Phosphate Rocks Chapter 38: Tuatara

Granulation looks easy: you add liquid and powder to one end of a drum. Make a porridge. Heat it up and spin it round. The acids and alkalis react and release heat. Lots of it. The water evaporates, plumes of fluffy white steam soaring from the top of a high chimney. Tilt the drum, and at the other end of the rotating cylinder, hot granules emerge to be screened and cooled. The oversized stuff is smashed and added to the dust, which is recycled back to the beginning, while the correctly sized granules are conveyed into the huge whalebone sheds. 

But the truth is that granulation is an art. The alkalis – liquid and gaseous ammonia, solid potash – must be balanced by the acids – phosphoric and nitric. The acid has to be just weak enough to provide water to wet the solids and bind them together as they react, but not so watery that they turn into gloopy slop at the end. 

In the art of granulation, which was nothing like the art of Dan Flavin, John was a master, a witch doctor. The Granny plant ran like clockwork on his shift. If he inherited a problem at the start of the shift – liquid emerging from the back end of the massive rotating cylinder, or the thump of giant boulders rattling and banging inside the drum – he would have fixed it by the time his shift left. 

John kept secrets. Knowledge was power, and John revelled in his pre-eminence, but the main reason for his silence was that much of his problem-solving involved actions which were, if not forbidden, at least frowned upon. And even where his conduct was logical and sensible, he lacked the vocabulary or science to fully explain what he did and how he knew what to do. 

Some of his directions seemed counter-intuitive. He would occasionally add more acid when the product was sloppy, increasing the liquid load further. Or when snowballs thumped around inside, he would add more solids. If he thought about it at all, he considered the granulator to be a giant gut. When you have diarrhoea, you need to drink more liquid; when you are constipated you ingest more fibre. It often worked. If the alkali ratio was too high, extra acid brought it back into balance, and the heat of reaction drove the water away. If boulders were forming instead of granules, adding more solids increased the number of potential nuclei available, and smaller granules developed. John had the benefit of experience, the courage of his convictions and the patience to wait while things got worse before they got better. John was a self-tuning control loop. His finest quality was his patience. What appeared to be an instinctive understanding was actually born of long experience, trial and error. If he made a change, he would not try anything else until success or failure was proven, a negative result being just as important as a positive one. He read the shift logs from his days off, cast an eye over the charts and log sheets. His eagle eyes and intimate knowledge of the authors allowed him to differentiate between fabrication and honesty, between numbers written down to satisfy management and what had really been done. He would file the trends away, synapses snapping as his brain constructed a huge internal model of the process. John was an analogue supercomputer in the days of binary control. A multi-dimensional digital array at a time of linear 3-15 psi pneumatics. 

You might think it easy. If you know the volume (in metres cubed) and you know the feed-rate (in metres cubed per hour) then you can calculate the residence time by dividing the former by the latter. The granulator was huge – four men, standing on each other’s shoulders, in diameter; forty men, lying head to toe, in length. So, if you feed an eighty-metre long, six-metre diameter granulator with two hundred cubic metres an hour, you might expect a residence time of about half a day, twelve hours. Although it is never more than a third full – the material tumbling from the walls as it rotates, an empty core of hot gas for the fertiliser to fall through – there is also no pure plug flow, no first in/first out fairness, but a considerable amount of back-mixing, the early particles going back and interacting with newer arrivals. So in normal circumstances the results of a change at the beginning of one shift are not known until near the end. The mistake that most people make is impatience. They make a change, see no improvement, make another. The change is too big, a new problem emerges, another change is made before the effects of the last one can be felt, and so on in a vicious cycle of instability. 

If there is insufficient reaction, the acids remain liquid. Without evaporation of water, the residence time changes, a slurry runs directly from one end to the other, failing to fill the rotating drum, failing to describe the complex helical path designed to pick and mix. As John liked to put it, diarrhoea – it could have passed through the eye of a camel. 

If there is too much reaction, then everything builds up inside. Instead of 30 per cent full the granulator becomes dangerously overloaded, building up until the whole thing sets solid. Constipation. 

John had a lizard’s patience, the attention span of a tuatara. Waiting, immobile, inflexible. Happy to dig and shovel with the rest of the team if push came to shove, but always trying to find a way to avoid waste. 

John also had a magic wand. A long-handled shovel that he kept hidden behind the long chute to the Granny screens. He used it to clear the sieves. The fine screen, designed to separate and recycle particles that were too small, often blinded, resulting in insufficient seed being sent back to the start of the process, and causing huge snowballs to form where Maltesers were required. 

He also used it when the product was too dusty. By shovelling the fine dust off the screens onto choking piles on the floor, he ensured that less seed returned to the granulator and the granule size increased. 

John’s understanding of both feedback and the more complex feed-forward control was highly developed. He had never seen a Nyquist diagram, never programmed a proportional, derivative and integral control algorithm, but he knew how to watch and wait and wasn’t afraid of a bit of mess in the middle.