The datasheet comes back with an orifice two letters larger than anything else on the exchanger, the inlet line is now bigger than the nozzle it bolts to, and the client rejects the package. Work backwards from the required relieving load and the fault is almost always the rupture basis, not the valve.
Three ruptured tubes gives you three times the open flow area, three times the mass flow, and roughly three times the required orifice area. That is where the oversize came from. Nothing in API forbids it — it is a conservatism you chose, not a code requirement — but you now own every downstream consequence of that choice, and the client is objecting to the cost of them.
Check 1: Does the Exchanger Need a Tube-Rupture PSV at All?
Start here, before you argue about tube count. If the low-pressure side is strong enough relative to the high-pressure side, the tube-rupture case drops out entirely and the whole argument disappears.
- Compare the low-pressure side design pressure against the high-pressure side design pressure. API 521 carries the long-standing 10/13 screening rule: where the low side is designed to at least 10/13 of the high side, the hydrotest margin covers the transient and dedicated tube-rupture relief is generally not required. Confirm the wording in the edition your project is contracted to — editions differ on how the rule is qualified.
- Check whether the low-pressure side can actually be blocked in. If it discharges to a large-volume system that cannot be isolated, the pressure rise may be bounded by something other than a PSV.
- Check whether the low side runs liquid-full. If it does, see Check 4 — a PSV may not be the correct mitigation regardless of how you size it.
If the case survives all three, you need the valve. Move on.
Check 2: Fix the Rupture Basis — One Tube, Two Open Ends
The conventional basis across process industry practice is a single tube failing by a complete circumferential break. That is not one hole. A clean break leaves two open tube ends, both discharging into the low-pressure side, so the flow area is:
A_flow = 2 x (pi/4) x d_i^2where d_i is the tube internal diameter. Engineers who assume a single open end under-size by half; engineers who assume three tubes over-size by three. Confirm which mistake you made before you touch anything else — the required load may be closer than you think.
For an incompressible, non-flashing release:
W = Cd x A_flow x sqrt(2 x rho x dP)
W [kg/s], A_flow [m2], rho [kg/m3], dP [Pa]Cd for a ragged tube break is typically taken in the 0.6-0.7 range; use the value fixed by your company or licensor standard rather than picking one. dP is the high-side pressure at the moment of rupture minus the low-side relieving pressure — settle whether your project uses high-side normal operating pressure or high-side relieving pressure, because that choice alone can swing the load 10-20%.
Three tubes is defensible only if you can point to a mechanism that fails three tubes simultaneously — a documented vibration failure history, a tube-sheet or baffle failure, an erosion pattern in the inspection record. "Being conservative" is not that mechanism. If the client is paying for it, the client gets to reject it.
Check 3: Confirm the Flow Model Before You Blame the Tube Count
An oversized result is not always a tube-count error. Read the relieving-condition physics before you resize:
- Is the high-side fluid flashing? A subcooled liquid that flashes across the break is two-phase at the low-pressure side. Sizing it as all-liquid understates the volumetric load; sizing it as all-vapour grossly overstates it. Use an appropriate two-phase method and state which one on the datasheet.
- Is the low side vapour-filled with a free vent path? Then the required relief is the vapour displaced plus the flash vapour, not the whole liquid mass flow.
- Did you take credit for the exchanger volume? The low-side hold-up absorbs part of the transient. Neglecting it is another silent conservatism stacked on top of the tube count.
Check 4: What the Oversize Is Actually Costing You
"It passes more than the minimum required flow, so what is the problem?" The problem is that relief valve performance is bounded on both ends.
| What you see | Mechanism | First reading to take |
|---|---|---|
| Inlet line pressure drop exceeds 3% | API guidance limits non-recoverable inlet loss to under 3% of set pressure evaluated at the valve's rated capacity, not at your required load. Grow the orifice, and the rated capacity grows with it, so the same inlet piping now fails the check. | Recompute inlet loss at rated capacity for every overpressure scenario on that valve |
| Inlet piping larger than the vessel nozzle | Direct consequence of the 3% check at the inflated rated capacity | Nozzle size vs. required inlet line size |
| Chatter, seat leakage, drifting set pressure after commissioning | A spring valve passing well under roughly a quarter of its rated capacity cannot hold stable lift. It cycles, hammers the seat, and loses tightness | Ratio of the governing (non-rupture) relief load to valve rated capacity |
| Flare or header re-rate demanded by the client | The inflated tube-rupture load propagates into the header hydraulics and flare tip design load | Contribution of this valve to the total simultaneous flare load |
| Built-up backpressure exceeding limits on adjacent valves | Larger discharge from this valve raises header backpressure on everything discharging with it | Header backpressure at the discharge of neighbouring conventional valves |
The chatter risk is the one people miss. Tube rupture usually governs orifice selection while a much smaller case — blocked outlet, thermal expansion, fire — is what the valve actually sees in service. Size for a three-tube fantasy and the valve is permanently oversized for every real event.
One more trap: if the low-pressure side is liquid-full and short, the rupture generates a pressure surge faster than any spring valve can lift. No PSV size solves that. Mitigation is a rupture disc, an uprated low-side design pressure, or accepting the case per an evaluated risk basis — not a bigger orifice.
Resize and Re-Rate: The Procedure
- Recompute the required relief load on a single tube, complete break, two open ends, using the flow model matched to the flashing behaviour from Check 3.
- Rebuild the scenario table for the valve. List every overpressure case with its required load and identify which case governs the orifice.
- Select the smallest standard orifice whose certified capacity at the relieving conditions exceeds the governing load. Use the manufacturer's certified
Kd, not a generic coefficient. - Check the ratio of the smallest credible relief load to the valve's rated capacity. If it lands well below about a quarter, evaluate a modulating pilot-operated valve or split the duty rather than accepting a chatter-prone installation.
- Recompute inlet non-recoverable loss at the new rated capacity, for every listed scenario, and confirm it stays under 3% of set pressure.
- Recompute built-up backpressure in the discharge line and confirm it is within the limit for the valve type — conventional, balanced bellows, or pilot.
- Re-run the flare or header hydraulics with the revised load and reissue the header contribution.
- Reissue the datasheet with the rupture basis,
Cd,dPdefinition, and flow model stated explicitly so the next reviewer does not repeat the argument.
Verify Before You Close the Item
- Inlet loss under 3% of set pressure at rated capacity, documented per scenario — this is the check the client will audit first.
- Reaction force at the outlet recomputed for the new orifice, and the support design confirmed against it.
- Set pressure verified against the low-side MAWP at the coincident design temperature, with accumulation within the allowed limit for the governing case.
- Bench test at the shop: set pressure, seat tightness, and blowdown. A valve that reseats cleanly on the test stand tells you the trim and spring match the selected orifice.
- Rupture basis, tube count, tube ID, and coefficient recorded in the relief study so the assumption is auditable rather than buried in a spreadsheet.
Stop and escalate when the low-pressure side is liquid-full and the surge outruns valve response, when the 10/13 screening result is borderline, or when the client's philosophy document conflicts with the licensor's basis on tube count. Take the certified discharge coefficient, stability limit, and minimum stable flow from the valve manufacturer's application engineering rather than a catalogue figure, and route any deviation from the owner's relief philosophy through the process safety authority who owns the study.
FAQ
How do I decide how many tubes to assume ruptured for a PSV sizing case?
Use one tube failing by a complete circumferential break, giving two open ends and twice the tube cross-sectional flow area. More than one tube is not prohibited, but you need a documented failure mechanism — vibration history, tube-sheet failure, erosion — to justify the cost it imposes.
How do I know if my tube-rupture relief valve is oversized?
Compute the inlet non-recoverable pressure loss at the valve's rated capacity; if it exceeds 3% of set pressure with reasonable piping, the orifice is too large. Also compare the smallest credible relief load to the rated capacity — a ratio well under about a quarter puts the valve in the chatter regime.
Is it acceptable to oversize a relief valve if it still passes more than the required flow?
Passing flow is only one half of the requirement. An oversized valve drives larger inlet piping to hold the 3% inlet loss, raises header backpressure on adjacent valves, inflates the flare design load, and chatters on the smaller scenarios it actually sees in service.
How do I avoid a tube-rupture PSV entirely?
Apply the 10/13 screening rule from API 521: where the low-pressure side design pressure is at least 10/13 of the high-pressure side, dedicated tube-rupture relief is generally not required. Verify the exact wording and qualifications in the edition your project is contracted to.
How do I handle a liquid-full low-pressure side where the PSV cannot respond fast enough?
Do not solve it with a larger orifice. Evaluate a rupture disc for faster opening, uprate the low-side design pressure, or address the surge through the mechanical design — a spring valve cannot lift within the timescale of a liquid-full pressure wave.