• Energy Research

In this paper, a comparison of power system frequency response is conducted for a simple modelled power system with primary frequency control being provided either by synchronous generators or by inverter-based Battery Energy Storage (BES) systems. Mathematical models of conventional governor and turbine are developed, representing conventional synchronous generator frequency control, and are used to illustrate system frequency response for a range of typical conventional generating units. A mathematical model of a power-electronics interfaced Li-ion BES system is developed and used to represent a non-synchronous inverter-based generator with primary frequency control capabilities. MATLAB/Simulink is used to build a model of a small power system, and simulations are carried out with at a range of typical conventional synchronous generating units with a nonsynchronous generating unit in the power system. The simulation results demonstrate that the BES can be used for primary frequency control providing a faster and better response. BES is also capable of almost eliminating frequency overshoot and reducing 70% of settling time while providing primary frequency control in power system with various types of conventional synchronous generating units and a nonsynchronous generating unit.

Due to single-phase connected loads and distributed generation, distribution networks often operate under unbalanced loading conditions. This implies that the assumption of a balanced network for the purpose of state estimation does not always hold. Numerous papers have been published on the topic of three-phase state estimation in distribution networks in order to solve this problem. However, there is no generic method to determine whether a three-phase state estimator is required or if a single-phase network representation can deliver a sufficient degree of accuracy. Therefore, in this paper a new method is presented that provides a generic method to choose between the single-and three-phase state estimator method, depending on the network topology, and metering configuration.

This paper presents a comparative assessment of the current control methods (CCM) of switch-mode converters for photovoltaic (PV) applications. In this paper, average current control, current programmed control, hysteresis current control and nonlinear carrier control methods are addressed considering input fluctuations and load variations for PV systems. Dynamic responses of PV systems are investigated and harmonic analysis is performed. Performances of the above current controllers are examined through simulation and the results are presented. The results show that the selection of different current control techniques depends on the working conditions and the area of applications.

Recently penetration level of converter-interfaced distributed generation (DG) systems has been increased in utility and will be more so in the future. To avoid grid instability, converter-interfaced DG should have low voltage ride through (LVRT) capability considering grid codes. An efficient scheme is proposed to enhance the LVRT capability of converter-interfaced DGs. The proposed LVRT technique employs a single phase controllable DC-link bridge type fault current limiter (CD-BFCL), which is placed in DC side of the converter. During normal condition, the CD-BFCL does not affect overall operation of converter-interfaced DG; whereas, during all AC faults including balanced, unbalanced, and transient repeated network faults, converter output AC currents are limited to a desired level. In this way, the converter-interfaced DG is capable to ride through faults even at the zero grid voltage during the AC grid faults. Analytics and modes of operation of the CD-BFCL are presented during switching intervals to demonstrate the efficiency of the proposed approach in limitation of converter over-current. The operation of the proposed method is proved through simulation studies in PSCAD/EMTDC. Also, a laboratory prototype is developed to validate the possibility of practical implementation of the CD-BFCL.

This paper provides an overview of forecasting problems and techniques in power system. Available forecasting techniques have been reviewed with the focus on data mining for wind power prediction. This paper also discusses forecasting issues associated with electricity price and load forecasting.

This paper presents a sensitivity analysis of neural network (NN) parameters to improve the performance of electricity price forecasting. The presented work is an extended version of previous works done by authors to integrate NN and similar days (SD) method for predicting electricity prices. Focus here is on sensitivity analysis of NN parameters while keeping the parameters same for SD to forecast day-ahead electricity prices in the PJM market. Sensitivity analysis of NN parameters include back-propagation learning set (BP-set), learning rate (c), momentum (a) and NN learning days (dN N). The SD parameters, i.e. time framework of SD (d = 45 days) and number of selected similar price days (N=5) are kept constant for all the simulated cases. Forecasting performance is carried out by choosing two different days from each season of the year 2006 and for which, the NN parameters for the base case are considered as BP-set=500, c=0.8, a=0.1 and dNN = 45 days. Sensitivity analysis has been carried out by changing the value of BP-set (500, 1000, 1500); c (0.6, 0.8, 1.0, 1.2), a (0.1, 0.2, 0.3) and dNN (15, 30, 45 and 60 days). The most favorable value of BP-set is first found out from the sensitivity analysis followed by that of c and a, and based on which the best value of dNN is determined. Sensitivity analysis results demonstrate that the best value of mean absolute percentage error (MAPE) is obtained when BP-set = 500, c= 0.8, a=0.1 and dNN = 60 days for winter season. For spring, summer and autumn, these values are 500, 0.6, 0.1 and 45 days, respectively. MAPE, forecast mean square error and mean absolute error of reasonably small value are obtained for the PJM data, which has correlation coefficient of determination (R2) of 0.7758 between load and electricity price. Numerical results show that forecasts generated by developed NN model based on the most favorable case are accurate and efficient. This paper presents a sensitivity analysis of neural network (NN) parameters to improve the performance of electricity price forecasting. The presented work is an extended version of previous works done by authors to integrate NN and similar days (SD) method for predicting electricity prices. Focus here is on sensitivity analysis of NN parameters while keeping the parameters same for SD to forecast day-ahead electricity prices in the PJM market. Sensitivity analysis of NN parameters include back-propagation learning set (BP-set), learning rate (c), momentum (a) and NN learning days (dN N). The SD parameters, i.e. time framework of SD (d = 45 days) and number of selected similar price days (N=5) are kept constant for all the simulated cases. Forecasting performance is carried out by choosing two different days from each season of the year 2006 and for which, the NN parameters for the base case are considered as BP-set=500, c=0.8, a=0.1 and dNN = 45 days. Sensitivity analysis has been carried out by changing the value of BP-set (500, 1000, 1500); c (0.6, 0.8, 1.0, 1.2), a (0.1, 0.2, 0.3) and dNN (15, 30, 45 and 60 days). The most favorable value of BP-set is first found out from the sensitivity analysis followed by that of c and a, and based on which the best value of dNN is determined. Sensitivity analysis results demonstrate that the best value of mean absolute percentage error (MAPE) is obtained when BP-set = 500, c= 0.8, a=0.1 and dNN = 60 days for winter season. For spring, summer and autumn, these values are 500, 0.6, 0.1 and 45 days, respectively. MAPE, forecast mean square error and mean absolute error of reasonably small value are obtained for the PJM data, which has correlation coefficient of determination (R2) of 0.7758 between load and electricity price. Numerical results show that forecasts generated by developed NN model based on the most favorable case are accurate and efficient.

This paper presents a messy genetic-algorithm-based optimization scheme for SVC planning, aimed at voltage stability enhancement of power systems under most critical and N-1 operation conditions. The SVC placement was formulated as a multi-objective optimization problem in terms of maximum worst-case reactive power margin, highest load voltages towards the critical operating points, minimum real power losses, and lowest SVC device costs. During the genetic algorithm search for the optimal solution, the most critical disturbance scenario was estimated with each SVC placement and the configuration of the original power system. The preselected candidate was further checked against the N-1 security criterion.

The preparation and training of power system dispatchers with the aid of a training simulator combining the problem-oriented teaching technique are described. The training simulator discussed belongs to a class of programmable simulators and contains a set of training procedures dealing with the most important emergency situation which is the characteristic conditions in bulk power systems. Using the same training procedures, it is possible to utilise a wide range of facilities for abandoning the current emergency practices. The scenarios for training procedures are built on the principles which form the basis of computer-assisted learning systems, consisting of a logical sequence of separate programs. The training is carried out in a dialogue mode with a computer. >

To improve the use of renewable energy sources and increase their penetration in the renewable energy market, different types of energy storage technologies have been introduced among which compressed air energy storage (CAES) systems offer more competitive feature. In this paper, a comprehensive exergy analysis of a one-stage, adiabatic compressed air energy storage (A-CAES) system is carried out for a wide range of parameters. The governing parameters on the system performance include, but not limited to, storage pressure, pre-set vessel pressure as well as isentropic efficiencies of the major components involved in the A-CAES systems most specifically compressor and expander. By performing the exergy analysis and using the concept of exergy destruction, which is a criterion for power loss due to the irreversibility, the magnitude and source of thermodynamic inefficiencies at each part of the system is identified. The obtained results reveal that the isentropic efficiency of the machine has a significant effect on the amount of exergy destructed by the machine. In addition, the results show that generally more exergy is lost in the expander than other components of the A-CAES system during the system operation. Consequently, to improve the system performance, utilising a high-efficient expander should be a priority. It is also shown that increase in the storage pressure improves the exergy efficiency of the compressor and expander. Using the obtained results, it would be possible to find the optimum working condition for the system leading to improvement of the system performance.