Battery Thermal Management and Design. Scott Stanton Xiao Hu ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary

Transcription

1 Battery Thermal Management and Design Scott Stanton Xiao Hu ANSYS, Inc ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary

2 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 2 ANSYS, Inc. Proprietary

3 WorkBench An Integrated Solution for Battery CFD Analysis DM : Geometry tool with full parametric capability Project page: Defines the work flow WB Mesher: Quality meshing with automation CFD post : takes advantage of CFX postprocessing capability 2010 ANSYS, Inc. All rights reserved. 3 ANSYS, Inc. Proprietary

4 Build-in What-if Study in WB More uniform temperature across cells with smaller gap due to higher velocity at the same mass flow rate inlet First and last cells have higher temperature due to lower velocity without the blockage effect ANSYS, Inc. All rights reserved. 4 ANSYS, Inc. Proprietary

5 HEV Battery Thermal Management Input with Variations Gap Thickness Cell Resistance Flow Rate Outputs with variations Max temperature Differential temperature Pressure drop 2010 ANSYS, Inc. All rights reserved. 5 ANSYS, Inc. Proprietary

6 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 6 ANSYS, Inc. Proprietary

7 Motivation of Using Model Order Reduction CFD as a general thermal analysis tool is accurate but Can be expensive for large system level repeated transient CFD analysis Can be cumbersome to couple with electrical circuit model for large system analysis 2010 ANSYS, Inc. All rights reserved. 7 ANSYS, Inc. Proprietary

8 Motivation of Using Model Order Reduction Seek reduced order models for system level transient analysis Thermal network (use thermal resisters and capacitors, etc) Compromised accuracy Needs careful calibration and calculation of thermal resistance, capacitance LTI method (Foster network, state space) Can be as accurate as CFD or even testing No need to calculate thermal resistance, capacitance Rely on linearity and time invariance I1 R1 C1 + V VM1 R2 C ANSYS, Inc. All rights reserved. 8 ANSYS, Inc. Proprietary

9 Reduced Order Model Extraction Process Using Simplorer Create step responses From CFD / Test 2. Generate.simpinfo file 3. Extract equivalent thermal model Use Simplorer 4. Simulate inside Simplorer 2010 ANSYS, Inc. All rights reserved. 9 ANSYS, Inc. Proprietary

10 Six Cell Test Case Geometry/Mesh Cell 4 Cell 5 Cell 6 Cell 3 Cell 2 Cell 1 Inputs: heat source to each battery Outputs: battery volume average temperature 2010 ANSYS, Inc. All rights reserved. 10 ANSYS, Inc. Proprietary

11 Foster Network for the Battery Module in Simplorer This Foster network can be automatically generated by Simplorer!! Some cross heating elements have negligible contribution (less than 0.1% compared with self heating) and thus no Foster network Reduce the computational effort ANSYS, Inc. All rights reserved. 11 ANSYS, Inc. Proprietary

12 Foster Network vs CFD Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6 Foster network and Fluent CFD give identical solution under arbitrary sinusoidal power inputs Systems represented by impulse response, Foster network, and the original thermal system are all equivalent systems ANSYS, Inc. All rights reserved. 12 ANSYS, Inc. Proprietary

13 LTI Approach for Flow Rate Change of 100% Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6 Power inputs are sinusoidal functions Flow rate changes at time of 1000 second. Results are excellent for the entire duration. A small difference is seen during transition period ANSYS, Inc. All rights reserved. 13 ANSYS, Inc. Proprietary

14 LTI Approach for Non-Linear Problems Non-linear CFD: Ideal gas law plus temperature dependent properties are used. Full Navier-Stokes equations are solved LTI: Assumes the system is linear and time invariant. A speed-up factor of 10,000 is observed. Huge time saving if the error, which is about 2%, is acceptable ANSYS, Inc. All rights reserved. 14 ANSYS, Inc. Proprietary

15 LTI Model Extraction for a General Motors Battery Module Example State space model gives the same results as CFD. State space model runs in less than 30 seconds while the CFD runs 2 hours on one single CPU. 1. X. Hu, S. Lin, S. Stanton, W. Lian, A Novel Thermal Model for HEV/EV Battery Modeling Based on CFD Calculation IEEE Energy Conversion Congress and Expo, Atlanta, Sep 12-16, X. Hu, S. Lin, S. Stanton, W. Lian, A State Space Thermal Model for HEV/EV Battery Modeling", 2010 ANSYS, SAE Inc All rights reserved. 15 ANSYS, Inc. Proprietary

16 Additional Information on LTI Separate training material exists for LTI method ANSYS, Inc. All rights reserved. 16 ANSYS, Inc. Proprietary

17 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 17 ANSYS, Inc. Proprietary

18 Electrical Circuit Model Motivation Simple enough for system level analysis Models based on detailed electrochemistry can be too complex and/or too time consuming for system level analysis Accurate enough for virtual prototyping Non-linear circuit voltage as a function of SOC Transient I-V performance Runtime prediction Discharge rate dependent capacity Temperature effect Accurate transient temperature prediction 2010 ANSYS, Inc. All rights reserved. 18 ANSYS, Inc. Proprietary

19 Battery Cell Electrical Model Ref: Chen et al* Accounts for non-linear opencircuit voltage Capable of predicting runtime Error less than 0.4% Capable of predicting transient I-V performance Error less than 30-mV Can be implemented easily in circuit simulator Implemented in Simplorer Results from Simplorer Results from Chen * Reference: M. Chen, G. A. Rincon-Mora, Accurate electrical battery model capable of predicting Runtime and I-V performance", IEEE Trans. On energy conversion, vol. 21, no. 2, June ANSYS, Inc. All rights reserved. 19 ANSYS, Inc. Proprietary

20 Experimental Observation Ref: Gao et al* Chen s model works OK compared with testing data. Under constant temperature and discharge rate Rate effect and temperature effect are important to consider Impact of discharge rate The discharge history is sensitized to rate of discharge and temperature through rate factor(α) and temperature factor(β) State of Charge: * Reference: L. Gao, S. Liu, and R. A. Dougal, Dynamic lithium-ion battery model for system simulation, IEEE Trans, Compon. Packag. Technol., vol. 25, no. 3, pp , Sep Impact of temperature 2010 ANSYS, Inc. All rights reserved. 20 ANSYS, Inc. Proprietary

21 Thermal Network Model for Liion Battery: 1 Cell Three node cell thermal network model Rcond Rcond T1 T2 Tptc Rconv Rconv Rconv htc Ambient Two temperature nodes for the battery Separate temperature node for Positive Temperature Coefficient (PTC) PTC has higher temperature under high load condition CFD can be used to provide heat transfer coefficient 2010 ANSYS, Inc. All rights reserved. 21 ANSYS, Inc. Proprietary

22 Electrical + Thermal Network Ref: Gao et al* Electrical circuit and thermal circuit are coupled Includes Positive Temperature Coefficient (PTC) Voltage PTC and Battery Temperature Normal Discharge with a load of 10 Ω Voltage PTC and Battery Temperature Overloading Discharge with a load of 2 Ω Simplorer Model with PTC and 3 T Nodes Implemented using VHDL-AMS * Reference: L. Gao, S. Liu, and R. A. Dougal, Dynamic lithium-ion battery model for system simulation, 2010 ANSYS, Inc. IEEE All rights Trans, reserved. Compon. Packag. Technol., vol. 25, no. 3, 22pp , Sep ANSYS, Inc. Proprietary

23 Complete Circuit Model for Li-ion Battery: 1 Module Electrical circuit and Foster network are coupled Electrical circuit provides power to Foster network Foster network provides temperature to electrical circuit Electrical/thermal interaction Battery1 Heat Battery2 Heat Battery3 Heat LTI Temperature1 Temperature2 Temperature ANSYS, Inc. All rights reserved. 23 ANSYS, Inc. Proprietary

24 Example: A Battery Module Coupled Analysis Voc=-1.031*exp(-35*(abs(IBatt.V/Vinit))) *(abs(IBatt.V/Vinit)) *(abs(IBatt.V/Vinit))^ *(abs(IBatt.V/Vinit))^3+0.3/30.0*(U1.Temp_block_1-273) Ccapacity IBatt VOC Rseries CT_S RT_S CT_L RT_L H00 CONST SIMPARAM1 C1 C4 0 0 I7 I8 E1 E2 R1 R5 R2 C2 R6 C5 R3 C3 R7 C6 RLoad H01 CONST H02 CONST Qcell1 Qcell2 Qcell3 Qcell4 Qcell5 Qcell6 Tambien Temp_block_1 Temp_block_2 Temp_block_3 Temp_block_4 Temp_block_5 Temp_block_6 C7 0 I9 E3 R9 C8 R10 C9 R11 H03 CONST C10 0 I10 E4 R13 R14 C11 R15 C12 H04 CONST Y1 [kel] TR Curve Info U1.Temp_block_1 C13 0 I11 E5 R17 R18 C14 R19 C15 H05 CONST Time [s] TR TR U1.Temp_block_3 U1.Temp_block_ ANSYS, Inc. All rights reserved. 24 ANSYS, Inc. Proprietary

25 System Level Circuit Model Li-ion Battery 60 Cells connected in matrix pack Packs are connected in matrix to final configuration 5 cells Peak voltage: 16 V (4 cells in series) Peak current: ~3.25 Amp (15 cells in parallel) 0.4 Amp for single battery case And yet runtime is ~doubled Estimated life: 0.4/(3.25/15)x8000 sec without rate factor consideration 2010 ANSYS, Inc. All rights reserved. 25 ANSYS, Inc. Proprietary

26 Battery in Control System with Motor Controller Solid-state driver chips Solid-state Controller Perm. Mag. Motor Battery Alternator/inverter Motor Performance Command Multi-disc clutch Torque limiter Linear Coupling Drive shafts Battery Performance 2010 ANSYS, Inc. All rights reserved. 26 ANSYS, Inc. Proprietary

27 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 27 ANSYS, Inc. Proprietary

28 Single Battery Cell Thermal Model The model is based on the work of: - Newman & Tidemann (1993); - Gu (1983) ; - Kim et al (2008)* Current i p = Current Vectors at Cathode plate Current J = Current Density J (t, x, y, T ) i n = Current Vectors at Anode plate J ( σ φ) = J = Y( φ φ U ) f ( T ) p n Transfer current U and Y are derived from experimentally obtained polarization curve, dependent on Depth of Discharge (DOD) & Temperature Cathode Anode * Reference: U. S. Kim, C. B. Shin, C. S. Kim, Effect of electrode configuration on the thermal behavior of a lithium-polymer battery, Journal of Power Sources 180 (2008) ANSYS, Inc. All rights reserved. 28 ANSYS, Inc. Proprietary

29 Results of a Prismatic Lithium-Ion Cell Geometry & Mesh Temperature Current Density 2010 ANSYS, Inc. All rights reserved. 29 ANSYS, Inc. Proprietary

30 Case Setup Integrated battery cell thermal setup panels 2010 ANSYS, Inc. All rights reserved. 30 ANSYS, Inc. Proprietary

31 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 31 ANSYS, Inc. Proprietary

32 Newman s 1d Electrochemistry Model in Simplorer Li + Jump Li + Lithium Ion Batteries Li + Li x C 6 e (εece ) t = 1 e e t F Li ( D c ) + j + Li + e Li x -Metal Metal-oxide Electrochemical Kinetics Solid-State Li Transport Electrolytic Li Transport Charge Conservation/Transport (Thermal) Energy Conservation Results from Simplorer Results from Newman 2010 ANSYS, Inc. All rights reserved. 32 ANSYS, Inc. Proprietary

33 Model Implementation Ref: Newman et al (1993, 1995, 1996) The model is entirely based Neman s 1d electrochemistry model (1993, 1995, 1996) x=0 x=l δ n δ s δ p 1 c s1,1 1 c s2,1 1 c s3,1 1 c s4,1 1 c s5, c s1,2 c s2,2 c s3,2 c s4,2 c s5, c s1,3 c s2,3 c s3,3 c s4,3 c s5,3 Negative Electrode Separator Positive Electrode Newman assumed constant diffusivity inside particles. In the current model, such an assumption is not used and the particle diffusion equations are solved numerically and thus allow for non-constant diffusivity. The makes the model really 2d rather than 1d. This model is also called pseudo-2d in literature c 1 c 2 c 3 c 4 c 5 φ 1,1 φ 2,1 i 1,1 i 2,1 x= 0 φ 1,2 φ 2,2 i 1,2, i 2,2 h φ 1,3 φ 2,3 i 1,3 i 2,3 φ 1,4 φ 2,4 i 1,4 i 2,4 φ 1,5 φ 2,5 i 1,5 i 2,5 φ 1,6 φ 2,6 i 1,6 i 2,6 x=δ n Reference: M. Doyle, T.F. Fuller, and J. Newman, Journal of Electrochem. Soc., 140, 1526 (1993) C. R. Pals and J. Newman Journal of Electrochem. Soc., 142 (10), (1995) 2010 ANSYS, M. Inc. Doyle, All rights J. Newman, reserved. Journal of Electochem. Soc., 143, 1890 (1996) 33 ANSYS, Inc. Proprietary

34 Sample Results in Simplorer Discharge Curve Particle Concentration It takes less than two days for an engineer to implement the model in Simplorer compared to months using inhouse methods Run time is a couple of minutes for a complete discharge curve of 100,000 seconds Electrolyte Concentration 2010 ANSYS, Inc. All rights reserved. 34 ANSYS, Inc. Proprietary

35 Newman 1d Model Single Insertion Simplorer s Results Newman s Results Reference: M. Doyle, T.F. Fuller, and J. Newman, Journal of Electrochem. Soc., 140, 1526 (1993) 2010 ANSYS, Inc. All rights reserved. 35 ANSYS, Inc. Proprietary

36 Newman 1d Model Dual Insertion Simplorer s Results Newman s Results Reference: M. Doyle, J. Newman, Journal of Electochem. Soc., 143, 1890 (1996) 2010 ANSYS, Inc. All rights reserved. 36 ANSYS, Inc. Proprietary

37 Newman 1d Model Quantitative Comparison x=l p +L s +L n x=l p +L s x=l p x=0 1/10 C 1/2 C 1 C 2 C 4 C 6 C 8 C 10 C Simplorer s Results White s Results Reference: Long Cai, Ralph E. White, Journal of Electrochem. Soc., 156 (3), A154-A161 (2009) 2010 ANSYS, Inc. All rights reserved. 37 ANSYS, Inc. Proprietary

38 Newman 1d Model - Quantitative Comparison Simplorer s Results White s Results Reference: Long Cai, Ralph E. White, Journal of Electrochem. Soc., 156 (3), A154-A161 (2009) 2010 ANSYS, Inc. All rights reserved. 38 ANSYS, Inc. Proprietary

39 Newman 1d Model - Isothermal Concentration profiles due to different temperatures Discharge curves due to different temperatures Reference: C. R. Pals and J. Newman Journal of Electrochem. Soc., 142 (10), (1995) 2010 ANSYS, Inc. All rights reserved. 39 ANSYS, Inc. Proprietary

40 Energy Equation Energy Equation + Heat Source Energy Equation(Default) (ρ c T) p = λ T q t + Heat Source (Source Term/reaction heat + joule heating) q = a sj i σ nj φs φe U φ φ j + T U T j ( eff eff κ φ φ + κ c φ ) eff + s s+ e e D ln e e Reference: W. B. Gu, C. Y. Wang, Journal of Electrochem. Soc., 147 (8), (2000) 2010 ANSYS, Inc. All rights reserved. 40 ANSYS, Inc. Proprietary

41 Newman 1d Model - Energy Reference: 2010 ANSYS, Inc. W. All rights B. reserved. Gu, C. Y. Wang, Journal of Electrochem. 41 Soc., 147 (8), (2000) ANSYS, Inc. Proprietary

42 3D Electrochemistry Modeling Li concentration in electrodes during discharge 2010 ANSYS, Inc. All rights reserved. 42 ANSYS, Inc. Proprietary

43 1D with FLUENT - Discharge 20[A] Li + concentration [mol/m 3 ] t=0[s] t=1000[s] Li concentration [mol/m 3 ] t=2000[s] t=4000[s] 2010 ANSYS, Inc. All rights reserved. 43 ANSYS, Inc. Proprietary

44 1D with FLUENT - Discharge 20[A] 4500 Concentration of Li Li + [mol/m 3 ] t=0 t=1000 t=2000 t=3000 t= x [mm] 2010 ANSYS, Inc. All rights reserved. 44 ANSYS, Inc. Proprietary

45 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 45 ANSYS, Inc. Proprietary

46 Introduction CFD as a battery thermal management tool needs heat generation as input and can produce temperature as output Battery electrical circuit model and electrochemistry model can produce heat generation but they themselves need temperature as input Coupling of electrical circuit model or electrochemistry model with CFD is the solution 2010 ANSYS, Inc. All rights reserved. 46 ANSYS, Inc. Proprietary

47 Simplorer FLUENT Co-Simulation Heat Dissipated Cell 4 Cell 5 Cell 6 Cell 3 Cell 2 Cell 1 Temperature Simplorer Battery Circuit Model FLUENT Battery CFD Model 2010 ANSYS, Inc. All rights reserved. 47 ANSYS, Inc. Proprietary

48 Simplorer FLUENT Co-Simulation Heat Dissipated Heat dissipation Temperature Temperature Discharge curve Simplorer Battery Circuit Model FLUENT Battery CFD Model 2010 ANSYS, Inc. All rights reserved. 48 ANSYS, Inc. Proprietary

49 Quasi-Steady CFD vs Transient CFD The previous CFD model uses quasi-steady assumption because thermal time scale is much smaller than electrical time scale Transient Simplorer coupled with steady state CFD Transient Simplorer coupled with transient CFD is also possible 2010 ANSYS, Inc. All rights reserved. 49 ANSYS, Inc. Proprietary

50 Comparison of Two Coupling Strategies Heat Dissipated Cell 5 Cell 6 Cell 4 Cell 3 Cell 2 Cell 1 Temperature Rseries RT_S RT_L H00 SIMPARAM1 Ccapacity C1 C4 C7 C IBatt I7 I8 I9 I10 VOC E1 E2 E3 E4 R1 R5 R9 R13 CT_S R2 C2 R6 C5 R10 C8 R14 C11 CT_L R3 C3 R7 C6 R11 C9 R15 C12 RLoad CONST H01 CONST H02 CONST H03 CONST H04 CONST Y1 [kel] Qcell1 Qcell2 Qcell3 Qcell4 Qcell5 Qcell6 Tambien Temp_block_1 Temp_block_2 Temp_block_3 Temp_block_4 Temp_block_5 Temp_block_6 TR Curve Info U1.Temp_block_1 The battery circuit model is the same for both approaches. The difference is on the modeling of the thermal side. C13 0 I11 E5 R17 R18 C14 R19 C15 H05 CONST Time [s] TR TR U1.Temp_block_3 U1.Temp_block_ ANSYS, Inc. All rights reserved. 50 ANSYS, Inc. Proprietary

51 ROM (LTI) vs Co-Simulation ROM Using LTI Faster Easier to use Recommendation: If temperature field is not needed, use LTI approach. If non-linearity is strong, use co-simulation. If flow rate changes constantly, use co-simulation Co-Simulation Can provide more detailed temperature information temperature field, max temperature, temperature gradient, etc. Can have non-linearity and non-constant flow rate 2010 ANSYS, Inc. All rights reserved. 51 ANSYS, Inc. Proprietary

52 Simplorer Electrochemistry Model and FLUENT CFD Model Co-Simulation x=0 x=l δ n δ s δ p Heat Dissipated Negative Electrode Separator Positive Electrode Temperature Simplorer Battery Electrochemistry Model FLUENT Battery CFD Model 2010 ANSYS, Inc. All rights reserved. 52 ANSYS, Inc. Proprietary

53 Simplorer Electrochemistry Model and FLUENT CFD Model Co-Simulation Heat Dissipated Temperature An Array of Simplorer Battery Electrochemistry Models FLUENT Battery CFD Model 2010 ANSYS, Inc. All rights reserved. 53 ANSYS, Inc. Proprietary

54 Outline Battery thermal management using CFD Battery system thermal management using Foster network approach Battery electric circuit model Battery single cell thermal model Battery electrochemistry Battery thermal management using cosimulation Battery thermal management with bus bar heating using CFD 2010 ANSYS, Inc. All rights reserved. 54 ANSYS, Inc. Proprietary

55 Bus Bar Heating for Battery For steady (or low frequency) currents under temperature dependent conductivity User σ= Defined ( Φ) f( T) σ = 0 Ohmic Loss= Scalar : r J 2 σ Loss Temperature Fluent Energy Solver User Defined Scalar for potential Temperature dependent conductivity Ohmic loss for energy solver 2010 ANSYS, Inc. All rights reserved. 55 ANSYS, Inc. Proprietary

56 Validation Maxwell vs Fluent Maxwell Results : Φ max = V Φ = V min Fluent Results : Φ = V Φ max min = V Property : σ copper = sm 7 1 Boundary Conditions : 7 current boundaries 2 voltage boundaries Maxwell FE Results and Mesh Fluent FV Results and Mesh Constant conductivity used for comparison 2010 ANSYS, Inc. All rights reserved. 56 ANSYS, Inc. Proprietary

57 Coupled Simulation in Fluent Temperature Distribution Current Density Magnitude Distribution Fluent Results : Φ max due changes from due to temperature impact V to V Property : σ copper = 1 α = ( [ 1.0+ α( T T )]) ref sm 1 Conductivity Distribution 2010 ANSYS, Inc. All rights reserved. 57 ANSYS, Inc. Proprietary

58 What if Only One-Way Coupling One way coupling means that Ohmic loss from constant conductivity is mapped to thermal solver without update of temperature from thermal solver. Temperature Distribution with Two-Way Coupling With only one-way coupling, the max temperature increase is 25K compared with that of 29K using two-way coupling, a difference of 15%. Temperature Distribution with Only One-Way Coupling Two-way coupling is necessary 2010 ANSYS, Inc. All rights reserved. 58 ANSYS, Inc. Proprietary

59 Validation and Coupled Simulation in CFX Same coupled analysis can be done in CFX with the same results ANSYS, Inc. All rights reserved. 59 ANSYS, Inc. Proprietary

60 Conclusion ANSYS is uniquely ready with full range engineering simulation solutions for the entire range of battery applications - from detailed electrochemistry to system level thermal management ANSYS technology provides the most comprehensive state-of-the-art battery solutions in the industry 2010 ANSYS, Inc. All rights reserved. 60 ANSYS, Inc. Proprietary

61 Thank You 2010 ANSYS, Inc. All rights reserved. 61 ANSYS, Inc. Proprietary