Battery life forecast: the invisible energy that keeps the digital and energy world running

Data centres, substations, electric vehicles and energy storage systems all rely on batteries that give no warning when they lose capacity. Knowing how much life they have left —their battery life forecast— starts with measuring their state of health (SOH). We look at why it matters, which criteria IEEE and IEC standards set for each chemistry and how to obtain the estimated remaining life with the Amperis online tool.

In an era when artificial intelligence demands data centres with uninterrupted availability, electric vehicles are transforming urban mobility and battery energy storage systems (BESS) offset the intermittency of renewables, the power grid depends on a component as crucial as it is silent: the battery. In critical infrastructure —from high-voltage substations to cloud computing centres— a battery rarely fails unexpectedly: it fails through neglected degradation.

That is why the battery life forecast is no longer just another technical indicator but the foundation of operational resilience. It is based on the state of health, usually abbreviated to SOH, which shows how much of its original capacity a battery retains and, therefore, whether it will still deliver the backup time it was designed for on the day it is needed, and for how much longer it can do so.

Why is state of health the definitive metric?

Unlike the state of charge (SOC), which reflects how much energy is available at a given moment, the state of health measures the battery's maximum usable capacity and the increase in its internal resistance compared with its as-new condition. Over time, cycling, temperature and the chemistry itself wear it down: in lead-acid, sulphation, grid corrosion and electrolyte loss; in lithium, loss of active lithium, growth of the passivation layer (SEI) and metallic lithium plating (dendrites). The result is always the same: less energy delivered when it is needed.

The consequence is stark: a battery at 100 % charge but with a state of health of 60 % delivers only 60 % of the backup it was designed for. In a data centre carrying AI loads of hundreds of megawatts, or in a substation that must trip its protection on a short circuit, relying on a degraded battery is like operating without a safety net.

How state of health is determined: capacity first, resistance as a warning

The primary state of health is the capacity-based SOH (SOH-C): the battery's actual capacity, obtained from a controlled discharge test at constant current or constant power down to the end voltage, corrected to the reference temperature and compared with the manufacturer's rated capacity at the same rate. It is the replacement criterion in every maintenance standard and the only one that proves the real backup time.

The resistance-based state of health (SOH-R) is an indirect indicator: the internal resistance, impedance or conductance of each cell, compared with its commissioning value or the manufacturer's value. It does not replace the capacity test, but it detects weak cells and trends without discharging the battery. In lead-acid, a 20 % increase over the baseline is already a warning sign (IEEE 1188), and 50 % is a replacement criterion according to Best Practices.

All capacity values refer to a reference temperature, normally 25 °C, because a cold battery delivers less. Best Practices applies a linear correction with a coefficient of 0.006 per °C for lead-acid and 0.002 per °C for nickel-cadmium; IEEE 450 provides its own tables for vented lead-acid, and IEEE 2962 those for lithium.

The regulatory framework: which standards require it

To ensure reliability at scale, the industry follows international standards that define how batteries are monitored, assessed and retired. These are the main ones. Those marked criterion are the ones the Amperis tool uses to reach its verdict; those marked reference complete the framework in terms of safety, installation and mobility.

On this basis, the tool applies the Best Practices criteria by default: a compilation, prepared by Amperis, of the recommendations of the international standards and of the most widely accepted battery manufacturers in the sector. It brings together in a single framework the end-of-life limits for capacity and resistance, voltage and specific gravity tolerances, temperature correction and minimum insulation. Where Best Practices and a specific standard differ, the tool applies Best Practices and reports the alternative given by the standard.

One chemistry, one criterion

Each technology ages differently and has its own standard. That is why a single threshold is not enough: float voltage, cell-to-cell tolerance, temperature correction and end of life change from one chemistry to another.

Lead-acid: vented (VLA) and valve-regulated (VRLA)

This is the most widespread technology in backup applications: substations, telecommunications and UPS. In vented batteries the electrolyte specific gravity is also measured and water is topped up; VRLA batteries (AGM and GEL) need no electrolyte maintenance, but they are very sensitive to temperature and overvoltage. IEEE 1188 calls for measuring their internal resistance every quarter, and IEC 60896-21 includes a thermal runaway test: excessive float voltage or an overheated room can trigger it. A bulging VRLA is grounds for immediate replacement.

Nickel: industrial nickel-cadmium and nickel-metal hydride

Nickel-cadmium is common in substations, railways and industry thanks to its robustness against cold and cycling. When its capacity falls below 80 %, the answer is not immediate replacement but reconditioning: full discharge, constant-current charge and a new test. Nickel-metal hydride is used in portable equipment; IEC 61951-2 sets its test conditions, at 20 ± 5 °C and without temperature correction.

Lithium and sodium: end of life depends on the application

In lithium, end of life is not a single number: a backup battery is retired at 80 %, a BESS or an electric vehicle usually at 70 %, and a second-life battery can operate down to 60 %. IEEE 2962 defines how to test them —with the battery practically full at the start— and how to correct for temperature. The SOH calculated by the BMS must be checked against periodic tests, and a difference of more than 30 mV between cells calls for balancing. Unlike lead-acid, capacity loss in lithium is irreversible: it cannot be regenerated. Sodium-ion, the emerging technology, follows similar criteria that are still under development.

From health to forecast: how much life is left

The state of health tells you how the battery is today; the battery life forecast answers the question that really matters for planning: how much time does it have left? To estimate it, three sources are combined and the least favourable is taken:

• The capacity trend. The capacity loss per year —between two tests or since commissioning— is projected to the end of life for the chemistry. It is the most reliable estimate when test data are available.
• The calendar. The manufacturer's design life, corrected for the average operating temperature, minus the age of the battery. In lead-acid, a room at 33 °C halves the expected life.
• The cycles. In cyclic service, the rated cycle life remaining at the current rate of use.

The result is an estimated remaining life, with the expected end-of-life year, a range and a confidence level. It is an indicative extrapolation: in lead-acid, capacity falls faster as it approaches end of life, so the test should be repeated according to the schedule in the standard and the forecast updated.

Good practice for effective monitoring

Manufacturers of batteries and battery management systems (BMS) agree that modern monitoring must go beyond measuring voltage. These are the practices that make the difference:

• Regular tracking of internal resistance. Its increase is one of the most reliable indicators of chemical degradation and gives warning before capacity loss becomes critical. It is only comparable when measured with the same type of instrument and against a baseline from the battery itself.
• Smart BMS and cloud models (digital twins). Combining real-time measured data with algorithms that emulate the degradation of each chemistry —LFP, NMC, sodium or solid-state— makes it possible to anticipate how the state of health will evolve depending on temperature and usage profile. Even so, the BMS SOH must be checked against tests.
• Capacity tests with controlled discharge. However far continuous analysis advances, periodic discharges under standardised parameters remain the reference for validating the real state of health: with the battery fully charged, recording the initial temperature and the voltage of each cell.
• Rigorous thermal control. Temperature is the leading cause of ageing. In lead-acid, every 8 °C above 25 °C halves battery life: keeping the room between 20 and 25 °C can double the service life of the bank compared with a room at 33 °C.
• Baseline and trend. Record the commissioning values and compare each test with the previous one. A loss of more than 10 points between tests is a reason for closer observation, according to IEEE 450 and IEEE 1188.

Try the Amperis Battery life forecast

To help you take the first step, Amperis has developed an online tool that applies all these criteria to your battery data. It runs in the browser, on desktop and mobile, and only assesses what you have measured: missing data are not penalised but appear as a pending test, together with the standard that requires it.

1. Choose the chemistry in its tab —lead-acid, nickel, or lithium and sodium— and the application: the standards, tolerances and end of life change accordingly.
2. Enter whatever data you have: the capacity test, internal resistances, cell voltages, temperatures, electrolyte, age and the findings of the inspection.
3. Click “Diagnose”. You will get the verdict, the state of health and the estimated remaining life, with its confidence level, one card per indicator with the applicable standard, actions ranked by urgency and the reference test plan.
4. Print the report or request a review by an Amperis engineer with the diagnosis attached.

The diagnosis also recommends the Amperis equipment suited to the findings —capacity analysers, testers, chargers or regeneration—. In each case it starts with the equipment that covers the need, continues with higher-performance models and shows which requirements of the standard each one meets and which it exceeds.

The result is indicative and is calculated solely from the data provided: it does not replace a thorough diagnosis by a qualified engineer.

Conclusion

In an electrified ecosystem where the digital economy, the energy transition and artificial intelligence converge, batteries are no longer auxiliary components but primary infrastructure assets. Knowing their state of health and forecasting their life in accordance with international standards not only prevents serious failures and substantial financial losses: it ensures that energy is available exactly when it is needed most.