Lithium Battery Cycling at Partial State of Charge
Most buyers read a lithium battery datasheet, find a cycle-life number, and build a duty plan around it. That number was measured at one depth of discharge, one temperature, one rate, and one end-of-life criterion that may not match your application. In eleven years of designing lithium battery packs for industrial customers, I have rarely met a deployment that actually cycles 0 to 100 percent. Peak-shaving banks, hybrid generator sites, solar self-consumption systems, and telecom buffers all live in a partial state of charge band, and that band is the biggest lever you have on pack lifetime. This article is the framework I use to turn that lever deliberately.
What Partial State of Charge Cycling Actually Means
Partial state of charge cycling, usually shortened to PSOC cycling, means the battery repeatedly swings between two charge levels that never touch the extremes. A bank that runs daily between 40 and 70 percent SOC is PSOC cycling with a 30 percent depth of discharge. A bank held at full charge and discharged only during outages is not cycling at all; it is floating, and it degrades by a different mechanism. A drone battery that dumps 85 percent of its charge per sortie and recharges fully is deep-cycling, nearly the opposite regime. The distinction matters because each regime attacks a different weak point in the cell.
Three parameters define a PSOC duty cycle. Depth of discharge is the height of the swing. Band placement is where that swing sits on the SOC scale, for example 40 to 70 versus 80 to 95. Dwell time is how long the pack sits still at any point inside the band, because degradation is driven by both throughput and time at voltage. A specification that quotes only “30 percent DoD” is missing two thirds of the story, and the missing third is often the one that kills the pack.
Why Shallower Cycles Last Longer
The physics is mechanical before it is chemical. Every lithium ion that enters a graphite anode or an LFP cathode particle makes the crystal lattice expand; every ion that leaves makes it contract. A full cycle forces the largest possible strain excursion on every particle in the electrode. Repeated large strains fatigue particles, open cracks in the active material, and expose fresh surface to the electrolyte. Fresh surface consumes lithium to rebuild the passivation film, the slow loss you eventually measure as aging. At 20 percent depth of discharge the strain excursion is five times smaller, so the fatigue damage per cycle falls much faster than the cycle count rises.
Voltage adds a second, calendar-driven mechanism. At high SOC, especially with dwell, the cathode sits at a potential that slowly oxidizes the electrolyte, thickening the positive-electrode film and releasing gas. At very low SOC the anode potential climbs toward the copper collector dissolution threshold, which is why storing a pack near zero percent for months is dangerous. There is also the plating risk: charging cells when the anode is cold deposits lithium as metal instead of intercalating it, permanently consuming cyclable lithium. Shallower cycles spend less time in the high-voltage and low-voltage corners where these reactions run fastest, which extends life beyond the fatigue argument alone.
Turning Cycle Counts Into Throughput
Comparing cycle counts across depths of discharge is meaningless without a common currency. The one I use is equivalent full cycles, or EFC: multiply the cycle count by its depth of discharge. A pack rated for 5,000 cycles at 80 percent DoD delivers 4,000 EFC of lifetime throughput; a pack rated 25,000 cycles at 20 percent DoD delivers 5,000 EFC. The EFC figure tells you how much energy the chemistry can move before end of life, and comparing EFC across DoD points shows whether shallow cycling genuinely buys anything.
For a typical large-format LiFePO4 cell at 25 degrees C and 0.5C, the relationship follows a power law with an exponent near 1.1:
- 100 percent DoD: roughly 4,000 cycles, about 4,000 EFC of throughput
- 80 percent DoD: roughly 5,100 cycles, about 4,100 EFC
- 60 percent DoD: roughly 7,000 cycles, about 4,200 EFC
- 40 percent DoD: roughly 11,500 cycles, about 4,600 EFC
- 20 percent DoD: roughly 23,000 cycles, about 4,600 EFC
- 10 percent DoD: roughly 50,000 cycles, about 5,000 EFC
Notice that the EFC column is nearly flat: for healthy LFP chemistry, lifetime throughput is close to constant, and the “extra” life from shallow cycling comes back mostly as calendar aging relief. The cycle count alone misleads the other way: quoting 23,000 cycles at 20 percent DoD sounds like a sixfold improvement but moves only 15 percent more total energy through the cell. For NMC the exponent is higher, near 1.35, and the EFC curve genuinely rises as DoD falls, because NMC suffers more from high-voltage dwell. That difference is exactly why you cannot copy an LFP duty strategy onto an NMC pack or vice versa.
Where You Put the Window Matters More Than How Wide It Is
Two duty profiles with identical depth of discharge can differ by a factor of two in lifetime, and the difference is band placement. Cycling 45 to 75 percent is gentle; cycling 90 to 100 percent at the same 30 percent swing parks the cell at full charge for hours daily, where electrolyte oxidation and positive-electrode film growth run fastest. When I model customer duty profiles, I plot the time-at-voltage histogram before saying anything about cycle life. The number that matters most is usually hours per day spent above 95 percent SOC for LFP, or above 4.1 volts per cell for NMC.
This yields a design rule that surprises buyers: shrink the top of the window first, not the bottom. For LiFePO4 the voltage curve is nearly flat between about 20 and 90 percent SOC and turns sharply upward at the knee near 95 percent, so cap the upper band at or just below that knee unless a load case truly needs the last few percent. Reserve the extreme top for balancing events and rare full-charge days. The low end is more forgiving: sitting at 20 to 30 percent SOC is benign, and the real floor is set by state estimation quality, because below roughly 15 percent the LFP curve starts to move and the fuel gauge becomes trustworthy again. A band of 20 to 85 percent with quick passage through the top region will outlast a 10 to 70 percent band that gets topped up to 100 percent every weekend.
The Hidden Cost: State Estimation and Balance Drift
PSOC duty has a cost that never appears on the cycle-life curve, and I have watched it strand more field systems than DoD damage ever has. A battery management system estimates state of charge mainly by integrating current, called coulomb counting, and corrects the accumulated error against open-circuit voltage when the pack rests. Full charging anchors the BMS at 100 percent. If the duty cycle never reaches either end, the correction never happens and integration error accumulates without bound.
Put numbers on it: a shunt or Hall sensor with a half percent offset and a half percent gain error, integrated over thirty days of partial cycling, can drift the reported SOC by 8 to 15 percent. Now add the LFP problem: in the flat region the open-circuit voltage changes only 2 to 5 millivolts per percent of SOC, while sensor and cell tolerances are easily 10 to 20 millivolts. A resting voltage reading cannot resolve SOC in the middle of the band. The pack reporting 55 percent may actually hold 40 or 70, and the first symptom is usually a runtime complaint followed by a surprise shutdown.
Cell balancing fails the same way. Passive balancing bleeds small currents through resistors and can only tell which cells are high when their voltages separate, near the top of charge. A pack that never approaches the knee never balances, so divergence grows silently for months. The fixes are procedural, and any serious custom battery solution includes them in firmware and the operating plan:
- Schedule a full absorption charge every 7 to 14 days for LFP systems to re-anchor SOC and run balancing.
- Terminate charges by low-current tail detection, typically when current falls below C/20, rather than by voltage alone, so the anchor is real.
- Keep 10 to 15 percent of balancing headroom at the top of the normal band, treating the weekly full charge as the exception, not the rule.
- For long series strings above roughly 16 cells, specify active balancing with at least 200 milliamperes of transfer capability.
- Use an extended Kalman filter or equivalent model-based estimator so voltage restraint limits drift between anchors.
Sizing and Economics: Buy Capacity to Buy Shallowness
Once you accept that the band, not the nameplate, defines the usable energy, sizing becomes an economic optimization. Suppose your load needs 100 kWh of daily throughput. Option A sizes a 111 kWh nameplate pack and cycles it 90 percent deep, yielding about 4,300 cycles, roughly 430 MWh delivered over its life. Option B buys a 200 kWh pack and cycles it 45 percent, which yields about 9,600 cycles at 90 kWh per cycle, roughly 866 MWh delivered. At a battery price of 180 dollars per kWh, Option A costs about 20,000 dollars and 46 dollars per delivered MWh; Option B costs about 36,000 dollars and 42 dollars per delivered MWh.
Two second-order effects widen the gap in favor of the bigger pack. Shallower cycles generate less ohmic heating, so the same cooling holds a lower average cell temperature, and calendar aging, which usually dominates anyway, slows. The larger pack also reaches its daily peak current at a lower C-rate, reducing plating risk on cold mornings. The gap closes or reverses when floor space, structural weight, or shipping volume is expensive, precisely the situation in drones, vehicles, and marine craft. For stationary plants, buy the shallower duty; for weight-limited platforms, buy the right chemistry and defend the band you can afford.
How to Specify PSOC Life in a Purchase Contract
A cycle-life number without its test conditions is a marketing sentence, not a specification. When I write procurement documents for a lithium battery pack, I require the vendor to state five conditions next to every cycle count: depth of discharge, the SOC band including its center, the charge and discharge rates, the test temperature, and the end-of-life definition, which should be 80 percent of initial capacity or a stated internal resistance increase. A cell quoted at 6,000 cycles to 80 percent at 25 degrees C and 0.5C can easily deliver 2,500 cycles in a 45 degrees C container with 1C pulses, and the datasheet was never wrong.
Anchor the specification to test standards so there is no argument about method. IEC 62619 covers safety and cycle testing of the cell design; UL 1973 addresses stationary and light-rail applications; IEC 61427-1 is particularly valuable because it defines a partial-state-of-charge duty test for renewable energy systems, exactly the regime most industrial packs live in. IEEE 1679.2 governs how characterization data should be reported. Then make acceptance real: require cycle-life data at your actual depth of discharge and band, not the vendor’s favorite 70 percent point, and for large orders reserve the right to witness a 300 to 500 cycle accelerated block with checkpoints every 100 cycles.
Finally, demand clarity on how cycle and calendar aging combine. They run in parallel in any real deployment, and the correct accounting uses rainflow counting of the actual load profile rather than a simple cycle tally. A supplier who can explain the combination model in one paragraph is a supplier whose numbers you can plan around.
Duty-Cycle Playbook for Common PSOC Regimes
Four regimes cover most industrial deployments, each with a different failure mode to manage.
- Peak shaving and solar self-consumption: daily swings of 20 to 40 percent, typically 20 to 85 percent band. This is the kindest realistic duty. Manage the weekly full-charge anchor and keep the top reserved for backup.
- Hybrid generator sites: the engine covers peaks and the battery absorbs transients in a 40 to 70 percent band. Watch state estimation drift hardest here, because these sites run unattended for months. Anchor with a scheduled generator-driven full charge.
- Frequency regulation and grid services: thousands of micro-cycles per day at 5 to 15 percent DoD. DoD is nearly irrelevant; the pack dies of throughput fees and calendar age. Contract on EFC delivered, not cycles.
- UPS float duty: technically not PSOC at all, since the pack sits at 100 percent. The degradation is high-voltage dwell, and the mitigations are lower float voltage, periodic exercise cycles, and honest derating of the warranty.
For contrast, consider a drone battery fleet. Sorties demand 80 to 90 percent of the charge in minutes and recharge fully between flights, so there is no partial-cycling luxury. The strategy there is high-power cell selection, aggressive thermal management, and a shorter replacement interval, which is why I never let a stationary-storage duty plan migrate onto an airborne platform or the reverse.
Frequently Asked Questions
Does partial state of charge cycling damage a lithium battery?
No. Kept away from the extremes, shallow cycling is the gentlest realistic duty for lithium-ion chemistry. The genuine risks of PSOC duty are silent: state estimation drift and cell imbalance, both managed with scheduled full charges and proper balancing, not by cycling deeper.
What is the best SOC window for a LiFePO4 battery in daily cycling?
For most stationary applications I specify 20 to 85 percent, with the top at or just below the voltage knee near 95 percent. If you can tolerate it, 15 to 80 percent is even kinder. The key discipline is minimizing hours parked at full charge rather than obsessing over the bottom edge.
How often should I fully charge a battery that cycles partially?
For LFP systems, every 7 to 14 days is a workable anchor interval; NMC tolerates 2 to 4 weeks because its steeper voltage curve restrains the estimator better. The right interval depends on current sensor quality, so log estimation error after maintenance events and adjust from data.
Why do vendor cycle-life numbers differ so much for the same cell?
Because the conditions differ and are often unstated. The same cell can show 6,000 cycles at 70 percent DoD and 25 degrees C but fewer than 3,000 at 90 percent DoD, 45 degrees C, and 1C pulses. Always read the five test conditions before comparing numbers.
Is it cheaper to buy a bigger battery and cycle it shallower?
For stationary plants, usually yes. The worked example above came out about 10 percent cheaper per delivered megawatt-hour, before counting lower heating and slower calendar aging. The conclusion reverses when weight and volume carry a real cost, as in drones and vehicles.
Can sodium-ion batteries handle partial state of charge cycling the same way?
Yes, and the hard-carbon anode tolerates partial SOC duty well. The sodium-ion voltage curve is even flatter than LFP, so coulomb counting drifts faster and scheduled full-charge anchors matter more. A bonus: sodium-ion cells ship and store safely at zero volts.
Further Reading
References
