A Finite-Element Method for the Solution of Turbine-Generation End-Zone Fields

Abstract

The growth of demand for electrical power has led to large increases in the unit ratings of turbine-generators; typically, maximum 2-pole ratings have increased from 100 MW 20 years ago, to 660 MW today, with 1300 MW units being seriously considered. The growth in unit output has been made possible by major improvements in design, particularly the introduction of direct cooling of the windings, which has allowed the electric loading to be greatly increased. The end-region flux densities have increased in consequence, and core end-heating and end-winding forces are major factors to be considered in the design. It is thus essential to be able to calculate end-zone fields accurately. This thesis describes a finite-element method of predicting fields in the air space, allowing for complexities of boundary geometry. To economise in computer store, it is assumed that all functions vary sinusoidally in the peripheral direction, so that the 3-dimensional problem can be treated by a "quasi-3-dimensional" method, involving radial and axial distances only. The solution is obtained as a scalar potential distribution, from which the component flux densities can be derived. Following proving of the method by application to problems for which analytic solutions exist, flux density distributions have been calculated for a short-core replica of a 500 MW generator and a production 660 MW generator. Agreement with test results is good. Finally, the value of the method has been demonstrated by application to problems arising in the end-zone design of superconducting field and fully slotless generators.