Views: 0 Author: J-VALVES Publish Time: 2026-10-27 Origin: Site
Almost every high-pressure valve that leaks in service leaks at the same place: the body-to-bonnet joint. On a Class 1500 gate valve that joint is asked to hold a separating force in the meganewtons while sitting in a thermal gradient that swings several hundred kelvin between a cold start and full load. A bolted bonnet answers that question by adding bolts, and the answer stops working somewhere around Class 1500, not because the bolts are too weak but because the bolt circle becomes physically impossible to build.
The pressure seal bonnet is the industry's answer, and it is a genuinely different mechanism rather than a heavier version of the same one. It inverts the load path: internal pressure no longer tries to open the joint, it closes it. This article quantifies where that inversion pays, explains what the self-energising mechanism actually depends on, and sets out the maintenance rules and failure signatures that decide whether a pressure seal valve lasts twenty years or weeps at its first hydro test. Welded bonnets appear only as a comparison, because they answer a different question.
A bolted bonnet joint is a preloaded tension assembly. Studs clamp the bonnet flange down onto a gasket; the gasket is compressed until it conforms to the flange faces; the stud preload must exceed the force that internal pressure generates over the joint area, plus whatever the gasket needs to stay seated, plus margin for relaxation over the service life. Pressure acts in the one direction that is unhelpful, opening the joint, and the only thing resisting it is stretched steel.
A pressure seal bonnet removes the tension member from the pressure load path. The bonnet sits inside a conical seat machined into the body neck. A seal ring — a graphite or metal ring with a triangular or trapezoidal section — occupies the gap between the bonnet cone and the body cone. A segmented retaining ring, sitting in a groove machined in the body, stops the bonnet from moving axially. When pressure enters the bonnet cavity it pushes the bonnet outward against that retaining ring, and in doing so drives the bonnet cone into the seal ring, which is squeezed radially into the body cone. The higher the pressure, the harder the ring is squeezed.
Three consequences follow, and they explain almost everything about how these valves behave in a plant.
First, the fasteners stop being the pressure-retaining element. The bonnet screws on a pressure seal valve only pull the bonnet up against the retaining ring during assembly and hold it there at zero pressure. They are typically an order of magnitude smaller than the studding a bolted joint of the same class would need, and they do not see the process load. The retaining ring does, and it carries that load in bearing and shear at a diameter well outside the seal, which is a far more efficient place to put it.
Second, bolt relaxation stops mattering. The classic bolted-bonnet failure at temperature is preload loss: the stud relaxes, the gasket recovers elastically, the joint opens a few hundredths of a millimetre, and the leak begins. On a pressure seal valve the small screws can relax all they like; as long as the bonnet cannot move past the retaining ring, the seal stress is set by pressure and geometry, not by fastener tension.
Third, the maintenance burden moves. You have traded a preload problem for a geometry problem. Everything now depends on the taper surface, the ring, and the groove. Those three items have no adjustment: they either fit or they do not, and a mistake in any of them shows up as a leak that cannot be tuned out with a wrench.
Pressure seal maintenance goes wrong when the technician treats the joint as a gasket change. It is a five-part assembly with a defined load path, and each part has its own failure mode. The table below is the vocabulary the rest of this article uses.
| Part | Function in the seal | What destroys it |
|---|---|---|
| Body taper seat (neck bore cone) | Reacts the seal ring radially; the female half of the wedge. Part of the pressure boundary and not replaceable. | Scoring from a bonnet withdrawn at an angle, pitting during wet standby, erosion if the ring chatters at low pressure |
| Bonnet with male cone | Transmits the pressure-generated axial force into the ring as radial squeeze | Bent or galled cone, damage to the stem backseat from lifting the bonnet by the stem |
| Pressure seal ring (the gasket) | The only compliant element in the assembly; crushed to fit the gap on first pressurisation | Reuse, over-compression, extrusion past its temperature limit, oxidation of the graphite binder, wrong height |
| Segment ring (gasket retainer, usually 2 to 4 pieces) | Takes the entire axial pressure load and transfers it into the body groove in bearing and shear | Incomplete engagement in the groove, segments refitted in the wrong order, groove flank damage, corrosion of the groove root |
| Bonnet screws or yoke clamp | Assembly and zero-pressure retention only; not a pressure-retaining member | Being torqued as if they were flange studs, which distorts the bonnet cone and unseats the ring |
You can identify a pressure seal valve in about thirty seconds on a loading dock, and it is worth teaching the receiving inspector to do it. There is no bonnet flange and no ring of large studs. The bonnet enters the body neck as a spigot, and what holds it is a thin ring at the top of the neck plus a set of screws that look far too small for the pressure class stamped on the body. The neck is a stepped cylinder rather than a flange, so the valve is taller and narrower than a bolted equivalent. If a supplier ships a "pressure seal" valve that has a conventional bolted bonnet flange with a spiral wound gasket in it, it is a bolted bonnet valve with a marketing description.
The claim that pressure seal construction is "necessary" above Class 900 is usually stated as a rule of thumb. It is worth doing the arithmetic once, because the result is more dramatic than the rule suggests and it also tells you how far down the class range the argument still holds.
Take a DN 200 valve and assume a bonnet joint with a 300 mm mean gasket diameter, so the pressure acts on 0.0707 m². Multiply by the ASME B16.34 rating at 38 °C for each class to get the separating force the studding must resist. Then divide by an allowable stress of 120 MPa, which is representative of A193 B16 studding at around 540 °C — the temperature at which power plant bonnet joints actually operate, and the reason the arithmetic bites. Divide again by the 817 mm² stress area of an M36 stud.
| Class | Rating at 38 °C (MPa) | Separating force on a 300 mm joint (MN) | Stud area at 120 MPa allowable (mm²) | M36 studs, minimum | Practical outcome |
|---|---|---|---|---|---|
| Class 600 | 10.2 | 0.72 | 6,010 | 8 | Comfortable; bolted bonnet is the right answer |
| Class 900 | 15.3 | 1.08 | 9,010 | 11 | Workable but heavy; the crossover zone |
| Class 1500 | 25.5 | 1.80 | 15,020 | 19 | Flange becomes the dominant feature of the valve |
| Class 2500 | 42.5 | 3.00 | 25,030 | 31 | Bolt circle geometrically saturated |
Those are minimum counts before any allowance for gasket seating, wrench access or the fact that studs come in even numbers. Run the last row properly and the geometry fails. Thirty-two M36 studs at a realistic wrench clearance of 60 mm pitch need a bolt circle of about 610 mm, and once edge distance is added the bonnet flange is roughly 750 mm across on a valve whose pipe is 200 mm. The flange then has to be carried by the body, which means a heavier neck, a heavier transition and a bending moment at the body-neck junction that grows with every thermal cycle. The studding alone weighs close to 90 kg.
Put the same 3.00 MN through a pressure seal joint and the load path changes character completely. The segment ring on a DN 200 Class 2500 valve might sit at 270 mm diameter with an 18 mm bearing height, giving about 15,300 mm² of bearing area and a bearing stress near 200 MPa — high, which is exactly why segment rings are wide, grooves are deep, and groove-root damage is treated as a body-level defect. The bonnet screws, meanwhile, might be ten M20 at 40 kN each, 0.4 MN in total: about one-seventh of the pressure load, and as established, they do not actually carry it.
Two qualifications keep this honest. The crossover is not a fixed class number: it moves down as bore increases, because separating force scales with the square of the joint diameter while bolt circle capacity scales with diameter, so a DN 400 Class 900 valve hits the wall sooner than a DN 50 one. It also moves down as temperature rises, because allowable stress for the studding falls faster than the rating does: the same A193 B16 studding that carries 320 MPa at ambient carries roughly 120 MPa at 540 °C, which nearly triples the bolt area required for the same joint. That triple, not the pressure alone, is why the practical boundary sits at Class 900 for large bore and Class 1500 universally.
Thermal movement is the quiet failure mode of bolted bonnets, and it can be estimated in one line. Take a bonnet neck 200 mm long from the joint face to the yoke, a mean neck temperature 500 K above ambient, and a coefficient of thermal expansion of 12.5 × 10⁻⁶ per kelvin for chrome-moly steel. The neck grows 1.25 mm. A spiral wound graphite-filled gasket has perhaps 0.2 to 0.4 mm of elastic recovery available. The growth exceeds the gasket's recovery range by a factor of three to six, so something has to give: either the gasket is crushed further, or — more commonly, because the studs sit outside the insulation and run cooler than the bonnet — the joint gap opens and preload is lost.
Nothing in ASME B16.34 prevents this. The standard fixes the pressure-temperature rating, the minimum wall and the test requirements; it does not prescribe a bonnet joint design, and it does not rate joints for thermal cycling. Two valves with the same class number and the same material can therefore have wildly different joint reliability, which is why the class number tells a buyer nothing about the part that actually leaks.
A pressure seal joint is indifferent to this movement because the bonnet is free to float axially. The neck can grow 1.25 mm, or 2 mm on a bigger valve; the bonnet slides within the body neck and the retaining ring takes up nothing, because its job is only to stop the bonnet leaving. Each time pressure rises, the ring is re-energised against whatever gap exists at that moment. This is the real reason pressure seal valves dominate main steam and feedwater isolation in cycling plants, and it also explains the one thing they cannot do: with no pressure, there is no energising force, and the joint depends entirely on the assembly preload built into the ring when it was crushed. That is not a design flaw, but it is a maintenance obligation, and the next section is about what happens when it is missed.
Because the energising force is proportional to pressure, the joint has two sealing regimes. Above some threshold, the ring is being squeezed harder than the process can push past it, and the joint is effectively self-sealing. Below that threshold, sealing depends on the interference that was built in at assembly. A correctly assembled valve seals across both regimes. A badly assembled one can seal at rating and leak at atmospheric-plus-a-bit, or the reverse.
This matters because of how valves are tested. A code shell test at 1.5 times the rating proves the joint holds at the top of its range, which is precisely where a pressure seal joint is strongest. It tells you almost nothing about the low-pressure regime, where the ring either seated or did not. A valve can pass its shell test with a ring that was reused, seated high, or slightly cocked, and then weep at the first low-pressure leak check on site — or worse, never be checked at low pressure at all and show the symptom months later during a cold start.
The practical response is to test both ends of the range. Run a low-pressure hydrostatic tightness check at 0.5 to 1.0 MPa before the code shell test, then the shell test, then repeat the low-pressure check after the joint has been pressurised and the bonnet screws re-torqued to the manufacturer's table. Three data points cost an hour and they separate an assembly error from a material problem. The acceptance criteria for the shell test itself are set out in our API 598 pressure testing guide; the point here is that passing it is not the same as proving the joint.
The ring is the only part of the assembly with a genuine material choice, and it is also the part most often bought on price. Three properties matter: compressive strength at temperature, resistance to relaxation (the ring must keep pressing after a thousand hours hot), and chemical compatibility with the process. Graphite is the default because it does all three well up to about 540 °C; it fails in oxidising service, where the binder burns away and the ring simply disappears.
| Ring construction | Typical temperature use | Best suited to | Do not use for |
|---|---|---|---|
| Die-formed flexible graphite with stainless insert | -200 to 540 °C | Steam, hydrocarbon, general high-pressure process; the standard choice | Hot air, oxygen, fuming nitric acid; any oxidising service |
| Flexible graphite with Inconel 718 insert | to about 600 °C | High-pressure steam where relaxation resistance matters | Cost-sensitive utility orders; long-lead spares |
| Soft iron or low-carbon steel ring | to about 400 °C | Dry gas, oil, clean non-corrosive service; cheap and stiff | Wet standby, condensate, marine atmosphere — it rusts into the groove |
| 316 stainless ring, often silver plated | to about 540 °C | Mildly corrosive process; higher seating load available | Chloride service; unplated rings gall the taper |
| Inconel 718 ring | to about 650 °C | Highest creep relaxation resistance, superheated steam | Budget projects; check availability before promising a delivery date |
| PTFE or filled PTFE | Not applicable | Nothing in this application | Pressure seal joints — the material cold-flows and has no elastic reserve, so the joint cannot hold preload |
Treat that table as a screening aid, not a rating. Published limits differ between manufacturers by tens of degrees, and the number that governs your valve is the one on the OEM data sheet for that specific body-and-bonnet combination. A ring that is dimensionally correct but made to a different supplier's specification will not necessarily develop the same seating stress, which is why rings are usually bought serial-matched to the valve.
Pressure seal is one of three answers to the same problem, and choosing between them is easier when you compare what each one does when it fails.
| Bolted bonnet | Pressure seal bonnet | Welded bonnet | |
|---|---|---|---|
| Sealing mechanism | Stud preload compresses a gasket; pressure works against the joint | Pressure-driven cone squeezes the ring; pressure works for the joint | No joint; a weld replaces the seal entirely |
| Practical class range | Class 150 to 900, occasionally 1500 in small sizes | Class 900 to 2500, standard at 1500 and above | Any class where disassembly is not required |
| Mass at Class 1500, same bore | Heavy: large flange, long studs, thick nuts | Typically 30 to 50% lighter | Lightest of the three |
| Can be opened in line | Yes, routinely, with hand tools | Yes, but cold, depressurised, with lifting gear and vertical clearance | No |
| Typical failure | Preload loss after thermal cycling; gasket creep; leak grows with temperature | Ring not seated, reused or extruded; leak worst at low pressure | Weld crack or heat-affected zone corrosion; no warning |
| Repair path | Re-torque, then re-gasket | New ring, taper inspection, controlled reassembly | Cut out and reweld with post-weld heat treatment |
The welded bonnet exists because some services cannot tolerate a joint at all — hydrogen at high pressure and temperature, very toxic media, and anything where a fugitive emission is a reportable event rather than a maintenance ticket. It is not a competitor to the pressure seal design so much as the option you take when you have decided that the joint itself is the risk. The price is that every trim inspection becomes a cutting and welding job with a post-weld heat treatment, which is why welded bonnets are common on small forged valves and rare on large ones.
Before any of the steps below, the safety case has to be settled, because this is the one valve type where the retaining ring is the only thing separating stored energy from the person holding the wrench. The stored energy in a pressurised liquid volume is roughly E = V × ΔP. A DN 200 Class 2500 valve with about 40 litres of cavity at 42.5 MPa holds something like 1.7 MJ. That is the energy required to lift the 60 kg bonnet assembly almost three kilometres. It is released in milliseconds if a segment ring is loosened while the valve is live. Lock-out, verified zero pressure on both sides, cavity drained, and metal temperature below 60 °C are not paperwork items.
Measure the vertical clearance before planning the job. The bonnet has to come straight up and out, and it is longer than a bolted bonnet. Insufficient headroom is the most common reason a site abandons a pressure seal repair after the line is already open.
Remove the actuator or gearbox, then the yoke. On many designs the yoke clamps the bonnet, so the clamp has to come off before anything else can move.
Remove gland nuts, gland flange, gland and packing. Replace the packing while you are here; it is a separate seal and it will fail next if you do not.
Release the bonnet screws. Understand what they do first: they draw the bonnet up against the retaining ring. Removing them does not release pressure; it only removes the assembly preload.
Break the seal downward. Tap the bonnet sharply with a soft-faced hammer so it drops against its stop and uncovers the segment ring. Never pull the bonnet up with the stem or the handwheel: the load passes through the backseat, and an overloaded backseat deforms or cracks.
Remove the segments. Start with the segment that has the small drilled hole; the groove in the ring's top face is there to take a pry bar, and on larger valves the segments can be pushed out through radial holes in the body neck. Keep the set together and identified — segment rings are matched to their groove.
Withdraw the bonnet squarely. Check squareness by measuring the gap between the ring face and the body neck face at four points before you lift, and again as it rises. Slings go on the gland retainer, never on the stem. Any binding means it has cocked: stop, lower it, re-check. A scored taper scraps the body, and a body is worth an order of magnitude more than the ring you were trying to save.
Inspect before you clean. Measure ring height against nominal — more than about 1 mm of crush indicates over-compression and points to a torque or geometry problem rather than wear. Look for radial cracks and extrusion. Check the body taper by dye penetrant or bluing, and check the groove root the same way. The ring is single use, without exception, even if it looks perfect.
Prepare the taper properly. Lint-free cloth, high-temperature anti-seize on the seating surfaces, never general-purpose grease. The finish target is around Ra 1.6 µm or better; light scratches can be polished out with 400 then 600 grit, but pitting or scoring deeper than about 0.5 mm cannot, and polishing must not alter the cone angle.
Reassemble and test in both regimes. New ring, correct orientation with the shoulder facing the process side on most designs — confirm on the OEM drawing rather than by habit. Lower the bonnet concentrically, engage the segments fully in the groove, torque the screws to the published table. Then the low-pressure tightness check, the shell test at 1.5 times rating, re-torque of the bonnet screws after first pressurisation as most OEM manuals require, and a repeat check.
Pressure seal valves produce a small set of distinct symptoms, and the symptom usually names the cause. The single most useful diagnostic is the relationship between leak rate and pressure, which is the opposite of what a bolted bonnet does.
| What you see | Most likely cause | How to confirm | Action |
|---|---|---|---|
| Weep at the joint that gets worse as pressure rises | Ring extruded, cracked, or wrong material for the temperature | Ring height and section after strip-down; check material against the data sheet | New ring of the correct specification; inspect taper |
| Weep at the joint at low pressure that stops at rating | Ring not seated: reused, wrong height, or bonnet cocked on assembly | Low-pressure tightness check reproduced in the workshop | New ring, full reassembly procedure with squareness checks |
| Leak that appears only after a thermal cycle and then stops | Ring relaxation at temperature; marginal seating stress | Log the leak against load and temperature | Upgrade to a higher-relaxation-resistance ring; verify screw torque |
| Leak at the gland, not the joint | Packing, not the pressure seal | Soap solution applied separately to gland and joint; shroud the area | Repack. Do not touch the segment ring |
| Bonnet appears to lift, or the ring walks out of its groove | Segments not fully engaged, wrong segment order, groove flank damaged | Engagement depth and groove profile by measurement | Stop operation; strip down before pressure is reapplied |
| Crack indicated at the groove root | Thermal fatigue at a stress concentration, often after a taper repair | Dye penetrant, then ultrasonic sizing of the indication | Body replacement; this is not repairable in the field |
One warning belongs with that table. Do not attempt to stop a joint leak by tightening the bonnet screws with the valve in service. They are not the load path; tightening them cannot close a leaking taper, and the act of tightening can unseat a segment that was marginally engaged. In plants where a fitter has done this, the usual outcome is not a stopped leak but a bonnet that moves.
Because the design is described as the high-pressure solution, it gets specified where the mechanism cannot work. Five cases to reject it outright:
Vacuum service. The energising force comes from internal pressure. Below atmospheric there is nothing to push the bonnet into the cone, and the joint depends solely on assembly preload, which is exactly the condition the design is not built for. Use a bellows seal valve or a bolted bonnet.
Class 600 and below. The bolt-circle argument does not apply, so you pay for a taper, a matched ring and a difficult strip-down in exchange for nothing. A bolted bonnet is cheaper, lighter to maintain and universally understood.
Valves opened every turnaround. If the plan is to lap seats and change trim at every shutdown, a joint that requires cold conditions, lifting gear, vertical clearance and a single-use matched ring is the wrong choice regardless of class. A bolted bonnet gate valve is the honest answer for that maintenance philosophy.
Small bore. Below about DN 40 the bolt loads are trivial, and a welded bonnet forged valve per API 602 does the same job with fewer parts. The pressure seal design's advantage scales with joint diameter.
Cryogenic service. The problem there is not joint preload but heat leak down the bonnet, and the answer is an extended bonnet with the correct vapour column height, not a self-energising joint. Combining the two adds a taper that iced-up maintenance cannot inspect properly.
There is also a project-management case: do not specify pressure seal bonnets on a site whose maintenance team has never opened one, unless the order includes the assembly drawing, the torque table and a handover session. The failure mode of an inexperienced reassembly is not a graceful degradation. It is a joint that passes a shell test and weeps at the first cold start.
Most disputes about these valves come from datasheets that name a class and a size and leave the joint undefined. The clauses below close the gaps that matter, in rough order of how often they are missing.
| Clause | Say this, not that | Why it matters |
|---|---|---|
| Governing document | Name ASME B16.34 for rating and wall, plus the project piping specification for the joint. If the datasheet says API 600 on a pressure seal valve, ask which document actually governs the joint. | API 600's scope is written around a bolted bonnet joint. A class number alone does not define a self-energising seal. |
| Taper finish and protection | Ra 1.6 µm or better, dye penetrant on the taper, protective cap fitted for shipment | The taper is the seal face and is not replaceable. Transit damage is common and invisible under paint. |
| Ring specification | Material, construction, batch certificate, positive material identification, serial-matched to the valve, one spare ring shipped with each valve | An unmatched ring may not develop the designed seating stress; a missing spare turns a two-hour job into a six-week one. |
| Segment ring | Material and hardness, segment identification, groove-root dye penetrant, engagement depth stated | Segments are matched to their groove; mixed sets are a known cause of incomplete engagement. |
| Fasteners | Stud grade and coating, published torque in N·m, and the instruction to re-torque after first pressurisation | "Tighten securely" is not an instruction. Most OEM manuals specify a re-torque step, and most sites skip it. |
| Testing | Low-pressure tightness at 0.5 to 1.0 MPa, then shell test at 1.5 × rating, both with recorded medium, temperature and duration | Two regimes, two tests. The shell test alone cannot detect an unseated ring. |
| Documentation | Assembly drawing showing ring orientation and stack-up height, spare parts list, OEM lead time for rings stated in the offer | Orientation errors are the most common reassembly mistake, and ring lead time drives maintenance planning. |
The honest summary is that a pressure seal bonnet is not a better bonnet in the abstract; it is a different load path, and it is the correct one above Class 900 because bolt circles stop being buildable and stud preload stops being stable. What you buy is a joint that gets tighter as pressure rises and does not care about fastener relaxation. What you pay is a joint you can only open cold, a ring you can only use once, and a taper you can only damage once. Buyers who understand that trade get decades of service. Buyers who treat it as a gasket change get a leak at the first cold start and a bill for a body.
For the wider context of how these valves fit into a high-temperature isolation package, see the globe valve range, the gate valve packing and gland leakage guide for the adjacent stem seal, our forged versus cast valve body comparison for the body route that decides whether a taper can be machined at all, and the special material valve rangewhere chrome-moly and stainless bodies for these classes are listed.
