🎓 Lesson 2
D2
Lithium-Ion Cell Operation & Voltage Profiles
A lithium-ion cell is like a rechargeable energy shuttle: it stores electricity when charging and releases it as usable power when discharging, with voltage changing predictably as its stored energy goes up or down.
🎯 Learning Objectives
- ✓ Explain how electrode material chemistry governs the shape and range of a Li-ion cell’s voltage profile
- ✓ Calculate state of charge (SoC) from measured open-circuit voltage using empirical OCV-SoC lookup tables or polynomial fits
- ✓ Analyze voltage hysteresis and polarization effects during charge/discharge cycles to diagnose degradation or thermal imbalance
- ✓ Design a BMS voltage-sampling strategy that resolves critical inflection points in the OCV curve for SoC accuracy < 2% error
📖 Why Voltage Profiles Dictate Real-World Battery Performance
In mining and remote energy storage—where diesel gensets are replaced by battery–solar–inverter hybrids—understanding voltage profiles isn’t academic: it determines whether your 2.5 MWh site battery shuts down unexpectedly at 15% SoC (due to misestimated voltage sag) or safely delivers full runtime. A single-cell voltage curve tells you *when* the battery is full, *how much energy remains*, and *whether it’s degrading*. Get this wrong, and you risk over-discharge, thermal runaway, or premature replacement—costing $200/kWh in capital and downtime.
📘 From Electrochemistry to Voltage Signature
The cell voltage arises from the difference in Gibbs free energy between cathode and anode active materials. During discharge, Li⁺ ions de-intercalate from the anode (oxidation) and insert into the cathode (reduction); the resulting OCV follows Nernstian behavior but is modulated by solid-state diffusion kinetics, phase transitions (e.g., two-phase plateaus in LFP), and entropy changes. Graphite/NMC cells show a smooth, sloping OCV (~3.0–4.2 V), while LiFePO₄ exhibits a flat ~3.2–3.3 V plateau due to its olivine two-phase reaction mechanism. Temperature shifts the entire curve (~−0.3 mV/°C per °C), and aging reduces voltage hysteresis width and flattens plateaus—early indicators of capacity loss.
📐 OCV–SoC Polynomial Fit for State Estimation
Empirical OCV–SoC relationships are modeled using low-order polynomials for real-time BMS implementation. A 4th-order fit balances accuracy and computational load in embedded systems. This formula converts measured resting OCV into SoC—critical for duty-cycle planning in off-grid mine sites where load profiles vary hourly.
💡 Worked Example
Problem: A rested LFP cell measures 3.285 V after 2-hour rest at 25°C. Use the industry-standard 4th-order OCV–SoC fit: OCV(SoC) = 3.200 + 0.126×SoC − 0.172×SoC² + 0.112×SoC³ − 0.031×SoC⁴ (SoC ∈ [0,1]). Solve for SoC.
1.
Step 1: Define f(SoC) = 3.200 + 0.126×SoC − 0.172×SoC² + 0.112×SoC³ − 0.031×SoC⁴ − 3.285 = 0
2.
Step 2: Use Newton-Raphson or solver (e.g., MATLAB fzero) with initial guess SoC₀ = 0.6
3.
Step 3: Converge to SoC ≈ 0.792 → 79.2% SoC; verify residual |f(0.792)| < 10⁻⁵ V
Answer:
The result is 79.2% SoC, which falls within the typical accurate range of 10–90% SoC where polynomial fit error < 0.5% SoC.
🏗️ Voltage Profile Failure in Pilbara Off-Grid Hybrid System
At Rio Tinto’s Yandicoogina solar–battery–diesel microgrid (Western Australia), early BMS units used linear OCV–SoC mapping. During monsoonal cooling (Tₐₘb ↓ 12°C), cells drifted 15 mV below nominal OCV—causing 8% SoC overestimation at 20% true SoC. Result: three inverters tripped during peak load due to undetected low-voltage cutoff. Retrofitting with temperature-compensated 4th-order LFP OCV tables reduced SoC error to <1.2% across −5°C to 45°C—extending battery life by 18 months and avoiding $1.2M in unscheduled diesel fuel.