Vacuum System Leak Rate: How to Set and Test an OEM Acceptance Limit
HCEM Engineering TeamShare
A vacuum system leak rate is the amount of gas that gets into an evacuated system per second, written as pressure × volume ÷ time, usually in mbar·L/s. You measure it by pumping the system down, closing a valve and timing the pressure rise: Q = V × Δp ÷ Δt. For an OEM machine, the acceptance limit comes from the pump: the leak must be small compared with what the pump can remove at the working pressure, and small enough that the vacuum holds for as long as the machine needs.
This guide covers the leak-rate acceptance test for a complete machine: the units, the calculation, how to tell a leak from outgassing, how to set the limit and the test conditions, and a checklist you can put in a test procedure. Testing the pump on its own (blank-off vacuum, pump-down time, flow) is covered in our guide to how to test a vacuum pump. The worked examples use datasheet values from our own oil-free piston vacuum pumps.
What is a vacuum system leak rate?
It is the gas load that enters the vacuum side of a machine from outside: through fittings, seals, valve seats, hose walls and joints. Because the gas arrives at low pressure, the leak is expressed as a pV flow: pressure multiplied by volume per unit time. A leak of 1 mbar·L/s raises the pressure of a 1 L volume by 1 mbar every second, or of a 10 L volume by 0.1 mbar every second.
Two things follow from that definition:
- The leak rate belongs to the system, not to the pump. The same pump can sit on a tight machine or a leaky one. A pump that reaches its datasheet vacuum when blanked off says nothing about the hoses and seals downstream.
- A pressure-rise test measures everything that adds gas, not only true leaks: moisture released from surfaces, gas trapped in blind holes, permeation through elastomers and temperature drift. A good test procedure is designed to separate these.
Which units are used for vacuum leak rate?
Europe and most vacuum datasheets use mbar·L/s. The SI unit is Pa·m³/s. American documents often use Torr·L/s or atm·cc/s, and flow meters read sccm (standard cubic centimetres per minute). The conversions are exact for the pressure units and depend on the reference conditions for sccm:
| Unit | mbar·L/s | Pa·m³/s | Torr·L/s | sccm |
|---|---|---|---|---|
| 1 mbar·L/s | 1 | 0.1 | 0.750 | 59.2 |
| 1 Pa·m³/s | 10 | 1 | 7.50 | 592 |
| 1 Torr·L/s | 1.333 | 0.1333 | 1 | 78.9 |
| 1 atm·cc/s | 1.013 | 0.1013 | 0.760 | 60.0 |
| 1 sccm | 0.0169 | 0.00169 | 0.0127 | 1 |
The sccm figures use the common standard of 0 °C and 1013.25 mbar, with the gas at the same temperature. If your flow meter uses a 20 °C standard, or the gas is warmer than the reference, the figure moves by a few per cent: 1 mbar·L/s of gas at 20 °C is about 55.2 sccm on a 0 °C standard. State the reference in the test procedure.
A useful sense of scale: 1 mbar·L/s is about 0.059 L/min of free air, and 1 L/min of free air is about 17 mbar·L/s. So a pump rated at 120 L/min free-air flow moves about 2,000 mbar·L/s at the inlet when it is open to atmosphere. For flow unit conversions, use our CFM to L/min converter or see what CFM means.
How do you calculate leak rate from a pressure rise?
Pump the system down, close the isolation valve, wait for the reading to settle and then log the pressure. The leak rate is:
Q = V × Δp ÷ Δt
where V is the isolated volume, Δp the pressure rise and Δt the time it took. With litres, mbar and seconds, Q comes out in mbar·L/s. Use absolute pressure, or gauge readings taken on the same gauge in the same session (a pressure difference is the same on either scale, as long as the barometer does not change during the test).
Worked example (example numbers, not a measured result). A machine has a 10 L reservoir and 2 L of manifold and hose, so V = 12 L. It is pumped down to about 150 mbar absolute and the valve is closed. After a 5-minute settling period the gauge reads 154.0 mbar; 10 minutes (600 s) later it reads 159.0 mbar.
- Q = 12 L × 5.0 mbar ÷ 600 s = 0.10 mbar·L/s
- = 0.010 Pa·m³/s ≈ 5.9 sccm
- Rate of rise: 5.0 mbar ÷ 10 min = 0.5 mbar/min
The result is only as good as the volume. Work it out from the CAD model (reservoir, manifold bores, valve bodies) and add the hoses: an 8 mm bore hose holds about 0.05 L per metre. If the internal volume is uncertain, measure it by gas expansion: connect a vessel of known volume Vref at atmospheric pressure patm to the evacuated system at ps, wait for the temperature to settle, and read the final pressure pf. Then Vsys = Vref × (patm − pf) ÷ (pf − ps).
Is it a leak or outgassing?
A pressure-rise curve has a shape, and the shape tells you what is adding the gas. Log a reading at least every minute and compare the rate in the first and second half of the window.
| Source | What the curve does | How to confirm |
|---|---|---|
| Real leak from atmosphere | Straight line: the same rise every minute | Rate is the same after longer pumping; it changes when a fitting is resealed |
| Outgassing (moisture on surfaces, residues) | Fast at first, then slows and levels off | Rate falls after longer pumping or after drying the parts; worse after washing or on humid days |
| Virtual leak (gas trapped in blind threaded holes, between double seals) | Quick rise that dies away over minutes | Disappears with longer pumping; vented screws and single seals cure it |
| Permeation through elastomer hose and seals | Slow, steady, rises with temperature | Changes with hose material and length; hard to tell from a small leak |
| Temperature drift | Follows the temperature, can even fall | Log system temperature next to pressure |
A real leak from atmosphere is close to constant while the system stays below about half an atmosphere, because the inflow through a small opening no longer depends on the pressure inside. That is why a leak gives a straight line and why the test pressure should sit near the machine's working pressure.
Temperature matters more than people expect. In a closed volume, pressure is proportional to absolute temperature, so at 155 mbar a change of 1 K moves the reading by about 0.53 mbar. In the example above that is more than a tenth of the 5 mbar rise. Hold the system temperature steady within about 0.5 K during the window, or record it and correct each reading by p × Tref ÷ T.
What leak rate is acceptable for an OEM machine?
There is no universal number. The limit comes from two questions: how much spare capacity the pump has at the working pressure, and how long the vacuum must hold when the pump is not removing gas. Work out both and use the tighter one.
1. The pump's spare capacity at working pressure
A pump's throughput at a given pressure is its pumping speed at that pressure multiplied by the pressure. For an oil-free piston pump, the speed falls towards zero as the pressure approaches the ultimate vacuum. A simple estimate is throughput ≈ S0 × (pw − pu), where S0 is the free-air flow, pw the working pressure and pu the ultimate pressure, both absolute.
Take a standard HC580A: 120 L/min free-air flow (2.0 L/s) and −92 kPa ultimate vacuum, about 93 mbar absolute. At a working pressure of −80 kPa (213 mbar absolute):
- Throughput ≈ 2.0 L/s × (213 − 93) mbar = 240 mbar·L/s, about 14 L/min of free air.
- That has to cover everything: the process load (suction cups that do not seal fully, porous parts, purge flows) and the leaks.
- If you allow leaks 10 % of it, the leak budget is 24 mbar·L/s. With that leak the pump's best vacuum on the system rises from about 93 to 105 mbar absolute (93 + 24 ÷ 2.0), and pump-down to −80 kPa takes about 5 % longer. The 10 % is a design choice, not a standard.
For a machine that pumps continuously, this limit is generous. Its value is to show quickly whether a leak is a real problem for the pump, and how much room is left for seals to age over the machine's life. For the speed-versus-pressure behaviour behind it, see ultimate pressure vs working vacuum.
2. How long the vacuum must hold
Many machines hold vacuum with the pump stopped or a valve closed: a reservoir that the pump tops up, a clamped fixture, a sealed pack. Here the leak sets the time between restarts, and the limit is usually far tighter.
Take a 24 V DC machine with an HC480D (100 L/min free-air flow, −92 kPa ultimate vacuum), a 10 L reservoir and 1 L of lines, so V = 11 L. The pump stops at −85 kPa (163 mbar absolute) and restarts at −70 kPa (313 mbar absolute), a band of 150 mbar. If the pump should restart no more than once every 15 minutes (900 s) because of leakage alone:
- Qmax = V × Δp ÷ t = 11 × 150 ÷ 900 = 1.8 mbar·L/s (about 110 sccm).
- As a gauge criterion: 1.8 ÷ 11 = 0.17 mbar/s, about 10 mbar per minute of rise at the stop pressure.
- Each restart takes about (11 ÷ 1.67) × ln[(313 − 93) ÷ (163 − 93)] ≈ 7.6 s of running to recover.
That limit is about 60 times tighter than the pump's throughput at the same pressure, 1.67 L/s × 70 mbar ≈ 117 mbar·L/s. Restart frequency matters for the pump's life and the motor; see our guides to duty cycle and start-stop frequency and reservoir sizing.
Set the production acceptance limit below whichever budget is tighter, so that a new machine leaves room for seal wear and for the leaks that appear after shipping and installation.
How do you run a vacuum leak rate acceptance test?
The test conditions decide whether two results can be compared. Write them down and keep them the same for every machine.
- Define the boundary. List what is inside the test: from which valve to which ports, with which cups, nozzles or fixtures capped. Decide whether the pump is inside or outside.
- Check the pump first. A blank-off test at the pump inlet confirms the pump reaches its datasheet vacuum, so a poor result can be put on the system. The procedure is in how to test a vacuum pump.
- Condition the system. Parts clean and dry, assembled to the drawing, then pumped down for a fixed time before isolation. Freshly washed parts and new elastomer hose outgas the most.
- Isolate. Close a valve close to the system side and put the gauge on the system side of it. Decide what happens on the pump side: leaving the pump running keeps vacuum on both sides of the valve seat; stopping it usually lets the pump side return to atmosphere and tests the valve seat against the full pressure difference. Choose one and write it down.
- Wait for stabilisation. A fixed settling time (for example 5 minutes for a small metal system, longer for systems with a lot of hose or plastic) lets the gas temperature settle and the fastest outgassing die away.
- Log the rise. Read at least every minute through a fixed window. Make the window long enough that the rise is many times the gauge resolution: with 0.1 mbar resolution, aim for a rise of at least 2 mbar.
- Check linearity. Compare the rate in the second half of the window with the first. If it is clearly lower, outgassing is still part of the result: extend the conditioning and test again.
- Calculate and record. Q = V × Δp ÷ Δt in the agreed unit, against the agreed limit, with the serial number, date, readings, temperature and barometer.
Use a digital gauge, ideally an absolute one, with a resolution of 0.1–1 mbar. A mechanical −100 to 0 kPa gauge is fine for the pump's blank-off check but too coarse for a rise of a few mbar. When the barometer changes during a long test, a gauge-pressure reading drifts with it; an absolute gauge avoids that (see altitude and ambient conditions).
When a machine fails, find the leak by isolating sections one at a time and repeating the rise test, or by using an ultrasonic leak detector on larger machines. Typical culprits are threaded fittings, push-in connectors, valve seats, O-rings, cup lips and hose ends. Soapy water works poorly under vacuum, because the bubbles are drawn inwards; if the parts are rated for it, a low overpressure test with leak-detection fluid finds the larger leaks faster. A leak that only shows with the valve open and the pump stopped is air returning through the pump: see backflow prevention.
When do you need helium leak testing?
Only when the specification is tighter than a pressure-rise test can resolve. In a rough-vacuum machine of around 10 L, outgassing, temperature and gauge resolution put the practical floor of a pressure-rise test at roughly 0.01 mbar·L/s. Most machines built around oil-free piston pumps, working between −60 and −98 kPa, have leak budgets well above that, so a pressure-rise test is the right method.
Helium leak testing is a tracer gas method: a mass-spectrometer leak detector, sensitive only to helium, is connected to the evacuated system while helium is sprayed on suspect joints, or the part is filled with helium and a sniffer probe is moved around the outside. It detects leaks many orders of magnitude smaller than a pressure-rise test and shows where they are. It also needs its own equipment, a pumping system to suit the detector and a written procedure. Consider it when the end product demands it (sealed medical, refrigeration or high-vacuum components, for example) or when you need to locate a small leak quickly in production. Acceptance criteria for a helium test come from that product's own specification, not from the pump.
Vacuum leak test acceptance checklist
Use this as the skeleton of the test procedure. The example entries are illustrations; replace them with the values for your machine.
| Item | What to specify | Example entry |
|---|---|---|
| Boundary | What is inside the test: valves, manifold, reservoir, hoses, capped ports; pump in or out | From valve V1 to all cup ports, capped; pump outside |
| Volume | Isolated volume and how it was found | 12.0 L, CAD plus hose, checked by gas expansion |
| Test pressure | Absolute pressure at isolation, near working pressure | 150 ± 10 mbar abs |
| Conditioning | State of parts, pumping time before isolation | Parts dry; 10 min pumping after assembly |
| Isolation | Valve, gauge position, pump running or stopped | Ball valve V1; gauge on system side; pump running |
| Stabilisation | Wait between isolation and first reading | 5 min |
| Window and sampling | Logging time and interval | 10 min, one reading per minute |
| Instrument | Gauge type, range, resolution, calibration | Absolute digital, 0–1100 mbar, 0.1 mbar, in calibration |
| Temperature | Ambient range and allowed drift in the window | 20–25 °C; drift ≤ 0.5 K |
| Calculation and unit | Formula and unit of the result | Q = V × Δp ÷ Δt, mbar·L/s |
| Linearity | How a still-falling rate is treated | Second-half rate within 20 % of first half, or retest |
| Limit | Pass value and where it came from | ≤ 0.5 mbar·L/s, from the hold-time budget |
| Failure and retest | What may be changed before a retest; who approves | One reseal and retest; parts changed recorded |
| Record | What goes into the test report | Serial no., date, readings, temperature, barometer, result |
The leak test fits into the wider acceptance of the vacuum system; our startup and commissioning checklist covers the rest, and the vacuum pump specification checklist covers what to send a supplier. If the leak budget is tight, check pump-down time as well: see evacuation time sizing.
Common questions
What is a good vacuum leak rate?
One that is small compared with the pump's throughput at the working pressure and lets the vacuum hold as long as the machine needs. Only your own budget calculation gives the number: in the two examples above the limits came out at 24 mbar·L/s for continuous pumping and 1.8 mbar·L/s for holding vacuum between restarts.
How do I convert mbar·L/s to sccm?
Multiply by 59.2 (0 °C, 1013.25 mbar standard). To go from Pa·m³/s to mbar·L/s, multiply by 10; from Torr·L/s, multiply by 1.333.
Is a pressure-rise test the same as a pressure decay test?
They use the same formula. A pressure-rise test works under vacuum and measures air coming in; a pressure decay test pressurises the part and measures air going out. For a vacuum machine, test under vacuum near the working pressure, because seals and lips can behave differently when the pressure pushes the other way.
Does the pump's ultimate vacuum prove the system is leak-tight?
No. Ultimate vacuum is measured at the pump inlet with nothing else connected. It shows the pump is healthy. The system's leak rate needs its own test with the valves, hoses and seals in place.
What are helium leak test acceptance criteria?
A maximum helium leak rate, in mbar·L/s or Pa·m³/s, set by the product's own specification, together with the test method (vacuum or sniffer mode), helium concentration and detector calibration. There is no general value; it depends on what the part must hold.
Browse our oil-free vacuum pumps, or send us your flow, working vacuum, system volume, supply voltage and duty cycle, and our engineers will suggest a pump with the spare capacity your leak budget needs, including OEM versions for your machine. For background on the pumps themselves, see how vacuum pumps work.