28-06-2012, 04:13 PM
Analytical Thermal Models for Small Induction Motors
Analytical Thermal.pdf (Size: 1.92 MB / Downloads: 78)
INTRODUCTION
A range of small TENV induction motors, one of which is
shown in Fig 1, has been modeled with analytical network
thermal models having different number of nodes. The models
with a small number of nodes are fast to run but can be
difficult to set up accurately. The models with a large number
of nodes can be programmed to give an accurate thermal
prediction. This is done by dividing individual heat transfer
phenomena into separate thermal resistances for which
mathematical algorithms exist for their calculation, e.g. thermal
resistances to represent heat transfer through composite
components such as the winding and bearings, the conduction
through the solid components such as the teeth and back iron
and convection and radiation from the internal and external
surfaces of the machine.
SIMPLIFIED MODELS WITH BETWEEN 5 AND 10 NODES
A more complex thermal model employs between 5 and 10
nodes to represent important features within the machine. For
instance the 5 node network shown in Fig 3 has nodes to
represent ambient, stator lamination, housing, winding and
rotor. The internal thermal resistances between winding to
stator back iron and stator back iron to housing can be
calibrated from the internal temperature drops within the stator.
The difference in stator and rotor temperature is defined as a
temperature difference in this simple model. The convection
and radiation from the housing and endcap surfaces is
calculated using an effective heat transfer coefficient (h –
W/m2.K) in the model. This is used together with the frame
surface area FArea to calculate a frame-to-ambient thermal
resistance:
DETAILED MULTIPLE-NODE THERMAL MODELS
Detailed analytical thermal models tend to have 20 or more
nodes to model complexities in the heat transfer path such as
interface gaps between components, the composite material
nature of the winding, cooling of the active and end-winding
components, etc. Such detail could be modeled using
numerical techniques such as finite-element analysis (FEA)
and computational fluid dynamics (CFD). However, a detailed
analytical model has large advantages in terms of calculation
speed. The near instantaneous calculation capabilities of the
analysis technique make it possible to run "what-if" scenarios
in real time. It also facilitates sensitivity analysis on the main
thermal unknowns to evaluate their importance on the thermal
performance, i.e. effective interface gaps, impregnation
goodness, etc. The main strengths of the numerical techniques
are in the visualization of flow and in the development of
convection formulations for use in lump-circuit analysis, rather
than carrying out the thermal circuit optimization itself.
RESULTS DISCUSSION
Two small TENV induction motors have been modeled.
Both motors operate at 50Hz and have 4-pole. Motor A has 3-
phases, while Motor B is a permanent-split capacitor motor.
Rated torque is 0.2Nm for both motors. Fig. 1 illustrates one of
the analyzed motors. Table I shows the results for the steadystate
temperature rise for the stator winding. Model 1 is based
on the DegCW equivalent thermal circuit shown in Fig. 2 with
thermal resistance from Motor-CAD calibration. Model 2 is
based on the Hot10 equivalent thermal circuit model shown in
Fig. 4. Again, the parameters within the model are set up
according to the automated Motor-CAD calibration process.
Model 3 is the original Motor-CAD model. As expected all
three models give very similar results as they are calibrated
using the same Motor-CAD model.
CONCLUSIONS
Simplified and/or detailed analytical thermal models can be
successfully used in predicting the temperature rise in small
induction motors. The level of detail and accuracy of these
models strongly depends on the number of nodes and how the
thermal resistances are set up. The calibration process for
various reduced nodal model has been successfully described.
Once calibrated the reduced node thermal models give both
satisfactory accuracy and allow very fast and robust thermal
calculations.