One fault, three bills, and nobody who owns it

Two meters that should agree and do not

Zarqun measures water twice. The intake meter at the gate read 58,940 cubic metres in the year to 30 June 2026. The outfall meter on the effluent plant read 46,120.

The gap is 12,820 cubic metres, or 21.75% of everything drawn.

Most factories stop here, and the two available conclusions are both wrong. One is "we have a huge leak", which sends a fitter round the site for a fortnight. The other is "water evaporates, this is normal", which is true and explains nothing.

Rasha did the third thing. She wrote down every term until the leftover was small enough to be interesting.

The stenter is the long oven that dries and sets the fabric to width. The gap is not the finding. The residue is. 1,180 cubic metres is 2.00% of intake, and it is the only number on that table that nobody can explain.

The loss the meter is blind to

1,180 cubic metres a year is 0.13470 cubic metres an hour. Lesson 3 has already shown that no comparison window shorter than 9.7 years could prove a loss that small against this site's variability.

But there is a second and worse problem, and it belongs to the instrument rather than to the arithmetic.

A water meter is specified with a minimum flow rate, below which its error limits do not apply, and a starting flow, below which the mechanism does not turn at all. Below that threshold the meter does not read approximately zero. It reads exactly zero, forever, while being fully within specification and correctly calibrated.

So the night-flow test — read the intake meter at 22:00 on a Sunday and again at 04:00 with nothing running — gives 0.00 cubic metres. Everybody concludes there is no leak, and the conclusion is drawn from a reading that could not have said anything else.

The leak was a cracked underground main between the intake meter and the softener house. It was found by digging up the one section where the balance had been drawn twice, once with a valve shut. It cost USD 1,109.20 a year in raw water, and it never reached the effluent plant, which is exactly why the outfall meter could not see it either.

USD 1,109.20 is not much money. The method is worth having anyway, because the same balance drawn monthly is what catches the ten-cubic-metre-an-hour version on the day it starts.

The fault that is on three bills

Now the large one, and it hides in a completely different way.

Zarqun's boiler turns 6,259 cubic metres of water into steam in a year. Only 2,879 of them come back to the feed tank as condensate — the hot water steam turns into once it has given up its heat — which is 46.0%. The rest leaves as vapour: a failed trap group on the stenter, and a return line put on manual dump after a heat exchanger leaked dye liquor into it two years ago and was never put back.

That is the 3,380 cubic metres in the balance above. Now price it.

The heat term is where the money is, and here is how it is built, so it can be checked. Condensate returns to the feed tank at 88 degrees Celsius. Make-up water arrives at 24. Replacing 3,380 cubic metres therefore means heating 3,380,000 kilograms through 64 degrees, which is 251,532.09 kilowatt-hours of delivered heat. Zarqun's own gas meter and heat balance give it 1.19 kilowatt-hours of gas bought for each kilowatt-hour of heat delivered, and it buys gas at USD 0.0412 a kilowatt-hour. So the gas is 299,323.19 kilowatt-hours and USD 12,332.12.

A condensate problem is a fuel problem wearing a water label. Everybody in the factory calls it a water issue, and the water is 18.92% of it.

Now the reason it survived two years. The water line appeared on the facilities budget, where it was inside normal variation. The chemical line appeared on the dyehouse lab's ledger, where it was inside normal variation. The gas line appeared on the boiler house's fuel account, where it was inside normal variation. Three departments each saw a third of a fault, each correctly judged their third to be noise, and no report anywhere added them up. Nobody was wrong and nobody owned it.

The repair — a trap survey, six trap elements, and putting the dump valve back under control — was USD 2,140.00.

The saving that landed on another meter

Muntaha Barghash in the dyehouse ran a water-reduction project that worked exactly as designed. The final rinse of a pale shade is held and used as the first rinse of the next lot. Over a year it took 4,180 cubic metres out of both the intake and the outfall.

Zarqun's effluent authority charges a volume rate and, separately, a rate on the chemical oxygen demand discharged — a measure of how much pollutant the water carries. The rinse re-use removed water and removed no pollutant at all. The same load now leaves in less water. Concentration rose from 4.6531 to 5.1168 kilograms a cubic metre, or 9.97%, which took the biological stage outside the range it was set up for. Tahseen Farhat had to dose harder and run the aeration longer, and the plant went over its permit twice, which brought a re-test regime with it.

A water saving in the dyehouse became an electricity and chemical cost in the effluent plant, and the two are on different budgets. The project was right. The arithmetic that was never done is that the treatment plant sits downstream of every water decision on the site, and is the one place where all of them arrive together.

Re-tuning the plant — a longer anoxic stage and a revised dosing curve — cost USD 4,400.00 once and brought the extra running cost down to USD 1,940.00 a year. The steady-state net is USD 4,539.00 a year, positive, eleven months after it looked like a failure.

What you should be able to do now

1. Build a water balance with every term written down, and treat only the residue as a finding.
2. Repeat it four times before you believe it, because an estimating error changes sign.
3. Know your meter's starting flow, and stop treating a zero night-flow reading as proof of anything.
4. Price a utility fault across every ledger it lands on, and expect the largest share to be on a bill nobody associates with it.
5. Ask, of any water-reduction proposal, what it does to the concentration arriving at the treatment plant.