A 3" line running to a 1-1/2" safety valve inlet forces a concentric reducer at the last joint, and on natural gas at 10 barg that single fitting can eat the entire 3% allowance before the pipe friction is even added. Two separate questions are hiding inside that result: whether the reducer belongs inside the accounting boundary at all, and whether the number the flow calculation printed is actually a non-recoverable loss. Work them in that order.
Boundary of the Inlet Piping
The 3% applies to the inlet piping, and inlet piping means everything from the protected equipment down to the inlet flange of the relief valve. There is no carve-out for the fitting that adapts line size to valve size. A concentric reducer bolted or welded immediately upstream of the valve is inlet piping, not part of the valve, and its non-recoverable loss goes into the budget along with the vessel nozzle, the pipe, the elbows, the block valve, and any rupture disk.
The reason the boundary sits at the valve inlet flange is mechanical, not bureaucratic. The valve senses pressure at its own inlet. Anything that depresses pressure between the vessel and that sensing point subtracts from the force holding the disc open once flow starts. The reducer sits inside that path.
- Print the isometric of the relief line and mark two points: the vessel nozzle face and the PSV inlet flange face.
- List every element between those marks, including reducers, spool pieces, tees, and the entrance geometry at the nozzle.
- Confirm nothing on that list has been silently assigned to the valve datasheet.
Do not move on until the element list and the isometric agree item for item.
Loss Budget and the Flow Rate It Is Evaluated At
Fix the number the calculation has to beat before touching the hydraulics. At a 10 barg set pressure, 3% gives 0.30 bar (300 mbar) of allowable non-recoverable loss for the whole inlet path. Check your governing document for whether the percentage is taken on set pressure alone or on set pressure minus superimposed back pressure; the second form applies when a constant back pressure is present and it shrinks the budget.
The flow used in the calculation is the rated capacity of the installed valve, not the required relief rate. This is where most over-budget surprises get manufactured in reverse: an engineer sizes the orifice on a modest required flow, then computes inlet loss on that same modest flow and declares 1%. The valve does not know your required flow. When it lifts, it passes what the certified orifice will pass at relieving conditions, and the inlet piping has to hold 3% at that flow.
| Input | Value to use | Common error |
|---|---|---|
| Flow rate | Rated capacity of the selected orifice at relieving pressure | Using the required relief rate |
| Density | Gas density at relieving pressure and relieving temperature | Normal or design-condition density |
| Method | Incompressible Darcy-Weisbach plus fitting K-factors | Compressible/isothermal solver that folds in expansion effects |
| Budget | 3% of set pressure = 0.30 bar at 10 barg | 3% of relieving pressure |
Gas is handled with the incompressible equation deliberately. Density is evaluated once at relieving conditions and held constant, so the calculation returns friction only and does not smuggle in the expansion-driven pressure change that a compressible solver would report.
Stripping Recoverable Loss Out of the Reducer Number
A 3" to 1-1/2" contraction quadruples the flow area ratio, so the velocity head at the small bore is sixteen times the velocity head in the 3" pipe. Most flow programs report the total static pressure change across that fitting, and the dominant term in it is Bernoulli acceleration, which is recoverable. Only the loss coefficient term is non-recoverable and only that term goes into the 3%.
Static change across a contraction (constant density form):
dP_static = (rho/2)*(v2^2 - v1^2) <- acceleration, RECOVERABLE, exclude
+ Kc*(rho*v2^2/2) <- friction/turbulence, INCLUDE
v1 = velocity in the 3" pipe
v2 = velocity in the 1-1/2" bore
beta = 1.5/3 = 0.5
Kc referenced to the DOWNSTREAM velocity head (v2)
Read KcFor a sudden contraction at beta = 0.5 the classical correlation 0.5*(1 - beta^2) lands near 0.375; a gradual concentric reducer with a shallow included cone angle is far lower, commonly an order of magnitude below the sudden case. If your software modeled the reducer as an abrupt area change, it has already overstated the recoverable-corrected loss on top of overstating the total.
Recompute the reducer by hand with the two terms separated. If the 3% collapses to a fraction of a percent once the acceleration term is removed, the piping was never the problem and the calculation was. That result is the same one that shows up on liquid services where the total static change exceeds 3% while the frictional component sits under 1%.
One liquid-specific caution: elevation head between the vessel nozzle and an elevated valve is also excluded from the 3%, because it is present at no-flow as well as at flow and therefore does not create a flowing-versus-static differential. Account for it in the set pressure instead.
The Element-by-Element Loss List
Build the total from discrete terms so each one can be defended and each one can be attacked if the budget is tight.
| Element | Counts toward 3%? | Reference velocity |
|---|---|---|
| Vessel nozzle entrance (sharp-edged or projecting) | Yes | Line velocity, v1 |
| Straight pipe friction, f*(L/D) | Yes | Line velocity, v1 |
| Elbows, tees, full-bore block valve | Yes | Local line velocity |
| Rupture disk in series, if fitted | Yes | Per disk manufacturer's resistance |
| Concentric reducer at the valve | Yes, Kc term only | Small-bore velocity, v2 |
| Acceleration / velocity-head change | No | — |
| Elevation head (liquid) | No — handle in set pressure | — |
| Losses downstream of the valve inlet flange | No — that is back pressure | — |
dP_inlet = Ke*(rho*v1^2/2)
+ f*(L/D)*(rho*v1^2/2)
+ sum(Kf)*(rho*v1^2/2)
+ Kc*(rho*v2^2/2)
Accept when dP_inlet <= 0.03 * P_set(gauge) [0.30 bar at 10 barg]
On a long run — and a double-wall vessel forces a long run, because the nozzle has to cross the annulus before it reaches open air — verify the friction term separately from the fitting terms. Knowing which of the two dominates decides the next section.
Fixing an Over-Budget Inlet
Work the options in this order; each one is cheaper or lower-risk than the one after it.
- Correct the calculation first. Separate recoverable from non-recoverable, use the real reducer geometry, confirm the rated-capacity flow. A large share of over-budget results die here.
- Shorten the run or remove fittings. Every elbow deleted and every metre removed is direct budget recovered at v1^2.
- Delete the reducer by running the small line all the way. If the valve inlet is 1-1/2", run 1-1/2" pipe and flanges from the vessel to the valve. The contraction then happens at the vessel nozzle, where an entrance loss was already being charged, and the discrete reducer term disappears. With a threaded male NPT valve inlet, the last flange can be a blind flange with a tapped hole, which turns a small threaded valve into something that mounts like a 1-1/2" flanged valve while keeping the flange joint accessible for maintenance.
- Re-check friction after downsizing. Smaller bore means higher v1 and higher friction per metre. This trade only wins when the flow is modest — which it is when the required relief rate allows a small orifice — or when the run is short. Recompute the whole line; do not assume the swap is an improvement.
- Change the valve. A smaller certified orifice that still covers the relief case lowers the rated flow and therefore the loss. A pilot-operated valve with a remote pressure pickup taken at the vessel removes the inlet loss from the sensing path entirely.
- Perform a documented engineering analysis. Where the recognised standards permit an alternative to the 3% screening rule, the analysis compares computed inlet loss against valve blowdown and system dynamics. Treat it as the last resort, and keep the calculation in the relief file.
A balanced-bellows trim does not buy relief here. The bellows addresses variable back pressure on the outlet side; the inlet loss still subtracts from the pressure the disc sees.
Why Inlet Loss Destabilizes a Conventional Valve
Before flow starts, the valve inlet sees vessel pressure. The instant the disc lifts, flow accelerates through the inlet line and the non-recoverable loss appears, so the pressure at the valve inlet drops below vessel pressure by exactly the amount just calculated. If that drop exceeds the valve's blowdown, the spring reseats the disc, flow stops, the loss vanishes, inlet pressure recovers to vessel pressure, and the valve pops again. The result is rapid cycling — chatter — at a frequency set by the acoustics of the inlet line rather than by the process.
Acceleration loss does not participate in that mechanism. Static pressure falls at the small bore because velocity head rose; that energy is still in the stream and it reappears as the flow decelerates in the valve body. On top of that, the fully open disc is not held by static pressure difference alone. Momentum from the jet impinging on the disc face carries a large share of the lifting force — the same reason a water jet at atmospheric static pressure still pushes hard on whatever it hits.
| Symptom | Likely cause | Check |
|---|---|---|
| Rapid cycling immediately on lift, seat and guide damage | Non-recoverable inlet loss exceeds blowdown | Recompute at rated capacity; compare to blowdown setting |
| Calculation flags >3% but friction alone is <1% | Velocity head included in the reported static drop | Separate the two terms by hand at the reducer |
| Loss passes on paper, chatters in service | Sized on required flow, not rated capacity | Re-run with certified orifice capacity |
| Reducer dominates the budget | Modeled as sudden contraction, or Kc referenced to v1 | Use gradual-reducer K, referenced to v2 |
Two features of this installation work in favour of stability but do not license exceeding the budget: a valve that stays open until pressure falls well below set gives a wide blowdown margin, and a vessel of several cubic metres repressurizes slowly after a discharge, so re-lift is not immediate. Note both in the relief file; justify the design on the 3% number.
End-to-End Verification
- Confirm the calculation boundary runs from vessel nozzle face to PSV inlet flange face, reducer included.
- Confirm the flow input equals the rated capacity of the selected orifice at relieving conditions.
- Confirm density is taken at relieving pressure and temperature, and that the method is incompressible Darcy with K-factors.
- Hand-check the reducer: acceleration term removed,
Kctaken for the actual cone geometry, referenced to the 1-1/2" velocity head. - Sum the element table and compare against 0.30 bar for the 10 barg set pressure. Record the margin, not just the pass.
- Compare the computed inlet loss against the valve's blowdown. Loss must sit below blowdown with margin, not merely below 3%.
- Walk the installed line against the isometric before commissioning: no unlisted reducer, no partially open or non-full-bore block valve, no undocumented spool.
- File the calculation, the K-factor sources, and the rated-capacity basis with the valve datasheet, and re-run the sum whenever the orifice, the trim, or the line routing changes.
FAQ
Why does the concentric reducer at the PSV inlet count toward the 3% limit?
Inlet piping is defined as everything from the protected equipment to the inlet flange of the relief valve, and the reducer sits inside that boundary. The valve senses pressure at its own inlet, so any non-recoverable loss upstream of that flange — including the adapting fitting — subtracts from the pressure holding the disc open.
Why does my static pressure drop exceed 3% when the frictional loss is under 1%?
The reported static change across a contraction includes the Bernoulli acceleration term (rho/2)*(v2^2 - v1^2), which is recoverable and excluded from the 3% rule. Only the Kc*(rho*v2^2/2) loss term counts. On a 3" to 1-1/2" step the velocity head rises sixteen-fold, so acceleration dominates the raw number.
Why must inlet loss be calculated at rated capacity instead of the required relief flow?
When the valve lifts it passes whatever the certified orifice will pass at relieving conditions, regardless of what the relief case demanded. Computing loss on the smaller required flow understates it by roughly the square of the flow ratio and can turn a real 4% into a paper 1%.
Why does excessive inlet loss cause a conventional safety valve to chatter?
Once flow starts, the non-recoverable loss depresses pressure at the valve inlet; if that drop exceeds the valve's blowdown the spring reseats the disc, flow stops, pressure recovers, and the valve pops again. The cycle repeats at line-acoustic frequency and destroys the seat and guides. Verify computed loss against blowdown, not only against 3%.