| Contemporary power systems exist under heavy stress, caused by higher asset utilization in electric power transmission. As networks are operated nearer to their limits, new stability issues have arisen. One of the more destructive problems is voltage instability, where large areas of an electrical network may experience reduced voltages or collapse because of high reactive power demand.; Voltage stability margins may be improved through the adjustment of the system operating position, which alters the power flow profile of the transmission network. Furthermore, the margins may be optimized through the application of nonlinear programming, if they are quantified using an index of voltage collapse proximity. This dissertation details the maximization of the eigenvalues of the reduced reactive power-voltage matrix, in an effort to increase voltage security.; The nonlinear optimization was solved using four different techniques. First, a conventional optimal power flow was applied to the problem, which solved linear approximations of the original problem and maintained feasibility for the intermediate points. This method was augmented to include a quadratic model of the objective function. In addition, the feasibility requirement was relaxed to produce a third solution technique. Finally, the stability optimization problem was solved using a quadratic model without the feasibility requirement.; Tests of all four methods were performed on three sample power systems. The systems included six, 14, and 118 bus examples. In all three cases, each of the four methods effected improvement in the stability margin, as measured by a variety of indicators. The infeasible linear solution provided the best results, based on runtime and the relative stability improvement. Also, the results showed that the additional quadratic approximation did not provide any measurable benefit to the procedure. Moreover, the methods that specified feasibility at each step were inferior compared to the algorithms which relaxed that restriction.; Finally, the modeling of the reactive power loads in the electrical network was altered to provide insight into the application of the stability improvement algorithm to realistic systems. Tests showed that the effects were not large, even as the model varied from constant reactive power demand to a pure reactance representation. |
