A primer on Lithium-ion cells and battery management
A battery management system monitors and controls key parameters of a battery, optimizing the system’s longevity, performance, and safety. We often refer to a battery as a pack of cells, wired in series and parallel to achieve target voltage and capacity. To understand the roles of a BMS, let’s start with the fundamental building block - a single cell.
We’ll be using Lithium-ion cells as an example, the most common cell chemistry for EVs and rechargeable devices.
Anatomy of Li-ion cells (cross-sectional view):
Cathode: The electron-emitting electrode.
Anode: The electron-accepting electrode.
Electrolyte: The medium through which ions can move between the anode and cathode.
Separator: A polymeric membrane positioned between the anode and cathode preventing them from shorts.
Key Processes (extremely simplified)
Cells store chemical energy and release electrical energy, utilizing the electrochemical potential. Standard reduction potential is the tendency for a material to lose electrons. Lithium is so popular for cathodes because it has one of the lowest standard reduction potentials and the highest energy density.
Oxidation refers to when atoms lose electrons, and reduction refers to when they gain electrons. Standard reduction potential is a numerical value indicating an element’s tendency to gain electrons. Lithium’s low reduction potential just means it’s very easy for it to lose electrons and reduce.
When charging, lithium ions flow from the cathode to the anode through the electrolyte, which creates free electrons in the anode. When discharging, lithium ions flow from the anode to cathode, delivering electrical current to the load. During the charge and discharge cycles, the cell transforms into a complex electrochemical system where its safety, lifespan, and functionalities can be easily compromised without a BMS.
Roles of a BMS
SOC(state of charge) measuring
SOC measures the remaining usable energy in a battery. Soc estimation is done by using OCV, impedance spectroscopy combined with Kalman filters, and AI/ML to improve model accuracy.
Ensure the cell is in SOA(Safe operating area)
The safe operating area is where a cell can operate without failure or safety issues, it’s bounded by current, voltage, and temperature.
Current limiting
When the current is too high the BMS simply stops the charging or discharging.
Cell balancing
Ensuring cells in a battery pack have the same level of voltage, preventing overcharging or discharging for a subset of the cells in the pack.
Active balancing
Active balancing involves using electronic components (such as switches and inductors) to transfer energy between cells actively. The BMS actively moves charge from higher-voltage cells to lower-voltage cells.
Passive balancing
Passive balancing relies on dissipative elements, typically resistors, to equalize the voltage of cells. Resistors are connected in parallel with higher-voltage cells to create a discharge path, allowing excess energy to be dissipated as heat.
A well-designed BMS should have the ability to disconnect any single faulty cell, preventing it from further damage to the whole battery pack.
Thermal management
The BMS dissipates heat when it is too hot and provides heat when it is too cold, usually by controlling the pack’s cooling system, whether it’s coolants or fans. Accurate thermal profiling is key to informing the respective interferences from the BMS.
Thermal profiling is done through a network of thermistors. There are two main types of thermistors:
NTC(negative temperature coefficient), meaning the resistance is inversely proportional to temperature
PTC(positive temperature coefficient), meaning the resistance is directly proportional to the temperature
NTC’s nonlinear change in resistance provides much more sensitive responses to variations in temperature in the regions we care about: 0-45c for charging, -20-60c for discharging. See the comparison in resistance and temperature graph below. Asymptotic high-temperature region is improved by implementing linearization models.
Data logging for post-mortem analysis and ML training for prognosis algorithms and better SOC and SOH estimation models.