A challenging transformation is set to take place on Sweden’s energy infrastructure over the next 25 years as the country moves toward achieving its energy policy linked to global efforts on reversing the forecasted climate change. Sweden has set objectives to reduce emissions from transport by 70% by 2030, have 100% fossil-free generation by 2040 and release zero net greenhouse gases into the atmosphere by 2045. Achieving these goals presents a huge challenge considering the current energy and transport infrastructure, especially in urban areas.
Currently, the Stockholm region is facing capacity limitations as Svenska Kraftnät (SvK), the transmission system operator (TSO), has notified Ellevio AB, the local distribution network operator (DNO) in the city of Stockholm, that there is an upper power availability limit of 1.5 GW until 2030.
With the limit on the available power from SvK in mind, there was a need to find out whether the existing distribution network had sufficient surplus capacity to supply the increase in demand likely to be imposed by electric vehicles and heat pumps. Therefore, Ellevio AB took part in a research project that had been awarded to KTH Royal Institute of Technology. KTH and Ellevio AB collaborated within the framework of the research project, called “Smart and Robust Electricity Infrastructure for the Future.”
The Study
The selected area of study on the distribution network focused on Hammarby Sjöstad, a southeast neighborhood of Stockholm. Hammarby Sjöstad has been recognized globally as a highly successful urban renewal project for its approach on integrating sustainability into city planning. Today, about 70% of the area is developed. Based on estimated completion data, it will have 12,000 residential units with 31,000 inhabitants and employ some 10,000 people.
As a result of its success story, the neighborhood is used as a demonstration platform for a number of test beds in smart energy, sustainable transport and sharing economy. This makes Hammarby Sjöstad a natural fit for being an early adopter of various distributed energy resources (DERs), thus making it an ideal development and test case for Ellevio’s study.
The citizen-driven initiative ElectriCITY has been established for some years within the area to engage the tenant-owner associations and residents in developing concrete energy solutions on climate issues. Two projects to highlight from ElectriCITY are the Charge at Home and Energy at Home initiatives.
The former project makes use of policy incentives to install a charging infrastructure in households with the intended outcome of motivating residents to buy plug-in electric vehicles (PEVs). The latter project focuses on energy consumption reductions and related savings for the housing associations. As a result, an increasing number of customers have decided to disconnect from the traditional heating network to take advantage of the potential economic savings offered by the installation of residential heat pumps (HPs) in their buildings.
The Network Model
The Hammarby Sjöstad network consists of nine 12/0.4-kV distribution substations supplied by seven 12-kV underground cables routed from the existing 30/12-kV Hammarby Sjöstad primary substation. The distribution substations were each categorized based on the predominance of connected customer types as six residential (RS1 to RS6), one commercial (CM1) and two mixed (MX1 and MX2) substations. The 12/0.4-kV transformer capacity installed in each of the nine distribution substations is as follows:
- Two 800 kVA at RS1, RS4, RS5 and RS6
- Two 1000 kVA at RS2 and RS3
- Two 800 kVA at CM1
- Two 800 kVA at MX1 and MX2.
A model of the distribution grid in Hammarby Sjöstad was developed using a Python-based tool for load-flow analysis of networks. To build a relevant representation of its network, Ellevio provided data on transformer ratings, low-voltage underground cable properties, connected customer characteristics, load-flow measurements, maximum transformer loading and system topology. The percentage of available transformer capacity with respect to the peak load demand for each type of substation within the area corresponded to 21.1% for the mixed substations, 10.5% for commercial and 68.4% for residential.
Based on an analysis of hourly loads on the network provided by the DNO, six critical hours were chosen. Then the critically loaded hours were selected based on high overall demand on the medium-voltage (MV) feeder (global maximum) or peak loads on the particular substation (local maximum). These occurred during the months of January, September and December.
Maximum Load Scenarios
The study calculated the maximum level of penetration of additional loads the distribution network could supply without any reinforcement and violation of voltage limits for all MV/low-voltage (LV) substations and for each critical hour identified. To quantify the additional load in terms of numbers of PEVs and HPs, the following assumptions were made:
- PEV charging was assumed to be performed from a standard single-phase 230-V, 16-A outlet, with the charger having a power factor of 0.95.
- HPs had an average annual coefficient of performance of 3.2 reference to a building with an area requiring 170 kWth of heating capacity (space heating and water heating).
Uncontrolled Loads
Using the critical loads as a starting point, the first assessment was made around the critical situation of simultaneous loading. In essence, how many DERs — that is, PEVs or HPs — could be connected in a critical hour simultaneously? This had to be acceptable without resulting in any overloaded underground cables and transformers or violating statutory voltage limits.
The results confirmed the maximum number of PEVs or HPs that could be connected varied according to the existing load on the distribution substation. However, overall, a range of 1793 PEVs to 2496 PEVs or 120 HPs to 170 HPs could be connected to the network at the same time.
Load Management
This assessment provided a benchmark to understand an extreme case of uncontrolled and simultaneous loads, but no consideration was given to behavioral assumptions or load management strategies. Therefore, a comparison was undertaken to determine the level of integration of PEVs and HPs that could be achieved when managing and controlling these loads.
Given the predominance of residential substations in the Hammarby Sjöstad network and the similarities in load characteristics on each substation, six of the nine substations were included in the study. Among the critical hours studied so far, the daily load profile of Jan. 24, 2016, was selected for daily analysis. Based on the two ongoing Charge at Home and Energy at Home projects in the Hammarby Sjöstad neighborhood, daily load profiles were generated for vehicle charging and household heating.
When generating these profiles, two scenarios were tested. The first scenario assumed the number of HPs and PEVs connected were the same as those calculated in the first assessment, where the assumption was the loads were behaving in a business-as-usual (uncontrolled) manner. The second scenario consisted of controlling the loads based on price improvements while, at the same time, increasing the level of penetration of the loads. For this scenario it was assumed customers in the network would be linked directly to electricity prices in the Nord Pool.
For each type of load, the scenarios were compared regarding increased levels of penetration, system losses and costs. This was done by running the model of the distribution network with the different profile scenarios and verifying no limits were surpassed. As the load profile behavior for PEVs and HPs differ, separate assumptions had to be made for each case.
PEV Load Management
In the uncontrolled PEV charging profile, charging takes place once the vehicle is connected to the power outlet, which in this case is assumed to be in the early evening after residential customers have returned home. Considering an average daily journey of 36 km (22.38 miles) over 44 minutes, this comes to an energy consumption of 0.17 kWh/km (0.27 kWh/mile) with a charging efficiency of 88%. A two-hour charging period was assumed for all PEVs.
To generate the price-controlled profile, the same charging period was used as in the uncontrolled scenario. However, the charging period was shifted to take place when the electricity price was minimal. Because this charging period coincided with when the load on the distribution network was low, the number of vehicles connected to each substation could be increased in this scenario compared to the uncontrolled case.
The feasibility and impact of these two profiles on the distribution network, including an evaluation of the system losses and charging costs, were evaluated in the power flow model.
For PEVs, it was possible to increase the number of vehicles by a factor of 3, but this entailed the price-controlled scenario having 144.6% more system losses and 134.4% higher costs than charging in the uncontrolled scenario. With respect to the total energy consumed, the losses corresponded to 0.11% and 0.24% in the uncontrolled and controlled cases, respectively. With respect to the total electricity costs, the charging costs corresponded to 5.27% in the uncontrolled and 11.54% in the controlled. The specific charging costs per vehicle was €0.18 (US$0.198) in the uncontrolled case and €0.14 (US$0.154) in the price-controlled case.
HP Load Management
In the uncontrolled HP profile, the heating load follows the total load profile for the residential substations. As the number of HPs installed for each substation on the distribution network corresponds to the peak hour demand, the number of pumps connected simultaneously was scaled linearly with the base load for each hour.
The price-controlled profile for the HP loads was based on the principle of load shedding. During the peak demand hours, the HP loads were disconnected, but to counterbalance this and maintain a comfortable temperature in the spaces being heated, loads were increased in the hours prior to the peak load when the price for electricity was cheaper. Because the load on the distribution network is lower during these hours, the number of HPs connected to the network can be increased.
The impact of HPs on the distribution network, system losses and heating costs was evaluated in the power flow model.
For HPs, it was possible to increase the number of connected pumps by a factor of 1.3, while the price-controlled scenario had 8.5% less system losses and 24% less costs than in the uncontrolled scenario. With respect to the total energy consumed, the losses corresponded to 0.61% and 0.56% in the uncontrolled and controlled cases, respectively. With respect to the total electricity costs, the heating costs corresponded to 30.6% in the uncontrolled and 23.9% in the controlled. The specific costs per HP compared were €21.80 (US$24) in the uncontrolled case and €12.70 (US$14) in the price-controlled case.
Sharing the Results
The results of the study showed Ellevio the number of PEVs and HPs that could be connected simultaneously to the distribution network without the need for investment. However, this study was based on the assumption both the PEVs and HPs would be placed on charge in an uncontrolled manner during the most critical hour.
The results of this study were shared with residents through their ElectriCITY forum. This has helped to make them aware of the current limitations on the distribution network and the utility’s plans to overcome this problem.
Currently, PEV car ownership in Stockholm is about 370 cars per 1000 residents, so it is estimated there could be 7400 PEVs in Hammarby Sjöstad when fully developed with 20,000 residents.
Future Studies
In this study, the distribution network in the Hammarby Sjöstad district of Stockholm was analyzed for six critically loaded hours on a critically loaded day to determine whether the existing infrastructure could accept the mass introduction of DERs, specifically PEVs and HPs. The results provided Ellevio the number of PEVs or HPs that could be connected simultaneously during critical hours without further investment.
The maximum numbers serve as a reference for planning the distribution network capacity that would be needed from the external grid under these circumstances. However, for selected distribution substations within the area (for example, RS1), the results were marginal, indicating there might be a need for early reinforcement.
The results also showed that, by adopting load management strategies, it was possible to increase the level of penetration of DERs, with a decrease in the cost per unit of additional load. In the case of PEVs, the level of penetration was increased by a factor of 3 in the uncontrolled scenario. However, these results serve to identify possible limitations and capacity shortages in the distribution network to ensure smooth planning of a robust energy infrastructure for the future.
Finally, this study was performed considering PEV and HP loads in isolation, when the increased connection of these loads in reality will happen simultaneously. While the number of PEVs and HPs found within this study serve as a baseline, future studies will need to examine the security of the distribution network in terms of the N-1 criterion.
Monika Topel ([email protected]) is a post-doctoral researcher at the KTH Royal Institute of Technology located in Stockholm, Sweden. She is employed in the energy technology department within the smart energy infrastructure group and has responsibilities for the electricity infrastructure of the city of Stockholm and how to achieve the upcoming climate goals.
Josefine Grundius ([email protected]) is a long-term planner at Ellevio AB, and her main focus is risk analysis for the urban and rural distribution networks as well as geographic information system and spatial analysis. Grundius previously worked on investment projects linked to the distribution network in Stockholm.
For more information:
Ellevio | www.ellevio.se
Nord Pool | www.nordpoolgroup.com
Svenska Kraftnat | https://svk.se
