Method, device and computer program product for battery memory lifetime estimation

Abstract

The lithium-ion single cells after formation are measured using high precision coulometry in order to relatively quickly group the lithium-ion single cells into a grade according to their expected aging, wherein one or more criteria including coulombic efficiency, energy efficiency, average capacity loss, and effective single cell internal resistance are used for the grouping.

Description

TECHNICAL FIELD

[0001] The present application relates to a method, a device and a computer program product for battery storage life estimation, in particular for determining an expected capacity loss of a battery storage. BACKGROUND

[0002] Lithium-ion batteries are used as energy storage in many mobile and stationary applications due to their high power and energy density. In this new context, lithium-ion batteries are increasingly used in electrically driven vehicles and as stationary intermediate storage for power supply in buildings. In the following, the batteries are referred to as cells or battery storages in everyday language, for example in these technical application areas.

[0003] Battery storages for power supply of vehicles or for grid connection must have a voltage of 400 V to 1000 V. For this purpose, a larger number of several hundred cells are usually connected in series. In addition, in order to increase the capacity or current-carrying capacity, not individual cells, but also blocks consisting of a plurality of cells connected in parallel, respectively, can also be connected in series.

[0004] Here, the disadvantage is that the behavior of such a battery storage is determined by the worst cell or the worst block in the series connection, respectively. If, for example, only one cell in the series connection ages significantly faster than the other cells, the entire battery will appear to age significantly.

[0005] Therefore, it is usually attempted to classify the cells after manufacture according to their expected life (i.e. expected aging). This step is assigned to the so-called "End-of-Line Test" in the process chain and is declared, for example, as a "self-discharge test" process step. Such a self-discharge test can be carried out, for example, by storing the cells after formation for a longer period of several weeks, wherein the cell voltage is measured at regular intervals. The aging of the cells is then evaluated from the self-discharge rate. The disadvantage of this is the long, cost-intensive storage and measurement of the cells and the fact that the aging prediction from the self-discharge test does not have a high accuracy. Therefore, the individual course of aging of the cells has a high bandwidth under the same classification result. SUMMARY

[0006] The technical problem addressed by the present application is to provide a method and a device with which the classification of cells can be carried out more accurately.

[0007] The above technical problem is solved by a method having the features of the present application and by a device having the features of the present application.

[0008] In the method for classifying a battery cell according to the application, a plurality of load cycles of the battery cell are measured by means of a high-precision Coulomb counter device, wherein the measurement results comprise a plurality of current values. Here, a working point is preferably selected for the load cycles which is as effective as possible but at the same time also sparing, which working point in particular has a state of charge of less than 50%.

[0009] The measurement is carried out until a termination criterion is met, and from the measurement results a first value and a second value for the discharge capacity of the battery cell are obtained or determined by means of first and second calculation rules, wherein the calibration of the current measurement enters into the first and second calculation rules in different ways, and an optimization method is carried out in which a calibration of the current measurement is obtained (ermitteln), which calibration is used to achieve the greatest possible agreement or matching of the obtained first and second discharge capacities

[0010] Finally, an aging criterion of the battery cell is determined from the measurement results, and the battery cell is classified into one of a plurality of classification regions in accordance with the aging criterion.

[0011] The device according to the application is designed to carry out the method according to the application and comprises a high-precision Coulomb counter device and a calculation unit with a memory for recording the results of a series of measurements of a battery cell, wherein the calculation unit is designed to carry out the optimization method and to classify the battery cell into one of a plurality of classification regions.

[0012] With the application, the possibility is advantageously created of being able to classify battery cells relatively quickly in terms of their expected service life. In other words, a new method for accelerating the quality testing of lithium-ion batteries after formation and improving the classification accuracy is created.

[0013] Whereas a duration of several weeks can be required for quality determination according to the prior art, this duration is reduced to a few days by means of the application, whereby the capital investment for storage during the test can be significantly reduced.

[0014] Advantageous design options of the application can be gathered from the preferred embodiments of the application. Here, the above-mentioned embodiments according to the application can be combined with the features of one of the preferred embodiments, or preferably also with the features of a plurality of preferred embodiments. Thus, the following features can also be provided additionally:

[0015] The Coulomb efficiency of the battery cell can be determined as the aging criterion. Alternatively or additionally, the energy efficiency and / or the effective cell internal resistance and / or the capacity loss per cycle of the battery cell can also be determined as the aging criterion.

[0016] If two or more of the mentioned parameters are used, it is advantageously no longer only a single characteristic parameter, i.e. the self-discharge rate, which is used for quality assessment by using the application, but rather a plurality of independent Key Performance Indicators (KPIs) determined in a single measurement. The robustness of the Quality Management (QM) process is thereby increased, the effectiveness is increased, and new possibilities for integrating data backflow from the field use of the produced battery cells are opened up.

[0017] The aging criteria can be stored in a database together with an identification mark for the battery cell. This storage enables an evaluation of the data collected in this way afterwards, for example in order to identify differences relating to different production batches. Furthermore, the storage allows a comparison with aging data obtained in the field use of the battery cell. It is particularly advantageous if real aging data from a battery cell in use are received in the battery memory and the classification region is adapted in accordance with the real aging data and the stored aging criteria. It is also possible to determine a changed operating point in accordance with the real aging data, which enables a future improved prediction accuracy.

[0018] The load cycle can comprise a discharge below 40%, in particular below 25%. Furthermore, the load cycle can work with a C-factor between 0.5 and 1.5, in particular between 0.8 and 1.2. Thereby, together with the charging, an advantageous operating point for the measurement of the battery cell is created, on which the State of Charge (SOC) as well as the Depth of Discharge (DOD) are not too high in order to reduce the resulting current costs, and on the other hand, the discharge is sufficient for an effective measurement. Such an advantageous operating point is 30% SOC, 20% DOD and 1C.

[0019] For at least a part of the load cycle, one or more other operating points different from the already mentioned operating point can be used. Thereby, the measurement is made more effective for different operating situations occurring in the actual operation of the battery cell.

[0020] Here, the load cycle preferably comprises a first discharge, in which a first charge quantity from a first state of charge to a second state of charge is measured, a subsequent first charge, in which a second charge quantity from the second state of charge to a third state of charge is measured, a second discharge, in which a third charge quantity from the third state of charge to a fourth state of charge is measured, wherein the charges and discharges of the load cycle are carried out between a lower voltage and an upper voltage of the battery memory.

[0021] Here, the first charge offset can be determined by means of the difference between the fourth charge state and the second charge state, and the second charge offset can be determined by means of the difference between the third charge state and the first charge state. Furthermore, the capacity loss can be determined from the difference between the first charge offset and the second charge offset, and the capacity loss can be averaged based on at least two capacity losses of different load cycles.

[0022] The capacity loss acquired in this way can be used as a termination criterion. Here, in particular, the relative change in the capacity loss in two or more load cycles following one another can be taken into account. If this relative change is sufficiently low, the battery cell can be considered to have stabilized and the measurement can be terminated.

[0023] The capacity loss acquired in this way can also be used as an aging criterion for the battery cell. Thus, in addition to the already mentioned aging criterion, another such aging criterion can be used which improves the effectiveness of the measurement.

[0024] The termination criterion can be selected depending on a classification derived from the existing measurement results. Thus, for example, if it emerges for the battery cell that it has to be considered as a reject or is classified into a poor category, the measurement can be terminated.

[0025] The termination criteria used can be combined with one another. Thus, for example, the measurement can be terminated if a minimum number of load cycles measured so far is exceeded and the deviation of the capacity loss is below a threshold value.

[0026] For the present application, there can be a computer program product which can be loaded directly into the memory of a programmable computing unit. The computer program product comprises program code means for implementing the method according to the present application when the computer program product is implemented in the computing unit.

[0027] The computer program can be integrated into a higher-level process control and quality assurance of the battery production.

[0028] Here, if the power electronics used for the formation (formation circuit) can also be used for the HPC measurement or can be extended accordingly, the device can also advantageously be integrated into the existing production without additional hardware. Alternatively, the device can comprise a completely new station in the production process, which can replace or supplement the existing measurement stations, for example for the self-discharge test. BRIEF DESCRIPTION OF DRAWINGS

[0029] Further features, properties and advantages of the present application result from the following description with reference to the drawings. In the drawings:

[0030] Figure 1 a device for classifying battery cells is shown,

[0031] Figure 2 A schematic diagram of a method for classifying a battery cell by means of the device is shown,

[0032] Figure 3 A voltage-time diagram of a load cycle is shown,

[0033] Figure 4 A voltage-charge diagram of a load cycle is shown,

[0034] Figure 5 A diagram showing the capacity loss during a load cycle is shown. DETAILED DESCRIPTION

[0035] Figure 1 A device 1 for classifying a battery cell 2 is shown. The device 1 comprises a high-precision coulomb meter device 4 and a temperature-controlled chamber 3. The battery cell 2 to be classified is introduced into the temperature-controlled chamber 3 and connected to the high-precision coulomb meter device 4 by means of a power cable 11.

[0036] The high-precision coulomb meter device 4 is connected to a computing unit 10 by means of a data cable 12. The high-precision coulomb meter device 4 records the charge-time diagram of the battery cell 2 with very high accuracy. Here, the battery cell 2 is operated with a periodic load cycle 100.

[0037] The computing unit 10 comprises a computer program 13 which processes the data transmitted by the high-precision coulomb meter device 4. The computer program at least temporarily stores these values.

[0038] Figure 2 A schematic diagram of a method for classifying a battery cell 2 is shown, which is carried out using the device 1 shown in Figure 1 Here, the computational aspects of the method are carried out by the computer program 13 on the computing unit 10.

[0039] In a first step 201, the battery cell 2 is introduced into the temperature-controlled chamber 3 after formation and temperature-controlled. A constant temperature is thus achieved during the subsequent measurement of the load cycle 100. In some embodiments, the first step 201 can be dispensed with, depending on the manner in which the battery cell 2 is fed to the temperature-controlled chamber 3, while in other embodiments, this step 201 can last for several hours.

[0040] The second step 202 follows the first step 201, in which the battery cell 2 is measured by means of the high-precision coulomb meter device 4. Here, the already mentioned load cycle 100 is operated and at least a current measurement is carried out, which allows the determination of the charge quantity.

[0041] The second step 202 further comprises a third step 203 in which a current calibration is performed using the acquired data. Furthermore, the second step 202 further comprises a fourth step 204 in which a decision is made as to whether a termination criterion is present. If the termination criterion is not present, the measurement of the battery cell 2 is continued with another load cycle 100.

[0042] Here, the third and fourth steps 203, 204 can be performed after each load cycle or always after a determinable number of further measured load cycles. Here, the third and fourth steps 203, 204 are in principle independent of one another, but can also be performed at the same time. In the present example, the third and fourth steps 203, 204 are implemented after 10 load cycles 100, respectively.

[0043] In this embodiment, a combination of different factors is used as a termination criterion. A first factor is a minimum number of load cycles 100, in this case 100. Experience shows that the behavior of the battery cell does not reach a steady state before this number of load cycles 100. A deviation of the capacity loss between load cycles 100 following one another is used as a further factor. This will be described further below. If this deviation is below 5%, a steady state of the battery cell 2 can be derived. Furthermore, the termination criterion comprises giving a classification for the battery cell 2, provided that the data to date is taken into account after the minimum number of load cycles 100. If this classification is such that the battery cell is to be regarded as scrap or is to be classified into a lower aging class, the further measurement is terminated. Conversely, if the battery cell 2 has a high expected quality in terms of aging, the measurement is continued.

[0044] If the termination criterion is met, the measurement of the battery cell 2 is ended. In a fifth step 205, a final aging criterion is then determined. In this example, in the fifth step 205, the library efficiency and the average capacity loss are determined and stored. The acquired values are stored in a database together with an identification mark, for example a serial number in respect of the battery cell 2.

[0045] In a sixth step 206, the battery cell 2 is finally classified on the basis of the acquired data. Here, the classification is made in predetermined classes, which are subsequently used as quality criteria in order to use battery cells 2 of the same quality together as far as possible. Thereby, it is avoided that the properties of a battery storage are determined too much by individual "bad" battery cells 2. The classification thus groups the battery cells 2 into classes with a similar expected aging.

[0046] Since the aging criteria determined in the fifth step 205 are stored together with the identification mark, in a seventh step 207 the prediction values acquired in this way can be compared with actual aging data about the battery cell 2. For this purpose, such actual aging data of the battery cell 2 during its use in an environment such as an electric vehicle, a locomotive or a traction battery are recorded and compared with the aging criteria. Here, if a systematic deviation from the expected aging, i.e. a systematic deviation from the final said classification, is shown, which is related to one or more of the aging criteria, the classification is corrected, for example by changing the weight of one of the aging criteria. The changed weight leads to an improved classification of the new battery cell 2.

[0047] The actual measurement of the battery cell 2 is carried out by means of the high-precision Coulomb counter device 4. Figure 3 A voltage-time diagram recorded by the high-precision Coulomb counter device 4 during a periodic load cycle 100 of the battery cell 2 is shown. The load cycle 100 comprises a discharge from a first charge state 21 to a second charge state 22, wherein the first charge state 21 is at an upper voltage 25 and the second charge state 22 is at a lower voltage 26. The battery storage 2 is then charged from the second charge state 22 to a third charge state 23 in the load cycle 100. As a next step, the third charge state 23 is discharged to a fourth charge state 24 in the load cycle 100. In each individual charging / discharging step, the upper voltage 25 and the lower voltage 26 are maintained as voltage limits. The charging lasts for a charging period t C . The discharging lasts for a discharging period t D .

[0048] Based on the measurements shown in Figure 3 , it is now possible to acquire how much cumulative charge quantity has flowed in the individual charging and discharging steps as shown in Figure 4 . Figure 4 A diagram is shown in which the voltage of the battery storage is plotted with respect to the cumulative charge quantity Q. The load cycle 100 starts again in the case of the first charge state 21. The battery storage 2 is discharged in the case of the first discharge 31 until the second charge state 22. Here, a first charge quantity Q1 is taken out of the battery storage 2. The first charge quantity Q1 can be calculated by means of formula 1, wherein I denotes the current, t D denotes the discharging period:

[0049]

[0050] Within the load cycle 100, the battery storage 2 is then charged from the second charge state 22 to the third charge state 23 by means of the first charge 32. A second charge quantity Q2 is loaded into the battery storage 2. Q2 can be calculated by means of formula 2:

[0051]

[0052] In the load cycle 100, the battery storage 2 is then discharged from the third charge state 23 to the fourth charge state 24 by means of a second discharge 33. Again, the amount of charge Q3 taken out can be calculated from the discharge period and the associated current, analogously to equation 1.

[0053] The first charge offset di between the first charge state 21 and the third charge state 23 can now be acquired. Furthermore, the second charge offset d2 between the second charge state 22 and the fourth charge state 24 can be acquired. The capacity loss dKap for the load cycle 100 can now be acquired from the difference between the first charge offset di and the second charge offset d2 by means of equation 3.

[0054] dKap = d2 - di equation 3

[0055] The capacity loss for each load cycle with regard to 250 load cycles is shown in Fig. 1 1. Here, the load cycle number Z, i.e. the number of runs of the respective load cycle 100, is located on the x-axis, and the capacity loss dKap for each load cycle 100 is located on the y-axis. Figure 5 The capacity loss for each load cycle with regard to 250 load cycles is shown in Fig. 1 1. Here, the load cycle number Z, i.e. the number of runs of the respective load cycle 100, is located on the x-axis, and the capacity loss dKap for each load cycle 100 is located on the y-axis. Figure 5 It is explained that a transient oscillation phase PI first occurs during the load cycles 100 following one another. The length of the transient oscillation phase PI depends on the working point and the history of the battery cell 2.

[0056] The average capacity loss dKap Mittel is determined as the measure of the method in that the values of the capacity loss dKap are fitted linearly smoothed and the smallest slope in the linear equation produced in this way is determined. Based on the fit of all values of the capacity loss dKap, i.e. values 1 to values 250, the data set is continuously shortened and new straight lines are produced (2 to 250, 3 to 250, etc.) are fitted. The fit is performed until a determined minimum residual length of the data set, for example 10% of the total length. The linear equations are then sorted in ascending order, in particular according to the values of their parameter slope. The measure can be considered valid if at least two slopes have a value in terms of magnitude that is less than 10% of the average of the last 10% of the capacity losses dKap. For example, in particular when at least 200 capacity losses are measured, if the average of the last 20 capacity losses is 5 mAh per load cycle, the slopes of the two best fitting tangents ("fit") should be less than 0.5 mAh per load cycle.

[0057] Otherwise the measurement has to be repeated, in particular with a higher number of support points, because a sufficiently stable state of the system has not yet been reached. From the classification a certain number is selected, for example 3% of the total length of the data set or a minimum number of two measurement values, and the corresponding start indices of the straight lines after the fitting are taken. For each section taken in this way the average capacity loss dKap Mittel is determined as the average of the individual capacity losses averaged.

[0058] If a sufficiently stable, i.e. essentially constant, capacity loss has not yet been reached, the measurement of the load cycle is repeated. Then again a certain number is selected from the classification, for example 3% of the total length of the data set or a minimum number of two measurement values, and the corresponding start indices of the straight lines after the fitting are taken. For each section taken in this way the average capacity loss dKap Mittel is given as the arithmetic mean of the capacity losses contained. However, it is also possible to take the value of the average capacity loss dKap Mittel as the average of the arithmetic mean capacity losses.

[0059] Figure 5 It is also stated that after the transient oscillation phase P1 is the acquisition phase P2. These phases can be offset during the evaluation of the capacity loss dKap.

[0060] In addition to the capacity loss, the coulomb efficiency can also be used as an aging criterion. The coulomb efficiency is calculated as

[0061] CE = Q3 / Q2 Formula 4 For the current calibration performed in the third step, it can be used that the discharge capacity Q0 which can be assigned to the discharged state, i.e. for example the second charged state, can be calculated in two different ways, wherein the current calibration of the high-precision coulomb meter device 4 enters into the two calculations in different ways. Thus, Q0 according to the first rule is

[0062]

[0063] In addition, Q0 can be calculated from the initial discharge capacity Q0A assigned to the previous load cycle 100 and the capacity losses between said previous load cycle 100 and the current load cycle:

[0064] Q 0m = Q 0A +∑dKap Formula 7 In the case of ideal, i.e. error-free, current measurement, these two values are identical, i.e.

[0065] Q0 = Q 0mFormula 8 In reality, however, the two values differ from each other due to the not entirely accurate current calibration present in the current measurement. The greater the difference in values, the less correct the current calibration.

[0066] Formula 8 is used in the form f = Q0- Q 0m as a basis for optimization in which the function value f is minimized. The parameter that is changed for the optimization forms the current calibration. The current calibration is a mapping from the measured current value to the corrected measurement value. If a great degree of agreement between the values is achieved through the optimization, the corrected measurement value corresponds very accurately to the actual current. The optimization is performed in the computing unit 10 by the computer program 13.

[0067] List of reference signs

[0068] 1 device

[0069] 2 battery cell

[0070] 3 temperature control chamber

[0071] 4 high-precision Coulomb counter device

[0072] 10 computing unit

[0073] 11 power cable

[0074] 12 data cable

[0075] 13 computer program product

[0076] 21 first charge state

[0077] 22 second charge state

[0078] 23 third charge state

[0079] 24 fourth charge state

[0080] 25 upper voltage

[0081] 26 lower voltage

[0082] 100 load cycle

[0083] t C charging period

[0084] t D discharging period

[0085] 201...207 first step...seventh step

Claims

1. A method for classifying a battery cell (2), in which method - measuring a plurality of load cycles (100) of the battery cell (2) by means of a high-precision Coulombmeter device (4), wherein the result of the measurement comprises a plurality of current values, - the measurement is performed until a termination criterion is met, - a first value and a second value for the discharge capacity of the battery cell (2) are obtained from the result of the measurement by means of a first calculation rule and a second calculation rule, in which the calibration of the current measurement enters into the first and second calculation rules in different ways, and an optimization method is performed in which a calibration of the current measurement is obtained, which is used to achieve the greatest possible agreement of the obtained first and second discharge capacities, - an aging criterion for the battery cell (2) is determined from the result of the measurement, - the battery cell (2) is classified into one of a plurality of classification areas on the basis of the aging criterion.

2. The method of claim 1, wherein, The coulombic efficiency of the battery cell (2) is determined as an aging criterion.

3. The method of claim 1, wherein, The energy efficiency and / or the effective cell resistance and / or the capacity loss per cycle of the battery cell (2) is determined as an aging criterion.

4. The method of claim 1, wherein, The aging criterion is stored in a database together with an identification mark for the battery cell (2).

5. The method of claim 4, wherein, Real aging data from a battery cell (2) in use is received in a battery memory, and the classification areas are adapted on the basis of the real aging data and the stored aging criterion.

6. The method of claim 1, wherein, The load cycle (100) comprises a discharge of less than 40%.

7. The method of claim 1, wherein, The load cycle (100) comprises a discharge of less than 25%.

8. The method of claim 1, wherein, The load cycle (100) works with a C factor between 0.5 and 1.

5.

9. The method of claim 1, wherein, The load cycle (100) works with a C factor between 0.8 and 1.

2.

10. The method of claim 1, wherein, For at least a part of the load cycle (100), a plurality of operating points are used.

11. The method of claim 1, wherein, The load cycle (100) comprises a first discharge in which a first charge quantity from a first state of charge (21) to a second state of charge (22) is measured, a subsequent first charge in which a second charge quantity from the second state of charge (22) to a third state of charge (23) is measured, a second discharge in which a third charge quantity from the third state of charge (23) to a fourth state of charge (24) is measured, wherein the charging and discharging of the load cycle (100) takes place between a lower voltage and an upper voltage of the battery cell (2).

12. The method according to claim 11, characterized in that - a first charge offset is determined by means of the difference between the fourth state of charge (24) and the second state of charge (22), and a second charge offset is determined by means of the difference between the third state of charge (23) and the first state of charge (21), - a capacity loss is determined from the difference between the first charge offset and the second charge offset, - an average capacity loss is obtained on the basis of at least two capacity losses of different load cycles (100).

13. The method of claim 12, wherein, The relative change in capacity loss in two or more load cycles (100) following one another is used as a termination criterion.

14. The method of claim 1, wherein, The termination criterion is selected in dependence on a classification derived from existing measurement results.

15. A device (1) for carrying out the method according to any one of claims 1 to 14, comprising - a high-precision Coulombmeter device (4), - a computing unit (10) having a memory for recording the results of a series of measurements on the battery cell (2), wherein The computing unit (10) is designed to carry out the optimization method and the classification.

16. A computer program product (13) which can be directly loadable into the memory of a computing unit (10) which can be programmed, said computer program product having program code means in order to carry out the method according to any one of claims 1 to 14 when the computer program product (13) is implemented in the computing unit (10).

Images