Renewable Energy Sources (RESs) such as wind and photovoltaic are environment-friendly energy sources for power generation. However, the largely varying output energy of RESs is a major obstacle to their integration into the power grid. In addition, variations in consumer activity increase the risks of power fluctuations. The ability of power systems to deal with power fluctuations caused by renewable sources and dynamic demand has to be improved to keep the stable operation of a power system. This paper proposes a novel robust energy balancing concept to reduce energy imbalance due to the mismatch between fluctuating generation and demand. The proposed concept targets the long-term operation of a power system and aims to guarantee its safe operation under any dynamic fluctuations of RESs and demands. The key ideas behind this concept include the periodic power level patterns of fluctuating sources and loads, and their upper bound envelops and lower bound envelops, which enable us to analyze the long-term system behavior by the analysis of one cycle operation. The extracted non-trivial upper bound envelops and lower bound envelops provide us with a novel interpretation of the collaboration between storage systems and controllable sources/loads in power balancing, as well as their minimum sizes.
This paper presents an optimisation method for the direct energy exchange between two electric vehicle (EV) charging stations located in the UK. Each EV charging station consists of solar panels, hydrogen and battery energy storage systems (SHBS). The stations are interconnected through a energy exchange system that enables the transfer of excess energy from one station to the other. The objective function of SHBS charging stations is to minimize the capital and operation and maintenance costs of the stations. The system constraints are the power output of individual components, as well as the power balance between SHBS charging stations and EV charging demand. Genetic Algorithm is used to optimize the system, considering various factors such as the size of the solar panels and hydrogen storage tanks, the capacity of the electric vehicle chargers, and the amount of energy exchanged between the two stations. The optimized system yields substantial cost savings.
The high penetration of converter-interfaced renewable energy sources reduces the inertia levels of modern power systems, jeopardizing grid stability. Therefore, to ensure the safe and reliable operation of their grids, power system operators shall monitor close to real-time the overall inertia levels. Towards this objective, in this paper an inertia estimation technique, based on the modal sensitivity concept, is formulated and validated on heterogeneous multi-machine power systems. Several cases are considered and examined to demonstrate the applicability range of the developed method.
The availability of realistic grid topologies is of central importance for the usability of grid models in calculations. However, especially in distribution grids, poor documentation, limited digitization and insufficient data quality can lead to faulty topologies, which result in errors in subsequent calculations. Methods of topology identification (TI) offer possibilities to use measurement data of grid quantities such as voltages to validate and, if necessary, correct the topology used for calculations. This paper presents robust methods for TI in medium voltage networks based on an adapted form of Prim's algorithm, which is based on measurements of node voltage magnitudes. Methods to include spatial priors in the process are investigated. The developed method is applied to realistic models of medium-voltage networks and its robustness to unadapted graph-based methods is shown. The correlation of injection correlations, as well as noise on the measured data, is addressed to show that the proposed methods can handle these conditions.
This study deals with the implementation of recorded lightning current waveforms in ATP-EMTP. These can be employed as excitation in fast-front transient simulations. A generalized code is developed in MODELS language to import arbitrary waveforms expressed in the form (time, voltage/current) pairs. An investigation is presented on the influence of the time interval used for digitizing recorded lightning current waveforms and of the interpolation method applied between points. ATP-EMTP simulations of direct lightning strikes to 150 kV overhead transmission lines are performed. Fast-front overvoltages are computed across line insulators and the minimum backflashover currents are determined. Cases with Line Surge Arresters (LSAs) are also treated and the energy stressing the LSAs is computed. It is shown that the digitization time interval is not important as long as it reproduces the main features of the recorded waveforms. This applies also for the interpolation method.
This study presents the investigations of visual images originating from various corona discharge forms occurring under +DC and −DC voltages. The experimental setup included two hemispherical tip rod electrodes with a tip radius of 2 mm, and the gap spacing between them was set to 5 cm to observe corona discharge forms. The voltage was increased in steps, and a corona camera was used to determine the corona inception voltage. The transitions between different forms of corona discharges were observed. Images of corona discharges were taken at each voltage level using a digital single-lens reflex (DSLR) camera. Furthermore, the pulses generated by the corona discharges were examined using a high-frequency current transformer (HFCT) and a shunt resistor. The main objective of this investigation was to analyze the severity of corona discharge on the system based on the obtained corona images. Also, the usability of the HFCT and shunt resistors as measurement sensors was discussed.
Due to the harsh environmental issue and hard accessibility, offshore wind turbines (WTs) have more challenges for operation and maintenance (O&M). Thus, it is crucial to develop effective condition monitoring (CM) methods for WT fault prediction to detect incipient faults before their occurrences, thus preventing durable downtimes. In this paper, eight specific faults are classified for fault prediction using status information from Supervisory Control and Data Acquisition (SCADA) data. The classification steps are based on fault prediction from 10 to 210 minutes prior to faults. By embedding a model-agnostic vector representation for time, Time2Vec (T2V), into Gated Recurrent Unit (GRU), a novel deep learning neural network model, T2V-GRU, is applied for fault classifications. As a result, T2V-GRU successfully predicts over 84.62% of faults and outperforms its counterpart, vanilla GRU, in both overall and individual fault predictions in terms of accuracy, recall scores and F-scores.
There are many global factors that are challenging the colossal transition to Zero Carbon Economy, ranging from regional conflicts, possible new cold wars, inflation to rising interest rates. The climate challenge is, de facto, an energy transition challenge, which historically takes generations. Governments all over the world are working to implement policy that encourages society to foster clean energy and low-carbon technologies. It is a fine balance between supply and demand of energy networks, whilst maintaining energy security. This was evident in Ireland during the winter of 2022 which witnessed several Systems Alerts, from the Transmission System Operator (TSO), EirGrid, mainly due to sub-zero temperatures, low wind, and system capacity. In line with the European Union, member states are moving towards the decarbonisation of energy systems. This will require a holistic behavioural change in the way society provide, transport and consume energy. The large scale expansion of low-carbon technologies, namely Air Source Heat Pumps (ASHPs) and the electrification of buildings, using a low-carbon intensity electricity grid is generally accepted and predominantly uncontroversial. This paper aims to analyse datasets produced from a Power and Energy Data Logger which consisted of time series data recorded at ten minute intervals from two different load sources. The first dataset monitored an ASHP's electrical energy and the second dataset monitored a residential building's electrical energy over a period in winter 2022. This data allowed the author undertake comparative analysis between different scenarios, such as the ASHP's electrical consumption load profile and the Outside Air Temperature (OAT). Furthermore, comparisons were made between the TSO's demand profile and the building's new electrical consumption load profile incorporating an ASHP. This paper's main findings are that ASHP's electrical energy profile fluctuates considerably throughout the day, due to continuously changing OAT. Finally, the comparative analysis between the actual heat pump data collected and the previously predicted profile shows clear variations between the two models.
Grid connected inverters can suffer harmonic instability when connected to the grid due the interaction between the converter output impedance and the grid impedance seen at the point of common coupling (PCC). To counteract this it could be useful if the inverter were able to measure the impedance at the PCC over a wide frequency range. This can be done by injecting a current disturbance signal to the grid and measuring the resulting voltage disturbance. This paper investigates the use of a chirp signal injection in combination with Welch's method for signal analysis, as a means to measure the grid impedance up to 2 kHz. The accuracy of the approach is first investigated using MATLAB/Simulink simulations for a single phase inverter. The results demonstrate a good ability to accurately determine grid impedance with relatively low amplitude chirp signal injection. Experimental results show that the amplitude of signal injection needs to be higher in order to improve accuracy. There is however a trade-off between accuracy of estimation and total harmonic distortion of the current.
The Department for Business Innovation and Skills (BIS) in the UK has recognized ensuring a good supply of talented Power Electronics engineers as a challenge. Inability to recruit high-quality engineers would drive companies out of the UK. The use of outdated or inappropriate curriculums at universities has been identified as a gap to address this challenge. Some academic institutions have well-recognized power electronics, machines and drives (PEMD) programs where their undergraduate courses are also linked to their research interests. However, other academic institutions do not provide that depth of knowledge required by the PEMD industry, considering it as optional knowledge and do not have suitable training materials. This paper reviews the current state of Power Electronics curriculums and contribute to filling the gaps in skills, talent and training for the PEMD industry by developing a framework for academic curriculum, which is supported by industrial-oriented knowledge and inputs. The developed framework has been designed to fit other disciplines also achieving wider awareness.
A purpose built battolyser, using the combined technologies of a battery and an electrolyser, has been developed for energy storage and hydrogen production at a lower cost than current electrolysis technology. Various acidic conditions and electrode materials were used to evaluate the performance of each battolyser configuration. Only low-cost, abundant, low toxicity and low environmental hazard materials were selected. Performance was evaluated by durability and degradations tests and the results demonstrated that a vanadium redox flow battery configuration could achieve high hydrogen yields, good efficiency, and electrical storage compared to an iron flow cell and sodium-ion manganese hybrid cell. Materials deposited on the electrodes after cycling were characterized using power x-ray diffraction.
This paper solves the Non-Intrusive Load Monitoring problem by using two machine learning based models: Xgboost and Recurrent Neural Network. We utilize and develop models using a publicly available dataset. To improve the performance we have implemented hyperparameter optimization using grid search. The numerical simulation results show that proposed Xgboost model outperforms the RNN based model. With the implementation of hyperparameter optimization an improved numerical accuracy is obtained.
This paper proposes a system, CosyGrid, to achieve the objective of Renewable Energy Communities (RECs), such as maximizing self-consumption, through Peer-to-Peer (P2P) trading. In P2P trading, each order from the peer is matched with an order (or orders) from any other peer (peers) in the same REC marketplace. But it is often not possible to fill all orders of peers with matching orders from the other peers. Those orders from the peers that remain unfilled at the start of the delivery period but cannot be cancelled (i.e. are not flexible) are matched with the peer's retailer as per the supply agreement, based on actual volumes imported or exported. In this proposed design for CosyGrid, contracts are instead settled based on contracted volumes of consumption and production and the difference (imbalance) between actual and contracted volumes, and the retailer contracts for unfilled orders are settled at a higher spread between buy and sell price than the supply agreement. These adapted supply agreements allow for stronger p2p price signals and are referred to in CosyGrid as Framework agreements. But the resulting difference between supply and Framework agreements can negatively impact the effectiveness of P2P trades to achieve the REC's objective. To minimize the impact, CosyGrid settles imbalances such as to ensure that the aggregated value in Framework agreements is the same as the aggregated value in Supply agreements. A price per unit of imbalance called the Imbalance price, is calculated, and applied to every end-user proportional to the individual imbalance. The proposed approach will incentivize end-users to respond to price changes and motivate them to use accurate forecasts for their orders, which will further improve the effectiveness of P2P trading.
In this paper, we showed that using the nonlinear optimization to estimate the PAR(p) model coefficients can, sometimes, improve adherence to historical data when generating synthetic inflows, which is suit for the Stochastic Dual Dynamic Programming implemented by NEWAVE in the Brazilian energy operation planning. The used solvers were IPOPT and the Python toolbox SciPy. The accuracy of the model was checked by generating inflows via Monte Carlo simulation and outperformed, in some cases, the well established Box-Jenkins algorithm for obtaining these coefficients.
Signalized intersections are significant spots of energy consumption because of frequent stop-and-go behavior. Eco-driving aims to reduce energy usage by optimizing driving behavior. Researchers have reviewed optimization-based method while lack of them reviewed the learning-based approaches. This work critically reviewed two different types of approach. In addition, one well-known rule-based car-following model and two state-of-the-art optimization-based and learning-based methods are selected to test in a signalized intersections environment with the metrics of energy consumption, travelling time and algorithm execution time. The experiment results show that the travelling time of three algorithms are similar, while the energy consumption of the learning-based method and optimization-based method are 30.72% and 51.82% less than that of the rule-based method respectively. However, due to algorithm execution time, the optimization-based method is not suitable to be used in real-time.
This paper investigates the effectiveness of ensemble modelling for time series forecasting using Autoregressive Integrated Moving Average (ARIMA) models. In recent years, ensemble modelling has become a popular approach for improving forecasting accuracy by combining multiple models to achieve better performance than individual models. However, there is still limited research on the effectiveness of ensemble models for time series forecasting using ARIMA models. In this paper, we tested simple averaging of ARIMA models and investigate their performance in comparison to individual models. We conducted experiments using real-world datasets and evaluated the models' performance using metrics such as Mean Absolute Percentage Error, and Root Mean Squared Error. In this paper, we provided experiments on both short and long datasets to evaluate the performance of ensembled models compared to individual models. For the short datasets, our results clearly demonstrated the advantages of using ensembled models over individual models. The ensemble of models consistently outperformed the individual models in terms of accuracy. Our findings suggest that ensemble modelling can be a tool for time series forecasting and can provide improvements in accuracy. By leveraging the strengths of different models, ensemble models can effectively capture the underlying patterns in the data and make more accurate predictions.
A real-time troubleshooting guide was developed in this paper for the maintenance support system hydropower plant. Root cause analysis of emergency cases has to use real-time data in the application program interface (API) via the Internet of Things (IoT) and Accumulated knowledge of power plant logic, drawing, equipment manuals, test reports, corrective maintenance reports, and history events. After that, develop a root cause analysis using the fault tree analysis (FTA) method and the maintenance and operation team's expertise to diagnose alarms and emergency situations. after which create the platform for a real-time troubleshooting guide. Finally, all data will prepare for the logic flow and graphic user interface design by Node-red program and dashboard in troubleshooting guide pattern for assist in the decision and solve emergency problem events more effectively.
Active distribution networks (ADNs) are increasingly assuming an important role in future power system operations. Due to incremental phasing out of thermal power plants, a shift of ancillary services provision from the renewables is underway. Therefore, increased focus on the renewable rich distribution grid level is of prime importance. Active and reactive power flexibility (PQ-flexibility) quantification from the underlying distribution grid at the vertical interconnection to the overlaying grid is a topic of current research. A two dimensional PQ-flexibility map at the vertical interconnection serves as a basis for flexibility provision between grid operators. A terminology adapted in current research is the Feasible Operating Region (FOR) of the underlying distribution grid. The task of flexibility aggregation is further complicated when renewable power injection uncertainties are considered. The two dimensional PQ-flexibility map or FOR requires adjustments considering the probable generation scenarios. Therefore, a reliability parameterized flexibility aggregation segregated into confidence intervals is practical. The undertaken study adapts a method for generating spatially correlated renewable generation uncertainties from wind power plants (WPP) and photovoltaic generation. A corresponding statistical analysis is performed for a reliability parameterisation of the PQ-fexibility seggregated into confidence intervals. Subsequently, a FOR determination adhering to the determined confidence intervals is proposed. Results present multiple reliability parameterized two dimensional PQ-flexibility maps, classified according to the confidence intervals.
Clustering techniques play an important role in analysing complex networks such as electrical power systems grids. They can help in identifying congestion bottlenecks and other planning and operation activities. This paper presents a clustering method for partitioning a power network by grouping the nodes and edges based on their reachability to or from the various sources and sinks in the system. These clusters can be discussed using familiar terminology from river networks, such as tributaries and distributaries. A goal is to concretely visualise how nodes and edges are related, and their operational dependencies within the network. The clustering results may give power system operators new insights into the system's structure, enabling situational awareness, fault/vulnerability analysis, and planning. The proposed methodology is applied to two sample grids to demonstrate the types of clusters it can identify.
In this work, different battery control strategies are modelled, simulated and applied for the optimal operation of a sector-coupled distribution grid. For the evaluation of the impact on the electrical grid, a Model Predictive Controller (MPC) is chosen as a predictive control algorithm which is compared to a typical non-predictive controller. The resulting line and transformer loading of a centralized and a decentralized MPC approach, as well as a non-predictive, heuristic controller method, are analyzed. The grid equations and technical limits are implemented in the centralized optimization. The minimization of the quadratic power exchange with the medium voltage grid is chosen as the objective function to achieve peak shaving. For the selected benchmark grids from the literature, it has been shown that the central MPC achieves the best performance with respect to avoiding grid overloads and the optimal usage of flexibilities.
