Programme and method

The fictional loan programme and the consultancy, the two phases of the commission, and then the reference material every design is measured and costed against.

Teaching fiction. The Household Resilience Loan Programme and Common Ground Engineering do not exist. They are a frame for the engineering: they give every design a client, a deadline, a budget and a reviewer. The government is imaginary. Your water balance still has to add up.

The scenario

A government programme in this course's fiction lends to homeowners for retrofits that make a household more self-reliant in food, water and waste. Loans are selective: an application stands or falls on an engineering proposal showing that the retrofit meets the household's own service needs, costs what it says, and will still be working in five years.

You are a graduate engineer at Common Ground, a small consultancy that prepares those applications. Each graduate takes one household from intake to proposal under a senior reviewer, played by the teaching team, who reads your drafts the way a programme assessor will. Nobody applies for a real loan, and nobody waits a real year.

The two phases

1. Weeks 1 to 6 — work out what you would need to know. The homeowner supplies an intake dossier: the property as it stands, what they already own, what they will and will not allow, and the service targets they care about. You work through each candidate system far enough to know what it would need, and you produce a costed twelve-month measurement programme. That is Assignment 1, due in week 7.
2. Week 7, then weeks 8 to 12 — decide, and argue for it. Assignment 1 goes in on the Friday. The story then jumps twelve months and the specialist team hands over its analysed findings for your home. You compare the existing baseline with three feasible retrofit options and recommend one. That isAssignment 2.

The specialists' measured year is synthetic, generated to be consistent with each home and labelled that way everywhere it appears. It is the same for every student who took that home, whatever their Assignment 1 said. Specialists analyse; they never choose the system.

What you produce, and what you do not

What an application gets from you is the technical annex and executive recommendation. It is not a loan approval and it is not a construction-certified design: pipe sizes, wiring, structural detail and certification belong to licensed trades later. You specify performance and how to check it has been achieved. There are no interest rates, credit scores or income tests anywhere in this course, and nothing here is financial advice.

The programme's rules

Borrowing more is allowed up to the limit, but a larger loan is a larger exposure, so the programme asks a stronger question of it: what does the extra money buy that a cheaper option does not? That is an engineering question about alternatives, and it is the question Assignment 2 answers.

The reference material

Everything below this line is reference. Come back to it all semester.

• What is measured — the five capped indicators and their definitions
• Reference bands — where the per-site thresholds come from, and what they do not establish
• How stable the bands are — the sensitivity testing, published with its limits
• The screening model — what it contains and what it leaves out
• Measurement primer — instrument classes, units, frequencies and honest accuracies
• Costing and energy — the price schedule, and which loads count
• Worked comparison — one design idea through all five sites
• Datasets — the synthetic data, and what it is not

What is measured

Five indicators, all capped at 100 %, all counted as used rather than produced, and all net of imports. That last rule is the one that does the work: a system running on bought-in feed or carted water scores as what it is.

Water closure never appears alone

Water closure is on-site water used divided by all water used. Water that is never used enters neither term — so a design can post a very high closure figure by leaving its garden dry. Alice Springs' excellent reference design reaches 96 % water closure with 16 % garden water satisfaction. The two are published together everywhere, without exception.

Waste-stream recovery is deliberately not banded

Every banded indicator takes its thresholds from three constructed reference designs run through the model. There is no reference calculation for waste-stream recovery, so a band for it would have nothing behind it. It is assessed from your submitted mass balances against the rubric instead. Saying so is better than inventing a number and calling it a measurement.

Reference bands, and what they borrow from NatHERS

NatHERS could not use one national energy target across Australian climates because a single figure "would bias results", so it sets climate-specific star bands. This course borrows the idea and states plainly that it is not the method: NatHERS derives its bands from a population of housing simulations, while these come from three constructed reference designs — baseline, competent and excellent — run through the course's screening model at each site.

The reference recipe, identical at every site

One recipe, five climates. Everything that differs between sites' thresholds comes from the climate and the site inventory, never from a different design being used as the yardstick.

Your site's thresholds

Each cell reads baseline / competent / excellent, as percentages.

Rounding. Bands are evaluated against the model's unrounded reference values. The percentages above are a rounded display of those values, not the thresholds themselves — a design at 5.51 % against a threshold displayed as 5.5 % is above it.

How stable are the bands?

The three reference designs were re-run under eleven perturbations of the model's most uncertain constants — reference yield at ±33 %, domestic demand at 120 and 180 L/person/day, runoff coefficient, crop coefficient, feed conversion ratio and tank cost. For each, the question is where the competent design lands inside its site's baseline-to-excellent band. If the method is fair, that position should be similar at every site.

Read that honestly. Food energy and protein are stable. Garden water satisfaction is the least stable indicator in the set, and both it and water closure have their worst case under the same perturbation, domestic demand at 120 L/person/day. Excluding that one case, no indicator spreads by more than 0.17.

Garden water satisfaction was only brought inside this test in September 2026. Before then the stability claim covered four indicators and was quietly assumed to extend to the fifth. It does not, and the figure above is what that assumption was hiding.

What the bands do not establish

• They are not grades. A band is performance evidence inside the marking criteria, contributing to Calculations and Loop integration. Nothing converts a band into a mark.
• They do not prove equal grading outcomes. The testing above shows three constructed designs behaving consistently. It says nothing directly about how real student designs, which vary in ways the model does not represent, will distribute.
• Narrow bands are noisy. Alice Springs' food energy band spans about three percentage points end to end. That is why bands are ordinal — four levels — rather than a continuous score.
• They must be recomputed whenever a course constant changes. Household demand, allowances, the price schedule, model coefficients or the reference recipe: any of them, and every threshold on this page is republished.

The screening model behind all of it

Monthly, transparent, and built only from published formulas so that a student can reproduce any number by hand. It has no drought years, no pests, no soil types and no labour. Water temperature is approximated by air temperature. Food yield rests on a single reference figure of 1,500 kcal/m²/year inside a published range of 500 to 3,000.

It is a screening model, not a validated engineering design, and its outputs are not predictions of real yields. Everything it produces on this site is labelled with its basis.