Raymond D. Findlay

A New Model of AC Resistance in ACSR Conductors

An accurate new model for calculating AC resistance of ACSR conductors has been developed and verified for single, two and three layer conductor designs. Current redistribution was found to be far more important in determining the AC/DC resistance ratio than hysteresis losses in the steel core except in single layer conductors. The model suggests ways of reducing the AC resistance and promises to be a useful design tool.

Methods for predicting rotational iron losses in three phase induction motor stators

Iron losses in electrical machines are difficult to predict. Discrepancies between tested and calculated no-load core losses are sometimes large. In induction motors, a large portion of the stator core is subjected to flux that rotates in the plane of the laminations. Losses due to rotating flux differ from those observed under alternating flux conditions. This paper describes direct approaches for coupling rotational iron loss measurements with finite element analysis (FEA) results that yield the distribution of rotational flux in induction motor cores. Using these methods, the prediction of no-load core loss can be made to include the effects of rotational iron losses.

Stray Losses in Induction Machines: Part II, Calculation and Reduction

The definition, origin and measurement of stray load losses in induction machines have already been addressed in a companion paper. This paper completes the study by examining the mechanisms for prediction of those losses as well as some techniques for reduction. It also suggests some directions for future work.

A Neural Network Based Short Term Electric Load Forecasting in Ontario Canada

Accurate and reliable load forecasting is necessary to ameliorate energy management. Short-term load forecast plays a crucial role in economic and secure system operation. This paper presents a practical method for short-term electric load forecast problem using an artificial neural network with a powerful Levenberg-Marquardt training algorithm approach. The applications of real load from Ontario, Canada with hourly load, daily load, and weekly load predictions have been successfully achieved. Both visual comparison and statistical test are discussed and analyzed to validate training and testing phases of the neural network.

A new approach using artificial neural network and time series models for short term load forecasting

This paper presents a new approach for short-term load forecasting (STLF). Artificial neural network and time series models are used for forecasting hourly loads of weekdays as well as weekends and public holidays. In addition to hourly loads, daily peak load is an important data for systems operators. Most of the common forecasting approaches do not consider this issue. It is shown that the proposed approach provide very accurate forecast of the daily peak load. The input variables of the models have been selected based on their correlation coefficients. In addition, a new technique for selecting the training vectors is introduced. The valuable experience of expert operators is included in the modeling process. The model is simple, fast, and accurate. Obtained results from extensive testing on Ontario load data confirm the validity of the proposed approach. The mean percent relative error of the model over a period of one year is 2.066% including holidays.

Improved Fuzzy Neural Network Control for a Pneumatic System Based on Extended Kalman Filter

Although pneumatic systems are used in many applications such as robotics and manufacturing field, accurate control for such systems is difficult to be achieved due to their inherent nonlinear dynamics. This paper presents the favored results of fuzzy neural network (FNN) control for a pneumatic system based on extended Kalman filer (EKF). To optimally design a FNN controller for the pneumatic system, back- propagation (BP) algorithm is used to update the parameters of membership functions on-line. The partial derivative of the plant output with respect to the input, which is required by the learning process of FNN, is approximately estimated with a feed-forward neural network trained by recursive EKF. With the designed FNN controller for the pneumatic system, precise steady-state response and good dynamic tracking are obtained, which demonstrate that the nonlinear dynamics of the pneumatic system are efficiently overcome.

Accurate Assessment of Thermal Field and Ampacity of Underground Power Cables

A newer approach to cable thermal field and ampacity computations using a proposed concept of perturbed finite-element analysis is formulated. This approach involves the use of derived sensitivity coefficients associated with various cable parameters of interest and using such sensitivity coefficients to achieve optimal cable performance. The technique is applied to both design phase and operational aspects of power cable buried in a complex media of soils, heat sources and sinks as well as variable boundary conditions

A Novel Finite-Element Optimization Algorithm with Applications to Power Cable Thermal Circuit Design

This paper presents the results of a recent study to develop an optimization model for underground power cable thermal circuit based on generated gradient approach. A new concept of perturbed finite-element analysis is utilized, which involves the use of derived sensitivity coefficients associated with various cable parameters of the interest. A subsequent utilization of such sensitivities as gradients of objective functions is realized in a general framework of power cable performance optimization. Therefore, based on the work of this paper, it is now possible to optimize the thermal circuit parameters including the thermal conductivities, boundary conditions and heat generation with respect to cable temperatures defined in a desired objective function and/or constraints. This enables more effective dealing with the nonlinearity of such temperatures, as implicit functions, using the more accurate perturbed finite element method. The proposed method minimizes the objective function value, without sacrificing the modeling accuracy in order to suit other exiting traditional methods. The developed algorithm was applied to various practical utility cable systems of 132-kV XLPE and 380-kV oil filled cables with their actual in-service configurations and for different practical cable performance optimization objectives demanded by the power utility operators.

Prediction of Ontario Hourly Load Demands and Neural Network Modeling Techniques

Accurate and reliable load forecasting is necessary to ameliorate energy management. For the purpose of load demands prediction, this paper develops an artificial neural network model, which adopts Levenberg-Marquardt method as training algorithm, both visual comparison and statistical techniques as validation methods. With the built neural network model, the hourly load demands of Ontario in 2004 have been successfully forecasted

Skin Effect Factor in the Bar Extension of Large Two-Pole Induction Motors by Three-Dimensional Finite-Element Simulations

An accurate estimate of the skin effect factor in the rotor bars of squirrel cage induction machines is important for the calculation of starting torque since the developed torque is directly proportional to the bar resistance. Past studies have given good estimates of this factor for the portion of the bars embedded in the rotor iron, but little is known about the effect in the rotor bar extension beyond the rotor core before entering the end ring. Depending on the length of the bar extension, the effect on the starting torque calculations has the potential to become significant. In this study, the end region of the machine was modeled using a commercial three-dimensional finite-element package in order to gain some insight into this phenomenon