Detecting, diagnosing and controlling degradation in lithium ion battery packs

Transcription

1 Detecting, diagnosing and controlling degradation in lithium ion battery packs Yu Merla, Ian Hunt, Yan Zhao, Marie-Therese v. Srbik, Yatish Patel, Monica Marinescu, Vladimir Yufit, Billy Wu, Ricardo Martinez-Botas, Nigel Brandon, Gregory J Offer. Senior Lecturer, Mechanical Engineering, Imperial College London Gregory.offer@imperial.ac.uk

2 What is the problem?

3 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

4 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

5 How does pack design affect degradation? Temperature affects impedance exponentially non-linear temperature dependence on charge transfer resistance Two common types thermal mgmt. Surface cooling, GM Volt (left) Tab cooling, AVL (right) Thermal gradients are inevitable -Z imag [m ] C -5 C 5 C 15 C 25 C 35 C 45 C 55 C 10 Hz Hz Decreasing temperature 0.1 Hz Z real [m ] Troxler et al. Journal of Power Sources, 2014, Vol 247, Pages

6 Would you keep cells at different temperatures? Parallel cells have unequal currents External resistances Internal impedance» Mfg. differences» Different temperature» Different SOC» Different SOH Wu et al. Journal of Power Sources, 2013, Vol 243 Pages

7 Aging experiment under different thermal management Tab Cooled Two test rigs 1000 cycles 6C discharge and 2C charge Surface Cooled Hunt et al. Journal of the Electrochemical Society, Vol 163, Pages A1846-A1852

8 Aging experiment under different thermal management Tab Cooled Two test rigs 1000 cycles 6C discharge and 2C charge Tab cooling is better Useable and lifetime capacity Tab Cooled Surface Cooled Surface Cooled Hunt et al. Journal of the Electrochemical Society, Vol 163, Pages A1846-A1852

9 There are bad and less bad thermal gradients Tab cooling Different impedance within layer Each layer behaves same Minimal feedback Surface cooling Different impedance between layers Layers behave differently Positive feedback Hunt et al. Journal of the Electrochemical Society, Vol 163, Pages A1846-A1852

10 Conclusion (so far ) Inhomogeneity is bad Thermal gradients are very common Parallel cells are common Significant root cause of accelerated degradation Any other inhomogeneity could do the same Solutions Good pack design, surface cooling is risky Diagnostic techniques Models of degradation Unequal Degradation Models Thermal Gradients Unequal Current Pack Design Unequal Impedance Diagnostics

11 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

12 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

13 Differential Thermal Voltammetry (DTV) Novel in-situ battery diagnosis method for tracking degradation Cheap and easy Needs surface temperature and voltage readings dt/dv plotted against cell voltage Evolution of peaks acts as footprint of cell SOH Complimentary to ICA/SRCV dv dt dt dv dt dt B. Wu et al. Journal of Power Sources, 2015, Volume 273, Pages

14 Experiments 2 cells aged differently in incubator Graphite NMC 5Ah commercial cell Red cell = 1C cycling Promote SEI layer growth Black cell = 4.2V hold SEI layer growth followed by cathode cycle 300: same capacity loss => Looks the same but one cell will behave very differently in the future! Y. Merla et al. Journal of Power Sources, 2016, Volume 307, Pages

15 What do the peaks represent? Peak fitting similar technique to Dubarry et al. for ICA Peak = region of electrode stability; Peak to peak = phase change NMC 2 hexagonal phases Graphite 5 stages of lithium intercalation (3 observable) Stoichiometric drift -> change in peak parameters (position, width) Y. Merla et al. Journal of Power Sources, 2016, Volume 307, Pages

16 Using information for diagnosis Cycle 300 Cycle 300 Peak parameters can be used to monitor SOH Stoichiometric drift shown by shift in peak width 1*❷ and cycle 300 (8% cap. loss) DTV can now tell you that they are not the same! Aged in different ways = one cell is more likely to fail early Y. Merla et al. Journal of Power Sources, 2016, Volume 307, Pages

17 DTV in pack application 4 cells in parallel, with 1 artificially aged, 0.5-2C Whilst being cooled, and for charging Still works! Y. Merla et al. Journal of Power Sources, Vol 331, 2016, Pages

18 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

19 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

20 ECN electrochemical model A model is needed to interpret the results and make an informed decision Ultimately to embed in a microcontroller inside a BMS Our model Variable double layer capacitance All parameters function of SOC, I, T SEI layer growth degradation included M-T von Srbik et al. Journal of Power Sources, 2016, Vol 325, Pages

21 Model capabilities Accurate to 20C Thermally coupled Models capacity fade and SEI growth Able to reproduce diagnostics for cells under degradation M-T von Srbik et al. Journal of Power Sources, 2016, Vol 325, Pages

22 2 Empirical version Built using the Simscape library in MATLAB Simulink o Each circuit element still has physical meaning o Easier parameterisation (2-3 weeks testing for a new cell/chemistry)* o Flexible model setup easy addition of: Further degradation/aging mechanisms, Increasing number of electrode particles/cell layers. Can model inhomogeneities easily, thermal/physical o Fast simulation few seconds to simulate thousands of seconds o Low computational power requirement aimed at on-board application * Degradation can take much longer

23 2 Required preparation data OCV curve > Taken from battery manufacturer or a slow C-rate discharge (e.g. C/25) Half cell curve (min. one electrode) > Taken from literature for the specific cell chemistry (e.g. graphite) EIS under load > To find series, high frequency and mid frequency resistances at various SOCs Pulse loading > To find total over potentials at various SOCs to determine low frequency resistance

24 Systems Engineering Solution BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

25 Ongoing work BMS Hardware (Accuracy, Resolution, Precision) Cell Voltage, Current, Temperature Pack Design (Systems Sensors thermal mgmt series/parallel Cell Design (Cell Layers, tabs, thickness, voltages BMS Software Requirements (Micro-processor Memory) Data Processing Fitting, etc System Controller De-rating Voltage De-rating Current Thermal Mgmt Advanced Reduced Order Model Kalman Filters Model States / Parameter Estimation Battery States (SOC, SOAP, SOH) Control Intervention

26 Thank you

27 Acknowledgements EPSRC & Innovate UK for funding various projects that contributed towards this work Climate KIC, Jaguar Land Rover, & Ricardo Ltd for sponsoring PhD students that did much of this work