Thermal modelling for electrical machines fed with low voltage: First approach of a reliability model

https://doi.org/10.1016/j.matcom.2006.02.020Get rights and content

Abstract

Permanent magnet electrical motors are more and more currently used in car manufacture. A very constraining utilization reduces their life duration. Authors of this paper propose a new way to define reliability criterions of these low power machines. Most of internal fragility locations are winding insulations, ball bearings so that the system collector and brushes. Nowadays, statistical determination methods of the reliability criterions are very costly. For this reason, authors propose a new method consisting in the development of thermal models for different surrounding temperatures. It will also take into account degradation laws of constituting materials as functions of reached temperatures and life duration. This article will detail the initial validation of the thermal models for surrounding temperatures of 25 and 85 °C.

Introduction

Reliability constraints imposed by automobile manufactures for electrical machines are always more important, especially for new electrical machines fed with continuous current under low voltage. The studied machine is a low power one and is used as rotation motor of the fan. It permits also to chill the cooling water of the thermal engine by ventilation of the air-water thermal exchanger. Finally, the life duration of the electrical machine has to be similar as the life duration of the thermal engine and also the vehicle. Its reliability strongly depends on the manufacture process as well as the utilization speed of the driver. One of the many possibilities for studying the reliability of these motors consists in a statistical approach thanks to survival laws as Weibull or exponential laws among others things. Then, for this following methodology, automobile manufacturers have to realize a very high number of particularly constraining tests. Exception done for endurance classical tests (about 3 months of tests in a climate chamber), manufacturers require more and more tests bringing to the breakdown of the motor and allowing also to characterize all kinds of failure. All these realized tests destroy a very high number of prototypes 5–10 nowadays and 10–15 in a close future, with surrounding conditions (external conditions) which could be different case of classical endurance tests. These many tests will then have as consequence an important increase of technological auxiliaries as climate chambers, feeding systems or measurement systems. Authors of this paper intend to develop a new approach for this complex problem by proceeding to a thermal modelling of the motor. Indeed, failures of these machines come very often from heavy utilizations created also internal overheating. It is then fundamental to define internal heat and mass transfers. So to define and to valid this thermal modelling tool for different external surrounding temperatures, experimental phases and corresponding modelling have to be associated [2]. This association have to be obtained for surrounding temperatures of 25 and 85 °C. The numerical tool of calculus by mesh network is developed thanks to the software MATLAB. Experimental tests are realized in a specific climate chamber with machines instrumented with thermocouples of 100 μm diameter, conceived and realized in our laboratory. Obtained temperatures by measurements permit also to valid the numerical models. Many parameters as the rotating speed, the imposed torque so that the feeding current and voltage are then obtained. Used boundary conditions are Dirichlet ones which correspond to the surface temperatures. It can be measured thanks to thermocouples or thanks to a short waves infrared camera permitting to estimate simultaneously the temperatures on all the external surfaces. Internal thermal power sources are quantified, localized and separated thanks to different electromechanical tests realized with a test bench especially created in this way [1]. Fig. 1 shows the followed methodology which permits to valid the thermal model. Finally, a faithful knowledge of the thermal model and its evolution as function of the time could permit to define and to characterize some internal constraints at the origin of the failure. Then, a second experimental phasis consist in intentional and prematurely damages of the prototypes and bring to the determination and the integration of the failure laws as functions of time in the developed thermal model.

Section snippets

Mesh network

To develop a thermal model with a simulation tool imposes to characterize with a very good accuracy the heat and mass transfers inside the motor [5]. The internal structure is then separated (spatial discretisation) in three-dimensional isothermal volumes. Each volume V is located around a central position characterized by a central i node. This element is characterized thanks to its thermal conductivity (thermophysical parameter λ), eventually an internal thermal power source Q and the

Test bench

Fig. 2 presents a scheme representing the test bench. The tested motor can be seen in the climate chamber. Fig. 3 shows the machine with its fan which the role is to cool the hot water coming out of the thermal engine. Two similar tests are realized for two surrounding temperatures of 25 and 85 °C corresponding to criterions of validation of the manufacturer and European standards.

Different thermocouples are located in the studied prototypes on the two brushes (contactors on the ring of the

Experimental results

Fig. 4 shows the evolutions of overheating of the magnet, brush+ and smooth bearing as function of time. As seen in Fig. 4, temperature profiles seem to be linear at different ranges of temperature from 25 to 85 °C. The climate chamber is drived thanks to integrated software named SIRPAC. Beyond 25 °C and for taking into account the steady state thermal behaviour of the system fan-motor, the surrounding temperature of the climate room has been programmed each 10 °C for a 3 h duration.

Finally,

Validation of the model for a surrounding temperature of 85 °C

The model is considered as valid when obtaining a good convergence between the internal computed temperatures with experimental temperatures at the same geometrical locations. Table 1, Table 2 give and compare experimental and computed temperatures case of a three-dimensional model validated at 25 and 85 °C as surrounding temperatures. The final differences (experimental and simulated temperatures) for all the thermocouples locations constitute the validation criterion thanks to a minimum

Degradation laws

As shown on Fig. 5, thermal network model is solved thanks to the convergence of the vector of temperatures on two following calculus steps. After having validated the thermal model at t = 0 corresponding to a top operating system, it is then possible to solve it for t ≠ 0. This operating duration is traduced thanks to the evolution (degradation) of the thermophysical parameters at t ≠ 0. The taking into account of these degradation laws is shown in Fig. 4 with the integration of vector and matrix of

Conclusion and outlook

Thermal model has been validated from 25 to 85 °C. This model allows understanding the internal heat and mass transfers. It is then possible to locate with a good accuracy the location of internal and local overheating as function of internal and external constraints. This validation for the steady state thermal behaviour has needed to determine all the internal conductances of the studied structure for the 178 isothermal volumes. All these conductivities concern homogeneous materials, internal

References (5)

  • R. Bernard, Modélisation thermique par éléments finis en trois dimensions. Application aux machines électriques de...
  • R. Bernard et al.

    Modelling of the steady state thermal behaviour of a permanent magnet direct current motor with commutator. 3D finite elements study

    Eur. Phys. J. Appl. Phys.

    (1999)
There are more references available in the full text version of this article.

Cited by (4)

View full text
Copyright © 2006 IMACS. Published by Elsevier Ltd. All rights reserved.