The Dose Was 400 Millilitres a Day. The Smallest Pump We Could Buy Was Twenty-Two Times That.

Four grease traps under one commercial kitchen. A hundred millilitres of conditioner into each of them overnight. Four hundred millilitres a day, all four traps together. That was the entire duty, and I spent an afternoon discovering that the smallest industrial metering pump I could actually buy moves about eight and a half litres a day. Twenty-two times the whole daily requirement, before anybody adds a safety margin on top.

Last week I wrote about what is in our grease trap conditioner and why there is no enzyme in it. That post was about the bottle: what the chemistry is, what form it takes, why the powder and the sachet were both rejected. This one is the other half of the same job, and it is a hardware problem rather than a chemistry one. Once you have decided what the liquid is, something has to put it in the trap at two in the morning when nobody is in the kitchen. That turned out to be harder to specify than the product was to formulate, and the thing that finally solved it was not a better pump.

The duty is too small for the industry that makes the equipment

Metering pumps exist to feed chemicals into water treatment plants, and the plants they were designed around are bigger than a grease trap by a wide margin. On the sewage side we dose alkalinity correction and coagulant in litres per hour, and the pumps we buy for a sewage treatment plant are sized for that. A grease trap wants a hundred millilitres, once, at night.

The smallest unit in the range I priced first is rated at 2.3 gallons a day, which is 8.7 litres. The next brand up starts higher again and climbs to 25 litres an hour, which is a different world entirely. There is one premium industrial unit with a turndown wide enough to meter this duty properly at the low end, and it is the most expensive thing in the catalogue. So the choices were: a pump that cannot be turned down far enough, or a pump that can be turned down far enough and costs more than the whole dosing problem is worth.

This is a recurring shape in our work and it is worth naming, because it comes up in small plants generally. The equipment market is built around a middle size. Go far enough below it and you are not buying a smaller version of the same thing — you are buying the same thing and throwing most of it away.

The oversizing stopped mattering when we stopped metering by flow rate

The mistake was mine, and it was a framing mistake. I had been reading the pump's rating as a dose: eight and a half litres a day against a four hundred millilitre requirement, therefore twenty-two times too big, therefore wrong. But the pump does not run all day. It runs when the timer says so.

Eight point seven litres a day is six millilitres a minute. A hundred millilitres at six millilitres a minute is about seventeen minutes. Four traps, run one after the other, is a little over an hour. The kitchen is closed, there is no flow through the traps, and nothing is competing for the dose. The pump is not oversized once the controlled variable is time rather than flow rate. It is simply a pump that runs for seventeen minutes instead of all night.

I like this kind of answer because it costs nothing. It is the same instinct behind reading a settling cone instead of buying an MLSS meter: the measurement you can already take, used differently, is usually cheaper than the instrument you were about to specify. It also means the pump stops being a precision instrument and becomes a timer with a tube on it, which is a much easier thing to keep running in a hotel basement.

Then one pump had to feed four traps

Four traps, four tubes, one pump. The first architecture I drew put a single pump in an indoor cabinet with four solenoid valves and a sequencing timer: valve one opens, pump runs seventeen minutes, valve one closes, valve two opens, and so on. All the electrical work stays indoors, away from the wet area, and four lengths of tubing leave the cabinet. On paper it is the cheapest possible answer. One motor, one controller, four cheap valves.

I drew it, priced it, and then threw it out. Every solenoid valve is a thing that can fail closed, and a timer sequencing four of them is a thing that can fail in a way nobody notices for a month. Nothing downstream of a failed valve tells you it failed. The trap simply stops being dosed, and the first sign is a crust, by which point you are three weeks into a problem and arguing about whether the product works.

That is a design argument we make in other places and I had not applied it to our own product. It is the same reasoning behind needing less equipment rather than better equipment, and behind the way we now draw the clarifier with no weir plate in it — an adjustment nobody can reach is not an adjustment, and a valve nobody can see fail is not a control. The cheapest architecture on paper lost to the one with fewer things in it.

What replaced it, and the wording we had to correct in our own quotation

The replacement is a multi-channel peristaltic pump: one motor, one controller, three or four independent tube channels running on a single rotor. Each channel has its own tube and its own outlet, and because they all sit on the same rotor turning at the same speed, every trap gets the same dose by geometry rather than by adjustment. There is nothing to balance and nothing to set per trap.

Our own earlier quotation described this as a four-point distribution manifold, and that was wrong. A manifold is one flow split four ways, and one flow split four ways does not divide evenly. The tube runs are different lengths, the traps sit at different levels, the back-pressures differ, and the split drifts further out of balance as the tubing ages. The trap at the end of the longest run gets the least product and is the one that crusts first. We corrected the wording in the quotation rather than leaving a term in the document that described a different piece of equipment from the one we intended to install.

I mention it because a quotation is read as a specification, and a client comparing two documents cannot tell the difference between a manifold and a multi-channel head unless one of us writes it down properly. Loose wording in our document is our problem, not the reader's.

The tubing chose the pump, not the flow curve

The thing that actually eliminated most of the candidates had nothing to do with capacity. Our conditioner is a surfactant blend and it contains butyl glycol. Silicone tubing swells in contact with glycol ethers. It softens, the bore changes, the dose changes with it, and eventually the tube fails at the rotor.

Most of the small multi-channel pumps sold at this size are aquarium and laboratory units, and they ship with silicone tubing and nothing else. The industrial brands sell three or four different tube chemistries for exactly this reason, and that — not the flow curve, not the turndown ratio — is what kept them on the list. If you specify a dosing pump by its millilitres per minute and never ask what the tube is made of, the pump will meter beautifully for a few months and then quietly stop doing so.

We have had the same lesson from the other direction. Our own dosing pumps on sewage plants last about a year, and the wear item is almost always the wet end rather than the motor. A pump is a motor, a head and a consumable, and the consumable is the part that decides whether the plant is still dosing correctly in month nine.

One line in the drawing that is there to stop somebody killing the pump

A peristaltic pump is a positive displacement machine. It does not care that the outlet is shut. It will go on squeezing the tube against a closed line until something gives, and what gives is the tube.

So in any version of this that keeps valves in it, the control has to guarantee that at least one path is open whenever the motor is running. It sounds obvious written down. It is not obvious at two in the morning inside a controller somebody reprogrammed, which is part of why the valve architecture lost. The multi-channel version has no valves at all and therefore cannot be made to deadhead by a sequencing error, which is one more line of reasoning in its favour rather than a separate one.

The price pulled the whole thing back to earth

The first configurations I priced came in at three hundred thousand pesos and upwards, and I stopped the exercise there. That is sewage treatment plant money spent on getting four hundred millilitres of surfactant into four concrete boxes. We went back to a different manufacturer at roughly nine thousand pesos a pump — the same construction quality as far as I can judge it — with the tubing changed out for a chemistry that suits the product.

That is the decision in the end: buy the pump for its build and its tube options, not for its catalogue turndown, and change the consumable to suit the chemical. The conditioner itself is about five hundred pesos a litre and the digesting culture about two hundred and fifty, both sold and dosed by the litre, so a four-trap kitchen is spending roughly two hundred pesos a day on product. A dosing system that costs more than a year of product is not a dosing system, it is a monument. As always, your own numbers will differ — trap count, trap size, how much fat the kitchen actually produces and whether the water is hard all move the dose, and they move the hardware with it. The price estimator and the design generator will give you the shape of the larger plant, though neither of them sizes a grease dosing line.

Most kitchens still get a bottle and a line on the wall

Everything above applies to a kitchen with four traps and a maintenance team. A single trap behind a restaurant gets a five litre dispenser mounted on the wall and a pour bottle marked at every hundred millilitres, and the closing staff pour it before they lock up. There is no pump, no controller and no tubing chemistry to get wrong.

I am not being modest about that. The measured pour has fewer failure modes than anything with a motor in it, and the only thing it depends on is a person who is already in the room at the right time. We put the pump in when the trap count makes pouring tedious enough that it stops happening, or when the traps are somewhere nobody goes at closing. That threshold is about behaviour, not engineering, and it is why the same commercial kitchen can justify automation in one building and not in the next one along.

What the dosing line cannot do, whichever way it is built

It is worth repeating the boundary from the chemistry post, because a pump makes people more optimistic than a bottle does. Dosing disperses fat and feeds a culture that digests it. It does not turn a grease trap into a grease interceptor, it does not do the work of a dissolved air flotation unit, and it will not rescue a trap that was built too small at the front end. If the crust in the trap is calcium soap rather than fat, no dose and no bacteria will touch it and the answer is upstream in the water, not in the pump.

It also does not make anybody compliant on its own. What arrives at the treatment plant downstream still has to be treated, and the parameters you are held to come from the receiving water, not from the kitchen. Our pages on discharge compliance and the parameters that matter by industry cover that side, and the nutrient field guide covers what happens when fat reaches an aeration tank and starts showing up as brown greasy foam or a clarifier carrying solids over.

Where this stops transferring

Several of the decisions above are ours and would be somebody else's differently.

If your conditioner has no glycol ether in it, the tubing argument collapses and silicone is fine, which puts the cheap multi-channel units back on the table and changes the price by an order of magnitude. If the traps are dosed during the day while the kitchen is running, metering by time stops working, because there is flow through the trap carrying the dose away and the controlled variable has to go back to being rate. If there is one trap rather than four, the whole multi-channel argument disappears and a single small pump is the obvious answer. And if you have a maintenance contractor already visiting weekly, the pour bottle may be more reliable than anything automatic, simply because somebody is checking.

The rejection of the valve architecture is a judgement rather than a calculation, and I will say so plainly. Solenoid valves work. Plenty of properly built systems use them and run for years. We rejected them for a product that gets installed in buildings we do not operate, maintained by staff who change, where a silent failure costs us a reputation rather than costing the owner a day. In a plant with an operator who reads a logbook, I would weigh it differently. Who is actually going to be standing in front of the equipment changes what is a reasonable thing to install, more than the equipment itself does.

The prices are from quotations we have actually handled and they move with the supplier, the exchange rate and the year. Treat them as the shape of the problem, not as your budget.

If you are working through a grease or dosing problem in a kitchen or a plant in Cebu, we work across the region, you can see what we have built or who we are, send the site details through the request form, ask the PCO help centre if the question is a permitting one, read how we work with clients or how a plant gets built, or just book a free 15-minute consultation and tell me what the trap is doing.