Key takeaways
A battery pack under a vehicle spends its life in the worst environment on the car. Road spray, salt, standing water, pressure washing and thermal cycling all act on the same enclosure, and what it contains reacts badly to water.
Leak testing is how a plant confirms the seal on every pack. It sounds like a simple measurement and is one of the more technically demanding tests on the line, because the thing that most affects the result is not the leak.
This article covers what the enclosure has to achieve, why ingress ratings cannot be tested in production, the available methods and the errors that corrupt them. The wider test sequence is covered in the guide to end-of-line testing for battery packs.
What the Enclosure Has to Do
The requirement is not simply to be sealed, and this is the first thing that makes battery leak testing unusual.
The pack must keep liquid water and contaminants out, to a standard typically expressed as an ingress protection rating covering dust and temporary immersion, and frequently a higher rating covering high-pressure washing.
At the same time it must not be gas-tight. Air inside the pack expands and contracts with temperature and altitude, and a fully sealed enclosure would build a pressure differential across its seals every time the vehicle warmed up or climbed a hill. More seriously, if a cell fails and vents, the pack needs a controlled path for that gas to leave rather than pressurizing until the housing bursts.
Packs therefore carry a pressure equalization element, a membrane that passes air and blocks liquid, and frequently a separate emergency vent for a thermal event. The enclosure is deliberately not sealed, which has direct consequences for how it is tested.
Why Ingress Ratings Cannot Be Tested in Production
An ingress protection rating is defined by a functional test: submerge the object under specified conditions and confirm no harmful ingress. That is a valid laboratory verification and it is impossible on a production line, since immersing every pack would be slow, would wet a unit full of electronics, and would tell you nothing about a marginal seal that lets water in on the two hundredth cycle rather than the first.
Production therefore measures something else, a leak rate, and relies on a correlation established during validation.
The correlation is built by testing sample packs with deliberately introduced defects of known size, subjecting them to the immersion or spray condition the rating requires, and identifying the leak rate at which water actually enters. The production limit is then set below that figure with margin.
That process is worth understanding because it explains a common confusion. A leak rate limit is not derived from a formula; it is derived from experiment on that specific enclosure design. A limit borrowed from a different pack, or set by the test equipment supplier without correlation work, is a number with no relationship to whether water gets in.
The Methods
| Method | Sensitivity | Locates the leak? | Practical notes |
|---|---|---|---|
| Pressure decay | Moderate | No | Simplest and cheapest; highly sensitive to temperature and needs stabilization time |
| Differential pressure decay | Good | No | Compares the pack against a sealed reference volume, which cancels ambient temperature effects |
| Vacuum decay | Good | No | Evacuates rather than pressurizes, useful where positive pressure would deform a large flat enclosure |
| Mass flow | Good | No | Measures the flow needed to hold pressure, giving a direct leak rate rather than an inferred one |
| Tracer gas, accumulation | High | No | Pack filled with helium or forming gas inside a chamber; detector measures what escapes |
| Tracer gas, sniffer probe | High | Yes | Operator or robot traces the seam to find where the leak is; used for diagnosis rather than pass or fail |
| Ultrasonic | Low | Yes | Detects turbulent flow at gross leaks only; a fault-finding aid rather than a production test |
Pressure decay
The pack is pressurized slightly, isolated, allowed to stabilize, and the pressure monitored for a defined period. Any fall indicates gas leaving. It is inexpensive, needs no consumables and is the default choice for most production applications.
Its weakness is that pressure inside a fixed volume responds to temperature as readily as it responds to a leak, which is covered below.
Differential pressure decay
The same principle measured against a sealed reference volume of similar size. Both are pressurized, then a differential sensor measures only the difference between them. Ambient temperature changes act on both sides and largely cancel, so the measurement isolates the leak far better than absolute decay does. For most battery pack applications this is the sensible default.
Mass flow
Rather than watching pressure fall, the instrument measures the gas flow required to hold pressure constant. The output is a leak rate directly, in standard volume per unit time, which is easier to relate to a specification than a pressure change. Cycle times can be shorter than decay methods on large volumes.
Tracer gas
Where sensitivity beyond pressure methods is needed, the pack is filled with a tracer, either helium or a forming gas mixture of hydrogen in nitrogen, and a detector measures what escapes. Forming gas is considerably cheaper than helium, easier to source and non-flammable at typical tracer concentrations, which has made it the more common choice in automotive production.
Tracer methods split by how the escaping gas is captured. Accumulation in a chamber gives a pass or fail figure at high sensitivity. A sniffer probe traced along the seam finds where the leak is, which is a diagnostic tool rather than a production measurement.
Temperature: The Dominant Error
This is the single most important practical point in battery leak testing.
Gas in a sealed volume obeys the gas laws, so pressure varies with temperature whether or not anything is leaking. At a test pressure of around 100 millibar gauge, the absolute pressure inside the pack is roughly 1,113 millibar, and a one degree Celsius rise raises that by close to 3.8 millibar. Expressed against the gauge pressure being measured, that is a 3.8 percent change from a single degree.
Typical decay limits are a fraction of a millibar. A temperature change of a tenth of a degree can therefore swamp the leak signal entirely.
That has direct consequences on the line. A pack arriving warm from electrical testing is still cooling, so its internal pressure is falling for reasons that have nothing to do with a leak, and it fails. A pack that has been sitting in a cool area and is warming will show pressure rising, and can pass while genuinely leaking. Neither result is about the seal.
Three responses are used, usually in combination. Thermal stabilization time before the measurement, which costs cycle time. Differential measurement against a reference volume, which cancels ambient effects. And controlling the environment around the test station so packs arrive at a consistent temperature.
Volume, Deformation and Cycle Time
Two further physical realities shape what is achievable.
A given leak produces a smaller pressure change in a larger volume, so test time rises with enclosure size for the same sensitivity. Battery packs are large, which is why leak testing them takes longer than testing a small component and why the test frequently sets the cycle time in an end-of-line station.
Large flat enclosures also flex when pressurized. That flexing increases internal volume during the fill and continues for a period afterward as the structure creeps, and increasing volume looks exactly like a leak to a decay measurement. The response is a fill phase followed by a stabilization phase before measurement begins, and cutting the stabilization to save cycle time produces false failures that appear random.
“Every battery pack must prove its sealing performance before release, making leak testing, calibration, traceability, and controlled test conditions fundamental parts of end-of-line quality.”
See it in action
The Pressure Equalization Vent
The membrane that lets the pack breathe is, from a leak test perspective, a designed leak path. Two approaches exist and each has consequences.
Testing with the vent plugged measures the enclosure and seals alone, which gives a clean measurement of the thing the assembly process actually controls. It requires an extra fixture operation, and it does not verify the vent installation.
Testing with the vent fitted measures the pack as it will ship, which is closer to reality. The difficulty is that the membrane has its own permeability with a tolerance band, so that variation adds to the measurement and the limit has to accommodate it. On a tight specification the vent tolerance can consume most of the available budget.
There is no universally correct answer. What matters is that the choice is made deliberately during validation, and that the correlation work described earlier is performed under the same configuration the production test uses.
Setting the Limit
Bringing those together, a defensible leak specification requires four things.
Fixture and Calibration
Two sources of error sit in the test system rather than in the pack.
The fixture that seals against the pack and connects it to the instrument is itself a potential leak, and a deteriorating fixture seal produces a rising failure rate that looks like a production problem. Regular verification with a known-good master part identifies it quickly.
Calibration uses a master leak, a calibrated orifice of known rate, introduced into the circuit so the system can confirm it detects a leak of the expected size. Running that check at defined intervals is what distinguishes a leak test that is measuring from one that is merely producing numbers, and it is the same discipline that applies to any measurement on the line.
Where It Sits in the Line
Leak testing belongs at end of line, after the enclosure is closed and before the pack is marked as good. Cybernetik builds this into a dedicated end-of-line zone within battery pack assembly automation, separated from the assembly stations that feed it.
| Cybernetik end of line capability | Specification |
|---|---|
| Leak testing | Air leakage testing performed on every pack |
| Test zone | Zone 3 of the pack assembly line, separated from module and pack build |
| Sequence position | After BMS programming and TCU mounting, before laser marking |
| Electrical verification | End of line electrical testing on the completed pack |
| Marking | Laser marking applied only to a pack that has passed |
| Upstream sealing stages | Telematics gasket assembly, bottom plate gasket and BMS mounting plate, top cover closing and cleaning |
| Traceability | Barcode and RFID capture with MES connectivity and process logging |
| Line rate | Up to 6,000 cells per hour through the upstream cell handling stages |
| Cell formats | Cylindrical 18650, 21700 and 32140, plus prismatic, pouch and blade |
The sequencing detail is worth noting. Air leakage testing sits after BMS programming and telematics mounting and before laser marking, so a pack that fails the seal test never receives the identity that would allow it to ship. The gasket and cover operations that determine the result, telematics gasket assembly, bottom plate gasket and top cover closing, happen upstream in the pack assembly zone, which is where the seal is actually made rather than where it is verified.
Cybernetik has operated for more than three decades, is headquartered in Pune with additional facilities in Gujarat and Raigad and international offices in the United States and UAE, and has installed over 6,000 systems across 30 plus countries, including more than 400 custom robotic automation solutions. In battery manufacturing that work has been delivered for manufacturers including Hero MotoCorp, TVS Motor, Livguard and Matter. Further background is on the Cybernetik about page.