I cannot find it on Google Earth. It stuck one kilometer out into the bay and was one kilometer wide, so it should still be visible. I refer to a phosphogypsum tailings impoundment I designed for Louis Luyt at Richards Bay in South Africa in the late 1970s.

His company, Triomf Fertilizers, was going into the business of fertilizer manufacturing. They had a mine somewhere in the Transvaal and transported the raw rock to Richards Bay where they were building a factory. As a staff engineer with SRK I was called on to design and oversee construction of a tailings impoundment for the new factory.

Triomf Fertilizer seems to have disappeared – the web hints at skullduggery and illegality, but gives no details – and so too has my one kilometer by one kilometer tailings impoundment.

Unless they have reclaimed the bay around the pile and built beside it. On Google Earth I spy a piece of land protruding out to the south-east into the bay. But that is not the same size as my impoundment; the river is in the wrong place; and the bay seems smaller than I recall it. But the southern tip of the protrusion looks desolate and white in places.

That impoundment presented a challenge. The bay muds were soft, very soft. You sunk to your ankles and deeper if foolish enough to walk were there was no vegetation. I did some primitive shear vane testing and plugged the results into the even more primitive slope stability computer code up at the university.

In those days SRK had no in-house computer. Rather you punched out the run instructions and data on punch cards, drove up to the university, wondered into the computer center, and deposited your cards in the box for commercial runs that were done overnight. The next day back to the computer center to retrieve your cards and a paper printout. Sometimes there were results, but all too often you had an input error and so another 24 hours had to elapse before you got another run.

The first geometry I tried involved a two horizontal to one vertical slope of gypsum. Afterall at Chloorkop they were building on a liner at 1.5H:1V. The computer blew up. I tried three to one, and then four to one, and still the computer blew up. Except when it was polite enough to tell me that my factor of safety was indeterminate or less than one. I grew desperate: how could I propose a tailings impoundment slope flatter than four to one? Oscar Steffen was called in along with Hendrik Kirsten. They checked my shear vane results; they checked my computer cards. And Oscar pronounced that we would have to try five to one. Even that did not work.

While waiting for the daily computer output I had begun to read deep literature on clay shear strength. I knew about the difference between total strength and effective strength, but I really knew little about gain of strength resulting from stress increase, consolidation, and pore pressure dissipation. Now days it is all so simple, that I shudder to recall how complex and mysterious it seemed then. Skempton was the only person writing about it; and some fellow in Quebec who was building roads over soft muds.

By devious means I now forget, I came to realize that I had to load the clay foundations slowly, let the pore pressure dissipate, and let the clay gain strength. For if I could not do that, I would never get that silly gypsum embankment to stand up.

Luckily the rate of rise of the gypsum tailings was calculated to be no more than a few feet per year. Not really knowing the permeability of the clays I had no way of estimating if the clay strength gain would be sufficient to deal with the increasing height of the slope. And of course there was no computer code to even begin to analyze the conditions.

At Oscar’s suggestion I generated many computer card piles. Each for a different height of pile and for a different theoretical strength. Now I arrived at the university computer center with piles of cards and next morning retrieved ten to twenty runs. We had to get special dispensation to do so much computing each evening, and they charged us.

Gradually I built up graphs of strength versus pile height versus slope inclination. I bet that file is still somewhere in the SRK files in Johannesburg. Relying on engineering judgment, I settled on a slope inclination of five to one. And elicited a promise they would not increase the height of the impoundment more than a foot to three a year. The big problem was persuading them to install piezometers and making them promise to let us monitor the piezometers and let us decide if the pile could go higher.

Triomf was represented by a young chemical engineer—I seem to recall that his name was Les Batch. He was bright and I brought him along with me on this experiment. Nervously he let me proceed to construction.

I had met the representatives for Bidim in South Africa and they had convinced me to place a layer of geotextile over the clay before laying down the one meter sand layer that I proposed as the starter dike for the impoundment. I had no way of calculating the stress on the geotextile. So I choose the strongest available. Teams of Zulus cleared the vegetation and we rolled out the thirty-meter wide zone of Bidim. To connect the rolls and make sure they stayed together, we had other teams of Zulus sweating in the sun as they used every wire coat hanger in the land to “stitch” the individual strips together.

Then I let the scrapers go. Loaded with sand from the escarpment to the north, they bounced out over the Bidim. It became a game: could you go faster than the bow wave that developed ahead of the scraper.? Could you pull out of the depression around your machine? Could you remain sober as they scraper rose and fell by upto half and more of a meter?

Regularly the Bidim joins would fail and a volcano of soft mud would spray up and pool on the clean sand. We would pull back, bring in the backhoe, dig a bit, lay another strip of Bidim, pile the sand over it, and get back to running the scrapers.

Somehow meter by meter we advanced the sand berm out into the bay and around the corner and back to the escarpment. J.P Giroud heard about what we were doing and came to visit. Bidim brought people from all over the land to see. I became a virtual tour guide.

One fateful day the waste phosphogypsum began to flow into the impoundment. The sand berms held, the piezometers fluttered, and it was proclaimed a success.

Other things brought me to Tucson. Another engineer took over. He continued monitoring piezometers, refined the stability analyses, and made the decisions to install sand drains along the southern perimeter of the impoundment where the slope was planned to be highest, where the pore pressure were not dissipating and where outward movement was a little more than we cared for.

Now that is all forgotten. I cannot even find the pile for it must be pretty well vegetated and maybe built on. Now I have FLAC to analyze similar situation. There is a never-ending supply of different geotextiles and geogrids, and geocells to choose from, and I do not have to drop a bare electric wire down the piezometer to get a reading.

My phosphogypsum impoundment will never be a monument to engineering folly, courage, or innovation. But still I remember it fondly. And now I use the experience to go further and further across soft materials. It a wonky foundation, but it suffices.