Loss Minimization With Optimal Power Dispatch in Multi-Frequency HVac Power Systems

Abstract

Low-frequency high voltage ac transmission scheme has recently been proposed as an alternative approach for bulk power transmission. This paper proposes a multi-period optimal power flow (OPF) for a multi-frequency HVac transmission system that interconnects both conventional 50/60-Hz and lowfrequency grids using back-to-back converters with a centralized control scheme. The OPF objective is to minimize system losses by determining the optimal dispatch for generators, shunt capacitors, and converters. The OPF constraints include the operational constraints of all HVac grid and converter stations. The resulting mixed-integer nonlinear programing problem is solved using a proposed framework based on the predictor-corrector primaldual interior-point method. The proposed OPF formulation and solution approach are verified using a multi-frequency HVac transmission system that is modified from the IEEE 57-bus system. The results with the optimal dispatch from the proposed method during a simulated day show a significant loss reduction and an improved voltage regulation compared to those when an arbitrary dispatch is chosen.Index Terms-Back-to-back converter, Low-frequency AC transmission, Multi-frequency power systems, Optimal power flow, Interior point method. * , Q conv * power from/to BTB converters; Q sh * injected reactive power into grid from shunt capacitors at 1 pu;

Steady State Modeling for Variable Frequency AC Power Flow

“…In the design of a system where frequency can be selected, it is important to understand where these frequency ranges lie. This section demonstrates that the frequency regions can be determined analytically as intersections of the power flow equations (11,12) and engineering constraints (15)(16), dependent on branch parameters and constraint values. With these analytical solutions, appropriate selection of frequency can be combined with the utilization of existing branch conductors and design of new components to achieve optimal utilization of the transmission system.…”

“…In [14], this power flow model was extended to accommodate networks composed of multiple areas with frequency conversion at the area interfaces, allowing for distinct frequencies, including DC, in each area. In [15], an optimal power flow problem was formulated with multiple areas of distinct frequencies. The frequency was not considered variable in the formulation but fixed before the calculation.…”

Power System Sensitivity Matrix Estimation by Multivariable Least Squares Considering Mitigating Data Saturation

“…By matrix transformations, the estimator B is given by: In some cases, weights to the residual errors from different variables are the same, i.e., (11) where w is the each of column vectors in W. Then (9) can be simplified as: (12) The implementation procedure of the WMLS is: first, build the sample regression model by (4) with N sample units of the online-measured data X and Y; next, set the weighting matrix W; then calculate the estimator B by (9) or (12).…”

“…In conventional deterministic power systems, sensitivities can be obtained using offline models of the systems, such as the calculation by performing inversion of augmented Jacobian matrix [9], by the perturbation method [10], and by direct calculation using the power system offline equations [11]. However, in modern stochastic power systems, the sensitivity changes along with the time-varying operating conditions; therefore, owing to the ensuing model incompatible, the results from conventional methods may no longer be suitable in practice.…”

Power System Sensitivity Matrix Estimation by Multivariable Least Squares Considering Mitigating Data Saturation

“…In some cases, weights to the residual errors from different variables are the same, i.e., (11) where w is the each of column vectors in W. Then ( 9) can be simplified as: (12) The implementation procedure of the WMLS is: first, build the sample regression model by (4) with N sample units of the online-measured data X and Y; next, set the weighting matrix W; then calculate the estimator B by (9) or (12).…”