Elektrische Batteriespeichermodelle : Modellbildung, Parameteridentifikation und Modellreduktion

  • Electrical battery models : modelling, parameter identification and model reduction

Witzenhausen, Heiko; Sauer, Dirk Uwe (Thesis advisor); Kowal, Julia (Thesis advisor)

1. Auflage. - Aachen : ISEA (2017)
Book, Dissertation / PhD Thesis

In: Aachener Beiträge des ISEA 90
Page(s)/Article-Nr.: 1 Online-Ressource (xi, 266 Seiten) : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2017


This work deals with the electrical behavior of lithium-ion batteries.The aim is to gain as much information as possible by only monitoring the current and voltage behavior of a battery without being able to look inside. For this, a model representation of the processes inside a lithium-ion battery is developed. For all important phenomena, the influence on the electrical voltage response is derived and merged into a mathematical framework. For parameter identification based solely on the voltage curve, these model equations must be simplified in order to obtain the necessary computation speed for a parameter estimation algorithm. The parameters whose sole influences on the electric behaviour can not be separated without changing the system are summarized so that a model with purely electrical quantities is derived. Using different methods, the model order is significantly reduced while minimizing the approximation error. The model thus transformed, based on the model equations of porous electrodes according to Newman, can even be used for voltage prognosis or for parameter tracking in onboard diagnostics. Furthermore, a method is presented with which the individual processes can be quantitatively measured and separated without destroying the battery cell. In total, more than sixteen different types of batteries were extensively measured within the scope of the work. On the basis of the findings derived from this, the development of the parameterization procedure was carried out. Thus, it is now possible to separate the individual processes occurring in the battery and to support a hypothesis of the assignment to positive and negative electrodes on the basis of several indicia. For this purpose, the alternating current electrochemical impedance spectroscopy (EIS) and a correlation analysis are used to calculate the distribution function of the time constants (DRT) occurring in the impedance. Together with a series of tests with direct current excitation, a measuring method is obtained which, with the correct evaluation methodology, allows a broad characterization of the battery behavior. Finally, the described method is used in two studies. In the first, the possibilities of a current-voltage model derived from the measurements are demonstrated by fully parameterizing the battery cell of the first generation Mitsubishi iMiEV. The overvoltages of both electrodes can be viewed separately and the effects of so-called blend electrodes are also visible. The appendix of the work additionally includes the parameters of the cell of the first generation BMW i3. The second study examines the changes in the model parameters during the cyclic aging of batteries. Further information on the assignment of the processes to the two electrodes is derived from this data. This precise knowledge of the composition of the voltage and the assignment of the overvoltages to the individual electrodes can in the future allow a completely new control of the operation of a battery. The prognosis quality of the condition and the remaining lifetime of the system will be improved significantly by applying the presented methods.