State-of-charge (SOC) algorithm design methodology for implementation on Battery Management Systems (BMS) of industrial Li-Ion battery packs

Abstract:

The design of a full diagnostic and management system of the Battery Management System (BMS) of a lithium ion (Li-Ion) battery-pack entails multidisciplinary knowledge. Having as a core mathematical models and equations, it is surrounded by chemistry, electronics and thermal constraints that appear when it is implemented in a real battery-pack.

Academia mainly concentrates efforts to look for new mathematical solutions at laboratory level to improve the accuracy of diagnostic and management. Industry, nevertheless, requires treating the issue with an integral Overview. It requests to consider in addition to laboratory level approaches, real implementation constraints (impact of BMS hardware, cell performance and battery-pack design and Operation) in their design/selection process. In effect, Once Out from laboratory conditions a significant portion of final performance decrease of algorithms Often comes down to them. Beyond accuracy, moreover, there are other factors (characterization and parameterization complexity of the algorithms, required computational resources, etc.) that make them more Or less suitable for different applications.

Literature nowadays does not provide a procedure that considers all these aspects together for algorithm design/selection. In this framework, the main objective Of this thesis is the proposal and application of a novel multiobjective algorithm design/selection methodology. It considers both, generic diagnostic and management approaches and their most constraining factors. It aims at determining the most suitable solution for any application case according to specific key performance indicators (KPI). Main focus is placed on state-of charge (SOC) estimation strategies design/selection.

A generic-purpose application of the proposed methodology is realized for SOC estimation algorithms. The study is supported by more than 250 simulations where different algorithms are Guantitatively analyzed at specific load profiles and wide cell performance and BMS hardware conditions. The Objective is to Quantify their robustness, as well as adaptability to each application, required computational resources and, characterization and parameterization complexity, before concluding their suitability.

The study is further reinforced with experimental data concerning a sample of 18 Li-Ion cells that represents almost all the variety of Li-Ion cells nowadays in the market. The tested cells are of 5 different Li-Ion chemistries (NCM/C, LMO/LTO, NCM-LFP/C-LTO, LCO/C and LFP/C), wide range of capacity (1Ah-100 Ah), both high energy and high power characteristics, 3 formats (pouch, cylindrical, prismatic) and 8 different manufacturers. They are used to emulate all real casuistries that could happen when designing the SOC strategy of a battery-pack.

Finally, the proposed methodology is applied for the design of the SOC diagnostic system of a LFP/C (LiFePO) battery-pack Oriented to an elevation application. The designed algorithm is validated On a BMS.

Keywords: lithium ion, Li-Ion, battery management system, BMS, battery-pack, diagnostics, management, state-of-charge, SOC, state-of-health, SOH, cell balancing, OCV, Open circuit voltage, hysteresis, LiFePO LFP/C.