Why a Pressure Test and a Smoke Test Find Different Faults

Why this matters

Two tests get run on the same length of drainage pipe and they routinely disagree, which reads as one of them being unreliable. Neither is. They load the pipe at pressures a factor of several hundred apart, and those two pressures sit on opposite sides of a trap seal's rating, so they are not weak and strong versions of one test, they are two different tests of two different boundaries. The fault sets overlap, since a cracked cap shows up on both, but the outputs never do, and that is the whole distinction. One of them tells you how much the section leaks and cannot tell you where. The other tells you exactly where and cannot tell you how much. A shop that knows which question it is asking picks the right test the first time, and reads the disagreement between them as the finding rather than as an error.

What each test actually loads

A hydrostatic test fills the section with water and the pressure at any point is the height of water above it: 0.433 psi per foot. A 10-foot head puts 4.33 psi on the lowest joint and, by definition, zero on the joint at the water surface. That gradient is the defining property of the test. The model codes commonly handle it by requiring every joint to see a stated minimum head except the topmost portion of the system, and that requirement binds only as your jurisdiction adopted and amended the code.

A low-pressure air test plugs the section and pressurizes it with air to a few psi, then times a stated pressure drop. Pressure is essentially uniform through the section because air weighs almost nothing, so unlike the hydrostatic test it loads the top joint and the bottom joint alike.

A smoke test puts a blower on the system and pushes tracer smoke through it at a pressure on the order of a quarter inch of water column, which in psi is 0.25 x 0.0361 = 0.009 psi, a factor of about 500 below an air test at 4.58 psi. It is not trying to stress anything. It is trying to find every path from the inside of the pipe to the outside air, and it marks each one by coming out of it.

The trap seal is the dividing line, and the arithmetic says so

A trap seal is a column of water, commonly two inches deep, and what it can hold is exactly its own height: 2 inches of water column, which is 2 x 0.0361 = 0.072 psi.

• The smoke test applies 0.25 in w.c. against a seal rated at 2 in w.c. That is one eighth of what the seal holds. Every intact trap in the building stays intact.
• The air test at 4.58 psi is 4.58 / 0.0361 = 127 in w.c., which is 63 times the seal. Every trap connected to that section is blown out in the first second and the sewer gas behind it goes into the room.

That single comparison decides which test can be run on an occupied, connected system and which cannot. A pressure test is run on isolated, plugged, unconnected pipe, before fixtures are set or with the section positively isolated. A smoke test is run on a live system in an occupied neighbourhood precisely because it works below the seal's rating.

It also has a consequence that gets missed in both directions. A trapped connection is invisible to a smoke test and is a leak on an air test. A trapped yard drain tied into the lateral holds its seal at 0.25 in w.c. and passes no smoke, so the smoke test does not know it exists; the same connection at 4.58 psi has no seal at all and leaks air like an open pipe. If a section fails an air test and the smoke test finds nothing at the failing rate, look for what is trapped before assuming the air test was wrong.

Rate against position

The last row is the practical summary. The sibling card on inflow and infiltration separates those two defect families by their time signature against rainfall; this is the testing consequence of that split, and the two cards agree on direction: discrete inflow connections are smoke-test findings, distributed infiltration is a pressure-test finding.

The hazard each test creates

The air test is the dangerous one and the danger is the plugs. Water at test pressure changes volume by a few thousandths of a percent, so when a joint or a plug lets go the pressure collapses at once and what you get is a splash. Air at 4.58 psig is at 19.3 psia absolute, and expanding back to 14.7 psia is a volume ratio of 19.3 / 14.7 = 1.31, so the gas expands by about 31 percent while it is still pushing, and it pushes a dislodged plug the whole way. That is the mechanism behind every mechanical-plug fatality in this work.

So: brace or restrain every plug mechanically rather than trusting friction, keep every person out of the trench and away from both ends while the section is under pressure, never exceed the test pressure the specification states, and bleed the section down and confirm zero at a gauge before any plug is loosened. Relieving stored energy before disassembly is the core of 29 CFR 1910.147, and a pressurized test section is stored energy in exactly that sense. Many pipe and fitting manufacturers prohibit compressed-gas testing of plastic products outright, so read the manufacturer's documentation for the material in front of you before you choose air over water.

Where the test happens in an excavation, that excavation is construction work under 29 CFR 1926 Subpart P, with a protective system required at and beyond 5 feet under 1926.652 and a means of egress in trenches at and beyond 4 feet under 1926.651. Where a manhole is involved, it is a permit-required confined space: 29 CFR 1910 Part 1910.146 in general industry, 29 CFR 1926 Subpart AA on construction work. The smoke test's own hazard, smoke entering an occupied space through a barrier that was already missing, has its controls in the sibling HowTo; do not run one without them.

Worked example: a 90-foot section, tested both ways

A 90 ft lateral section, plugged at both ends for a low-pressure air test, groundwater standing 2.5 ft above the pipe.

The correction that has to be applied before the test pressure means anything. Groundwater outside the pipe is pushing back at 0.433 psi per foot, so the test pressure must be raised by that amount or the wall never sees the differential the specification intends.

• Raw target test pressure: 3.50 psig
• Groundwater correction: 2.5 ft x 0.433 psi/ft = 1.08 psi
• Corrected target: 3.50 + 1.08 = 4.58 psig, and the timed drop starts from that corrected figure.

Skip that line and the section is tested at 3.50 psig against 1.08 psi of external head, which is a real differential of 2.42 psi, about half the intended value. The test then passes sections it should have failed, which is the flattering direction and the reason the correction is written into the specification in the first place.

The result. The section fell 1.0 psi, from 4.58 to 3.58 psig, in 2 minutes 10 seconds. The project specification's table for this diameter and length required the section to hold that 1.0 psi drop for 3 minutes 10 seconds. It held 130 seconds against a required 190, which is 68 percent of the required time, failing by 60 seconds, about a third short.

What that gives you: a rate, and a fail. What it does not give you: any idea which of ninety feet is responsible, or whether it is one defect or twenty.

The smoke test on the same section. Blower at about a quarter inch of water column, and three observations:

1. Smoke from a cleanout box at 22 ft, out around a cracked cap.
2. Diffuse smoke through the ground surface over about a 6 ft patch centred on 61 ft.
3. Nothing at all from a yard drain grate 12 ft off the pipe line, which the plan shows tied into this lateral.

Reading the disagreement. Points 1 and 2 are positions, and both would have contributed to the air test failure, so between them they plausibly account for the 60-second shortfall. Point 3 is the interesting one. The plan says that yard drain is connected; the smoke test says nothing came out of it. That is not evidence the connection does not exist. A trapped yard drain holds against 0.25 in w.c. all day, so a live connection reads as an absence on a smoke test. On the air test, with 4.58 psi behind it, that same trap held nothing and the connection was one of the leaks in the rate.

So the two tests together say: at least two discrete openings at known positions, plus a trapped connection that is invisible to smoke and is contributing to the measured rate. Neither test alone produces that sentence.

Failure mode. The version that goes wrong fixes the cap at 22 ft and the joint at 61 ft and re-runs nothing. The next air test still fails, because the trapped connection was never on anybody's list, and the crew starts doubting the test rather than the inventory.

What changes which test is even valid

• Groundwater above the pipe does not just require the correction above; where it is deep enough that the corrected test pressure would exceed what the specification or the pipe allows, the air test stops being the right instrument and the section gets tested another way. Read the specification's limit rather than raising the pressure to suit.
• Live fixtures anywhere on the section rule out a pressure test entirely, by the trap-seal arithmetic above. Not "be careful": the test blows every seal on the section and pushes sewer gas into occupied rooms at 63 times what the seal can hold.
• A lined or cured-in-place section changes what a smoke test sees, because the liner can be intact while the annulus behind it and the original host pipe are not. Smoke emerging tells you the liner is holed; smoke not emerging tells you less than it does on a bare pipe.
• A section with no path to the surface, under a slab, a road base or a saturated clay lens, produces a smoke test with nothing to see and a pressure test that fails. That combination is a finding, not a null result.

Verifying you ran the right test

• State which question you were asked before choosing. "Does it leak and by how much" is a pressure test. "Where is storm water getting in" is a smoke test. A job needing both answers needs both tests, and running one and inferring the other is where the wrong repair gets sold.
• Write the corrected test pressure and the correction separately on the record: raw target, groundwater head, corrected target. A record showing only the final number cannot be audited later.
• Report the pass or fail against the specification's own criterion, in its own units, rather than converting it into a percentage that sounds worse or better than it is.
• Count the smoke emergence points and check the count against your summary, and record the absences too. The yard drain that produced nothing is a line in the report, not a blank.

References

• 29 CFR 1910.147, control of hazardous energy, for relieving stored pressure before disassembling a test section
• 29 CFR 1926 Subpart P, excavations: 1926.652 for protective systems, 1926.651 for egress and locating underground installations
• 29 CFR 1910 Part 1910.146, permit-required confined spaces, general industry; 29 CFR 1926 Subpart AA, the construction counterpart
• The model plumbing code as adopted and amended by your local jurisdiction, and the project specification on the job, which together set required test pressures, holding times and acceptance criteria
• Pipe and fitting manufacturer documentation on whether compressed-gas testing is permitted for that material
• See related: What Inflow and Infiltration Look Like From the Building Side; How to Run a Smoke Test Without Creating a Complaint; Why a Trap Seal Is the Only Barrier and What It Is Made Of