BLDC Motor for Automotive Cooling Fan Assembly: Heat Sink Optimization Davide Parodi Fluid Dynamic and Aeroacoustic Engineer, Automotive Product Group, Johnson Electric Asti S.r.l. Asti, Italy Email: davide.parodi@johnsonelectric.com Web: www.johnsonelectric.com
Johnson Electric Asti - GATE Powertrain cooling Reference Motors Families 2
CFD Simulations in JE - Asti Commercial software (ANSYS) Part of design procedure Component optimization Data exchange with customers Data exchange with suppliers Research activities 3
Recent investigation activities 4
Brushless Electric DC Motor 5
The thermal stabilization test Thermal Stabilization in Oven: Required by some Customers for Product validation Regularly performed internally for design validation 6
Purpose of analyses Starting point: Technically optimized configuration Issues: Not completely profitable for lower power classes Heavy for lower power classes Ways: Investigation on new heat sink design and materials Investigation about influence of different gap pad materials Constraints: Electronics temperatures under prescribed safety limits: T safe = 155 [ C] Final design: lighter, cheaper, reliable, feasible, etc. 7
Preliminary options Heat sink concept modification From aluminum mold heat sink to metal stamped lamination Preliminary metal options: iron, steel Materials thermo-physics characteristics strongly different (4 times) Material for traditional aluminum heat sink: Thermal conductivity: ~ 240 [W m -1 K -1 ] Material for new steel heat sink: Thermal conductivity: ~ 60 [W m -1 K -1 ] Points for differentiation on production process Material thickness reduction necessary Significant exchange surface reduction Planarity issues Internal coupling material modification From traditional gap pad to gap filler paste Heat transfer conductivity typically reduced 8
Heat transfer mechanisms modeled Heat transfer problem: balance between contributions of conduction, convection and radiation. Radiation: present at every temperature value but important for high temperature values Temperature range for the present problem: < 180 [ C] Radiation negligible! Heat transfer mechanisms solved: conduction, convection Key parameters on heat transfer analysis: 1. Geometry (thickness, exchange surface) 2. Material properties (specifically: thermal conductivity) 3. Flow field characteristics (local air speed, temperatures, etc.) Complex fluid path and geometry Problem focused on heat rejection Realistic and reliable models should consider all parts affecting fluid motion 9
Complete fluid-dynamic and thermal analysis Detailed thermal-fluid dynamic model needed Previous models exploited: fluid domain; solid domain; fan fluid domain Original Model Size: up to 15 x 10 6 of elements (TETRA + PRISM; HEXA), even exploiting symmetric properties of turbo-machinery for fan Fluid domain model simplification without affecting input flow field by means of a cutting plane 10
MOTOR ASSEMBLY Solid interfaces - Aluminum heat sink to be compared with metal cover - PCB Electronic components - Plastic cover (FRAME) All of them have been modeled in the current CFD model From CFD Post-Processing, their impact can be simplified as surface input temperature, as per the following description ANSYS Automotive Simulation World Congress - 2013 Frankfurt; 29-30 October 2013 11
Complete model reduction Fluid dynamic Inputs (1) Previous CFD analysis used for fluid dynamic inputs to simplify models by means of user functions 95% similar geometry and same upstream thermal-fluid-dynamic field (Electric Power: 400 450 [W]) New input surface New geometrical model 12
Complete model reduction Fluid dynamic Inputs (2) Old velocity profile splitted on its three components v x, v y, v z plus T, TKE, TurbEddyDiss Original CFD model New geometrical model Old velocity profile re-used for the new model v x v y v z T air TKE Turbulence Eddy Dissipation 13
Plastic Frame-Motor Support Interface & Internal Air Material Aluminum motor support plastic frame interface: constant wall temperature Specific heat flux: 19 [W m -2 ] Exchange surface: 9.72 x 10-9 [m 2 ] Heat exchange: 1.847 x 10-7 [W] Negligible with respect to others Constant T wall Internal frame air domain: solid air Physics: natural convection! Max internal speed: 0.6 [mm s^-1] Small volume for fluid motion Air modeled as solid material 14
PCB Modeling Complex geometry and internal heat exchange Extreme anisotropy (composite materials, miniaturized, multilayer) Necessity to define resultant/corresponding material Proper materials thermo-physic characteristics modeled using both analytical approaches and fine tuning CFD analyses Input thermal powers derived from measured conditions International CAE Conference 21-22 October 2013 15
Thermal Inputs Main Materials DOMAIN MATERIAL Thermal Conductivity W mk Source Notes Gap Pad Gap Filler Paste Soft filled-polymer material Bi-Component resin + Spacer Beads 2.4 1.8 Supplier Data Sheet Board FR4 0.2 Supplier Data Sheet Coil Copper 401 ANSYS CFX Library Coil Core Ferrite 73 Supplier Data Sheet Capacitors Aluminum + PA 237 Modeled Driver Capacitors Aluminum + PA 237 Modeled MOS Fet Aluminum + Si + PA 237 Modeled Thermal VIAS Hybrid 8.94 Modeled Bus Bars Copper 401 ANSYS CFX Library Plastic Frame PBT 0.3 Supplier Data Sheet Aluminum Heat Sink Aluminum 237 ANSYS CFX Library Steel Heat Sink Iron Heat Sink Steel Iron 60 75 ANSYS CFX Library 16
Thermal Inputs Input Powers DOMAIN MOS Fet Input Power 2.77 [W] @ 150 [ C] @ 13 [V] @ 500 [W el ] 2.55 [W] @ 25 [ C] @ 13 [V] @ 500 [W el ] 3.24 [W] @ 150 [ C] @ 12 [V] @ 500 [W el ] 3.02 [W] @ 25 [ C] @ 12 [V] @ 500 [W el ] Specificic Dissipated Power W m 3 Source Notes 9.066 x 10 6 Calculated from real tests Coil 0.844 [W] @ 25 [ C] @ 13 [V] @ 500 [W el ] 0.987 [W] @ 25 [ C] @ 12 [V] @ 500 [W el ] 1.83 x 10 6 Calculated from real tests Capacitors 0.2 [W] Neglected Calculated from real tests Positive Bus Bar Negative Bus Bar 0.17 [W] @ 25 [ C] @ 13 [V] @ 500 [W el ] 0.2 [W] @ 25 [ C] @ 12 [V] @ 500 [W el ] 5.824 x 10 5 Calculated from real tests 0.2 [W] @ 25 [ C] @ 13 [V] @ 500 [W el ] 0.24 [W] @ 25 [ C] @ 12 [V] @ 500 [W el ] 8.097 x 10 5 Calculated from real tests Boundary Condition Averaged Input Temperature Source Notes External Environment 110 Thermal Stabilization Condition Motor Support/Frame Interface 130 Previous CFD Analysis 17
Aluminum Heat Sink simulation PRESENT CASE CHARACTERISTICS: Cells Numbers (TETRA+PRISM): ~ 10 x 10 6 Steady state analysis Turbulence Model (chosen after sensitivity studies): K - ϵ with Wall Function 18
Aluminum Heat Sink Post Processing 19
Aluminum Heat Sink - Thermal Stabilization Test 20
Aluminum Heat Sink simulation Post Processing Generally good agreement between real test and CFD analysis CFD Model satisfactorily representing real geometry CFD Analysis: Aluminum Heat Sink Q1 133.2 [ C] Q2 133.6 [ C] Q3 133.28 [ C] Q4 133.18 [ C] Q5 133.76 [ C] Q6 133.27 [ C] L1 ind 140 [ C] Coil temperature difference 1. Coil thermal power dependent on current 2. Conservativeness reason: current value simulated for coil corresponding to 500 [W], instead of 400 [W] 21
Metal Heat Sinks simulations (Steel and Iron) PRESENT CASE CHARACTERISTICS: Cells Numbers (TETRA+PRISM): < 10 x 10 6 Steady state analysis Turbulence Model: K - ϵ with Wall Function 22
Metal Heat Sinks Post-Processing (Steel and Iron) K Steel = 60 CFD Analysis: Steel Heat Sink Q1 183.66 [ C] Q2 189.54 [ C] Q3 185.52 [ C] Q4 183.96 [ C] Q5 189.06 [ C] Q6 182.21 [ C] L1 ind 176.25 [ C] W mk K Fe = 75 CFD Analysis: Iron Heat Sink Q1 178.88 [ C] Q2 184.36 [ C] Q3 180.7 [ C] Q4 178.39 [ C] Q5 183.07 [ C] Q6 176.92 [ C] L1 ind 172.28 [ C] W mk 23
Post Processing Results Aluminum, Iron and Steel K Al ~ 240 K Al = 237 K Steel = 60 W mk W mk W mk K Fe = 75 W mk Comparison trends in accordance with different material properties Necessity to improve heat rejection for components temperature reduction! 24
Reason of heat rejection decrease (Heat sink material) General conduction theory described FOURIER s EQUATION: q = Heat Flux [W] q i = KS T x i Simplified 1D form q l = KS T l R th = T q = l KS K = Thermal Conductivity of material [W m^-1 K^-1] S = Thermal Exchange Surface [m 2 ] T x i = Temperature Gradient along the i-direction [K m^-1] R th S = l K Δi = Material Thickness [m] R th = Specific Thermal Resistance [K W^-1] 2.0 10 3 R thal S = ~ 8.33 10 6 [K W^ 1] 240 0.5 10 3 R thst S = ~ 8.33 10 6 [K W^ 1] 60 Heat sink Material change alone not responsible of heat rejection decrease compensation effect in solid thermal resistance (reduced conductivity with reduced material thickness) 25
Reason of heat rejection decrease (Convection) (1) Convection mechanism: q = h A Heat Exchange T wall T air Heat path Simplified multilayer model T j = Junction temperature to be reduced q = T j T air R R = l T.V. K T.V. A T.V. + l G.P. K G.P. A G.P. + l H.S. K H.S. A H.S. + 1 h W A Heat Exchange S = Heat Source (MOSFet) T j = Junction Temperature [K] T.V. = Thermal Vias (conductivity K, Thickness l, Surface A) G.P. = Gap Pad (conductivity K, Thickness l, Surface A) H.S. = Heat Sink (with proper conductivity K, Thickness l, Surface A) T air = External Air Temperature q = Heat Flux moving from Source to air R = Global Thermal Resistance h = Convection coefficient T j ~ T air + q A l T.V. K T.V. + l G.P. K G.P. + l H.S. K H.S. + 1 h W Almost constant for convection 26
Reason of heat rejection decrease (Convection) (2) Shroud Vanes properly driving coolant flow Streamlines more detached from the surface of the heat sink 27
Considerations to Optimize CONDUCTION Different heat sink material Geometry thickness reduction Material higher conduction, cost compatibility Production feasible and easy solution T j ~ T air + q A + l H.S. K H.S. + CONVECTION Heat sink ventilation increase THERMAL EXCHANGE AREA Higher heat exchange surface Aluminum: 0.030402 [m 2 ] Steel/Iron: 0.0169583 [m 2 ] Resulting reduction: -44 [%] Geometry Heat sink ventilation increase ( flow driver ) Material Impact to be defined (costs, etc.) Production feasible and easy solution T j ~ T air + q A.. + 1 h W Geometry Highest feasible increase Material Choice affecting surface increasing Production feasible and easy solution T j ~ T air + q A.. 28
Convection improvement: first considerations Averagely 15 [ C] on external MOSFet (1, 2, 3) Averagely 10 [ C] on internal MOSFet (4, 5, 6) Around 12 [ C] on the coil CFD Analysis: Plastic Shroud Convective Vane Q1 169 [ C] Q2 173.83 [ C] Q3 170.72 [ C] Q4 174.75 [ C] Q5 182.03 [ C] Q6 172.96 [ C] L1 ind 163.77 [ C] 29
Important note on optimized shroud for Steel Heat Sink 1. Not completely enough 2. Not interchangeability Improved flow driving High positive impact Counter measurements 30
Optimization: needs and constraints - CONVECTION INCREASE Ventilation: Approaching to the correct flow driving effect Interchangeability - THERMAL EXCHANGE AREA Area increase: Lost surface difficult to be completely recovered Limitation due to process Increase with direct stamping on hot areas - CONDUCTION Different material: Higher conduction material: aluminum has been selected Parallel studies to find best trade-off between stamping process and minimum thickness feasibility Parallel studies to warranty reliable contact between PCB and heat sink 31
Sensitivity study for Heat Sink PCB interface geometry Gap pad and gap filler paste differently adapt to heat sink geometry Geometrical sensitivity: different filler shape, same filler material (K G.P. = K G.F. = 2.4 [W/m K]) Averaged Differences: < 0.18 % (Aluminum Heat Sink); < 0.75 % (Iron Heat Sink) Aluminum Heat Sink with Hybrid Model Q1 133.2 [ C] Q2 133.6 [ C] Q3 133.28 [ C] Q4 133.18 [ C] Q5 133.76 [ C] Q6 133.27 [ C] L1 ind 140 [ C] Aluminum Heat Sink with Gap Pad Model Q1 132.82 [ C] Q2 133.42 [ C] Q3 133.11 [ C] Q4 132.92 [ C] Q5 133.52 [ C] Q6 133.2 [ C] L1 ind 139.53 [ C] Iron Heat Sink with Hybrid Model Q1 178.88 [ C] Q2 184.36 [ C] Q3 180.7 [ C] Q4 178.39 [ C] Q5 183.07 [ C] Q6 176.92 [ C] L1 ind 172.28 [ C] Iron Heat Sink with Gap Pad Model Q1 178.25 [ C] Q2 183.73 [ C] Q3 179.7 [ C] Q4 175.85 [ C] Q5 181.64 [ C] Q6 175.14 [ C] L1 ind 170.95 [ C] 32
Sensitivity study for Heat Sink PCB interface : filler material K Fe = 75 W mk Averaged Temperature difference: < 1 [ C] Maximum Temperature difference: 1.5 2 [ C] Gap Filler Paste: Less conductive Gap Filler Paste: Thinner 33
Thinned Aluminum Heat Sink with integrated deflector Case 1: Minimum Thickness (0.5 [mm]) with Gap Pad (K G.P. = 2.4 [W m-1 K^-1]) Case 2: Higher Thickness (1.0 [mm]) with Gap Pad (K G.P. = 2.4 [W m-1 K^-1]) Heat Sink Material: STAMPED (0.5mm) ALUMINUM Q1 152.98 [ C] Q2 156.13 [ C] Q3 154.38 [ C] Q4 152.25 [ C] Q5 155.71 [ C] Q6 152.75 [ C] L1 ind 151.86 [ C] Heat Sink Material: STAMPED (1mm) ALUMINUM Q1 163.42 [ C] Q2 168.05 [ C] Q3 165.51 [ C] Q4 165.7 [ C] Q5 169.56 [ C] Q6 164.7 [ C] L1 ind 159.41 [ C] 34
Main CFD cases results comparison 35
Thinned Aluminum Heat Sink with integrated deflector: convection 36
Thinned Aluminum Heat Sink with integrated deflector: R2 (1) Bigger deflection vane Cleaner air passage Cross-checked for feasibility Necessity of gap filler paste (cohesion) Thickness: 0.8 [mm] (cracking) Reduced surface (cracking) 37
Thinned Aluminum Heat Sink with integrated deflector: R2 (2) CFD Analysis: Heat Sink, Integrated Vane (R2) 38
The new heat sink design (R2): real test 39
General conclusions Thanks to the present work a new evolution of a real motor cooling system has been designed with the following optimization: 1. Weight reduction 2. Reliability (internal electronics temperatures under safety limits) 3. Cost reduction (materials usage optimization) The use of CFD modeling has confirmed : Good reliability as simulation tool even with some simplifications due to the high complexity of the analyzed system Extreme utility in understanding main thermal exchange mechanisms importance Good tool to drive effective design decisions allowing to achieve a system configuration working and providing the searched requirements! 40