Optimization of Heat-sink Fin Features using ANSYS. Jing Heng Wen E-Z-GO, Textron

Transcription

1 Optimization of Heat-sink Fin Features using ANSYS Jing Heng Wen E-Z-GO, Textron 1

2 Background: Heat sinks are frequently used when heat exchange is required for electric circuits. In cases where the heat generation is evenly distributed, the height and spacing of fins are uniform. However, when heat generation is locally concentrated, uniform fin distribution is not the most efficient design. Tasks of Virtual Prototyping: Virtual prototyping is applied to the investigation of fin feature design in order to reduce the cost when re-tooling is required. (1) To make sure that the new design will increase the heat transforming efficiency; (2) To reduce the weight. Procedures: (1) Analyze original design as the baseline; (2) Investigating heat generation sources; (3) Simple method to see the possibility for weight cut; (4) Optimize Fin Space, Height following the structure/installation requirement. (5) Validation 2

3 Optimization of Heat-sink Fin Features using ANSYS Original Product vs. Optimized Product Heat Exchange: Improve ~10-15% Material Reduction: ~10% of Heat Sink ANSYS, Inc. August 11, 2011

4 Step 1. Original Model (Baseline) Ambient T=50C, Max C Ambient T=20C Max C 4

5 Step 2. Heat Generating Source Investigation The investigation of heat generation: Concentrate Distribution Wall 1, at Longer Side Center Wall Wall 2, at Shorter Side Total Heat Source: 88W (1) Center Wall: 12W+10W+10W+10W+18W=60W (2) Longer Side: 15W(R) +7W = 22W (3) Shorter Side: 6W (R) 5

6 Step 3. Simple Method (Conduct/Film) for the Possibility of Weight Cut (A) Test Model 1 Amb T=20C, Max C Removing 3 Fins at the end of one side, Cut fin 0.05*3=0.15M Test Model 2 Ambient T=20C, Max C Increasing the height of 3 Fins M. So total adding is 0.075M 6

7 Step 3 (B). Simple Method to Find the Possibility of Weight Cut Summary of Test Model 1, Test Model 2 and Original Model Description LC 1, T bulk = 50C LC 2, T bulk = 20C Max. T Min. T Max/Min Difference Max. T Min. T Max/Min Difference Model 1 Original 108.9C 77.5C 31.4C 75.6 C 43.5C 32.1C Model 2 Removed 3 side fins 111.6C 84.3C 27.3C 77.8C 49.6C 28.2C Model 3 Removed 3 side fins 109.4C 83C 26.1C 75.8C 48.6C 27.2C 1.5 X H of 3 center Fins The results suggests that changing the height of fin and density could reduce the weight by increasing heat transform efficient. We know the possible method. Now, we need to know how to determine their height and space (density) in detail. Optimization analysis by FEA is used. 7

8 Step 4. Fin Density (Space) & Height Change, Model with Straight Fin Optimization Analyzing Base Model w/straight Fin (Step 4-1, 4-2 & 4-3) Max C under Bulk 20 0 C Straight Fin Baseline, Max. M=0.631kg Max C under Bulk 50 0 C 8

9 Step 4-1: Fin Density (Space) Change, Model with Straight Fin (A) (Optimization Analysis Model with Parameters) Model Eight (8) Parameters P20: w1 P27(3): b7 P26(3): b6 P23(4): b3 P22(4): b2 P21(3): b1 P25(3): b5 P24(3): b4 9

10 4-1 (B): Fin Density (Space) Change, Model with Straight Fin (Optimization -- Response Surface Result Local Sensitivity) b2 b5 b4 b3 b1 Min. Temperature b7 b6 w1 Max. Temperature 10

11 4-1 (C) : Fin Density (Space) Change, Model with Straight Fin (Optimization Analysis Result Sensitivity Chart) Max. T vs.. b1 and b7 Max. T vs.. b1 and b2 Max. T vs.. b1 and b6 Max. T vs.. b1 and b5 Hold b1, see how T changes (sensitivity) following b7/b6 (Left) and b5/b2 (right)? 11

12 4-1 (D): Fin Density (Space) Change, Model with Straight Fin (Goal Driven Optimization Analysis Result Summary) Best: 71.9 vs C (base), M=0.631 Tradeoff Max. Temperature vs. Mass Reduced Samples 12

13 Step 4-2: Fin Height Change, Model with Straight Fin & Equal Space (A) (Optimization Analysis - Model) C: 0.075m Four Parameters in DesignModeler Model P30:l1 P28:h2 P31:h3 P29:h4 APDL Command in Mechanical (add another parameter ARG1) 13

14 4-2 (B): Fin Height Change, Model with Straight Fin (Optimization Response Surface Local Sensitivity) Mass Max. Temperature Max. T Mass h3 h4 h3 h4 l1 h2 l1 P29:h4 h2 14

15 Mass vs. h2 and l1 4-2 (C): Fin Height Change of Model with Straight Fin (Optimization Analysis Sensitivity Chart) Mass vs. h2 and h3 Mass vs. h2 and h4 Max. T vs. h2 and l1 Model with Parameters Max. T vs. h2 and h4 P29:h4 Hold h2, see how T & Mass change (sensitivity) following l1 (best l1 for T, Left low), h4 (right) and h3 (center)? 15

16 4-2 (D): Fin Height Change of Model with Straight Fin (Goal Driven Optimization Analysis Set & Result) Best: Mass vs (base) T= C Tradeoff Max. Temperature vs. Mass Result of Optimized Model, Height Change only 16

17 Step 4-3. Combining Fin Height and Space Change (A) Optimization Model with Dimension Parameters + 8+4=12 Parameters P29: h4 17

18 4-3 (B). Combining Fin Height and Space Change Optimized 12 Parameters, Mass=0.576kg, T= C (Base M:0.631Kg,T: C) Optimized Parameters Ambient T=50C, Max. T=107 0 C Ambient T=20C, Max. T= C Inacceptable gap 18

19 4-3(C). Combining Fin Height and Space Change Re-adjusted 8 Parameters Toward to Market-Manufacture Acceptable Model - Optimization Parameter Table Detail of Design of Experiments Note - h3: Shipping package; l1: optimized; b7: gap limit; w1: limited; TL: defined 19

20 4-3 (D). Combining Fin Height and Space Change, Goal Driven Optimization Results Sensitivity Manuf. Acceptable - Verification 1. Selected as response point Optimized Candidates b2 b1 b6 b5 Mass b2 b1 b6 b5 Max. T & Mass Trade-off Max. T 2. Acceptable /Round for Manufacture 3. Pints Verification, Mass=0.573Kg, Max. T= C 20

21 Step 4-3(E). Combining Straight Fin Height and Space Change (Manufacture) Results of ALL Accepted Parameters, Mass=0.579kg, T=74.7C (Base Mass=0.631,T= C) Model Geometry Optimized /Accepted Parameters (Edit in Parameter Set Table/Refresh) Update Ambient T=20C, Max. T= C Ambient T=50C, Max. T= C 21

22 Step 5: Final Model Validation (A) Summary, Optimized vs. Original (Geometry & Results) Original Heat Sink Redesigned Heat Sink (Space/Height/Installation/Thick/Profile/Wave) Model FEA Thermal Analysis LC 4 (50C) LC 5 (50C) Volume ( Fin L ) LC 1, Ta=50C, LC 2, Ta=20C, Thermal Thermal + V (Fin L) V (Fin L) # Detail Report Max. T Min. T Max. T Min. T Stress F 11 lbf cm^3 (cm) Reduction Model 1 Original A C 77.5 C 75.6 C 43.5 C N/A N/A V547.82(L116.3) Base Model C C, enforced C 82.7 C 75.3 C 45.3 C 14 Mpa 150 Mpa V484.44(L101.7) 11.50%(13%) 22

23 Ambient T= C, Max C Max-Min= C Optimization of Heat-sink Fin Features using ANSYS Step 5 (B): Final Model Validation -- Compared with Original in FEA Thermal Final Model of Thermal Analysis Results (T+5%) Original Model of Thermal Analysis Results Ambient T=21 0 C, Max C Max-Min=30 0 C Ambient T=50 0 C, Max C Max-Min= C Ambient T=20 0 C, Max C Max-Min= C 23

24 Step 5 (C). Final Model Validation -- Thermal Structure Stress Check Static Structure Analysis Thermal Structure Analysis 24

25 Step 5 (D). Final Model Validation -- Prototype Test Test PASS/FAIL Criteria: Test 2. Operating Temperature: C to C. The power supply shall operate without any evidence of degradation throughout a Normal Temperature range of 0 0 C to C. For thermal protection above C, its output will be reduced by 2.5% / 0 C until reaches 50% power at 60 0 C, above which it will shutdown Test Figure 1: Current Thermal de-rating curve Test Figure 2: Current in normal T (-20 0 C to C) Test Figure 3: Charge parts temperature distribution All prototypes passed test successfully. 25

26 RESULTS & CONCLUSION In summary, ANSYS, a virtual prototyping tool, is used for Heat Sink Redesign: (1) The electric charge has been in mass production years and is very successful. The finding has been submitted to and is pending in US patent office. (2) ANSYS is providing a powerful tool to help us to decide whether improvement is accessible; (3) Optimization tool help us to determine the dimension rapidly and accurately ; (4) Saved cost and valuable time. ACKNOWLEDGEMENTS Without the full cooperation in R&D lab and product design team, the strong support and leadership from the management in E-Z-GO Product Engineering Department, the virtual prototyping could not be implemented. The author is indebted to many colleagues from E-Z-GO, Textron. 26