Two Hundred Litres, Then Twenty Thousand: A Trial Protocol

The trial ran to plan: five doses in beakers, a clean curve, a clear optimum at 21 kilograms per cubic metre, and a report that went to the client in a week. Three months later the plant was running at 27 and still producing marginal batches. The beaker had not lied about the chemistry. It had lied about the mixing, the heat loss and the geometry, and those three things were worth 6 kilograms per cubic metre.

A trial is not a test of a chemical. It is a test of a chemical in a vessel, and the vessel has to resemble the one you will use.

Why a 1 Litre Beaker Over-Predicts

A beaker on a magnetic stirrer puts more energy per cubic metre into the liquid than almost any full-scale mixer. The stir bar creates a high-shear zone a few millimetres across, and the small volume means every part of the liquid passes through it many times a minute. Full-scale mixing is gentler and less uniform.

A beaker also loses heat differently. Two hundred millilitres in a glass beaker on a bench at 22 °C stays at 22 °C. Twenty cubic metres in a steel tank at 06:00 in November does not.

And a beaker has no dead zones. The full-scale vessel has corners, a bottom, a discharge point and a lid, and the material in those places does not see the same shear as the material in the middle.

The net effect is usually an over-prediction of how well a given dose will work, which shows up later as a plant running above the trial number and nobody trusting the trial.

The Three-Stage Protocol

Run it as three stages, each one an order of magnitude up, and do not skip the middle one.

Stage one, 200 litres. A 200 litre drum or a cut-down IBC, a paddle wide enough to move the whole volume, and five doses bracketing your estimate: 14, 17, 20, 23 and 26 kilograms per cubic metre. One batch each, three repeats at the dose that looks best. Mix for the same time you plan to use at full scale — not longer, because longer mixing flatters the result. Record temperature at the start and end, and read the free-liquid result at 10 minutes, 1 hour and 24 hours.

Stage two is where the surprises come out: at 24 hours, cut a lump in half. If the centre is dry or glossy, the batch is not homogeneous and the dose is not the problem.

Stage two, 2 cubic metres. The plant mixer, at a fifth of its working volume, with the plant’s powder addition method. If the site uses a hoist and a bulk bag, use it. If it uses a bag splitter, use it. This stage isolates the effect of the vessel, and it is the stage that most trials skip.

Expect the optimum to move up by 5 to 15 percent. If it moves by more than 20, the problem is the mixing method rather than the dose, and no amount of powder will fix it.

Stage three, full batch. Two batches at the stage-two dose, sampled at three positions and tested with the method your receiving facility uses. Only then write the setpoint on the board.

What to Record, and Why It Matters Later

• Polymer grade and lot number for every stage. A trial run across two lots is two trials.
• Liquor: pH, conductivity, ammonia, solids, temperature. One sample per stage, taken at the same point.
• Mixing: vessel, impeller type, speed, time, whether a top-up was used.
• Result: which test, at what time, read by whom, and photographs at 10 minutes and 24 hours.
• Weather, because a trial run in a cold snap will produce a number that does not hold in June.

The record is what lets you defend the setpoint when a new shift supervisor asks why 27 and not 21. Without it, the number looks arbitrary and gets changed.

The Cost of Doing It Properly

Stage one takes a day and about 150 kilograms of polymer. Stage two takes a day and a tonne. Stage three is two production batches. Total: three days and roughly 3,000 dollars at typical prices, against a setpoint that will run for a year on 12,000 cubic metres. A 10 percent error in the dose is 26,000 dollars. The trial is not the expensive part; skipping it is.

One caution: do not run the trial on a sample that has been sitting in a jug for a week. Leachate changes in a sealed container — solids settle, biology shifts, pH drifts. Sample the day you trial, and keep the sample cool but not cold.

The grade comparison data is the right starting point for choosing which polymer to put through the ladder, since trying three grades across three stages is nine days and nobody has nine days.