Methods for determining the aging of individual battery cells and methods for controlling battery pack aging

Abstract

This invention relates to a method for determining the aging of individual cells in a lithium-ion battery pack. The method according to the invention is characterized by actively charging or discharging the individual cells using power pulses, or detecting such power pulses during battery pack operation. This includes recording current pulses and the voltage response of the individual cells during idle periods with lower or no power following the power pulses, and evaluating the relaxation of the individual cells. At least a relaxation time constant (τ) is determined and stored. Then, based on the currently measured and stored relaxation time constant (τ), charging and discharging curves are formed with respect to the charge of the individual cells during charging or discharging with power pulses. The current capacity (K1) of the individual cell is inferred from the intersection of the charging and discharging curves.

Description

[0001] This invention relates to a method for determining the aging of individual cells in a lithium-ion battery pack. The invention also relates to a method for controlling a battery pack, the type of which is defined in more detail in the preamble of claim 11.

[0002] Lithium-ion batteries, currently commonly used as power sources for motor vehicles, are typically based on nickel-manganese-cobalt electrodes and graphite electrodes. Graphite electrodes, in particular, are susceptible to aging effects, which can impact the performance of the entire battery pack or its individual cells. To counteract these aging effects throughout the battery pack's lifespan, or at least prevent accelerated aging, it is essential to measure and accurately interpret the relevant indicators of these aging factors.

[0003] A commonly used analytical technique is differential voltage analysis (DVA). This technique can be used to determine the uniformity of lithium distribution within a battery pack. However, this technique requires full charging and discharging with the smallest and most constant current possible. This is nearly impossible in practical use, especially in motor vehicles, where charging is performed as quickly as possible, using high currents, thus always occurring just before the edge of the stabilization window, i.e., the region immediately preceding the formation of the anode capping layer. The discharge direction is also highly dynamic, frequently interrupted by short charging cycles, as very strong discharges often occur during acceleration, while recharge occurs during braking. Thus, there is no constant load, and the current direction is constantly changing. Therefore, these practical situations are, in principle, unsuitable for differential voltage analysis. This increases the difficulty of identifying aging mechanisms during operation, and without identification, these aging mechanisms cannot be effectively counteracted.

[0004] One of the inventors, Hust Friedrich Emanuel, published a paper titled "Physico-chemically motivated parameterization and modelling of real-time capable lithium-ion battery models: a case study on the Tesla Model S battery" at RWTH Aachen University in 2019 (DOI: 10.18154 / RWTH-2019-00249The paper proposes a physicochemically driven, impedance-based real-time lithium-ion model and parameterizes it using a Tesla Model S battery cell. The main focus is on the relaxation behavior of the battery cell, which is found to be essentially caused by the negative electrode (i.e., the graphite electrode). The paper hypothesizes that relaxation is caused by the redistribution of lithium within the graphite electrode. Based on these findings, further research can now be conducted using a lithium-ion battery model and further parameterization using other real-world cells.

[0005] Based on these fundamental findings, the object of this invention is to provide a method for determining the aging of individual cells in a lithium-ion battery, a method applicable to dynamic real-world operation, so as to enable the adjustment of the battery pack's operating strategy based on the identified aging behavior.

[0006] According to the invention, this objective is achieved by a method for determining the aging of individual cells in a lithium-ion battery, having the features of claim 1. Advantageous designs and improvements are given in the dependent claims. Furthermore, this objective is also achieved by a method for controlling a battery pack having at least one cell that has been determined to be aged, having the features of claim 11. In this respect, advantageous designs also arise in the related dependent claims.

[0007] In the method according to the invention, individual battery cells are actively charged or discharged using power pulses, or such power pulses for charging or discharging individual battery cells are detected during the operation of the battery pack. On one hand, the method can be actively initiated by triggering corresponding power pulses for charging and / or discharging at appropriate times (e.g., in the case of a battery pack installed in a vehicle, when the battery pack is not currently needed). However, in practice, such power pulses occurring during the normal operation of the battery pack (e.g., during acceleration or sudden braking) may be more relevant. This results in the discharge of the battery pack and its individual cells or charging due to energy recovery. Therefore, such power pulses during the normal dynamic operation of the battery pack can be easily detected and used in the method according to the invention. During the idle phase following such a power pulse, the voltage response of the individual battery cells is recorded in addition to the previously recorded current pulse. The idle phase can be a low-power phase or a no-power phase, such as when a vehicle stops at a red light after sudden braking.

[0008] Continuous recording of current and voltage is also particularly advantageous, whereby the recorded data can be evaluated in the sense described below, following the identified power pulse and the subsequent idle phase. Importantly, according to the invention, the relaxation time constant is determined and stored based on this voltage response. This is essentially also known from the paper on the prior art described at the outset, but is not applied in the sense described herein.

[0009] The determined and stored relaxation time constants can then be used to: generate charging curves for the individual cell's charge during charging or discharging, based on the most recently stored and currently determined relaxation time constants, and to generate discharging curves based on other data. The charging and discharging curves then intersect at a point from which the current capacity of the individual cell, particularly the graphite electrode, can be inferred. This establishes a first basis for determining the aging of individual cells, enabling a method for properly and carefully handling aged cells to prevent or at least minimize their aging process.

[0010] Preferably, this can be accomplished based on the transition from one stage to another, determined by the intersection points, of the open-circuit voltage curve of lithium to graphite with respect to the capacity of a battery cell or its graphite electrode, where the transition to the second stage can be detected particularly effectively. The current capacity of the battery cell can then be inferred from this. A characteristic diagram is obtained by plotting the open-circuit voltage of lithium to graphite with respect to the capacity of the battery cell, which is essentially known from the prior art and has been discussed accordingly in the aforementioned paper, particularly from page 40 onwards. The different stages within the open-circuit voltage curve are characterized. The current capacity of the battery cell can be inferred from the transition from one stage to another.

[0011] If, according to a highly advantageous design scheme based on the method according to the invention, the charging and discharging curves are plotted using a logarithmic plot of the relaxation time constant of the battery cell charge, the intersection point can be determined particularly effectively. This then essentially produces two curves, namely the charging curve and the discharging curve, which can be well approximated as straight lines. These curves form a defined intersection point through which the location of the transition, for example, the transition to the second stage in the open-circuit voltage curve, can be inferred with relatively high accuracy.

[0012] If these values ​​are now obtained, a model-based approach can be used to calculate the current open-circuit voltage curve of lithium against graphite based on the current capacity of the individual battery cell and a given open-circuit voltage curve of the cell at the beginning of its lifespan. For this purpose, according to a highly advantageous improvement of the method according to the invention, the current open-circuit voltage curve is achieved, in particular, by compressing the given initial curve of the individual battery cell at the beginning of its lifespan along the capacity direction. Specifically, this compression can be performed until the transition from the compressed initial curve to the second stage coincides with the corresponding transition of the current open-circuit voltage curve to the second stage. Therefore, this transition can be used as a reference point to obtain the current open-circuit voltage curve from the substantially known initial curve of the individual battery cell or its graphite electrode. Thus, the current open-circuit voltage curve can be modeled with relatively high accuracy based on simple measurements that can also be performed during battery pack operation (e.g., in a vehicle).

[0013] Furthermore, the current and stored relaxation time constants of the individual cells are then correlated with the simulated current voltage curve of lithium against graphite regarding the state of charge of the battery cells. Typically, the maximum values ​​occur in the transition regions between the stages, i.e., at the transition from stage 2 to stage 1, and possibly at the transition from stage 3L to stage 4L.

[0014] In the method according to the invention, the relaxation time constant itself can be determined directly or indirectly based on the minimum voltage, minimum charge conversion, the shortest time required for a cell to reach a substantially constant voltage over time, and / or based on the gradient, i.e., the slope, of the cell's voltage response to a power pulse, by means of a pre-given correlation between the gradient and the relaxation time constant, particularly one that can be stored in a lookup table. Actual measurement of the shortest time is, of course, the preferred solution here, but requires a relatively long idle period. Therefore, the slope of the cell's voltage response to a power pulse can be used primarily. Various voltages that have occurred can be assigned to a specific relaxation time constant based on historically known correlations; thus, for example, even in cases where the idle time is very short, a fairly reliable statement about the relaxation time constant can be made using values ​​stored in a table, with reference to the slope.

[0015] The current situation is that, according to a highly advantageous improvement of the method according to the invention, the currently determined relaxation time constant is compared with a previously determined and stored relaxation time constant, wherein an increase in the value of the relaxation time constant infers that the formation of the capping layer on the battery cell or its graphite electrode has become more intense. This capping layer formation, to some extent, restricts the accessibility of the active material and also restricts the lithium-ion balance, resulting in a corresponding increase in the relaxation time. Therefore, an increase in the relaxation time constant at several determined values ​​clearly indicates a more intense formation of the capping layer on the battery cell or its graphite electrode, thereby correspondingly identifying a first specific aging mechanism.

[0016] According to another highly advantageous design, it can be proposed that, particularly in the current open-circuit voltage curve of lithium against graphite with respect to the cell capacity, an increase in the loss of anode active material can be inferred from the shift of the relaxation time constant towards lower capacity. This shift, especially the shift towards lower capacity in the aforementioned current open-circuit voltage curve, i.e., the shift to the left along the x-axis in this curve, indicates an increase in the loss of anode active material. This second specific aging mechanism is also known as LAAM (Loss of Active Anode Material).

[0017] Another highly advantageous design of the method according to the invention also achieves, as an alternative or supplement to the first two aging mechanisms, an increase in lithium loss can be inferred from the decrease in the distance between the current relaxation time constant and the last stored relaxation time constant relative to the previously stored paired relaxation time constants, and particularly in the current open-circuit voltage curve of lithium to graphite with respect to the cell capacity. Therefore, when the distance between the various relaxation time constants in the graph decreases, an increase in lithium loss, also known as LLi (Loss of Lithium), can be identified. This means that the relaxation time constants move further closer to the capacity axis. Thus, a third aging mechanism can be specifically identified.

[0018] Now, in the model-based evaluation of the measured relaxation time constant, all three described mechanisms can be identified together or independently. Each aging mechanism has a different effect on the relaxation time constant, thus all three can be read from the measurements. If the specific aging mechanism present in the battery cell or its graphite electrode is now known, it can now be counteracted very specifically. For example, the power supplied to and drawn from the battery pack during charging and discharging can be adjusted during operation, or the battery cooling or temperature regulation can be altered to operate the battery cell more gently.

[0019] Referring to the current open-circuit voltage curve of the individual battery cells obtained from model analysis, the method according to the present invention for controlling the power supplied to and obtained from the battery pack can be adjusted as follows: charging and discharging of individual battery cells should be avoided as much as possible from exceeding the stage limits of the open-circuit voltage curve. Charging and discharging should occur within one stage whenever possible, and should only exceed that stage in special circumstances. This can thereby correspondingly prevent or minimize the aging process of the battery cells, particularly their graphite electrodes.

[0020] Other advantageous designs of these two methods according to the invention are also given in the embodiments and explained in more detail below with reference to the accompanying drawings.

[0021] in:

[0022] Figure 1 A schematic diagram of a method flow according to a possible embodiment of the method according to the present invention is shown;

[0023] Figure 2 A graph showing the correlation between charge and relaxation time constant is provided.

[0024] Figure 3 The basic open-circuit voltage diagram of the lithium graphite electrode is shown;

[0025] Figure 4 It shows that according to Figure 3 The open-circuit voltage plot, with additional plots of the open-circuit voltage curves of aged battery cells and additional plots of the relaxation time constants detected and stored under the condition of the formation of the capping layer;

[0026] Figure 5 It shows that according to Figure 3 The open-circuit voltage plots, with additional plots of the open-circuit voltage curves of aged battery cells and additional plots of the relaxation time constants detected and stored under conditions of loss of anode active material (LAAM);

[0027] Figure 6 It shows that according to Figure 3 The open-circuit voltage plot, with additional plots of the open-circuit voltage curves of aged battery cells and additional plots of the relaxation time constants detected and stored under lithium loss (LLi).

[0028] Figure 7 It shows the method for performing according to Figure 1 A schematic diagram of the buffer design for this method.

[0029] exist Figure 1 The diagram schematically illustrates a possible process flow for determining the aging of individual battery cells and variations in battery pack operation methods based on this. In box 100, for example, a strong current consumption, i.e., a power pulse, is detected in at least one of the battery pack or its individual cells. Now, in box 200, the current pulse and the voltage response of the individual battery cell are recorded. In a subsequent stage with less or no power extraction or supply (referred to here as the idle stage), the relaxation of the individual battery cells is evaluated. Then, in step 400, the relaxation voltage and relaxation time constant are calculated, as known in principle from the paper mentioned at the outset. Now, in step 500, a correlation is established between charge C and different relaxation time constants τ, which have been detected and stored and / or determined in real time. The following will... Figure 2 This will be discussed in more detail within the scope of [the relevant document / section].

[0030] Now, in step 600, aging can optionally be determined based on a model, where different aging mechanisms, such as capping, LAAM, and / or LLi, can be detected, which is particularly advantageous. This will... Figures 4 to 6 This will be discussed in more detail later. In the equally optional step 700, these identified aging mechanisms can now be used to select appropriate operating strategies for the battery pack, i.e., avoid specific operating states and operate the battery pack as gently as possible in terms of charging, discharging, and temperature control of the battery pack through the cooling medium, in order to prevent or at least slow down its aging process.

[0031] In step 400, the calculated relaxation time constants τ, for example, those occurring in power pulses used for charging or discharging, are calculated and stored. According to step 500, in... Figure 2 The relaxation time constants τ for charge C are plotted. Thus, with the relaxation time constant T expressed on a logarithmic scale, a straight line is formed along the black dot for the charging cycle and a straight line is formed along the white dot for the discharging cycle. Then, the intersection of these two lines is obtained, and the capacity K of the battery cell or the graphite electrode (anode) that essentially determines the aging mechanism at the current time is inferred from the charge C1 determined therefrom.

[0032] Figure 3 A schematic open-circuit voltage curve R0 of lithium to graphite is shown for the capacity K of a single cell or its graphite electrode. Here, the open-circuit voltage V is plotted on the y-axis, and the single-cell capacity K of the lithium-ion graphite electrode is shown on the x-axis, with stage 3L or stage 4L as the standard (see below). Figure 3 The open-circuit voltage curve R0 in the diagram represents the open-circuit voltage of a new battery cell or graphite electrode. The initial capacity of the battery cell or its graphite electrode at the start of its lifespan is denoted as K0. A zero open-circuit voltage at the initial capacity K0 indicates a fully charged state, where the discharge direction E of the entire battery cell is shown from right to left. The open-circuit voltage curve R0 has characteristic stages, also called phases. These stages are labeled from right to left along the capacity as Phase 1, Phase 2-1, Phase 2L-2, and Phase 3L / 4L.

[0033] Now, the open-circuit voltage curve R1 with the current capacity K1 can be modeled into the open-circuit voltage curve R0 with the known initial capacity K0. The open-circuit voltage curve R1 with the current capacity K1 is generated by correspondingly compressing the initial curve with the initial capacity K0 in the direction of capacity K. Therefore, the aged capacity K1 will be lower than the initial capacity K0, and different conditions can now be read at different positions of these open-circuit voltage curves R0, R1.

[0034] It is scientifically known that loss of anolyte active material (LAAM) causes the capacity K to contract from the right side, thus allowing for faster attainment of fully lithiated graphite electrodes. On the left side, loss of lithium (LLi) limits the capacity K. The following discussion will cover how these two effects influence the relaxation time constant τ, and also their impact on capping layer formation.

[0035] Therefore, in Figures 4 to 6 Two schematic open-circuit voltage curves, R0 and R1, can be seen in the image. The open-circuit voltage curve R0 for the unaged graphite electrode is shown as a dashed line (similar to...). Figure 3The open-circuit voltage curve R1 for aging is shown as a solid line. Here, the y-axis represents the relaxation time constant τ in seconds on one side and the open-circuit voltage V for lithium on the other. The x-axis again represents the capacity K, with a standard of 3L / 4L. In the case of aged graphite electrodes, the position of the open-circuit voltage curve R1 is determined by the voltage across the graphite electrode. Figure 2 The charts explained in the text show the current charge C1 and the aging capacity derived from it.

[0036] exist Figure 4 In the chart, in addition to the open-circuit voltage curves R0 and R1 of the new and old graphite electrodes, the relaxation time constant τ is also shown. The relaxation time constant τ of the new graphite electrode is represented by a dashed line extending from zero to the corresponding relaxation time constant τ, similar to the representation of the open-circuit voltage curve R0, while the relaxation time constant of the aged battery cell is represented by a solid line.

[0037] from Figure 4 As can be seen from the graph, the relaxation time constant τ of the aged graphite electrode is higher than that of the nascent graphite electrode at the corresponding location. This increase in the relaxation time constant τ indicates an increase in the formation of the capping layer, which is one of the aging mechanisms in the graphite electrode.

[0038] exist Figure 5 In, with Figure 4 Similar to the illustration in the diagram, this shows the effect of LAAM on the relaxation time constant τ. As the capacity K decreases due to LAAM, the relaxation time constant τ shifts to a lower state of charge (relative to the initial capacity). The relaxation time constant τ is again represented by data points. The dashed data points represent the original (fresh) state, and the solid data points represent the relaxation time constant τ shifted due to aging.

[0039] exist Figure 6 In the middle, with the preceding Figure 4 and Figure 5 Similar to the illustration in the previous section, this shows the effect of LLi on the relaxation constant τ. Lithium loss increases the slope of the increase in the relaxation time constant τ. If the relaxation constant τ is now determined at multiple points during discharge, the extent of lithium loss can be deduced from the slope of the change in the time constant. The relaxation time constant τ of the newly detected graphite electrode is again shown as a dashed line, while that of the aged graphite electrode is shown as a solid line. Due to aging, the distance between values ​​within the group decreases, making the group appear as if it has been pushed together.

[0040] Therefore, three attached Figure 4 , Figure 5 and Figure 6The mechanisms shown can be easily and automatically identified in the modeling data based on measured values. Here, all three mechanisms are usually superimposed on each other, but this does not cause interference because the three aging mechanisms have very different effects on the relaxation time constant τ. Therefore, even in the case of their superposition, these mechanisms can be easily and effectively read from the existing data.

[0041] Once in Figure 1 In the process flow, once a strong current consumption is detected (box 100), recording begins 200 of the current value passing through the battery pack or individual battery cells, as well as the voltage value of the individual battery cells. Advantageously, this recording 200 is continuous and exhibits high dynamic range. This can be achieved... Figure 7 The recording is performed in the highly dynamic first buffer 210, schematically shown. After an event suitable for evaluation based on a power pulse and subsequent idle phase is detected, the recorded data is transferred from the first buffer 210 to the second buffer 220 and stored there as a copy 221. Recording 200 then continues in the second buffer 220 with lower dynamics, shown here by region 222. This also allows for the retrospective evaluation of the current pulse and results in lower memory consumption during periods of lower dynamics (relaxation periods).

[0042] Recording continues until a termination criterion is met. This criterion could be, for example, the minimum voltage change per unit time, the minimum charge conversion (only higher conversions are meaningful for non-uniformity), or the shortest time.

[0043] The relaxation voltage and relaxation time constant τ can then be calculated. The relaxation time constant τ can then be entered into a table, for example, relating charge C, state of charge (SOC), or capacity K to the relaxation time constant τ. This can be done for both discharge and charge pulses. Then, using several of these relaxation time constants τ, the behavior of the discharge and charge pulses can be correlated (e.g., using regression analysis) to find the intersection point between the maximum relaxation time constant τ in the discharge and charge directions. Figure 2 This point is the location of the transition point and can be used to evaluate LLi and LAAM. This can be done in stages 3L / 4L, as well as stages 2 and 1, to quantify the development of different aging effects.

[0044] With the help of this quantification, operating limits and operating strategies can now be adjusted in step 700 of the method. For example, charging or discharging current limits related to aging can be adjusted, or the operating strategy can be adjusted to use a medium operating window. It is also conceivable that higher requirements for cooling or heating and corresponding adjustments to the thermal strategy can be derived from the calculated aging variables.

[0045] This method is particularly useful when the data for the entire fleet is aggregated and evaluated on a backend server. In addition to conventional evaluation criteria (capacity, resistance), it can also assess the uniformity of lithium distribution at different stages.

Claims

1. A method for determining the aging of individual cells in a lithium-ion battery pack. Its features are, The battery cells are actively charged or discharged using power pulses, or such power pulses are detected during the operation of the battery pack. During a low-power or no-power idle period following the power pulse, the current pulses and the voltage response of the battery cells are recorded, and the relaxation of the battery cells is evaluated. At least a relaxation time constant (τ) is determined and stored. Then, based on the currently measured and stored relaxation time constant (τ), charging and discharging curves are formed with respect to the charge of the battery cells when they are charged or discharged by the power pulses. The current capacity (K1) of the battery cell is inferred from the intersection of the charging and discharging curves. Based on the current capacity (K1) of the battery cell, the current open-circuit voltage curve (R1) of lithium against graphite with respect to the capacity (K) of the battery cell is calculated from the given open-circuit voltage curve (R0) of the battery cell at the beginning of its lifespan.

2. The method according to claim 1, Its features are, The current capacity (K1) of the battery cell is inferred from the transition from one stage to another on the lithium-to-graphite open-circuit voltage curve with respect to the capacity (K) of the battery cell.

3. The method according to claim 1, Its features are, The charging curve and the discharging curve are formed by a logarithmic plot of the relaxation time constant (τ) with respect to the charge of the battery cell.

4. The method according to claim 1, Its features are, The current open-circuit voltage curve (R1) is determined by compressing the given open-circuit voltage curve (R0) along the capacity (K) direction until the transition between two stages is consistent with the corresponding transition of the current open-circuit voltage curve (R1).

5. The method according to claim 1, Its features are, The current and stored relaxation time constant (τ) is correlated with the current open-circuit voltage curve (R1).

6. The method according to any one of claims 1 to 5, Its features are, The relaxation time constant (τ) is determined directly or indirectly based on the minimum voltage, the minimum charge conversion, the shortest time required for the battery cell to reach a voltage that is substantially constant in time, and / or based on the gradient of the voltage response of the battery cell, by means of a predetermined correlation between the gradient and the relaxation time constant (τ).

7. The method according to any one of claims 1 to 5, Its features are, The currently determined relaxation time constant (τ) is compared with the previously determined and stored relaxation time constant (τ), wherein the increase in the value of the relaxation time constant (τ) indicates that the formation of the capping layer in the battery cell has become more intense.

8. The method according to any one of claims 1 to 5, Its features are, The increase in loss of anode active material (LAAM) is inferred from the shift of the relaxation time constant (τ) toward the lower capacity (K).

9. The method according to any one of claims 1 to 5, Its features are, Based on the fact that the distance between the current relaxation time constant (τ) of the capacity (K) and the last stored relaxation time constant (τ) is smaller than the distance between the previously stored pairs of relaxation time constants (τ), it is inferred that the lithium loss (LLi) is increased.

10. The method according to claim 6, characterized in that, The correlation between the gradient and the relaxation time constant (τ) is given in advance by a table.

11. The method according to claim 8, characterized in that, Based on the shift of the relaxation time constant (τ) toward the lower capacity (K), an increase in the loss of anode active material (LAAM) is inferred from the current open-circuit voltage curve (R1) of lithium to graphite with respect to the cell capacity (K).

12. The method according to claim 9, characterized in that, Based on the fact that the distance between the current relaxation time constant (τ) and the last stored relaxation time constant (τ) for the capacity (K) is smaller than the distance between the previously stored pairs of relaxation time constants (τ), an increase in lithium loss (LLi) is inferred from the current open-circuit voltage curve (R1) of lithium to graphite for the cell capacity (K).

13. A method for controlling a battery pack having at least one battery cell, wherein aging is determined for the battery cell according to any one of claims 1 to 12, characterized in that, Adjust the power supplied to the battery pack during charging and the power drawn from the battery pack during discharging during operation to at least not reinforce the identified aging mechanisms.

14. The method according to claim 13, characterized in that, Adjusting the power supplied to the battery pack during charging and the power drawn from the battery pack during discharging during operation to counteract identified aging mechanisms.

15. The method according to claim 13 or 14, Its features are, To the greatest extent possible, the charging and discharging of each individual battery cell should be avoided from exceeding the phase limits of its open-circuit voltage curve for lithium to graphite with respect to the battery cell capacity.

16. The method according to claim 13 or 14, Its features are, Adjust the thermal management of the battery pack during operation to at least not exacerbate the identified aging mechanisms.

17. The method according to claim 16, Its features are, Adjust the thermal management of the battery pack during operation to counteract the identified aging mechanisms.

Images