' batteries and the grid. The aim was to explore the potential of V2G to perform peak power demand shaving by discharging electricity to the grid. The objective was to compensate for the overloadings caused by the HPs.
The technical performance simulation over a representative critical day demonstrated that, with a 100% penetration of EVs, the available power for discharging could cover all the overloadings' amplitudes. However, because of the time mismatch between the cars' availability and the need for balancing, only some cables could be completely relieved. This means that only the buildings connected to these cables could install residential HPs.
This conclusion is strongly dependent on the minimum SoC set for discharging. The study was conducted with a minimum SoC value of 80%. It was further shown that lowering this parameter to As a more general conclusion, it can be stated that the approach presented in this paper can play a relevant role as a decision support tool for city planners, energy utilities, and engaged citizens. A district-level perspective linked to a sector-coupling approach (heating, electricity, and transport) can unlock the implementation of innovative technologies like active thermal mass and V2G.

Conclusions
In this paper, a district level perspective was applied on an integrated energy infrastructure problem where the electricity, heating, and transport sectors were interconnected.
Hammarby Sjöstad, a residential neighborhood in Stockholm, was selected as a relevant case. A previous study assessed that the plan of installing distributed domestic HPs will overload the local electricity distribution grid. The criticality of this situation can be improved, but not be solved, by using the thermal mass of the buildings as TES.
In the present work, V2G was presented as a new solution to further support the grid. This technology enables a bi-directional flow of electricity between the EVs' batteries and the grid. The aim was to explore the potential of V2G to perform peak power demand shaving by discharging electricity to the grid. The objective was to compensate for the overloadings caused by the HPs.
The technical performance simulation over a representative critical day demonstrated that, with a 100% penetration of EVs, the available power for discharging could cover all the overloadings' amplitudes. However, because of the time mismatch between the cars' availability and the need for balancing, only some cables could be completely relieved. This means that only the buildings connected to these cables could install residential HPs.
This conclusion is strongly dependent on the minimum SoC set for discharging. The study was conducted with a minimum SoC value of 80%. It was further shown that lowering this parameter to 70% could help solving the overall overloading problem. However, this is expected to have an impact on the following charging requirements, which should be further studied.
As future research, a model predictive control logic should be investigated in order to solve the time mismatch challenge. Furthermore, the validity of the results should be tested against potential daily EVs' driving patterns variations. A sensitivity analysis regarding the penetration level of EVs and their SoC boundaries for charging is also suggested.
Finally, the approach used in this study shows that city planners, energy utilities, and engaged citizen can benefit from taking a district-to-city level perspective on integrated energy systems.