By Jody Muelaner
EV production has made the transition from a niche market into true high volume automotive mass market. The battery is the most important element in determining the value of an EV — it largely determines the vehicle’s performance, especially in terms of range, and the battery accounts for around one-third of the vehicle’s cost. Automotive manufacturers are increasingly turning to software to enhance vehicle performance, because this provides an opportunity to scale competitive advantage with minimal increased production cost. The smart connected battery leverages state of the art condition monitoring and AI algorithms to significantly improve battery performance with negligible additional hardware cost.
“The idea of the cloud connected smart battery is basically that you bring the battery to life from the day the cell is born and keep the battery on learning over lifetime by doing basically or by establishing basically a digital twin of the battery in the cloud combine it with machine learning and artificial intelligence and then leveraging basically big data or swarm data not only of that one particular car where the battery is built in no but of a lot of cars of an entire fleet of an entire pool and that really significantly improves the diagnostic features of the battery,” said Jens Hinrichsen, EVP Advanced Analog at NXP.
Smart connected batteries have a digital twin
A digital twin is a virtual representation of a physical product, combining traditional engineering simulation models into a more integrated model of the product, and updating this with sensor data. A digital twin therefore represents best current understanding of the physical product. Digital twins often also use Internet of Things (IoT) sensors and other Industry 4.0 technologies such as big data analytics and artificial intelligence to condition data before the digital twin is updated.
Wild claims are sometimes made about digital twins simulating every process within the lifecycle of a product. However, while complex multi-physics models can simulate interactions between electrochemical reactions, electromagnetic forces, mechanical stress, and thermal loads, these models made simplifying assumptions and run much slower than real-time.
Practical digital twins often model components and sub-systems as black boxes, only considering the inputs and outputs, which are significant. While a traditional block representation of a system may use first-principles analytical equations to map inputs to outputs, a digital twin will fit the equations to observed data. This might mean starting with analytical equations, and then tuning parameters to best fit the observed outputs. Using machine learning, it is also possible to start with an empty black box and let the algorithm learn the relationships without any initial assumptions. The big advantage of a machine learning approach is where large numbers of inputs and outputs, with complex interactions, are combined with large datasets. In such situations, where analysts would find it very difficult to model the interactions from first principles and the resulting models may be very cumbersome and slow, deep learning can efficiently produce accurate models with minimal human input.
Conventional simulation and modelling tools have evolved into digital twins along two dimensions. One dimension is increasing integration of multiple models, and the other dimension is increasing interaction between digital models and the real world. Multiphysics models integrate simulation models of different physical domains, predictive maintenance updates models of component failure using condition monitoring data from the real world. Digital twins combine both of these technologies.
Smart connected batteries have sensors that monitor key parameters at the cell level, such as charge/discharge rates, charge state, and temperature. This enables the digital twin for an individual EV to maintain the best possible understanding of the current performance of the battery and its expected life.
Unleashing the potential of the smart connected battery
When data is aggregated across all the vehicles produced by a manufacturer, the real potential of the smart connected battery is unleashed. The data can be used to refine the digital twin model, allowing it to become an extremely high fidelity representation of current and future performance.
Aggregated data combined with this highly refined digital twin then also provides an understanding of the impact of different vehicle usage patterns, battery management control updates and even different battery chemistries. With it being possible to apply software updates remotely across a fleet of vehicles, improved understanding of battery management can enhance the performance and life of vehicles already in service. It would even be possible to run experiments, with different battery management updates applied to some vehicles and the effects quantified.
When battery management is better understood, it not only allows the life of the battery to be improved, it is also possible to squeeze more performance out of the same battery without damaging it. This can mean increased range, lower weight, and faster acceleration.
“The more precise data I can provide the more you can leverage and unleash the full power out of the battery without damaging it. Also… you can define the residual value of the particular battery,” said Hui Zhang, Group Vice President, NIO.
The range of uses for battery digital twins include:
- State-of-X estimation and cell balancing: State-of-charge, state-of-health and state-of-power are important parameters used by the battery management system (BMS). Users need to know the state-of-charge as it indicates the range available before the EV needs to be charged. The state-of-power sets safe limits for power during acceleration or regenerative braking.
- Fault diagnosis and prognosis: Although the onboard BMS provides basic detection for common faults such as over-discharge, over-charge and short-circuit, this is typically a simple comparison of individual variables with threshold values. A digital twin can provide much more advanced fault diagnosis and prediction, using multi-physics models to monitor the mechanisms and processes that may lead to faults.
- Remaining useful life (RUL) estimation: While state-of-health only provides the current state of the battery, RUL predicts future degradation of the battery. RUL is typically expressed as the number of remaining charge/discharge cycles.
- Predictive maintenance: While EVs require regular service and maintenance, fixed schedule maintenance is inefficient since it generally results in work being carried out either before it is actually needed, or after expensive damage has already occurred. Good predictive maintenance caries out the work optimally when it is actually needed to reduce the risk of failure. Battery maintenance costs have been show to reduce by up to 62% when predictive maintenance is used.
- Battery repurposing (second-life): A major cost in repurposing EV batteries for second-life use is the need to disassemble and test each cell. If state-of-health and RUL estimations are already available at the cell level, the cost of repurposing can be significantly reduced.
- Battery sharing/swapping services: If the state-of-health and RUL for a battery are known, this facilitates the exchange of batteries, without concerns about loss of value during the exchange.
- Energy optimization: The improved optimization methods that digital twins make available to the energy management system can improve range, battery life and acceleration.
- Vehicle-to-grid (V2G) operation: V2G could support decarbonization by making EV batteries available to buffer the supply and demand of power in a variable renewables intensive grid. Smart connected batteries can enable V2G with health-aware charging.
“You can really significantly improve the performance of the car and therefore basically the range and while you leverage the maximum power out of the battery you do this in a very gentle way without damaging the battery you increase also longevity of the battery you can really define and also protect the residual value of the battery at any point of time in the life cycle,” saide Jens Hinrichsen, EVP Advanced Analog at NXP.
Fully leveraging the potential of the smart connected battery requires a high degree of collaboration across the manufacturing supply chain and vehicle operators.
Conclusions
Smart connected batteries combine condition monitoring at the cell level with cloud based digital models of the battery, to provide a wealth of benefits. As data is shared between digital twins for the batteries in millions of vehicles, very high levels of confidence in the predictive capability of the models can be assured. Each vehicle therefore will have a clear understanding of the residual value of its battery. At the fleet level, this enhanced understanding will enable greatly improved battery performance and life.
You may also like:
Filed Under: Batteries