Home / Valve Test Benches / Valve leakage rate classes
Valve Leakage Rate Classes — API 598 and ISO 5208, explained for test reports
The test pressure says how hard the valve is pushed; the leakage rate class says how much may pass the seat while it is pushed. This page puts the full ISO 5208 rate table for liquid and gas on one chart, maps API 598 onto it, works the numbers for real valve sizes, and lists what a bench needs to measure each rate — so the class printed on your test report means the same thing to you, your inspector and your customer.
Quick answer
| Rate A | No visible leakage |
|---|---|
| Rate D liquid | 0.1 × DN mm³/s |
| Rate D gas | 30 × DN mm³/s |
| DN100 / Rate D / water | 10 mm³/s = 0.6 ml/min |
| 1 bubble/s (gas) | ≈ 65 mm³/s |
The ISO 5208 rate table, liquid and gas
ISO 5208:2015 (Table 4) defines ten closure leakage rates. Every limit scales with the valve's nominal size DN, and the same rate letter has a different number for liquid and gas — a specification that says “Rate D” without naming the test medium is incomplete.
| Rate | Liquid test mm³/s | Liquid test drops/s | Gas test mm³/s | Gas test bubbles/s |
|---|---|---|---|---|
| A | No visually detectable leakage for the duration of the test | |||
| AA | 0.006 × DN | 0.0001 × DN | 0.18 × DN | 0.003 × DN |
| B | 0.01 × DN | 0.00016 × DN | 0.3 × DN | 0.0046 × DN |
| C | 0.03 × DN | 0.0005 × DN | 3 × DN | 0.0458 × DN |
| CC | 0.08 × DN | 0.0013 × DN | 22.3 × DN | 0.3407 × DN |
| D | 0.1 × DN | 0.0016 × DN | 30 × DN | 0.4584 × DN |
| E | 0.3 × DN | 0.0048 × DN | 300 × DN | 4.5837 × DN |
| EE | 0.39 × DN | 0.0062 × DN | 470 × DN | 7.1293 × DN |
| F | 1 × DN | 0.016 × DN | 3,000 × DN | 45.837 × DN |
| G | 2 × DN | 0.032 × DN | 6,000 × DN | 91.673 × DN |
Values from ISO 5208:2015, Table 4. Leakage rates apply when the test fluid discharges to atmosphere. Confirm against the edition named in the enquiry or purchase order.
How API 598 maps onto ISO 5208
API 598 does not use rate letters. It gives visual pass or fail criteria per valve type and size — drops per minute for metal-seated valves, zero visible leakage for resilient-seated valves — and couples them with the familiar pressures: shell test at 1.5 × the 38 °C rating, seat test at 1.1 ×.
ISO 5208's own table note states the correspondence, which is worth keeping on the wall of the test house:
| API 598 acceptance | Corresponds to ISO 5208 |
|---|---|
| DN 50 and below | Rate A |
| Soft-seated valves (other than check valves) | Rate AA on gas, Rate CC on liquid |
| Metal-seated check valves | Rate EE on gas, Rate G on liquid |
So a customer who writes “API 598, zero leakage seat” on a DN80 ball valve is, in ISO terms, asking for roughly Rate AA gas / Rate CC liquid — not Rate A. Quoting the wrong letter either overbuys or underdelivers, and the difference only appears at final inspection.
Worked examples at real sizes
| Valve | Class & medium | Maximum allowable leakage | How it is judged |
|---|---|---|---|
| DN50 gate valve | Rate A, water | No visible leakage | Visual inspection over the full hold time, good lighting, dry surfaces |
| DN100 globe valve | Rate D, water | 0.1 × 100 = 10 mm³/s (0.6 ml/min) | Graduated vessel over the timed hold |
| DN150 soft-seated butterfly valve | Rate CC, water | 0.08 × 150 = 12 mm³/s | Graduated vessel; gas equivalent judged by bubble count |
| DN200 trunnion ball valve | Rate D, gas | 30 × 200 = 6,000 mm³/s (≈ 0.36 m³/h) | Too fast for bubble counting — flow meter or pressure-decay method |
| DN150 valve, air bubble test | Rate B, gas | 0.3 × 150 = 45 mm³/s (≈ 0.7 bubbles/s) | Count bubbles per second in an immersion tank |
Two things these numbers make obvious. First, the tightest rates are visual judgements, and a visual judgement is only as good as the lighting, the water clarity and the hold time. Second, the loose gas rates produce hundreds of bubbles per second at large DN — counting them by eye is impossible, which is why the measuring method belongs in the bench specification, not in the operator's hands.
What the bench needs to measure a leakage rate
- A separate gas circuit. Bubble counting and flow-meter tests run on air or nitrogen at low pressure, typically a few bar — not on the 40 MPa water circuit. A dual-medium bench such as the SYJ-600SQ carries water to 40 MPa and air to 4 MPa on one frame, so shell test and gas closure test happen without re-fixture.
- A transparent immersion tank. Rates A to C on gas are judged by counting bubbles per second at the stem, the seat or the weld-adjacent joints. The tank, the lighting and the wait for the first bubble to form are all part of the method.
- Graduated measuring gear. Liquid rates between AA and G are verified by collecting leakage over a timed interval. A bench with a graduated vessel and a PLC-timed hold turns the number into a record; a wet floor and a stopwatch does not.
- Automatic hold and logging. The leakage rate is defined over the specified hold time. A PLC bench applies the time, logs pressure, temperature, hold duration and result against the valve serial number, and prints the report your customer's QA department files.
- Throughput for the judgement time. Bubble tests and timed collection are slow per valve. A dual-station layout such as the DFS2-600 lets one valve sit through its hold while the next is already clamped.
Frequently asked questions
What is the difference between API 598 and ISO 5208 leakage classes?
API 598 gives pass or fail criteria per size and seat type, judged visually in drops or bubbles per minute. ISO 5208 defines ten closure leakage rates (A, AA, B, C, CC, D, E, EE, F, G) with a numeric maximum scaled by DN. ISO 5208's table note gives the correspondence: Rate A for DN 50 and below, Rate AA gas / Rate CC liquid for soft-seated valves, Rate EE gas / Rate G liquid for metal-seated check valves.
Is ISO 5208 Rate A the same as zero leakage?
No. Rate A means no visually detectable leakage for the duration of the test. It is a visual acceptance criterion, not a molecular zero, and it still depends on the test pressure, hold time and medium named in the specification.
How is a leakage rate measured on a valve test bench?
Liquid: collect the leakage in a graduated vessel over the timed hold and convert to mm³/s. Gas at tight rates: count bubbles per second in an immersion tank — from the ISO 5208 values, one bubble per second corresponds to roughly 65 mm³/s. Gas at loose rates (Rate E and above): use a flow meter or a pressure-decay method instead of counting.
Which Changxun bench performs gas closure tests?
SYJ-600SQ runs water to 40 MPa and air to 4 MPa on one frame, so the high-pressure shell test and the low-pressure gas closure test happen on the same fixture. DFS2-600 adds a second station for judgement-time throughput.
Which leakage rate should be specified for a valve purchase?
The rate named by the valve product standard or the purchase order. Metal-seated valves are designed with a controlled seat leak, so specifying Rate A for them conflicts with the product standard; soft-seated valves are normally specified Rate A on gas. When in doubt, write the letter, the medium and the size on one line — that is what the test house needs.
Does the bench record the leakage value or only pass and fail?
Both are possible. With a PLC bench plus a graduated vessel or flow meter, the measured value is logged against the valve serial number and printed with the report. A visual-only configuration records pass or fail against the class criterion. State which your QA system requires when you ask for a quotation.
Benches for pressure and leakage testing
DFS2-600
- DN50 – DN600, dual station
- Judgement on one, loading on the other
- Water and air circuits
For the test pressures behind these classes — shell at 1.5 × PN, seat at 1.1 × PN, hold times by size — read the valve hydrostatic test pressure chart. For the wider selection logic, read the valve pressure test bench buyer guide.
Two questions that come up at this point — which test standard the bench follows on paper, and whether a certificate is issued per valve — are answered in the valve test bench FAQ.
Send the leakage class your reports must state, the DN range and the medium. You get a model recommendation, the measuring method that rate demands, a datasheet and a price within 12 working hours.