The primary objective of this work is to develop and demonstrate a steady-state stochastic simulation method to estimate the PV hosting capacity of a given distribution, based on the ANSI voltage regulation standard. The work discusses the key factors that determine the voltage rise due to distributed PV. Load demand analysis is done to determine statistically representative minimum daylight load demand for PV analysis. And lastly, the steady-state, stochastic simulation method is discussed and implemented to estimate the PV hosting capacity for small-scale and large-scale PV Deployments. / text

This study aims to examine the hosting capacity (the maximum amount of distributed generation possible to add to a current grid) of Mälarenergi’s distribution networks. The three areas examined are a rural network, a modern suburban grid and an older suburban grid. The networks are modelled in PowerWorld Simulator with data mainly from Mälarenergi’s NIS (Network Information System). The basic models include calculated minimum loads based on load profiles, combined with 0 kW, 2 kW, 4kW or 5 kW installed photovoltaics (PV) systems at each consumer. The compensating models are based on the previous ones but with reduced transformer voltages to lower the risk of grid over-voltages. A high load case is also examined to make sure there are no under-voltages for these models. The results show that the rural network is strong enough to handle the biggest available PV system at 5 kW, if the transformer voltage is lowered from 1.03 p.u. to 1.005 p.u. The modern suburban grid can host 4 kW solar panels together with a lowered voltage level of 1.005 p.u. The larger package of 5 kW leads to overloading at the transformer when used. The older suburban grid has the largest issues with both overvoltages and overloading and can only handle 2 kW distributed generation with a voltage reduction to 1.005 p.u. The models are fairly sensitive because of assumed transfomer parameters and, in the case of the modern suburban grid, some lines being removed due to limitations in the software. Regardless, the results are robust enough that they can be considered correct.

The study examines the potential for extensive connection of photovoltaic (PV) production in the Swedish rural power grid, considering the case distribution grid (10kV) of Herrljunga Elektriska. Hourly PV production is calculated using radiation and temperature data together with information regarding building roofs in the studied area. Furthermore, hourly customer load data is aggregated, enabling detailed power flow simulations of the grid resulting in hourly voltages and currents for all nodes during 2014. Three cases with varying PV production are studied, using different thresholds for minimum annual radiation. Thus, roofs with lower annual radiation are excluded from the simulations, limiting PV production. The three cases considers annual radiation greater than 0 kWh/m2 , 700 kWh/m2 and 1000 kWh/m2. Simulations show that the distribution grid in Herrljunga maintains acceptable performance with respect to voltages and currents for the 1000 kWh/m2,year case, yielding an annual production of 30 % of consumption. The hosting capacity, which is an estimate of the amount of PV that can be connected to the grid, is therefore 30 %. In order to further examine grid limitations, weak parts of the grid are identified. These are situated in the peripheral parts of the grid, which is in accordance with earlier studies of intermittent power production in distribution grids. Additionally, low voltage grids in connection to these weaker parts of the distribution grid are simulated, showing no further limitations for hosting capacity.

With the growing acceptance of photovoltaic (PV) systems globally including Sweden, an increasing number of PV systems has continually been installed all through Sweden. In this study, a selected residential area in Uppsala Sweden is considered as a possible site for PV system installation. Due to the intermittent nature of this energy source and the disturbances it causes to the grid, there is therefore the need to determine the amount of PV that can be accommodated by the already existing grid without any adverse effect to it. This is known as hosting capacity (HC). The HC signifies the level of risks the network operator as well as customers are willing to take with regard to the stability of the grid network. Many possibilities exist by which this can be done which are well discussed later in this report. However, the deterministic method is used in this work. For an effective determination of the HC, the DIgSILENT Power Factory 2021 is used to simulate a synthetic network assumed to be similar to that of the area studied. A parametric analysis is also done to as certain the impact some network variables would have on the limit of PV systems a grid network can accommodate. The results of the study showed that under 0 % penetration, the grid network is stable and no violations witnessed neither from the network with 800 kVA transformer considered as the main network in this study nor that with 500 kVA used for the parametric studies. However, the loadings on the two transformers varied by about 15 %. Integrating PVs into the grid network introduced some level of instability which increases as the size of the PVs increases. This shows the need to peg the PVs’ size to a certain maximum to be able to have control over the grid network. Customers are allowed to install lower sizes but not more than this maximum. The results also showed that changing the transformer size do not affect the nodal voltages nor the cable loadings in the network provided all other parameters remained the same. The only impact is on the transformer loading, with smaller transformers experiencing higher loading condition. This study serves as a basis to determine the initial range of PV sizes customers are allowed to install in the area studied in Uppsala Sweden.

With the integration of renewable and distributed energy resources (DER) and advances in metering infrastructure, power systems are undergoing rapid modernization that brings forward new challenges and possibilities, which call for more advanced learning, analysis, and planning tools. While there are numerous problems present in the modern power grid, in this work, this work has addressed four of the most prominent challenges and has shown that how the new advances in generation and metering can be leveraged to address the challenges that arose by them. With regards to learning in power systems, we first have tackled power distribution system topology identification, since knowing the topology of the power grid is a crucial piece in any meaningful optimization and control task. The topology identification presented in this work is based on the idea of emph{prob-to-learn}, which is perturbing the power grid with small power injections and using the metered response to learn the topology. By using maximum-likelihood estimation, we were able to formulate the topology identification problem as a mixed-integer linear program. We next have tackled the prominent challenge of finding optimal flexibility of aggregators in distribution systems, which is a crucial step in utilizing the capacity of distributed energy resources as well as flexible loads of the distribution systems and to aid transmission systems to be more efficient and reliable. We have shown that the aggregate flexibility of a group of devices with uncertainties and non-convex models can be captured with a quadratic classifier and using that classifier we can design a virtual battery model that best describes the aggregate flexibility. For power system analysis and planning, we have addressed fast probabilistic hosting capacity analysis (PHCA), which is studying how DERs and the intermittency that they bring to the power system can impact the power grid operation in the long term. We have shown that interconnection studies can be sped up by a factor of 20 without losing any accuracy. By formulating a penalized optimal power flow (OPF), we were able to pose PHCA as an instance of multiparametric programming (MPP), and then leveraged the nice properties of MPP to efficiently solve a large number of OPFs. Regarding planning in power systems, we have tackled the problem of strategic investment in energy markets, in which we have utilized the powerful toolbox of multiparametric programming to develop two algorithms for strategic investment. Our MPP-aided grid search algorithm is useful when the investor is only considering a few locations and our MPP-aided gradient descent algorithm is useful for investing in a large number of locations. We next have presented a data-driven approach in finding the flexibility of aggregators in power systems. Finding aggregate flexibility is an important step in utilizing the full potential of smart and controllable loads in the power grid and it's challenging since an aggregator controls a large group of time-coupled devices that operate with non-convex models and are subject to random externalities. We have shown that the aggregate flexibility can be accurately captured with an ellipsoid and then used Farkas' lemma to fit a maximal volume polytope inside the aforementioned ellipsoid. The numerical test showcases that we can capture 10 times the volume that conventional virtual generator models can capture. / Doctor of Philosophy / With the integration of renewable and distributed energy resources (DER) and advances in metering infrastructure, power systems are undergoing rapid modernization that brings forward new challenges and possibilities, which call for more advanced learning, analysis, and planning tools. While there are numerous problems present in the modern power grid, in this work, this work has addressed four of the most prominent challenges and has shown that how the new advances in generation and metering can be leveraged to address the challenges that arose by them. With regards to learning in power systems, we first have tackled power distribution system topology identification, since knowing the topology of the power grid is a crucial piece in any meaningful optimization and control task. We next have tackled the prominent challenge of finding optimal flexibility of aggregators in distribution systems, which is a crucial step in utilizing the capacity of distributed energy resources as well as flexible loads of the distribution systems and to aid transmission systems to be more efficient and reliable. For power system analysis and planning, we have addressed fast probabilistic hosting capacity analysis (PHCA), which is studying how DERs and the intermittency that they bring to the power system can impact the power grid operation in the long term. We have shown that interconnection studies can be sped up by a factor of 20 without losing any accuracy. Regarding planning in power systems, we have tackled the problem of strategic investment in energy markets, in which we have utilized the powerful toolbox of multiparametric programming to develop two algorithms for strategic investment. We next have presented a data-driven approach in finding the flexibility of aggregators in power systems. Finding aggregate flexibility is an important step in utilizing the full potential of smart and controllable loads in the power grid and it's challenging since an aggregator controls a large group of time-coupled devices that operate with non-convex models and are subject to random externalities.

Made available in DSpace on 2016-08-31T13:33:44Z (GMT). No. of bitstreams: 1 EdnardoPR_DISSERT.pdf: 2329567 bytes, checksum: ab9bc3145d3823983047e9d57956c8d0 (MD5) Previous issue date: 2015-01-30 / Fundação de Apoio a Pesquisa do Estado do Rio Grande do Norte / The connection of wind generators might cause significant influences in the profile voltages, voltage unbalance, loading and the electrical losses in radial power distribution systems. This fact requires a specific study, called Hosting Capacity, which aims to analyze the maximum limit of the power increase on the network that makes the performance of the system acceptable to the established quality limits. This work shows an IEEE radial distribution system behavior, composed of 13 bus, in steady state, when a synchronous wind machine is engaged on the bus 680 in two ways: directly connected to the network and connected through the frequency converter. For each type of connection the power factor was varied from 0.9 capacitive to inductive 0.9. The parameters analyzed in connection bar were the degree of voltage unbalance, the losses in the system and the profile of voltages on the bus 680 for each case . The simulations were performed using the program Distribution Network Analysis with Generation Aeolian-Electric - ANAREDGEE, which was developed and validated with own results of the IEEE. It was found that there was a reduction in the degree of unbalance for all simulated situations, when compared to the original value of the system status, not exceeded the limits determined by imbalances entities NEMA, ANSI, IEEE and ANEEL. Regarding the profile of voltages in the various system buses, there was a slight decrease in some situations. In the original system, the voltage level of the phase B in the bus 680 exceeded the 5% voltage given by ANEEL, with a value of 1.0529 p.u. With the machine connected directly, there was an increase of this value in all simulations with different power factor, while the connection through frequency converter might causa a reduction of the phase B voltage levels to below 1.05 pu in situations where the power factors were equal to 0.9 capacitive and unitary. The system losses were reduced in all cases, but showed lower values when the synchronous machine was integrated into the system by frequency convertor. It was also observed a significant reduction in the degree of system imbalance after connecting the synchronous machine, this reduction, in most cases, occurred in proportion to the increase in power injected into the connection bar / A conexão de geradores eólio-elétricos pode causar influências significativas no perfil de tensões, desequilíbrio de tensão, no carregamento e nas perdas elétricas em sistemas de distribuição de energia radiais. Este fato requer um estudo específico, denominado Hosting Capacity, que tem por finalidade analisar o limite máximo do incremento de potência na rede que torna a performance do sistema aceitável para os limites de qualidade estabelecidos. Este trabalho demonstra o comportamento do sistema de distribuição radial do IEEE, composto de 13 barras, em regime permanente, quando uma máquina eólica síncrona é acoplada na barra com o maior nível de tensão do sistema, de duas formas distintas: diretamente conectada à rede e conectada por meio de conversor de frequência. Para cada tipo de conexão o fator de potência foi variado de 0.9 capacitivo a 0.9 indutivo. Os parâmetros analisados na barra de conexão foram o grau de desequilíbrio de tensão, as perdas no sistema e o perfil das tensões. As simulações foram realizadas através do programa Análise de Redes de Distribuição com Geração Eólio-Elétrica - ANAREDGEE, que foi desenvolvido e validado com resultados próprios do IEEE. Verificou-se que houve uma redução no grau d e desequilíbrio para todas as situações simuladas, quando comparadas ao valor da situação original do sistema, não superado os limites de desequilíbrios determinados por entidades como NEMA, ANSI, IEEE e ANEEL. Com relação ao perfil das tensões nas diversas barras do sistema, houve uma ligeira diminuição em algumas situações. No sistema original, o valor de tensão na fase B da barra 680 superava os 5% de sobretensão determinado pela ANEEL, apresentando um valor de 1,0529 p.u. Com a máquina diretamente conectada houve um aumento deste valor em todas as simulações com diferentes fatores de potência, enquanto que a conexão por meio de conversor de frequência possibilitou uma diminuição dos níveis de tensão da fase B para valores abaixo de 1,05 p.u. nas situações em que os fatores de potência eram iguais a 0.9 capacitivo e unitário. As perdas no sistema foram reduzidas em todos os casos, entretanto apresentaram menores valores quando a máquina síncrona era integrada ao sistema por meio de conversor de frequência. Foi observado também uma redução considerável do grau de desequilíbrio do sistema após a conexão da máquina síncrona, esta redução, na maioria dos casos, se deu de forma proporcional ao aumento da potência injetada na barra de conexão

The demand for electrical power is growing due to factors such as population growth, urbanisation, and the transition from fossil fuels to renewable energy sources. To be able to keep up with the changes in electricity demand, the Swedish power grid must connect more renewable power generation, but also  increase its transmission capacity. Traditionally, power grids are expanded to increase the transmission capacity which requires a lot of time and investments. In order not to hinder the electrification of society, it is important to adequately estimate the current transmission capacity and plan the expansions accordingly. In the past, the generation of electrical power was primarily based on dispatchable energy sources, and the planning of new connections to the grid was assessed according to the stable and controllable nature of the electricity supply. However, renewable sources like solar and wind power are affected by weather variations. Therefore, the traditional methods of planning the power grid are no longer sufficient. Instead, there is a need to develop and implement new methods that account for the variable nature of renewable energy sources, and also the possible complementarity between different renewable power sources. This can possibly allow more connection of renewable power generation to the grid, without the need of expanding it. The aim of this thesis is to investigate two different methods for analysing how much renewable power generation that can be connected to the power grid, so-called hosting capacity methods. The first method is a deterministic method which is traditionally used in power system analyses since it is a fast, simple and conservative method. This method does neither consider the intermittent nature of solar and wind power, nor any complementarity. The second method is a time series method which considers the complementarity and intermittency of solar and wind power but requires much data. The methods are compared in regards to assessed hosting capacities, risks and reliability of results. The study is performed on a regional grid case in the middle of Sweden. Solar and wind power plants with different capacities are modeled at ten buses in the power grid. The power grid is analysed in PSS/E with loading of lines and voltage levels determining the assessed hosting capacities. A correlation map presenting the temporal correlations of solar and wind power over the grid case area is also created in order to evaluate the complementarity in the area and its possible effects on the assessed hosting capacities.  The results show that the time series method is more reliable than the deterministic method. This is due to the difficulties in identifying accurate worst case hours that are used for the deterministic method. The time series method is also preferred as it considers complementarity between solar and wind power. However, the correlation map argues that the grid case area has weakly positive correlations, meaning low complementarity between solar and wind power. This suggests that the differences in hosting capacity between the two methods are more likely dependent on the temporal variations in existing load and power generation. The differences in assessed hosting capacity between the ten buses in the power grid are probably not due to the local complementarity either, but rather the structural differences of the grid in terms of components, local loads and existing power generation.

abstract: Nowadays, the widespread introduction of distributed generators (DGs) brings great challenges to the design, planning, and reliable operation of the power system. Therefore, assessing the capability of a distribution network to accommodate renewable power generations is urgent and necessary. In this respect, the concept of hosting capacity (HC) is generally accepted by engineers to evaluate the reliability and sustainability of the system with high penetration of DGs. For HC calculation, existing research provides simulation-based methods which are not able to find global optimal. Others use OPF (optimal power flow) based methods where too many constraints prevent them from obtaining the solution exactly. They also can not get global optimal solution. Due to this situation, I proposed a new methodology to overcome the shortcomings. First, I start with an optimization problem formulation and provide a flexible objective function to satisfy different requirements. Power flow equations are the basic rule and I transfer them from the commonly used polar coordinate to the rectangular coordinate. Due to the operation criteria, several constraints are incrementally added. I aim to preserve convexity as much as possible so that I can obtain optimal solution. Second, I provide the geometric view of the convex problem model. The process to find global optimal can be visualized clearly. Then, I implement segmental optimization tool to speed up the computation. A large network is able to be divided into segments and calculated in parallel computing where the results stay the same. Finally, the robustness of my methodology is demonstrated by doing extensive simulations regarding IEEE distribution networks (e.g. 8-bus, 16-bus, 32-bus, 64-bus, 128-bus). Thus, it shows that the proposed method is verified to calculate accurate hosting capacity and ensure to get global optimal solution. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2018

As renewable energy plays an increasingly important role in the power system, the addition of PV systems to the distribution network has become a major trend in the current power system development. However, if a PV system with excessive capacity is added to the distribution network, voltage problems may occur in the system. Hence, it is important to determine the capacity of the PV system that can be added at the distribution system. The thesis aims to identify the major factors that affect the PV hosting capacity of distribution systems. The thesis studies various scenarios for the IEEE-123 test network PV system and evaluates the PV hosting capacity of the distribution system based on simulation tools including Matlab and Opendss software.

The ambitious governmental policies, particularly in Europe, in pursuit of established energy targets require an increase in distributed generation. As a result, photovoltaic (PV) technologies have emerged, predominantly at residential Low Voltage (LV) feeders. However, PV rich LV feeders are highly likely to pose technical challenges such as significant voltage rise and thermal overloading. This inevitably limits the volume of PV systems that can be hosted on LV feeders. Therefore, the deployment of solutions that can enable feeders to accommodate greater volumes of PV systems without having any technical issues is crucial. This thesis, consequently, thoroughly investigates one of the potential solutions: transforming the radial operation of LV feeders into ring operation. European-style LV feeders are typically operated in a radial fashion and yet are designed to be reconfigurable with neighbouring feeders. It is, therefore, essential to identify the best pairing option (of PV rich LV feeders) in a practical and straightforward manner due to the large number of existing LV feeders in a given Distribution Network Operator (DNO) area. This thesis proposes a generic innovative methodology to enable DNOs to straightforwardly identify the best pairing feeder; a decision-making tool to facilitate the rapid uptake of PV systems. To accomplish this goal, an impact assessment of a set of real residential LV feeders is carried out to identify the first technical issue/constraint that limits their hosting capacity. Next, regression analyses are carried out to gain an understanding of the relation between this first occurrence of technical issue/constraint and the corresponding level of PV penetration. The most practical and adequately accurate metric needs to be chosen. Feeders are then classified based on the range of metrics to cover all possible pairing cases. Finally, the ring operation of feeders in each class is analysed and hosting capacities are compared to those of radial ones. This process creates a practical matrix from which DNOs can easily identify the best pairing feeders. DNOs are likely to be hesitant to adopt permanent ring operation as it is not typically adopted in traditional LV feeders. Therefore, the switch located between feeders can be operated over time (i.e., dynamic ring operation) to reduce the duration for which ring operation is in place. It is, however, challenging to identify the most favourable control strategy. This thesis proposes different strategies for dynamic ring operation. Note that the most preferable control strategy is that which preserves the benefits of permanent ring operation with the minimum duration of ring operation and minimum number of switching. To achieve this, four different control strategies are explored-using different control cycles and considering hosting capacity, duration and switching. The best control strategy is found to be able to increase hosting capacity as permanent ring operation, reduce switching actions and minimise duration of ring operation compared to other proposed strategies and, crucially, operate ring operation only when it is truly needed. Finally, this thesis investigates the use of ring operation with an LV on-load tap changer (OLTC) as this is recently available voltage control technologies and is increasingly drawing the attention of DNOs. Two approaches are investigated to increase hosting capacity and limit ring operations: the use of the switch and OLTC are controlled separately using local measurements (i.e., localised) and their simultaneous control at the LV transformer level (i.e., centralised). The latter gives the priority to the OLTC to minimise the duration of the ring operation. The assessments are extended to cover an integrated medium and low voltage network to obtain more realistic results. The results show that centralised approach provides better performance considering hosting capacity, the number of switching/tap actions and the duration of ring operation.