2 - Thermal Analysis of Electric Machines (1788)
2 - Thermal Analysis of Electric Machines (1788)
2 - Thermal Analysis of Electric Machines (1788)
11
Topics
Background
Thermal analysis methods
CFDvs Thermal Network
Motor-CAD capabilities
Cooling of Electric Machines
Fundamentals of heat transfer for electric
machine application.
2
Introduction
The power output of an electrical machine is
strongly affected by its thermal performance
because machine operating temperature limits the
electric loading, q = f(I)
Machine life
“10C half-life rule", every 10 degree Cincrease in
operating temperature cuts insulation life by half.
permanent magnet, the performance of magnets
decreases with temperature
the magnets will be demagnetized if theyare
overheated.
Copper loss, temperature dependent
Mechanical, elevated temperatures can induce
mechanical stresses due to thermal expansion
Bearing failure
Thermal fatigue, variable speed operation.
3
Traditional Motor Sizing Methods
sizing based on single parameter
thermal resistance Thermal Resistance
housing heat transfer coefficient
winding current density RTH [oC/W]
specific electric loading P [W]
thermal data from
simple rules of thumb Twinding Tambient
5 A/mm2, 12 W/m2/C etc.
tests on existing motors
competitor catalogue data Heat Transfer Coefficient
can be inaccurate h [W/(m2.oC)]
single parameter fails to describe
complex nature of motorcooling
poor insight of where to concentrate
design effort
next slide rules of thumbexamples
4
Typical Rules of Thumb Air Natural Convection
h = 5-10 W/(m2.C)
Air Forced Convection
h =10-300 W/(m2.C)
Liquid Forced Convection
h = 50-20000 W/(m2.C)
5
Need for electric motor thermal analysis
Simple motor sizing based on traditional methods such as
limiting winding current density or specific electric loading are
not good for optimisation
Depend on experience from the manufacturing process and material
used, so tend to become very inaccurate when trying something new
in a design
Give designer poor insight into where to concentrate design effort to
reduce temperature rise
6
More Sophisticated ThermalAnalysis
Techniques
Two methods are available:
Numerical Method
Finite Element Analysis
Computational Fluid Dynamics (CFD)
includes flow
7
Analytical Lumped Circuit Analysis
lumps together the machine parts with
similar temperature into discrete nodes
in anetwork.
Thermal resistance, heat transfer and
temperature difference are analogous to
electrical resistance, current and voltage.
The heat flow between the nodes is
modelled using thermal resistances, i.e.
conduction, convection andradiation.
Proven formulations for difficult heat
transfer mechanisms such as conduction
through bundles of wire in slot, complex
convection (empirical correlations),
interface gaps, etc.
Power losses are placed at nodes where
heat is generated.
8
Analytical Lumped Circuit Analysis
thermal resistances placed in the circuit to model heat transfer
paths
L
Conduction, th kA
R
emissivity () & view factor (F)from surface finish & geometry
thermal capacitances for transient analysis
Capacitance = Weight Specific Heat Capacity
9
Thermal Analysis Model Order
Thermal Model Order Elements/nodes Computational Note
time
Computational fluid High Millions hours/days Good to obtain convective
dynamics (CFD) heat transfer coefficient,
predict flowdistribution of a
cooling system, difficult to
perform thermal transients
10
CFD vs Thermal Network
For thermal network method, the machine is represented by an equivalent thermal
network based upon the machine’s geometry, materials, cooling methods, losses, etc.
For CFDmethod, the conjugate heat transfer problem is solved numerically based upon
the conservation of mass, momentum andenergy.
CFD,prepare clean geometry, generate high quality mesh, select turbulence model to
resolve the heat transfer and rotating flow problem.
Level of details
This makes thermal network method useful in the beginning of the machine design
process – to identify the important parameters that affect the thermal performance,
while CFDis used to fine tuning the thermal design for complex issues, e.g. flow
distribution, pressure drop, heat transfer for hairpin winding, waftereffects, etc.
11
CFD, FEA vs Thermal Network
We need to know ifthe complexities of heat transfer
can be modelled using thermal resistances.
Forced flow in cooling channels, e.g. developing flow,
noncircular duct, coolant type, etc.
effects of rotation in rotorcooling channels, airgap.
Flow resistance model the flow distribution in a cooling
system (fan characteristic and system flow resistance)
Winding heat transfer
12
Design process of electricalmachines
Prototype – cost to build is
very expensive Design Specification
13
Motor-CAD
Ultra-fast analytical network analysis
software dedicated to thermalanalysis of
electrical machines
input machine geometry using
dedicated parametric editors
select cooling type, materials, etc.
calculate steady state or transient
temperatures
all difficult heat transfer datais
calculated automatically
easy to use as motor designers may
not be heat transfer specialists
18 year of practical experiencebuilt
into the heat transfer and flow
algorithms
14
Motor-CAD Motor Types
Brushless Permanent Magnet (BPM)
Outer Rotor BPM
PMDC
Induction
Synchronous
Switched Reluctance
Claw Pole
15
Cooling Types Investigated
Motor-CAD includes proven models for an extensive range of
cooling types:
TENV: Totally enclosed non-ventilated
natural convection from housing
TEFC: Totally enclosed fan cooled
forced convection from housing
Through Ventilation
TE with Internal Circulating Air
Internal air circulating path
water jacket as heat exchanger
Open end-shield cooling
Water Jackets
axial or circumferential
Submersible cooling
Wet Rotor & Wet Stator cooling
Spray Cooling
e.g. oil spray cooing of end windings
Direct conductor cooling
e.g. Slot ducts with oil
16
Many Geometry Options
17
Flow Network Analysis
18
18
Numerical Thermal Analysis
Subdivide full machine into smaller models (high quality mesh formore
accurate numerical solutions):
Finite element analysis (FEA)
Only useful for conduction heat transfer
Computational fluid dynamics (CFD)
Can also simulate complex fluid flow
CFD
FEA
Motor-CAD Fluent
19
Motor-CAD Integrated Thermal FEA Solver
Motor-CAD fully integrated FEA thermal solver allows fast
calibration of winding thermal resistance network and
improved understanding of winding hotspot location.
20
Computational Fluid Dynamics (CFD)
Can be time consuming to
construct models and make
calculations
Useful to filter designsusing
Motor-CAD and just do CFD
on final candidate.
Fluent
Then use CFDresults to
calibrate Motor-CAD model,
convection and flow
formulations.
Assistance in CFDmodel setup
using links to Motor-CAD
being developed Fluent
21
Winding Temperature Resolution
The number of nodes in the lumped circuit model for the winding
can be varied depending on the temperature resolution required
and if there is an uneven distribution of losses in the winding
Improved hotspot prediction
22
Live Demo of Motor-CAD
23
Sensitivity Analysis
machine design involves some uncertainty due to the manufacturingprocess,
nature of turbulence, electromagnetic design,etc.
Sensitivity analysis is recommended to gain an in-depth understanding of the
main restrictions to cooling for a givendesign
Design effort can then be focused on the main cooling issues
24
2
4
Cooling of Electrical Machines
There are three mechanisms by which heat
(energy) is transferred:
Conduction
Convection
Thermal radiation
25
Conduction heat transfer
Heat transfer mode in a solid due to molecule vibration.
Good electrical conductors are also good thermal conductors.
Metals have large thermal conductivities due to their well ordered
crystalline structure,
k is usually in the range of 15 – 400W/m/K.
Solid insulators have not well ordered crystalline structure and areoften
porous.
k is typically in the range of 0.1 – 1W/m/K. (better than air with k
0.026W/m/K)
In an electrical machine, it would be desirable tohave also materials that
are good electrical insulators and have good thermal conductivity.
material research to try to achieve this.
the calculation of conduction is complex forcomposite
components such as winding, bearings, etc.
26
Conduction heat transfer
Heat transfer is transferred from high temperature to lowtemperature.
𝑘𝐴 𝑇ℎ𝑜𝑡 − 𝑇𝑐𝑜𝑙𝑑
𝑄 =
𝐿
𝑇ℎ𝑜𝑡−𝑇𝑐𝑜𝑙𝑑
Effectiveness of heat flow is determined by R, 𝑄 = .
Analogous to Ohm's law, 𝐼= 𝑉
𝑅.
Conduction thermal resistance is calculated using:
𝐿 𝐶
𝑅= [𝑢𝑛𝑖𝑡: ]
𝑘𝐴 𝑊
L = path length [m]
A = heat transfer cross section area [m2]
k = material thermal conductivity [W/m/K]
27
Heat extraction through conduction
high slot fill factor solutions (packing as
much wire as technological possible in the
slots of the electricalmachines).
28
Heat extraction through conduction
Sometimes slot liner material has significant impact on the
temperature rise between the winding and statorlamination.
significant research efforts have led to the development of
insulation materials that have a higher thermal conductivity
value.
Slot
liner
29
Heat extraction through conduction
30
Heat extraction through conduction
31
Radiation Heat transfer
heat transfer mode from a surface due to energy transferby electromagnetic
waves.
Radiation thermal resistance is calculated using:
1 𝐶
𝑅𝑟 = [𝑢𝑛𝑖𝑡: ]
ℎ𝑟𝐴
hr can be calculated using the formula:
𝜎𝜀𝐹1−2 𝑇4 −𝑇4
1 2
ℎ𝑟 =
𝑇1−𝑇2
hr = radiation heat transfer coefficient [W/m2/K]
A = area of radiating surface [m2]
σ = Stefan-Boltzmann constant (5.669x10-8 W/m2/K4)
T1 = absolute temperature of radiating surface[K]
T2 = absolute temperature of surface radiated to (ambient)[K]
ε = emissivity of radiating surface (ε < 1)
F1-2 = view factor (F1-2 ≤ 1) – calculated from geometry
32
Typical Emissivity (ε) Data
33
Convection heat transfer
Heat transfer mode between a surface and a fluid due to fluidmotion.
Natural Convection – fluid motion due tobuoyancy forces that arise from change
in density of fluid due to the temperature difference.
Forced Convection – fluid motion due to external forces (fan andpump)
Two types of flow
Laminar flow, streamline flow at lower velocities
Turbulent flow, turbulent eddies created at higher velocities
Convection heat transfer can be computed as:
Air Natural Convection
𝑄 = ℎ 𝑐𝐴 𝑇 𝑤𝑎𝑙𝑙 −𝑇 𝑓𝑙𝑢𝑖𝑑 h = 5-10 W/(m2.C)
Convection thermal resistance is calculated using: Air Forced Convection
1 𝐶 h =10-300 W/(m2.C)
𝑅𝑐 = [𝑢𝑛𝑖𝑡: ] Liquid ForcedConvection
ℎ𝑐𝐴
h = 50-20000 W/(m2.C)
hc = convection heat transfer coefficient [W/m2/K]
A = wall surface area [m2]
Convection depends on the heat transfer coefficient hc that is determined
empirically from test data or from CFDanalysis.
34
Convection heat transfer coefficient
As convective heat transfer is non-dimensionalized with Nusselt number (Nu),
hCcan be calculated using empirical correlations based on dimensionless
numbers (Re, Gr, Pr):
ℎ𝑐 𝐿
𝑁𝑢 = = 𝑓 Re, Gr, Pr
𝑘
L= characteristic length (m)
k = fluid thermal conductivity (W/m/K)
Dimensionless numbers allow the same formulation to be usedwith
different fluids, dimensions and models of dynamic and geometric
similarity to those used in the original experiments.
A lot of correlations are available inthe literature.
Motor-CAD automatically selects the most appropriate correlation that
matches the cooling type and geometry shape.
Cylinder, flat plate, open/enclosed channel,etc.
35
Convection Dimensional Analysis – Dimensionless Number
Reynolds number, 𝑅𝑒 = 𝜌𝑈𝐿
𝜇
𝑔𝛽𝜌 2 𝑇𝑤𝑎𝑙𝑙−𝑇𝑓𝑙𝑢𝑖𝑑 𝐿3
Grashof number, 𝐺𝑟 = 𝜇2
𝜇𝑐𝑝
Prandtl number, 𝑃𝑟 =
𝑘
μ = fluid dynamic viscosity[kg/m/s]
ρ = fluid density [kg/m3]
k = thermal conductivity of the fluid[W/m/K]
cp = specific heat capacity of the fluid[J/kg/K]
U = fluid velocity [m/s]
L= characteristic length[m]
β = coefficient of thermal expansion of fluid[1/K]
g = acceleration due to gravity[m/s2]
36
Natural Convection
General form of natural convectioncorrelation:
𝑁𝑢 = 𝑓 𝐺𝑟,𝑃𝑟 = 𝐶 𝐺𝑟 𝑃𝑟 𝑛
C& n are curve fitting constants.
Rayleigh number, Ra = Gr Pr
Transition from laminar to turbulent flow: 107 < Ra <109
𝑇𝑤𝑎𝑙𝑙+𝑇𝑓𝑙𝑢𝑖𝑑
All fluid properties are evaluated at the film temperature, 𝑇𝑓 =
2
The heat transfer coefficient is temperature dependent and it is needed tobe
solved iterative in Motor-CAD.
37
Horizontal Cylinder (Natural Convection)
Fluid properties at mean film temperature (average of surface and bulk fluid
temperatures)
38
Vertical Cylinder (Natural Convection)
formulation for average Nusselt number of a vertical
cylinder of height L:
Laminar flow, 104 < GrLPr < 109,
𝑁𝑢 = 0.59 𝐺𝑟𝐿𝑃𝑟 0.25
39
Vertical Flat Plate (Natural Convection)
formulation for average Nusselt number ofa
vertical flat plate of heightL:
Laminar flow, 104 < GrLPr < 109,
𝑁𝑢 = 0.59 𝐺𝑟𝐿𝑃𝑟 0.25
40
Horizontal Flat Plate (Natural Convection)
Upper face:
Laminar flow, 105 < GrLPr < 108, upper
𝑁𝑢 = 0.54 𝐺𝑟𝐿 𝑃𝑟 0.25
Turbulent flow, GrL Pr > 108,
𝑁𝑢 = 0.14 𝐺𝑟𝐿 𝑃𝑟 0.33
Lower face:
Laminar flow, 105 < GrLPr < 108,
𝑁𝑢 = 0.25 𝐺𝑟𝐿𝑃𝑟 0.25
41
Vertical Fin Channel (Natural Convection)
formulation for Nusselt number of U-shaped vertical channels
(laminar flow):
𝑟×𝐺𝑟 𝑃𝑟 0.75
0.5
𝑁𝑢 = 𝐿
× 1 − exp −𝑍 𝑟
𝑍 𝐿 𝐺𝑟 𝑃𝑟
1−0.483×exp −0.17/𝑎
𝑍 = 24× 3 3
1+𝑎 1+ 1−exp −0.83𝑎 ×9.14 𝑎×exp −465×fin spacing −0.61
2
42
Horizontal Fin Channel (Natural Convection)
Horizontal Fin Channels
0.44 0.17
−7640
𝑁𝑢 = 0.00067 ×𝐺𝑟𝑠 𝑃𝑟 × 1 −exp 𝑃𝑟
𝐺𝑟𝑠
s = fin spacing used as characteristic dimension
ℎ= 𝑁𝑢 𝑘/ 𝑠
Ref: Jones, C.D., Smith, L.F. : Optimum Arrangement of Rectangular Fins on Horizontal Surfaces for Free-Convection Heat Transfer, Trans. ASME, Feb1970.
43
Live Demo of Motor-CAD
44
Forced Convection
General form of forced convectioncorrelation:
𝑁𝑢 = 𝑓 𝑅𝑒,𝑃𝑟 = 𝐶 𝑅𝑒 𝑚 𝑃𝑟 𝑛
C, m and n are curve fittingconstants.
Reis a measure of inertial forces to viscousforces.
For external flow, laminar/turbulent transition Re 5 x 105
For internal flow,
The flow is assumed to be fully laminar when Re< 2300 in Circular/Rectangular
Channels and when Re< 2800 in Concentric Cylinders
The flow is assumed to be fully turbulent when Re> 5000 (in practice the flow may not
be fully turbulent until Re>10000)
A transition between laminar and turbulent flow is assumed for Revalues between
those given above.
45
Flat Plate Forced Convection (External Flow)
flow
formulation for average Nusselt number of flat plate (or horizontal cylinder) of length L:
For laminar flow, Re< 5 x 105,
Nu = 0.664 Re 0.5 Pr 0.33
For turbulent flow, Re> 5 x 105,
Nu = (0.037 Re0.8 – 871) Pr 0.33
h = Nu × k / L
Fluid properties at mean film temperature (average of surface and bulk fluid temperatures)
Ref: Incropera, F.P& DeWitt, D.P.: Introduction to Heat Transfer, Wiley,1990.
46
Enclosed Channel Forced Convection (Internal Flow)
Enclosed channel – Internal flow heat transfer correlations
𝑅𝑒 = 𝜌𝑈𝐷 ℎ
𝜇
h = Nu k / Dh
47
Enclosed Channel Forced Convection
Laminar Flow
Concentric Cylinders (adaptation of formulation for parallel plates which includes entrance length effects):
The 2nd term in the above equation is the entrance length correction which accounts for entrance lengths where the
velocity and temperature profiles are not fully developed.
Circular Channels (which includes entrance lengtheffects):
48
Enclosed Channel Forced Convection
Turbulent flow
Gnielinski's formula for fully developed turbulent flow (3000 < Re< 1×106)
𝑓 8 𝑅𝑒 − 1000 𝑃𝑟
𝑁𝑢 =
1 + 12.7 𝑓 8 1 2 𝑃𝑟2 3 − 1
𝑓 = 0.79ln 𝑅𝑒 − 1.64 −2
49
Enclosed Channel Forced Convection
Transition from Laminar to TurbulentFlow
It is noted that h increases dramatically as the flow regime changes from being laminar to turbulent
flow.
For transitional flow heat transfer, Nu is calculated for both laminar and turbulent flow using the
above formulations. A weighted average (based on Re) is then used to calculate Nu in the transition
zone.
50
Enclosed Channel Forced Convection
Useful hints
If users cannot model a particular duct using Motor-CAD existing
geometry options, users can model the duct using its hydraulic diameter.
As the actual duct is not physically modelled, then users need to correct
the convective surface area for thermal resistance, i.e. R=1/hA.
51
Enclosed Channel – Slot Water Jacket
52
Live Demo of Motor-CAD
53
Enclosed Channel – Housing Water Jacket
54
Enclosed Channel – Shaft Cooling
55
Spray / jet impingement Cooling
Copper loss is the major loss component
Copper loss (active) can be dissipated by housing
water (liquid) jacket through conduction heat
transfer
Direct end winding cooling is proposed
Electric traction drive machine application
Automatic transmission fluid (ATF) is available in
the vehicle’s transmission
Impinging automatic transmission fluid (ATF)jets
onto the end winding
Jet impingement – high heat transfer coefficient
The ability to use ATFto cool the rotor andend
winding directly has significant advantages in
reducing their temperatures.
[K. Bennion and G. Moreno, “Convective heat transfer coefficients of automatic transmission fluid jetswith
implications for electric machine thermal management,” in ASME 2015 International Technical Conference and
Exhibition on Packaging and Integration of Electronic and Photonic Microsystems (InterPACK2015), 2015, pp.1–9.
56
Spray Cooling
57
Spray Cooling
Formulation used in Motor-CAD for calculation the heat transfer coefficients of thespray cooling.
Very complex cooling method due to the fact of the multiphase flow, even difficult to be modelled
using CFD,good customer feedback.
Rough estimation of the impact of spray cooling on machine thermalperformance
Sensitivity analysis on spray cooling could be useful.
58
Spray / jet impingement Cooling 18AWG
26AWG
59
5
9
Experimental study of jet impingementcooling
For baseline sample (smooth)
The variation of convective htc with jet
impingement velocity
The experimental results are compared
against the available correlations in the
literature.
Reasonable agreement
The jet inlet temperatures haveminimal
influence on convective htc for smooth
surface
At jet velocity of 10m/s, the htc is about
8000 – 10,000 W/m2K.
60
Experimental study of jet impingementcooling
50°C inlet temperature 70°C inlet temperature 90°C inlet temperature
61
Experimental study of jet impingementcooling
62
Experimental study of jet impingementcooling
Differential pressure drop across the nozzle versus the jet velocity
The pressure drop decreases with increasing fluid temperature (lower fluid density)
1
Pressure loss coefficient, K= ΔP/ (2 𝜌𝑉 )2
63
Limitations / Issues / Complexities
On a larger scale, the heat transfer coefficient varies because not all of the endwinding
is directly impinged upon by the jet.
The variation in heat transfer (Nu) is due partly to the variation in the local heat transfer
coefficient away from the stagnation point of the jet impinging on the surface.
64
Open Fin Channel – Forced Convection
The empirical correlation used in Motor-CAD for open finchannels
is that of Heiles[1]:
0.214
𝐿0.946 𝑘𝐴𝑖𝑟
𝑚 = 0.1448 × ×
𝐷1.16 𝜌𝐴𝑖𝑟 ×𝑐𝑝𝑎𝑖𝑟 ×𝑉𝐴𝑖𝑟
D = Hydraulic Diameter = 4 x Channel Area / Channel Perimeter (including
open side)
L= Axial length of cooling fin
Air properties at the film temperature = (Tfree-stream +Twall)/2
It is assumed that the flow is always turbulent due to the fact that
the radial fans and cowlings used in such machines create
turbulence.
The h is directly multiplied by a Turbulence Factor in the rangeof
1.7 – 1.9 based on the tests, which seem independent of theflow [1] Heiles, F., "Design and Arrangementof
velocity. Cooling Fins", Elecktrotecknik und
Maschinenbay, Vol. 69, No. 14, July1952.
65
Axial Fin Channel – Air Leakage
For Blown Over (TEFC)machines, Motor-CAD assumes the air
leaks out of the open channels causing the local Air Velocity
to be lower at the drive end than at the non-drive end.
Leads to reduction in convection cooling further from the fan
66
66
Axial Fin Channel – Air Leakage
67
67
Axial Fin Channel – Air Leakage
68
68
Calibration of Flow Calculations using CFD
We are currently developing automated links from Motor-CAD to
CFDto allow automated calibration of such complex flow situation
as blown over semi-open channel leakage, end winding cooling,
water jackets, etc.
Fan editor been incorporated into Motor-CAD so we can export to CFD
69
MOTORDESIGNLIMITED
Thank you for your attention!
70
70