Three Depths, Three Different Doses: Sampling a Stratified Sump

The bailer came up with a sample at 14.2 millisiemens per centimetre, the batch was dosed to that number, and the load was rejected two days later. The sump was 4.6 metres deep. The top metre read 9.8, the middle read 14.2, and the bottom 600 millimetres read 21.6 with an ammonia figure nearly four times the top. The sample had come from the middle because the bailer was a metre long and that is where it ended up when the rope was let go.

A sump is not a stirred tank. It stratifies, and the stratification is stable enough to survive a week of normal draw-off.

Why the Layers Separate

Three mechanisms, usually together. Density: a stronger, saltier liquor sits lower, and in a deep sump it stays there because the density difference suppresses mixing. Temperature: warm leachate from a fresh cell is less dense and floats above cold liquor from a capped cell, even when it is dirtier. And settling: solids fall, carrying adsorbed organics and metals with them, and the sludge blanket at the bottom has its own chemistry entirely.

The practical consequence is that a single sample from a stratified sump describes one layer and misrepresents the other two, always in the direction that makes the batch look easier to treat.

The Numbers From One Sump

A batch drawn from the bottom of that sump at the middle-layer dose would produce a soft, wet mass. A batch drawn from the top at the same dose would be over-treated and stiff. Same sump, same day, same recipe.

How to Sample Three Depths Without Buying Anything

A bailer with a check valve on a marked rope is enough. Lower it slowly — 0.3 metres per second or slower — and the valve closes on the way down so it fills at the depth you stop at rather than on the way through. Mark the rope at half-metre intervals with tape.

Take three: 300 millimetres below the surface, mid-depth, and 300 millimetres above the sludge blanket. Note the blanket level from the same exercise, because it moves, and a rising blanket changes which layer your draw-off valve is pulling from.

Sample before mixing and again after 30 minutes of recirculation. If the two sets converge, the sump mixes and your problem is scheduling. If they do not, the sump does not mix and your problem is the draw-off point.

What to Do With Three Answers

You have four options, and the right one depends on the sump.

Dose to the worst layer. Simple, safe and expensive, because you are paying for the bottom layer on every batch even when the draw is mostly from the top.

Draw from a fixed depth and dose to that layer. Cheap and predictable, provided the draw-off point is at a fixed height and the sump does not turn over.

Recirculate before drawing. Thirty to sixty minutes of mixing with a submersible on a stand, then sample once. This is the best answer for most sites, and it costs a pump and an hour.

Treat the layers separately. Only worth it if the bottom layer is a small fraction of the volume and needs a very different recipe, which is common in sumps with a deep sludge blanket.

The Failure That Keeps Happening

Weighted bottles. A sample bottle weighted with a length of chain and lowered on a rope fills as it descends, so the contents are an average of the whole column weighted toward the top, where it spends longest. The result is a plausible number that describes nothing. If the sampler is a bottle with a stopper on a separate line, it works; if it is a bottle with a weight, it does not.

The second recurring failure is sampling from the draw-off line itself, which is fine in principle and misleading in practice if the line has been sitting full for a week. Run it for two minutes before taking the sample, and record that you did.

Records That Show You Looked

Log the three depths on the batch sheet, not just the number you dosed to. When a load is questioned, a sheet showing 9.8, 14.2 and 21.6 with a note that the batch was drawn at 2.4 metres tells a coherent story. A sheet showing a single number tells an inspector nothing about whether the site understands its own sump.

The sampling plan guidance covers depth sampling alongside the composite and grab options, including how to decide when a composite is actually worse than three separate numbers.