Decentralized Battery Management System for Improved Reliability and Optimized Battery Operation

Kurzfassung

In this thesis, a decentralized architecture for heterogeneous battery systems is presented and the corresponding design and control challenges are addressed considering the system objectives of increased reliability and system availability as well as optimized battery operation. With the integration of renewable energy sources and the electrification of the transportation sector, the demand for battery storage systems is increasing. In safety-critical applications such as electric vehicles or home storage systems, the reliability, safety and availability of the battery system play a crucial role. The increased demand for batteries is leading to constant new developments with the aim of improving performance and capacity. Furthermore, with the increased use of battery systems, the number of second life batteries is rising. Second life batteries are batteries which, due to increased internal resistance and reduced capacity, can no longer be used in applications with high requirements in terms of dynamics and storage capacity, such as electric cars. These batteries can still be used in applications such as home storage systems with lower requirements in terms of dynamics. Heterogeneous battery systems combine batteries with differences in cell chemistry, actual battery capacity, state of charge, state of health, and permissible operating range in one system. They enable the exploitation of different battery characteristics, for example the combination of power- and energy-dense batteries, and provide and offer the additional option of integrating second life batteries. For sufficiently high battery currents and adequate battery capacity, several battery modules are connected in parallel. Different battery modules with various cell chemistries have deviating terminal voltages, whereby these also in some cases depend on the state of charge. The direct parallel connection of batteries with deviating terminal voltages leads to uncontrolled, potentially very high, circulating currents between the battery modules. This results in uncontrolled charging and discharging of the batteries and, in the worst case, to battery operation outside the safe operating area resulting in serious safety risks. Furthermore, the batteries supply each other and the actual control objective, i.e. the reliable supply of a connected load, is not achieved. In this thesis, a DC system consisting of a varying number of various batteries, variable loads and generators is considered. For improved reliability and availability, a decentralized architecture is proposed, whereby all system components (batteries, generators and loads) are equipped with local control units. The local control units include a microcontroller with communication interfaces, current and voltage sensors, a relay that can be opened in case of a fault or for maintenance purposes, and DC-DC converters. In order to realize the power flow in charging and discharging direction, bidirectional DC-DC converters are used for the batteries. The system control is distributed among the functional equivalent, equally local control units, thus reducing potential single points of failures. Despite the numerous advantages that arise from the combination of different batteries and the decentralization of the system, there are several open challenges to be addressed in terms of the coordination of the system tasks, the control and the implementation. Communication between the nodes is required in order to enable information exchange with higher level system components and to achieve the completion of the system tasks in a collaborative manner. System-wide data consistency, low latency with few user data bytes, low computational effort and low power consumption are communication requirements that arise from the chosen architecture. A central control unit is required for synchronization between the nodes and the allocation of system tasks without endangering the system reliability by introducing potential single points of failures. For effective task coordination in a system consisting of networked, functionally equal nodes, a dynamic, criterion-based leader election algorithm is proposed. One of the battery nodes is temporarily elected as the leader depending on a defined election criterion. It controls the execution of the system tasks and synchronizes the local measurements of the individual nodes. The main objectives of the system control is the stable regulation of the DC line voltage to a predefined set point under variable loads and consumers. For a safe operation of a heterogeneous battery system it is necessary to consider the different safe operating areas of the diverse batteries. In order to combine diverse batteries while maintaining the safe operating areas, a battery-state dependent load current distribution is necessary in all operating modes. Therefore, two different control strategies are presented for the realization of the battery-state dependent load current distribution, which differ in the communication requirements, in the control accuracy and dynamics, in robustness, and in the possibility to acount for battery degradation. In addition, a system-wide unambiguous state evaluation of different batteries is required, taking into account their permissible operating ranges. For the implementation, dedicated DC-DC converters are required which have a variable input voltage range at the low voltage side for the support of different batteries, an adjustable voltage at the high voltage side for supporting different applications, whereby low voltage systems in a range of 18V to 24V are considered here, and a current limitation which can be adjusted in active operation. For initial tests of the control system and for scalability considerations, simulations of the DC-DC converters are necessary. Furthermore, a special test environment is required which takes into account the communication line in the control loop. Special algorithms are implemented for the microcontrollers to fulfill the communication-, coordination-, measurement- and control-tasks. A hardware test setup consisting of four local control units was developed to verify the functionality of the decentralized architecture and the fulfillment of the defined control objectives. Experimental data under consideration of different load scenarios exhibits the desired behavior. Initial tests to demonstrate robustness, e.g. failure of a node, faulty measurements, were also carried out successfully.