The month the tank runs dry#
SLOP4761 · Week 9
Water security and storage
You have two curves#
Supply, from week 7. Twelve rows, in kilolitres.
Demand, from week 2. Twelve rows, in kilolitres.
Storage is what turns one into the other.
The balance#
V(t) = min( capacity, V(t−1) + inflow − demand )
Three-year spin-up. Report year three.
Otherwise you are reporting the volume you guessed.
Three rules#
- Domestic first. A shortfall is carted in, and carted water is an import.
- Garden gets greywater, then the tank. A garden shortfall lowers garden water satisfaction, not water closure.
- Full tank overflows. Overflow is reported. You cannot spend it.
Darwin, on paper#
| On paper | Volume |
|---|---|
| Roof yield | 411 kL |
| Domestic demand | 219 kL |
| Annual surplus | +192 kL |
Comfortable. Obviously.
Darwin, by the month#
| By the month | Volume |
|---|---|
| Storage | 76 kL |
| Imported | 48 kL |
| Overflowed | 245 kL |
1.4 %#
Darwin’s share of annual rain falling in its driest four months.
Throws away 245. Buys 48.#
In the same year.
With a surplus, on paper, of 192 kL.
The annual balance is not imprecise here. It is the wrong question.
So build a bigger tank?#
Test it. Step the spend and plot the curve.
$4,000 → $9,000 buys 38 more kilolitres.
Water closure moves by 0.06.
And that is Darwin, the best case#
At Canberra, Alice Springs and Adelaide the curve is flat.
$2,000 or $10,000 — the same water closure.
Those sites are supply-limited, not storage-limited.
A tank cannot hold water that never fell#
Canberra’s roof: 142 kL.
Its household alone: more than that.
Every tank empties, every year, at any size.
A flat curve is information.
It says the money belongs somewhere else.
This course got it wrong#
The reference designs used to spend $9,000 on storage at every site.
Costed against the full schedule: $3,704 to $9,545 over budget.
Past the knee at four sites out of five.
Find your own knee#
Step the tank in 10 kL increments.
Plot the indicator against capacity.
Spend where it is still rising. Stop where it is not.
Then break it#
The average year contains no droughts.
A design tuned to it is tuned to a year that never happens.
The dry-year dataset#
Synthetic. Scaled from your site’s Bureau statistics toward a low decile.
Not a historical drought. Never describe it as one.
Its job is to make your design fail somewhere visible.
What leaves this tutorial#
A sized storage system.
The month it empties, named.
Volume imported. Volume overflowed.
The same figures re-run on the dry year.
A design that never fails on the dry year#
has not been tested hard enough.