‘Charge to 80 per cent, not 100’ is the most widely repeated instruction in electric vehicle ownership, and it is also one of the least precise. It is correct for some cells, harmless for others and actively counterproductive for a third group, and the difference is not a matter of opinion but of crystal structure, electrode potential and estimation mathematics. A traction pack is an electrochemical system whose ageing depends on the voltage at which its cells dwell, and a software system whose knowledge of its own state depends on whether those cells give the software anything to measure. The two constraints point in opposite directions for nickel-rich and iron-phosphate chemistries. This article deconstructs both, from the lattice up to the Battery Management System (BMS), and then converts the result into charging and long-term storage profiles for commuters, highway users and seasonal fleet yards.
Figures given below are indicative values drawn from the general behaviour of commercial cells. Individual cell designs, electrolyte additives, pack thermal management and vendor software vary considerably, so the manufacturer’s published guidance for a specific vehicle always takes precedence. The purpose of the analysis is to explain why the guidance differs between chemistries, so that owners and fleet managers can apply it intelligently.
The Voltage Window: Where Each Chemistry Lives
A lithium-ion cell stores energy as a difference in chemical potential between two electrodes. At the cell terminals, that appears as a voltage, which rises as lithium is moved from the cathode to the anode during charging. A conventional nickel-rich cell using a layered oxide cathode (nickel manganese cobalt, NMC, or nickel cobalt aluminium, NCA) and a graphite anode is charged to an upper limit of around 4.2 V and has a nominal voltage of roughly 3.6 to 3.7 V. A lithium iron phosphate (LFP) cell is charged to roughly 3.6 to 3.65 V, with a nominal voltage of about 3.2 V. The difference matters in two ways. The energy density of a cell scales with its voltage, which is why LFP has lower energy per kilogram. And the stress on the materials scales with the electrochemical potential at which they are held, which is why nickel cells are far more sensitive to high states of charge.
State of charge (SoC) is not a measured quantity. It is an inference from measurements of current, voltage and temperature, and its accuracy depends on how much information those measurements carry. For nickel cells the open-circuit voltage (OCV) changes by roughly 8 to 12 millivolts for each percentage point of SoC across the mid-range, so a voltage measurement says a lot about charge. For LFP the OCV is nearly constant across most of the range, so a voltage measurement says very little. That single observation explains why the charging advice for the two chemistries diverges.
High-Voltage Hold States in NMC and NCA Cells
Lattice instability and oxygen release
Layered oxide cathodes store lithium between slabs of transition-metal oxide. As lithium is extracted during charging, the electrostatic screening between the oxygen layers weakens and the lattice responds by changing its interlayer spacing. In nickel-rich materials, the lattice expands slightly during the first stages of delithiation and then contracts sharply as the state of charge approaches the top of the window. This contraction, associated with a phase transition often labelled H2 to H3 in the literature, imposes an anisotropic strain on the primary particles. Over many cycles the repeated expansion and contraction opens intergranular micro-cracks in the secondary particles, which expose fresh surface to the electrolyte and fragment electrical contact within the electrode.
Equally important is the behaviour of oxygen. At high degrees of delithiation the oxidation state of nickel rises toward Ni4+, and the oxygen anions in the lattice become oxidised in turn. Oxygen can be released from the surface and near-surface region of the particle, transforming the layered structure into a disordered spinel-like and eventually rock-salt phase that has poor lithium diffusivity and high impedance. The released oxygen, in singlet form, is aggressive towards the carbonate electrolyte, accelerating its decomposition and generating gas and heat. Operating cells above about 4.1 V, equivalent to something like 80 to 85 per cent SoC for a typical nickel-rich cell, moves the surface into this regime for a larger fraction of its dwell time. The damage is nonlinear: dwelling at 4.2 V is far more harmful per hour than dwelling at 4.0 V.
Transition-metal dissolution and anode poisoning
Transition-metal ions, particularly manganese and, to a lesser degree, nickel and cobalt, dissolve from the cathode surface into the electrolyte, especially when the surface is highly delithiated and when trace hydrofluoric acid from electrolyte salt hydrolysis is present. The dissolved ions migrate through the separator and deposit at the anode, where they are reduced to metallic or low-valent forms. There they catalyse further reduction of the electrolyte, thickening the solid electrolyte interphase (SEI) and consuming cyclable lithium. A single manganese ion has been shown to trigger the consumption of many lithium ions in this catalytic cycle, which is why small amounts of dissolution produce disproportionate capacity loss.
SEI growth and the anode potential
The SEI is a thin passivation layer formed on graphite from reduction products of the electrolyte. It is necessary, because without it the graphite would continue to react, but it is also a sink for lithium and an added resistance. Its growth rate depends on the anode potential and the temperature. At high SoC, the graphite anode is highly lithiated and sits at a potential of only about 0.1 V above lithium metal, which is far into the region where electrolyte reduction is thermodynamically driven. At low SoC the anode potential is higher and the driving force is smaller. The consequence is that calendar ageing, the capacity loss that occurs even when the car is stationary, is much faster at a high state of charge, and that it is accelerated by temperature roughly according to an Arrhenius relationship, so a hot car parked at 100 per cent loses capacity fastest of all.
For planning purposes, one can think in terms of an indicative scale. Calendar capacity fade for a nickel-rich cell stored at 25 degrees Celsius might be in the region of 1 to 2 per cent per year at 50 per cent SoC, perhaps 2 to 4 per cent per year at 80 per cent, and noticeably more at 100 per cent, with the rate roughly doubling for every 10 degrees of temperature increase. These numbers are not guarantees; they illustrate the shape of the curve, which steepens toward the top of the window. Cycling adds its own fade, with the depth of discharge and the charging rate as the main parameters.
A related mechanism is lithium plating on the anode. When the anode is near full lithiation, or when charging at high current or low temperature, the local anode potential can fall below that of lithium metal and lithium deposits as a metallic film instead of intercalating. Plated lithium is partly irreversible, consumes inventory and, in the worst case, forms dendritic structures. Holding charge ceilings lower reduces the likelihood of plating because the anode retains headroom, which is one reason that high-SoC charging in cold conditions is particularly hazardous for nickel packs.
Why LFP Is Different
The olivine lattice and its two-phase plateau
Lithium iron phosphate crystallises in an olivine structure in which the iron-oxygen octahedra and phosphate tetrahedra form a robust three-dimensional framework with one-dimensional channels for lithium. The phosphorus-oxygen bonds are strongly covalent, so the oxygen is locked into the phosphate polyanion and is not released at high states of charge or moderate temperatures. The material changes between two phases, lithiated LiFePO4 and delithiated FePO4, and the coexistence of those phases over most of the composition range pins the electrochemical potential at a nearly constant value of about 3.4 V against lithium, which appears at the cell level as a plateau of about 3.2 to 3.3 V against graphite. The same structural stability that produces the plateau underlies LFP’s long cycle life and its relative tolerance of high states of charge at the cathode.
The consequence is that the high-voltage degradation mechanisms of the nickel cathode are largely absent. There is no oxygen release at the cathode surface in normal operation, little transition-metal dissolution (iron dissolution is much smaller than manganese dissolution, though not zero), and a small volume change of roughly 6 per cent between the phases, in a lattice that tolerates it. The residual degradation sits mainly at the anode, where SEI growth continues as in any graphite cell, and in the loss of lithium inventory. Because LFP is typically charged to a lower upper voltage, the anode potential at the top of charge is also somewhat less aggressive, though graphite is still deeply lithiated.
The flat curve and the estimation problem
The same plateau that protects the cathode starves the BMS. Across the middle of the SoC range, from roughly 20 to 80 per cent and sometimes beyond, the OCV of an LFP cell changes by only a few tens of millivolts. The slope, dOCV/dSoC, can be well below one millivolt per percentage point. Voltage measurement in a traction pack has a finite accuracy, typically one to a few millivolts per cell, and the terminal voltage under load is further contaminated by the polarisation voltage, which depends on current, temperature and history. An uncertainty of 2 millivolts divided by a slope of 0.5 millivolts per percentage point translates into an SoC uncertainty of about 4 percentage points, and the real uncertainty is larger because of hysteresis. LFP exhibits a characteristic voltage hysteresis between charge and discharge of the order of 10 to 30 millivolts that depends on the path by which the cell reached its present state. In effect, the same SoC corresponds to different voltages depending on whether the cell was last charged or discharged, and across a flat plateau this ambiguity exceeds the entire signal.
The BMS must therefore rely mostly on Coulomb counting, which integrates the measured current over time to track the charge entering and leaving the cell. Coulomb counting is accurate in the short term but accumulates error: current-sensor offset and gain error, finite sampling, temperature effects and the unknown efficiency of each cycle all add small biases that integrate into drift. A sensor with an offset of only 0.1 per cent of full scale, integrated over many hours, produces SoC errors of several per cent. Left uncorrected, the displayed state of charge and the range estimate move away from reality, and the usable capacity is mis-estimated.
The correction comes from the points on the curve where voltage is informative. At the top of charge, when the cell approaches about 3.6 V, the OCV rises steeply as the lithiated phase is exhausted at the cathode and the anode approaches full lithiation; the voltage then changes by tens of millivolts for each percentage point of SoC. A BMS that observes the cell reaching the voltage knee at a known low current can declare the cell full, reset the Coulomb counter to 100 per cent and recompute the capacity from the charge accumulated since the last reset. A similar knee exists at the bottom of discharge, but it is much less used in practice because customers rarely run LFP packs to empty. This is the reason that manufacturers instruct LFP drivers to charge to 100 per cent at regular intervals, often weekly: not because 100 per cent is good for the cell, but because the software needs the cell to reach the knee in order to relearn where it is.
Cell balancing provides a second reason. In a series string, cells differ in capacity and self-discharge, and the weakest cell sets the pack’s usable energy. Most vehicles use passive balancing, which bleeds energy from higher-voltage cells through resistors, and which works effectively only when the cells are at a voltage at which their differences in SoC appear as differences in terminal voltage, that is, near the top of charge. In the flat region, voltage differences are too small to act upon. Without periodic top-of-charge balancing, the spread between cells widens, the usable pack capacity shrinks to the capacity of the weakest cell, and the apparent range declines even though no cell has been damaged.
This calibration logic does not mean that LFP should sit at 100 per cent every night. The cathode is robust, but the anode is not: at the top of charge the graphite is almost fully lithiated and sits at a potential of only about 0.1 V above lithium metal, and SEI growth at that potential is faster, particularly when the pack is warm. A cell held at 100 per cent for many hours a day therefore accumulates calendar fade through SEI thickening and lithium-inventory loss even though its cathode is unaffected. The efficient compromise is to cap routine charging at 80 per cent, which keeps the anode away from its most aggressive potential, and to schedule one deliberate saturation charge each week so that the BMS reaches the voltage knee, resets its Coulomb counter and balances the string. The saturation charge should finish shortly before departure, so that the full-charge dwell is measured in minutes and not in days.
| Estimation Method | Principle | Behaviour on NMC and NCA | Behaviour on LFP |
|---|---|---|---|
| Coulomb counting | Integrates measured current over time from a known starting point | Accurate in the short term; drift is corrected by frequent OCV references | Drifts without correction because voltage references are rarely informative; depends on periodic top-of-charge reset |
| OCV lookup after rest | Reads open-circuit voltage after a relaxation period and maps it to SoC | Reliable: a slope of roughly 8 to 12 mV per percentage point gives a sharp estimate | Unreliable on the plateau: a slope below 1 mV per percentage point plus hysteresis creates errors of several percentage points |
| Extended Kalman filter (equivalent-circuit model) | Fuses current integration with a voltage model, correcting using measurement residuals | Converges quickly; model parameters track ageing | Converges weakly across the plateau because the voltage residual carries little information about SoC; relies on model accuracy and hysteresis modelling |
| Data-driven and hybrid estimators | Neural networks or regression models trained on fleet data, often fused with a filter | Improves robustness to ageing and temperature | Helps by learning hysteresis and history effects, but still anchored by the voltage knee at the top of charge |
Thermal and Structural Delta Between the Chemistries
Safety and longevity are linked through the same bond chemistry. In a charged nickel-rich cell, the delithiated cathode is a strong oxidant. When a cell is heated, abused or internally short-circuited, the layered structure becomes unstable at temperatures that can begin around 150 to 220 degrees Celsius depending on composition and state of charge, releasing oxygen that reacts exothermically with the electrolyte and the lithiated graphite. The heat released raises the temperature further and drives a self-sustaining thermal runaway. Higher nickel content generally lowers the onset temperature and raises the energy released.
In an LFP cell, the phosphate framework holds oxygen tightly, and decomposition of the cathode does not begin until much higher temperatures, often above 250 degrees Celsius, with a markedly smaller heat release. Thermal runaway can still occur because the electrolyte, separator and anode are common to both chemistries, but the cascade is slower, less energetic and generates less oxygen, and propagation between cells in a module is easier to prevent. At pack level, this allows simpler thermal barriers, reduces the safety margin that must be built into the structure and permits denser cell-to-pack designs.
The lifecycle consequence is large. LFP cells are commonly specified to retain 80 per cent of their capacity after somewhere between 3,000 and 6,000 full equivalent cycles, whereas nickel-rich automotive cells are more commonly specified for roughly 1,000 to 2,000. For a vehicle driven 12,000 miles a year with 4 miles per kWh and a 60 kWh pack, annual throughput is about 3,000 kWh, or 50 full equivalent cycles. Over 15 years the pack completes fewer than 800 equivalent cycles. In that regime the limiting factor for both chemistries is calendar ageing and not cycle ageing, which is why dwell voltage matters so much for nickel and so little for LFP. For heavy commercial vehicles and taxis, which can accumulate several hundred cycles a year, the cycle life advantage of LFP becomes the dominant economic variable.
| Cell Chemistry Variant | Recommended Daily Charge Cap | Primary Microscopic Degradation Mechanism at High SoC | BMS Telemetry Tracking Challenge | Long-Term Storage Target SoC (Parking >3 Weeks) | Low SoC Vampire Drain Hazard |
|---|---|---|---|---|---|
| Lithium iron phosphate (LFP) | 80 per cent on routine days; one scheduled 100 per cent saturation charge each week to recalibrate | Mild but real: SEI thickening on the deeply lithiated graphite anode and slow loss of lithium inventory; the olivine cathode itself is stable at its upper voltage | Very flat OCV plateau (about 3.2 to 3.3 V) plus 10 to 30 mV hysteresis; SoC drifts without a top-of-charge reset | 40 to 50 per cent, even though LFP tolerates high SoC better; the flat curve makes low SoC hard to observe | Continuous telematics drain over weeks can carry cells towards the low-voltage cut-off; below about 2.0 V the anode potential rises far enough to dissolve copper |
| Lithium manganese iron phosphate (LMFP) | About 80 per cent routine; periodic 100 per cent per manufacturer guidance, as the chemistry is less standardised | Manganese dissolution (mitigated by coatings and electrolyte design) alongside graphite SEI growth | Two-step voltage profile gives slightly more information than LFP but is still flatter than layered oxides | 40 to 50 per cent | Same copper-dissolution boundary; manganese-containing cathodes add sensitivity to deep discharge |
| Mid-nickel NMC (such as 532 or 622) | About 80 per cent for routine daily use; 100 per cent only for long journeys | Surface reconstruction and electrolyte oxidation above about 4.1 V; moderate transition-metal dissolution | Steeper OCV curve makes SoC estimation straightforward; the main challenge is ageing-model accuracy | 40 to 50 per cent | Over-discharge below about 2.0 V per cell risks copper dissolution and an internal short on recharge |
| High-nickel NMC (such as 811) and NCA | 70 to 80 per cent for routine daily use; 100 per cent sparingly and only just before departure | Lattice contraction and micro-cracking from the H2 to H3 transition, oxygen release and accelerated SEI growth | Accurate SoC from voltage is possible, but impedance growth makes power estimation more complex with age | 40 to 50 per cent; never store full, particularly in warm conditions | Highest consequence: a deeply discharged high-nickel pack has the least tolerance for copper dissolution and subsequent plating |
Long-Term Parking and Storage: The Two Boundaries
Daily charging policy is only half of the lifecycle problem. A vehicle left standing for three weeks or more, whether it is a second car, a seasonal vehicle, a demonstrator or a fleet asset waiting for a contract, faces a different pair of hazards at opposite ends of the voltage window. The upper hazard is parasitic chemistry at high potential; the lower hazard is irreversible damage when the pack drifts towards empty. The safe storage window sits between them, at roughly 40 to 50 per cent SoC for every lithium-ion chemistry, and it applies to LFP as well as to nickel packs.
The upper boundary: storage at 100 per cent
Parasitic side-reactions consume lithium and thicken the SEI whenever the cell is at a high potential, and they do so whether or not any current is flowing. Their rate follows an Arrhenius relationship, increasing roughly exponentially with temperature, so the damage from a full pack in a hot environment is multiplicative and not additive: a rule of thumb is a doubling of the reaction rate for every 10 degrees Celsius rise. A car stored at 100 per cent in a closed car park at 35 degrees Celsius can lose several times as much capacity per month as the same car stored at 50 per cent in a cool garage. In nickel-rich cells the cathode contributes as well, with electrolyte oxidation and transition-metal dissolution continuing at the surface of a highly delithiated lattice. In LFP cells the cathode is largely spared, but the graphite anode is at its lowest potential and SEI growth proceeds. For this reason a long-stored pack should be left near the middle of the window, where both the anode and the cathode potentials are moderate.
The lower boundary: vampire drain and copper dissolution
A parked electric vehicle is not inert. The telematics unit, the cellular modem, the security system, the 12-volt battery charger and the BMS itself all draw power, and features such as remote climate control, camera monitoring and cabin overheat protection add to the load. A sleeping drain of around 20 watts, a plausible figure for a connected vehicle, amounts to roughly 0.5 kWh a day, or about 0.8 per cent of a 60 kWh pack per day. Cell self-discharge adds a further 2 to 3 per cent a month. Over six weeks, a pack that started at 50 per cent can fall to somewhere near 10 to 15 per cent; a pack that started at 20 per cent can reach zero in under a month. Actual figures vary widely between models and settings and should be measured, not assumed.
As the pack approaches empty, the weakest cell in the string approaches the low-voltage limit first. Below about 2.0 V per cell the graphite anode, which is normally held near 0.1 V, rises towards a potential at which the copper current collector oxidises. Copper dissolves as Cu+ and Cu2+ ions into the electrolyte. The damage may be invisible at the time, but when the pack is recharged the dissolved copper is reduced and deposits as metallic copper on the anode and through the separator, where it can form dendrites that create a soft or hard internal short circuit. The result can range from elevated self-discharge and reduced capacity to a thermal event, and the failure often appears only on the first recharge after a long deep discharge, which is why it is described as a latent hazard. The damage is also asymmetric: a cell that has suffered copper dissolution cannot be restored by recharging.
The BMS and vehicle architecture provide protection, but it is not unconditional. Contactors open and the system enters a protective state when voltage thresholds are reached, but the 12-volt auxiliary battery that powers the wake-up logic continues to discharge, and if it fails the high-voltage pack may be left without monitoring or the means to be recharged. Some vehicles will refuse to charge a pack that has fallen below a deep-discharge threshold and require service attention. The robust approach is to avoid reaching the protective region at all.
A practical storage protocol follows from these mechanisms. Set the pack to 40 to 50 per cent before parking, using the vehicle’s own storage or transport mode if one exists; disable cabin overheat protection, sentry or camera monitoring, remote wake-ups and scheduled pre-conditioning; keep the vehicle in a cool, dry and ventilated place; check the state of charge at intervals of two to four weeks and top it up to the target if it has fallen below 30 per cent; maintain the 12-volt battery with a suitable maintainer; and log each check. For fleet yards where vehicles are parked for a season, the log should record the date, the state of charge and the ambient temperature, so that the asset manager can demonstrate compliance with the warranty and the supplier’s storage guidance. When the vehicle is returned to service, charge it to the operational ceiling, and for an LFP vehicle complete a full saturation charge to recalibrate the state-of-charge estimate after the long Coulomb-counting gap.
Operational Advice by Duty Cycle
Matching the chemistry to the duty cycle and the tariff structure turns the preceding science into a practical policy. Three profiles are described below, each with an indicative cost consideration based on an off-peak tariff of 7p per kWh and a standard-rate tariff of 24.5p per kWh; readers should substitute their own rates.
Urban commuter cycles
A commuter covering 20 to 40 miles a day uses well under half of a typical 60 kWh pack. For a nickel-pack car, the optimum is to set the charge limit to 70 or 80 per cent, schedule charging to run in the cheapest tariff window and finish shortly before departure, and recharge when the state of charge falls to a moderate level instead of keeping it topped up. The saving is not just in cell life: the energy not stored at the top of the pack costs nothing. A reduction of ten percentage points of SoC represents 6 kWh, worth 42p at off-peak rates and £1.47 at standard rates, per full top-up. For an LFP car, the recommended pattern is a daily cap of 80 per cent, which limits SEI thickening at the anode, with one targeted 100 per cent saturation charge each week, preferably on the cheapest night and finishing shortly before departure, so that the BMS reaches the voltage knee and recalibrates.
Long-distance highway corridors
On a long journey, the constraints change. Fast charging power is limited by the cell’s charge-acceptance curve: as the SoC rises, the BMS tapers the current to hold the voltage below the upper limit and to avoid plating. For nickel packs, this taper begins at around 50 to 60 per cent and becomes steep above 80 per cent, so the time to add the last 20 percentage points can equal the time to add the first 60. The efficient pattern is to arrive at the charger with 10 to 20 per cent remaining, charge to 70 or 80 per cent and leave, accepting more frequent but shorter stops. Charging to 100 per cent at a rapid charger is justified only when the next leg exceeds the usable range, in which case it is best done as close to departure as possible, and with the battery preconditioned so that it is warm enough to accept charge without plating. For LFP packs, the taper is generally less steep in the mid-range but the cells are more sensitive to low temperature, so preconditioning matters even more. An LFP user on a long trip has the advantage that charging to 100 per cent at the end of the journey also resets the state-of-charge estimate.
Stationary and unmanaged fleet yards
In a depot, vehicles may sit for days or weeks between duty periods, and the charging policy is often set by a depot management system with little attention to chemistry. The dominant risk for nickel vehicles is prolonged dwell at a full state of charge, which is the worst combination for calendar ageing. The policy should hold stationary nickel vehicles at 50 to 60 per cent and complete the charge to the operational target only in the hours before the vehicle is needed. For LFP fleets, high-SoC dwell is less damaging at the cathode but still ages the anode, so the depot should cap routine charging at 80 per cent, shift charging into the cheapest tariff window, and schedule one full saturation charge for each vehicle every week. For seasonal or unmanaged yards, where vehicles may stand for more than three weeks, the policy should switch to the storage protocol described above: 40 to 50 per cent, telematics features disabled, and a logged check every two to four weeks. Fleet managers should log the date of each vehicle’s last full charge, because a vehicle that is never charged fully will exhibit growing range-estimate error and reduced usable capacity as the balance of its cells deteriorates.
| Duty Cycle | Nickel-Rich Packs (NMC and NCA) | LFP Packs | Cost Logic |
|---|---|---|---|
| Urban commuter | Daily limit 70 to 80 per cent; charge in the cheapest window and finish before departure | Daily limit 80 per cent; one scheduled 100 per cent saturation charge each week | Smart off-peak tariffs reward timed charging; unused SoC headroom costs nothing |
| Long-distance highway | Arrive at 10 to 20 per cent; charge to 70 to 80 per cent; precondition the battery | Precondition in cold weather; charge to 100 per cent at the journey’s end if a recalibration is due | Rapid charging is the most expensive energy; the taper above 80 per cent increases cost per minute |
| Seasonal stationary or unmanaged yard (more than 3 weeks) | Park at 40 to 50 per cent; disable telematics extras; check every 2 to 4 weeks; top up if below 30 per cent | Park at 40 to 50 per cent; same checks; complete a full saturation charge on return to service | Avoiding a deep-discharge failure is worth far more than the few pence of energy held in storage |
Software Implications for Systems Architects
Several design points follow for battery software engineers and automotive systems architects. First, the user interface should expose the charging ceiling as a chemistry-aware setting. A single slider that runs from 50 to 100 per cent hides the fact that the right default differs by pack, and an LFP vehicle should actively prompt for periodic full charges and report when the last reset occurred. Second, the BMS should treat the top-of-charge knee as a first-class calibration event: record the voltage, current and temperature at the knee, update the capacity estimate, and flag the estimator’s confidence so that downstream range prediction can widen its uncertainty band when no reset has occurred for a long time.
Third, estimation algorithms for LFP benefit from explicit modelling of hysteresis, such as a state-dependent hysteresis term in the equivalent-circuit model, and from the use of differential voltage and incremental capacity analysis, which can identify the electrode-specific features that survive the plateau. Fourth, the charge-scheduling layer should coordinate with the tariff and grid signals so that a full-saturation charge for LFP is aligned to the lowest-cost window, and the ‘finish by departure’ logic for nickel packs minimises dwell at high voltage. Fifth, fleet telematics should aggregate the time spent above defined SoC thresholds and the temperature during that time, since this ‘stress exposure’ metric is a better predictor of calendar fade than the average SoC.
For asset managers, these data feed directly into residual-value models. Battery state of health is increasingly a determinant of resale price and of eligibility for second-life applications. A fleet with logged charging behaviour that follows chemistry-appropriate profiles can support warranties and secondary-market valuations, whereas an undocumented fleet carries a discount for uncertainty. Asset-optimisation in this sense means treating the charging policy as a controllable variable in the total cost of ownership, alongside energy tariffs and utilisation.
Conclusion
The 80 per cent rule is a heuristic for one family of cathodes, not a law of electrochemistry. For nickel-rich cells, holding high voltage drives lattice strain, oxygen release, transition-metal dissolution and SEI growth, and a daily ceiling of 70 to 80 per cent, with charging timed to finish just before departure, measurably extends life. For LFP cells, the olivine lattice tolerates full charge, but the anode still ages at high SoC, so an 80 per cent daily cap with one weekly saturation charge is the balanced policy: a flat voltage curve leaves the BMS dependent on the voltage knee at about 3.6 V to reset its Coulomb counter and balance its cells, and the weekly full charge protects the accuracy of the software and the usable capacity of the pack. For every chemistry, long-term parking belongs at 40 to 50 per cent SoC: high enough to survive weeks of vampire drain without approaching the roughly 2.0 V boundary at which copper dissolves, and low enough to avoid the exponential parasitic reactions of a full, warm pack.
The engineering message is to match the policy to the chemistry and to treat the BMS as part of the system under optimisation. Owners should consult the vehicle handbook, which states the manufacturer’s tested guidance for the specific cells and software in the car. Fleet managers should record the chemistry of every vehicle and encode its charging profile in the depot management system. And systems architects should recognise that the most important question in battery management is not only how much energy the pack stores, but also how well the software knows how much it has.