Tim Engstrom explains how long term battery reliability is determined as much by the battery management system as the cell itself.
Batteries are still something of a mystery even within the automotive industry, with uncertainty around longevity continuing to be a point of nervousness amongst consumers and a source of anxiety for OEMs assessing warranty liabilities. With batteries making up 30-50% of the value of a car, knowing more about what is going on inside these sealed boxes becomes vital rather than a ‘nice to know’.
Currently, significant challenges exist when trying to understand what the present or future state of a battery will be. This creates a trust issue for customers and for fleets who are wary of being stuck with cars that nobody wants to buy due to the battery not performing as advertised.
Part of the issue is the relative lack of sophistication seen in the software of many battery management systems (BMS), which has not kept pace with the rapid improvement in cell constructions and chemistries. Most battery software is designed around keeping within values of current, voltage and temperature that the cell manufacturer has specified from lookup tables. This doesn’t take into account what is actually taking place inside of each battery cell.
We often think of the battery as a component in the car, but really, it’s a system, encompassing a whole universe of mechanics, electronics, chemistry and physics combined. OEMs have traditionally designed around keeping the battery within simple operating parameters specified by the cell partner; but with this simplicity comes the risk of leaving significant performance potential on the table.
Current BMS challenges
State of health (SoH) in batteries is a constant topic of debate, with OEMs placing long warranties, including guarantees on SOH for new vehicles to reassure fleets and private customers. But accurately assessing it is not trivial; we have to interpret the data that the sensors in the pack collect. We find analytically that estimation of SoH from partial charge/discharge data is inaccurate and worsens over time.
As the battery degrades, State of Charge (SoC) error develops as its dependency to the measured open circuit voltage (OCV) becomes corrupted and morphed over time. In most cases, SoC is overestimated in a degraded battery; even at 5% SoC, it can be overestimated by as much as 10%, representing a direct risk to unexpected power loss. In a phone or laptop, such a state of health miscalculation can cause annoyance to the customer, but for a car the consequences could be much greater on the road.
Testing has shown huge sensitivity found between daily usage patterns (i.e. the typical window of utilised SoC) and the accuracy of the reported SoH. Even estimating the health from wide ranges, equivalent to charging to full from 10% SoC, may suffer from more than 5% SOH error. For most BMS systems, changes to the daily operating SoC window has significant sensitivity on reported SoH. We have seen that a driver starting the charge at 50% instead of 30% SoC has the effect of masking the majority of the capacity loss detected by the BMS, which leads to knock-on inaccuracies in range estimation.
The need for intelligent software
Battery Intelligence software, which combines state of the art data science, electrochemical modelling and AI can demystify the true state of the battery, with smarter interpretation of the data already being collected by existing sensors. By blending model-driven and data-driven approaches, it is possible to accurately predict what will happen to the battery state of health in the future, not just what degradation has occurred already.
By understanding and forecasting the future risks with a prognostic, rather than diagnostic approach, it is possible to predict non-linear degradation in a battery. This is termed a “knee point” where the previously latent degradation becomes limiting to capacity. The impact of this is a change in the trajectory of the battery SoH, radically altering the lifespan of the vehicle.
For an OEM with Battery Intelligence software on the BMS, they can take action to change the charging and discharging behaviour of the battery to avoid the knee point that the battery would otherwise be destined for. For a fleet operator using cloud BMS software, it would instead be possible to change charging and usage behaviour, for instance by avoiding keeping the battery fully charged, changing charging strategies, or modifying usage patterns.
For the fleet operator using cloud software this new insight allows them to manage a large part of the value of its asset better. This can allow optimisation of fleet utilisation through allocating vehicles to duty cycles appropriate to their current SoH and compatible with achieving the target lifetime for the vehicle. It can also bring the possibility of financing vehicles over longer periods if battery state of health is going to remain at a level that is appropriate to their usage.
For the OEM, using Battery Intelligence software - both embedded and in the cloud - allows them to understand how batteries are really performing in use and make changes as necessary to improve safety, increase lifespan and unlock more performance, without changing anything physically in the battery design or manufacturing processes.
Tim Engstrom is the Manager – Advanced Battery Technologies at WAE.