Study of Temperature Fields in Enclosed Fan-Cooled Electric Motors
Abstract Generated abstract
This paper examines temperature fields and thermal loading in enclosed fan-cooled low-power asynchronous motors, using AO2-series machines of several frame sizes and pole numbers. The authors embedded numerous thermocouples in stator windings and frames, repeated rated-load and overload tests, and analyzed temperature distributions and root-mean-square scatter across operating conditions, stator circumference, and individual machines. The results show higher temperatures in end windings than in slot portions, especially on the drive side, identify significant circumferential nonuniformity in frame temperature, and indicate that the frame-to-air temperature drop accounts for a major share of winding temperature rise. The study also proposes physical model testing for external ventilation and gives an empirical Nusselt-Reynolds relation for the average frame heat-transfer coefficient.
Full Text
Study of Temperature Fields in Enclosed Fan-Cooled Electric Motors
V. A. Zhadan, D. I. Sannikov, R. Ya. Klyain
(Presented by the scientific seminar of the Departments of Electrical Machines and General Electrical Engineering)
At present, most low-power asynchronous motors are manufactured in an enclosed fan-cooled design. It is therefore very important to know the pattern of distribution of the temperature field and the thermal loading of the stator winding for the further improvement and perfection of motors.
Investigations of the temperature field of the stator winding, as well as of the frame, were carried out on AO2-series motors of the 3rd and 4th frame sizes, both four- and six-pole. During winding, from 38 to 50 thermocouples were embedded in the stator winding, which made it possible to obtain a sufficiently reliable picture of the temperature field, and also to determine the temperature scatter between identical points around the stator circumference, the temperature scatter from one test to another, and the temperature scatter between individual motors. Thermal tests were conducted both at rated load and under overload at $I = 1.2 I_{\mathrm{n}}$. Heating of the motors to steady-state temperature was monitored by 15 thermocouples from the stator winding and 10 thermocouples from the frame. For greater reliability of the temperature field, the investigations were carried out on 12 electric motors, and the thermal regimes were repeated many times.
Figures 1 and 2 give the mean values of the temperature field of motors of the 3rd and 4th frame sizes. Analysis of temperature distributions shows that the end-winding parts are hotter than the slot parts by an average of 6–8%, which is explained by better heat transfer from the winding to the frame through the stator iron, compared with heat removal from the end-winding parts to the frame through air. The internal thermal resistance in the end-winding parts, whose coil ends form a monolithic body from the thermophysical point of view, acts in the same direction.
The hottest part is the end winding on the drive side. Its overheating exceeds the overheating of the opposite side by 5–9%. This is explained by the higher temperature of the frame on the drive side, and also by the presence of intercoil connections and, in this connection, reinforced insulation.
The temperature in the slot part increases from the bottom of the slot toward the wedge by 6–7% of the mean overheating under the influence of the heat flow from the rotor through the stator teeth to the frame, as well as the heat-insulating effect of the wedge.
Dips in the temperature curves are observed at the point where the sections leave the slots. This is explained by the following:
Fig. 1. Temperature field of the AO2-42-4 motor under overload.
O — in the upper part of the slot,
X — at the bottom of the slot.
-
The end sheets of the stack are under more favorable conditions from the point of view of heat removal to the frame, since they are located closer to the free parts of the frame (not in contact with the stack).
-
At the exit of the sections from the slots, radial channels are formed, through which air is driven intensively by the rotor blades.
The volume of experimental data obtained makes it possible to estimate the scatter of temperatures. Three types of temperature scatter may be distinguished: scatter of temperatures by operating conditions \(\sigma_{\mathrm{r}}\); scatter of temperatures around the circumference of the stator \(\sigma_{\mathrm{t}}\); scatter of temperatures among machines \(\sigma_{\mathrm{m}}\). The magnitude of the scatter is estimated by root-mean-square deviations. The calculation of the root-mean-square deviations is performed with 95-percent probability.
The scatter of temperature between operating conditions is characterized by the following data:
for the end-winding part, \(\sigma_{\mathrm{r}} = 1.7 \div 4.6\%\),
for the slot part, \(\sigma_{\mathrm{r}} = 1.0 \div 2.9\%\).
The scatter of temperature between thermocouples of one position at different points is characterized by the following data:
for the end-winding part, \(\sigma_{\mathrm{t}} = 4.5 \div 8.2\%\),
for the slot part, \(\sigma_{\mathrm{t}} = 4.5 \div 7.6\%\).
Fig. 2. Temperature field of motor AO2-32-4.
O — in the upper part of the slot,
X — at the bottom of the slot.
The temperature scatter between machines is characterized as follows:
for the end-winding part \(\sigma_M = 5 \div 11\%\),
for the slot part \(\sigma_M = 1.9 \div 3.5\%\).
The greatest scatter is observed in the end-winding parts of the winding, especially the scatter among machines.
Figure 3 shows the temperature distribution around the circumference of the frame. The temperature distribution around the circumference has several peaks.
The temperature difference between the right and left parts of the frame averages 27% of the winding temperature rise. Evidently, it is precisely this difference that causes the nonuniform heating of the end-winding parts on both sides of the stator. The reduction of the temperature skew in the winding to 5–9% is explained by its axial thermal conductivity. The average temperature rise of the frame is 50% of the winding temperature rise for 4-pole machines and 58% for 6-pole machines.
As for the stator winding, the temperature scatter around the circumference of the frame was determined. The root-mean-square deviation of the temperature between operating conditions is
\[ \sigma_p = 2.8 \div 5.8\%. \]
The root-mean-square deviation of the temperature between points around the circumference
Fig. 3. Distribution of temperature rises around the circumference of the frame.
\[ \sigma_{\mathrm{t}} = 12.8 \div 18.3\%. \]
The root-mean-square deviation of temperature between machines is
\[ \sigma_{\mathrm{m}} = 1.5 \div 3.0\% \]
as a percentage of the average temperature rise of the frame. As a percentage of the temperature rise of the winding it is 1.7–2.0 times smaller, i.e., of the same order as the scatter of the winding temperature, with the exception of the scatter among machines, which proves to be significantly smaller.
The temperature drop from the frame to the surrounding air constitutes, in machines of AO2 frame sizes 3 and 4, the largest share among the other drops that make up the temperature rise of the stator winding (from 50 to 70%). Therefore, reducing this drop on the basis of in-depth studies of external ventilation is promising. These studies are most conveniently carried out on physical models; more specifically, on electric-motor frames equipped with a heater and a fan driven in rotation by an auxiliary motor. This reduces the expenditure of time and resources for preparing and conducting experiments, achieves high accuracy in determining heat losses, and provides greater convenience when introducing various design changes into the ventilation system.
The disadvantages of models include the impossibility of investigating transient conditions.
To measure the temperature, 30 thermocouples were built into the frame according to the same scheme as in testing actual motors.
On the basis of the experimental data obtained, a dependence was derived for the average heat-transfer coefficient of the frame surface \(\alpha_{\mathrm{k}}\), referred to the full surface, on the average velocity of the air flow as it exits from under the casing,
\[ v_{\mathrm{avg.}} = \frac{V_{\mathrm{p}}}{S_{\mathrm{v}}}, \]
where \(S_{\mathrm{v}}\) is the cross-sectional area of the annular channel between the surface of the frame and the casing, taking into account the presence of ribs, pockets, and feet.
The empirical dependence was obtained in criterion form:
\[ \mathrm{Nu} = 0.004\,\mathrm{Re}^{0.65}; \qquad (0.3 \cdot 10^{5} < \mathrm{Re} < 3 \cdot 10^{5}) \]
\[ (\text{error not more than } 5\%) \]
\[ \mathrm{Re} = \frac{v_{\mathrm{cp}}D}{\nu} \quad \text{— Reynolds criterion,} \]
\[ \mathrm{Nu} = \frac{\alpha_{\mathrm{k}}D}{\lambda} \quad \text{— Nusselt criterion.} \]
As the characteristic dimension \(D\), the outside diameter of the ribbed frame was taken; \(\lambda\) and \(\nu\) are the coefficients of thermal conductivity and kinematic viscosity of air at ambient temperature.