Chapter 9 of 10 · 10 min read
Inspection and pressure testing
Choose visual, radiographic and ultrasonic examination for pipe welds, sample so a struggling welder is found fast, measure the repair rate, and compare the energy stored in water and gas tests.
A pipe weld is accepted twice. First, its welds are examined: every weld looked at, and some or all of them examined through the wall by radiography or ultrasonic testing. Then the finished system is pressure tested, filled and pressurised above its design pressure, to prove that it holds without leaking or failing. The examination finds defects in the welds; the test proves the whole system, welds, pipe, fittings, flanges and all. Neither replaces the other, and a line that passed its test can still contain welds that will fail later by fatigue or corrosion, if the examination missed them.
This chapter compares the examination methods, explains how sampling works and why it should find a struggling welder quickly, shows how the contractor measures its repair rate, and compares hydrostatic and pneumatic pressure tests by the energy stored in them, which is why a gas test is treated so differently from a water test. Testing Welded Joints by the Numbers covers the principles of testing in more depth.
The methods
Pipe welds are examined by several methods, each finding different defects.
| What it finds well | What it misses or struggles with | Notes | |
|---|---|---|---|
| Visual | surface defects, profile, undercut, cracks, arc strikes, misalignment | anything below the surface; inside root unless accessible | every weld; first and cheapest |
| Radiography | porosity, slag, lack of penetration, burn-through, root profile | planar defects at unfavourable angles; fine cracks | radiation safety; film or digital record |
| Ultrasonic | lack of fusion, cracks, planar defects through the wall | small porosity; thin walls; coarse-grained stainless | no radiation; skilled operator; can be recorded |
| Surface crack testing | surface cracks by magnetic particle or dye penetrant | anything below the surface | caps, roots accessible, attachments |
| Borescope | inside root profile, colour, penetration on small bore | defects within the wall | hygienic lines; recorded video |
Visual inspection is applied to every weld and is the most useful single method, because most defects that affect a pipe weld’s life show at the surface or at the profile. Radiography, the traditional method for pipe, makes an image of the weld on film or a digital detector and gives a permanent record; it finds volume defects such as porosity and slag well, and lack of penetration very well, but can miss planar defects such as lack of side fusion if they lie at an angle to the beam. Ultrasonic testing, increasingly used in place of radiography, especially in phased-array forms that scan and record the whole weld, finds planar defects well and needs no radiation exclusion zone, but requires skilled operators and is harder on thin and coarse-grained material.
How much to examine
Codes set the extent of examination by the line’s class: its fluid, pressure, temperature and the consequences of failure. A line carrying a hazardous fluid at high pressure may have every weld radiographed; an ordinary utility line a small percentage, chosen at random; some low-risk lines none beyond visual inspection and the pressure test. The percentage is set by the code and the contract, not by the contractor’s preference. Where a percentage is examined, the sample must represent the work fairly: each welder’s welds, each size and each period of the job, not just the easiest welds or the ones near the scaffold.
Sampling that follows the welder
Random sampling by itself can let a struggling welder make many defective welds before one is picked. Most codes therefore add progressive sampling: when a sampled weld is rejected, further welds by the same welder are examined, and if those fail, more again, until either the welder’s work is shown to be acceptable or all of it has been examined. That rule concentrates the examination where the problems are, and it gives the contractor a strong reason to sample a new welder’s first welds early.
- Welds made
- 220
- Random sample
- 22 radiographed
- Rejected in the sample
- 2, both by one welder
- random sample repair rate = 2 ÷ 22 ≈ 9%
- progressive: 4 more of that welder's welds radiographed; 1 rejected
- across the 26 examined, 3 rejected ≈ 12%; across other welders' welds, 0 of 18
The arithmetic shows why a single overall rate can mislead. The overall rate was high in the first month, but all the rejections came from one welder, whose fit-up practice turned out to be the cause. With help on the fit-up, and progressive examination of their work for the next fortnight, their rate fell to match the others. A contractor that looks only at the overall figure misses the cause; one that looks at the rate by welder, by procedure and by defect type finds it quickly.
When to examine
Timing matters. A weld examined as soon as it is finished may not yet show a hydrogen crack that will form over the following hours, so for hardenable and high-strength steels procedures and codes may require a delay between welding and the final examination. Examination after post-weld heat treatment, not before, is required in many cases, because the treatment itself can reveal or cause defects. At the other extreme, waiting until a whole system is finished before examining any welds means that a struggling welder’s defects are found only after dozens more joints have been made the same way. The contractor examines a sample of each welder’s work within the first few days, then keeps examination running steadily behind the welding, so that results come back while the cause is still fresh.
Acceptance criteria
An indication found by an examination is not automatically a defect. The code sets acceptance criteria for each method and inspection class: which kinds of imperfection are allowed, how large, how many in a given length, and which are never acceptable, such as cracks. Porosity within the allowed size and spacing is accepted; lack of fusion of any size usually is not. The examiner reports what was found and judges it against the criteria; the inspector reviews the judgement. Arguments about whether an indication is acceptable are settled by the code’s words and the examination record, never by how good the weld looks from outside.
Interpreting results
Radiographs and ultrasonic records need skilled interpretation. Surface irregularities in the cap and root show on a radiograph and can mask or mimic defects within the weld; geometric features of the joint, such as the root bead and the counterbore, produce ultrasonic signals that are not defects. The examiner compares the image or signal with the joint’s geometry, and where there is doubt, asks for the cap to be dressed, a second shot at a different angle, or a second method. A contractor that grinds caps smooth and keeps roots even makes the examiner’s job easier and gets clearer answers.
Records of examination
Every examination produces a record: the weld’s number, the method and technique, the examiner and their certification, the date, the result, and for radiography and recorded ultrasonic testing the image or data itself. Those records are linked to the weld log, so that each weld’s history, from fit-up to acceptance, can be traced. Digital radiography and phased-array ultrasonic records are easier to store, search and review than film, and are increasingly what clients ask for.
The cost of a repair
A repair is far more costly than the original weld. The defect must be located, the weld ground out to sound metal or cut out completely, the joint re-prepared, refitted and rewelded, with preheat and purge if they apply, and then examined again, often by the same method and sometimes with additional examination of nearby welds. On a hygienic line or an alloy line with heat treatment, a cut-out can cost several times the original joint. The contractor’s records put a typical carbon steel repair at about three times the cost of the original weld and a stainless cut-out at about four, which is why a few percentage points of repair rate make a large difference to the job’s margin.
Pressure testing
When a system is complete, it is pressure tested to prove its strength and tightness. The code sets the test pressure, usually a stated multiple of the design pressure, the test medium, the holding time and the examination during the test. In a hydrostatic test, the system is filled with water, vented of air, and pressurised with a pump; in a pneumatic test, it is pressurised with air or nitrogen. Water is strongly preferred, and codes usually permit pneumatic testing only where water cannot be used, for example where the line cannot carry the weight of water, or water would contaminate the process.
Why water is preferred
The reason is the energy stored in the test. Water is nearly incompressible, so very little energy is stored in pressurising it; if the system fails, the pressure drops almost instantly and the water spills out. A gas is highly compressible, so a great deal of energy is stored, and a failure releases it explosively, throwing fragments and pipe.
- Internal volume
- π ÷ 4 × 0.1023² × 120 ≈ 0.99 m³
- Test pressure
- 2.4 MPa gauge
- water: 2.4² × 10¹² × 0.99 ÷ (2 × 2.2 × 10⁹) ≈ 1,300 J, about the energy of a 20 kg bag of cement dropped from about 6 m
- air: about 3.7 MJ, nearly three thousand times as much
- 3.7 MJ is roughly the energy of 0.8 kg of TNT, released if the line fails
The comparison explains why a pneumatic test is treated as a hazardous operation in its own right: the code usually limits its pressure, requires an exclusion zone, staged pressurisation with holds, and sometimes additional examination of the welds before the test. A hydrostatic test still needs care, with venting of all air, because trapped air pockets store energy like a pneumatic test, and protection against the water’s weight and any failure; but its stored energy is small.
Running a hydrostatic test
A hydrostatic test begins with isolating the section to be tested, removing or blanking equipment that cannot take the test pressure, such as instruments, relief valves and some pumps, and fitting a test pump, a calibrated pressure gauge or recorder at the lowest and highest points, and vents at every high point. The section is filled from its low point, vented until water flows clear of air from every vent, then pressurised in stages to the test pressure and held for the time the code requires, while the welds and joints are inspected for leaks. The water must be clean and, for stainless lines, low enough in chloride that it does not cause pitting while standing; the line is drained and dried promptly after the test. The gauge or recorder’s chart, the pressure, the time, the section tested and the witnesses go into the test record.
Leak testing hygienic and low-pressure lines
Lines that operate at low pressure, or hygienic lines that must stay dry, are sometimes tested with a lower-pressure gas leak test, with leaks found by soap solution or more sensitive methods, under the code’s provisions for such tests. The low pressure keeps the stored energy modest, but the same precautions against over-pressure apply, with a relief valve set just above the test pressure on the test supply.
Who examines
Examination is done by qualified examiners: radiographers and ultrasonic technicians certified for the method, and welding inspectors for visual inspection, under the code’s requirements. On many projects, the client or an independent inspection body witnesses tests and reviews records. The contractor’s own welders and supervisors also inspect their work visually as they go, because finding a defect before the next pass is laid costs a fraction of finding it on a radiograph.
The running case: the test packs
This is an illustration. The contractor divided the dairy’s piping into test packs, each a section of a system that could be isolated and tested in one go, with its drawings, line list, weld log extract, examination reports and a marked-up diagram of vents, drains, blanks and gauges. A test pack could be tested only when every weld in it had been examined as required and accepted. Steam, condensate, compressed air and cooling water packs were hydrostatically tested. One small instrument air section, which could not be dried reliably after a water test, was pneumatically tested at the reduced pressure the code allowed, with an exclusion zone and staged pressurisation planned by the supervisor and approved by the client’s engineer.
A failure: the trapped air
This is an illustration. On an earlier job, a hydrostatic test was started on a section with a high point that had no vent, so a large pocket of air was trapped under the pressure. When a gasket at a temporary blank failed during the test, the expanding air drove the blank and a jet of water across the area with far more violence than a water test should produce.
Nobody was hurt, because the area had been cleared, but the incident showed that a hydrostatic test with trapped air is partly a pneumatic one. The contractor’s test pack diagrams now show a vent at every high point, and the test does not start until each vent has run clear.
Exercise: the repair cost and the test energy
The stainless product lines have 300 welds. If 2% need cut-outs, at about A$450 per original weld and four times that per cut-out, what do the repairs cost? Then estimate the energy stored if a 20 m length of 50.8 mm tube with a 1.6 mm wall were tested with water at 1.5 MPa gauge.
At 2%, about 6 welds need cut-outs, each about 4 × 450 = A$1,800, about A$10,800 in repairs, excluding GST.
The bore is 50.8 − 3.2 = 47.6 mm, so the volume is about π ÷ 4 × 0.0476² × 20 ≈ 0.036 m³. The stored energy in water is about (1.5 × 10⁶)² × 0.036 ÷ (2 × 2.2 × 10⁹) ≈ 18 J, trivial, which is why a properly vented water test of a small line is so safe. The same line tested with gas at the same pressure would store tens of kilojoules, enough to cause injury if it failed.
