The pH Was Falling and Nobody Had Changed Anything. He Asked Which Chemical to Buy.

A plant we look after started drifting on pH. Not dramatically. The operator's log showed 7.4, then 7.1, then 6.8 over about three weeks, and nothing on the site had changed — same building, same occupancy, same blowers, same sludge schedule. He asked me which chemical he should start dosing to hold the pH up.

That is the right question asked one step too late. The pH was not the thing that moved first. The alkalinity had been falling for weeks before the pH followed it, and the first thing I wanted to do was not buy a chemical. It was to check whether the plant was already able to make some of its own buffer back. This post is about that order of operations. We have written before about the arithmetic that decides whether you need to buy carbon for denitrification, and about the four quiet things that stop a plant nitrifying, where alkalinity is one line among four. This one is the alkalinity line on its own: where the buffer goes, where we get it back for free, which three chemicals we have actually used when free is not enough, and why how you introduce it turns out to matter more than which one you pick.

The buffer goes before the pH does

Nitrification is an acid-forming reaction. Every milligram of ammonia nitrogen the bacteria oxidise consumes roughly 7.14 milligrams of alkalinity as calcium carbonate. That number is standard and you can look it up, but what it means on a working plant is less obvious than it looks on paper.

Alkalinity is the plant's shock absorber. While there is buffer in the water, acid production does not move the pH much at all — the buffer takes it. The log looks stable. Then the buffer gets thin, and the same acid production that was invisible last month starts walking the pH down a tenth at a time. By the time the pH is visibly falling, you are not at the start of a problem. You are near the end of a reserve.

This is also why I am wary of judging a plant on a single instantaneous reading of anything, which is a habit we have had to learn the hard way — we once left a dissolved oxygen of 1 mg/L alone because everything else about that tank said it was working.

So a pH trend is a late indicator. On the plant I opened with, the pH log was three weeks of bad news that had already happened. If the operator had been reading alkalinity rather than pH, he would have seen it coming in week one.

And the consequence is circular, which is what makes it worth catching early. Everything we design has to hold its parameters without somebody watching it daily, and a slow buffer drawdown is the failure mode that most easily escapes a daily look. Low pH slows the nitrifiers down. Slower nitrifiers mean ammonia starts to appear in the effluent. The plant that was comfortably inside its ammonia limit now is not, and the cause does not look like ammonia at all. If ammonia is the parameter you keep failing, this is one of the places we look first.

Before we dose anything, we check what the plant can give back

Destroying nitrate returns alkalinity. Roughly half of what nitrification consumed comes back when nitrate is reduced in an anoxic tank. That is not a chemical you buy; it is a reaction that happens in a tank with the air switched off, fed by water you already have.

Which is why, on plants where nitrate is nowhere near its limit, we still build the anoxic stage, still install the return pump, and still run a return at about one times the incoming flow. The pump is not there to chase nitrate on those sites. It is there to carry alkalinity back to the front of the plant. We said this in one paragraph in the carbon post; it deserves to be said properly, because it is the cheapest alkalinity any plant will ever get and it is almost always available.

The case that makes it concrete is a plant with weak ammonia. Say the influent ammonia is around 10 mg/L and the nitrate limit is 14. Convert all of that ammonia and the nitrate still lands inside the limit without doing anything about it. On that plant, running the return at three or four times flow — which is what we would do if nitrate were the problem — buys nothing. We run it at about one times, and the only reason the pipework and the pump are in the design at all is the buffer.

Clients sometimes read that line in a quotation as padding. A recirculation pump on a plant whose nitrate was never going to fail does look like something you could delete. It is one of the items I defend most often, and the defence is that deleting it is how you end up buying a chemical and a dosing pump instead, both of which cost money every month, which is the part of a plant that actually decides what it costs to own.

Three chemicals we have actually used

When the free half is not enough, we dose. We have used three things, and we are not loyal to any of them.

Lime is the cheapest, and on a plant where cost per cubic metre is the argument, it is usually where the conversation starts. It is also the most awkward to handle, it settles, it fouls lines, and it is the one most likely to end up as a bucket and a stick rather than a working dosing system — which is a comment on us as designers, not on the chemical.

Soda ash is also cheap, and I will say plainly that it has been a lifesaver on more than one plant. It goes into solution properly, it is easy to make up, and it does the job without the handling problems. Where I have a choice and the cost difference is not decisive, this is usually what we end up running.

Sodium bicarbonate is the gentle one. It is harder to overshoot with. On a plant where the operator is part-time, or where the dosing is manual, a chemical that is difficult to get badly wrong has a value that does not show up in the unit price. It is the same reason we default to fixed-film media — not because it is clever, but because it is forgiving. We have written before about designing around who will actually run the plant, and this is that same reasoning applied to a chemical choice.

None of the three is hard to find in the cities we work in, which matters to us. We do not let compliance depend on a chemical that might not be on the shelf in the city the plant sits in.

How you introduce it matters more than which one you chose

This is the part I cannot give you a rule for, and I would rather say so than invent one.

Some plants take a direct dose into the aeration tank and simply absorb it. Others do not. Put the same daily quantity in as a slug and you get a local pH spike where the dose lands, which the nitrifiers in that part of the tank feel, and then a slow slide back down over the day. Put it in slowly over hours and the same quantity holds the plant flat. We have had to find out which behaviour a plant has by trying it, watching, and changing the introduction rather than the chemical.

So when we commission a dosing line, the arrangement is deliberately adjustable: we would rather have a cheap means of slowing the dose down than a precisely calculated feed rate that assumes the tank mixes the way the drawing says. The mixing in a real tank is not the mixing in a calculation.

If I had to generalise from what we have seen — and I am generalising from a small number of plants — the plants that tolerate a direct dose are the ones with real mixing and plenty of volume ahead of the point where it matters, and the ones that need it slow are the small plants where the dose is large relative to the tank. That is a hypothesis, not a finding. Nobody has measured it on our plants.

Why alkalinity dosing appears in our quotations as an estimate, and marked "if required"

Our operating cost pages carry the alkalinity chemical as an estimated figure and explicitly as conditional. That is not us hedging. It is the honest state of the information at the time we write the document.

Whether a plant will need alkalinity correction depends on the alkalinity of the incoming water and the ammonia load, and on most jobs we do not have a measured influent result to work from — we ask for one on every job and almost nobody has one. We can assume, and our assumptions are deliberately safe. But "safe assumption" and "measured" are different things, and a client reading a firm monthly chemical cost would reasonably believe we knew. We do not, until the plant runs.

The commercial consequence of that honesty is that our quotation looks slightly more uncertain than one that prints a confident number, and I have lost at least one comparison on that appearance. I would still rather write the line that way. The alternative is a figure the client holds us to for something we had no way of knowing in advance, or a line quietly left out entirely, which is worse — it shows up as a surprise in the first month of operation. When two quotations are side by side, the difference is usually scope before it is price, and an omitted chemical line is exactly that kind of difference.

What we actually watch on site

Alkalinity is cheap to measure and almost nobody measures it. It is a titration with a test kit, it takes a few minutes, and it tells you about next month rather than this morning. It is also the kind of reading that makes a monthly visit worth something, which is most of what our operations support actually consists of. We would rather have a weekly alkalinity reading than a daily pH reading, and if we can only have one number from a plant between lab visits, that is the one.

We take it alongside the readings we already take rather than adding an instrument nobody will use — the same thinking that has us reading a settled-sludge cone rather than chasing MLSS results we would wait a week for. The readings we take on a site visit, and what each one is telling us, are written out in our ammonia, nitrate and phosphate field guide.

Two situations where we watch it harder than usual. The first is a new plant in its first weeks, when the nitrifier population is still building and the demand on the buffer rises as it does — we watched alkalinity through exactly that on a plant that failed its first test while being perfectly healthy. The second is after any interruption to aeration, because ammonia accumulates while the air is off and then gets oxidised in a rush when it comes back, and that rush draws down buffer faster than the plant's normal day does. When a blower stopped for three days, the thing we left alone was the chemicals and the thing we watched was the alkalinity.

What we did on the plant I opened with

We checked the return first. The pump was running, but the recirculation had been throttled back at some point — valve, not design — and the return was well under the one times flow it was built for. Opening it up recovered part of the buffer within a fortnight and the pH stopped falling.

It did not come all the way back, and that plant now doses soda ash as well, modestly, introduced slowly. But the dose we are buying is a fraction of what it would have been if we had answered the operator's original question directly and sized a dosing system against the full deficit. The free alkalinity was sitting in a valve somebody had closed.

I do not tell this story to make the operator look careless. He had a pH problem and he asked a sensible question about chemicals. The reason nobody looked at the valve is that the recirculation pump on that plant has never had an obvious purpose — its purpose is a reaction that produces no visible result. An item like that is exactly what gets quietly turned down. We now explain what the return is for in the handover rather than only on the drawing, which is part of a broader habit of treating what happens after we leave as a design problem.

Where this does not transfer

Everything above assumes a plant that is nitrifying and has somewhere to send nitrate. A plant with no anoxic stage and no space to make one has no free alkalinity to recover, and the whole first half of this post collapses into "you will be dosing" — which is fine, but it changes the economics and it belongs in the design, not in the first month of operation.

It also assumes influent that is not already strongly buffered. Some groundwater in the Philippines is hard and carries plenty of alkalinity with it, and a plant drawing on that may never need any of this. That is one of the facts we ask for as a routine survey item rather than discovering later.

Weak-ammonia plants are the ones where the one-times return is the whole answer. A plant with influent ammonia at 200 mg/L is a different problem entirely: there the buffer demand is large enough that the return is a contribution, not a solution, and dosing is in the design from the start. The choice between the three chemicals is ours from a small number of jobs, in one climate, mostly on small and medium plants — somebody running larger plants elsewhere may have good reasons to disagree with all three preferences. And the paragraph about how a plant responds to a direct dose is observation, not measurement. If somebody has measured it properly I would rather read that than defend my hypothesis.

If your pH is drifting and nothing on the site has changed, the useful next step is an alkalinity reading rather than a chemical order. If you are still at the stage of working out what a plant should cost at all, our STP price estimator will give you an indicative figure from your flow.

If you want a second pair of eyes on a plant that is doing this, or on a design before it is built, book a free 15-minute consultation.