Physics-based battery model parametrisation from impedance data
Noël Hallemans, Nicola E. Courtier, Colin P. Please, Brady Planden, Rishit Dhoot, Robert Timms, S. Jon Chapman, David Howey, Stephen R. Duncan · 2024
Read Physics-based battery model parametrisation from impedance data by Noël Hallemans, Nicola E. Courtier, Colin P. Please, Brady Planden, Rishit Dhoot, Robert Timms, S. Jon Chapman, David Howey, Stephen R. Duncan (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-10896.txt.
Introduction
Noël Hallemans and a team of co-authors present a 2024 paper on setting up physics-based battery models from impedance data. Such models can guide battery design, but their many internal values are hard to measure directly. Impedance tests probe a cell with small signals and reveal signs of its inner processes. The team built free software that computes model impedance quickly using automatic differentiation.
They compared three model types and picked the one that kept the key features of real lithium-ion data with the fewest parts. They then grouped its values and fit eighteen of them to both simulated and measured data from a real cell. Values that shape the response could be pinned down and assigned to the right electrode. The authors stress that fitting needs data across many charge levels at once. Their method offers a practical, non-invasive route to model setup.
Catalog record
| CREATOR | Noël Hallemans, Nicola E. Courtier, Colin P. Please, Brady Planden, Rishit Dhoot, Robert Timms, S. Jon Chapman, David Howey, Stephen R. Duncan |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-10896
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Physics-based battery model parametrisation from impedance data Noël Hallemans, Nicola E. Courtier, Colin P. Please, Brady Planden, Rishit Dhoot, Robert Timms, S. Jon Chapman, David Howey, Stephen R. Duncan Abstract Non-invasive parametrisation of physics-based battery models can be performed by fitting the model to electrochemical impedance spectroscopy (EIS) data containing features related to the different physical processes. However, this requires an impedance model to be derived, which may be complex to obtain analytically. We have developed the open-source software PyBaMM-EIS that provides a fast method to compute the impedance of any PyBaMM model at any operating point using automatic differentiation. Using PyBaMM-EIS, we investigate the impedance of the single particle model, single particle model with electrolyte (SPMe), and Doyle-Fuller-Newman model, and identify the SPMe as a parsimonious option that shows the typical features of measured lithium-ion cell impedance data. We provide a grouped-parameter SPMe and analyse the features in the impedance related to each parameter. Using the open-source software PyBOP, we estimate 18 grouped parameters both from simulated impedance data and from measured impedance data from a LG M50LT lithium-ion battery. The parameters that directly affect the response of the SPMe can be accurately determined and assigned to the correct electrode. Crucially, parameter fitting must be done simultaneously to measurements across a wide range of states-of-charge. Overall, this work presents a practical way to find the parameters of physics-based models. Characterisation techniques are essential for understanding physical processes in batteries and monitoring their state. In many applications, such as health estimation in battery management systems (BMS) and recycling facilities, characterisation should be non-invasive and based on current and voltage data. Characterising a battery is often performed by parametrising a model (a relation between terminal voltage and applied current) from measured data, which can then be used for physical interpretation and simulation of the battery. The choice of the model may be challenging. Often ad hoc equivalent circuit models [ 1 ] are used to fit measured data [ 2 , 3 ] , but these have the disadvantage that the parameters are not always physically meaningful and simulation may be inaccurate. Instead, we choose to parametrise physics-based models [ 4 , 5 ] . The standard physics-based modelling approach for lithium-ion batteries is the Doyle-Fuller-Newman (DFN) framework [ 6 , 7 ] , based on porous-electrode theory, consisting of a set of coupled nonlinear partial differential equations that are typically too expensive to simulate in a BMS. There also exist reduced-order models, derived from the DFN model, such as the single-particle model (SPM) [ 8 ] , single-particle model with electrolyte (SPMe) [ 9 , 10 , 11 ] , or multi-particle model [ 12 ] , all of which may be more practical for BMS. Model parameters are only identifiable from measured datasets that are sufficiently informative, containing information at the relevant time-scales of the physical processes in the model—which may be orders of magnitude apart. One approach uses constant-current or pulse data during cycling to fit the model to the time-domain voltage response, which we refer to as the “time-domain voltage method” [ 13 ] . However, this is often not sufficiently informative to fully parametrise a physics-based model as the excitation may not excite all time-scales and requires large datasets. Electrochemical impedance spectroscopy (EIS) is therefore often used as a complementary source of data, providing compressed information over a wide range of time-scales (or, equivalently, frequencies) [ 4 , 14 , 15 , 16 , 17 ] . Conventional EIS measurements [ 18 , 19 , 20 ] consist of taking an equilibrated battery at a fixed operating point (state-of-charge (SOC) and temperature) and applying a small sinusoidal current (or voltage) at set amplitude and frequency. The resulting voltage (or current) is measured and the Fourier component of this, at the same frequency as the excitation, is used to calculate the impedance. The process is then repeated at several frequencies. Measuring impedance over a wide frequency range allows physical processes occurring at different time-scales to be distinguished, and including impedance data for model parametrisation has been shown to improve the identifiability of parameters [ 21 ] . The frequencies are typically spaced logarithmically to give appropriate weights to the different physical processes in the impedance data. To parametrise a physics-based model from impedance data, we need to be able to compute the impedance response of the model. There are two basic approaches, one that solves the problem in the time-domain and the other in the frequency-domain. In the first method, the impedance can be computed directly in the time-domain (the “brute-force" approach) by simulating the conventional EIS measurement procedure described above on a model. The advantages of this are that it is easy to implement, the simulation is true to the practical method, and additional information can be extracted, if required. For example, the higher harmonics can be examined for use in nonlinear EIS (NLEIS) [ 22 , 23 , 24 ] . The main drawback of the brute-force method is relatively high computational cost. In the second approach, the model is first linearised about an operating point before being transformed into the frequency-domain, which can sometimes be performed analytically from the model equations. Early examples of this approach include the impedance calculations in Lasia et al. [ 25 ] and Meyers et al. [ 26 ] , and the impedance mode of the classical pseudo two-dimensional Dualfoil code [ 27 ] . More recently, Song and Bazant analytically calculated the diffusion impedance of battery electrodes for different particle geometries [ 28 ] , Bizeray et al. calculated the impedance of the full SPM [ 29 ] , and Plett and Trimboli studied the impedance of the DFN [ 4 ] . However, practically extending this to more complicated models is challenging, so numerical approaches are adopted on the discretised model in space (usually using finite differences or finite volumes). One method, referred to as numerical differentiation, uses a finite difference approximation for the derivatives arising from the linearisation, such as in Zhu et al. [ 30 ] . Alternatively, here we create exact expressions for the derivatives using automatic differentiation , leveraging the open-source battery simulation software PyBaMM [ 31 ] , which stores models as analytical equations. Automatic differentiation—implemented in JAX [ 32 ] , for instance—computes derivatives by applying the chain rule to the algebraic operations and analytical functions within the model, and has been used previously on specific models [ 33 , 34 ] . Using these expressions, the impedance of the model can be computed numerically at any operating point at a set of frequencies, and we have developed an open-source Python package, PyBaMM-EIS , that enables simulation of EIS with any PyBaMM model. This “numerical frequency-domain” impedance method has a much lower computational cost than the brute-force method and is therefore suitable for model parametrisation. Using PyBaMM-EIS , we can easily compare the impedance of many commonly used models (SPM, SPMe, and DFN). We found that including electrolyte dynamics adds a “diffusion bump” impedance feature, often present in measured battery data. We study the impedance of the SPMe in more detail and provide a grouped-parameter model to explain how its impedance