"Only a 25% Increase": The Sump Doing a Job Nobody Had Written Down

In the mid-1990s I was the Metallurgical Supervisor on a South African gold plant, responsible, among other things, for the flotation section. We floated a pyrite concentrate out of the gold plant tails, partly for the additional gold and partly to feed sulfuric acid production. It had been a commercial success, and the decision was taken to expand it to treat a higher proportion of the plant's tails: roughly a 25% increase in throughput. The installed flotation cells had spare capacity, so the expansion was mostly a matter of upgrading pumps and piping.

“A South African gold plant’s flotation expansion of about 25% was approved with its existing froth sumps unchanged, on the assurance that 25% was too small to matter. The sump was doing a job that appeared nowhere in the design basis: giving the air in the froth time to escape before the concentrate could be pumped, with the centrifugal pump quite possibly losing suction when it couldn’t. A chance conversation on the plant floor led to a redesign with about 50% more volume, a sloped floor and fixed froth breakers, and no further overflows in normal operation. The lesson is to ask, before sign-off, what existing equipment is actually doing, and to ask someone who knows how the plant behaves and has no stake in the plan as approved.”

I wasn't on the project team. The plan was approved by everyone who needed to approve it, and the engineer running it was a young colleague of mine. Because it was seen as a modest brownfields expansion, the existing concentrate sumps were to stay exactly as they were. It was only 25%, and the team had been assured by plant management that there was no problem with only 25%.

A walk through the plant

One day I was walking through the section and found the project engineer standing on the intermediate floor, looking down into the rougher concentrate sump, the largest of the froth collection sumps. We talked about the project plans, and I pointed out what the sump was doing right then. At existing throughput, its level rose, then suddenly fell, and every so often it rose far enough to overflow onto the plant floor, which was built to catch the spill and send it back to the flotation feed. He asked why it did that.

Flotation works by blowing air through a slurry so that the valuable particles attach to bubbles and rise as a froth. That froth carries a lot of air, and the air has to come out before the concentrate can be pumped away. Mechanical or static froth breakers and defoaming chemicals can all help it out faster, but this circuit had none of them. All it had was time, and gravity slowly squashing the froth down. The sump, without anyone having designed it to, was providing that time. Some of the swings in level were the air escaping. Quite possibly the main cause, though, was the centrifugal pump: when air-laden froth that hadn't had time to break reached its suction, the pump lost suction, the level in the sump climbed until suction recovered, and then dropped suddenly as the pump started pumping again. It looked like a storage vessel on the drawings. In practice it was also a deaeration vessel, a job that appeared nowhere in the design basis.

He asked what I would suggest. I sketched a different design on the spot: around 50% more volume, a steeply sloped floor, and fixed froth breakers. He took me at my word, redesigned that sump and the other froth sumps as well, and once they were installed we never had another overflow in normal operation.

What I can and can't say about it

I can't say exactly what would have happened if the original sumps had been left in place for the expanded throughput, because nobody ran that experiment. What I saw was a sump that was already surging and occasionally overflowing at the old throughput, and a plan that would have given the same air less time to escape and more froth to deal with. That is an inference, but not a difficult one.

What I can say with more confidence is how the problem was found. Nothing in the approval process would have found it. Everyone who signed the plan was working from a reasonable description of what a sump is for, and "only 25%" sounded too small to be worth a second look. What found it was a chance conversation between an engineer who asked why and a supervisor who had watched that sump every shift. He gets the credit, because he asked, listened, and changed a signed-off design.

Chance is not a mechanism. The question that did the work was a simple one: what is this equipment actually doing? It is worth asking deliberately, by someone who knows how the existing plant behaves and has no stake in the plan as approved, before sign-off rather than during a walk past a sump. It matters most when the change looks small, because a small percentage is exactly where nobody expects the equipment to be doing hidden work.

The drawings said what the sump was for. The people who operated it every shift knew what it actually did, and nobody had asked them.