Periodic Reporting for period 1 - BatCon (Lithium-ion battery control for faster charging and longer life)
Reporting period: 2020-11-01 to 2022-10-31
Current road transport heavily relying on fossil fuels has caused severe public concerns about energy affordability, air quality, and environmental friendliness. In the foreseeable future, the only possible way to dispel these concerns is via the electrification of vehicle powertrains integrated with renewable and sustainable energy. Fulfilling such a fossil-fuel-free vision greatly depends on the continued advancement of battery technology, which is still the most expensive and perhaps the least understood vehicle component. Although having been recognised as today’s leading commercial energy storage source among various cell chemistries, lithium-ion (Li-ion) battery represents a limiting factor in energy density, “refuelling” time, and cycle life, compared to fossil fuels burned in internal combustion engines. Despite this, available battery systems are compromised even further with conservative designs, with 20-50% underutilised energy capacity, or aggressive de-rating, in order to present safety issues and premature ageing. This in turn considerably reduces the cost-benefit of the battery systems. One way to improve the cost-benefit is through better understanding leading to less conservative safety factors enabled by advanced management systems for safe and optimal use of Li-ion batteries.
The fast charging problem can be roughly described as determining appropriately applied current, which recharges a battery from an initial state of charge (SOC) to a predefined SOC level in the shortest time but does not cause dangerous events like thermal runaway and does not excessively degrade the battery state of health (SOH). In this process, battery electrochemical, thermal, and ageing dynamics will simultaneously evolve and interact with each other over multiple timescales and length scales. The corresponding mathematical structure is overly complicated and in fact, there is no individual model in the literature that is capable of accurately capturing all characteristics of a battery over its lifetime for practical implementation. The lack of accurate prediction battery models for practical use forms one of the biggest technical challenges in the field of advanced charging management. In addition, none of the internal battery states is measurable in onboard vehicle applications through currently available in-situ sensing techniques. This means, to real-time monitor and control a battery’s distributed behaviour associated with chemical reactions and diffusion, one has to develop estimation techniques in the presence of limited and noisy measurements, which are current, voltage, and temperature, at best. Third, as large current rates will be included in the fast charge process, dangers can be easily triggered if the management strategies are not designed meticulously and appropriately. How to always maintain an acceptable safety and health margin while archiving optimal performance is another big challenge that was focused in this project.
The fast charging problem can be roughly described as determining appropriately applied current, which recharges a battery from an initial state of charge (SOC) to a predefined SOC level in the shortest time but does not cause dangerous events like thermal runaway and does not excessively degrade the battery state of health (SOH). In this process, battery electrochemical, thermal, and ageing dynamics will simultaneously evolve and interact with each other over multiple timescales and length scales. The corresponding mathematical structure is overly complicated and in fact, there is no individual model in the literature that is capable of accurately capturing all characteristics of a battery over its lifetime for practical implementation. The lack of accurate prediction battery models for practical use forms one of the biggest technical challenges in the field of advanced charging management. In addition, none of the internal battery states is measurable in onboard vehicle applications through currently available in-situ sensing techniques. This means, to real-time monitor and control a battery’s distributed behaviour associated with chemical reactions and diffusion, one has to develop estimation techniques in the presence of limited and noisy measurements, which are current, voltage, and temperature, at best. Third, as large current rates will be included in the fast charge process, dangers can be easily triggered if the management strategies are not designed meticulously and appropriately. How to always maintain an acceptable safety and health margin while archiving optimal performance is another big challenge that was focused in this project.
In WP1, we have developed a scheme to reformulate and simplify these models into an electrical circuit network that added physical insights into internal battery states, such as overpotentials, electrolyte concentrations, and local temperature. To investigate the generalizability of these model order reduction techniques to the rapidly upgrading battery materials, we have investigated applying these techniques to some new types of batteries for next-generation battery systems, e.g. all solid-state batteries (ASSB), and have achieved the expected high performance. These works have been presented in three journal articles and one conference paper.
In WP2, a general multi-timescale estimation framework has been proposed to enable a rigorous analysis of the effects of timescale separation and then applied to solve the Li-ion battery problem. We develop a robust estimation algorithm for fast timescale states based on the ensemble transform Kalman filter (ETKF). Furthermore, the techniques to estimate the power capability are explored by proposing a novel method to prevent the battery from moving into harmful situations during its operation for its health and safety. The outcomes of WP2 have been presented in two journal articles and one conference paper.
In WP3, we first developed a non-optimisation scheme that considers only a single constraint to be activated at a time. Then, a health- and safety-aware charging protocol is proposed based on successively applied proportional-integral (PI) control actions. The proposed physical constraint-triggered PI charging control strategy is evaluated and compared with several prevalent alternatives. It shows that the derived controller can achieve close to the optimal solution in terms of charging time and trajectory, as determined by a nonlinear model predictive controller, but at a drastically reduced computational cost. We also proposed a novel nonlinear control approach by converting a constrained optimal control problem into an output tracking problem with multiple tracking references. An optimisation-free nonlinear model inversion-based control algorithm is developed for charging the batteries. We demonstrate the efficacy of our method using a spatially discretised high-fidelity P2D model with thermal dynamics. The outcomes of WP3 have been presented in two journal articles and one conference paper.
Furthermore, two relevant special sessions and one tutorial in international conferences are organised to disseminate the works of all three WPs.
In WP2, a general multi-timescale estimation framework has been proposed to enable a rigorous analysis of the effects of timescale separation and then applied to solve the Li-ion battery problem. We develop a robust estimation algorithm for fast timescale states based on the ensemble transform Kalman filter (ETKF). Furthermore, the techniques to estimate the power capability are explored by proposing a novel method to prevent the battery from moving into harmful situations during its operation for its health and safety. The outcomes of WP2 have been presented in two journal articles and one conference paper.
In WP3, we first developed a non-optimisation scheme that considers only a single constraint to be activated at a time. Then, a health- and safety-aware charging protocol is proposed based on successively applied proportional-integral (PI) control actions. The proposed physical constraint-triggered PI charging control strategy is evaluated and compared with several prevalent alternatives. It shows that the derived controller can achieve close to the optimal solution in terms of charging time and trajectory, as determined by a nonlinear model predictive controller, but at a drastically reduced computational cost. We also proposed a novel nonlinear control approach by converting a constrained optimal control problem into an output tracking problem with multiple tracking references. An optimisation-free nonlinear model inversion-based control algorithm is developed for charging the batteries. We demonstrate the efficacy of our method using a spatially discretised high-fidelity P2D model with thermal dynamics. The outcomes of WP3 have been presented in two journal articles and one conference paper.
Furthermore, two relevant special sessions and one tutorial in international conferences are organised to disseminate the works of all three WPs.
The proposed new intelligent management system architecture offers a milestone in Li-ion batteries for speeding up the charging rate and prolonging the cycle life. In addition to EVs, the research outcomes of the project would also benefit another emerging field, i.e. smart grids, through the technologies developed through the research activities on system modelling, estimation, and intelligent control. As an example, we have used the developed health-aware modelling framework to investigate home energy management systems to achieve the least-cost operations with the aim of reducing battery degradation while increasing daily income by feeding solar power into the grid systems. In view of environmental and social impacts, the research outcomes will positively contribute to the energy use and environmental effects of Li-ion batteries. Rapid market penetration of EVs will help to reduce pollution and CO2 emission, particularly within the big cities, and thus contributes to the development of a more sustainable society and the planet.