A 1 1/2" x 2" conventional spring-loaded safety valve, set at 180 bar, 1500# inlet flange, 300# outlet flange, discharging back to the suction of a piston-type positive displacement pump running at 145 bar. The recurring question on this arrangement is whether the relieved liquid pressurises the suction hard enough to hold the valve open and lock the system into a loop. It does not, and the reason is worth following hop by hop.
Follow the fluid. The nozzle throttles the jet, the bowl expands it, the outlet flange feeds a pipe, and that pipe terminates somewhere with a defined pressure. The pressure at the outlet flange is set by what is downstream of it, not by what is upstream. Everything below is a check sequence against that path.
Check 1: Where does the outlet line actually terminate?
Read the P&ID, not the valve datasheet. On this system the outlet line ties into the pump suction line. That single fact fixes the answer to the outlet-pressure question before any calculation is done: pressure at a point is pressure at a point. Where the discharge line joins the suction line, the two are at the same pressure. The outlet flange therefore sees the suction-line pressure, plus the friction loss of the outlet pipe at relieving flow, plus or minus the static head between the flange and the tie-in.
P_outlet_flange = P_suction_tie_in + dP_friction(outlet line @ relieving rate) + rho*g*h
Total back pressure = superimposed (valve closed) + built-up (valve open)Superimposed back pressure is what the outlet flange sees with the valve shut — here 0.7 bar, i.e. the static suction pressure. Built-up back pressure is the additional rise caused by the relief flow itself passing through the outlet pipe — here 6.1 bar. Total at full lift: 6.8 bar.
Branch: if the outlet line terminates in a closed or restricted system, or shares a header with other valves that can lift simultaneously, the superimposed component becomes variable and Check 3 changes outcome. If it terminates at an open suction vessel or a low-pressure suction line, superimposed back pressure is essentially constant.
Check 2: Is 50 bar a pressure or a rating?
This is where the analysis usually goes wrong. A 300# outlet flange is a pressure-temperature rating — the maximum the joint may see, taken from the ASME B16.5 rating table for the material group and design temperature. It is not a prediction of operating pressure. Confusing the two produces the false conclusion that the outlet will deliver roughly 50 bar into the suction line and re-lift the valve.
The measured numbers say otherwise: 6.8 bar total against a rating of tens of bar. The rating is a containment envelope with a wide margin, not a design point.
| Quantity | Value on this system | What sets it |
|---|---|---|
| Set pressure | 180 bar | Spring, MAWP of protected equipment |
| Normal operating pressure | 145 bar | Pump discharge system resistance |
| Superimposed back pressure | 0.7 bar | Suction-line static pressure |
| Built-up back pressure | 6.1 bar | Outlet pipe friction at relieving rate |
| Total back pressure | 6.8 bar | Sum of the above |
| Outlet flange class | 300# | API 526 tabulated pairing / B16.5 rating |
Two outlet sizes appear in the field notes for this valve — 2" and 3". Settle it from the nameplate and the certified drawing before any hydraulic calculation, because the outlet bore drives the built-up back pressure directly.
Check 3: Why is the outlet class lower than the inlet class?
API 526 fixes the inlet and outlet flange size-and-class combinations for flanged steel pressure relief valves by orifice designation. Manufacturers do not calculate a new outlet class per order; they read the table. A 1500# inlet in the small orifice range is paired with a 300# outlet as standard.
The physical justification is the expansion across the nozzle. Set pressure exists only at the nozzle bore. Downstream of the disc the jet expands into the body bowl, which has an order-of-magnitude larger flow area, and the static pressure collapses — the drop across that expansion can reach 90% of set pressure. The bowl and outlet flange never see inlet pressure while the valve is passing flow.
Do not turn that 90% into a design number. It describes the nozzle-to-bowl expansion, not the pressure at the end of your outlet pipe. The outlet flange pressure comes from Check 1 and nothing else. If the outlet pipe were fully blocked, the bowl would eventually equalise toward inlet pressure and the 300# flange would be the weak point — which is exactly why isolation valves in relief discharge lines are prohibited or car-sealed open.
Check 4: Is a conventional valve still valid at 6.8 bar back pressure?
A conventional spring-loaded valve has its bonnet vented to the discharge side, so back pressure acts on the top of the disc and adds directly to the spring force. Two consequences follow:
- Superimposed back pressure shifts the opening pressure upward, roughly one-for-one. With 0.7 bar superimposed, the valve opens near 180.7 bar at the inlet.
- Built-up back pressure fights the lift. Once the valve cracks, the pressure rise in the outlet line pushes the disc closed. If built-up back pressure is large relative to the allowable overpressure, the valve cannot achieve or hold full lift and it chatters.
Standard practice limits built-up back pressure on a conventional valve to about 10% of set pressure; confirm the exact allowable against API 520 Part I and the manufacturer's certified back-pressure correction curve for this trim. Here, 6.1 bar on a 180 bar set is 3.4%, and total back pressure is 3.8%. The conventional design is the correct selection for this service.
Branch: if a revamp adds outlet line length, a header tie-in, or another valve that can lift into the same line, recompute built-up back pressure at the rated relieving rate. Cross the 10% region and the answer changes to a balanced-bellows or pilot-operated valve.safety valve for all of these regardless of application or design — the API terminology split does not exist in that standard, which matters when reading a European datasheet.
Check 5: Can the suction-pressure rise re-lift the valve?
Take the worst case that was feared: relief flow raises suction pressure by some amount ΔP. A positive displacement pump does not develop a fixed discharge pressure; it develops whatever pressure the discharge system demands, on top of whatever suction pressure it is given. So discharge pressure rises by the same ΔP, and the PSV inlet sees 145 + ΔP.
But that same ΔP is superimposed back pressure on a conventional valve, and it raises the opening pressure by ΔP as well. To first order the two track and cancel. There is no runaway lift-up mechanism from suction repressurisation on this configuration. At the observed 6.8 bar, the inlet would reach roughly 152 bar against an opening pressure near 187 bar — nowhere near lift.
What does bite on suction recirculation is thermal, not hydraulic. A relief valve dumping full pump flow back to suction converts the entire pump shaft power into heat in a closed circuit with no external sink. Liquid temperature climbs, viscosity falls, vapour pressure rises, and the pump cavitates. Relief-to-suction is acceptable for a short-duration excursion; if the credible relief case is a sustained blocked discharge, route the discharge to a vessel with volume and cooling, or fit a temperature switch that trips the driver.
Check 6: Is the set-pressure margin adequate for a piston pump?
180 bar over 145 bar is about a 24% margin. The working rule for pulsating PD pump service is that set pressure should sit at least 40% above maximum normal operating pressure — 145 x 1.4 = 203 bar on this system. The configuration breaches it, and that is the single most likely source of the operating trouble.
The mechanism is peak versus mean. Your discharge gauge is damped and reads the mean. A reciprocating piston pump superimposes a pressure wave on that mean whose amplitude depends on plunger count, single or double acting, and the acoustic response of the discharge piping. A simplex or duplex pump can put peaks tens of percent above the mean at the valve inlet. When those peaks reach the simmer point of a metal-seated conventional valve, the disc lifts momentarily, the seat leaks, and the leak erodes the seat until it leaks continuously.
Decision path when the margin is short:
- Confirm the set pressure ceiling. Set pressure cannot exceed the MAWP of the protected equipment — check the pump and piping data before assuming you can raise it.
- If headroom exists to MAWP, raise set pressure toward the 40% margin and re-verify the outlet class and back-pressure percentages against the new set point.
- If it does not, lower the operating pressure, or attack the pulsation amplitude directly: correctly sized suction and discharge pulsation dampeners with verified charge pressure, shortened unsupported pipe runs, and an acoustic study per the reciprocating pump pulsation-control requirements of API 674.
- Do not rely on a dampener to make an inadequate margin acceptable. Charge pressure drifts, bladders fail, and the margin disappears silently.
Check 7: Vibration and seat condition
Excessive vibration on relief valve branch connections drives fatigue cracking at the weld toe of the inlet nozzle — a small-bore connection cantilevered off a pulsating line is a classic failure geometry. Brace the outlet line for the relief reaction force as well; calculate it per API 520 Part II and anchor to structure, not to adjacent piping.
| Symptom | Most likely cause | Reading that confirms it |
|---|---|---|
| Seat leaks continuously below set | Operating pressure too close to set; pulsation peaks reaching simmer | High-response transducer at PSV inlet, not the damped gauge |
| Valve rattles/chatters at lift | Built-up back pressure too high, or valve oversized for actual pump rate | Built-up back pressure as % of set; required rate vs certified capacity |
| Scored disc and nozzle faces | Solids harder than the seat metal carried through at lift | Lap the seat and inspect; strainer differential trend |
| Cracks at inlet branch weld | Pulsation-driven fatigue on cantilevered small-bore connection | Vibration survey, dye penetrant at the weld toe |
| Pump cavitation after a relief event | Thermal buildup in the closed relief-to-suction circuit | Suction temperature during and after relief |
On the seat itself: whether dirt scratches the faces depends on the hardness of the particles relative to the seat metal. Harder particles will score both disc and nozzle, and once scored the valve leaks and the leak accelerates the damage. Where the service carries solids or the valve sees vibration, a soft-seated (elastomer or PTFE insert) design tolerates the duty far better than metal-to-metal and holds tighter close to set. Confirm the soft seat material is compatible with the fluid and the design temperature before specifying it.
Procedure and verification
- Confirm outlet size and class from the valve nameplate and certified drawing; resolve the 2" versus 3" discrepancy in the records.
- Recalculate built-up back pressure at the rated relieving rate using the confirmed outlet bore, the actual pipe run, and every fitting to the suction tie-in. Add the measured superimposed pressure.
- Express total back pressure as a percentage of set pressure. Compare against the conventional-valve limit in API 520 Part I and the manufacturer's back-pressure correction curve. Below roughly 10% built-up, keep the conventional valve; above it, respecify as balanced bellows or pilot-operated.
- Verify the outlet flange class against the ASME B16.5 rating table at design temperature, and confirm there is no isolation valve or removable spool in the discharge path.
- Install a high-response pressure transducer at the PSV inlet and log the true peak, not the gauge mean, over a full operating cycle.
- Compare peak to set pressure. If the peak exceeds roughly 90% of set, act on the margin: raise set pressure within MAWP, lower operating pressure, or fix the pulsation source.
- Pop-test the valve on the bench, record cracking and reseating pressure, and inspect the disc and nozzle faces. Convert to a soft seat if the faces show particle scoring.
- Log suction temperature through a controlled relief event of realistic duration to size the thermal exposure of the recirculation loop.
Final verification: with the valve reinstalled and the pump at full rate, log inlet pressure at the PSV for one hour at full sampling rate and confirm no sample exceeds 90% of the certified set pressure, and that seat leakage measured at the outlet flange tap is zero.
FAQ
How do I calculate the pressure at a PSV outlet flange?
Add the terminal pressure at the discharge line's tie-in point, the friction loss through the outlet pipe at the rated relieving flow, and the static head between the flange and the tie-in. On a suction-return arrangement with 0.7 bar superimposed and 6.1 bar built-up, the flange sees 6.8 bar — not a fraction of the 180 bar set pressure.
How do I know if a 300# outlet flange is acceptable on a 1500# inlet valve?
API 526 tabulates the standard inlet-and-outlet size and class pairings by orifice designation, and a 1500# inlet in the small orifice range is paired with a 300# outlet by that table. Verify the calculated total back pressure stays well inside the ASME B16.5 rating for the outlet class at design temperature, and confirm there is no block valve that could dead-head the discharge.
How do I decide between a conventional and a balanced-bellows relief valve?
Express built-up back pressure as a percentage of set pressure. Below roughly 10%, a conventional valve is valid; at 6.1 bar on a 180 bar set that is 3.4%, so the conventional design is correct here. Above that limit, or where superimposed back pressure varies, move to balanced bellows or pilot-operated and check the manufacturer's back-pressure correction curve.
How do I set PSV set pressure for a reciprocating positive displacement pump?
Target at least 40% above maximum normal operating pressure to stay clear of pulsation peaks — 145 bar operating calls for about 203 bar set. A 24% margin leaves the disc simmering on pressure peaks that a damped discharge gauge never shows, which is what erodes the seat.
How do I stop a PSV from re-lifting when it discharges into the pump suction?
On a conventional valve it will not re-lift from that path: the suction pressure rise raises pump discharge pressure and the valve opening pressure by the same amount, so they cancel. Watch the thermal case instead — log suction temperature during a sustained relief event, and route the discharge to a cooled vessel if the blocked-discharge case can last more than a few minutes.