Inlet flow is metered to the litre a second and the strength of it is sampled and tested, because both numbers are a permit condition. At ten past three the flow rises by eleven litres a second, and the ammonia in it is the ammonia of a town waking up.
I am a process operator at a water recycling works. Fourteen years, mostly nights.
What gets measured, because the whole of this is two numbers.
The inlet flow goes through a flume with a level sensor over it, which gives litres a second continuously, and that figure is logged and reported. Our dry weather flow is about ninety litres a second at three in the morning and about four hundred at half eight.
The strength is sampled by an autosampler, which takes a measured aliquot at set intervals into a fridge, and those bottles go for analysis: ammonia, biochemical oxygen demand, suspended solids.
Between the two you know how many people are upstream and what they are doing. A catchment has a shape to its day that you could draw from memory: the morning peak, the lunch bump, the evening, the long decline to about four in the morning.
At ten past three the flow rises by eleven litres a second for about forty minutes, and then it is gone.
Eleven litres a second for forty minutes is about twenty six cubic metres.
Twenty six cubic metres at three in the morning is not nothing and it is not a flood. It is the waste water of about four hundred people getting up.
That is not a guess. The autosampler catches it, because I reprogrammed the autosampler to catch it, and the bottles are in the log.
The ammonia in that forty minutes is at the concentration of morning sewage, not the weak dilute stuff you get from a catchment asleep, and the suspended solids match.
What it is not, and these were all checked by people senior to me between 2016 and 2019.
Rain. It happens on dry nights and wet ones, and a rainfall derived infiltration surge is dilute, which is the opposite of this.
A pumping station kicking in. This was everybody's answer, and it is a good one: a station with a tank on it will collect all night and pump on level, and what you see at the works is a slug. Every station in this catchment telemeters its pump starts and its flows, and the telemetry has been pulled for thirty one of these nights.
Nothing pumps at ten past three. The combined flows of every station upstream are flat.
A tanker discharge. We have a tanker point and it is locked, metered and on camera, and trade effluent tankering is booked in. No bookings at that hour, ever.
A trade discharge. There are four consented trade premises on this catchment and none of them runs a night shift, and I have been round all four and met the people.
What the catchment is.
Two villages and the edge of a town. Population about nine thousand on the figures we are given, and those figures come from the council and the water company's own billing and they agree with each other.
Nine thousand people produce the day I have described. The morning peak is nine thousand people.
Four hundred of them, an hour and a half before anybody else, every night, is a population that is not in anybody's numbers.
What happened when I tried to find them.
In 2018 the company put a temporary flow monitor in a manhole at the village end of the main sewer for three weeks, as a favour to a supervisor who believed me.
The surge is there, at the same time, the same size.
In 2019 they moved it a kilometre further up, to the last manhole before the sewer comes out of the old part of the village.
It is there too. Same time, same volume, and arriving from upstream.
Upstream of that manhole is a hundred and forty houses and a chapel.
A hundred and forty houses is about three hundred and fifty people, and if every single one of them got up at ten past three and used the toilet and a tap, it would look like this, which I know because I did the arithmetic twice.
What the paperwork shows.
It is in the works log as a recurring nocturnal flow event, unattributed, and it has been in the annual process review three times.
The 2020 review says the event does not breach any permit condition, does not affect final effluent quality, and that further investigation is not justified on cost grounds.
That is correct. We treat it. It goes through the works, it gets treated, the final effluent is inside consent, and the regulator has never had cause to look.
What I do.
I am on the plant anyway, so it is one line in the log and a sample bottle.
The thing I changed is that I do the two o'clock round in one direction and the four o'clock round in the other.
The works is laid out in a line: inlet and screens at the top end, then the tanks, and the final effluent and the river outfall at the bottom. You walk down and back up.
The screens are at the inlet and that is where what comes in is visible, because a screen is a rake lifting the solids out of the flow onto a belt.
In 2021 I was at the screens at about twenty past three, which I had been avoiding by accident for a couple of years without really noticing, and the flow was up and the rake was working.
What comes off a screen at the morning peak is what you would expect, and what comes off it at three in the morning is usually almost nothing.
It was a normal morning's worth. Wipes, paper, the ordinary refuse of a town getting ready for work, going up the belt in the dark with nobody in any of those houses awake, and I stood and watched it for a while and then I went and did the bottom of the works and came back up at five.