The Netherlands is committed to reducing carbon emissions in all sectors. This will cause the electricity production to be dominated by wind and solar power, and sectors such as mobility and industry will be heavily electrified. This poses significant challenges for the electricity grid. In addition, the government plans to phase out the use of natural gas in buildings by 2050. Local governments are now choosing alternative heating sources for specific neighbourhoods, and heat pumps are one of the options. As these use electricity for heating, it is important to consider their impact on the grid.

Therefore, this thesis analyses the impact of different types of heat pumps on a real Dutch low-voltage grid, focussing on transformer overload and voltage limit problems. A simulation model based on Powerfactory and Python is built, where different types of heat pumps are considered, including air- and ground-source heat pumps, as well as electric-resistive and natural gas hybrid heat pumps. Different control methods are also considered. In addition to heat pumps, public and private electric vehicle chargers and photovoltaic systems are also included in the model. Additionally, the effects of dynamic energy prices and domestic hot water production are investigated. The simulation is performed on an hourly basis using real data from the two coldest weeks of the winter of 2021.

The results show that air-source heat pumps have a significantly higher impact on transformer overload and voltage limit violations than ground-source heat pumps, and modulating heat pumps have a slightly lower impact on the grid than on-off controlled heat pumps. Gas-hybrid heat pumps are found to cause almost no grid problems on the low-voltage grid studied and have a significantly lower impact than the all-electric air-source heat pump. The presence of an electric-resistive backup heater in the heat pump significantly increases transformer overload and voltage limit problems. Scenarios for 2030 and 2050 show that problems are unlikely to occur in 2030, but the current grid will not be sufficient in 2050 if air-source heat pumps dominate the grid. More research is needed to determine whether the current grid is sufficient when ground-source heat pumps are dominant.

It was also found that dynamic energy prices will not be a problem for the grid at current rates of heat pump and home charging for electric vehicles. However, if every household has a heat pump, a penetration rate of 20\% or more of dynamic energy contracts will substantially increase transformer overloads and voltage level violations. Finally, domestic hot water production does not significantly increase overloading and voltage limit problems.

In conclusion, heat pumps can cause significant problems for the low-voltage network. However, the impact can be reduced by choosing the right type of heat pump. It is recommended that future research considers cooling, heat and electricity storage, thermal ratings of transformers, and variable capacity tariffs.

In the context of increasing renewable energy integration and the push for decarbonization, this research addresses the need for long-duration electricity storage to ensure the reliability and adequacy of the future European power system. The variability of renewable energy sources such as wind and solar, coupled with the challenges posed by climate change, requires advanced storage solutions to balance periods of low energy generation and high demand.

This thesis explores how long-duration energy storage (LDES) and short-duration energy storage (SDES) can effectively contribute to resource adequacy by assessing various operational strategies, storage capacities, and system conditions. Using the Calliope energy system modeling framework, the research simulates a future interconnected European grid in 2050, considering different weather conditions and storage configurations to evaluate metrics such as Expected Energy Not Served and Loss of Load Expectation.

The results demonstrate that increasing the energy storage capacity and discharge duration of LDES significantly reduces LOLE, particularly during extended periods of low renewable output. However, SDES plays a complementary role, addressing short-term reliability challenges by reducing EENS during shorter disruption events. This combination highlights the need for a balanced deployment of both SDES and LDES to optimize grid performance and ensure resource adequacy. Through these findings, the thesis contributes valuable insights for energy planners and policymakers, aiming to foster a resilient, low-carbon electricity grid capable of meeting the demands of a renewable-dominated future.

The growing integration of renewable energy sources presents significant challenges for the reliability of the energy infrastructures in The Netherlands. To address these challenges, dynamic electricity pricing has been proposed as a potential solution to align the electricity demand with the variable supply. While dynamic pricing can incentivize the shifting of flexible loads, its impact on the low-voltage grid remains under-investigated. This thesis explores the impact of dynamic pricing on grid congestion, especially in combination with EV charging and PV systems.


It is hypothesized that while dynamic pricing can effectively shift electricity consumption, it may also lead to concentrated periods of high demand. When consumers synchronize their electricity usage during low-price times, grid congestion can occur.
The study creates a simulation model using Python and PowerFactory to analyse the effects of dynamic pricing on the low-voltage grid. The simulations are based on real-world data from 2023, making use of hourly intervals of loads, EV charging profiles, and solar energy generation across three representative weeks. The model evaluates the voltage levels and power flows to assess the potential for grid congestion under various scenarios of EV and PV adoption.

The results indicate that dynamic pricing can lead to concentrated EV-charging loads, resulting in voltage dips and power spikes. These findings show that the current low-voltage grid may not be equipped to handle the increased demand and variability caused by dynamic pricing schemes. Thus, while dynamic pricing can offer a promising tool for demand-side management, additional measures are necessary to ensure grid stability and to avoid future congestion issues.

In conclusion, the study underscores the dual nature of dynamic pricing as both a valuable tool and a potential risk for the low-voltage grid. Both the voltage and power limits of the grid were found to be causes of congestion, highlighting the need to investigate both. Future research should focus on optimizing dynamic pricing schemes and developing robust grid infrastructure to fully leverage the benefits of renewable energy integration while ensuring grid reliability and resilience.

Nowadays, the rise in energy consumption and the integration of renewable energy resources (RES) have introduced significant challenges to the existing power grid, making it necessary to upgrade the current power system. However, considering the high variability in load and RES and the large number of possible investment options, the power system expansion planning problem results in a large mixed integer linear problem (MILP), or in some cases, a non-linear problem, impractical to solve for real-world scenarios. Time-series aggregation (TSA), capturing representative load and RES patterns, has emerged to reduce the temporal complexity, making the power system expansion planning model much easier to solve while providing similar final results.

Current TSA methods mainly rely on passive clustering, focusing on the statistical proximity between the representative periods and the full-space time-series. However, this approach does not guarantee a satisfactory solution for the final expansion planning problem in general cases, even with predefined extreme periods implemented such as days with maximum load and minimum available RES. The operation of the power system and extreme conditions are highly sensitive to the specific power system configurations, making the standalone TSA method unreliable in practical applications.

To improve the time-series aggregation in terms of the power system operation, firstly, the methods of assessing the performance of the estimated investment decision are explored directly in terms of the objective function. By introducing the full-space operational cost model for the power system with the estimated investment decision, the operational cost error made by representative periods can be obtained, referred to as the operational estimation error. The actual objective difference between the estimated investment decision and the optimal investment decision found by the full-space expansion planning model can also be evaluated, denoted as the optimality gap. It is found that the optimality gap is bounded by the difference in the operational estimation error of representative periods for the power system with the two investment decisions. As the operations of the power system with the estimated and optimal investment decisions are better estimated, the simplified model can provide a closer investment decision.

Subsequently, looking into the operational estimation error, the performance of representative periods in estimating the full-space operational cost is highly unevenly distributed among the full-space time-series. Representative periods fail to accurately estimate a minor portion of the full-space time-series, causing extremely high operational estimation errors that contribute to the majority of the total operational estimation error. This uneven distribution remains as the number of representative periods increases, indicating that standalone TSA methods are not capable of capturing the extreme conditions of the specific power system operation.

Therefore, to improve the time-series aggregation in terms of its ability to better estimate the operation of the power system, bad-performing representative periods and original periods with high operational estimation error can be identified and prioritized, forming a feedback enhancement loop. Case studies show that the feedback enhancement with re-clustering on the bad-performing representative periods improves the optimality gap by more than 50% compared with the standard mean-based clustering method.

In recent years, the rapid increase in energy demand and the threat of climate change have pushed the energy sector towards electrification. Despite this being the first step to a more sustainable future, a new problem arises with the growing demand: grid congestion. This project explores innovative strategies within energy system design and operational framework to enhance operation efficiency in large urban energy systems. A model was established that allows the optimization and analysis of the performance of a residential energy community with energy systems supplied by renewable sources. The design focuses on the flexibility of the system to operate with an integrated planning framework exploiting batteries, thermal storage and shared Electric Vehicles (EVs). By investigating these technologies' design and operational conditions, the project seeks to optimize the system's functioning and minimize its impact on the electricity grid to mitigate congestion. Focusing on a new residential area in Utrecht (NL), the study investigates the role of flexibility for new capacity connection challenges in a highly populated urban area with a vulnerable infrastructure. This research provides useful insights that can guide the development of more adaptable energy infrastructures capable of meeting the growing demands of modern cities. The findings aim to inform future urban energy areas to facilitate the transition towards a more sustainable residential landscape.

As part of making a power grid simulator, a hardware controller is needed. This thesis specifically describes the design of such a controller. The objective is to design it in such a way that it is self-explanatory and interactive. Therefore, the following user controls and feedback are made possible. A small solar panel makes it possible to provide input for a solar generator in the simulation. A slider gives input for a wind farm in the simulation. This is visualized by adding a small physical turbine to the hardware, going faster or slower based on the input.
A small knob to go back or forward in the simulation, if pressed, pauses or resumes the time.
A RGB LED indicates the situation on the grid. Based on the loading of transmission lines, the color of the LED changes. These user-control actions and feedback from the simulation are processed on an Arduino Mega Rev3 development board. This uses an ATMega2560 microcontroller. To interact with the simulation, bidirectional serial communication is needed between the Arduino and the computer running the simulation with the Godot game engine. These components are implemented on a PCB to finalize the design.

This thesis presents the process of design and implementation of a power grid simulator, with focus on the power grid configuration and load forecasting as part of a larger project. The overarching project aims to address the increasing complexity of power grids and the occurence of grid congestion. Its primary objective is to visualize these issues for educational purposes. For this subsystem, the specific goal is to provide a grid configuration, all necessary data to simulate it, and load predictions for the grid.

This report explains the processes behind the design, construction and testing of a high-voltage power grid simulator and visualizer. For this, a system based on data found in a paper was created. This paper describes an aggregated grid for the Dutch high-voltage power grid in 2018. The main use case of this system would be for educative purposes, where interested students and potential students can come, play and learn at the same time while making interactive decisions and seeing how their actions are influencing the system as a whole. The subsystem in this report is tasked with the integration of the data from the forecasting & scenario module and the inputs from the hardware module, furthermore it calculates the power flow using a Gauss-Seidel algorithm and visualizes its results alongside the forecasting in Godot

The energy transition towards fossil-free systems presents significant challenges in power grid management, particularly in handling distributed and intermittent energy resources. The Netherlands faces substantial grid congestion due to rapid Distributed Energy Resource (DER) integration without adequate flexibility solutions.

While Energy System Optimisation Models (ESOMs) enable robust system planning, they often oversimplify or neglect Demand Response, particularly load-shifting capabilities. This thesis validates modeling frameworks for load-shifting and proposes a novel framework for modeling aggregated load-shifting in large-scale ESOMs. This approach accurately accounts for load-shifting characteristics without requiring extensive process-specific constraints.

Case study results for the Netherlands demonstrate that load-shifting provides significant value in a fossil-free power grid by adding flexibility and reducing overall system costs, particularly in battery energy storage investments. However, controllable flexibility assets remain crucial for peak demand management and facilitating renewable energy integration. Furthermore, power grid configurations incorporating Demand Response showed greater robustness against weather and demand variations, suggesting that networks utilizing implicit flexibility are more economical and risk-averse than those relying solely on explicit flexibility solutions.

Grid congestion, which results from a larger demand for electricity transport than the available transport capacity, is a challenge in the Dutch energy transition. The growing pace of decentralized renewable energy development in less densely populated areas and the electrification of demand leads to congestion, which potentially hinders the connection of new renewable projects to the grid and slows down the energy transition.

Due to the intermittent behaviour of renewable resources, the grid connection capacity is not used to its full capacity at all times. Cable pooling is introduced as a possible solution to congestion, allowing an existing and a new renewable resource to share a grid connection and improving the utlilisation of the current grid infrastructure.

This thesis’ main objective is to develop a calculation framework to assist developers to evaluate the economic potential of a cable pooling location by optimizing the Net Present Value (”NPV”) of a shared grid connection, considering technical, regulatory, legal, and financial aspects. It aims to provide developers with a tool for making a preliminary decision on whether to continue development for a potential cable pooling location.

The Dutch context is used to identify different grid connecting possibilities, different combinations of wind and solar to form a hybrid farm and different revenue streams. These are used in a methodology to find the optimal installed capacity of the added resource and the impact of on-site storage to one of the hybrid farms. The approach considers the influence of market prices on the technology specific cable pooling business case at an hourly level and accounts for long-term market developments. Other factors accounted for in the tool are the hourly export capacity, defined as the residual space after the export of the existing farm, land size and costs of installation.

A case study of a solar farm oriented to the east-west is used to test the tool and draw conclusions on the potential of cable pooling for this case study. The additions of a wind and solar resource with either south or east-west generation all yield positive NPV values with the highest values seen for the wind addition.

Sensitivity tests for technical and economic inputs show that the results are sensitive to weather data and various economic inputs, but the results of this case study are robust. The case study however considers a relatively large grid connection. Generalizing the results by considering a smaller grid connection shows the value of complementary production patterns.

The impact on the cost-benefit framework by adding a battery is considered. The addition of a battery adds value by peak-shifting the produced energy, but not enough to cover the costs of the battery without any subsidy or alternative revenue streams.

In conclusion, cable pooling shows potential for this case study, for different revenue streams and additions. The tool used to obtain the results can be tailored to different case studies and input scenarios and shows the economical attractiveness of a cable pooling location, based on the Dutch context. The Dutch context shows a promising potential for cable pooling as a method to deal with congestion, but not many projects are present yet. Considering different parties sharing one grid connection, coming to an agreement on the terms can form a hurdle. Therefore, the introduction of more transparent cable capacity calculations by DSOs and a separate subsidy for cable pooling projects could help incentivize the development of more cable pooling projects. The impact of developments such as the ”Use-itor-lose-it” on the cable pooling potential should be monitored closely. Other solutions to dealing with congestion, such as using the fault reserves, should not be disregarded.

Electricity grids worldwide are experiencing increased peak demands and decreasing simultaneity due to higher shares of Renewable Energy Sources (RES). It is expected that many grids will soon reach their limits. One solution to mitigate these issues is exploiting flexibility in e.g. electric vehicles. In this work, a shrinking horizon model predictive controller is constructed to optimally charge and discharge EVs with respect to the day ahead electricity price or grid carbon intensity. The model takes into account that users with a dual rate electricity plan only want to charge during their off-peak hours. A feature to implement a household PV setup in the optimization is included. The possible consequences in terms of associated carbon emissions, utility costs and user costs are analysed using a simulation based on data from 4279 charging sessions that took place between June 23, 2021 and June 23, 2022. The sessions are split in 2855 weekday sessions (duration between 4 and 24 hours) and 1424 weekend sessions (duration between 4 and 60 hours). It is found that using current circumstances, minimizing the carbon emissions using bidirectional charging results in a higher price (5.5 %) for the utility than using the current state of the art, unidirectional charging minimizing the wholesale electricity cost. Bidirectional charging minimizing the wholesale electricity cost results in higher emissions compared to unidirectional charging (2.8 %), and even compared to uncontrolled charging (0.9 – 3.6 %). The reason for this seems to be a negative correlation between carbon intensity and wholesale price during the times that vehicles are typically connected although this needs further investigation to be confirmed.

Electrical faults in the distribution network can lead to interruptions in the power supply of the customers. Therefore penalties are applied to the DSOs if they overcome the benchmark set based on all the DSOs reliability performance. Hence, the fast restoration of the power supply is crucial for the grid operator in order for the operational costs to be decreased. The traditional way of fault locating is performed with the help of so-called Fault-current Passage Indicators. This can be improved by automatic estimation of the fault location using analysis of the current and voltage signals during the fault, which also depends on the accurate fault classification. However, the existence of distortions and instabilities in some of the fault waveforms leads to an unreliable fault loop-impedance/-reactance. As a result, the location of the fault has to be performed in the traditional way which leads to a delay in the restoration of the power supply.

The objective of this thesis is to explore the potential of modern signal processing and machine learning (ML) techniques for the development of an explainable classification scheme of faults in the distribution network. ML explainability is a huge part of the research held for this thesis, as it is very important for the fault analysis department to understand the model and for the grid operators to trust the results of the model. The problem is divided into two parts. The first part concerns the construction of an explainable ML-based classification model that can accurately differentiate between types of stable faults. The second part concerns the identification of suitable criteria for single-phase and multi-phase fault stability. Combined with a fault classifier based on the first part research, this will lead to a classification and location scheme that can incorporate all fault types.

First, a literature study is done. The pre-modeling explainability stage is the most suitable one and based on that, the Short-time Fourier Transform (STFT) is chosen as the signal processing technique since it can lead to more explainable features than other techniques, e.g. the Wavelet Transform (WT). Also, the Support Vector Machines (SVMs) are selected as the supervised learning technique, as it performs well in classification problems and it is not a method as complicated as for example the Artificial Neural Network (ANN). Next, three different feature sets are shaped appropriately from the three-phase fault current and voltage signals, which are the inputs for the SVM classifier model, so then this model can be tuned and evaluated. The model with the symmetrical feature set is selected serving suitably both the performance and explainability requirements. Following that, a set of stability rules are developed to better identify the stable faults that previously were classified as unstable. These rules are tuned based on a set of initially labeled unstable faults. In the end, the ML classification and the stability model combined with section selection rules constitute the final classification scheme. This is tested on hidden data and the results indicate good performance in classifying faults, locating faults and identifying stable faults that initially would be classified as unstable. In particular, the performance of the proposed scheme on that test set has f1-score of 95.1% exceeding 93.3% of the current algorithm tested on the same data, errors regarding the fault loop-impedance/reactance are located within the acceptable limits and 4 stable faults correctly identified that previously were seen as unstable.

Reinforcement learning (RL) has grown tremendously over one and a half decades and is increasingly emerging in many real-life applications. However, the application of RL is still limited due to its low training efficiencies and surplus training cost. The sampling and computation complexity normally depends on the size of the state space and splitting the state space can distribute computation and accelerate learning. State abstraction as a form of data-centric method shrinks the state space and reduces learning time, however, it is challenged by the fact that abstraction throws away information and might result in a sub-optimal solution. In this thesis, we propose the hierarchical clustering-based state grouping (HCSG) method to split the ground state space into clusters and train multiple agents for each cluster without changing the dimension of the state space. This approach allows us to distribute computation and improves training efficiency without losing the overall performance, and was also shown to outperform baseline and other state-of-art data-centric methods.

Batteries are an essential tool for energy transition. They provide the ability to reduce imbalances by absorbing the forecast errors for consumers and renewable energy sources. This has the added benefit that congestions occurring due to supply and demand mismatches would be prevented. Furthermore, they allow generating a stable power consumption for a longer period by utilizing the available reserves. The electricity markets are all power markets, where the products grant an amount of power for a certain period. Therefore, the introduction of a battery would make it easier to buy power for a portfolio of consumers. Currently, it is not yet feasible to use a battery for the consumer segment as the revenues generated are simply too small to pay the battery back before it reaches the end of its lifetime.

This report aims to describe a machine learning-based battery operating algorithm that can be utilized for the consumer market. The operating algorithm will be trading on the day ahead and the voluntary aFRR market while satisfying the demand of the consumer(s). The trading is done based on the forecast generated by different machine learning models. For the day ahead market, two forecasting models are tested, namely an XGBoost algorithm and a Dense Neural Network. For the imbalance market, a simple Dense Neural Network and a Long Short-Term Memory (LSTM) are compared and for the consumer forecast, an XGBoost model is compared to a Lasso. All forecasts are combined to create a profit-maximizing algorithm. Lastly, a comparison is made between having an individual battery for every homeowner and aggregating a group of consumers together in a larger battery is compared.

The results show that forecasting an aggregated group of consumers has a significantly lower error compared to forecasting individual demand patterns. The large errors in the individual forecast result in a large opportunity cost, based on the Value of Lost Load (VoLL), which renders the benefits of owning a battery useless again. However, the aggregated group of consumers combined into a single large battery is profitable. Considering the current prices (22Q1) the battery generated enormous profits, such that the battery would be break-even in less than 2 years.

Whereas in the past, the Distribution System Operator (DSO) almost never encountered congestion in their grids, nowadays, with the increase of connected renewable energy sources, this will become more prevalent. To forecast congestion on transformer stations, with the goal of mitigating it, the Dutch DSO Stedin uses machine learning models with standard loss functions for regression. These fall short in predicting congestion peaks, as loss functions like MSE are unaware of the prediction goal, which is minimizing the cost associated with the congestion. Without knowledge of this goal, a loss function will not put any extra importance on finding congestion peaks and is more likely to miss them.

In this thesis, the use of a a custom loss function is explored, which incorporates the different costs associated with congestion mitigation. This function aims to improve upon existing congestion forecasts by having a loss function in line with the congestion forecasts goal. For this research, the case of substation Middelharnis is used. This substation encounters congestion due to connected wind parks and distributed photovoltaics. The congestion encountered will be solved by grid expansion in 2024, but until then will have to be mitigated using the redispatching market GOPACS.

A custom function was generated using two cost components: a congestion fee, which needs to be paid if wind parks are disconnected and GOPACS costs, which are the price of buying flexible power on the redispatch market. The function takes into account two-time horizons: costs associated with buying GOPACS power day-ahead acting on a prediction of the load, and the resulting cost on the day itself if any remaining congestion was resolved via the congestion fee. The contribution of this function is that it depends on the difference between the costs of the prediction and the cost of the realization, instead of the cost of the difference between prediction and realization.

Four models were trained, each with a different loss function: MSE, Pinball, and Cost. The cost model is trained twice and for each model, a different value for GOPACS cost is used. The results show that the cost models outperform the Pinball and MSE-trained models, when the performance is evaluated on the final cost metric. However, if the GOPACS price is high, the cost model becomes conservative in predicting congestion peaks. It is not expected that this will be an issue in the future, as with more renewable energy, the energy price and subsequently the GOPACS price will be lower.

Electrical faults in the distribution system can lead to interruptions in customer power supply resulting in penalties that are borne by the distribution system operator. Accurate fault classification is an important step in locating the fault to achieve faster network restoration times. This reduces the operational costs of the system operator by reducing the duration of interruptions in the power supply. Furthermore, it has been found from practical experience that distortions or instabilities in fault waveforms can result in their classification and subsequent localisation being delayed, causing the loss of valuable customer minutes. The objective of this thesis is to investigate the potential of modern supervised learning techniques for the classification of faults in the distribution network. The problem is split into two parts: one aspect is to study the ability of a supervised learning classifier to differentiate between types of stable faults, and in this pursuit, to also identify better criteria for stability.

First, after a review of pertinent literature, it is found that discrete wavelet transforms and support vector machines (SVM) are suitable for fault signal processing and classification, respectively. Next, identifying characteristics, or features, from the three-phase fault current and voltage waveforms are engineered with the help of the db4 wavelet. The features are used as inputs to an SVM classifier model, and the model is tuned and validated to ensure optimal classification performance. Results showed that the developed SVM can differentiate between real-world instances of single-phase, two-phase and three-phase stable faults with a classification accuracy of 95%. A set of business rules are also developed to characterise instability by performing a windowed Fourier analysis and studying the strength of the fundamental frequency component of fault waveforms. The rules are tested on a set of fault data whose stability is uncertain, and it is found that the developed rules are able to improve upon the older method of stability analysis by increasing the rate of stable fault identification.

Transition from fossil fuels to sustainable sources of energy like wind and solar is the need of the hour. All over the globe, plans are in motion to achieve this goal. This implies addition of new elements to the grid in the form of Distributed Energy Resources (DERs). These affect the working of distribution grids and to ensure reliable as well as safe operation, it is important to keep a track on the grid’s state regularly which is essential to a Distribution System Operator (DSO). For this very reason, Distribution System State Estimator (DSSE) has been introduced and has been a prominent topic of interest which has been discussed in literature over the past two decades. Because of lack in observability of the network owing to unavailability of measurements and the stochastic load profiles of the distribution network, DSSE poses its own challenges. For this very reason, it is necessary to validate the working of a suitable DSSE that is affected by the continuous changes in the grid. By selecting a suitable algorithm, this thesis attempts to solve the observability issue by introduction of pseudo-measurements. The work in this thesis comprises of sensitivity analysis of Weighted Least Squares(WLS) algorithm tested on two networks in the Netherlands; a synthetic model of an anonymized distribution network and a part of Stedin’s distribution network which has limited measuring devices data available. The percentage of pseudo-measurements is varied to determine the state estimator’s accuracy. Also the statistical validity of the algorithm against measures like bias and consistency is determined with the help of scenarios generated by using Latin Hypercube Sampling (LHS). The results obtained prove the effectiveness of the selected algorithm for DSSE and are important for the DSOs to make critical decisions when needed for grid operation.

The growing fleet of offshore wind farms in the North Sea imposes a lot of stress on the energy grid due to its volatile power production. Accurate forecasting models that can predict the power generation of these wind parks could help grid operators to keep the energy grid stable and reduce the re-dispatch of energy. Therefore the objective of this thesis was to find how to make an accurate power forecasting model for an offshore wind turbine park. To find the most suitable model, five different models were made. The first model tested was a physical model which calculates the power output through a deterministic approach. Four other models were made through the use of machine learning algorithms. A Gradient Boosting model, a support vector regressor, a feed-forward neural network, and a recurrent neural network were built to find out which one was most suitable. The next aspect looked at was what input data could be beneficial to make the power predictions, the numerical weather prediction model ICON-EU was chosen as input data. The support vector regressor algorithm showed the best results with a score of relative mean absolute error 14.35\%, which is the product of the mean absolute error divided by the difference between the minimum and maximum power production of the period in question. Since the performance of the model met the requirements it can be concluded that the forecasting model can be beneficial for the grid operator to reduce the amount of re-dispatched energy.

The Distribution System Operator (DSO) Alliander has the ambition to explore possibilities of improving its predictive demand modelling applications. This research aims to contribute to one of these, the Advanced Net DEcision Support (ANDES) model, which provides detailed predictions for long-term capacity planning. This thesis pursues two main objectives: Firstly, the creation of a probabilistic power demand model which reflects the volatile nature of real customer demand in an adequate manner. Secondly, the development and evaluation of methods to estimate certain model output quantities of interest in a computationally efficient manner. To this end, variance reduction techniques have been investigated. Self-tuning Importance Sampling (IS) methods giving different weight to time steps and/or customer load profiles have been developed. In order to make the optimisation stage superfluous, additionally an approach to find a generalised asset distribution has been investigated. The essential finding from evaluating all considered methods was that their performance in terms of efficiency and accuracy depends mainly on two variables – the order of magnitude of the estimated quantity and the number of customers connected to an asset in question. For small assets and an estimated overload probability of the order 1e-5 or smaller, all profile IS methods and especially the generalised bin probability IS showed the strongest performance with average speed-ups of 5-30 times with respect to the reference method of sampling full annual traces. For assets with more than 80 customers and small overload probabilities, the profile IS methods were found to frequently produce estimates of a significantly too small order of magnitude. Conventional Monte Carlo (MC) sampling and time step IS, in turn, produced reliable estimates regardless of the number of customers. For assets of all sizes with an estimated overload probability of the order 1e-4 and larger, conventional MC sampling showed the best performance with speed-ups above 5 times. Overall, conventional MC sampling performed robustly in all circumstances, while IS demonstrated its potential to significantly increase the estimation efficiency of rare event probabilities in certain cases. To determine which magnitudes demand maxima and minima can potentially reach, Extreme Value Theory (EVT) has been applied. The computationally more efficient methods and extreme value inference were considered not compatible, sampling full annual traces appears to be required for a reliable estimation of maximum and minimum demand return levels. Based on the findings of this thesis, a flexible algorithm could be investigated in future research which employs IS for rare event probabilities and conventional MC sampling otherwise. For an integrated evaluation of all risk metrics, the algorithm could initially sample 200 entire annual traces to be used for extreme value inference.

Advancements in the field of Information and Communication Technologies (ICT) has enabled the possibility to utilize the flexibility offered by responsive assets in a better way by employing Demand Response (DR) schemes. This thesis analyzes the performance of one such DR scheme developed at TU Delft called Forecast mediated Market Based Control (F-MBC), which aims to coordinate such flexible assets by communicating "self-fulfilling forecasts" [1]. The main aim of the project is to investigate the applicability of this method in real-world settings. To do so, several simulation scenarios were formulated to understand how well F-MBC coordinates heterogeneous populations of uninterruptible time shiftable loads over an extended time horizon, both from the system perspective and devices' perspective. The thesis also proposes an approach to test the mechanism in a rolling horizon setup. First, the performance of F-MBC is examined under several combinations of deferrable loads having identical deadlines. Then, its ability to coordinate devices with dynamic load profiles under a complex realistic setting is investigated. Trade-offs adopted when simulating such a setup is also highlighted. Results indicate that while F-MBC achieves good overall performance when coordinating devices with uniform power consumption profiles, its performance in scheduling heterogeneous populations of devices with dynamic load profiles was quite variable. When devices that consume high power when they start was considered for coordination, F-MBC was able found to allocate the devices in such a manner that steered towards overall cost minimization. However, its performance if used to schedule devices which consume low power when it starts was found to be undesirable. Hence, several recommendations were provided to deduce better conclusions about the applicability of the mechanism in reality. References: [1]Hazem A. Abdelghany, Simon H. Tindemans, Mathijs M. de Weerdt, Han la Poutré, Distributed coordination of deferrable loads: A real-time market with self-fulfilling forecasts, Sustainable Energy, Grids and Networks, Volume 23, 2020, 100364, ISSN 2352-4677. http://www.sciencedirect.com/science/article/pii/S2352467720302952