Home / Valve Test Benches / Testing by valve type
Testing gate, ball, butterfly and check valves on one bench — what changes between them
Buyers often ask for one bench that covers everything in their catalogue, from DN15 gate valves to DN600 butterfly valves, and then discover that the bench was never the problem: the adaptors and the test sequence were. This page sets out, valve family by valve family, where the pressure is allowed to go, how many seats have to be proven, and what has to change on the fixture — so one bench really does cover the range you sell.
Quick answer
| Gate valve | 2 seats · shell half open · seat both ways |
|---|---|
| Ball valve | 2 seats · seat both ways · vent the cavity |
| Butterfly valve | 1 seat · shell disc open · seat closed |
| Globe valve | 1 seat · test in the closing direction |
| Check valve | 1 seat · flow direction only |
| What changes | The adaptor, not the pump |
Four things that change from one valve family to the next
Every pressure test on every valve asks the same two questions: does the body hold, and does the closure hold. What differs is how many closures there are, where the pressure has to be allowed to go for the body test, and what the valve looks like where it meets the bench. Those three questions, plus the seat material, explain almost every difference between the procedures.
- How many seats. A gate valve and a ball valve have two sealing faces and both have to be proven, one direction at a time. A butterfly, globe or check valve closes on a single seat.
- Where the pressure goes in the shell test. The closure has to be off its seat so the pressure reaches the whole body cavity. Half open for a gate, open for a butterfly, held off its seat for a check valve.
- What the closure is made of. A resilient seat (PTFE, EPDM, NBR) is proven against a leakage rate that is usually no visible leakage at all. A metal seat is proven against a measured allowance — the classes are on the leakage rate classes page.
- How it meets the bench. Flanged, wafer, lugged or weld-end. This is the part that turns a single bench into six, and it is the part that is cheapest to solve when it is planned at the quotation stage.
Ball valves: two seats, one moving ball
A ball valve looks like the simplest valve on the bench and it is the one that most often produces a misleading result, because the ball is not fixed. On a floating ball design the ball is free to move a millimetre or two between the two seats, and the line pressure is what pushes it onto the downstream seat to seal. Test from one end only and you have proven the downstream seat; the upstream seat has been held shut by the seat ring itself, not by the ball. Reverse the pressure and the ball travels across, and the other seat is the one under load. Two directions is not a formality for a ball valve, it is the whole test.
On a trunnion-mounted ball the ball is held on bearings and does not move, and each seat has its own spring or piston assembly that seals independently. Both seats still have to be tested, but the result is far less sensitive to which end you pressurise first, and the valve will usually hold at much lower pressures than a floating design.
Three ball valve details that catch people out
- Vent the body cavity. A double block and bleed ball valve traps the cavity between two closed seats. Heat from the test water, or pressure seeping past a seat, can raise the cavity above the line pressure. Fill and vent the cavity before the seat test, and never leave a DBB valve pressurised with the cavity isolated.
- Watch the stem seal. The stem is the third sealing path on a ball valve and it is tested by the shell test, not the seat test. A drip at the stem gland on a new valve is a packing or O-ring problem and it will not go away by re-torquing if the stem is scored.
- Orientation matters at low pressure. Seal is made by pressure on a floating ball, so at a very low seat test pressure there is very little force holding the ball against the seat. A few drops there is not the same finding as a few drops at rated pressure; check what the product standard requires before writing a reject.
Most ball valves arrive with a lever or a gearbox, and the bench has to let the operator work the lever with the valve clamped — roughly a quarter turn of clear space on the operator side. A valve tested with the lever removed proves the seat, not the valve as shipped.
Butterfly valves: one seat, and it is soft
A butterfly valve closes with a disc rotating into a rubber or PTFE liner. That single seat carries the whole seat test, and it is also the part most easily damaged by a careless shell test. The sequence that protects it is simple: shell test with the disc open, so the pressure acts on the body and the liner backing rather than on the seating line, then seat test with the disc closed at the lower seat pressure, read from the open end.
The fixture is where butterfly valves differ most. A wafer butterfly valve has no flange of its own: it is designed to be clamped between two pipe flanges, and its liner is what seals against them. Clamp it between two plain steel plates and the liner is crushed between flat faces with nothing to seat into, which produces a leak that belongs to the fixture and not to the valve. The plates need a raised sealing face or a gasket ring of the right bore, and the clamping force has to be enough to seal the joint without deforming the liner.
- Wafer bodies — clamped between two plates with sealing rings; the bench's daylight has to be short enough to close on a body that is only a few centimetres thick.
- Lugged bodies — bolted on both sides; the bench needs through-bolts or a bolted adaptor, and the lugs are not designed to take clamping force, only bolt load.
- Flanged bodies — the same adaptors as a gate valve of the same DN and class.
One more rule: a soft seat has a pressure limit that is usually below the body rating. Do not run the shell test through a closed disc to save time — it is the fastest way to turn a good valve into a leaking one, and the damage will not show until the valve is on the line.
Check valves and globe valves: one seat, one direction
A check valve closes in one direction by design, so the seat test is run in the flow direction only, with the closure free to seat. The shell test still applies to the body, and the disc or flap has to be held off its seat during that test so the pressure reaches the whole cavity instead of being stopped at the closure. Reverse tightness is a separate characteristic: test it when the purchase order or the product standard asks, not by default.
A globe valve seats in one direction too, and the direction is set by the design — flow over the disc or under it. Test it in the closing direction, with the disc seated by the stem, and note on the report which direction was used, because an inspector reading the report needs to know that the seat was loaded the way it will be loaded in service. The stem packing on a globe valve is under shell pressure, so a gland leak during the shell test is a genuine finding.
Gate valves, briefly
A gate valve is the family with two metal or resilient seats and a long body, and it has its own page because the sequence and the failure modes are specific: the gate goes half open for the shell test, closed for each of the two seat tests, and a one-sided leak usually points at one seat face. If gate valves are the bulk of your production, start there: how to test a gate valve.
The adaptor matrix: what one bench needs to cover five families
| Valve family | End connection | What the bench needs |
|---|---|---|
| Gate valve | Flanged RF / RTJ | Raised-face or ring-joint plates, long daylight, end clamp |
| Ball valve | Flanged or butt-weld | Flange adaptors plus weld-end spigots; cavity vent; operator clearance |
| Butterfly valve | Wafer / lugged / flanged | Sealing-ring plates, short daylight, through-bolts for lugged bodies |
| Globe valve | Flanged | Standard flange plates, moderate daylight, gland access |
| Check valve | Flanged or wafer | Flange plates with a closure hold-off, flow-direction marking |
One bench, one pump and one control system; the adaptor set is what changes. Price the adaptor set with the bench, not after the first order arrives.
Sizing one bench for mixed production
- Size the daylight for the longest body, not the largest bore. A DN300 gate valve is longer than a DN400 butterfly valve and it is the gate valve that sets the clear span. The reasoning behind the numbers is on selection by DN.
- Size the pressure for the highest class, not the average. One PN100 valve in the catalogue sets the pump, the seals and the gauge range for everything else.
- Add an air circuit if you sell soft-seated valves. The low-pressure air seat test is a separate requirement from the hydrostatic test, and it needs its own regulator and gauge — the water circuit cannot be reused for it.
- Think about changeover time. Five families on one bench means five adaptor changes a day. Quick-change plates and a second clamping station usually pay for themselves before the second month, which is why dual-station benches exist.
SYJ-600SQ · large bore, water + air
- DN50 – DN600
- Water 40 MPa / Air 4 MPa
- 90-ton clamping
If your catalogue mixes families, the useful thing to send us is not one size but the shape of the range: which valve types, which DN bands, which pressure classes, and how many of each per week. That is enough to name the bench, the plate set and the adaptors. The buyer guide covers the rest of the selection logic, and the eight-step hydrostatic procedure is the general sequence that all of the above sit inside.
Questions we get asked about testing different valve types
Can one bench handle gate, ball, butterfly and check valves?
Yes, within the same size and pressure envelope — but the clamping plates alone are not enough. Each family needs its adaptor set: sealing-ring plates for wafer butterfly valves, RF or RTJ plates for flanged gate and globe valves, through-bolts for lugged bodies, weld-end spigots for butt-weld ball valves. The bench is the frame and the pump; the adaptor is what makes a family testable.
Why test a ball valve in both directions?
Two seats sit either side of one moving ball. On a floating design the pressure pushes the ball onto the downstream seat, so a one-direction test only proves that seat; reverse the flow and the ball crosses over. A trunnion ball holds position but still has two independent seat assemblies, and each one has to be proven.
Open or closed disc for a butterfly shell test?
Open, or at least clear of the seat. The shell test is about the body and the liner backing; the seat test is run closed at the lower seat pressure. Holding shell pressure against a closed soft seat damages the seating line on valves that would otherwise pass.
Why does a soft-seated valve weep at low pressure but hold at rated pressure?
Because the seal is made by pressure. At very low differential pressure there is little force holding the closure on the seat, so a few drops can pass. The product standard or the purchase order names a minimum seat test pressure for exactly this reason — do not reject a valve tested below it.
Do check valves get a two-direction seat test?
Normally not. The seat test runs in the flow direction, because that is the direction the valve is built to close. Reverse tightness is tested only when the specification asks for it, and the body shell test is run with the closure held off its seat.
One bench can cover a mixed catalogue; what it needs is the right plates. You get a model recommendation, the datasheet, the adaptor list and a price within 12 working hours.