Showing papers by "Jovica V. Milanovic published in 2014"

Probabilistic Framework for Assessing the Accuracy of Data Mining Tool for Online Prediction of Transient Stability

Abstract: The paper presents a generic probabilistic framework for assessing the accuracy of online prediction of power system transient stability based on phasor measurement unit (PMU) measurements and data mining techniques. It allows fair comparison of different data mining models in terms of the accuracy of the prediction. To illustrate the concept, a decision tree (DT) method is used as an example of a data mining technique. It is implemented in a 16-machine, 68-bus test power system. Generator rotor angles and speeds provided by PMUs during post-fault condition are chosen as predictors. The performance of the DT based prediction method is tested using a wide variety of disturbances with probabilistically modeled locations, durations, types of fault and the system loading levels. The accuracy of prediction is approximately 98.5% immediately following the fault clearance and can increase to almost 100% if the prediction is made 2.5 s after the fault clearance.

Tuning of a Damping Controller for Multiterminal VSC-HVDC Grids Using the Probabilistic Collocation Method

Abstract: This paper presents a robust probabilistic controller tuning method to improve the damping of critical system modes through the modulation of active power injected by a voltage-source converter-based multiterminal high-voltage direct current (VSC-MTDC) grid. This methodology first establishes the probabilistic locations of the critical modes based on the known variation in power system operating conditions. Following this, the modal linear quadratic Gaussian (MLQG) controller structure is tuned for a set of probabilistic values of critical eigenvalues. The controller's performance following small disturbances in the network for wide-ranging operating conditions is compared with the conventionally tuned MLQG controller designed for the nominal system operating point. The probabilistic collocation method is shown to facilitate robust probabilistic tuning without the need for large numbers of full system linearizations. The test system used incorporates a large wind farm with variable power output connected to the meshed ac network through the VSC-MTDC grid.

International Industry Practice on Power-Quality Monitoring

Abstract: Monitoring of voltages and currents at system buses gives the network operators information about the performance of their network, both for the system as a whole and for individual locations and customers. There is also demand from the customers and the regulatory agencies to provide information on the actual power-quality (PQ) level. Developments in enabling technology have made it possible to monitor at a large scale and to record virtually any PQ parameter of interest. While many network operators are installing monitoring equipment and while more and more monitors are available, there is a lack of knowledge and agreement on a number of aspects of the monitoring process and on processing the recorded data. As a response to this lack of uniformity in approach, data acquisition, and processing, in February 2011, CIGRE and CIRED established the Joint Working Group C4.112: “Guidelines for Power quality monitoring-measurement locations, processing and presentation of data.” In order to identify the current international industry practice on PQ monitoring, the group carried out a survey in 43 countries across the world. This paper summarizes the key findings from 114 responses to the questionnaire and identifies prevalent industrial practice in PQ monitoring around the world.

Probabilistic small-disturbance stability assessment of uncertain power systems using efficient estimation methods

Abstract: This paper presents comparative analysis of the performance of three efficient estimation methods when applied to the probabilistic assessment of small-disturbance stability of uncertain power systems. The presence of uncertainty in system operating conditions and parameters results in variations in the damping of critical modes and makes probabilistic assessment of system stability necessary. The conventional Monte Carlo (MC) approach, typically applied in such cases, becomes very computationally demanding for very large power systems with numerous uncertain parameters. Three different efficient estimation techniques are therefore compared in this paper-point estimation methods, an analytical cumulant-based approach, and the probabilistic collocation method-to assess their feasibility for use with probabilistic small disturbance stability analysis of large uncertain power systems. All techniques are compared with each other and with a traditional numerical MC approach, and their performance illustrated on a multi-area meshed power system.

Reliability Based Framework for Cost-Effective Replacement of Power Transmission Equipment

Abstract: This paper presents a probabilistic framework for making decisions about the replacement of ageing power equipment. The framework involves three steps: first, to identify the most important and critical components of the system for overall system reliability; secondly, to perform Pareto analysis to relate the replacement of the components to the effect on system reliability indices; finally, to determine the optimum scenario for replacement based on a comparison between the cost of unreliability due to deferring the replacement and the saving on reinvestment cost. The proposed approach is illustrated on a meshed test system modeled using U.K. transmission system parameters, a representative transformer age profile and regulatory energy not supplied values. The results demonstrate the feasibility of the framework for application in the area of power system reliability, and show its feasibility for informing replacement decisions.

Probabilistic Indicators for Assessing Age- and Loading-Based Criticality of Transformers to Cascading Failure Events

Abstract: The paper investigates the effect of a transformer unplanned outage on the failure probability of the remaining transformers in the network. The probability of consequential dependent failures of transformers might ultimately lead to a multiple or cascading failure. New probabilistic indicators based on the transformer's unavailability under outage conditions have been formulated for individual power transformers and transformer sites to help with this assessment. The effect of thermal stress of transformer on its unavailability due to the new loading condition is taken into account as well as the age and the original loading of the transformer. The Arrhenius-Weibull life-stress model was adopted for assessment of the transformer unavailability. The study identifies both transformers whose outage could initiate cascading failure and those that are the most vulnerable to consequential failure. The effectiveness of the proposed probabilistic indicators in identifying the most critical transformers for multiple failure events is demonstrated on a realistic transmission test system with 154 power transformers.

Guidelines for Power quality monitoring - Results from CIGRE/CIRED JWG C4.112

Abstract: This paper summarizes major results of the CIGRE/CIRED JWG C4.112 - “Guidelines for Power quality monitoring - measurement locations, processing and presentation of data”. The results presented in this paper cover the work of JWG from January 2011 to January 2014. The paper provides information about the current PQ monitoring practices and makes recommendations for future monitoring campaigns considering monitoring locations, selection of monitoring parameters, and presentation of monitoring results.

Accuracy of ANN based methodology for load composition forecasting at bulk supply buses

Abstract: Accurate prediction of load composition at bulk supply points can significantly improve power system planning, electricity market analysis and demand side management. This paper discusses an artificial neural network (ANN) based approach to forecasting load composition at the bulk supply bus based on RMS measurement of voltage, real and reactive power and local forecasted weather. Probabilistic distributions and confidence levels of the prediction under different prediction error intervals have been derived and analysed. It is demonstrated that the approach yields prediction of load composition with errors typically less than 10%.

Practical Implementation of Coordinated Q-V Control in a Multi-Machine Power Plant

Abstract: This paper presents details of the practical implementation of a coordinated reactive power-voltage controller (CQVC) developed for a multi-machine steam power plant. The CQVC performs optimal coordination of the synchronous generators' reactive power outputs in order to maintain the total reactive power delivered by a steam power plant (SPP) or the voltage at a steam power plant HV busbar. This is required to achieve slow preventive control of generated reactive powers in order to maximize reactive power reserves along with slow high side voltage control. The CQVC is designed to act through existing automatic voltage regulator inputs, thus no modification of an existing excitation system was made. In order to increase the robustness to changes in operating conditions and to achieve the desired shape of the reactive power response, the two-step predictor corrector method is implemented. A newly developed algorithm for steady state detection ensures non-interference of reactive power voltage control with excitation system response. The paper also presents practical design solutions to overcome limitations posed by application to a real system. Acceptance and functional tests carried out during the commissioning of the CQVC are also discussed in this paper.

Methodology for Evaluation of Risk of Subsynchronous Resonance in Meshed Compensated Networks

Abstract: This paper proposes a methodology for evaluation of risk of subsynchronous resonance (SSR) in meshed compensated AC networks. The developed methodology takes into account the severity of SSR problem in different network configurations caused by fixed series compensation and probabilistic behavior of the power systems due to random outages of the lines. SSR risk evaluation considers severity of SSR problem, probabilities of different network configurations for different contingencies and different operating conditions of turbine generator. Probability of each contingency is determined through line outage model while the probabilities of different operating conditions are calculated from multilevel load duration curve. Using developed risk matrix, it is demonstrated that the degree of risk of SSR that the generators in the network are exposed to can be evaluated for different network configurations, different contingencies and different line compensation levels.

Identification of power system dynamic signature using hierarchical clustering

Abstract: This paper applies Hierarchical Clustering to identify the dynamic signature of power system, within database of post-disturbance system responses obtained by Monte Carlo simulation. Two different approaches are proposed to cut off the dendrogram so that generators can be grouped based on the similarity of their rotor angle behavior for a large number of contingencies automatically. The application of the method is illustrated on a 16-machine, 68-bus test system. Hierarchical Clustering provides accurate results in terms of generator grouping. 8 patterns of system responses are identified from the database. This work can be used to label the training data in the problem of on-line prediction of dynamic signature.

Potential interactions between VSC HVDC and STATCOM

Abstract: This paper analyses the dynamic behaviour of a power system with both FACTS and VSC HVDC, in particular, possible potential interactions between a STATCOM and a VSC in a point-to-point HVDC link. The investigation considers different STATCOM locations and VSC control strategies. A generic linearized mathematic model is firstly developed, where a combined method involving relative gain array (RGA) and modal analysis is applied to identify the interactions within the plant model and the outer controllers. The interactions identified are further analysed by creating a set of scenarios integrating both the STATCOM and VSC HVDC into a dynamic equivalent realistic AC system. Results show a collaborative operation between the STATCOM and the closely located VSC with reasonably tuned controllers.

CIGRE/CIRED JWG C4.112 ? Power Quality Monitoring

Abstract: In a response to the renewed interest in power quality monitoring and recognising cross-boundary relevance of power quality monitoring, CIGRE Study Committee C4 and CIRED established, in late 2010, the Join Working Group (JWG) C4.112: “Guidelines for Power quality monitoring – measurement locations, processing and presentation of data”. The JWG started work in February 2011 with the aim to address the application aspects of power-quality monitoring, in particular what to measure, how to measure and how to handle recorded data. This paper presents some of the results of the JWG achieved between February 2011 and December 2013, provides recommendations with respect to power quality monitoring depending on identified objectives of monitoring and identifies the areas requiring further development and research in order to comprehensively address the issue of power quality monitoring in contemporary and future power networks.

Interactions between STATCOM and VSC HVDC in dynamic GB system

Abstract: This paper analyses the operation and interaction of a STATCOM and a VSC HVDC link located in close electrical proximity in an AC system, taking into consideration of various controller configurations and operating conditions for both components. It is identified that the dynamic behaviour of one component can be substantially affected by a parametric event in the control of the other component whereas little interaction is shown between the two components for an AC system short circuit event. Results also indicate that the operating point of the DC link can deteriorate the transient performance of both the STATCOM and VSC. The study is performed based on a multi-machine dynamic GB system developed from a real reference load flow case to achieve higher fidelity.

Evaluation of classification methods for on-line identification of power system dynamic signature

Abstract: The paper investigates the use of Decision Tree (DT), Ensemble DT and multiclass Support Vector Machine (SVM) for on-line prediction of post-fault system dynamic signature based on Phasor Measurement Unit (PMU) measurements. The performance of these multiclass classification techniques is compared in terms of i) how fast the prediction about generator grouping can be made after the clearance of transient disturbance and ii) the accuracy of prediction. The application of these methods is illustrated on a 16-machine, 68-bus test system. Results indicate that the Ensemble DT method performs the best by achieving accuracy of close to 90% using 10 cycles data of post-disturbance generator rotor angles as predictors and over 90% using 30 cycles data of rotor angles as predictors.

Probabilistic assessment of the impact of distributed generation and non-linear load on harmonic propagation in power networks

Abstract: The paper investigates the uncertainties involved in the assessment of annual harmonic performance of a distribution network with distributed generation and non-liner loads. Considering a random variable locations and injections of harmonic sources, both generation and load, a probabilistic assessment of harmonic propagation through the network is performed. The influence of variable and diverse distributed generation in particular is clearly documented.

Guidelines for monitoring power quality in contemporary and future power networks – results from CIGRE/CIRED JWG C4.112

Abstract: Guidelines for monitoring power quality in contemporary and future power networks – results from CIGRE/CIRED JWG C4.112

Modelling transformer short-circuit reliability using multi-stress accelerated test data

Abstract: The high level of reliability experienced by power transformers has led to the situation that a large number of aged units are still operating in the power systems. However, uncertainty on how transformers behave when aged causes concerns over the existence of possible hidden failure modes, which have yet to be revealed by failure data. In order to establish whether aged transformers are able to maintain their operational capability when exposed to system events, such as short-circuits, multi-stress parameter analysis is required. In this paper the procedure of modelling multi-stress accelerated test data is demonstrated by using the results of a previously published paper concerning the combined effect of ageing and short-circuit forces. The results are subsequently used to determine the short-circuit reliability of transformers, through the use of a cumulative damage model. It is shown that reliability models can be constructed and justified based on an understanding of the failure mode. However, application of the results remains tentative due to the recognised shortcomings of the modelling procedure.

On accuracy of demand forecasting and its extension to demand composition forecasting using artificial intelligence based methods

Abstract: Accurate prediction of the load plays an indispensable role in power system planning and electricity market analysis. Load forecasting based on artificial intelligence (AI) techniques received a significant attention in the past and it is rapidly developing because of its high accuracy. Some of the AI based methodologies for load forecasting have already been adopted and widely used by the industry. This paper presents for the first time comparative analysis of the state of the art artificial neural network (ANN) based and adaptive neuro-based fuzzy inference system (ANFIS) based load forecasting methodologies in the same operation environment. Furthermore, the paper implements the extension of forecasting tools to forecast hourly load composition in addition to overall load at a bus. It is confirmed that either approach is very effective in load forecasting and that they have comparable performance providing appropriate setting of relevant parameters. It also proves that the approach can be successfully extended to hourly load composition forecasting and the load composition forecasting error is less than 10% at most of the time during the day.

Grid inertia supporting by energy storage inverters

Abstract: The new approach to increase the amount of inertia in power system with dispersed renewable generation and energy storage inverter is proposed. The inverter capability to deliver frequency support is presented and analyzed. The basics of the new method are in the inverter modulation scheme related to the generation of the inverter reference voltage trough a specific delay line. Derived results are verified by MatlabSimulink simulation results, comparing two cases, one with increased generator inertia and the other with energy storage inverter employed as proposed. Key-Words: energy storage integration, power-frequency control, grid inertia, inverter control

Investigation of maximum likelihood percentile estimates for transformer asset management

Abstract: To achieve a suitable balance between investment and reliability, it is necessary for utilities to be able to understand the long-term behaviour of asset populations in association with the `wear-out' stage of the so called `bathtub curve'. For the case of power transformers this is made difficult due to the presence of excessive amounts of censored data. Censored data presents partial information only, e.g. transformer survival times, inhibiting the ability of asset managers to correctly identify the long-term behaviour of the population through the use of traditionally statistical methods such as the Maximum Likelihood Estimation ({MLE}) procedure. This paper investigates the ability of the {MLE} to estimate strategic percentiles of asset populations that are used to assist asset management decisions. The analytical study is performed through a series of Monte Carlo simulations and statistical measures to determine the `quality' of estimates of the 2.5 and 97.5 percentiles of the Normal and Weibull distributions, under scenarios with different sample size and percentages of censored data. The results demonstrate the ability of the {MLE} procedure to identify percentiles in the near tail of the distribution with reasonable accuracy, and the pessimistic nature of percentiles in the far tail when the dataset contains levels of censoring normally found in transformer populations. The results are further verified through the calculation of approximate 95\% confidence intervals.

Estimation of percentage of controllable load in total demand at bulk supply point

Abstract: This paper presents an Artificial Neural Network (ANN) based approach to estimate percentage of controllable load in overall demand at bulk supply point at any given time based on standard voltage, real and reactive power measurements at the substation. Monte Carlo Simulation (MCS) is used to generate the training and validation data. The estimated controllable and uncontrollable load percentages are compared with the targets in the validation process, and the probability distribution and the confidence levels of load participation estimation errors are obtained. When all inputs are available, the most probable absolute error of estimation of controllable and uncontrollable load percentage is approximately 4.3%, with about 60% of all estimations having absolute errors below 10%. The robustness of the methodology with respect to missing input data is also evaluated. It demonstrates that the absence of an input, especially the absence of the reactive power, can reduce the confidence level of estimation with the same estimation error.

The effect of temporal and spatial variation of harmonic sources on annual harmonic performance of distribution networks

Abstract: The increasing penetration of converter connected distributed generation and the electric vehicles in today's distribution network, introduce new uncertainties to the harmonic performance evaluation. The common practice of day/week evaluation of harmonics may not be feasible for long term planning applications due to the spatial (EV) and temporal (DG) variation in harmonic sources. This paper proposes a probabilistic methodology to evaluate the annual harmonic performance of distribution network, considering the uncertainties involved due to the connection of distributed generation and electric vehicles including their temporal and spatial variation.

Risk-based framework for establishing and visualising operational constraints of power systems

Abstract: This paper demonstrates a risk-based framework that can be used to establish operational constraints of power systems. The application of this methodology is illustrated on the areas of small disturbance stability and sub-synchronous resonance (SSR) in order to determine risk levels associated with various operating regions (conditions). By visualizing these risk areas, guidance for planning and operation of the power system can be provided. Furthermore, the way in which these risk areas are affected by various system contingencies, is also investigated to gain understanding of the risk performance of the network. The proposed methods are demonstrated using a meshed power system with capacitor series compensated AC tie lines.

Applying process immunity time to assess annual industrial process trips

Abstract: The effect of voltage sag to customer's process has been studied by different authors. They propose methods in order to assess annual indices related to process trip due to sags and long duration interruptions (LDI). However, these methods don't consider the process immunity time (PIT) during the estimation of these indices. This paper introduces a procedure where PIT is included in order to increase the accuracy of annual indices. The procedure is applied to a representative distribution network. The results show that (i) the difference, between the values obtained considering PIT or not, is approx. on average 50% and (ii) only 6% the total of event between sag and LDI affects the customer's process.