Search

Round conductive bars embedded in semi-closed slots are frequently used in the construction of electric machinery, for instance in the squirrel cage of induction motors. Frequently, such as in the study of induction motor steady-state performance at rated slip, it is useful to estimate the slot leakage inductance of these conductors under the hypothesis of no eddy currents and no magnetic saturation. This is usually done through simple approximated analytical formulas available in the literature. In this paper, an improved explicit ready-to-use leakage inductance expression for circular bars embedded in semi-closed slots is derived by solving Poisson's equation in the slot domain. The precision of the proposed formulation is assessed against Finite Element Analysis (FEA) for various slot geometries and is shown to always give very accurate results, with errors below 2%. Conversely, approximated simplified formulas available from the literature are demonstrated to possibly give large errors, which exceed 20% for some explored slot geometries.

Squirrel-cage induction motors are the most widely used type of electric machinery in industrial applications thanks to their rugged, cheap and robust construction. For the fast prediction of induction motor steady-state performance without the need for time-consuming finite-element analyses, the equivalent circuit parameters of the machine need to be calculated, including rotor slot leakage inductances. This paper, in particular, proposes a new improved analytical formula to determine the rotor slot leakage inductance in squirrel-cage induction motors with rectangular-shaped bars. The formula is obtained by solving Poisson's equation in the slot domain. Results are assessed by comparison against Finite Element Analysis (FEA). The precision of the presented methodology is also compared to that of approximated analytical models available from the literature. The comparison shows that the accurate formula proposed is in excellent agreement with FEA with errors below 2%, while the simplified model generally leads to errors above 10% which increase as the height-to-width bar ratio decreases.