The Pennsylvania State University. The Graduate School. Department of Mechanical and Nuclear Engineering

Transcription

1 The Pennsylvania State University The Graduate School Department of Mechanical and Nuclear Engineering IMPROVING LOW TEMPERATURE PERFORMANCE OF LITHIUM ION BATTERIES THROUGH MUTUAL PULSE HEATING A Thesis in Mechanical Engineering by Kamiar Salehi 2014 Kamiar Salehi Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science December 2014

2 ii The thesis of Kamiar Salehi was reviewed and approved* by the following: Chao-Yang Wang William E. Diefender, Chair of Mechanical Engineering Director of the Electrochemical Engine Center Thesis Advisor Chris Rahn Professor of Mechanical Engineering Karen Thole Department Head of Mechanical and Nuclear Engineering Professor of Mechanical Engineering *Signatures are on file in the Graduate School

3 iii ABSTRACT Increase of greenhouse gas emissions over the past few years have initiated the motion towards battery powered vehicles, however only 4% of the total vehicles sold in the U.S. in 2014 were electric. This is because consumers are concerned with electric vehicles (1) Long refueling time (2) High cost and (3) Limited cruising range. This limited range becomes even further reduced when vehicles are operated at subzero temperatures. Studies has shown that due to performance drop of Li-ion batteries at low temperatures, the cruise range of an EV can see a 57% drop. The cause of lithium ion poor performance at low temperatures is attributed to the following three reasons: (1) Low diffusivity of lithium ion within graphite active material (2) Reduced ionic conductivity of the electrolyte and (3) Increase in charge transfer resistances. Experiments using a 13 Ah NCM-Graphite pouch cell also revealed that in comparison with room temperature, at -25 C the discharge capacity, discharge energy, and discharge power is reduced by 33%, 43%, and 40% respectively. It was also observed an exponential increase in DC resistance as temperature drops below 0 C. Researches are being done from the material and chemical stand point to create less temperature sensitive Li-ion cells. In the meantime, the simple solution of heating the cells can be implemented to overcome the poor performance, so EVs can become reliable and comprise a larger percentage of the vehicles on the road. Heating must be done fast and energy efficiently. Mutual Pulse Heating which was first introduced and theoretically modeled by Dr. Ji and Dr. Wang utilizes the increase in internal resistance at low temperatures to warm up the battery from the inside. This method splits the battery pack into two groups and both groups undergo short charge and discharge pulses. Most of discharge power of the discharge group gets stores in the charge group while some of it gets consumed to generate heat. Mutual pulse heating which has theoretically been shown to be effective was experimented in this paper for the first time. The

4 iv results showed that using a constant power of 180W to charge and discharge and pulses of 1 second, the test cell temperature increased from -25 C to 0 C in 139 seconds while consuming 0.59 Ah (4.5%) of battery capacity. The results of this experiment validates the practicality of mutual pulse heating and introduces it as an effective and high potential method for vehicle applications. That is because (1) Uniform temperature distribution owning to internal heating (2) Fast heating time (3) Low capacity consumption and (4) Low cost since no additional components or systems (i.e. heaters, flow channels, etc.) are needed. It was also shown that this method is more effective at mid SOC levels where charge and discharge powers are comparable, and more work must be done to improve its efficiency at low and high SOCs.

5 v TABLE OF CONTENTS List of Figures... vi List of Tables... ix Acknowledgements... x Chapter 1 Introduction Background Previous Studies In This Work Chapter 2 Characterization of Low Temperature Performance of Li-Ion Cells Experimental Setup and Procedures Results Discussion Chapter 3 Li-Ion Cell Performance at Low Temperatures Using Mutual Pulse Heating Previous Work Experimental Setup Results Mutual Pulse Heating: Preliminary Test Mutual Pulse Heating: High Discharge Rate Mutual Pulse Heating: Matched Discharge and Charge Power Discussion Chapter 4 Conclusions and Future Work References Appendix Additional Data Poor Performance of Lithium Ion Cells at Low Temperatures Non-Ohmic Behavior of Lithium Ion at Various C-Rates at Low Temperatures Mutual Pulse Heating Additional Data Heat Generation During Charge and Discharge Ah Test Cell Charge Capacity Mutual Pulse Heating Tests Key Results... 86

6 vi LIST OF FIGURES Figure 1-1: Vehicles Sold in U.S. from 2007 to Figure 1-2: Vehicle Market Share in Figure 1-3: PHEV16 and PHEV65 Electrical Power Train Cost Breakdown Figure 1-4: Battery Pack Cost Breakdown Figure 1-5: Baseline and EV Range Figure 1-6: AAA Range Test at Various Temperatures Figure 1-7: Discharge Capacity at Different Temperatures Figure 1-8: Voltage-Capacity at Different Temperatures for (a) LiPF6 and (b) LiBF Figure 1-9: Relative Capacity and Ionic Conductivity as a Function of Temperature Figure 1-10: EIS Plot of Li-ion cell and the Equivalent Circuit Figure 1-11: Temperature vs. Resistance Components at (a) 3.87V and (b) 3.45V Figure 2-1: 13Ah Test Cell in comparison with Cell and Coin Cell Figure 2-2: Cell Holder. Left: Top View, Right: Front View Figure 2-3: Cell Holder with Terminals and Separator Figure 2-4: Wire Connections, White and Green for Voltage Figure 2-5: Schematic of Experiment System Figure 2-6: Cell Inside Environment Chamber Figure 2-7: Voltage vs. Capacity. 1C Discharge at Various Temperatures Figure 2-8: Temperature vs. Voltage Profile. 1C Discharge at -25 C Figure 2-9: DC Resistance vs. Temperature Figure 2-10: Voltage vs Capacity. C-Rate at 25 C Figure 2-11: DC Resistance vs. C-Rate at 25 C Figure 2-12: Voltage vs Current During Non-Ohmic Behavior at 25 C Figure 2-13: Voltage vs. Capacity. C-Rate at -25 C... 26

7 vii Figure 2-14: Voltage vs. Capacity Zoomed. C-Rate at -25 C Figure 2-15: Voltage and Temperature Profile vs. Capacity Figure 2-16: DC Resistance vs. C-Rate at -25 C Figure 2-17: HPPC at Room Temperature Figure 2-18: HPPC at -25 C Figure 2-19: Discharge Power vs. DOD Figure 2-20: Charge Power vs. DOD Figure 3-1: Voltage and Temperature evolution during Mutual Pulse Heating Figure 3-2: Voltage Profile of Cell 1 During the first 20 Seconds Figure 3-3: Voltage Profile of Cell 2 During the first 20 Seconds Figure 3-4: Lithium Ion Concentration on Graphite Particle Surface Figure 3-5: Voltage and Temperature. 80% 5C Discharge 4.3V Charge Pulse Heating Figure 3-6: Charge and Discharge Power. 80% 5C Discharge 4.3V Charge Pulse Heating Figure 3-7: Voltage and Temperature. 50% 5C Discharge 2.1C Charge Pulse Heating Figure 3-8: Charge and Discharge Power. 50% 5C Discharge 2.1C Charge Pulse Heating Figure 3-9: Voltage and Current Profile. 50% 5C Discharge 2.1C Charge Pulse Heating Figure 3-10: Voltage and Temperature. 80% 7C Discharge 4.3V Charge Pulse Heating Figure 3-11: Charge & Discharge Power. 80% 7C Discharge 4.3V Charge Pulse Heating Figure 3-12: Voltage and Temperature. 50% 7C Discharge 4.3V Charge Pulse Heating Figure 3-13: Charge & Discharge Power. 50% 7C Discharge 4.3V Charge Pulse Heating Figure 3-14: Voltage and Current Profile. 50% 7C Discharge 4.3V Charge Pulse Heating Figure 3-15: Voltage and Temperature. 80% CP 120W Pulse Heating Figure 3-16: Charge and Discharge Power. 80% CP 120W Pulse Heating Figure 3-17: Voltage and Temperature. 50% CP 180W Pulse Heating Figure 3-18: Charge and Discharge Power. 80% CP 120W Pulse Heating... 57

8 viii Figure 3-19: Voltage and Current Profile. 50% CP 180W Pulse Heating Figure 3-20: Voltage vs. Capacity. 1C Discharge at -25 C Using Various Protocols Figure A-1: Cycling Performance of Half-Cell at Various Temperatures Figure A-2: Discharge capacity vs Cell Voltage Figure A-3: Ionic Conductivity of LiPF6 and LiBF4 at Various Temperatures Figure A-4: Temperature vs. Relative Resistances (a) R b (b) R sei (c) R 24 ct Figure A-5: Voltage vs. Resistance of Impedance Segments Figure A-6: Voltage vs Current During Non-Ohmic Behavior at Figure A-7: Comparison of Heating Strategies Figure A-8: Voltage and Current. 80% SOC 5C Discharge 4.3V Charge Pulse Heating Figure A-9: Voltage and Current. 80% SOC 7C Discharge 4.3V Charge Pulse Heating Figure A-10: Voltage and Current. 80% SOC CP 120W Pulse Heating Figure A-11: Voltage and Temperature. 80% 3C Discharge 4.3V Charge Pulse Heating Figure A-12: Charge & Discharge Power. 80% 3C Discharge 4.3V Charge Pulse Heating Figure A-13: Voltage and Current. 80% SOC 3C Discharge 4.3V Charge Pulse Heating Figure A-14: Voltage and Temperature. 50% 4C Discharge 4.3V Charge Pulse Heating Figure A-15: Charge & Discharge Power. 50% 4C Discharge 4.3V Charge Pulse Heating Figure A-16: Voltage and Current. 80% SOC 3C Discharge 4.3V Charge Pulse Heating Figure A-17: Heat Generation Rate vs SOC at 25 C (a) 1C Charge (b) 1C Discharge

9 ix LIST OF TABLES Table 1-1: Comparison of Battery Systems Table 2-1: Test Cell Specifications Table 2-2: Discharge Energy at 1C at Different Temperatures Table A-1: Test Cell Nominal Capacity Table A-2: Mutual Pulse Heating Tests Results... 86

10 x ACKNOWLEDGEMENTS I would like to first extend my thank you to Dr. Chao-Yang Wang for his support, guidance, and encouragements. I was able to be involved with some extraordinary projects during my two years, and I will always cherish the given opportunities. Also I would like to express my appreciation to Dr. Guangsheng Zhang for his continuous help not only with this thesis, but also with all the other projects that we worked on. I learned a lot from him and I m sure the experiences gained will be to my advantage in both my personal and professional life. I am grateful to have been able to collaborate with the very talented Dr. Yan Ji. He has guided me greatly throughout this thesis, and it was a pleasure to work with him on further advancement of his novel ideas. Next I am thankful for my lab mates, Poowanart Poramapojana, Dr. Lei Cao, and Man Su Park who helped and supported me during the last two years that we worked together. Finally, I would like extend my deepest gratitude to my loving family. My sister was always there for me when I needed her help, and I would have never been able to accomplish any of my achievements had it not been for the never ending support and sacrifices of my parents. This thesis and my degree is surely dedicated to them.

11 1 Chapter 1 Introduction 1.1 Background The increase in gas prices due to shortage of fossil fuels, and increase in greenhouse gas emissions have initiated the search for alternative source of energy for vehicle applications 1, 2. Even though vehicle emissions have reduced significantly over the past 50 years, further reduction is becoming an impossible task with internal combustion engines 3. To solve these economic and environmental problems, the vehicle industry has started development of hybrid electric vehicles (HEV), extended range or plug-in hybrid vehicles (PHEV), and electric vehicles (EV). Among the choices of electrochemical energy storage for these vehicles, lithium ion battery is the most promising one available due to its high power and energy density. Table 1-1 shows the comparison of a few battery chemistries. Table 1-1: Comparison of Battery Systems 4 Battery Practical Specific Energy (Wh/kg) Practical Energy Density (Wh/L) Alkaline Lead-Acid Ni-Cd Ni-MH Li-Ion

12 Numbers Sold in Million 2 Although EVs and HEVs offer high energy efficiency and low environmental impact 5, they have yet to comprise a significant portion of the total vehicle sales. As can be seen in Figure 1-1 and Figure 1-2, the total number of battery powered vehicles sold in 2014 in U.S. is less than 3.6% of the total sale 6. Even though this number has increased over the past years, it indicates that electric vehicles have not yet been widely recognized as a suitable alternative to conventional vehicles. Some of the concerns that consumers have when it comes to electric vehicles are long refueling time, high cost, and limited range Year * Conventional Vehicles HEV PHEV BEV *As of September 2014 Figure 1-1: Vehicles Sold in U.S. from 2007 to % 0.35% 0.35% 2.88% Conventional Vehicles HEV PHEV BEV Figure 1-2: Vehicle Market Share in

13 3 Due to the fast pace life style of today, people prefer petroleum cars which can be refueled within 10 minutes compared to several hours charge time of electric vehicles 8. Some solutions to this issue has been implemented, including Tesla Motors supercharger which can charge the vehicles to 80% state of charge (SOC) in 40 minutes, and Tesla Motors battery swap which will replace the battery pack with a fully charged one in less than 90 seconds 9, 10. The next concern mentioned is the high cost of electric vehicles. Electric vehicles and conventional vehicles are similar in all aspects except for the power train. As so, the reason behind their cost dissimilarity is due to the cost of their power trains. International Renewable Energy Association (IREA) has studied the cost-breakdown of PHEV16 and PHEV65 power trains shown in Figure It can be seen that the lithium ion battery pack is the most expensive component in electric vehicles power train. Inside the battery pack, lithium ion cells are the reason behind the high cost. Figure shows that the lithium ion batteries contribute to 50% of the battery pack cost, and 13% of the cost goes towards warranty and margin which are also battery related. Hence reduction in the price of lithium ion battery will sure reduce the overall cost of electric vehicles. Currently the price of lithium ion battery is around $500/kWh, but solutions such as Tesla Motor s Gigafactory is aiming to reduce this cost by 30% 13, 14. This will reduce the price to $350/kWh which is even lower than the Department of Energy target of $405/kWh by

14 4 Figure 1-3: PHEV16 and PHEV65 Electrical Power Train Cost Breakdown 11 Figure 1-4: Battery Pack Cost Breakdown 12

15 5 The third major concern of consumer with electric vehicles is their limited cruise range. Figure 1-5 shows the EPA driving range of different EVs on one charge, and it compares them with the similar conventional or PHEV models 16. As shown, EVs driving range is 20% to 25% of the baseline models which is the reason behind range anxiety among electric vehicle drivers. TESLA MODEL S 265 Baseline EV NISSAN LEAF* % 378 HONDA FIT % 371 TOYOTA RAV % 398 FORD FOCUS % 397 CHEVROLET SPARK % *Compared with Nissan Versa Note Range (Miles) Figure 1-5: Baseline and EV Range 16 Range anxiety becomes even more significant as electric vehicles lose their reliability when operated at low temperatures. A recent study done by AAA Automotive Research Center has shown that EVs experience 57% range decrease when driven at 20 F (-6.7 C) in comparison with driving at 75 F (23.4 C) 17. According to AAA, all vehicles tested performed great at any temperature. It simply took extra energy from the lithium ion battery to counter the cold or hot extremes and therefore reduced driving range. A small portion of range reduction can be linked to the use of climate control when driving at low temperatures, but the main reason behind the limited range comes from the poor lithium ion performance. Throughout the rest of this thesis, effects of low temperature on performance of lithium ion batteries will be studied, and a practical solution to improve the performance will be presented.

16 6 AVERAGE F 95 F 75 F CAR CAR CAR Range (Miles) Figure 1-6: AAA Range Test at Various Temperatures Previous Studies The low temperature performance of Li-ion battery has been the subject of many studies in the past. This can be attributed to the desire of using li-ion batteries at cold temperatures for vehicle applications, for military use, in aerospace projects as well as space programs Before finding solutions to the poor performance of lithium ion batteries at low temperatures, it is important to understand the cause of such behavior. While the origin is still under debate, it can be summarized that the low performance is due to: 1. Reduced diffusivity of lithium ion within graphite 2. Reduced ionic conductivity of the electrolyte 3. Increased charge transfer resistance

17 7 Zhang et al. (2002) 23 mentioned that one of the causes for poor low temperature performance is limited diffusivity of lithium ion within graphite when temperature falls below 0 C. This conclusion was followed after analyzing the results obtained from testing Li/Graphite half-cell at different ambient temperatures. It was observed that the half-cell capacity reduced as temperature decreased. It was also noted that at -20 C, graphite could retain 94% of its delithiation capacity when lithiated at room temperature, but it could only retain 12% when lithiated at -20 C. Both findings agreed with reduced lithium ion diffusivity in graphite at low temperature. Next cause of poor performance at low temperatures was reduced ionic conductivity of the electrolyte. Smart et al. (2002) 24 conducted experiments to test the performance of threeelectrode, O-ring-sealed, MCMB-Li x Ni y Co 1-y O2 cells using different LiFP 6 -based electrolytes at -40 C. The best performing electrolyte (1.0 M LiFP 6 EC+DEC+DMC+EMC (1:1:1:3)) was then chosen to be tested at room temperature as well as -20 C, -30 C, and -40 C. The result, as shown in Figure 1-7, demonstrated that even with such electrolyte that has good ionic conductivity, only 59% of the room temperature discharge capacity was delivered at -40 C. Smart has also mentioned that the improved low temperature performance obtained by using the said electrolyte has come at the cost of reduced performance at room temperature due to poor lithium kinetics.

18 8 Figure 1-7: Discharge Capacity at Different Temperatures 24 Zhang et al. (2002) 25 studied the use of LiBF 4 as an alternative electrolyte chemistry in order to increase the low temperature performance. LiBF 4 has lower ionic conductivity than LiPF 6, but it has shown good performance between 0 and -20 C. However it suffers a significant drop in performance below -20 C as the electrolyte froze. To increase its operating temperature down to -40 C, LiBF 4 was dissolved in 1:1:3 PC/EC/EMC. Cells with LiBF 4 and LiPF 6 electrolytes were then tested at different ambient temperatures. Two important conclusions were made from the results shown in Figure 1-8 (a) and (b). First, as the temperature decreased so did the operating voltages. This was due to high polarization that was caused by increase in the overall cell electric resistance, which includes electrolyte, electrodes, and SEI. Second conclusion was that ionic conductivity of the electrolyte is not the dominant limitation of low temperature performance; as long as the solvent does not freeze or salt precipitation does not occur. This was concluded after seeing that in comparison to 20 C, LiBF 4 retained 86% of its capacity at -30 C and near 20% at -50C, while LiPF 6 with higher ionic conductivity retained only 72% at -30 C and completely failed at -50 C.

19 9 Figure 1-8: Voltage-Capacity at Different Temperatures for (a) LiPF6 and (b) LiBF4 25 To complement his previous work, Zhang et al. (2003) 26 investigated the impact of charge transfer resistance on cell performance. The curiosity was arise after plotting cell capacity and ionic conductivity on the same plot as a function of temperature shown in Figure 1-9. It was noticed that although ionic conductivity had a smooth decline as temperature decreased (white dots), the relative capacity (black dots) underwent a rapid decline below -10 C. It was then concluded that there must be a more dominant limitation to low temperature performance. Figure 1-9: Relative Capacity and Ionic Conductivity as a Function of Temperature 26

20 10 Using electrochemical impedance spectroscopy (EIS), cell resistance of li-ion was obtained. The plot was consisted of two semi circles at high and medium frequencies and a straight inclined tail at low frequencies. A simple equivalent circuit was used to explain what each section represented, as shown in Figure R b is bulk resistance of the cell which includes electric conductivity of the electrodes, separator, and electrolyte. R sei correspond to the resistance of the SEI layer, and R ct is the faradic charge-transfer resistance. Figure 1-10: EIS Plot of Li-ion cell and the Equivalent Circuit 27 It was clear from the EIS plot that charge transfer resistance was the dominating segment of the total cell impedance, so experiments were conducted to study its effect at low temperatures. In order to measure the total impedance of the lithium ion cell and the contribution of each electrode, an asymmetric lithium ion cell as well as symmetric cells with graphite/graphite and cathode/cathode were produced. Separate plots for each impedance components (R b, R sei, and R ct ) were used to compare the relative resistance of li-ion cell, symmetric cells, and electrolyte as a function of temperature. The results showed that the increase of R b and R sei is consistent with that of electrolyte as temperature decreases, and they all tend to increase faster around -40 C. On the other hand R ct showed a rapid increase around -10 C, which explains the drop in capacity that

21 11 was seen in Figure 1-9. Zhang et al. (2004) 27 then supplemented his work by running a similar test using li-ion cells. Figure 1-11 shows the result of his experiment at voltages of (a) 3.87 V and (b) 3.45 V. These two voltages were chosen due to the effect of cell voltage on the EIS. It was shown that in both cases, charge transfer resistance accounts for almost 100% of the total cell resistance as temperature drops below -20 C. Figure 1-11: Temperature vs. Resistance Components at (a) 3.87V and (b) 3.45V 27 There have been many solutions suggested to the above low temperature problems such as formulating new solvent mixtures 27, replacing the existing LiPF 6 salt 27, or reducing graphite particle size 1. However no method has been able to fully fix the issues and revive the performance to its above 0 C state. Given the need to use lithium ion batteries, perhaps the easiest and quickest solution at this time is to simply heat up the cells! 1.3 In This Work As stated above, there are many phenomena that contribute to the low temperature performance of lithium ion batteries, and the debate is still ongoing even though much research

22 12 has been done on the topic. Under such circumstances, a short term solution is to simply heat up the cells to recover the performance. Of course if battery is the only source of energy, heating it comes at the cost of losing capacity. That is why an efficient method is needed to warm the cells by using the minimum amount of energy needed. A method called Mutual Pulse Heating 5 is considered to be a promising solution. This technique utilizes the increased cell impedance at low temperatures to internally generate heat as it charge and discharge the cell in short pulses. More details of this method will be explained in Chapter 3. This technique however has only been studied theoretically so far and all the data are based on modeling. The goal of this paper is to experimentally verify the effectiveness of this heating strategy. In this work, a novel procedure is developed to physically experiment mutual pulse heating for the first time. Prior to that in Chapter 2, the low temperature problem of li-ion cells will be validated by testing a large-format 13 Ah prismatic pouch cell at different ambient temperatures. The resulting data will not only show the decrease in performance of li-ion cells at low temperatures, but it also provide a set of baseline data which will later be used in Chapter 2 to compare the cell performance with and without Mutual Pulse Heating.

23 13 Chapter 2 Characterization of Low Temperature Performance of Li-Ion Cells 2.1 Experimental Setup and Procedures In order to better understand the behavior of lithium ion cells at low temperatures, and to provide baseline data for verification of Mutual Pulse Heating Effects, the following tests were conducted: 1. 1C Discharge at various ambient temperatures 2. C-Rates discharge at room temperature and -25 C 3. Hybrid Pulse Power Characterization (HPPC) test at room temperature and -25 C All the above tests were completed using a 13 Ah prismatic pouch cells with Nickle Cobalt Manganese (NCM) as cathode and graphite as anode. It is important to note that unlike most researches that use coin cells for their experiments, full size cells that can be used in vehicle applications were tested during the experiments of this paper. Figure 2-1 shows the cell used for testing as well as an and a coin cell for comparison. The cell was designed by EC Power 28 and manufactured at its facility in China. The chemical reaction of NCM cell is shown below in equation 2-1 4, and some of the key parameters of the cell are outlined in Table 2-1.

24 14 (2-1) Table 2-1: Test Cell Specifications Parameter Value Dimensions Width x Length x Thickness (mm) 140x70x12 Volume (L) Weight (Kg) Capacity (Ah) 13 Specific Energy (Wh/Kg) Energy Density (Wh/L) High Frequency Resistance (mω) 1.97 Heat Capacity (J / kg K) 1000 Figure 2-1: 13Ah Test Cell in comparison with Cell and Coin Cell

25 15 Given the high capacity of the cell, it was important to ensure that all the tests were conducted in a safe manner. To make sure no accidental shorting occurs during setup and test, a cell holder was designed from non-conductive polycarbonate (PC) plate to secure the cell. Two 0.25 inch copper bars were then mounted on the PC as positive and negative terminals. The copper bars offered larger area for the cell terminals and test wires to be mounted on. Good and low resistance contact between the copper bars and the cell tabs was ensured by the use of three ¼ -20 screws. Using lathe machine, a small groove was carved on the PC in between where the cell terminals would be. The groove was for a removable PC plate to prevent any accidental contact between the positive and negative terminals during assembly. Lastly two holes were punched on each of the cell tabs, allowing them to be fastened to the copper bars. Cell holder, terminals, and the separator can be seen in Figure 2-2 and Figure 2-3. Figure 2-2: Cell Holder. Left: Top View, Right: Front View

26 16 Figure 2-3: Cell Holder with Terminals and Separator Battery tester machine, Arbin BT2000, was utilized to cycle the cell as well as to record temperature data. Each channel of this battery tester could handle maximum of 5V and 20A. During some of the tests, the current was scheduled to go as high as 7C, thus 8 channels were combined in parallel to provide the needed current rating. To prevent error during data acquisition, the voltage wires of the battery tester were connected directly to the cell terminals on the hole closest to the cell as seen in Figure 2-4. This was to minimize the voltage drop caused by current rushing through. The simplified connection schematic is shown in Figure 2-5. Figure 2-4: Wire Connections, White and Green for Voltage

27 17 Figure 2-5: Schematic of Experiment System The cell was placed inside an environment chamber, Tenney model TJR, with temperature range of -75 C to +200 C, which allowed the tests to be done at the desired ambient temperatures. Two T type thermocouples were used to measure the cell and the chamber temperatures, and the data was recorded through the auxiliary channels of the battery tester. Figure 2-6: Cell Inside Environment Chamber

28 18 Before each test, the cell was fully discharged and then charged at room temperature following the CCCV charge protocol below: 1. Discharge the cell at constant current of 13 A (1C) until voltage of 2.75 V 2. Rest for 30 minutes 3. Charge the cell at constant current of 13 A (1C) until voltage of 4.2 V 4. Charge the cell at constant voltage of 4.2 V until current drops below 0.65 A (C/20) Fully discharging was done to keep the test consistent and also to provide an accurate measure of the cell capacity for each test. Charge capacity of 10 tests were then averaged to obtain a nominal cell capacity of 13.4 Ah. This value will be used when calculating percentage capacity consumed, unless otherwise is noted. earlier: After fully charging the cell, the following protocols were used for the tests mentioned 1C discharge at ambient temperature: 1. Cool the cell to 25 C/ 0 C/ -10 C/ -20 C/ -25 C/ -30 C/ -40 C and let it soak for 3 hours 2. Discharge at constant current of 13 A (1C) until the voltage reaches 2.75 V C-Rate at room temperature: 1. Cool the cell to 25 C 2. Discharge at constant currents corresponding to 0.2C/ 0.5C/ 1C/ 2C/ 3C until the voltage reaches 2.75 V

29 19 C-Rate at -25 C: 1. Cool the cell to -25 C and let it soak for 3 hours 2. Discharge at constant currents corresponding to 1C/ 2C/ 3C/ 4C/ 5C until the voltage reaches 2.75 V Basic HPPC at room temperature and -25 C: 1. Adjust the cell temperature to 25 C/ -25 C and let it soak for 3 hours 2. Discharge at 1C to 90% SOC 3. Rest for 5 minutes 4. Discharge pulse of 10 seconds at 10C or until voltage of 2 V is reached 5. Rest for 20 seconds 6. Charge pule of 10 seconds at 4C or until voltage of 4.3 V is reached 7. Rest for 2 minutes 8. Discharge at 1C to the next SOC level (80%, 70%, 60%, etc.) 9. Rest for 5 minutes 10. Repeat steps 4 through 9 until fully discharged In the next section, the results of each test will be presented and analyzed. It will be shown that both temperature and C-rate have notable impact on cell performance. 2.2 Results Figure 2-7 shows voltage vs capacity of the experimental cell during 1C discharge at different ambient temperatures. The most obvious observation is the decrease in discharge capacity as temperature decreases. At -30 C the discharge capacity was only 60% of room

30 20 temperature capacity. This agrees with our expectation and previous studies since internal resistance increases as temperature decreases. Note that this cell completely failed to perform at -40 C. That means due to high internal impedance, the overpotential at -40 C was great enough to reduce the cell voltage from open circuit voltage (OCV) to less than the 2.75 V minimum voltage limit. Another interesting phenomenon is the initial voltage drop followed by increase in voltage that can be seen at temperatures below -10 C. Zhang et al. (2014) 29 observed similar results, and they explained that the initial decrease is due to the high internal resistance of the cell at low temperature. However this high resistance yields to faster irreversible heat generation at the beginning. As temperature increases, the cell resistance decreases which allows the cell voltage to increase, until it starts to decrease again due to SOC reduction. This explanation can be verified by looking at the temperature profile shown in Figure 2-8 during 1C discharge at -25 C. It can be seen that during the first 5 minutes of the test the temperature increases at a much higher rate, which is also the region where voltage is increasing. From minute 8 forward, the temperature rise become much more gradual and voltage starts to decrease with SOC.

31 Voltage (V) Temperature ( C) Voltage (V) degC -10degC -25degC -40degC 0degC -20degC -30degC Capacity (Ah) Figure 2-7: Voltage vs. Capacity. 1C Discharge at Various Temperatures Voltage Temperature Time (min) Figure 2-8: Temperature vs. Voltage Profile. 1C Discharge at -25 C -30

32 Internal Impedence (mω) 22 From the 1C discharge results, the cell s DC resistance after 1 second of current flow at each temperature can be obtained. DC resistance can be calculated using Ohm s law shown in equation 2-2, and the difference of OCV and voltage after one second of current passing through. The calculation is shown in equation 2-3 and the results have been plotted in Figure 2-9. The increase of internal resistance as temperature decreases can be clearly seen in the plot; the resistance at -40 C is more than 17 times higher than the resistance at 25 C. (2-2) (2-3) y = e x R² = Temperature ( C) Figure 2-9: DC Resistance vs. Temperature

33 Voltage (V) 23 Next set of data was obtained by discharging the cell at room temperature using different C-rates. Figure 2-10 shows the comparison of discharge capacities at different C-rates. It can be seen that higher c-rate has caused decrease in both cell voltage and discharge capacity. This can be explained by looking at the Ohm s law which suggests the voltage drop due to ohmic and kinetic resistance should increase as current increases. Such result was expected and in agreement with previous study by Zhang et al. (2013) 30. Note that if we look at the DC resistance at different C-rates shown in Figure 2-11, we see that the cell resistance is lower at higher C-rates. This is a non-linear and non-ohmic behavior as it can be seen from Figure The ohmic line in the Figure 2-12 was obtained using Ohm s law and cell resistance at 0.2C. It is believed that the non-ohmic effect at high C rates is due to the large concentration polarization and sluggish kinetics C 0.5 1C 2C 3C Capacity (Ah) Figure 2-10: Voltage vs Capacity. C-Rate at 25 C

34 ΔV (V) Impedence (mω) C-Rate Figure 2-11: DC Resistance vs. C-Rate at 25 C Non-Ohmic Ohmic Current (A) Figure 2-12: Voltage vs Current During Non-Ohmic Behavior at 25 C

35 25 Similar results were achieved when the C-rate test was done at -25 C. Figure 2-13 and Figure 2-14 show the voltage vs. discharge capacity at five different C-rates. Three key differences can be seen between the discharge at 25 C and -25 C. First, the cell failed to perform at 3C, 4C, and 5C at -25 C. This is due to the high internal resistance which combined with high C-rate has imposed an overpotential that surpassed the 2.75 V voltage limit. Another difference can be seen during 2C discharge where a second dip in the curve appears. Lastly, the discharge capacity at 2C was slightly greater than discharge capacity at 1C. The improved capacity at higher discharge rate is due to higher heat generation. Figure 2-15 illustrates the temperature profile of each C-rate. It is clear that temperature rise during 2C discharge was more than twice of 1C discharge. Hence leading to decrease in resistance and increase in discharge capacity. The appearance of the second dip during 2C discharge is suspected to be linked to further and sudden reduction of resistance due to rapid heating. Figure 2-15 shows that voltage began to increase again when cell temperature was around -15 C. It is possible that a segment of internal resistance showed a notable improvement at this temperature which allowed the voltage to increase again. Also the second dip was not observed during 1C discharge because -15 C was not reached until the end of the test. Further investigation of the cause is beyond the scope of this paper and subject to future work. Figure 2-16 shows the DC resistance of each C-rate at -25 C. Similar non-linear and non-ohmic behavior that was observed during 25 C test can be seen here, as the resistance decreases with increasing current.

36 Voltage (V) Voltage (V) C 2C 3C 4C 5C Capacity (Ah) Figure 2-13: Voltage vs. Capacity. C-Rate at -25 C C 2C 3C 4C 5C Capacity (Ah) Figure 2-14: Voltage vs. Capacity Zoomed. C-Rate at -25 C

37 Impedence (mω) Voltage (V) Temperature ( C) C Voltage 2C Voltage 3.6 1C Temperature 2C Temperature Capacity (Ah) Figure 2-15: Voltage and Temperature Profile vs. Capacity C-Rate Figure 2-16: DC Resistance vs. C-Rate at -25 C

38 28 The data presented thus far clearly showed the decrease in performance of li-ion battery at low temperatures. Another way to better demonstrate the poor performance is to look at the discharge energy and the discharge power. Both those quantities are of greater importance to vehicle industry, since energy determines the range of electric vehicles and power is needed during acceleration and regenerative braking. Table 2-2 lists the discharge energy of this cell at various temperatures, and it compares them to 25 C discharge energy as reference. The values in the table clarify that discharge energy is reduced even further as temperature goes below 0 C. For instance at -30 C the discharge capacity is 60% of the discharge capacity at 25 C, but the available discharge energy is only 50% of the room temperature discharge energy. Table 2-2: Discharge Energy at 1C at Different Temperatures Temperature ( C) Discharge Energy (Wh) Dis. Energy / Dis. Energy 25 C % % % % % % % To obtain the power comparison data, basic HPPC test was conducted at room temperature and -25 C. Figure 2-17 and Figure 2-18 shows the results of the tests at each temperature respectively.

39 Voltage (V) Current (A) Voltage (V) Current (A) Voltage(V) Current(A) Time (min) Figure 2-17: HPPC at Room Temperature V Voltage(V) Current(A) Time (min) Figure 2-18: HPPC at -25 C

40 30 It can be seen from the figures that at -25 C only 7 HPPC cycles were able to be completed before the cell voltage dropped below 2.75 V during 1C discharge. Another observation is that at every cycle the cell voltage reached its upper limit of 4.3V and lower limit of 2V during charge pulse and discharge pulse respectively. Consequently every pulse lasted only for a fraction of a second, unlike HPPC at room temperature in which the upper and lower voltage limit were never reached and every pulse lasted 10 seconds which was the full duration of the pulse. Hence it is fair to argue that the HPPC test at -25 C failed to complete. Even though the pulses during HPPC at -25 C were less than one second, comparison of the discharge and charge power versus the depth-of-discharge (DOD) at both temperatures have been made in Figure 2-19 and Figure 2-20 respectively. It can be seen that discharge power at -25 C is on average 40% less than discharge power at 25 C. Once again keep in mind that the power was available for 10 seconds at 25 C, but only for tenths of a second at -25 C.

41 Discharge Power (W) Discharge Power 25 C Discharge Power -25 C DOD % Figure 2-19: Discharge Power vs. DOD Charge power results on the other hand shows a random behavior at -25 C. Further investigation showed that the duration of the pulse during charge was around 0.03 seconds on average (average duration of discharge pulse was.2 seconds and maximum of 0.8 seconds). Hence the results shown in Figure 2-20 are most likely the initial battery tester output power. During charge pulse, the battery tester outputs a small amount of power and adjusts it to match the current of 4C. In none of the cycles the battery tester was able to adjust its output power to reach 52 A (4C) because the cell voltage has already reached the upper limit of 4.3 V, and therefore the pulse was ended abruptly.

42 Discharge Power (W) Charge Power 25 C Charge Power -25 C DOD % Figure 2-20: Charge Power vs. DOD 2.3 Discussion The results outlined in the previous section of this chapter clearly confirmed the issue of poor performance of li-ion batteries at low temperature. It was presented that the NCM-graphite cell used during the experiments lost 30% of its discharge capacity at -25 C. Higher C-rate during discharge amplified the problem since the cell was not able to discharge above 2C. Discharge energy was also reduce at this temperature by around 45% in comparison with room temperature. This finding is crucial for vehicle industry since discharge energy is in direct relation with electric range, and also explains the result of the test done by AAA Automotive Research Center which claimed that electric range of EVs decreased by 57% at temperature of -7 C 15. The HPPC test also showed that the discharge power at -25 C was 40% less than the room temperature discharge

43 33 power. Also it must be noted that every pulse was ended in less than 1 second due to lower voltage limit, while at 25 C each pulse lasted the full scheduled time of 10 seconds. The underlying cause of all the discussed problems, from the system level point of view, is the increase of internal impedance with decrease of temperature. As seen from the DC resistance vs temperature plot, at -25 C the internal resistance is 10 times higher than that of room temperature. Chemical and material scientists are working hand in hand with battery engineers to make lithium ion batteries reliable at all temperature, but for now simple and efficient solutions are urgently needed. In the next chapter we will see how the method of Mutual Pulse Heating uses the increase of internal resistance to its advantage, and heats up the cell in around 2 minutes by using less than 5% of its capacity.

44 34 Chapter 3 Li-Ion Cell Performance at Low Temperatures Using Mutual Pulse Heating 3.1 Previous Work By now it is clear that cold temperature has a tremendous negative effects on the performance of li-ion batteries, which is a key setback for electric vehicles. To address this issue, vehicle manufacturers have been using different heating strategies such as liquid heating in Chevrolet Volt and Tesla, or passive air for Nissan leaf 31. More and more research is also being done around the topic of thermal management with respect to heating at low temperatures In general, there are 4 requirements that heating methods must meet in order to be considered effective: 1. Fast Heating Time. Comparable to conventional vehicles, electric vehicles must also be drive-ready in extreme cold temperatures within a few minutes of being turned-on. 2. Uniform Temperature Distribution. Temperature variation between the cells of battery pack could lead to different performance and charge/discharge behavior which will eventually effect the cycle life of the pack. 3. High Efficiency. Heating strategy must be as efficient as possible to prevent waste of energy that is scarce at low temperatures. This is particularly important in the case of EVs where the battery is the only source of energy.

45 35 4. Cost. Although the thermal management system is key to having safe and reliable electric vehicles, the cost of it should not make a significant increase on the already high price of electric vehicles. One method is Mutual Pulse Heating which has been developed at Electrochemical Engine Center 37 of Penn State University, and theoretically explored and presented by Ji and Wang (2013) 5. The working principles of mutual pulse heating is simple. First the modules inside the battery pack will be divided into two groups through the use of contactors or relays. Then each group charge and discharge the other group in short pulses repeatedly until the desired temperature is achieved. Small portion of output power of the discharge group will heat the charge group through internal resistance, and the rest of the power will be stored in the charge group cells. Higher voltage that is needed for charging can be obtained by the use of DC-DC convertor, which most EVs and PHEVs already have on-board. In this chapter Mutual Pulse Heating will be experimentally demonstrated for the first time in order to further advance this novel method. Mutual Pulse Heating method is of interest due to its effectiveness and potential for industry use. That is because this method meets all the mentioned requirements. It will be shown that mutual pulse heating method is capable of increasing the cell temperature from -25 C to 0 C in less than 2 minutes. Uniform temperature distribution is guaranteed since all cells are being heated internally. Vlahinos and Pesaran (2002) 35 have also mentioned that in compare to using electric heaters

46 around cells or module, or using hot fluids around cells, internal heating of battery by utilizing the increased cell impedance yields the most uniform heating. 36 Efficiency of this method is greatly dependent on the efficiency of the DC-DC convertor. However during the experiments where 100% DC-DC efficiency was assumed, 25 C temperature rise was achieved using as low as 5% of cell capacity. Lastly, this method is low cost since it does not require any additional equipment or sophisticated flow channels for fluids. Contactors that can split the battery pack in half are inexpensive, and as mentioned earlier the DC-DC convertor that is needed to charge each group is already available on most EVs and PHEVs. The DC-DC convertor however must be capable of outputting power at high frequencies, and it must have variable output voltage setting. These could add to the cost of it, but it is still expected that the overall cost of mutual pulse heating to be lower than the cost of liquid or air heating. Before analyzing the experimental data, the theoretical data that was obtained by Ji and Wang must be explored to gain an expectation of the results. During their simulations, they discharged one cell at constant voltages of 2.2 V, 2.5 V, and 2.8 V while charging the second cell using the power that was obtained from discharge. Both cells were modeled to be at 64% SOC and at ambient temperature of -20 C. Pulse charge and discharge duration was 1 second each and was continued until cell temperature reached 20 C. The results of his simulations showed that the quickest temperature rise (less than 80 seconds) was achieved when the lowest discharge voltage (2.2 V) was used. Figure 3-1 shows the voltage and temperature profile at each discharge voltage.

47 37 Figure 3-1: Voltage and Temperature evolution during Mutual Pulse Heating 5 The first 20 seconds of the heating process were magnified, and are shown in Figure 3-2 and Figure 3-3 corresponding to cell 1 and cell 2 respectively.

48 38 Figure 3-2: Voltage Profile of Cell 1 During the first 20 Seconds 5 Figure 3-3: Voltage Profile of Cell 2 During the first 20 Seconds 5

49 39 It was noted that the low discharge voltage has led to the highest charge voltage of around 4.75 V due to higher current and power during discharge. This high voltage during charge could cause lithium plating, especially at low temperatures. Hence investigation was done around lithium ion concentration on the graphite particle surface to see if lithium plating is of concern. Figure 3-4 shows the results of the simulation that was done with three different pulse rates of 0.1 second, 1 second, and 10 seconds. Also the starting voltage has been increased from 3.8 V (64% SOC) to 4.0 V to examine the cells at higher SOCs. It can be seen that only during the first 10-second pulse the lithium concentration at the anode came close to 1. It was then concluded that occurrence of Li plating is less likely at higher frequencies. Figure 3-4: Lithium Ion Concentration on Graphite Particle Surface 5 The results of the theoretical simulations clearly show why this method is preferred. First uniform temperature distribution is obtained through internal heating. Unlike heating by

50 40 discharging cell which has a limited allowable discharge rate that yields to long heating time, this method allow for high discharge rates for short durations. Finally most of the energy will be passed back and forth between the charging and discharging group, causing small capacity consumption to heat up the cells. 3.2 Experimental Setup The setup to the one introduced in Chapter 2 is used for Mutual Pulse Heating experiments. A single 13 Ah NCM-Graphite cell represents the first cell and the battery tester was used to act as the secondary cell. Prior to each test, calculations were done so that the charge power from the battery tester be less than or equal to the discharge power of the cell in order to simulate a 2 cell mutual pulse heating. The main goal of the experiments were to show the practicality of mutual pulse heating rather than validating the simulation results. Therefore some of the test criteria were changed. First, the cell was cooled to -25 C and then heated up to only 0 C. Zero degrees was chosen to increase the efficiency. As shown in previous chapters, the cell resistance increases slowly from room temperature to 0 C, but suffers an exponential increase at sub-zero temperatures. Hence most of lost battery performance will be regained at around zero, and further increase of temperature would only increase the heating time and capacity consumption without significant performance improvement. Note that discharge capacity at 0 C and -25 C is 90% and 67% of discharge capacity at room temperature as shown earlier in Figure 2-7.