Views: 0 Author: J-VALVES Publish Time: 2026-11-03 Origin: Site
An API 598 test report covers three tests, and buyers often read them as one. They answer different questions. The shell test asks whether the pressure boundary holds. The backseat test asks whether the stem seal can be isolated. The closure test asks how much passes a closed seat. A valve can pass one and fail the next, and the duration on the certificate belongs to a specific test, not to the valve as a whole.
For the shell test the valve is part open, both ends are blanked, and water is brought up to 1.5 times the rated pressure at 38 degrees Celsius. With the wedge or disc off its seat, no differential exists across the seat, so this test says nothing about leakage. What it loads is every wall, every body to bonnet joint and the gland. Visible leakage through a wall or a fixed joint rejects the valve. Leakage past an adjustable stem seal does not reject it automatically, but the seal still has to be shown to hold rated pressure afterwards, while a fixed seal such as an O-ring must show nothing at all.
Duration is measured from the moment the specified pressure is reached and stable, not from the moment the pump starts. Filling, venting trapped air and bringing the body up to pressure are outside the count. This matters at small sizes: a two inch valve is held for 15 seconds, and a rig that takes four minutes to fill while the gauge creeps will pass a leak that only opens once the body is warm and fully stressed.
| Size | Shell test | Backseat test | Closure, other valves | Closure, check valves |
|---|---|---|---|---|
| DN 50 and smaller | 15 s | 15 s | 15 s | 60 s |
| DN 65 to 150 | 60 s | 60 s | 60 s | 60 s |
| DN 200 to 300 | 120 s | 60 s | 120 s | 120 s |
| DN 350 to 600 | 300 s | 60 s | 120 s | 120 s |
| Above DN 600 | 600 s | 120 s | 240 s | 240 s |
The high pressure closure test uses 1.1 times the rated pressure with the valve closed, one side loaded and the other open to atmosphere. A bidirectional valve tested from both sides owes the full duration twice, not half each. A double block and bleed tested with both seats loaded at once owes twice the listed time. The low pressure closure test is a separate run at 5.5 plus or minus 1.5 bar gauge with air, nitrogen or an inert gas, and it finds the seat defects that water at high pressure pushes shut.
Backseating is the shoulder where the stem meets the bonnet when the valve is fully open. The backseat test runs at 1.1 times rated pressure with the valve fully open and the stem drawn back onto that shoulder. Where the procedure allows, the gland is relaxed first, so what is being measured is the metal to metal shoulder and not a packing that happens to be tight. Many operators never use the backseat, and many purchasers never ask for the test. It is still the only evidence that a valve can be repacked with the line under pressure.
Shop tests run cold, on clean water, with no pipe load and no thermal gradient. None of that matches a line at 300 degrees Celsius with a flange pulled out of alignment. A seat that seals cold may lift when the body grows faster than the trim, and a bonnet joint that held at 20 degrees may weep after the first thermal cycle because the bolt load relaxed. A shell test also says nothing about how the valve behaves after a hundred operations. It is a boundary check on a new casting, and it should be read as one.
Test water stays in the body after the test, in the bonnet cavity and behind the wedge. On a carbon steel valve that means rust before dispatch unless it is drained and dried. On a stainless body the chloride in tap water is the bigger risk, because residual droplets that evaporate concentrate chloride on a warm surface, which is the condition that starts pitting and stress corrosion cracking. Low chloride water and forced drying are cheap; a rejected stainless valve on site is not.
Gas at the same pressure stores far more energy than water. A pneumatic shell test is run only when the purchase order asks for it, at 1.1 times rated pressure, behind a barrier, with an approved procedure, and it still permits no visible leakage. Nobody should sign for one on the strength of a hydrostatic procedure with the medium changed.
A report is only as good as the fields it carries: test pressure, medium and temperature, the actual duration per test, the acceptance figure applied to the seat, gauge identification and calibration date, and the inspector. Ask for EN 10204 3.1 when the mill and the shop are the same legal entity, and 3.2 when an independent body has to sign. Our quality control sequence runs shell tests at 1.5 times and seat tests at 1.1 times rated pressure on every valve from 1/2 to 56 inch (DN15 to DN1400), Class 150 to 2500, and the seat acceptance rates themselves are set out in our note on seat leakage classes. What the casting behind the test has to guarantee is a material question, covered in our guide to WCB, WCC, LCB and LCC valve bodies.
How long does an API 598 shell test take? It depends only on size: 15 seconds to DN 50, 60 seconds from DN 65 to 150, 120 seconds from DN 200 to 300, 300 seconds from DN 350 to 600, and 600 seconds above DN 600. The count starts when the specified pressure is reached and stable, so filling and venting do not add to it.
Is a pneumatic test required after the hydrostatic test? Only where the purchase order asks for it. The standard sequence is a hydrostatic shell test at 1.5 times rated pressure followed by closure tests, one at 1.1 times with liquid and one at 5.5 plus or minus 1.5 bar with gas. A pneumatic shell test at 1.1 times is an optional extra that carries a much larger stored energy hazard.
Why does a valve pass the shop test and leak on site? Because the shop test is cold, clean and unloaded. Thermal growth moves the seat relative to the trim, pipe alignment puts bending into the body, and the process fluid carries particles that the test water did not. None of those conditions exists at 20 degrees on a test bench, so a pass certifies the new casting, not the installed valve.
