How to Separate a Belly From Roots From a Partial Collapse

Why this matters

These three faults produce the same customer complaint and take three incompatible repairs, and one of them punishes you for guessing. Jetting a partially collapsed line washes out the bedding that is still holding the pipe up. Cutting roots in a collapsed section can drop the remaining crown. Cabling a belly does exactly what it is supposed to and changes nothing. So the order of work is not a preference: rule out collapse before you put pressure or a cutter into the line, because that is the only one of the three where the diagnostic tool itself can make the fault worse.

Each of the three writes a different signature on three independent observables. Any one observable alone is a guess. Two agreeing is a working diagnosis. Three agreeing is something you can put on a quote.

Read this before anything goes into the ground

A sewer, a manhole, a lift station wet well, a large interceptor and a septic tank are permit-required confined spaces, and the requirement runs under 29 CFR 1910.146 in general industry or 29 CFR 1926 Subpart AA when the work is construction. Name which Part your job falls under before anyone opens a lid, because the programs differ.

The killer is the atmosphere, not the water. Hydrogen sulfide is detectable by smell at tiny fractions of a part per million, and it deadens the olfactory nerve at concentrations around 100 parts per million, which is also the NIOSH immediately-dangerous-to-life-or-health value. So the exposure that can kill you is the one you stop smelling. Your own nose is the least reliable instrument on the site, and a smell that fades is the warning sign rather than the all-clear. OSHA's general industry limits for hydrogen sulfide sit in 29 CFR 1910.1000 Table Z-2 as a ceiling with a short peak allowance; the construction limits are set separately in 29 CFR 1926.55. Oxygen deficiency and methane sit in the same space and neither one has an odour at all.

Nobody enters on a nose or a look. Test the atmosphere from outside the space with a calibrated multi-gas instrument reading oxygen, combustible gas, hydrogen sulfide and carbon monoxide, in that testing order, before entry and continuously during it, with the permit, the attendant, the retrieval system and the rescue arrangements the standard requires.

Address the rescue in advance, because this is how these incidents multiply. The most common second casualty in a sewer is the coworker who goes in after the first. A person standing at the lid is an attendant, not a rescuer, and an attendant who enters has left the space unattended and become a second victim. Rescue is non-entry retrieval where the geometry allows it, or a rescue service arranged before the entry began.

On the tools themselves. A drain machine's cable can loop and take a hand: run the machine on its foot control with the manufacturer's specified glove, never a rag or a loose wrap. A jetter nozzle is a high-pressure injection hazard; a stream at working pressure can inject through skin, and an injection injury is a surgical emergency that looks like a pinprick for the first hour. Keep the nozzle in the pipe before pressure comes up, and treat any skin contact as an emergency referral rather than a first-aid case.

And the one the last tech left you. Ask whether a chemical drain product went down the line in the past week and look for the bottle. Standing caustic above a blockage arrives with the water when a cap comes off: sealed goggles, face shield and chemical-resistant gloves rated for the product class from its safety data sheet, cap cracked from the side of the fitting, body out of the discharge arc.

Observable 1: the recurrence pattern

Pull the service history and lay the dates side by side. The sibling article on the line that blocked every few months owns this test and its reasoning; take its result as one input here rather than re-deriving it.

The third row is the one that carries urgency. A fault that gets worse on its own is structural, because neither a sag nor a root mass accelerates by itself.

Observable 2: the response to volume versus to mechanical work

Two tests, and they answer different questions.

The volume test. Run water into the upstream cleanout at a measured rate and watch the level. Do this before any tool goes in, because it is the only test that reads the line in its found condition.

• A belly takes water and returns it late: the level rises, then falls to a static level and stops there, because the pool is full and the pool does not drain.
• Roots take water at a reduced but steady rate: the level falls continuously and slowly, and it keeps falling, because the restriction is porous.
• A partial collapse takes water erratically: it may take a slug quickly and then back up sharply as material shifts, and the return water carries fines.

The mechanical test. What the cable meets tells you what is there. Roots give: the head meets a springy resistance that yields and then advances. A collapse does not give: the head meets a hard stop, or worse, it passes and then hangs on withdrawal, which is the head catching on a displaced edge. A belly offers no resistance at all; the cable runs the length of the line and comes back with nothing to report, which techs read as a clear line and is actually a positive finding.

If the mechanical test produces a hard stop or a hang, stop. Do not upsize the head and do not bring the jetter up to pressure. That is the point at which the diagnosis has told you the tool is now a risk.

Observable 3: what comes back

• Belly: fine sediment, grease, soft sludge. No structure to it. Often smells strongly sulfurous, because a static pool goes anaerobic.
• Roots: fibrous hair root, sometimes woody, with sheared ends where the cutter took it. The volume returned is the useful part; a cupful and a bucketful are different problems.
• Partial collapse: soil, sand or gravel, and shards. A curved fragment of clay or a piece of cast iron scale with a fracture face on it is definitive. Return water that runs turbid after cutting, and stays turbid, is bedding material entering the pipe, which is the tell that the line is taking in soil rather than just water.

Worked example: a lateral that got worse three times

The house. Single-family, 4 inch clay lateral, roughly 70 feet to the main, mature trees on the property line.

Observable 1. Four blockages spanning 14 months. Intervals between the clearings: 8 months, then 4 months, then 2 months. Three intervals, each half the one before it, which is monotonic shortening rather than a stable or a seasonal pattern. The middle interval spans autumn and winter, when a root pattern would have stretched rather than halved. Reads as partial collapse, and the direction is unambiguous.

Observable 2, the volume test, with its arithmetic. Water in through the cleanout at a measured 7.5 gallons per minute, taken by timing a 5 gallon container filled in 40 seconds: 5 gallons divided by 40 seconds, times 60 seconds per minute, is 7.5 gallons per minute. The level in the 4 inch riser rose. Riser volume per foot: the full bore area of a 4.0 inch internal diameter is 12.57 square inches, times 12 inches of riser is 151 cubic inches, which at 231 cubic inches per gallon is 0.65 gallons per foot.

Level fell from 3.0 feet to 1.0 feet in 95 seconds after the water was shut off. Fall: 2.0 feet, which is 2.0 x 0.65 = 1.30 gallons, over 95 seconds, which is 0.82 gallons per minute of drain-down.

The error line, because a two-foot fall read off a wet riser is not a precise number. Each level call is good to about 1 inch, or 0.083 feet, and that is an independent random spread rather than a fixed offset, so the two readings combine in quadrature: 0.083 times the square root of two is about 0.12 feet on the difference. That is 6 percent of the 2.0 foot fall, so the drain-down rate is 0.82 gallons per minute within roughly 6 percent. Precise enough to say it is about a tenth of the 7.5 gallons per minute going in, which is the finding: the line is passing roughly 11 percent of what it is being given.

Then the shape of the fall matters more than the rate. The level dropped fast for the first foot, stalled for about 20 seconds, and dropped again. A belly gives a smooth fall to a static level and then nothing. Roots give a slow, even fall. This was erratic. Reads as partial collapse.

Observable 2, the mechanical test. The cable met a hard stop at 52 feet that did not give under moderate feed. Work stopped there rather than upsizing the head.

Observable 3. The head came back carrying wet sand and one curved fragment of fired clay about the size of a thumbnail. Not root. Not sludge. Reads as partial collapse.

The call. Three observables, three agreements, and the sand plus the shard is the one that would stand on its own. Recommendation is a camera pass to locate and extent the damage, then a point repair or a lining decision, and explicitly no jetting of that section beforehand, because water at working pressure into a pipe that is already admitting bedding removes the support holding the remaining crown up.

The case where they disagree, and what it means. A different property: observable 1 reads seasonal, which says roots. Observable 3 returns a small volume of fine root hair together with a lot of soft sludge, which says roots plus belly. Observable 2 shows a smooth fall to a static level, which says belly.

Two of three say belly, one says roots, and the honest reading is that both are present and they are related. A sag concentrates its movement at a joint and opens it; an open joint in wet soil is exactly what a root finds. The companion reference on why roots find a joint carries that mechanism. Practically it changes the quote: cutting roots on that line buys a season, and the reason it only buys a season is the sag, so a root-only proposal is a maintenance plan being sold as a repair.

Confirming you got it right

• Say which observable was weakest and why you still went with the call. A three-way agreement is rare; naming the weak leg is what lets the next tech re-open the diagnosis honestly if the repair does not hold.
• Re-run the volume test after the repair, at the same measured input rate, from the same cleanout, and compare drain-down against the pre-repair figure. Same instrument, same setup, so the level-reading offsets cancel in the comparison.
• Check the return water for turbidity a week later. A point repair that stops the blockage but leaves the line taking in fines has not stopped the mechanism.
• Confirm nothing was jetted or cut before collapse was ruled out. If it was, note it in the record, because the condition you photographed afterwards is partly a condition you produced.

References

• 29 CFR 1910.146 (general industry) and 29 CFR 1926 Subpart AA (construction) for permit-required confined space entry, attendants, retrieval and rescue arrangements
• 29 CFR 1910.1000 Table Z-2 (general industry) and 29 CFR 1926.55 (construction) for hydrogen sulfide exposure limits; NIOSH for the immediately-dangerous-to-life-or-health value
• Safety data sheet for any drain-cleaning product identified on the line, and the drain machine and jetter manufacturer documentation for glove, foot-control and pressure requirements
• See related: The Line That Blocked Every Few Months in the Same Place; Why Roots Find a Joint and What That Tells You About the Pipe; Sewer Camera Inspection Reference