In multi-cell batteries, because of the larger number of cells used, we can expect that they will be subject to a higher failure rate than single cell batteries. The more cells used, the greater the opportunities to fail and the worse the reliability.
Batteries such as those used for EV and HEV applications are made up from long strings of cells in series in order to achieve higher operating voltages of 200 to 300 Volts or more are particularly vulnerable. The problems can be compounded if parallel packs of cells are required to achieve the desired capacity or power levels. With a battery made up from n cells, the failure rate for the battery will be n times the failure rate of the individual cells.
All cells are not created equal
The potential failure rate is even worse than this however due to the possibility of interactions between the cells. Because of production tolerances, uneven temperature distribution and differences in the ageing characteristics of particular cells, it is possible that individual cells in a series chain could become overstressed leading to premature failure of the cell. During the charging cycle, if there is a degraded cell in the chain with a diminished capacity, there is a danger that once it has reached its full charge it will be subject to overcharging until the rest of the cells in the chain reach their full charge. The result is temperature and pressure build up and possible damage to the cell. With every charge - discharge cycle the weaker cells will get weaker until the battery fails. During discharging, the weakest cell will have the greatest depth of discharge and will tend to fail before the others. It is even possible for the voltage on the weaker cells to be reversed as they become fully discharged before the rest of the cells also resulting in early failure of the cell. Various methods of cell balancing have been developed to address this problem by equalising the stress on the cells.
Unbalanced ageing is less of a problem with parallel chains which tend to be self balancing since the parallel connection holds all the cells at the same voltage and at the same time allows charge to move between cells whether or not an external voltage is applied. There can however be problems with this cell configuration if a short circuit occurs in one of the cells since the rest of the parallel cells will discharge through the failed cell exacerbating the problem.
See Interactions Between Cells for more details.
The problems caused by these cell to cell differences are exaggerated when the cells are subject to the rapid charge and discharge cycles (microcycles) found in HEV applications.
While Lithium batteries are more tolerant of micro cycles they are less tolerant of the problems caused by cell to cell differences.
Because Lead acid and NiMH cells can withstand a level of over-voltage without sustaining permanent damage, a degree of cell balancing or charge equalisation can occur naturally with these technologies simply by prolonging the charging time since the fully charged cells will release energy by gassing until the weaker cells reach their full charge. This is not possible with Lithium cells which can not tolerate over-voltages. Although the problem is reduced with Lead acid NiMH batteries and some other cell chemistries, it is not completely eliminated and solutions must be found for most multicell applications.
Once a cell has failed, the entire battery must be replaced and the consequences are extremely costly. Replacing individual failed cells does not solve the problem since the characteristics of a fresh cell would be quite different from the aged cells in the chain and failure would soon occur once more. Some degree of refurbishment is possible by cannibalising batteries of similar age and usage but it can never achieve the level of cell matching and reliability possible with new cells.
Equalisation is intended to prevent large long term unbalance rather than small short term deviations.
The first approach to solving this problem should be to avoid it if possible through cell selection. Batteries should be constructed from matched cells, preferably from the same manufacturing batch. Testing can be employed to classify and select cells into groups with tighter tolerance spreads to minimise variability within groups.
Large versus small cells
The high energy storage capacities needed for traction and other high power battery applications can be provided by using large high capacity cells or with large numbers of small cells connected in parallel to give the same capacity as the larger cells. In both cases the large cells, or the parallel blocks of small cells, must be connected in series to provide the required high battery voltage.
- Using large cells keeps the interconnections between cells to a minimum allowing simpler monitoring and control electronics and lower assembly costs. Until electric vehicles conquer a substantial percentage of the transportation market, the large cells they need will continue to be made in relatively small quantities, often with semi-automatic or manual production methods, resulting in high costs, wide process variability and the consequent wide performance tolerance spreads. When the cells are used in a serial chain, cell balancing is essential to equalise the stress on the cells, caused by these manufacturing variances, to avoid premature cell failures.
There are also safety issues associated with large capacity cells. A single 200 AmpHour Lithium Cobalt cell typically used in EV applications stores 2,664,000 Joules of energy. If a cell fails or is short circuited or damaged in an accident, this energy is suddenly released, often resulting in an explosion and an intense fire, known euphemistically as an “event” in the battery industry. When such an event occurs in a battery pack there is a strong likelihood that the fire and pressure damage resulting from a cell failure will cause neighbouring cells to fail in a similar way, ultimately affecting all of the cells in the pack with disastrous consequences.
- Using small cells connected in parallel to provide the same voltage and capacity as the larger cells results in many more interconnections, greater assembly costs and possibly more complex control electronics. Small, cylindrical, 2 or 3 AmpHour cells, such as the industry standard 18650 used in consumer electronics applications, are however made in volumes of hundreds of millions per year in much better controlled production facilities without manual intervention on highly automated equipment. The upside is that unit costs are consequently very low and reliability is much higher. When large numbers of cells are connected in a parallel block, the performance of the block will tend towards the process average of the component cells and the self balancing effect will tend to keep it there. The parallel blocks will still need to be connected in series to provide the higher battery voltage but the tolerance spread of the blocks in the series chain will be less than the tolerance spread of the alternative large capacity cells, leaving the cell balancing function with less work to do.
On the safety front, the more reliable low capacity cells are much less likely to fail and if a failure does occur, the stored energy released by any cell is only one hundredth of the energy released by a 200 AmpHour cell. This lower energy release is much easier to contain and the likelihood of the event propagating through the pack is much reduced or eliminated. This is perhaps the most important advantage of designs using lower capacity cells.
See also What a Joule can do
Another important avoidance action is to ensure at all times an even temperature distribution across all cells in the battery. Note that in an EV or HEV passenger car application, the ambient temperature in the engine compartment, the passenger compartment and the boot or trunk can be significantly different and dispersing the cells throughout the vehicle to spread the mechanical load can give rise to unbalanced thermal operating conditions. On the other hand, if the cells are concentrated in one large block, the outer cells in contact with ambient air may run cooler than the inner cells which are surrounded by warmer cells unless steps are taken to provide an air (or other coolant) flow to remove heat from the hotter cells. After cell selection, equalising the temperature across the battery pack should be the first design consideration in order to minimise the need for cell balancing. See also Thermal Management (Uniform heat distribution)
To provide a dynamic solution to this problem which takes into account the ageing and operating conditions of the cells, the BMS may incorporate a Cell Balancing scheme to prevent individual cells from becoming overstressed. These systems monitor the State of Charge (SOC) of each cell, or for less critical, low cost applications, simply the voltage across, each cell in the chain. Switching circuits then control the charge applied to each individual cell in the chain during the charging process to equalise the charge on all the cells in the pack. In automotive applications the system must be designed to cope with the repetitive high energy charging pulses such as those from regenerative braking as well as the normal trickle charging process.
Several Cell Balancing schemes have been proposed and there are trade-offs between the charging times, efficiency losses and the cost of components.
Active cell balancing methods remove charge from one or more high cells and deliver the charge to one or more low cells. Since it is impractical to provide independent charging for all the individual cells simultaneously, the balancing charge must be applied sequentially. Taking into account the charging times for each cell, the equalisation process is also very time consuming with charging times measured in hours. Some active cell balancing schemes are designed to halt the charging of the fully charged cells and continue charging the weaker cells till they reach full charge thus maximising the battery's charge capacity.
- Charge Shuttle (Flying Capacitor) Charge Distribution
With this method a capacitor is switched sequentially across each cell in the series chain. The capacitor averages the charge level on the cells by picking up charge from the cells with higher than average voltage and dumping the charge into cells with lower than average voltage. Alternatively the process can be speeded up by programming the capacitor to repeatedly transfer charge from the highest voltage cell to the lowest voltage cell. Efficiency is reduced as the cell voltage differences are reduced. The method is fairly complex with expensive electronics.
- Inductive Shuttle Charge Distribution
This method uses a transformer with its primary winding connected across the battery and a secondary winding which can be switched across individual cells. It is used to take pulses of energy as required from the full battery, rather than small charge differences from a single cell, to top up the remaining cells. It averages the charge level as with the Flying Capacitor but avoids the problem of small voltage differences in cell voltage and is consequently much faster. This system obviously needs well balanced secondary transformer windings otherwise it will contribute to the problem.
Dissipative techniques find the cells with the highest charge in the pack, indicated by the higher cell voltage, and remove excess energy through a bypass resistor until the voltage or charge matches the voltage on the weaker cells. Some passive balancing schemes stop charging altogether when the first cell is fully charged, then discharge the fully charged cells into a load until they reach the same charge level as the weaker cells. Other schemes are designed continue charging till all the cells are fully charged but to limit the voltage which can be applied to individual cells and to bypass the cells when this voltage has been reached.
This method levels downwards and because it uses low bypass currents, equalisation times are very long. Pack performance determined by the weakest cell and is lossy due to wasted energy in the bypass resistors which could drain the battery if operated continuously. It is however the lowest cost option.
The voltage on all cells levelled upwards to the rated voltage of a good cell. Once the rated voltage on a cell has been reached, the full current bypasses fully charged cells until the weaker cells reach full voltage. This is fast and allows maximum energy storage however it needs expensive high current switches and high power dissipating resistors.
A crude way of protecting the battery from the effects of cell imbalances is to simply switch off the charger when the first cell reaches the voltage which represents its fully charged state (4.2 Volts for most Lithium cells) and to disconnect the battery when the lowest cell voltage reaches its cut off point of 2 Volts during discharging. This will unfortunately terminate the charging before all of the cells have reached their full charge or cut off the power prematurely during discharge leaving unused capacity in the good cells. It thus reduces the effective capacity of the battery. Without the benefits of cell balancing, cycle life could also be reduced, however for well matched cells operating in an even temperature environment, the effect of these compromises could be acceptable.
Recent developments have produced a superior way of cell balancing by means of software control which is both simpler and lossless and avoids the different problems of each of the above methods. See the Software Configurable Battery.
All of these balancing techniques depend on being able to determine the state of charge of the individual cells in the chain. Several methods for determining the state of charge are described on the SOC page.
The simplest of these methods uses the cell voltage as an indication of the state of charge. The main advantage of this method is that it prevents overcharging of individual cells, however it can be prone to error. A cell may reach its cut off voltage before the others in the chain, not because it is fully charged but because its internal impedance is higher than the other cells. In this case the cell will actually have a lower charge than the other cells. It will thus be subject to greater stress during discharge and repeated cycling will eventually provoke failure of the cell.
More precise methods use Coulomb counting and take account of the temperature and age of the cell as well as the cell voltage.
Redox Shuttle (Chemical Cell Balancing)
In Lead acid batteries, overcharging causes gassing which coincidentally balances the cells. The Redox Shuttle is an attempt to provide chemical overcharge protection in Lithium cells using an equivalent method thus avoiding the need for electronic cell balancing. A chemical additive which undergoes reversible chemical action absorbing excess charge above a preset voltage is added to the electrolyte . The chemical reaction is reversed as voltage falls below the preset level.
For batteries with less than 10 cells, where low initial cost is the main objective, or where the cost of replacing a failed battery is not considered prohibitive, cell balancing is sometimes dispensed with altogether and long cycle life is achieved by restricting the permitted DOD. This avoids the cost and complexity of the cell balancing electronics but the trade off is inefficient use of cell capacity.
Whether or not the battery employs cell balancing, it should always incorporate fail safe cell protection circuits.