Alumni
- Fink, MichaelHide
- Hahn, MarkusHide
-
Title of thesis:
2022
Supervising Chair:
Electrifying the energy market and the transport sector based on renewable energies is indispensable to meet the objective of the Paris Agreement: Limiting the human-caused climate change to 1,5 °C. In both sectors, efficient energy storage at a low cost is required to fulfill the demand for mobility and to overcome the volatility in power generation. For this purpose, lithium-ion batteries are increasingly used in manifold applications. Alongside many advantages, this technology also comes with challenges. One of those is the limitation of its fast-charging capability, which defines the time required to recharge a vehicle in analogy to the re-fueling of a combustion vehicle. This thesis aims to develop methods and tools in the field of engineering to precisely determine those limitations and to extend those limits through design modifications.
After discussing the structure and the working principle of lithium-ion batteries, the characterization method Distribution of Relaxation Times is described, refined and the impact of its meta parameters is analyzed. The method allows for the separation, identification, and quantification of processes in arbitrary electrochemical systems without any a priori knowledge, assumptions, or models and is based on standard measurement procedures. The characterization of lithium-ion batteries is typically used to develop and parameterize battery models. These models can describe and predict the behavior at various load conditions and can furthermore be used for fast-charging algorithms. However, the state-of-the-art models do not meet the requirements of a charge controller based on local states inside the battery. Furthermore, zero-dimensional models which lack spatial information are not capable of characterizing and quantifying the behavior of porous electrodes. Therefore, the aim of this thesis is to adapt a frequency domain model originating from energy transmission and communications engineering as an electrode model for use in the time domain with a real-time capable solution. Finally, the model shall be applied for fast charging.
This transmission line model characterizes single electrodes and enables a spatial resolution alongside the thickness of the electrode. As a mixed conducting network, the discrete electrochemical model structure characterizes ionic as well as electronic transport processes. Based on the processes identified by the distribution of relaxation times, the model structure is developed using concentrated electrical network elements. The model is described mathematically and implemented for the time and frequency domain, respectively. Spatial discretization is introduced to allow for the transformation into the time domain. The model structure leads to a classification between zero-dimensional, phenomenological models and physical-chemical models.
The frequency domain variant can be calculated and solved iteratively with an analytic transfer function. In contrast, the time domain variant is solved by a sophisticated numerical algorithm since the resulting differential-algebraic equation system demands high stability of the solver. Through an efficient implementation, the model can be executed on a real-time system at a 10 ms cycle. The parameterization is carried out in the frequency domain using electrochemical impedance spectra of half-cells made from commercial electrode material which is brought into a three-electrode experimental cell setup. The achieved model precision and parameters are strongly dependent on the discretization in the frequency domain. For a well-interpretable result with a low deviation between measurement and model, at least several hundred discrete elements are required. Caused by the small number of model parameters, the parameterization is efficient and unambiguous. In the time domain, validation measurements show a high agreement with the model for static and dynamic load profiles at a significantly coarser discretization. Several model extensions, developed by students during their theses, demonstrate a further reduction of the computational effort as well as combinations of an anode and a cathodemodel to a full-cell model.
The frequency domain model is applied for the analysis of the electrode design of the anode. Besides the effect of varied model parameters on the impedance, various design properties are investigated through simulations. These include the electrode thickness, the porosity, and the active material particle size. Summarizing the findings, the commercial electrode is designed well considering its purpose in a high-energy battery. Furthermore, the model is proven to be capable of characterizing varying electrode properties. Therefore, it is suitable for the efficient model-based design and optimization of electrodes. This includes identifying rate-limiting factors during charging, allowing design changes to improve the fast-charging capability.
Finally, the time-domain model variant is used for model-predictive fast charging control using different cost functions. One of the proposed methods controls the potential at the electrode surface to a pre-defined value to avoid damage to the electrode by lithium deposition. The charging capability can be fully used without causing significant aging. A safety margin, which is used in state-of-the-art algorithms, can be omitted as local, internal states are used instead of the clamp behavior. A second approach allows for a user-defined trade-off between an even shorter charging time and the induced, increased aging. This allows for an adequate charging of electric vehicles at a minimum duration. The charge algorithm is proven stable against parameter uncertainties. State estimation is required for the application in a battery management system. At the time of submission related research is ongoing.
The proposed model copes with the requirements regarding fast charging for both the electrode development process and the battery operation. This is an enhancement compared to the state of the art. Unlike established models, the approach proposed in this work combines distinct parameterability and real-time capability based on the electrical network structure with spatially resolved, electrochemically-physically well-interpretable states inside the electrode. However, the model is not limited to the mentioned applications. Subsequent research based on this thesis is working on including a thermal sub-model. Detailed aging studies can be carried out, resulting in an aging sub-model. Furthermore, the model will be used and adapted to evaluate novel, solid electrolytes and their application within porous electrodes. - Karg, AndreasHide
- Katzer, FelixHide
- Koller, JanHide
- Michlik, TobiasHide
- Nazarenus, TobiasHide
- Rosenbach, DominicHide
- Rüther, TomHide
-
Title of thesis:
Elektrochemische Charakterisierung und systemtechnische Diagnose von Lithium-Ionen-Batterien
2025
Supervising Chair:
In light of the growing prevalence of lithium-ion batteries, it is imperative to investigate alternative approaches to recycling and to enhance the longevity and performance of these batteries. To develop appropriate solutions, it's essential to have a thorough understanding of the electrical characteristics of both individual cells and modules. For this reason, a detailed method for process characterization, the Löwner-method, is used in this thesis. It is validated through the utilization of both equivalent circuit models and an established method for process characterization. It offers a promising and innovative approach for the analysis of a wide variety of electrochemical systems. Moreover, the Löwner-method is integrated with alternative techniques to enable a thorough characterization of 92 lithium-ion batteries, surpassing the limitations of traditional approaches like equivalent circuit parameter analysis or direct variable measurement.
The identified processes are examined for their distribution function, where the general assumption of a normal distribution can be declined for some processes. In addition, a correlation analysis reveals three discrete process correlation groups that are assigned to cell winding, surface processes, and diffusion processes. The results of the cell-to-cell-variation study are then used to model the energy and pulse power capability of battery modules in series and parallel connection. To account for parameter variation and correlation, multivariate normal distributions are used. A simulation is then employed to examine the impact of cell sorting and an inhomogeneously aged cell on the performance variables of energy and pulse power capability. The results show that cell sorting, despite its significant sorting effort, only marginally enhances performance.
In contrast, the performance exhibits a significant dependence on the inhomogeneity, particularly in the serial connection. Consequently, a method for detecting inhomogeneities in serially connected battery modules is being developed. For this purpose, the impedance of the entire module is simulated and analyzed using various impedance characteristics. It turns out that the low-frequency minimum is the most suitable feature for this purpose. It is therefore validated by measurements. This feature enables the detection of inhomogeneities in a series connection of up to ten cells.
This work therefore makes a significant contribution to the understanding of the electrical properties of battery modules and their performance. Moreover, it serves as a foundation for the future development of decision support algorithms in the field of circular economy. - Schadeck, UlrichHide
- Weiß, SebastianHide
- Zander, JudithHide