Why is mathematics important when a battery seems to have only two states: charged or flat? Real batteries are not fuel gauges with perfect markings. They store charge through electrochemical reactions, deliver current at changing voltage, lose energy as heat, age differently under different temperatures and loads, and must be estimated from measurements that contain uncertainty. Mathematics turns those changing signals into useful ideas such as capacity, energy, state of charge, efficiency, power and state of health.
This matters in everyday life. A phone that reports 40 per cent remaining is making an estimate, not looking into a tiny tank. An electric vehicle predicts range from available energy, recent consumption and operating conditions. A solar-storage system balances generation, demand, charge limits and reserve. In each case, the percentage is the visible result of a model.
The mathematics includes unit conversion, proportional reasoning, integration, graphs, exponential and piecewise models, statistics, calibration and optimisation. It is useful because it lets us ask better questions: 40 per cent of which measured capacity? At what current and temperature? Under what voltage limits? With what uncertainty?
This article uses simplified worked examples to show the mechanisms. It is educational, not a charging, repair or safety manual. Battery chemistries, packs and management systems differ. For real devices, follow the manufacturer’s instructions, use approved equipment and seek qualified technical support when a pack is damaged, swollen, hot or otherwise abnormal.
Choose the battery question you want to answer
- To understand the basic quantity, begin with charge, current and time.
- To estimate runtime, connect ampere-hours to the load while keeping assumptions visible.
- To compare devices fairly, distinguish charge capacity from energy capacity.
- To understand the percentage display, study state of charge as an estimated state.
- To see why heavy loads matter, separate energy, power and internal loss.
- To understand ageing, distinguish reversible state changes from long-term capacity fade.
- To judge a model, look at calibration, uncertainty and changing operating conditions.
- To build skill, measure safe low-voltage examples or analyse public data rather than opening a battery.
The first habit is simple: write the quantity and unit beside every number. Amperes, ampere-hours, volts, watts and watt-hours are related, but they are not interchangeable.
Charge is current accumulated through time
Electric current is the rate at which charge moves. In symbols, I=dQ/dt. If current is constant, charge transferred is Q=It. One ampere is one coulomb per second. Engineers also use ampere-hours: one ampere-hour equals 3,600 coulombs because an hour contains 3,600 seconds.
Suppose a small load draws 0.40 A for 2.5 hours. The transferred charge is 0.40×2.5=1.00 Ah, or 3,600 C. This calculation is exact only under the constant-current assumption. If current changes, add the contributions over small intervals or integrate the current function.
This accumulation idea is broader than batteries. It is the same mathematical relationship that turns speed into distance, flow rate into volume and power into energy. A rate becomes a total by summing it over time.
Capacity is a tested quantity, not the battery’s physical volume
Battery capacity commonly describes how much charge can be delivered under stated test conditions before a defined endpoint. A label such as 3,000 mAh means 3.0 Ah, not that the cell always supplies exactly 3.0 A for one hour in every situation.
The usable result depends on chemistry, current, temperature, ageing, voltage limits and the testing method. A management system may also reserve part of the electrochemical range to support safety or life. Capacity therefore needs conditions, just as a sports time needs a distance and a laboratory measurement needs a procedure.
The U.S. Department of Energy notes that battery power and capacity fade with cycling and calendar time, while the mechanisms depend on chemistry, operating profile and ambient conditions. Mathematics helps separate the tested definition from the tempting but inaccurate image of a fixed tank.
Milliampere-hours must be converted before calculation
Prefixes create many avoidable mistakes. One ampere is 1,000 milliamperes, so 2,500 mAh is 2.5 Ah. If a device draws 500 mA, that is 0.5 A. Keeping both quantities in matching units gives the idealised time 2.5 Ah ÷ 0.5 A=5 h.
The word idealised matters. That division assumes the labelled capacity is available at the stated load and conditions, the current stays constant, conversion electronics have no loss, and the cutoff permits use of the whole tested capacity. A real runtime can be shorter or occasionally differ for other reasons.
A useful classroom check is dimensional cancellation: Ah divided by A leaves h. If the units do not reduce to time, the setup needs attention.
Energy adds voltage to the picture
Charge capacity alone cannot compare batteries with different voltages. Energy is related to voltage and charge. Under a simple constant-voltage approximation, E=VQ. When V is in volts and Q in ampere-hours, E is in watt-hours.
A nominal 3.7 V, 3.0 Ah cell has an approximate nominal energy of 3.7×3.0=11.1 Wh. A 12 V, 3.0 Ah battery has approximately 36 Wh. Both say 3.0 Ah, yet their energy differs because each unit of charge moves through a different potential difference.
Voltage changes during discharge, so a more accurate calculation uses E=∫V(t)I(t)dt. Nominal voltage is useful for comparison, but it compresses a changing curve into one representative number.
Power describes how quickly energy is delivered
Power is energy per unit time. Electrically, P=VI at an instant. A 10 V device drawing 2 A uses 20 W. If it stayed at 20 W for three hours, the energy would be 60 Wh.
Energy and power answer different questions. Energy helps estimate how long an activity can continue. Power helps determine whether the battery and electronics can support the required rate. A battery might contain plenty of energy yet be unsuitable for a short, high-power demand.
Students can compare this with water. Volume resembles stored energy; flow rate resembles power. A large tank with a narrow outlet may deliver slowly. The analogy is imperfect because voltage, current and electrochemistry behave differently, but it highlights why “how much” and “how fast” must not be collapsed.
A worked runtime estimate needs efficiency
Imagine a 48 Wh battery powering a device that requires 12 W. An ideal division gives 48÷12=4 h. If the conversion path is 85 per cent efficient and all 48 Wh is otherwise usable, energy reaching the device is 0.85×48=40.8 Wh, giving 40.8÷12=3.4 h.
This is still a model. The load may vary, efficiency may depend on power, voltage may reach the cutoff early, and temperature may reduce usable energy. The calculation is valuable because it exposes assumptions instead of hiding them inside a confident single number.
A better answer is often a range. If usable energy is between 38 and 42 Wh and average load is between 11 and 13 W, plausible runtime spans roughly 38÷13=2.92 h to 42÷11=3.82 h. Sensitivity analysis shows which uncertainty matters most.
State of charge is an estimated hidden state
State of charge, often abbreviated SOC, expresses remaining charge relative to a reference usable capacity. A simple definition is SOC=remaining usable charge ÷ reference capacity. The numerator cannot usually be observed directly while a battery is operating.
The management system therefore estimates it from measurable signals such as current, voltage and temperature, together with a model and calibration history. This is a state-estimation problem: use imperfect observations to infer an internal condition.
The percentage is not a universal chemical truth. It depends on how the system defines zero and full, how the reference capacity has changed with age, and how the estimator corrects drift. Two devices using the same cell could display percentages differently while each follows its own control policy.
Coulomb counting is an accumulation model
One common idea is to update charge by integrating current. With discharge current treated as positive, a simplified update is Qremaining(t)=Qstart−∫I(t)dt. In discrete time, add rectangular pieces: ΔQ≈I×Δt for each interval.
Suppose a 4.0 Ah reference begins at 75 per cent, so the estimate is 3.0 Ah. The device draws 0.8 A for 30 minutes and 0.2 A for the next hour. Estimated charge removed is 0.8×0.5+0.2×1=0.6 Ah. Remaining charge is 2.4 Ah, or 60 per cent of the 4.0 Ah reference.
This simple bookkeeping is transparent. Its weakness is that sensor bias and uncertain starting values accumulate. NREL documentation on ampere-hour counting explains the importance of starting from a known state and periodically resetting the count against a known state of charge.
Small current errors can become large state errors
Assume a current sensor reads 20 mA too high. Over ten hours, the estimated extra discharge is 0.020 A×10 h=0.20 Ah. On a 2.0 Ah reference, that is a ten-percentage-point error.
Random noise may partly cancel when integrated; a systematic offset does not. This distinction between precision and bias is central to measurement. More decimal places do not repair a biased sensor.
Battery estimation therefore needs calibration and correction opportunities. A system may use voltage behaviour, known rest conditions, model comparison or other evidence to reduce accumulated drift. The exact method is product-specific, but the mathematical lesson is general: integration magnifies persistent bias.
Voltage is informative but not a perfect fuel gauge
Open-circuit voltage can relate to state of charge, but the relationship depends on chemistry, temperature, ageing and how long the cell has rested. Under load, terminal voltage also includes voltage drops associated with internal resistance and dynamic electrochemical processes.
A simplistic rule that maps voltage linearly from “empty” to “full” may work poorly when the true curve has flat regions. In a flat region, a small voltage measurement error can imply a large SOC change. During load transients, voltage can dip and recover without the same amount of chemical charge disappearing.
The useful mathematical object is a calibrated curve or model, not a universal straight line. Interpolation within a verified table can be sensible; extrapolation beyond it is risky.
Internal resistance creates voltage drop and heat
A basic equivalent-circuit model writes terminal voltage during discharge as Vterminal≈Voc−Ir, where Voc is an open-circuit-like voltage, I is current and r is internal resistance. The resistive heat rate is Ploss=I²r.
If r=0.08 Ω and I=1 A, the drop is 0.08 V and resistive loss is 0.08 W. At 3 A, the drop is 0.24 V and the loss is 9×0.08=0.72 W. Tripling current multiplies this idealised loss by nine.
Real batteries need richer dynamic models, but the square relationship explains why high-current operation can be disproportionately demanding. It also shows why a voltage cutoff can be reached earlier under heavy load even when some charge remains.
Series connections add voltage
When identical cells are connected in series, the same current passes through each and their voltages add. Four 3.6 V nominal cells in series give approximately 14.4 V nominal. The series string’s ampere-hour capacity remains that of one matched cell under the simplified model.
Energy increases because voltage increases: four 3.6 V, 2.5 Ah cells contain about 4×3.6×2.5=36 Wh nominally. It would be incorrect to multiply both voltage and ampere-hours by four for a series-only connection; that would count the added cells twice.
Series packs also require attention to cell balance. The weakest cell can reach a limit first. Pack behaviour is not captured by the average cell alone.
Parallel connections add charge capacity
Matched cells connected in parallel share current and keep approximately the same voltage while their charge capacities add. Four 3.6 V, 2.5 Ah cells in an ideal parallel group give 3.6 V and 10 Ah, again about 36 Wh.
The series and parallel examples contain the same four cells, so total nominal energy is the same. The arrangement changes voltage, current sharing and system design.
Real parallel packs require appropriate engineering because unequal resistance, state and temperature can produce unequal current sharing. The arithmetic explains the first-order relationship; it does not replace protection design or manufacturer requirements.
Pack notation is a compact multiplication map
A pack described conceptually as 4s2p has four cell groups in series and two matched cells in each parallel group. With 3.6 V, 2.5 Ah cells, ideal nominal pack values are 4×3.6=14.4 V, 2×2.5=5.0 Ah, and 14.4×5.0=72 Wh.
The same energy appears by counting eight cells: 8×3.6×2.5=72 Wh. Reaching the answer by two routes is a powerful check.
Notation alone does not specify safe limits, cell chemistry, balancing, enclosure, fusing, thermal design or certification. A number model is useful only inside the physical and safety context.
State of health needs a reference
State of health, or SOH, is not one universally measured property. A capacity-based metric might be SOHcapacity=measured full capacity ÷ rated or beginning-of-life capacity. A power-oriented metric may involve resistance or maximum deliverable power.
If a battery originally delivered 60 Ah under a stated test and later delivers 48 Ah under comparable conditions, capacity-based SOH is 48÷60=0.80, or 80 per cent. Changing the test conditions would weaken the comparison.
The denominator matters. Rated capacity, first measured capacity and modelled initial capacity can differ. A careful report names the test, temperature, rate, endpoint and reference rather than presenting “80 per cent health” as self-explanatory.
Capacity fade and charge level are different
A well-used battery can be fully charged relative to its present capacity yet store less than when new. If a battery has faded from 50 Ah to 40 Ah, a displayed 100 per cent may correspond to about 40 Ah of present usable reference, not 50 Ah.
This is an important distinction between a fast-changing state variable and a slowly changing parameter. SOC moves during each charge and discharge. Capacity-based SOH changes gradually with ageing and use.
Confusing them leads to statements such as “the battery will not charge past 80 per cent” when the display actually reaches 100 per cent but runtime has fallen. Mathematics gives separate names and ratios to separate phenomena.
Cycle life is not simply a count of plug-ins
A cycle is often discussed in equivalent full-cycle terms rather than literal charging events. Two discharges of 50 per cent of capacity can sum to approximately one equivalent full cycle, although degradation is not determined by cycle count alone.
Suppose daily use removes 30 per cent of current usable capacity, then recharges it. Over ten days, throughput is about three equivalent full cycles. This calculation helps compare usage patterns, but it does not predict life by itself.
DOE material emphasises that degradation depends on chemistry, operating profile and ambient conditions. Temperature, time at certain states, charge and discharge rates, and depth of cycling can all matter. A cycle counter is one explanatory variable, not a warranty clock.
Calendar ageing continues even when the battery rests
Some degradation processes depend on time as well as cycling. This is called calendar ageing. A battery stored unused is therefore not frozen in its original condition.
A simple model might fit capacity as C(t)=C0−kt over a limited period. Another might use a square-root-of-time or Arrhenius-inspired temperature relationship. The correct form depends on evidence and range. Extrapolating a short straight line over many years can produce nonsense.
The broader lesson is that time series can contain multiple mechanisms. Separating cycle-related throughput from time-related ageing requires experimental design or careful observational data, not a single before-and-after reading.
Temperature is a variable, not background scenery
Temperature affects electrochemical kinetics, resistance, usable power and ageing. A cold battery may show reduced performance during use, while high temperature can accelerate unwanted reactions in many chemistries.
When comparing tests, temperature should be controlled or recorded. If one capacity test occurs at 25°C and another at a very different temperature, attributing the whole difference to permanent ageing may be wrong.
Mathematically, temperature can enter a model as a predictor, an interaction or a boundary condition. Practically, users should follow manufacturer guidance. The article does not turn a fitted temperature curve into permission to heat, cool or modify a battery.
Efficiency has more than one definition
Coulombic efficiency compares charge out with charge in. Energy efficiency compares watt-hours out with watt-hours in. Because charging and discharging voltages can differ, the two percentages are not identical.
If 10.0 Ah enters and 9.8 Ah later leaves under a stated test, coulombic efficiency is 98 per cent. If 40 Wh enters and 34 Wh leaves, energy efficiency is 85 per cent. Both can be true.
Round-trip system efficiency may also include converters, wiring, control electronics and standby consumption. A claim about “battery efficiency” should name the system boundary. Mathematics makes the boundary visible in the numerator and denominator.
Rate capability changes what is usable
A battery’s measured capacity can depend on discharge rate. For some chemistries, higher current reduces the capacity obtained before the endpoint; the relationship can be described empirically, such as with Peukert-type models for lead-acid batteries.
It is unsafe to copy one chemistry’s empirical exponent into another. The concept to learn is conditional measurement: the same physical battery can yield different tested capacity under different rates because losses and dynamics change.
A graph of delivered capacity against current can show the pattern more honestly than a single label. Students should identify the test range and avoid extrapolating beyond observed currents.
Variable loads require summing energy over time
A laptop may draw 8 W while reading, 25 W during heavy computation and 3 W while sleeping. Runtime cannot be estimated from one instantaneous value unless that value represents the future average.
For a schedule of one hour at 8 W, half an hour at 25 W and two hours at 3 W, energy demand is 8×1+25×0.5+3×2=26.5 Wh. If the usable battery energy is 53 Wh and other losses are already included, that pattern could repeat twice.
This piecewise calculation teaches integration without calculus notation. Each rectangle has height power and width time; its area is energy.
Energy consumption per task can be more useful than hours
When workload varies, measure energy per completed task. If a drone inspection, data-logging session or portable medical check consumes 4 Wh, a 40 Wh usable budget supports ten such idealised tasks before reserves and uncertainty.
The task metric connects mathematics to planning. It also prevents misleading comparisons where one device finishes work faster but draws more power. Energy is power multiplied by time, so higher power can still use less energy if the job ends much sooner.
A responsible plan includes a reserve and does not treat the last theoretical watt-hour as guaranteed availability. The size of that reserve depends on consequences, variability and verified operating guidance.
Battery percentages should not be averaged blindly
Suppose two independent packs show 80 per cent and 20 per cent. Their average is 50 per cent only if the percentages refer to equal usable capacities and compatible definitions.
If pack A has 10 kWh usable capacity and pack B has 2 kWh, remaining energy is 0.8×10+0.2×2=8.4 kWh out of 12 kWh, or 70 per cent. The unweighted mean gives the wrong fleet-level result.
This is the same weighted-average principle used in grades, surveys and portfolios. Percentages carry denominators. Before averaging, recover the underlying amounts.
Range estimates combine battery energy with consumption
If an electric vehicle has 54 kWh usable and recent consumption is 18 kWh per 100 km, a simple estimate is 54÷18×100=300 km. If only 70 per cent remains, energy is 37.8 kWh and the same rate gives 210 km.
Actual consumption changes with speed, traffic, temperature, terrain, payload, accessories and driving. The estimate should update as conditions change. It is a quotient of an uncertain stock and an uncertain rate.
The mathematics is valuable even when it cannot promise a distance. It explains why range displays move and why recent driving may influence the prediction.
Uncertainty belongs in the result
Assume usable energy is estimated as 45±2 Wh and average power as 9±1 W. The central runtime estimate is 5 h. A conservative interval from endpoint combinations is 43÷10=4.3 h to 47÷8=5.875 h.
This is not a formal probability interval unless the uncertainty statements are probabilistic. It is a scenario range. The distinction should be stated.
Students can also use relative-error approximations or Monte Carlo simulation when distributions are justified. The main habit is to stop reporting a model output with more certainty than its inputs support.
Regression can model degradation without proving its cause
Suppose monthly tests show capacity trending downward. A linear regression may estimate a slope in ampere-hours per month. Residual plots may reveal temperature effects, abrupt changes or curvature.
The fitted slope describes association over the observed range. It does not prove that time itself caused every change. Usage intensity, storage conditions, software updates and measurement procedures may have changed together.
Good analysis records covariates, repeats tests under comparable conditions and distinguishes prediction from causal explanation. A smooth line is a summary, not a microscope into battery chemistry.
Outliers may be faults, events or information
One capacity reading far below the trend could result from an incomplete charge, altered cutoff, cold test, sensor problem, unusual load or real damage. Automatically deleting it would hide evidence; automatically accepting it as permanent failure would also be premature.
A disciplined workflow checks the raw data, procedure, timestamps, units and operating notes. If the value is excluded, the reason is documented. Results can be reported with and without it.
This is data literacy in action. An outlier is a question to investigate, not a nuisance to erase.
Optimisation balances competing objectives
A battery system may aim to maximise available service, minimise degradation, reduce cost, keep temperature within limits and preserve reserve for unexpected demand. These goals can conflict.
Mathematically, designers express decision variables, constraints and an objective. A simple scheduling problem may decide when to charge a storage system given varying demand and generation, subject to SOC and power limits.
The “best” schedule depends on chosen priorities and forecasts. Optimisation does not decide values for society; it finds good decisions under the values, limits and data supplied.
Battery management is a control problem
A battery management system observes voltage, current and temperature, estimates internal state, and applies limits or commands. That is a feedback loop: measurement, estimation, decision and action repeat over time.
If the model underestimates resistance, a predicted voltage under load may be too high. If the current sensor drifts, SOC may drift. Robust control considers model error and protective margins.
The mathematics connects algebra, calculus, statistics and computing. It also creates careers in electrical engineering, control, data analysis, materials science and software, but mathematics alone does not guarantee entry. Students also need physics, chemistry, practical judgement, communication and safe professional practice.
A safe student investigation
Students can analyse manufacturer-published data or a safe supervised low-voltage setup without dismantling cells. One project is to log the runtime of an approved rechargeable device under two ordinary workload settings.
Record starting display percentage, task type, elapsed time and ending percentage across repeated trials. Do not force deep discharge or alter charging equipment. Plot percentage change against time, compare slopes, and explain why the display is not a laboratory charge measurement.
The aim is not to “test the battery to destruction”. It is to practise defining variables, controlling conditions, recording uncertainty and separating observed patterns from conclusions.
Common misconceptions to repair
- “Ampere-hours are energy.” They measure charge capacity; voltage is needed for a watt-hour comparison.
- “A 100 per cent display means new-battery capacity.” It usually means full relative to the system’s current reference and limits.
- “Two 50 per cent packs average to 50 per cent.” Only if their usable capacities and definitions match.
- “A bigger battery always gives proportionally longer runtime.” Load, efficiency, cutoff and rate effects may change.
- “Cycle count alone predicts failure.” Calendar time, temperature, depth, rate and chemistry also matter.
- “A fitted trend proves the cause.” Regression summarises data; causal claims need stronger evidence.
- “More decimal places make the estimate accurate.” Sensor bias and model error remain.
A practical learning path for students
Start with unit fluency: convert milliamperes to amperes, minutes to hours and watt-hours to joules. Next, solve constant-rate problems and check dimensions. Then handle piecewise loads, weighted percentages and energy efficiency.
After that, plot voltage, current and estimated SOC over time. Learn why numerical integration accumulates both signal and bias. Compare linear, exponential and piecewise degradation models, then inspect residuals rather than trusting one score.
Students interested in engineering can continue into circuits, differential equations, estimation and control. Those interested in data can study uncertainty, regression and time series. The transferable skill is to connect a number on a screen to the measurement and model that produced it.
What parents can encourage
Parents do not need specialist equipment to support this learning. Ask the student to explain the units on a charger label, distinguish watts from watt-hours, and estimate the energy cost of an ordinary task using approved public information.
Encourage answers that name assumptions. “About four hours if the average load remains 12 W and 48 Wh is usable” is stronger than “four hours exactly”. Praise a student who notices that percentages need denominators.
Keep safety boundaries firm. Curiosity belongs in calculation, public datasets and supervised approved activities, not puncturing cells, improvising chargers or bypassing protection.
Did you know? The same integral appears across technology
Charge is the integral of current. Energy is the integral of power. Distance is the integral of speed. Water volume is the integral of flow rate.
These are not four unrelated formulas. They share one mathematical idea: a changing rate accumulates into a total. Once students recognise that structure, mathematics begins to transfer between electricity, transport, water and mechanics.
That is one reason the benefits of learning mathematics extend beyond examinations. A student learns to identify what is changing, what is accumulating, what the units mean and which assumptions make the calculation valid.
Frequently asked questions
Why does my battery percentage sometimes fall quickly and then slow down?
The display is an estimate based on a model, measurements and a mapping from estimated state to percentage. Voltage response, load and temperature can change during use. The displayed scale need not fall linearly with time, especially when power demand varies.
Is 1 Ah always equal to 1 Wh?
No. Ampere-hours measure charge, while watt-hours measure energy. Under a constant-voltage approximation, watt-hours equal volts times ampere-hours. A 1 Ah battery at 3.7 V is about 3.7 Wh nominally; at 12 V it is about 12 Wh.
Can I calculate exact runtime from the label?
Usually not. You can build an estimate, but usable capacity, voltage, efficiency, load, cutoff, temperature and ageing matter. State the assumptions and use a range when consequences matter.
Why does high current cause more heating?
In a simple resistive model, heat power is I²r. Doubling current quadruples that component of loss. Real batteries have additional electrochemical dynamics, so professional models are richer.
Does charging more often mean more cycles?
Not necessarily one full cycle per plug-in. Partial throughput can be combined into equivalent full cycles. Ageing also depends on time and operating conditions, not only the count.
Which school mathematics matters most?
Ratio, percentages, units, graphs and algebra are useful first. Functions, calculus and statistics support changing loads, accumulated charge, estimation and degradation models. Physics and chemistry provide the mechanism.
Does learning battery mathematics guarantee an engineering career?
No. It builds useful foundations for several pathways, but careers also depend on wider study, practical skills, communication, safety practice and opportunities. Keep options open while building transferable capability.
Useful next reading
Continue with Why Mathematics? | Household Electricity, Power and Energy Use to strengthen the distinction between power and energy. Compare generation and storage through Why Mathematics? | Solar Panels, Sun Angles and Energy Yield and Why Mathematics? | Wind Turbines, Blade Angles and Power Curves.
For the official evidence behind the battery-ageing cautions, see the U.S. Department of Energy technology assessment on battery life and degradation, the DOE article Extending the Life of Lithium-Ion Batteries, and the NREL discussion of ampere-hour counting from a known state of charge.
The practical conclusion is optimistic: a battery percentage is not magic. It is a careful meeting of measurement, units, accumulated rates, models and uncertainty. Learning that mathematics helps students read technology with more confidence and less guesswork.