Online SOC Estimation for Vanadium Redox Flow Batteries Using OCV, Coulomb Counting, and ORP Fusion

Introduction

Accurate state-of-charge (SOC) estimation is critical for the safe and efficient operation of vanadium redox flow batteries (VRFBs). Unlike lithium-ion batteries, VRFBs experience significant self-discharge due to cross-contamination of vanadium ions across the membrane, leading to cumulative errors in traditional ampere-hour integration. This guide presents a hybrid algorithm that fuses open-circuit voltage (OCV), coulomb counting, and electrolyte oxidation-reduction potential (ORP) measurements to achieve robust online SOC estimation for flow battery BMS.

The proposed method leverages the inherent linear relationship between OCV and SOC in VRFBs at equilibrium, while ORP sensors provide real-time insight into the electrolyte oxidation state. By combining these with ampere-hour integration, the algorithm dynamically compensates for self-discharge losses and sensor drift.

Challenges of SOC Estimation in Vanadium Redox Flow Batteries

VRFBs store energy in vanadium ions dissolved in sulfuric acid. The SOC is defined by the ratio of V^2+/V^3+ (negative electrolyte) and V^4+/V^5+ (positive electrolyte). Self-discharge occurs when vanadium ions cross the membrane, reducing the usable capacity. This makes pure coulomb counting unreliable over long cycles.

Additionally, OCV readings require the battery to be at rest (zero current) for several minutes to reach equilibrium, which is impractical during normal operation. ORP electrodes, on the other hand, can provide continuous signals proportional to the vanadium ion ratios, but they suffer from calibration drift. A fusion approach is therefore essential.

Open-Circuit Voltage Method

The Nernst equation relates VRFB OCV to SOC: E = E0 + (RT/nF) * ln(([V^5+]/[V^4+])/([V^2+]/[V^3+])). In practice, a linear approximation holds for SOC between 10% and 90%. The BMS periodically initiates a rest period (e.g., during idle or low-load intervals) to obtain a reference OCV-SOC point.

This reference is used to reset the cumulative error from coulomb counting. However, frequent rest periods reduce system availability. Therefore, OCV is used as a periodic correction, not a continuous estimator.

Ampere-Hour Integration (Coulomb Counting)

Coulomb counting integrates the current over time: SOC(t) = SOC(0) + (1/C) ∫ I dt, where C is the measured capacity. In VRFBs, the capacity degrades over cycles and self-discharge adds an unmeasured current. Without correction, errors accumulate linearly.

To mitigate this, the BMS must periodically reinitialize SOC using OCV or ORP. The integration step also requires high-precision current sensors and temperature compensation for capacity changes.

Electrolyte ORP Fusion for Continuous Correction

ORP electrodes placed in the electrolyte recirculation loops provide voltage signals that correlate with the V^4+/V^5+ ratio in the positive half-cell and V^2+/V^3+ in the negative half-cell. Although ORP varies with temperature and electrolyte composition, it responds rapidly to SOC changes and can be used as a sliding correction.

In the fusion algorithm, the ORP-derived SOC is computed via polynomial calibration curves. A Kalman filter or complementary filter weights measurements from OCV (low frequency, high accuracy when available), ORP (mid frequency, moderate accuracy), and coulomb counting (high frequency, prone to drift). The filter adaptively reduces the influence of coulomb counting when self-discharge is detected (e.g., by divergence between ORP and integrated SOC).

Implementation in a Flow Battery BMS

Step 1: Install ORP sensors in both electrolyte tanks or circulation lines. Calibrate them against full-charge and full-discharge reference solutions. Step 2: During startup, measure OCV after a 10-minute rest to initialize SOC. Step 3: Run coulomb counting at 1 Hz. Step 4: At each BMS cycle (e.g., 1 second), compute ORP-based SOC using pre-defined curves. Step 5: Fuse three estimates using a weighted average: SOC = α * SOC_OCV + β * SOC_ORP + γ * SOC_Coulomb, with α+β+γ=1. Adjust β higher when self-discharge is suspected (e.g., after long idle periods).

Additionally, monitor the slope of ORP vs. time; a downward trend indicates increasing self-discharge rate. Use this to dynamically increase β. Periodically (every 10-20 cycles), if the battery is at rest, use OCV to correct the overall estimate and reset γ. This fusion reduces cumulative error to less than 3% under normal operation.

Conclusion

Combining OCV, ampere-hour integration, and ORP sensing provides a practical online SOC estimation method for VRFB systems. The ORP signal offers continuous, real-time awareness of electrolyte state, enabling the BMS to counteract self-discharge drift without interrupting operation. Implementation requires careful sensor calibration, noise filtering, and adaptive weighting logic. With this fusion algorithm, flow battery BMS can maintain accurate SOC throughout long charge-discharge cycles, improving system reliability and energy management.

Future work may explore machine learning to model self-discharge patterns and further optimize the fusion weights. For now, the described method represents a robust industrial solution for vanadium redox flow battery monitoring.