Supporting Relief Valve Piping — Reaction Forces, Stanchions and Load Path

A pressure relief valve that opens puts a large, sudden thrust on the weakest assembly in the system — a small-bore inlet nozzle, a valve body with its own flange-load limits, and an outlet flange that is often a lower pressure class than the inlet. Getting the piping right is a load-path problem: the reaction has to reach structural steel by the shortest, stiffest route without passing through the valve and its nozzle on the way. Code basis: ASME B31.3-2024 (¶301.5.5 discharge reactions, ¶322.6.2 pressure-relieving systems), ASME B31.1-2024 (Nonmandatory Appendix II, safety valve installations), API Standard 520 Part II (7th Ed., 2020), API Standard 521, and, where invoked, PIP PNC00004 (Feb 2025) and PIP PNC00001 (May 2023, Complete Revision).

First published 2026-09-23 · content last reviewed 2026-09-23 · page regenerated 2026-09-23 at build.

Written and reviewed by Matthew Norris, P.E. — active P.E. licensure in Arizona, California, Kansas, Missouri, North Carolina, Texas.

Why relief valve piping gets its own rules

Two things make a relief-valve discharge load unusual. It is large and sudden — the valve opens in milliseconds, faster than the piping can respond, and a suddenly applied load can produce up to twice the static response. And it acts on the weakest assembly in the system: a PSV typically sits on a short, often small-bore inlet nozzle, its outlet flange is frequently a lower pressure class than its inlet, and the valve body is a casting with its own flange-load limits. Put a moment arm between the discharge thrust and that inlet, and an overstressed nozzle, a leaking outlet flange, or a valve body distorted enough to stop reseating are all real failure modes.

ASME B31.3-2024 ¶301.5.5 requires piping to be designed, arranged and supported to withstand discharge reactions from let-down or discharge of fluids. ¶322.6.2, within the pressure-relieving-systems clause, states the point directly for relief devices: reactions on the piping system from actuation of a safety relief device must be considered, with adequate strength provided to withstand them. The Code sets the requirement and leaves the calculation method to you — API 520 Part II and B31.1 Appendix II fill that gap.

Where PIP PNC00004 is invoked, relief systems — open or closed, including tailpipes — sit on the stress-critical list that needs formal analysis (§3.4.1), and relief-discharge piping must be restrained for thrust using an owner-approved calculation method (§3.2.7). Where PIP PNC00001 is invoked, §4.2.8 calls for the reaction forces of safety relief devices to be evaluated and supported as needed, and separately flags that closed-system relief lines carrying two-phase flow should be restrained to control line vibration.

Getting the reaction force

Order of preference for the thrust value: a vendor-supplied reaction force for the selected valve, orifice and relieving conditions, recorded with its datasheet revision, is best when you have it. Absent that, calculate it with API 520 Part II (open systems) from the relieving rate and conditions, flagged in the calc as provisional until the vendor value arrives. For steam safety valves, B31.1 Nonmandatory Appendix II gives a dedicated method that includes its own dynamic load factor. Any thrust value — vendor or calculated — goes into the stress model as its own occasional load case; where a substantial thrust is expected and no value exists yet, estimate one from the flow rate, orifice size, pressure, temperature and valve model, and replace the estimate when the real number arrives.

API 520 Part II — open discharge, vapor or gas, US customary units

F = ( W / 366 ) · √( k·T / ((k + 1)·M) )  +  A_o · P₂

where, per ¶5.8.2.1, Eq. (1): F is the reaction force at the point of discharge, lbf; W is the flow rate through the valve (its rated capacity, for the reaction), lb/h; k is the ratio of specific heats at outlet conditions; M is the molecular weight of the flowing gas; A_o is the area of the outlet at the point of discharge, in²; and P₂ is the static pressure at the point of discharge, psig. T is the stagnation temperature at the pipe outlet, °R — the Standard notes the stagnation temperature is often unavailable and lets you approximate it, conservatively, with the relieving temperature, but cautions that this approximation is only good for the reaction-force calculation and must not be reused for other discharge-piping thermal design (auto-refrigeration or Joule-Thomson cooling, for instance).

The first term is the momentum flux at sonic velocity; the second is the pressure thrust from the exit static pressure above atmospheric, zero when the exit isn't choked. Use the valve's rated capacity, not the required relieving rate — the valve can pass its full rated flow whatever the process case needed.

Finding P₂ — standard compressible-flow theory

API 520 Part II ¶6.3.1 states that at a choked (sonic) discharge, the pressure within the exit of the pipe "is the calculated choking pressure," and points to an outside source for that calculation rather than publishing the equation itself. The ideal-gas isentropic relation below is the standard method for finding it:

P*(psia) = (W / 3600 / A_o) · √( R·g_c·T* / (k·M) ) / g_c    T* = T · 2/(k + 1),   R = 1545.35 ft·lbf/(lbmol·°R),   g_c = 32.174

If P* > 14.7 psia the exit is choked and P₂ = P* − 14.7 psig; if P* ≤ 14.7 psia the exit is subsonic and P₂ = 0 psig.

Dynamic load factor

API 520 Part II gives a steady-state force, but the load is still applied suddenly. B31.1 Appendix II — written for steam — treats the same physical problem with an explicit dynamic load factor: ¶II-3.5.1.3 states that for a one-degree-of-freedom, single-ramp idealization the DLF ranges between one and two, and that it shall never be taken less than 1.1. Common practice borrows that range for non-steam API 520-II reactions too — apply DLF = 2.0 to the API 520 force for an equivalent-static analysis as a common conservative default (ASSUMED unless your owner specification says otherwise, and never below the 1.1 floor), or run a dynamic analysis (time history or force spectrum) with the real opening time and support stiffnesses and use the computed response. A DLF below 2.0 needs justification: a known opening time, a computed natural period, or a time-history result.

PiperPSV today automates the API 520 Part I orifice sizing (gas/vapor, steam, liquid) and the B31.1 Appendix II open-discharge reaction force for steam service, with the DLF taken from ¶II-3.5.1.3 and the Table II-2.2.1-1 steam constants embedded. There is no API 520 Part II vapor/gas reaction card on the site yet — the choked-exit gas calculation in the worked example below is done by hand, the way you would run it today for a hydrocarbon-vapor relief case.

The load case, and the checks people forget

Apply the reaction as its own occasional load case, on top of sustained loads, at the discharge point and along the tailpipe axis:

B31.3-2024 ¶302.3.6(a):   S_L(sustained) + S_{(PSV reaction)} ≤ 1.33 · S_h

or, as many project specifications write it, W + P_o + R ≤ 1.33·S_h, where W is the stress from dead weight including operating contents and insulation, P_o is the stress from normal operating pressure, R is the stress from the relief discharge reaction including the DLF, and S_h is the basic allowable stress at design temperature (B31.3-2024 Table A-1, read from your own copy). Evaluate it separately from wind and seismic — relief discharge is its own occasional event.

Under B31.1-2024, ¶104.8.2 sets two duration-based occasional factors: k = 1.15 for loads acting no more than 8 hours at a time and 800 hours a year, and k = 1.2 for loads acting no more than 1 hour at a time and 80 hours a year. The clause states the two buckets; it does not say which one a safety-valve discharge belongs in — that classification is the designer's, or the owner specification's, based on the actual blow duration and annual frequency of the specific relief case. A brief, infrequent lift commonly falls in the 1-hour/80-hour, k = 1.2 bucket, but a service that cycles the valve repeatedly may not, and the classification belongs in the design basis with its stated reasoning, not left implicit.

The checks people forget

• Flanges at the PSV inlet and outlet. The outlet flange is often a lower class than the inlet — Class 150 on a Class 300-or-higher inlet is common. A widely used owner practice limits the flange moment from a discharge event to twice the allowable from ASME BPVC Section III, Division 1, Subsection NC, NC-3658 — a nuclear Class 2 piping rule some specifications borrow for non-nuclear flange screening because neither Section VIII nor B16.5 publishes an equivalent closed-form check. The commonly cited form is M_fd ≤ 6250·(S_y/36,000)·C·A_b — treat this as the shape of that owner practice, not a B31.1/B31.3 requirement, and take the equation, its terms and the factor of two from your own copy of Section III before applying it. See our bolted flange guide for the Kellogg equivalent-pressure screen and UG-44(b) as a Section VIII alternative.
• The valve body. Many PSV manufacturers publish allowable inlet and outlet flange loads; where they do, the stress report compares against them, and where they don't, ask. The body distorting enough to stop the valve reseating is a real failure mode.
• The vessel or header nozzle at the inlet. Check it with WRC 107 and WRC 297 — the pair PIP PNC00004 §3.3.5 names for vessel/nozzle local-stress evaluation within the bulletins' geometric limits — or WRC 537 or FEA where your specification permits going beyond them.
• Inlet pressure drop. Not a stress check, but decided by the same layout. API 520 Part II ¶7.3.4 states the total nonrecoverable pressure loss between the protected equipment and the PRV should not exceed 3% of the set pressure, with named exceptions for thermal relief valves, remotely sensed pilot valves, and an owner-approved engineering analysis per §7.3.6 — a recommendation with defined override paths, not an absolute prohibition, but one that pushes toward short, large-bore inlets, which also helps stiffness. Settle the layout with the process engineer before you add supports.

The support arrangement: put the steel in line with the thrust

The thrust acts along the tailpipe axis at the discharge point. Every inch of lever arm between that line of action and a support becomes moment on the valve and inlet, so: put the primary support where the thrust line meets the piping — under the outlet elbow, in the shadow of the pipe, so the thrust goes straight into steel; put a directional stop in the direction of the discharge thrust, since the stop carries the reaction and other restraints only guide; leave the thermal directions free, with gaps or a single stop per group of valves, so the support that resists thrust doesn't also lock in thermal expansion; and never use the valve body as a support, or load the inlet nozzle to carry the outlet.

Arrangement A — open discharge, vertical tailpipe

The common case: the PSV sits on a vertical inlet, discharges sideways into a short horizontal outlet, then turns up through an elbow into a vertical tailpipe that vents to a safe location. Flow exits upward, so the reaction acts downward along the tailpipe into the outlet elbow.

A common owner practice puts a stanchion (dummy leg or trunnion) welded to the outlet elbow, landing on steel directly below it, so the downward thrust goes axially through the stanchion into the structure while the short horizontal run between the valve and the elbow carries almost no bending. A directional stop at the stanchion base, in the horizontal direction of the outlet run, keeps any horizontal component — or the brief transient before flow is fully established — from swinging the tailpipe, with a small gap for thermal growth if the tailpipe is long. Keep the tailpipe short and straight, with a drain hole at the low point and a rain cap or weather hood that doesn't redirect the thrust — a cap that turns the flow sideways converts axial thrust into moment. The thermal interaction between the vessel and this stanchion is usually where designers get it wrong; see below.

Arrangement B — PSV in piping, supported at the inlet elbow

The PSV sits on a vertical inlet leg that rises from a horizontal line through an elbow. A common owner practice for this layout puts a hold-down and resting support, plus a stop in the direction of PSV discharge, on the inlet elbow below the relief valve.

One stop perpendicular to the discharge direction is typically added, with a gap where thermal movement needs it. Where several relief valves on one vessel form a set, the perpendicular stop commonly goes on only one of the inlet pipes, so the group isn't locked against differential thermal growth — the same supports may instead land on the vertical inlet leg below the valve rather than on the bottom elbow. If the inlet can't carry these stops, the outlet must, with an outlet-side stanchion and directional stop as in Arrangement A. The logic: the thrust reaches steel through a hold-down and stop that are rigid in the discharge direction and close to the valve, while the perpendicular direction stays free — or stops at one point only — for thermal growth.

Arrangement C — closed discharge to a relief header

The inlet comes off a header or vessel, and the PSV discharges through an outlet leg and elbow into a lateral that runs to the flare header.

A common owner practice puts trunnions or dummy legs in the shadow of the pipe at the inlet and outlet elbows, each with a guide, plus a directional stop on the outlet side in line with the discharge momentum. Header entry is typically a 45° lateral in the direction of header flow, sloped toward the header with no pockets, since collected liquid becomes a slug. During opening, a pressure wave travels down the discharge line and puts an unbalanced force on each pair of elbows in turn — roughly the pressure difference across the wave front times the pipe area — so restrain every long straight run between elbows axially, or show by dynamic analysis that the unrestrained response is acceptable. Size restraints for two-phase and slug loads where API 521 flags them; PIP PNC00004 §3.2.8 requires both hammer and slug flow to be considered where invoked, and PIP PNC00001 §4.2.8 separately notes that closed-system relief lines with two-phase flow should be restrained to control line vibration. The header tie-in also moves thermally (hot or cold flaring, sun, auto-refrigeration of flashing liquid), so the outlet supports have to allow that movement while still restraining thrust — usually a directional stop in the thrust direction with guides elsewhere, not a full anchor.

Thermal movement and support stiffness

The thermal trap: a stanchion on the wrong steel

A PSV on top of a vessel moves with the vessel — when the vessel heats up, its top nozzle rises. Suppose the outlet stanchion sits on a platform supported from the ground or a separate structure, not from the vessel: hot, the vessel lifts the PSV and outlet, and the stanchion lifts off its steel, with the model showing that support inactive. When the valve opens, the thrust has to push the outlet back down through the full thermal gap before the stanchion picks up load — the inlet and valve take the whole reaction until the gap closes, and then the pipe hits the support, adding impact to the dynamic load. Cold, or during steam-out, the reverse can happen: the stanchion can be overloaded holding the pipe against thermal contraction.

Fixes, in order of preference: support the stanchion from steel that moves with the vessel — a vessel-clip-supported platform or bracket — so relative movement stays near zero; choose the arrangement so the thrust acts in a direction with no significant thermal movement (horizontal thrust on a stop, for example, where the vessel's radial growth at that elevation is small) and give the stop a minimal gap; or show by nonlinear or time-history analysis that the gap stays closed in every operating state, or that the dynamic response with the gap is acceptable. Always run the installed (cold) and operating cases with the thrust applied, and report whether the thrust support is active in each state.

Stiffness decides how much thrust the stanchion actually takes

A stanchion only protects the valve if it is stiffer than the path through the valve and inlet — the two paths share the thrust in proportion to their stiffness. The stanchion's own axial stiffness (AE/L) is high, but it is in series with whatever it lands on: a stanchion on the midspan of a light platform beam can be softer than the valve and inlet assembly, in which case most of the thrust still goes through the valve.

Our support stiffness guide covers this in detail. Land thrust stanchions on stiff steel, at or near a column or a braced point, and model their stiffness — don't let them default to rigid. A rigid default will tell you the stanchion takes 100% of the thrust when it may take 20%.

Worked example — open-discharge vapor PSV with a stanchion at the outlet elbow

This is a hand calculation — the tools table below shows which pieces PiperPSV runs today and which are worked by hand here.

1 — Is the exit choked?

Mass flux: G = 150,000/3600/28.89 = 1.442 lb/(s·in²). Throat temperature: T* = 759.7 × 2/2.3 = 660.6 °R. √(R·g_c·T*/(k·M)) = √(1545.35 × 32.174 × 660.6 / (1.3 × 44)) = 757.8 ft/s. P* = 1.442 × 757.8 / 32.174 = 33.97 psia, above 14.7, so the exit is choked. P₂ = 33.97 − 14.70 = 19.27 psig.

2 — API 520 Part II reaction

Momentum term: (150,000/366) × √(1.3 × 759.7 / (2.3 × 44)) = 409.8 × 3.124 = 1,280 lbf. Pressure term: 28.89 × 19.27 = 557 lbf. F = 1,837 lbf (steady state). With DLF = 2.0 (ASSUMED, never below the 1.1 floor of B31.1 ¶II-3.5.1.3): F_design = 3,674 lbf. The thrust acts downward along the vertical tailpipe (flow up).

3 — Without a stanchion

The thrust reaches the valve and inlet through an 18 in arm: M = 3,674 × 18 = 66,130 in-lb (5,510 ft-lb). Bending stress in the NPS 4 inlet from the reaction alone: 66,130/3.21 = 20,600 psi, before any SIF, sustained stress or flange check. Against 1.33·S_h, that is likely to fail on its own, and will certainly fail once the nozzle or elbow SIF is applied — the valve body and a Class 150 outlet flange are very unlikely to accept 5,500 ft-lb either.

4 — With a stanchion under the outlet elbow

Stanchion: NPS 4 Sch 40, 36 in tall. Axial stiffness = 3.17 × 29×10&sup6; / 36 = 2.55 × 10&sup6; lb/in. Stiffness of the path through the valve and inlet: 50,000 lb/in (ASSUMED — get it from the model).

The finding. The stanchion only does its job on stiff steel. On a soft platform beam it takes less than a fifth of the thrust, even though a rigid-default model would say it takes all of it. Land it at a column or braced point, model its real stiffness, and the inlet bending from the discharge drops from about 20,600 psi to about 1,340 psi.

5 — Remaining checks

Stanchion: axial compression with buckling, plus the lateral moment from any eccentricity — the free AISC 360 steel member check. Stanchion-to-elbow attachment: local stress at the trunnion on the elbow needs a trunnion-on-elbow method or FEA, since WRC 107 does not cover elbows; PIP PNC00001 §4.2.3.2 says no reinforcing pads at elbows unless otherwise specified. Base plate and anchorage to the steel. Outlet flange: the NC-3658 occasional owner-practice screen (2×, confirmed in your own copy of Section III) or UG-44(b) — see the flange guide. Valve body: against the vendor's allowable flange loads. Thermal: confirm the stanchion is active in the operating case, per the thermal trap above.

PSV support checklist for the stress report

• System type stated (open or closed) and reaction method named: vendor, API 520 Part II (edition), or B31.1 App. II
• Rated capacity used, not required capacity; choked or subsonic exit shown
• DLF stated and justified (2.0 default, never below 1.1, or dynamic analysis)
• Reaction applied as a separate occasional case, checked to B31.3-2024 ¶302.3.6(a) (or the B31.1 duration factor, with the duration/frequency classification stated)
• Inlet nozzle local stress, inlet and outlet flanges, and valve-body allowable loads checked
• Thrust support in line with the thrust (stanchion under the outlet elbow, directional stop in the thrust direction)
• Support stiffness modeled for thrust supports, not left rigid
• Thrust support shown active in the operating and installed cases (vessel thermal growth, steam-out)
• Perpendicular stops limited to one per valve group where thermal growth needs it
• Closed systems: elbow-pair transient loads restrained or shown acceptable by dynamic analysis; slug and two-phase loads considered; header top-entry, slope and no pockets
• Provisional reaction flagged until the vendor value is received

Tools for this workflow

The API 520 Part II vapor/gas reaction with the choked-exit check worked through this guide is a hand calculation today — PiperPSV's live reaction card covers the B31.1 Appendix II steam case.

FAQ

Do I have to use a dynamic load factor with API 520 Part II reaction forces?

The API 520 Part II equation gives the steady-state reaction force; the valve applies it suddenly, and a suddenly applied load can produce up to twice the static response. Use an equivalent-static dynamic load factor of 2.0 as a common conservative default unless a dynamic analysis, or the ASME B31.1 Appendix II method for steam, justifies a lower value. Appendix II's own DLF rule, at ¶II-3.5.1.3, is written for a one-degree-of-freedom, single-ramp idealization: the factor ranges between one and two, and the Code states it shall never be taken less than 1.1, whatever the computed opening-time-to-natural-period ratio suggests. That floor matters because a system with a fast valve and a stiff, well-supported discharge line can compute a DLF close to unity from the ratio alone — the 1.1 floor exists precisely so that result is never taken at face value. A DLF below 2.0 for a non-steam API 520-II case needs its own justification: a known opening time, a computed natural period, or a time-history result, not just an assumption that the piping is stiff.

Should the relief reaction be combined with wind or seismic?

Not normally. Relief discharge is its own occasional event, checked as sustained plus the reaction against 1.33×S_h under ASME B31.3-2024 ¶302.3.6(a), or against k×S_h under B31.1-2024 ¶104.8.2, with k set by the actual duration and frequency of the discharge rather than by the Code itself — B31.1 doesn't say which of the 1.15 or 1.2 buckets a PSV lift falls into; that classification is the designer's, based on how long and how often the specific relief event actually runs. Combine the relief reaction with wind or earthquake only if your owner specification explicitly requires concurrent events; B31.1 Appendix II leaves that combination to the design specification rather than mandating it. Keep the occasional cases separate in the stress model too: a wind case, a seismic case in each direction, and the relief case are each their own load case built as sustained plus one event, not summed together into a single "everything at once" run. Bundling them produces a result nobody can trace back to which load actually governs, and a reviewer will ask you to split it out.

Why put the support under the outlet elbow rather than on the inlet?

The discharge thrust acts along the tailpipe axis at the point of discharge. A stanchion or trunnion placed directly under the outlet elbow puts structural steel on that line of action, so the reaction goes straight into the structure with almost no lever arm. Support only the inlet side instead, and the full outlet lever arm — the horizontal distance from the valve centreline to the tailpipe — acts as a moment on the valve body, the outlet flange and the inlet nozzle, which is exactly the load path relief-valve supports exist to avoid. Some owner-practice arrangements do support the inlet elbow instead, with a hold-down and a stop rigid in the discharge direction; that works because the stop is close to the valve and the outlet has no long lever arm to amplify. Either arrangement is acceptable engineering, but they solve the same problem differently — outlet support removes the lever arm at its source, inlet support removes it by stiffening the near end. What doesn't work is a support that is neither: an outlet with nothing under it and an inlet with no directional stop, which leaves the full moment on the valve casting.

My stanchion lifts off in the operating case. Is that a problem?

Yes, if that stanchion is the thrust support. A PSV mounted on top of a vessel moves with the vessel as it heats up, and if the outlet stanchion is supported from a platform that doesn't move with the vessel, the vessel's growth can lift the stanchion clear of its steel in the hot case. The stress model shows that support inactive — correct for the sustained case, but a warning sign for the relief case: when the valve opens, the thrust has to push the outlet pipe back down through the full thermal gap before the stanchion engages, so the valve and inlet nozzle carry the whole reaction until the gap closes, and then the pipe strikes the support, adding impact to the dynamic reaction. The fix, in order of preference: support the stanchion from steel that moves with the vessel — a vessel-clip-supported platform or bracket — so the relative movement stays near zero; failing that, orient the thrust direction where the vessel's growth is small and keep any gap minimal; or show by nonlinear or time-history analysis that the gap stays closed, or that the response with it open is acceptable.

Do closed relief systems need thrust restraint too?

Yes, mainly for the transient. In a closed system discharging to a flare or relief header, the steady-flow momentum forces are close to balanced once flow is established, but during the brief opening transient a pressure wave travels down the discharge line and loads each pair of elbows in turn with an unbalanced force, roughly the pressure difference across the wave front times the pipe area. PIP PNC00004 §3.2.8 calls for both hammer and slug-flow loading to be considered where it's invoked, and PNC00001 §4.2.8 separately notes that closed-system relief lines carrying two-phase flow should be restrained to control line vibration, distinct from the transient thrust itself. Restrain every long straight run between elbows axially — typically trunnions or dummy legs in the shadow of the pipe with a guide, plus a directional stop on the outlet side in line with the discharge momentum — or show by dynamic analysis that the unrestrained response is acceptable. The header tie-in also moves thermally, so the outlet supports need to allow that movement while still restraining the transient thrust — usually a directional stop with guides elsewhere, not a full anchor.

Does the code tell me which duration factor — 1.15 or 1.2 — applies to a PSV discharge under B31.1?

No, and that's a common misreading. ASME B31.1-2024 ¶104.8.2 gives two duration-based occasional-stress factors: k = 1.15 for occasional loads acting no more than 8 hours at a time and no more than 800 hours a year, and k = 1.2 for loads acting no more than 1 hour at a time and no more than 80 hours a year. The clause states the two buckets and their time limits; it does not say which one a safety-valve discharge belongs in. That classification is the designer's, or the owner specification's, based on the actual duration and annual frequency of the specific relief event — a brief, infrequent lift commonly lands in the 1-hour/80-hour, k = 1.2 bucket, but a valve protecting a process that cycles it repeatedly, or a case with an extended blowdown, may not, and the only way to know is to look at the real event rather than assume the category. State the classification and its basis in the design basis document, because it changes the occasional allowable by a small but real amount, and a reviewer who disagrees with the classification is really disagreeing with the assumed duration and frequency, not with the arithmetic.