Battery-integrated chargers (BIC) are essential for achieving fast charging in electric vehicles, and their development has therefore become a major focus of current research. The present thesis deals with sensitivity analysis (SA) of BIC. More precisely, the operational parameters defining a dynamic state-of-charge (SoC), temperature and recharge power control strategy of the internal battery shall be subjected to the investigation on their impact on the system performance, which is quantified by three different key performance indicators (KPI). The superior objective is the dimensional reduction of the input space for the subsequent parameter optimisation by only processing the most influential parameters identified by the SA. For this purpose four different SA methods were selected to gather the results generated by various approaches. Thus, a bivariate and a multivariate correlation analysis as well as a derivative-based and a derivative-variance hybrid approach were chosen to yield the desired factor prioritisation. The concrete procedures for the firstly mentioned method were developed by Pearson, Spearman and Kendall, the multivariate method is represented by the Canonical Correlation Analysis, while the Elementary Effect Test (EET) was selected as a derivative-based approach. Furthermore, the Derivative-Variance Hybrid Global Sensitivity Analysis was applied. The input-output data of the BIC’s simulation model were generated by two different sampling strategies. While an adapted version of Latin Hypercube Sampling was used for the correlation-based approaches, Morris Sampling as a tailored sampling strategy to the EET was used for the corresponding SA methods. Global and local SA determine similar rankings, but yield a different number of significantly important parameters, highlighting the impact of input space boundaries. Moreover, the influence of the set simulation duration as well as the influence of the applied load profile (LP) on the resulting KPI is subject to further investigations. The outcome thereby shows a moderate impact of the simulation duration, leading to some ranking shifts among closely ranked items while maintaining the principal relative magnitude of the importance measures. The SA’s reaction to different (extreme) LP infers a more varying factor prioritisation, resulting in the high priority of having a specific characteristic LP available. In conclusion, as a conservative recommendation eight operation parameters are proposed for further processing, represented by timestamps for changing between operation modes, SoC-levels and the parameters handling the thermal management during idle periods.
Optimisation of battery-integrated chargers -Selection of key operational parameters
Winklehner, X. (Author). 2025
Student thesis: Master's Thesis