Flow and Thermal Analysis of a Two Pole TETV Motor using CFD

GRD Journals- Global Research and Development Journal for Engineering | Volume 5 | Issue 6 | May 2020 ISSN- 2455-5703

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD Sireesha Baile Senior Engineer COE-Computational Fluid Dynamics Lab, Corporate R & D Bharat Heavy Electricals Limited

Pavitran Dynampally Deputy Manager COE-Computational Fluid Dynamics Lab, Corporate R & D Bharat Heavy Electricals Limited

Abstract Ventilation studies in Motors with complicated geometry are generally carried out with analytical methods during design stage.

However, analytical methods do not provide comprehensive information of flow and temperature fields inside the motor at a system level. Therefore, CFD techniques are being extensively employed by motor manufacturers to analyze motor cooling systems. In the present work, Flow & Heat transfer analysis of Totally Enclosed Tube Ventilated (TETV) motor was carried out using ANSYS CFX. The complete fluid domain was modelled and heat loss data was defined on heat generating components. The ventilation flow circuit and calculated temperatures on critical components have been studied and end winding temperatures are compared with physical test data. The CFD results were within 5% of the tested data.

Keywords- Motors, CFD, Computational Fluid Dynamics, Thermal Analysis, Electrical Machines

I. INTRODUCTION Modern electric motors have more complicated mechanical and electromagnetic structures over conventional motors for gaining better performance, higher operation reliability and efficiency. The improvements are being primarily aided by modern simulation tools and high performance computers. Computational Fluid Dynamics (CFD) techniques are being extensively used for this purpose. As motor designs and processes grow in sophistication, motor-cooling problems have become too complex to solve analytically. This led engineers to perform numerical simulations to gain insight into the details of fluid flow and heat transfer processes in motors [1]. T.Bäuml et al. [2] presented the thermal simulation models for a totally enclosed fan cooled induction machine using thermal equivalent circuits.

As motor manufacturers are targeting to make motors compact, heat generation problems in electric motors increased.

Limited cooling capabilities cause degradations of the motor performance and operation reliability. CFD has played an important role to understand and overcome these problems by aiding designer to gain detailed insights of physical aspects of motor cooling.

The flow fields and temperature distributions even in the most inaccessible locations of a complex motor geometry parts can now be quantified and visualised from the detailed results of CFD simulation [1].

Most of the previous studies in the ventilation and cooling of the motors have been carried out on a section of the parts.

The rotor and stator parts are analysed separately by considering a sector by applying symmetry and periodic boundary conditions [3]. Sector analysis was primarily done due to meshing and computational time constraints [4]. Similar methodologies have been used by various researchers to carry out the ventilation studies in electrical machines [4-17]. Marco et al. [5] carried out thermal analysis by considering a 60° sector for a PM machine using STAR CCM+. Maxmilian et al. [6] conducted heat transfer studies based on CFD inside electrical machines by taking a section of the stator and applying periodic boundary conditions. Shanel et al.

[7] has carried out thermal analysis on a 90° sector citing to computational limitations. Unai et al. [8] presented a review for thermal design and analysis of electrical machines using different tools. The ability of CFD tools to simulate fluid flow and heat transfer for overall machines is emphasized. Anderson et al. [9] has carried out CFD study of forced air cooling and windage losses in a high speed electric motor. Complete 360° model was considered for the analysis, however, only flow analysis was carried out.

Lukasz et al. [10] carried out CFD analysis in an electric motor by considering one quarter of the model and making many geometric simplifications such as modelling end winding region as a torus, not modelling the air gaps in the rotor, etc., Pirooz et al. [11] presented a predictive CFD approach assessed for the flow of cooling air in a generator by considering a one half sector of the model. Hettegger [12] performed CFD simulations on an electrical machine by considering 1/8th of the rotor model, 1/10th of the stator and stator end windings. D.-D.Dang et al. [13] carried out a CFD analysis in a hydro generator rotor-stator system with a simplified model of 20° sector for the rotating domain and 5° sector for stationary domain Maximilian et al. [14] performed numerical analysis of heat transfer and flow of one periodic stator duct model. Moradnia et al. [15] in his work discussed the flow analysis in an electric generator model, performing simulations on a half scale model. Kral et al. [16] compared two methods of numerical analysis for a totally enclosed fan-cooled induction machine.

In the present work, flow and heat transfer studies in a two pole “Totally Enclosed Tube Ventilated (TETV)” motor is performed using CFD by considering the entire motor (360°). Complete motor analysis is carried out to avoid the inherent uncertainties involved with sector analysis. Moreover, the advancements in computational hardware and improvements in commercial CFD solvers have enabled to analyse the entire motor [17].

All rights reserved by www.grdjournals.com

15

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

In the field of motor ventilation and thermal studies, the present work shows that the full motor model can be analyzed using CFD. This enables to avoid the uncertainties involved in sector analysis where geometry simplification and periodic boundary conditions are considered. The complete ventilation flow circuit is studied and the windage loss is estimated. Further, heat transfer analysis has been carried out and the temperatures are compared with measurements taken during physical test. The CFD study has been carried out on a Mega Watt rating motor whose dimensions are spread over a couple of meters. This work is a significant one citing to the volume of the machine attempted for complete 3D analysis.

II. DESCRIPTION OF THE MOTOR Totally enclosed tube ventilated motors have closed-circuit cooling with a concentric tube nest cooler, which is an integral part of the frame with shaft-mounted fans (external fan and two main compartment fans) as shown in Fig. 1.

Fig. 1: Longitudinal section and cross section of the investigated TETV motor with (a) rotor shaft, (b and c) fans of the inner cooling circuit, (d) rotor packets, (e) rotor cooling ducts, (f) air gap, (g) stator core packets, (h) stator cooling ducts, (i) stator winding, (j) end windings of the motor drive end, (k) end windings of the motor non drive end, (l) inlet of the external circuit, (m) fan of the external cooling circuit, (n) cooling tubes and (o) outlet of the outer cooling circuit. Inner cooling circuit consists of two symmetrical circuits (IC1 and IC2) shown in red and External cooling circuit (EC) is indicated in green

It consists of two air circuits: Internal and External circuit, as shown in Fig. 1 and named as IC1 & IC2 as the inner cooling circuit (red) and EC as external circuit (green). In the internal circuit, entrapped air circulates to cool the motor parts. The two internal fans (b and c) in the main compartment circulates the air in two symmetrical sub-circuits (IC1 and IC2). Circulation is intensified by the radial ducts (e) in the laminated rotor core (d). The primary air circulates through the ducts in the stator core packets (h) and end windings (j and k), from which the heat is picked up and carried to the cooling tubes (n). External circuit consists of a unidirectional external fan (m) and is located in the fan housing on the non-drive end (NDE). It draws the ambient air and enables it to flow through the cooling tubes of the stator frame.

As the internal circuit is totally enclosed, it is important to understand the distribution of flow in sub-circuits and the temperature profiles of the heat generating parts. The critical part is end winding, where the limiting temperature is decided by the insulation class, therefore it is important to monitor the temperature on the end winding. Since the convective heat transfer depends highly on the geometry of the end region [11], the complete scheme of the end winding has been modelled to predict the temperatures accurately. The objective of the study is to predict the flow and temperatures by applying actual physical conditions.

The motor has been analysed by a taking full scale model, with no geometrical assumptions and simplifications. The results have been validated with physical test measurements conducted in shop floor. The cool air inside the tubes picks up the heat from the surrounding hot air outside the tubes (shell side) by means of convection. The internal air is continuously circulated through internal fans (b &c). External circuit consists of a fan which throws cool air into the tubes stacked circumferentially (n) to cool the internal hot air by convection.

III. CFD ANALYSIS A. Pre-processing 1) Fluid Path Extraction The fluid domains in the various parts of the motor have been extracted as water tight bodies from the 3D solid model. This technique helped in faster extraction and accurate fluid model generation.

All rights reserved by www.grdjournals.com

16

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

2) Grid Generation Grid was generated using ICEM-CFD software. Segregated approach was followed i.e. mesh was generated in each component and then linked with General Grid Interface (GGI) in CFX Solver.

Tetrahedral elements easily fit into the complex geometries and are used for fast and automated meshing in CFD. Besides these advantages, tetra meshing compromises on computational accuracy and stability. For solving such complex and huge problems, it is necessary to use Hexahedral meshing to avoid computational instability [4],[13]. The meshing strategy followed in the present analysis is hybrid meshing – a combination of hexahedral and tetrahedral meshing based on the complexity of the domain.

Fig. 2: Hexahedral mesh of rotor core and stator core

A case study on meshing of the rotor to demonstrate the use of tetra and hexa meshing has been done. Firstly, a complete 360° model of the rotor has been meshed using tetra which resulted in 40 million cells. Subsequently, the radial ducts and rotor windings have been meshed using hexahedral elements and the complex rotor end plates using tetrahedral elements. The use of hexahedral meshing lowered the mesh count to 4 million, with better mesh quality. Fig. 2 shows the hexahedral mesh of the rotor and stator core. The hybrid meshing strategy is used for the rotor and stator core to reduce the mesh count and increase the computational stability.

A case study on meshing of the rotor to demonstrate the use of tetra and hexa meshing has been done. Firstly, a complete 360° model of the rotor has been meshed using tetra which resulted in 40 million cells. Subsequently, the radial ducts and rotor windings have been meshed using hexahedral elements and the complex rotor end plates using tetrahedral elements. The use of hexahedral meshing lowered the mesh count to 4 million, with better mesh quality. Fig. 2 shows the hexahedral mesh of the rotor and stator core. The hybrid meshing strategy is used for the rotor and stator core to reduce the mesh count and increase the computational stability.

Fig. 3: Mesh of end windings and the intricate gaps in the model

The fine mesh on end windings resulted in an average y+ of 14 (see Fig. 4).

Fig. 4: Y+ contour of end windings

All rights reserved by www.grdjournals.com

17

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

The heat transfer between the external and internal circuit is an important phenomenon. To predict heat transfer in a better manner, care has been taken to resolve the boundary layer with an average y+ around 13. Fig. 5 shows the mesh generated for the external circuit using tetrahedral and prism elements in the tube walls and the y+ contour is shown in Fig. 6.

Fig. 5: Tetrahedral mesh with prism layers of the tubes domain of external circuit

Fig. 6: Y+ contour of tubes of the external circuit

Mesh sensitivity analysis has been carried out with three different mesh sizes (coarse: 55, fine: 60 & finer: 80 million).

The results did not change significantly between the three mesh sizes. The simulations were carried out by considering the coarser mesh size (55 million). Most of the previous works in this field have reported a mesh count of 2 to 4 million cells [2] - [4] and recent simulations with a mesh count of 22.5 million cells [13] by considering only a sector for analysis.

B. Boundary Conditions All the meshed components are imported into CFX-Pre for defining the physics and boundary conditions. Domains are defined for each component, fluid-fluid type interfaces are defined between adjacent meshes. Rotor and Fan domains are considered as rotating domain by defining a constant rotational speed of 3000 rpm. Frozen rotor type interface was defined between stationary and rotating domains. Frozen rotor has been chosen over mixing plane as the later averages the fluxes at the Rotor Stator Interface and removes the flow structure at the interface. The Frozen rotor model has been validated in several test cases [14].

The heat losses generated in the machine have been taken from the experimental data and was applied as surface heat flux [W/m2] on walls of the heat generating areas. In the stator, iron, copper & stray losses are applied and in the rotor, stray & rotor copper losses are applied. The values have not been reported due to confidentiality reasons.

In the external circuit, mass flow rate [kg/s] at inlet and average static pressure [Pa] at outlet are defined. In addition to the above, thin wall boundary condition is applied to the tube walls of internal and external circuit to account for the heat transfer by defining the thickness of the tube.

The following solver settings were defined: – Steady state analysis – k-epsilon turbulence model – High resolution advection scheme.

– Convergence criteria 1E-05 was set for all the governing equations of fluid.

K-epsilon turbulence model was considered as it requires relatively higher first cell size (y+ ~ 20 to 30) in boundary layer, which results in reasonable mesh count [13]. Moreover, the results predicted using k-epsilon model was within acceptable deviation, which is discussed in results section. SST model was not considered as it significantly increased the total mesh count and due to limitations of our present hardware capability.

All rights reserved by www.grdjournals.com

18

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

C. Solver ANSYS CFD solver v15.0 was used to perform the simulations. The solver basically solves the fluid governing equations (1, 2 & 3) Continuity equation (Based on the law of conservation of mass)

  U i   0  t xi

(1)

Momentum equation (Based on the law of conservation of momentum)

 U i



 t



U iU j  P    x j  xi x j

 U i     ui u j   S  x  j   (2)

Energy equation (Based on the first law of thermodynamics)

 C p T



 t



 C pU i T    T k   u j t   S~t   x j  x j  x j 

(3)

ρ – Density [kg/m ] U – Time averaged velocity [m/s] Ui – Circumferential velocity [m/s] P – Static pressure [Pa] S – Additional momentum sources e.g. Coriolis and centrifugal forces for steady rotating frame of reference T – Temperature [K] µ - dynamic viscosity [Pas] Cp – Specific heat capacity [J/kgK] k – Thermal conductivity [W/mK] St – Source term [W] Subscripts i - 1, 2 and 3 corresponds to components in x, y and z directions respectively ANSYS CFD v15.0 is an implicit algebraic multi-grid solver. The computational time taken to solve the present 60 million mesh count problem was about 30 hours with 8 parallel processors on a High-Performance workstation [Intel Xeon CPU @2.60 GHz processors, 128 GB RAM].

3

Fig. 7: Ventilation flow circuit of the TETV motor with internal circuit (IC1, IC2) indicated in orange and external circuit indicated in green

All rights reserved by www.grdjournals.com

19

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

IV. RESULTS Flow and Temperature distribution of air inside the motor has been analyzed by plotting contours in various planes and discussed in below sections.

A. Flow Analysis The schematic of the ventilation circuit is shown in Fig. 7. Different domains have been indicated as per the divisions made while fluid domain extraction and meshing and the values of flow indicated in the figure are normalized values shown at the interfaces.

For the inner cooling circuit, the normalized mass flow rate ֹmNi has been calculated as the ratio of the mass flow rate at each interface ֹmi (in kg/sec) and mass flow rate entering the stator housing ֹmSH (in kg/sec)

(4) For the external circuit, the normalized mass flow rate, ֹmNe is the ratio of mass flow rate at boundary, ֹmb and mass flow rate at inlet, ֹmIN.

(5) The streamlines depicting the flow in the inner cooling circuit with its two sub circuits is shown in Fig. 8.

The line diagram in Fig. 7 shows that the two sub circuits in the inner flow circuit are not exactly symmetrical. Overall, they follow a similar pattern. Some legends and parameters are suppressed in the following figures due to proprietary reasons.

Fig. 8: Streamlines showing the flow in the inner cooling circuit – two sub circuits

Fig. 9: Velocity contour showing the flow in rotor packets (A) and stator packets (B)

The reason for the asymmetry can be attributed to the non-uniform distribution of the rotor packets (A) and different number of rotor and stator packets (B) shown in Fig. 9. This particular distribution of flow inside the motor could be predicted as the complete 360° has been modelled, which otherwise would not be possible through analytical methods and sector analysis [17].

B. Temperature Analysis The temperature of the windings is an important and critical parameter for the proper functioning of the motor. The maximum temperature of the windings should be within the limiting temperature of insulation class. The temperature on the windings is dependent on the flow around the windings. So, the end windings are modelled realistically, as per the manufacturing drawings [12]. The maximum temperature of the windings predicted by CFD is below the limiting insulation class temperature.

All rights reserved by www.grdjournals.com

20

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

C. Measurement and Validation The data considered for comparison has been taken from the type tested data of the motor on test bed. The physical test procedure of the induction motors has been done as per IEC 6034 standard. Due to confidentiality, the test layout and other details are not reported. The no load test on the squirrel cage induction motor is conducted to measure the rotational losses of the motor and to determine some of its equivalent circuit parameters. In this test, a rated, balanced ac voltage at a rated frequency is applied to the stator while it is running at no load.

The losses from the locked rotor test and no load saturation test have been given as input for the CFD analysis. The windage losses from CFD were compared with the test data and were found to be within 10%. Liu et al. [18] predicted windage losses using analytical methods and the predictions were within 15% of windage tests. The deviations were attributed to the simplified assumptions used in the prediction calculations. The improvement in the agreement of predicted CFD results with test data can be attributed to the complete realistic analysis of the flow regime.

During testing, the temperature of the windings are measured at the NDE side by placing the RTD’s (Resistance Temperature Detectors) at a distance of 20 mm from the stator core end and in between the top and bottom windings as shown in Fig. 10. Twelve numbers of RTD’s have been placed and distributed circumferentially (Section A-A of Fig. 10 show the approximate location of the temperature measurement of windings). These temperatures of windings from the physical testing is referred as test data. The temperatures at these measured locations are compared with CFD calculated temperatures and deviation was found to be within 5 %. Fig. 11 shows the comparison of temperatures between CFD and test data. A graph has been plotted to show the variation of the winding temperatures with respect to the slot locations. The winding temperatures plotted on Y axis are the normalized values, which is a ratio of the temperature of the winding slot (T) and maximum winding temperature from the test data (Tmax_test). The orange line represents the winding temperatures from testing and blue line from CFD. The trend agrees well.

Fig. 10: Temperature measurement in the motor – RTD locations shown in section A-A

Fig. 11: Comparison of winding temperatures – CFD and test data

All rights reserved by www.grdjournals.com

21

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004)

Fig. 12 and Fig. 13 show the velocity and temperature contours in mid plane of the TETV motor. Some legends and parameters are suppressed due to confidentiality reasons. It can be observed that the temperature of the air is higher where velocities are lower which is as expected. Such contours give detail insight to the designer for carrying out design optimization.

Fig. 12: Velocity contour along the motor in XY plane

Fig. 13: Temperature contour along the motor in XY plane

V. CONCLUSIONS This paper has presented the results of using the commercial CFD package to simulate the flow and heat transfer analysis in a TETV motor. The primary objective of this work was to establish a CFD methodology to carry out flow and thermal analysis in the entire motor. This study is significant as it provides the motor designer a means to assess the qualitative and quantitative aspects of flow & temperatures inside the motor. The enclosed internal circuit could be visualized and asymmetry of flow is observed in the sub-circuits. The windage loss predicted by CFD is within 10% of the test data. The winding temperatures calculated by CFD are found to be in good congruence with the test data with a maximum deviation of 5%, thus affording a conservative, correlated CFD model.

REFERENCES [1] [2] [3] [4] [5] [6] [7]

Aldo Boglietti, Andrea Cavagnino, David Staton, Martin Shanel, Markus Mueller, and Carlos Mejuto, “Evolution and Modern Approaches for Thermal Analysis of Electrical Machines,” IEEE Transactions on Industrial Electronics, vol. 56, no. 3, pp. 876-882, March 2009.

T.Bäuml, C.Kral, A.Haumer and H.Kapeller, “Enhanced Thermal Model of a Totally Enclosed Fan Cooled Squirrel Cage Induction machine,” IEEE International Electric Machines & Drives Conference, pp. 1054-1058, 2007 S J Pickering, D Lampard, M Shanel, “Modelling Ventilation and cooling of the rotors of salient pole machines” Electric Machines and Drives Conference 2001, pp. 806-808, 2001.

S J Pickering, D Lampard, M Shanel, “Ventilation and heat transfer in a symmetrically ventilated salient pole synchronous machine,” International Conference on Power Electronics, Machines and Drives, pp. 462-467, April 2002.

Marco Tosetti, Paolo Maggiore, Andrea Cavagnino and and Silvio Vaschetto, “Conjugate Heat Transfer Analysis of Integrated Brushless Generators for More Electric Engines,” IEEE Transactions on Industry Applications, vol. 50, pp.2467-2475, 2014.

Maximilian Schrittwieser and Oszkar Biro, “Validation of measurements with conjugate heat ransfer models,” The international journal for computation and mathematics in electrical and electronic engineering, Vol.32 no.5, pp. 1707-1720, 2013.

M.Shanel, S.J.Pickering, D.Lampard, “Conjuagate Heat transfer analysis of a salient pole rotor in an air cooled synchronous generator,” IEEE, 2003.

All rights reserved by www.grdjournals.com

22

Flow and Thermal Analysis of a Two Pole TETV Motor using CFD (GRDJE/ Volume 5 / Issue 6 / 004) [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

Unai SanAndres, Gaizka Almandoz, Javier Poza, Gaizka Ugalde, “Design of cooling systems using computational fluid dynamics and analytical thermal models,” IEEE Transactions on Industrial Electronics, Vol. 61, No.8, August 2014.

Kevin R. Anderson, Jun Lin, Chris McNamara, Valerio Magri, “CFD Study of Forced Air Cooling and Windage Losses in a High Speed Electric Motor,” Journal of Electronics Cooling and Thermal Control, 2015, 5, 27-44, June 2015.

Lukasz Slupik, Jacek Smolka, Luiz C. Wrobel, “Experimentally validated numerical model of coupled flow, thermal and electromagnetic problem in small power electric motor,” Computer Assisted Methods in Engineering and Science, 20: 133–144, 2013.

Pirooz Moradnia, Valery Chernoray, Hakan Nilsson, “Experimental assessment of a fully predictive CFD approach, for flow of cooling air in an electric generator,” Applied Energy 124, pp.223-230, 2014.

Martin Hettegger, Bernhard Streibl, Oszkr Biro and Harald Neudorfer, “Measurements and Simulations of the Convective Heat Transfer Coefficients on the End Windings of an Electrical Machine,” IEEE Transaction on Industrial Electronics, Vol.59, No.5, May 2012.

D.D.Dang, X.T.Pham, P.Labbe, F.Torriano, J.F.Morissette, C.Hudon, “CFD analysis of turbulent convective heat transfer in a hydro-generator rotor-stator system,” Applied Thermal Engineering 130, pp. 17-28, 2018.

Maximilian Schrittwieser, Andreas Marn, Ernst Farnleitner and Gebhard Kastner, “Numerical Analysis of Heat Transfer and Flow of Stator Duct Models,” IEEE Transactions on Industry Applications, Vol. 50, No.1, 2014.

Piroz Moradnia, Maxim Golubev, Valery Chernoray, Hakan Nilsson, “Flow of cooling air in an electric generator model – An experimental and numerical study,” Applied Energy 114, pp. 644-653, 2014.

Christian Kral, Anton Haumer, Matthias Haigas, Hermann Lang, Hansjorg Kapeller, “Comparison of a CFD analysis and thermal equivalent circuit model of TEFC Induction machine with measurements,” IEEE Transactions on Energy conversion, Vol. 24, No.4, December 2009.

Sireesha Baile, D Pavitran, “Ventilation flow analysis of 2-pole TETV motor,” Proceedings of ANSYS Conference 2015.

Hsing-Pang Liu, Mike Werst and Jonathan J.Hahne, “Prediction of Windage Losses of an Enclosed High Speed Composite Rotor in Low Air Pressure Environments,” ASME 2003 Heat Transfer Summer Conference, pp. 15-23, 2003.

All rights reserved by www.grdjournals.com

23