A GRP pipe does not have a burst pressure. It has a burst pressure at a stated time. The value printed on a datasheet as long-term hydrostatic strength is the end point of a statistical extrapolation from laboratory failures that last a few thousand hours, carried forward by more than a factor of forty in time to reach a fifty-year design life. Whether that extrapolation is worth anything depends entirely on how the regression was built: how many specimens failed, how they were loaded, whether the failures were true through-wall bursts or end-fitting leaks, and whether the slope of the time-to-failure line was steep or flat. This guide covers what ASTM D2992 and ISO 10928 actually require, why eighteen failure points is a floor rather than a target, how a tested strength becomes a rounded hydrostatic design basis and then a design stress, what a realistic set of numbers produces on a DN800 line, and how to tell whether a supplier's report describes a qualification or a hopeful curve fit.
1. Why a Pipe's Rating Is an Extrapolation, Not a Test Result
Metals are rated by a yield stress that does not change with time in service. A steel pipe that holds 4 MPa at commissioning holds 4 MPa twenty years later, provided it does not corrode. Glass-reinforced thermosetting resin does not behave that way. Under sustained hoop stress the resin matrix creeps, the load transfers progressively to the glass, and the laminate fails at a stress well below its short-term strength. The drop is not small. In the example used throughout this guide, a filament-wound vinyl ester laminate carries 95 MPa at 10,000 hours and reaches its fifty-year end point near 75 MPa: a loss of a fifth of its capacity, and a small one by the standards of the family.
The practical consequence is that every pressure class you have ever been quoted for a GRP pipe rests on an extrapolation from a few months of testing. Nobody has a fifty-year test. What the industry has instead is a set of procedures for making that extrapolation defensible, and a set of criteria for deciding when a data set is too thin to extrapolate from at all.
A steel pipe line is designed against a number the material will still have in fifty years. A GRP pipe line is designed against a number the material has never actually been observed to reach.
2. What ASTM D2992 Requires
ASTM D2992 is the standard practice that sets out how the hydrostatic or pressure design basis of a reinforced thermosetting resin pipe is obtained. The current edition supersedes earlier issues and keeps the same two procedures, which differ in the loading mode and in how much confidence the result carries.
| Item | Procedure A | Procedure B |
|---|---|---|
| Loading | Cyclic pressure: hold, release, hold again | Continuous static pressure to failure |
| Typical cycle rate | About 25 cycles per minute | Not applicable |
| Number of failure points | 18 minimum | 18 minimum |
| Time range to be covered | Four decades, roughly 103 to 107 cycles | Three decades, roughly 10 to 104 hours |
| Longest test required | At least one specimen failing beyond 10,000,000 cycles | At least one specimen failing beyond 10,000 hours |
| What it produces | Cyclic design basis; the more conservative of the two | Static design basis; the usual basis for water and process lines |
| Where it is mandated | Cyclic service, some oilfield specifications | ASTM pipe specifications, API 15HR, ISO 14692 qualification |
Two constraints in Procedure B are worth reading literally, because they are the ones suppliers most often soften. The first is that the wall must be thin relative to the diameter, so that the failure is a genuine laminate burst rather than a local bearing or end-effect artefact; a specimen with an outside diameter to wall thickness ratio below ten to one does not qualify. The second is that the internal pressure applied is held to roughly a fifth of the expected short-term hoop strength, so that the specimens fail by creep rupture over months rather than by overload in hours. A data set that has been built at higher pressures and shorter times will produce a flat line and an apparently excellent extrapolation, and it will be wrong.
3. Reading the Regression
The usual treatment is a log-log fit of failure stress against time to failure. Written out, the model is a power law:
- sigma(t) = sigma(t0) x (t / t0)-B, where B is the magnitude of the slope of the fitted line in log-log coordinates.
- The slope parameter quoted in resin data sheets is m = 1 / B. A large m means a flat line and slow strength loss; a small m means a steep line and a pipe that will not survive the extrapolation.
- For the laminate used in the worked example, B = 0.0625 and therefore m = 16.0 — a favourable slope, characteristic of a well-cured vinyl ester laminate with a high glass content.
The line is only usable if it is statistically real. The regression coefficient of a log-log fit through n failure points must exceed the critical value for the number of points available, and that critical value falls as the data set grows. The numbers below are the two-sided 99 per cent critical values, which is the level at which the correlation is conventionally accepted as not accidental.
| Failure points, n | Degrees of freedom | Critical regression coefficient |
|---|---|---|
| 18 (the minimum) | 16 | 0.5897 |
| 24 | 22 | 0.5151 |
| 32 | 30 | 0.4487 |
| 48 | 46 | 0.3683 |
The message in that table is easy to miss. With only eighteen specimens the fitted line must be very tight before it is accepted, because eighteen points can be made to look correlated by chance. A supplier who runs eighteen tests and reports an extrapolation is not cheating, but the effort has gone into satisfying the letter of the procedure rather than into buying confidence. A data set of thirty-two or more points is a much stronger claim, and it is the difference between a qualification and a formality.
4. From HDS to HDB
The extrapolated fifty-year stress is the hydrostatic design stress, or HDS. It is then rounded down to the next preferred number in the ISO R10 series to give the hydrostatic design basis, HDB. The rounding is deliberately downhill: the series exists so that purchasers and designers work in the same numbers, and stepping up to the next preferred value so that the datasheet looks better is not permitted.
| Quantity | Value for the worked example | Comment |
|---|---|---|
| Extrapolated 50-year stress, HDS | 75.0 MPa (10,878 psi) | Read off the fitted line at 438,000 hours |
| Next ISO R10 value above | 80.0 MPa | Not available; would sit above the tested strength |
| HDB, rounded down | 63.0 MPa (9,138 psi) | The published design basis |
| Rounding loss | 16.0 per cent | Paid for the convenience of a standard number |
| Design stress at a factor of 0.5 | 31.5 MPa | Used for all pressure calculations below |
The design factor applied to HDB is where two apparently identical pipes diverge in price. A factor of 0.5 is the usual choice for water and general process service where the consequence of failure is a leak and a repair. Lower factors are used where the fluid is hot, where surge pressures are not analysable, or where the pipe carries a hazardous inventory. A designer who accepts a datasheet's HDB without asking which factor the supplier expects it to be divided by has not finished specifying the pipe.
5. Where the Fifty-Year Number Comes From
The extrapolation target is 438,000 hours, which is fifty years of 24-hour days including leap years. ISO 10928 sets out two routes to it. Method A works on the mean regression and applies a statistical lower prediction limit, so it needs the scatter of the data to be quantified properly; it is the method to use when the scatter is wide. Method B works on a time-variable regression and is the simpler of the two. Either way the output is a characteristic value at 438,000 hours, not a single test result.
The intermediate values are worth seeing, because they show how much of the loss happens early:
- 1 hour: 101.3 MPa
- 10,000 hours (about 14 months): 95.0 MPa
- 100,000 hours (about 11 years): 82.3 MPa
- 1,000,000 hours (about 114 years): 71.2 MPa
Nearly half of the total fifty-year loss has already occurred by the time the first year in service is over. This is why a pipe that has survived its first year of a pressure test tells you almost nothing about its remaining life, and why a rising trend in a long-term regression is a bad sign rather than a good one.
6. The Worked Example: DN800
Take a DN800 line, 800 mm nominal bore, carrying water at a nominal pressure of 1.0 MPa, with a design stress of 31.5 MPa from the table above. Two separate checks decide the wall thickness, and they are frequently confused with each other.
Check one: ring stiffness. The ring stiffness class SN is defined through the flexural stiffness of the pipe wall per unit length, I = t3 / 12, divided by the cube of the diameter. Ring stiffness has nothing to do with internal pressure. It governs the pipe's resistance to deflection under soil and traffic load, and it is expressed as SN 2500, SN 5000 or SN 10000 in newtons per square metre. Rearranging for wall thickness with a hoop-direction modulus of 20 GPa gives the middle column of numbers below.
Check two: internal pressure. Hoop stress in a thin wall is sigma = P (D - t) / (2 t), so the wall thickness needed for a given pressure is t = P D / (2 sigma + P). This check is about burst and long-term rupture, not about deflection.
| Nominal bore | SN 2500 wall | SN 5000 wall | SN 10000 wall |
|---|---|---|---|
| DN400 | 4.58 mm | 5.77 mm | 7.27 mm |
| DN600 | 6.87 mm | 8.65 mm | 10.90 mm |
| DN800 | 9.16 mm | 11.54 mm | 14.54 mm |
| DN1000 | 11.45 mm | 14.42 mm | 18.17 mm |
| DN1200 | 13.74 mm | 17.31 mm | 21.81 mm |
Setting the two checks side by side for DN800 produces the useful result. At low pressure the stiffness requirement is the larger of the two and the wall is a stiffness decision. Above a crossover the pressure requirement takes over.
| Nominal pressure | Wall for pressure | Wall for SN 5000 | Which governs |
|---|---|---|---|
| 0.4 MPa | 5.05 mm | 11.54 mm | Stiffness |
| 0.6 MPa | 7.55 mm | 11.54 mm | Stiffness |
| 1.0 MPa | 12.50 mm | 11.54 mm | Pressure |
| 1.6 MPa | 19.81 mm | 11.54 mm | Pressure |
The crossover for an SN 5000 DN800 pipe sits at about 0.92 MPa. Expressed the other way round, the pressure capability of each stiffness class is 0.73 MPa for SN 2500, 0.92 MPa for SN 5000 and 1.17 MPa for SN 10000 — all at a design stress of 31.5 MPa. A specification that quotes DN800 SN 5000 for a 1.0 MPa line is therefore under-strength on the pressure check even though the stiffness is generous, and the error is invisible unless both checks are done. The hoop stress at that combination is 34.2 MPa against a design stress of 31.5 MPa, an overstress of about nine per cent, and only a factor of 2.2 against the fifty-year strength itself.
7. Reading a Supplier's Long-Term Report
A long-term hydrostatic strength report is a short document, and everything that matters is in four places. Ask for it as a document, not as a number.
- The failure table. Every specimen, its applied pressure, the time at which it failed, and the failure mode. Specimens that failed at an end fitting, a gasket or a saddle should be excluded and the exclusion explained. A table that contains only times and stresses, with no failure modes, cannot be checked.
- The regression plot. Time to failure on a logarithmic axis against hoop stress on a logarithmic axis, with the fitted line, the confidence bounds and the number of points marked. If the plot is not in the report, the number is not traceable.
- The slope and the correlation. B or m, and the regression coefficient against the critical value for the number of points. A fitted line with a correlation just above the minimum is a fragile extrapolation even though it passes.
- The failure mode at long times. Whether the longest specimens failed as genuine wall bursts or as weeps along a joint. The long-time end of the line is the only part that matters for a fifty-year life, and it is the part most easily masked.
Two further questions separate a qualification from a hope. Was the resin and cure system in the test coupons the same as the one that will be used in production, and was the liner construction identical? A change of catalyst, a switch from a cobalt-catalysed to a BPO/DMA-cured system, or a different veil can move the long-term end point materially while leaving the short-term properties untouched. If the answer to either question is no, the report describes a different pipe.
8. What Changes With the Service
Water at 20 degrees is the easiest case. Everything else moves the number, and usually down.
| Change in service | Effect on the long-term basis | What to do about it |
|---|---|---|
| Higher temperature | Strength and stiffness both fall; the regression shifts to a lower line | Require a qualification at the service temperature, not an ambient-derived number with a correction factor |
| Chemical attack of the resin | The liner, not the structure, sets the life | Resin selection data for the specific medium, concentration and temperature |
| Cyclic pressure | A static basis does not cover fatigue | A Procedure A cyclic basis, or a surge analysis |
| Buried installation | Ring deflection becomes a parallel design case | The bedding, backfill and deflection check alongside the pressure check |
| Potable or food contact | Approval requirements sit alongside the strength basis | Certification to the relevant national scheme |
Buried service deserves its own mention, because a pipe that passes every pressure check can still fail a deflection check. The two interact through the wall thickness: a thicker wall bought for pressure also raises the ring stiffness, and a thinner wall bought for stiffness may fail on hoop stress. Running both checks against the same wall thickness, and recording which one governs, is the only way to know what you have specified.
9. Putting the Long-Term Basis Into a Purchase Order
The long-term strength is the single most important piece of documentation attached to a GRP pipe order, and it is also the one most often omitted because it is not needed at the moment of purchase. By the time it is needed, the pipe is in the ground.
- State the design stress and the design factor, not only the pressure class. "PN 10 to a design stress of 31.5 MPa derived from an HDB of 63 MPa at a factor of 0.5" is a specification. "PN 10" is a label.
- State the stiffness class separately, and state which of the two checks governs at the specified diameter.
- Require the long-term report as a deliverable, with the failure table and the regression plot, and require the specimen count to be declared.
- Require the resin and cure system of the tested laminate to be identified, and treat any production change as invalidating the qualification unless re-tested.
- Require the test to have been run to the applicable procedure at the service temperature where the service is above ambient.
- Require the pipe to be marked with its stiffness class and pressure class, and check the marking against the delivery paperwork on arrival.
10. FAQ
Can a short-term burst test substitute for a long-term basis? No. A burst test measures strength at seconds to minutes, where the material is at or near its full short-term capacity. It confirms that the laminate was made as specified and that the joints do not leak. It says nothing about the fifty-year end point and it is not a substitute for a regression.
Is a higher stiffness class the same as a higher pressure class? No, and the worked example shows why. Ring stiffness is fixed by the wall's flexural stiffness and the diameter, and it has no pressure term in it. A SN 10000 pipe is stiffer underground and also somewhat stronger in hoop, but the two are consequences of the wall thickness rather than of each other, and one can govern while the other does not.
Why is the number lower after rounding? Because the ISO R10 series is a small set of preferred values and the tested result almost never lands exactly on one. Rounding down keeps the published basis inside the tested strength; rounding up would publish a number the material has not demonstrated.
How often does the qualification have to be repeated? Whenever the laminate changes materially — resin type or grade, cure system, glass content, winding angle, liner construction or the supplier's equipment. A qualification is a statement about a specific laminate, not about a company.
11. Specifying Long-Term Strength With Confidence
At Henan HuanGang we produce filament-wound FRP and GRP pipes for process and water service and pressure-rated FRP pipe across the bore range where the stiffness and pressure checks cross each other, which is exactly the range where the wall thickness stops being obvious. Our quotations are built around the two checks shown in this guide: the ring stiffness class required by the burial condition and the wall thickness required by the design stress, reported side by side so that you can see which one governs before the order is placed. If you have an existing specification, a datasheet with an HDB on it, or a service condition where the temperature or the medium has moved away from the water case, send it across and we will tell you what the numbers imply.
Related reading: GRP pipe specification and selection, FRP pipe diameter and wall thickness, GRP pipe bedding, backfill and ring deflection, FRP pipe installation and pressure testing, filament winding angle design, FRP pipe for fire water and deluge systems, FRP pipe for desalination and brine service, FRP pipe support spacing and hanger design and resin selection for FRP pipe.
Need help selecting FRP products?
Our engineers answer technical questions free of charge.
Ask an Engineer on WhatsApp