• Energy Research

Lithium-ion batteries are commonly employed in various applications owing to high energy density and long service life. Lithium-ion battery models are used for analysing batteries and enabling power control in applications. The Doyle-Fuller-Newman (DFN) model is a popular electrochemistry-based lithium-ion battery model which represents solid-state and electrolyte diffusion dynamics and accurately predicts the current/voltage response. However, implementation of the full DFN model requires significant computation time. This paper proposes a computationally efficient implementation of the full DFN battery model, which is convenient for real-time applications. The proposed implementation is based on spatial and temporal discretisation of the governing partial differential equations and a particular numerical method for solving the resulting discretised model equations, which is based on a damped Newton's method. In a simulation study, the numerical efficiency of the proposed implementation is shown.

Li-ion is the most commonly used battery chemistry in portable applications nowadays. Accurate state-of-charge sOC and remaining run-time indication for portable devices is important for the user’s convenience and to prolong the lifetime of batteries. A new SOC indication system, combining the electromotive force EMF measurement during equilibrium and current measurement and integration during charge and discharge, has been developed and implemented in a laboratory setup. During discharge, apart from simple Coulomb counting, the effect of the overpotential is also considered. Mathematical models describing the EMF and the overpotential functions for a Li-ion battery have been developed. These models include a variety of parameters whose values depend on the determination method and experimental conditions. In this paper the battery measurement and modeling efforts are described. The method of implementing the battery model in an SOC indication system is also described. The aim is an SOC determination within 1% inaccuracy or better under all realistic user conditions, including a wide variety of load currents and a wide temperature range. The achieved results show the effectiveness of our novel approach for improving the accuracy of the SOC indication.

Li-ion is currently the most commonly used battery chemistry in portable applications. Accurate state-of-charge (SOC) and remaining run-time indication for portable devices is important for user convenience and to prolong the lifetime of batteries. The actual SOC algorithms, which the main companies use in practice, make use of the so-called electromotive force (emf). In these SOC systems it is assumed that the emf of an Li-ion battery only depends on aging to a limited extent. In this paper, novel emf measurement and modeling efforts are presented as a function of battery aging. As will be shown, a better understanding of this dependence is useful for improving SOC accuracy.

Using electrochemistry-based battery models in battery management systems remains challenging due to their computational complexity. In this paper, we study for the first time the impact of several types of model simplifications on the trade-off between model accuracy and computation time for the Doyle–Fuller–Newman (DFN) model. As a basis for comparison, we consider, to what we refer as, the complete DFN (CDFN) model, which is a DFN model without any simplifications, and includes the concentration-dependency of parameters that have been studied in previous literature. Furthermore, we propose a highly efficient implementation of the CDFN model that leads to a considerable decrease in computation time, and is developed into a freely downloadable toolbox. This toolbox allows the user to easily toggle between the studied simplifications to make the desired trade-off between model accuracy and computation time. We compare several simplified DFN models to the single-particle-model and the CDFN model. Here, we show that with the proposed implementation, and by selectively making the proposed simplifications, as well as selectively choosing the grid parameters, a model can be obtained that has a minor impact on model accuracy, achieving a simulation time of over 5000 times faster than real-time.

Electric vehicle battery performance near end-of-life is limited by mismatched cell degradation, leading to an estimated 5-10% cell capacity variation across the pack. Active cell balancing hardware architectures incorporating a Low-Voltage (LV) bus supply have been introduced to unlock lost capacity due to cell imbalance at reduced cost, through elimination of the vehicle's 400-to-12V dc-dc converter. In this work, a hierarchical model-predictive control scheme is applied to a time-shared isolated converter active balancing architecture that incorporates LV bus supply. The proposed controller efficiently divides computation among the battery management system hardware components. The energy-buffering capability of the lead-acid battery, which is connected to the LV bus, is used to trade-off balancing and bus regulation objectives, reducing peak power and improving the system cost-effectiveness. Simultaneous state-of-charge balancing and LV bus regulation is verified in simulation and experiment using real-world drive and LV load data collected from a GM Bolt electric vehicle. Similar controller performance compared to a central scheme is achieved in simulation. The experimental setup includes a custom 12S2P, 3.9kWh, liquid-cooled Lithium Nickel Manganese Cobalt battery module with embedded battery management system. The controller performance is evaluated with an initial maximum state-of-charge imbalance of 6.8%.

Based on the concept of a defined sealed rechargeable NiCd battery, the mathematics of the various electrochemical and physical processes occurring inside the battery are described. Subsequently, these sets of mathematical equations are clustered and converted into an electronic network model. Introducing the relevant electrochemical and physical parameters, the one-dimensional model is shown to be capable of simulating not only the development of the cell voltage during (over)charging and (over)discharging, but also of simultaneously calculating the development of the internal gas pressure. Considering the thermal dependencies of the various electrochemical reactions and those of the battery environment, the temperature development and the mutual interaction with the voltage and gas pressure can also be calculated. Since the electronic network approach gives access to all partial currents flowing through the different reaction paths inside the battery, it is easy to visualize what processes are occurring during battery operation. This is, for example, illustrated for the two-step overdischarge process, indicating that, respectively, the Cd and O2 charge-transfer reactions play a dominant role under these conditions. Electronic network simulations are shown to be not only restricted to direct current applications but are also applicable to processes, like open-circuit voltage relaxation and self-discharge behavior.

In order to guarantee safe and proper use of Lithium-ion batteries during operation, an accurate estimate of the battery temperature is of paramount importance. Electrochemical Impedance Spectroscopy (EIS) can be used to estimate the battery temperature and several EIS-based temperature estimation methods have been proposed in the literature. In this paper, we argue that all existing EIS-based methods implicitly distinguish two steps: experiment design and parameter estimation. The former step consists of choosing the excitation frequency and the latter step consists of estimating the battery temperature based on the measured impedance resulting from the chosen excitation. By distinguishing these steps and by performing Monte-Carlo simulations, all existing methods are compared in terms of accuracy (i.e., mean-square error) of the temperature estimate. The results of the comparison show that, due to different choices in the two steps, significant differences in accuracy of the estimate exist. More importantly, by jointly selecting the parameters of the experiment-design and parameter-estimation step, a more-accurate temperature estimate can be obtained. In case of an unknown State-of-Charge, this novel method estimates the temperature with an average absolute bias of 0.4 degrees Celsius and an average standard deviation of 0.7 degrees Celsius using a single impedance measurement for the battery under consideration.