Search

• Risk-based procedure for managing a distribution transformer by its dynamic rating. • Formulation of an admissible risk level with an intuitive physical meaning. • Quantile regression forests to forecast transformer load and ambient temperature. • Numerical experiments are based on actual industrial data to validate the proposal. • Results are provided from energetic and economic perspectives. Electrical distribution networks are often inadequate to comply with the increasing load demand and to enable new installation of distributed energy resources. This inadequacy may result in upgrading existing transformers of in disconnecting the entire load during demand peaks, and into wasting a portion of the total renewable energy produced during favorable weather conditions. Increasing the capacity of electrical distribution network components is therefore mandatory in order to avoid these negative circumstances; however replacing lines and transformers is costly, and often the benefits are not worth the investments. Fortunately, the capacity of distribution networks can be unlocked by managing the electrical components by their dynamic ratings, allowing operating them even beyond their rated ampacity for a limited interval of time. This paper proposes a new risk-based procedure to manage distribution transformers by their dynamic rating. Probabilistic forecasts of loads and ambient temperatures are the inputs of the procedure, which are obtained in this paper from quantile regression forests. The procedure quantifies, in a probabilistic framework, the insulation thermal ageing due to the transformer (over)load in terms of the cost sustained for rewinding the transformer. The corresponding estimated risk is made available to the distribution system operator, who may therefore take a decision comparing it to an arbitrary level of risk coverage, favoring either the power delivery to customers or the reduction of the transformer ageing rate; this level is formulated in this paper having an intuitive physical meaning, thus enabling its setting also by non-expert operators. Comprehensive numerical experiments based on actual data are presented under different configurations and scenarios, in order to validate the proposal from an energetic and economic point of view. [ABSTRACT FROM AUTHOR]

Highlights • Multiple linear and support vector regression to forecast industrial reactive power. • Assessment of calendar, weather, and electrical variables as model predictors. • Two model selection techniques (cross-validation and lasso) are compared in the paper. • Numerical applications are performed on actual industrial data to test the proposal. • Results of forecasting models and of selection techniques are compared in the paper. Abstract Reactive power forecasting is essential for managing energy systems of factories and industrial plants. However, the scientific community has devoted scant attention to industrial load forecasting, and even less to reactive power forecasting. Many challenges in developing a short-term reactive power forecasting system for factories have rarely been studied. Industrial loads may depend on many factors, such as scheduled processes and work shifts, which are uncommon or unnecessary in classical load forecasting models. Moreover, the features of reactive power are significantly different from active power, so some commonly used variables in classical load forecasting models may become meaningless for forecasting reactive power. In this paper, we develop several models to forecast industrial reactive power. These models are constructed based on two forecasting techniques (e.g., multiple linear regression and support vector regression) and two variable selection methods (e.g., cross validation and least absolute shrinkage and selection operator). In the numerical applications based on real data collected from an Italian factory at both aggregate and individual load levels, the proposed models outperform four benchmark models in short forecast horizons. [ABSTRACT FROM AUTHOR]

In this paper, we address the problem of the probabilistic steady-state analysis of an electrical distribution system that includes wind and photovoltaic power plants. We propose a new method that takes into account the uncertainties due to the time variations of power load demands and the random nature of solar and wind energy. The new method extends a probabilistic approach proposed in the relevant literature for balanced electrical distribution systems to unbalanced electrical distribution systems, and it also takes into account the correlation among the random input variables. The method is based on Taguchi's orthogonal arrays. Numerical applications based on an IEEE test system provide evidence of the probabilistic performance of the proposed approach. The results obtained with the proposed method are compared with the results obtained using non-linear Monte Carlo simulation and the Point Estimate Method. [ABSTRACT FROM AUTHOR]

Electric power systems are increasingly influenced by meteorological conditions due the wide penetration of renewable energy plants. In particular, wind speed plays a key role in transmission and distribution systems, as it determines wind power production and it influences line rating; also, extreme values of wind speed (EWS) affect mechanical reliability of wind towers and blades, causing loss of lifetime and damages. Several physical models provide EWS forecasts on global or local scales; however, such models require accurate initial and boundary conditions, and enormous computational effort. Statistical models are good alternatives to achieve accurate forecasts, but they require assumptions on the statistical distribution of EWS in parametric frameworks. This paper proposes an Inverse Burr - Inverse Weibull finite mixture distribution for EWS characterization and its Expectation-Maximization (EM) procedure for parameter estimation. The proposed mixture model is compared to commonly-used EWS distributions. Actual sites are considered to evaluate the goodness of fitting in different, real conditions; the EM procedure is compared to the maximum likelihood estimation. Sensitivity and error analyses are performed to individuate the main features of the proposed model; numerical results confirmed the suitability of the proposed model and of the EM estimation in the most of considered cases. [ABSTRACT FROM AUTHOR]

In this paper, a probabilistic method is proposed to analyze the steady-state operating conditions of an active electrical distribution system with Wind (WD) and Photovoltaic (PV) generation plants. This method takes into account the uncertainties of power load demands and power production from renewable generation systems and combines Monte Carlo simulation techniques and multi-linearized power flow equations. The power flow equations include models of wind turbine and PV generation units and multi-linearization is accomplished by applying a criterion based on the total active power of the system. The method properly extends a probabilistic method proposed in the relevant literature for traditional passive electrical distribution systems to the field of an active electrical distribution system with WD and PV generation units. Numerical applications are presented and discussed with reference to a 17-bus test distribution system characterized by WD and PV systems connected at different busbars. The results obtained with the proposed algorithm are compared with the results obtained using a Monte Carlo simulation algorithm that included non-linear power flow equations. [ABSTRACT FROM AUTHOR]

Highlights: [•] Two multi-objective optimization methods are proposed. [•] Exponential Weighted Criterion (EWC) and Compromise Programming Method (CPM). [•] Suitable when computational time is an important issue, as in case of optimal control. [•] The EWC enables better exploration of the Pareto frontier by capturing more points. [•] The EWC and the CPM are robust with reference to the constraints. [ABSTRACT FROM AUTHOR]

Abstract: This paper applies the point estimate method to account for the uncertainties that affect the evaluation of the steady state operating condition of an unbalanced three-phase power system. Moreover, since the point estimate method requires that the input random variables are uncorrelated, a suitable adjustment to take into account the correlation is applied. Different point estimate schemes (2m, 2m +1 and 4m +1 schemes) are presented and tested. The accuracy of the proposed techniques is tested on a three-phase unbalanced IEEE 34-bus test system; the results obtained applying the Monte Carlo simulation are assumed as reference . Both correlated and uncorrelated input random variables are considered, and multimodal probability density functions are tested. The conclusion is that the 2m +1 scheme gives the best solution in terms of accuracy and computational efforts; in the case of correlated input random variables, an adequate procedure to take into account the correlation must be applied. [Copyright &y& Elsevier]