What Hydrogen Sulfide Does to a Sewer Above the Waterline

Why this matters

A sewer destroys itself from the top down. The sewage in the bottom of the pipe is not what eats it; the acid manufactured on the damp inside of the crown is, and that acid is made out of a gas that came off the flow a hundred feet upstream. This matters to anyone who reads pipe condition for a living, because the invert can look sound in a camera run while the crown above it has lost a third of its wall. Judge the pipe on what the camera lights up and you will call a structurally compromised line serviceable.

Before anyone gets near the headspace

Everything in this card describes conditions in the air above a flowing sewer, and that air is the hazard.

• A manhole, a wet well, a large interceptor and any sewer big enough to enter are permit-required confined spaces. Entry to inspect a crown happens under a written permit program with continuous atmospheric monitoring, an attendant outside, and non-entry retrieval rigged before anyone goes down: 29 CFR 1910.146 in general industry, 29 CFR 1926 Subpart AA on construction work. There is no inspection finding worth an unpermitted entry.
• Hydrogen sulfide deadens the sense of smell well below the concentrations that incapacitate, so smelling less of it as you work is a warning, not reassurance. A sibling card covers what that does to a technician's judgement; do not substitute your nose for an instrument. See Why a Sewer Is a Confined Space and What That Changes.
• Where crown measurement can be taken with a pole-mounted or remotely deployed gauge from the surface, take it that way. The most common reason a crown never gets measured is that measuring it was framed as an entry, and then the entry was skipped along with the measurement.

The mechanism, in five steps

Each step happens somewhere different, which is the whole reason the damage is not where the source is.

1. Sulfide is made below the waterline. The slime layer on the submerged pipe wall goes anaerobic. Sulfate-reducing bacteria living in it use sulfate from the wastewater as an oxygen substitute and put out dissolved sulfide. This needs three things together: sulfate, organic carbon, and no dissolved oxygen. Long detention time supplies the third, which is why flat oversized pipe, wet wells and force mains are sulfide factories and steep well-scoured pipe is not.
2. How much of that sulfide can leave depends on pH. Dissolved sulfide sits in equilibrium between molecular hydrogen sulfide, which is a gas that will escape, and the bisulfide ion, which is not and will not. The split turns near pH 7: around neutral, roughly half of the total is in the escapable form; a full unit lower and roughly nine tenths of it is; a full unit higher and most of it stays in the water. A pH reading is therefore a leading indicator of headspace load, not a water-quality footnote.
3. Turbulence strips it out. Gas does not leave a calm surface fast. It leaves at drops, at junctions, at weirs, and above all where a force main discharges into a gravity line, because that discharge dumps hours of accumulated sulfide into a single point of violent mixing. The release point is the address where the corrosion will be found.
4. It is oxidised back down onto the crown. The gas dissolves in the condensate film on the cool damp surfaces above the waterline, where a different family of bacteria oxidises it to sulfuric acid. This colony needs moisture and oxygen, and the crown has both.
5. The acid attacks the material. Cement-based materials lose it: concrete pipe, manhole benching and walls, cement-mortar linings inside ferrous pipe. Unprotected ferrous metal loses it. Vitrified clay and the common thermoplastics are essentially indifferent to it, which is why a run of clay and a run of concrete in the same trench, carrying the same flow, age nothing alike.

Sewer cross section, flowing about a third full.

              crown
        ..............
     ..:              :..     acid film works here,
    .:                  :.    fed by gas off the flow
   .:                    :.
   :        headspace      :  <- springline
   :                       :
   :.._________________..:     normal waterline
     :~~~~~~~~~~~~~~~~~:
      :...invert.....:         submerged, washed,
                               essentially unattacked

Why the invert survives

Three reasons, and they compound. The invert is submerged, so the acid-forming colony that needs oxygen cannot live there. Whatever acid does reach it is diluted into the bulk flow and neutralised by its alkalinity. And the invert is continuously scoured, so the softened surface layer that protects the crown material for a while is stripped and replaced with fresh flow rather than allowed to sit. The practical result is a pipe with a sound invert, a sound submerged springline, and a crown that has been thinning for twenty years.

What this changes about reading a camera run

A standard camera run lights the invert and the flow. It is excellent at finding the things a sibling card covers - offsets, root intrusion, a belly, a partial collapse - and it is weak at the exact defect this mechanism produces, because crown loss is a thickness change, not a shape change, until it is nearly through. Signs that should send you to a thickness measurement rather than a condition grade:

• Aggregate visible in a concrete crown while the invert is still smooth. That is the cement matrix gone and the stone left standing.
• A crown surface that changes texture at a manhole and changes back after the next one. That is a release point upstream of the first manhole.
• Manhole benching and the first few feet downstream eaten while the run between manholes looks better. Turbulence made the difference.
• Any run immediately downstream of a force main discharge, a drop, or a pump station, regardless of what the picture looks like.

Worked example: measuring crown loss without charging the manufacturing tolerance to corrosion

A concrete run downstream of a force main discharge. The question is whether it has lost enough wall to need a structural answer or only a rehabilitation lining, and the answer has to come from a number rather than from a picture.

Where to measure. Not at a randomly chosen manhole. The general rule above says release happens at turbulence, so the reading is taken in the first full section downstream of the force main discharge, which is where the load is highest and where a low reading rules out the rest of the line. Record the pH of the flow at the same visit, because the pH gate above is what makes the headspace load credible or not: this one reads 6.6, below neutral, so most of the dissolved sulfide is in the escapable form and a high headspace load is expected rather than surprising.

The naive method, and why it overstates. The pipe class sheet gives a nominal wall of 1.50 inches. Gauge the crown, read 0.90 inches, subtract, and call the loss 0.60 inches. That charges the entire manufacturing tolerance of the pipe to corrosion, because nominal is a specification and the pipe as cast is not exactly nominal.

The differential method. Take a second reading on the same pipe section, at the invert, which the mechanism above says is essentially unattacked. Confirm that reference is valid before trusting it: the flow here runs about a third full at low flow, so the invert is genuinely and continuously submerged. If it were not, the reference moves up to the springline below the normal waterline instead.

• Crown reading: 0.90 inches
• Invert reading, same section: 1.45 inches
• Crown loss, as a difference on one pipe: 1.45 minus 0.90 equals 0.55 inches

The manufacturing offset is a fixed systematic error common to both readings on that section, so it cancels in the subtraction and does not appear in the 0.55 at all. What remains is the gauge's own scatter.

What the error term actually is. The gauge's repeatability is a spread of about 0.01 inches per reading, independent between readings, not a percentage of the reading. Two independent spreads combine in quadrature, so the spread on the difference is 0.01 times the square root of two, about 0.014 inches. Against 0.55 inches of loss that is under 3 percent and changes nothing. Had the loss been 0.03 inches, the same 0.014 would have been half the finding and the measurement would not have supported a conclusion.

Rate. The as-built dates this section at 22 years in service.

• 0.55 inches divided by 22 years equals 0.025 inches per year

Say what that number is: an average over the interval, not the current rate. The colony takes time to establish and the flow regime has probably changed at least once in 22 years, so the real rate was lower early and is likely higher now.

Remaining wall, stated as a bound. 0.90 inches remain at the crown. At the observed average rate, thickness reaches zero in 0.90 divided by 0.025, which is 36 years. Structural adequacy is lost well before thickness reaches zero, and where that point sits is a question for a structural engineer looking at the pipe class, the cover and the live load. So the honest statement carries one inequality sign and no interval: remaining structural life is less than 36 years, and the actual figure requires an engineer. Writing it as 36 years plus or minus anything would claim a two-sided interval the measurement does not support.

What flips the recommendation. If the invert reference had come back at 1.20 inches rather than 1.45, the invert is not intact and the differential method has no valid reference; that pipe is being attacked below the waterline too, which points at an industrial discharge rather than at biogenic sulfide, and the answer becomes a source-control investigation before any rehabilitation. If the material is vitrified clay or thermoplastic, crown loss is not the mechanism at all and a thin crown reading means something else, most likely abrasion or a manufacturing variance. And if the run is short and the release point is a single drop manhole, fixing the drop is a cheaper answer than lining the pipe downstream of it.

What actually reduces it

Ranked by where they act rather than by cost, because they are not substitutes:

• Cut detention time. Sulfide is made during the hours the wastewater sits without oxygen. A force main sized for a growth projection it never reached is the most common cause and the hardest to undo.
• Cut turbulence at the release point. A drop that can be re-profiled, a discharge that can be brought in below the waterline rather than dumped into air, removes the mechanism at the one address where it does most of the damage.
• Chemical addition raises pH or binds the sulfide before it can leave solution. It treats step 2, works, and is a permanent operating cost with a permanent failure mode: it stops the day the dosing stops.
• Ventilating the headspace dilutes the gas in the pipe and moves it somewhere. That somewhere is a discharge point that now has a corrosion and odour problem, so ventilation is a relocation unless the discharge is treated.
• Acid-resistant lining or material change does not stop the acid, it just stops caring. It is the only measure on this list that survives the day everyone stops paying attention.

References

• 29 CFR 1910 Part 1910.146, permit-required confined spaces, general industry; 29 CFR 1926 Subpart AA for construction work
• Manufacturer documentation for the thickness gauge in use, for its stated repeatability and the basis that repeatability is quoted on
• Trade-standard practice for biogenic sulfide corrosion assessment in gravity sewers and at force main discharge points
• See related: Why a Sewer Is a Confined Space and What That Changes; Sewer Camera Inspection Reference; Drainage Pipe Materials Reference; Reading Rust and Corrosion Patterns