We Drew a Partition Inside a Sealed Tank. Then We Measured the Manhole.

We were asked to make an existing plant comply. Not to build one — to take a set of sealed fibreglass vessels that were already standing in a plant room, already piped, already full of water, and get a result out of them. The process we needed was clear enough on paper within about an hour. It needed one of those vessels split into two duties, with a partition across it.

So I drew the partition. Then somebody asked how we would get it in there, and I went and looked at the manhole.

You can build almost anything inside a concrete tank while the tank is still open to the sky. We have written at length about what we will and will not put inside a buried concrete chamber, and about how that list reopens when the tanks sit above ground. Both of those are about a tank we are building, where the question is whether a fitting can be reached and serviced later. This is a different question and a harder one. The tank already exists, it is a moulded sealed vessel, and the question is not whether a man can service the partition. It is whether a partition can be got in and bonded at all, through an opening of a fixed diameter, by a person who has to fit through the same opening.

The opening is a design constraint, and it is the first one to check

An internal partition in a sealed cylindrical vessel is not a plate you post through a hole. It is a structure that has to be cut to the curve of the shell, held in place, bonded along its whole perimeter, and then laminated over on both faces by somebody working inside the vessel with resin, in an enclosed space, with no second chance if the bond line is wrong. The panel has to pass through the manhole. The materials have to pass through the manhole. The person has to pass through the manhole, work in there, and come out. On most of the moulded vessels we meet, the manhole is sized for a man to climb through and nothing more.

A partition can be made to fit through in pieces and joined inside. We have seen it done. What we were not willing to do is write it into a scope of works and a price on the strength of having seen it done, on a plant that has to pass a test afterwards. A bonded joint inside a vessel nobody can inspect properly is the kind of item that holds for the commissioning period and fails in year two, which is the same after-sales reasoning we apply to every piece of equipment we specify.

So the rule we now carry is simple and it has nothing clever about it: on a retrofit into existing vessels, measure the access opening before you draw anything inside the vessel. Not after. The opening diameter belongs in the survey record next to the tank volumes, because it decides which designs are available to you. Working the other way around — drawing the process you want and then discovering the tank will not accept it — costs you the drawing and, if you have already quoted, costs you worse than that.

The fix was to stop subdividing and start assigning

Once the partition was off the table, the design got better rather than worse, which happens more often than I expect it to.

There were six vessels. Instead of two duties inside one vessel, each duty got a whole vessel. The process we wanted was available the whole time; it was just distributed across the tanks that existed instead of across partitions that did not. Nothing had to be cut, nothing had to be bonded, nothing had to be laminated by a man inside a cylinder. The scope became pipework and controls, which are things that can be installed, inspected and corrected from outside the tank.

That is worth saying plainly because the instinct on a retrofit is to try to recreate the drawing you would have produced for a new build. A new build and a retrofit are genuinely two different jobs, and the retrofit one is mostly an exercise in what the existing concrete and the existing vessels will permit. The volumes are the constraint you cannot argue with. The arrangement is the part you are allowed to be inventive about.

The sludge holding went outside the plant room, in a poly tank

The arrangement still needed somewhere to hold sludge. The obvious move was to give up one of the six vessels for it, or to find the volume inside one of them, which brought the partition problem back through the door.

We put it outside instead: a small above-ground polyethylene tank, standing on a slab outside the plant room, with the draw line run out to it. No civil works, no vessel modification, nothing cut into anything that currently holds water. And a second benefit we did not design for and will take anyway — the sludge now sits outside the enclosed space rather than inside it. Everything you remove from a plant has to go somewhere, and that destination is a drawing item. A tanker hose reaches an above-ground poly tank on a hardstanding in a way it does not reach a sump under a floor.

It is not elegant. It is a plastic tank on a slab next to a building, and nobody will photograph it. It is cheap, it is reversible, and it keeps a routine task out of a space that needs gas detection and a second person at the hatch every time somebody opens it.

Then we found the tanks were in parallel, not in series

This was the finding that changed the price, and it was not visible in anything we had been sent. The aeration vessels were piped in parallel. Influent arrived at a manifold and was shared between them, so every vessel was seeing full-strength raw wastewater at its inlet.

A parallel arrangement is not a mistake in every application. It is the wrong arrangement for what this plant had to achieve, because it destroys the substrate gradient. In a sequential train, the first tank sees the strongest water and does the bulk of the work, and each tank after it sees progressively weaker water, which is the condition the slower organisms need in order to get a look in. Share the raw load equally between four tanks and you have four tanks all doing the easy part and none of them doing the last part. It is the same reason ammonia is the parameter that fails while BOD passes comfortably: the organisms that remove it are the ones that need conditions arranged for them rather than merely adequate.

So the arrangement had to be repiped from a parallel manifold into a sequential train. That is a named scope item with a real cost, and we put it in the document as its own line rather than folding it into a general allowance for pipework. Every bidder on that plant should have been pricing the same repipe, and an owner comparing two quotations has no way of knowing whether they are looking at the same scope or two different ones unless the line is written down and visible.

Nothing wanted to flow downhill

Here is the part that took the longest, and it is the part I would most want another engineer to check.

A sequential train needs water to get from one vessel to the next. On a plant built as a train, that is arranged at design stage by setting each tank a little lower than the one before it. These vessels were all standing on the same floor, at the same elevation, because they had been installed as a parallel bank where it never mattered. There was no natural driving head anywhere in the arrangement. Water does not move through four tanks in sequence because you have drawn an arrow between them.

What we had to work with was freeboard. Each vessel is about 2.05 m in diameter, holding water to about 1.70 m, which leaves roughly 0.35 m of spare height above the water surface. That 0.35 m is the entire hydraulic budget for the whole train, and it has to be divided between every transfer in it.

Running the first vessel about 150 mm higher than its neighbours gives roughly 50 mm of fall per transfer across the train. Fifty millimetres is not much. The question is whether it is enough, and that is arithmetic rather than opinion, so we did the arithmetic rather than assuming it would be fine.

Fifty millimetres is enough for the forward flow and nowhere near enough for the recycle

Forward flow through the existing four-inch transfer pipes works out at roughly 15 mm of head loss. Against 50 mm of available fall, that is comfortable — not generous, but it works, and it works with the pipes that are already installed, which is the whole point of a retrofit.

The internal recycle is a different answer entirely. Recycle runs at a multiple of the forward flow — in this arrangement around three times it — and head loss does not scale with flow, it scales with roughly the square of it. Three times the flow through the same pipe is close to nine times the loss. The number came out at roughly 240 mm against 50 mm available. Not marginal. Not something you recover by cleaning the pipe or opening a valve further. It simply cannot be done on gravity through that pipework.

So the recycle gets a pump, and the recycle line gets upsized. Both of those are items in the quotation, and both of them exist because of a calculation rather than because somebody prefers pumps. That is the useful part of writing the arithmetic down: a pump that is in a quotation because 240 is bigger than 50 is a pump the owner can interrogate, and a pump that is there because the supplier always includes one is not. It is the same reason we publish where our flow number comes from and why the design rather than the build is what decides the cost — if the working is visible, the other quotation can be asked for its working too.

I will admit the thing I nearly got wrong. A 3Q recycle on gravity looked plausible to me at a glance, because the forward flow had checked out so easily. It only failed when the number was actually worked out. Forward flow passing a head check tells you nothing at all about the recycle passing one, because they are not the same flow and the relationship is not linear. On a plant with 0.35 m of freeboard to spend, a glance is not a method.

What we ended up selling was mostly arithmetic

The deliverable on that assessment was not a new plant. It was: use these six vessels, one duty each; repipe the manifold into a train; raise the first vessel; keep the existing four-inch transfers; pump and upsize the recycle; put the sludge in a poly tank outside. Almost none of it is equipment. Most of it is pipework, levels and an arrangement, and the reason it can be stated with any confidence is that volumes, diameters and elevations are facts you can measure and then compute with.

Equipment counts and operating histories on a plant you have not stood in are reports. The geometry is not. On another assessment we found an installed air capacity that looked more than adequate on the nameplates and was not being delivered at all, and the only part of that finding nobody could dispute was the tank volume arithmetic. Where we are handed somebody else's design, or asked to assess something already built, we anchor the findings on the measurements and treat everything we have been told as needing confirmation. That is also why a site visit is a gate on a quotation rather than a courtesy — the manhole diameter, the vessel elevations and the freeboard were all things one person with a tape measure settled permanently, and none of them were in any document we had been sent.

Where this does not transfer

Most of this is specific to a plant with sealed moulded vessels standing at one level, and plenty of plants are not that.

If the vessels are concrete and still open, a partition is ordinary civil work and none of the manhole argument applies — that is simply a wall drawn before the cover slab goes on. If the existing tanks sit at staggered elevations, the driving head is already there and the freeboard arithmetic is a formality. If there are two or three vessels rather than six, whole-tank-per-duty is not available and you are back to either a partition or a new tank. If the forward flow in your case is large, 15 mm of loss through a four-inch pipe will not be your answer either; our number is our pipe at our flow, and yours will be different.

And if the required process genuinely needs more stages than the site has vessels, none of this helps. We were lucky in the specific sense that the number of tanks happened to match the number of duties. On the next one it will not, and we will have to argue about a partition again — only this time we will measure the opening first.

The transferable parts are smaller than the story and probably more useful: measure the access opening before you draw anything inside an existing vessel; check whether a tank bank is piped in parallel or in series before you assume the arrangement matches the drawing; and do the head loss on the recycle separately from the forward flow, because the recycle is the one that fails.

If you have an existing plant that has to start complying and you are not sure whether what you have can be rearranged into something that works, book a free 15-minute consultation. You can also see how we approach wastewater treatment design, read the DENR discharge standards your result will be judged against, check the parameters that apply to your industry, get an indicative figure from the STP price estimator or the design generator, look at plants we have finished, start a scope with us on the get-an-STP page, and see how we handle operation and maintenance and ongoing support once a plant is running. If the plant is in Cebu, that is our home region.