The hosting capacity of low-voltage (LV) networks is influenced by existing consumption and production in adjacent LV networks under the same medium-voltage (MV) network. Since LV transformers typically lack automatic on-load tap changers, both voltage and current limits require a joint assessment covering the MV network and all underlying LV networks. This paper introduces a methodology to include different MV operating conditions, like reserve operating paths, in the calculation of the hosting capacity for new production at LV networks. The MV voltage profile before the connection of new production, called the background voltage, is modelled to enable such analysis. The methodology is applied to an existing MV/LV network. The hosting capacity was lower, for the studied reserve operating paths, and the limitation was due to overvoltage issues. The proposed methodology is a valuable tool for distribution network planning. It was also shown that it is essential to use an accurate model of the background voltage for hosting capacity calculation of distribution networks.

This paper presents an approach to estimate the hosting capacity for distribution networks considering the impact of PV penetration at different voltage levels. The estimation and the method were selected such that the results were most suitable for distribution system planning. A time-series based method was used as it covers significant aspects needed for prioritising network reinforcement. The MV background voltage was modelled varying in time, assuming the same penetration level in the other LV networks supplied by the same MV system. The hosting capacity is defined as the maximum acceptable PV size per customer for a given PV penetration. Based on the different possible combinations of PV location, the probability of overvoltage and overloading is used as a performance index. The planning risk is used as a limit for the performance criterion. The method can be automated for a large number of networks due to using an IEC 61970-based input format. It also enables linking DSO network models to customer smart metre databases. The severity and risks of limit violations are analysed with different metrics from the time-series simulations. The change in background voltage with increasing penetration is shown to impact the results significantly. When considering it, the estimated hosting capacity was reduced by 32 %, on average.

Assessing the influence of the medium voltage (MV) network on multiple low-voltage (LV) networks in hosting capacity studies becomes important with a large share of PVs. This paper introduces the concept of background voltage to describe the impact of varying production and consumption on hosting capacity studies. A time-series model for the background voltage is presented, which considers such effects. The background voltage model has two components: the voltage drop due to consumption and the voltage rise due to PVs in other LV networks. Both components impact the hosting capacity for LV networks. Different forms of modelling these components were used to assess their impact. Using time-series for both the voltage drop and rise gives a good representation of the voltage behaviour throughout the year, creating a realistic scenario, so that different solutions can be compared towards increasing hosting capacity.

This paper presents the results obtained for Hosting Capacity by applying a time-series-based methodology using real consumption, PV production and EV charging data. The background voltage from PV/EV integration at the MV level is modelled in the time domain. An 11 kV MV network with 57 underlying 400 volts LV networks is assessed. Results show the importance of using real data for modelling EV charging in order to get a correct coincidence factor of charging. Due to low coincidence in charging times, the voltage is not impacted significantly for EV, as it is for PV with higher coincidence. For PV studies, the MV background voltage is shown to be essential for an appropriate HC assessment. Time-series simulations including the background voltage allow the voltage behaviour to be realistically modelled and enable detailed studies.

The stochastic approach is applied to an entire concession area with 1264-LV distribution networks to estimate the hosting capacity. About 15 000-customers are connected to the individual secondary distribution networks supplied through 48-medium voltage 10 kV radial feeders. The hosting capacity assessment uses the end-customer voltage magnitude rise and transformer thermal overload. The hosting capacity is estimated by applying the "stochastic mixed aleatory-epistemic method" to determine the voltage magnitude rise and load flow with solar PV. The minimum power consumption is compared with the solar PV power infeed through the individual transformers. The hosting capacity estimation is done for three-phase connected solar PV sizes from 3 to 18 kW. At moderate PV penetration (25%-50%), the results showed that overvoltage would limit the hosting capacity more often than overload, but it becomes an issue only for LV networks studied with more than 8-customers. Considering all LV networks, most of the customers could install 6 kWp. Even when installing PV systems of 18 kWp (about twice the average size today and about the maximum area of a typical residential roof), two-thirds of houses would not need an upgrade to withstand SS-EN 50160 voltage limits. The latter customers can connect solar PV units with 18 kWp size without overvoltage or overload issues.

The amount of solar photovoltaics (PV) that can be connected to the grid is a question which, especially for distributed solar, is characterized by a strong information unbalance between, on the one side, the grid operators and on the other side those wanting to feed in electricity. Herein, how the grid impact of PV can be assessed by analyzing inverter voltage measurements that today are available from millions of online PV inverters is demonstrated. In the case where grid capacity is limited by voltage, online PV inverters are shown to provide a valuable complement to distribution system operators estimation of grid capacity. Using 25 million voltage samples from a thousand PV systems, a hosting capacity map is produced for 20 of 21 regions of Sweden. The map shows the amount of PV that typically can be installed without causing voltage rise that would lead to inverter cutoff and loss of production. By monitoring and following the voltage impact from solar PV over time, local saturation effects can be detected and the timing of grid investments better planned. The information could also enhance regulators in their role to monitor grid operators and assure the right level of grid investments.

The reliability of an electrical system depends on layered Protection, Automation, and Control (PAC) functions in which differential relays are considered a fundamental method to isolate the problem. By having a PAC architecture with functionality independent of hardware we hypothesize that the life cycle of PAC devices can be extended by using modern, powerful, and highly reconfigurable IoT devices. This can be achieved by applying modularity to hardware architecture, where an IoT device can be considered a mezzanine module. The Raspberry Pi and BeagleBone manufacturers are keeping the hardware the same size while constantly increasing its power. However, it is not clear how powerful the IoT device should be to meet some of the IEC 61850 requirements (which are usually included in PAC).In this article, we provide a step-by-step process for designing a system that mimics a single-phase transformer differential protection system. First, we provide a requirements analysis (Sampled Values rate, traffic synchronization, reaction time, etc.). Then we analyze the BeagleBone Black device (BBB). We give an example of an electrical system and finally analyze, in terms of IEC 61850 requirements, the performance of two software applications developed for the BBB (based on the operating system and on bare metal).

This paper presents a novel approach to estimate the impacts of Photovoltaic (PV) power production and electric vehicle (EV) charging over large geographical areas without the use of a full grid model and without end-customer consumption data, which is often not available externally to the distribution system operator. Instead spatiotemporal Markov based models of PV production and EV charging are used with building footprints, parking spaces and light detection and ranging (LiDAR) data taken from national land surveys and OpenStreetMap. Data of driving patterns in the city and weather data are used to calculate time-series that are added to the aggregated consumption and allocated to the simplified (70 kV) regional grid within the city. The method thus allows to evaluate the impacts of high PV and EV scenarios on the high/medium voltage substations. The method also enables assessment of the maximum 10-min power flow that can be allowed with full N-1 redundancy of the transformers at what PV and EV penetration levels transformer upgrade (often coupled to major substation refurbishment) would be required. A case study on a large Swedish city is presented made where 100% EV and PV penetrations with various charging powers and charger-accessibility were simulated. The impact of overloading the main transformers and the capacity of the reserve transformers is determined.

One of the most important elements of any AC power supply system is the transformer. Each stakeholder, such as Distribution Network Operators (DNOs), Transmission System Operators (TSOs), and other generators, is paying close attention to investigating possible negative impacts that could disrupt the normal distribution of electricity. One of them is short-term transient processes that occur in the network when the transformer core is turned on. These transients can cause significant distortion on the power line, which in turn can cause nuisance tripping of protection devices. For a small and isolated electrical system, this is not usually considered a major problem. However, as the electrical grid grows, this becomes more important. This is a problem that machine learning can help with. However, the problem is that there is currently no significant data on the transformer excitation phenomenon.This article gives a brief explanation of transformer excitation phenomena in an introductory section. With this information in mind, we developed a SIMULINK model and a MATLAB script to describe and automatically run a new experiment design and generate data. Finally, we use the generated data as a training dataset to fit it into a supervised convolutional neural network. The final data set that we created and used for the purposes of this work, as well as the SIMULINK model and the MATLAB script for automatically generating the experiment design, are in the public domain.

This paper presents a real-word implementation of a TSO-DSO-customer coordination framework for the use of flexibility to support system operation. First, we describe the general requirements for TSO-DSO-customer coordination, including potential coordination schemes, actors and roles and the required architecture. Then, we particularise those general requirements for a real-world demonstration in Sweden, aiming to avoid congestions in the grid during the high-demand winter season. In the light of current congestion management rules and existing markets in Sweden, we describe an integration path to newly defined flexibility markets in support of new tools that we developed for this application. The results show that the use of flexibility can reduce the congestion costs while enhancing the secure operation of the system. Additionally, we discuss challenges and lessons learned from the demonstration, including the importance of the engagement between stakeholders, the role of availability remuneration, and the paramount importance of defining appropriate technical requirements and market timings.

This paper proposes a number of deterministic and stochastic approaches to quantify the hosting capacity of the distribution network for solar photovoltaics (PV) units when that hosting capacity is limited by the loadability of feeder cables or distribution transformers. Such approaches should be used as input to investment and planning decisions by distribution-system operators. Distribution networks from two areas in Sweden supplying 309 MV/LV distribution transformers with 12,000 customers downstream have been used to illustrate the different approaches. By defining a break-even length, it is shown that both overload and overvoltage may limit the hosting capacity. The results obtained for the 309 networks show that the overload limit is more often exceeded for transformers than for feeder cables for smaller solar PV sizes. For larger solar PV sizes, feeder cables are likely to be overloaded before the transformers. A stochastic hosting capacity is calculated for 170 of the 309 distribution transformers and compared with a deterministic definition of the hosting capacity. It is shown that the deterministic definition corresponds to a 50% risk of overloading. Both deterministic and stochastic definitions show that transformer overloading is very unlikely as long as the probability of customers installing solar PV is less than 20%.

This paper presents a stochastic approach to single-phase and three-phase EV charge hosting capacity for distribution networks. The method includes the two types of uncertainties, aleatory and epistemic, and is developed from an equivalent method that was applied to solar PV hosting capacity estimation. The method is applied to two existing low-voltage networks in Northern Sweden, with six and 83 customers. The lowest background voltage and highest consumption per customer are obtained from measurements. It is shown that both have a big impact on the hosting capacity. The hosting capacity also depends strongly on the charging size, within the range of charging size expected in the near future. The large range in hosting capacity found from this study—between 0% and 100% of customers can simultaneously charge their EV car—means that such hosting capacity studies are needed for each individual distribution network. The highest hosting capacity for the illustrative distribution networks was obtained for the 3.7 kW single-phase and 11 kW three-phase EV charging power.

This paper applies a deterministic and stochastic approach to estimate the hosting capacity and likelihood of an overload at the cable cabinet or transformer due customers with solar PV. Distribution networks at 10 kV, supplying feeders with 1300 MV/LV transformers, have been studied. The paper also quantifies whether overvoltage or overload limits the hosting capacity first. Voltage and current measurements from a Northern Sweden distribution network located in a rural area have been applied in the paper. Illustrations are used to show the concept and important results are obtained. It is shown in the paper the hosting capacity is determined more often by overload than by overvoltage. It is shown in the paper the hosting capacity is determined more often by overload than by the voltage rise limit of the distribution cable. The results obtained for the data used show that the overload limit is exceeded more often for transformers than for cable cabinets. The stochastic approach applied to the cable cabinets yields a small probability to exceed the hosting capacity.

This paper presents the potential for prosumer batteries coupled to PV units to cover the national frequency balancing needs in Sweden. PV coupled residential batteries are found to be profitable with today's prices, if granted access to balancing markets. Simulations are based on national targets for solar PV production in 2040 (5-10 TWh, 5-10% of electric consumption) and current residential PV share of total installed PV capacity. In the study battery attachment rate was 50% and 15% of single family houses were equipped with 10 kW PV installation with a battery capacity of 6 kW / 7.68 kWh. In total, the battery PV systems constituted 25% of total installed capacity of PV in 2040. The results showed that 20% of the aggregated batteries capacity is sufficient to provide around 70-100% of each of the frequency reserves individually. The highest savings are gained for the households when both the primary frequency reserves, FCR-N and FCR-D, are provided by the aggregated batteries together with increasing the PV self-consumption, peak shaving and energy arbitrage. When providing frequency support the PV system payback time was reduced from 14 to 11 years when equipped with battery, compared to only installing PV.

Solar photovoltaics in electricity distribution networks is often limited by the rise in voltage magnitude. The pre connection voltage magnitude is an important factor that determines the hosting capacity. This paper studies to which extent details of the pre-connection voltage magnitude impact the hosting capacity. Extensive measurements of voltage magnitude and solar power production were obtained for a number of distribution networks with 10-minute resolution. The measured background voltage during the sunny-hours from the two-year measurements was used to obtain representative probability distribution functions. A guide for selecting the time-of-day (ToD) used is presented. The obtained probability distribution functions are applied to estimate the stochastic hosting capacity for a low-voltage distribution network with 83 customers. The impact of various details on the hosting capacity are studied. The results show that general knowledge about the range of the pre-connection voltage are essential for the hosting capacity estimation. Measurements over one year were shown to be sufficient to estimate the hosting capacity. The hosting capacity considering the entire day was underestimated by 11 % when compared to the 10 am - 2 pm sunny-hours. The proposed method is general and can be applied to other aleatory uncertainties and other types of hosting capacity studies.

The CoordiNet project within Horizon 2020 programme aims to demonstrate how Distribution System Operators (DSO) and Transmission System Operators (TSO) can act in a coordinated manner to procure and activate grid services in the most cost effective and reliable way. With 4 GWh of traded flexibility during winter of 2019/2020, the Swedish demo sites constitutes one of Europe's largest local flexibility markets for congestion management. The paper presents the experience from interviews with fifteen flexibility providers participating in the markets. Together they provide flexibility from a wide mix of different technologies including district heating, generation from waste disposal, aggregated consumer load, industrial heat pumps, gensets and residential housing blocks. Fifteen resource owners have made the journey to become flexibility providers. Understanding their experience and needs will help DSO develop products and business models for flexibility services as well as getting insight in the practical hurdles that hinder network customers to become a flexibility provider (FSP). The results from implementation of the platform, user interface, learnings on stakeholder interaction and initial evaluation of first two years operation of the market is also addressed.

With falling prices both the size and number of installations of single-phase and three-phase solar photovoltaics (PV) units is expected to continue to increase in distribution networks. The increase in solar PV units will have an impact on the power quality of the distribution network. There is a need to propose transparent methods that can be used to quantify the acceptable limit of solar PV penetration objectively, called the hosting capacity (HC). The fundamental methods for the quantification of hosting capacity include; deterministic, stochastic and time-series. Which of the latter method is the most appropriate for determining the hosting capacity? A comparison of the three methods based on the input data need, accuracy, uncertainties considered, time-related and models are done. Many academic and industrial studies have been published applying the hosting capacity method. Differences between the scientific and industrial application of the hosting capacity methods are presented in this paper. There is a need to expose the industrial application of solar PV hosting capacity methods.

This paper describes the requirements for a standardised hosting capacity method. It outlines the needs and requirements for a standardised methodology as well as ongoing and upcoming standardisation activities in IEEE and IEC. The paper outlines the criteria for consensus, transparency, interoperability and stability required from a standardised hosting capacity method. The use of IEC 61968/61970 Common Grid Model Exchange Specification is proposed as a basis for reproducibility and benchmarking of results across network operators, countries, regions and network design methods.

A fast screening method to estimate PV hosting capacity in a low voltage grid with respect to overvoltage and transformer overload is presented in this paper. The method uses the resistance of the direct path from the slack bus to customer and an assumed topology in order to avoid complete knowledge about the grid topology. Using an assumed topology an estimate of the minimum overvoltage hosting capacity is obtained for each penetration level and the transformer overload hosting capacity for any combination of customers having PV. The method is suitable for initial response to demand connection enquiries and showed good correspondence by comparison with power flow simulation using the actual grid topology. The method was implemented on four semi-rural low voltage grids and the results showed that these grids were able to handle PV infeed comparable to maximum winter load due to three-phase connections.

This paper proposes a stochastic method, ?mixed aleatory-epistemic?, for estimating solar PV hosting capacity (HC) of low-voltage (LV) distribution networks. The approach treats the aleatory and epistemic uncertainties in a different way. The HC is estimated by applying the transfer impedance matrix, ?which is only calculated once?, and the superposition principle to determine the voltage magnitude rise due to solar PV. By distinguishing between aleatory and epistemic uncertainties, the calculations are limited to the relevant hours (time-of-day or time-of-year) during which high solar PV production is expected. In this way, the random aleatory uncertainties (background voltage, solar PV production, local consumption) are modelled by their probability distributions during the selected time period. The distributions for the epistemic uncertainties (installed capacity per customer, number of customers with solar PV, phase to which single-phase units are connected) are created with simple models involving the interval value and possible occurrence. The stochastic approach proposed is applied to three LV distribution networks to illustrate the method. The results show that both types of uncertainties affect the HC. The need for distribution network planners to identify and distinguish between the types of uncertainties is emphasised.