Coulomb Counting, Voltage Measurement, and Kalman Filtering: How Battery Management Systems Estimate State of Charge and State of Health

One engineering camp argues that tracking every amp that flows in and out of a battery—a method known as Coulomb counting—is the only way to know exactly what remains in the tank of an electric vessel.

The opposing camp insists that current sensors inevitably drift, and that measuring the actual chemical potential of the cells via Open Circuit Voltage is the only ground truth. For a boat owner weighing a $15,000 lithium-ion refit in 2026, this is not an academic debate. When a vessel is 10 miles offshore, a dashboard that reads 20% must mean exactly 20%.

The tension between these two approaches defines how modern Battery Management Systems operate. "The battery management system (BMS for batteries) is an electronic control unit that controls, monitors and optimises battery operation," notes Zendure. Without it, a lithium-ion bank is a volatile chemical hazard. But its most critical job for the operator is estimating two invisible metrics: State of Charge (SOC), the immediate fuel gauge, and State of Health (SOH), the long-term degradation of the pack.

The most intuitive method is Coulomb counting. As the name suggests, it tallies the coulombs—the unit of electric charge—entering and leaving the battery. If a marine battery holds 300 amp-hours and the outboard motor draws 30 amps for an hour, the BMS subtracts 10%. It is a simple, real-time ledger.

But as engineers at SunLith Energy point out, this ledger is blind to reality: "Coulomb counting is the most widely used BMS SOC estimation method in real-time operation," but it requires periodic recalibration. It relies entirely on the accuracy of the current sensor. If the sensor is off by even 0.1 amps, that error accumulates over hours of cruising.[1]

To correct that drift, systems turn to Open Circuit Voltage. This method measures the resting voltage of the battery and maps it to a known discharge curve. For a lithium iron phosphate (LiFePO4) marine battery, a resting voltage of 13.3V might indicate a 70% charge. The advocates of OCV argue that it provides an absolute chemical truth, immune to sensor drift. "After measuring the OCV, you can find out the initial SOC from the voltage-capacity chart," explains Integra Sources.[2]

However, OCV has a fatal flaw for active boaters: it requires the battery to be completely at rest. The chemical reactions inside the cells must stabilize. "OCV is reliable when the battery has truly rested. A 30–60 minute rest lets the voltage fully settle after any charge or discharge event," according to SunLith Energy. Under load—while fighting a headwind or running navigation electronics—the terminal voltage fluctuates wildly. A BMS cannot use OCV to tell a captain how much range remains while the throttle is open.[1]

This is where the Kalman Filter enters the architecture. Originally developed for aerospace navigation and trajectory tracking, the Kalman Filter is a recursive mathematical algorithm that bridges the gap between the two imperfect methods.

It takes the real-time, noisy data from Coulomb counting and continuously compares it against a sophisticated mathematical model of the battery's expected behavior, as detailed in a comprehensive review published in MDPI. By treating sensor inaccuracies as an expected variable rather than a fatal flaw, the algorithm constantly refines its own predictions based on how the battery responds to the throttle.[3]

By minimizing the error between the predicted voltage and the actual measured voltage, the Kalman Filter dynamically corrects the SOC estimate. It does not require the boat to sit idle for an hour, nor does it fall victim to the slow accumulation of sensor drift.

For an owner, this means the dashboard gauge remains accurate even after three days of continuous cruising and variable solar charging. The computational overhead is higher, requiring a dedicated 32-bit processor on the BMS board, but the result is a fuel gauge that operators can trust implicitly.

Beyond the immediate fuel gauge, the BMS must also calculate the State of Health. A battery's capacity permanently degrades over time due to chemical wear. A pack that held 300 amp-hours when new might only hold 240 amp-hours after 800 cycles. If the BMS still calculates SOC based on the original 300 amp-hour capacity, the boat will run out of power while the gauge still reads 20%. As Battle Born Batteries explains, SOH "tracks long-term condition compared to when the battery was new."

To prevent this, advanced BMS units use genetically optimized Extended Kalman Filters to track the gradual increase in internal resistance and the fade in total capacity. Research published in MDPI demonstrates that combining model-based parameter identification with an Extended Kalman Filter guarantees the estimation accuracy of these critical metrics over thousands of cycles.

By continuously updating the SOH, the system recalibrates the "full" mark so the SOC gauge reflects the battery's current reality, not its factory specifications. For a buyer looking at a used electric vessel, the SOH reading is the single most important metric on the survey—it dictates whether the boat needs a $10,000 battery replacement next season.[4]

The architecture of the BMS directly dictates the reliability of the vessel. A basic system relying solely on Coulomb counting requires the owner to frequently charge the bank to 100% to reset the drift. A sophisticated system utilizing Kalman filtering manages the math invisibly. "Often called the brain of the battery, the BMS ensures your batteries operate safely, efficiently, and for as long as possible," summarizes Battle Born Batteries. For the marine market, that computational brain is the difference between a seamless electric transition and a stranded vessel.