Due to rising concerns of environmental issues and increasing greenhouse gas (GHG) emissions, the automotive industry is transitioning to more environmental and efficient driveline technologies. Multiple electric vehicles (EV) have been incorporated in the automotive market, encompassing the hybrid electric vehicle (HEV), plug-in electric vehicle (PEV) which comprises the plug-in hybrid electric vehicle (PHEV), extended range electric vehicle (E-REV) and battery electric vehicle (BEV). The high popularity of these vehicles is due to their low emissions and efficient manipulation of their well to wheel energy. These advantages can be achieved due to the battery packs that are included in the vehicles composed of rechargeable energy storage sources (RESS) which can provide electrical power to the electric motor. Different RESS technologies exist (lead-acid, nickel cadmium and nickel metal hydride batteries) but the energy and power that can provided by these cannot surpass lithium-ion batteries. To be able to incorporate lithium-ion batteries in these vehicles the proper characterization, modelling and sizing of these systems has to be performed. This is why academic and industrial research has been concentrated to evaluate lithium-ion batteries. In this PhD dissertation the in depth evaluation of different lithium-ion chemistries, modelling of the battery cells and their usage in a battery pack configuration is conducted.
An extended state of the art of the methodologies to simulate batteries and battery packs composed of lithium-ion batteries is initially performed. Lithium-ion batteries are electrochemical systems that have chemical, electrical, thermal and mechanical physical properties. Based on these properties different models can be established to investigate the functioning of the batteries under different environmental and operational conditions. Special consideration is given in this research work to investigate the electrical behaviour of lithium-ion batteries through the use of equivalent circuit models (ECM). Furthermore an overview of the characteristics and functions of battery packs and battery management systems (BMS) existing in current EV applications is investigated. Battery packs are systems that are incorporated with multiple lithium-ion batteries to meet the energy and power demand of EV applications. For these to function correctly the BMS has to be accompanied with reliable modelling tools and functions. Three different modelling methodologies have been identified which can provide this. They are categorized in i) aggregating a battery cell- level model ii) scaling up a battery cell-level model also called Generic Battery (GB) pack model and iii) simulating the battery pack as one complete system.
To acquire the operational characteristics of lithium-ion batteries a dedicated characterization methodology has been performed under multiple temperatures (5°C, 5°C, 15°C, 25°C, 35°C and 45°C), current rates (It/5, It/2, 1It, 1.5It, 2It and 2.5It) and load conditions. The lithium-ion chemistries that have been investigated are the lithium-ion nickel manganese cobalt oxide (NMC), the lithium-ion phosphate (LFP) and the lithium-ion titanate oxide (LTO). Based on the characterization procedures that have been established through international standards and scientific publications, new procedures called the extended advanced characterization (EAC) has been developed. Lithium-ion batteries have been identified as nonlinear systems exhibiting transient behaviours especially during the high (100% to 90%) and low (10% to 0%) state of charge (SoC) levels. Thus the main goal of the EAC procedure is to be able to acquire the battery characteristics for the complete operational range of 100% to 0% state of charge (SoC) with a special focus at the nonlinear controversial SoC ranges. This provides the possibility to employ additional energy from the batteries which can extend the range of an EV application. Through the EAC procedure the available capacity, specific energy, open circuit voltage and specific power characteristics exhibited by the batteries is acquired. The performed investigation showed that the NMC chemistry can be established as an energy based battery, the LTO chemistry as a power based battery and LFP as mid-range energy and power lithium-ion battery.
In the framework of this PhD dissertation particular attention has been given to the acquisition of accurate SoC estimation of lithium-ion batteries. The SoC estimation is a key parameter in the accuracy that can be achieved during the electrical modelling of these systems. Moreover the SoC plays an important role in the correct management of the electrical (energy balancing) and degradation mechanisms (capacity fading) that has to be performed by the BMS. A state of the art of the exist SoC estimation techniques found in the scientific literature are conducted to establish which can be employed in the electrical models developed. Two different SoC estimation techniques have been established, the Extended Coulomb Counting (ECC) providing fast calculations at the cost of accuracy and the Extended Kalman Filter (EKF) providing accurate results at the cost of computational effort.
Based on the EAC procedure and the developed SoC estimation techniques the electrical behaviour of lithium-ion batteries can be simulated. A dedicated modelling platform on a cell-level for multiple temperatures (-5°C to 45°C) and dynamic load profiles has been validated for all three investigated lithium-ion chemistries. A modelling methodology has been proposed to develop the required parameters through the EAC results and a nonlinear least square estimation method. Through this, the parameterization a 2nd order Thévenin ECM topology is performed. The main objective is to reach the highest accuracy (≤+/-5%) possible to acquire a holistic overview of the performance of the batteries. Thus different steps have been incorporated in the modelling methodology. It is possible to inspect the behaviour of the parameters through the “Parameter Smoothing” adjustment method and define in the model multiple levels of capacity values (It/5, It/2 and Itoptimized) to achieve the highest accuracy possible. Furthermore a statistical analysis is conducted on the modelling parameters to investigate the effect of current rates and SoC on the exhibited values. Through the performed simulations and parameter analysis an adequate accuracy is achieved of +/- 5% for the SoC range of 100% to 2% with increasing inaccuracies for the LTO and LFP chemistries after the 2% SoC mark. The NMC chemistry showed to have the most accurate simulations for the complete range of the SoC of 100% to 0%.
Following the development of the cell-level model it is possible to transition in the investigation of battery modules and battery packs that exist in current EV applications. A Generic Battery (GB) module and pack models have been developed which can simulate the appropriate battery topology and size. To inspect this, different simulations at module level are initially performed evaluating two modules of 5 NMC and 6 LFP batteries in series. The analysis showed an accuracy of +/-4.5% under the ECC SoC technique and +/-4 under the EKF SoC estimation techniques. Through the validated GB module level models an investigation is performed to size up different battery packs that can meet the energy and power operational requirements of EV applications. A vehicle simulation is established based on the Nissan Leaf electric driveline to acquire its energy and power requirements under different vehicle ranges (160km, 175km, 250km and 350km), temperatures (-5°C, 25°C, 45°C) and load profiles (NEDC, WLTP). Different battery pack configurations with batteries attached in a hybrid topology (series and parallel) have been established. The energy content, energy efficiency, weight and volume of each battery pack has been investigated to determine if these can substitute the battery pack found in the Nissan Leaf. Finally an investigation is conducted to check if the GB battery packs models based on the NMC chemistry can validate the voltage output of the Nissan Leaf battery pack under real driving conditions. The simulation performed reached an accuracy of less than +/-5%, showing the models and battery packs developed can be employed in this vehicle.