How these figures are derived

Anyone can publish a calculator. The question is whether its output is right, and almost nothing in this market gives a reader any way to check.

Sustainability Victoria publishes measured annual running costs for a Melbourne household — for water heating, for space heating and for cooling. It is the closest thing to an independent benchmark in Australia. The constants in both our models were fitted to reproduce it, across twenty-two separate published figures, and almost all of them land within two per cent.

Water heating

Space heating and cooling

Sixteen more published figures, across three house sizes and every common system. The table below is generated from the model every time the site is built, so the residuals in it are the residuals the build gate sees — not a number anybody typed.

The comparison is not a claim we made once. It runs as a build step (tools/build/check-model.js) every time the site is rebuilt, and the build fails if any residual drifts past five per cent. A future quarterly edit to an efficiency or a standby loss cannot quietly break the agreement this page asserts.

Space conditioning is harder to model than hot water, because the load depends on the building rather than on a fixed volume of water. The method is deliberately simple enough to state in full and check.

How much the house needs

An average pre-2005 Melbourne house needs about 63.2 kWh of delivered heat per square metre per year and 7.3 kWh of cooling. Both fall out of the benchmark fit, and everything else scales off them.

That ratio is the most surprising thing in this module: in Melbourne, heating is nearly nine times cooling for the same house and the same hardware.

Climate

Scaled by heating and cooling degree days on a base of 18 °C, derived from the same Bureau of Meteorology monthly mean temperatures the hot water calculator uses for inlet temperature — so the two modules can never disagree about the weather. Melbourne has 1,239 heating degree days and 173 cooling; Darwin has none and 3,508.

Cooling is scaled by the degree-day ratio raised to the power of 0.8 rather than linearly. A tropical climate needs latent as well as sensible cooling and the raw ratio overstates the difference; the exponent reproduces Sustainability Victoria's own warm-climate multiplier of about 6 within a few per cent.

Building fabric

Sustainability Victoria's own multipliers against a pre-2005 house: a 5-star home needs 40% of the heating energy and 75% of the cooling; a 6-star home 30% and 60%. Insulation helps heating far more than cooling, so a single "efficiency" figure for both is always wrong about one of them.

What the system actually conditions

A ducted system delivers 1.29 times as much conditioned energy as a multi-split in the same house; a single split about half. One factor covers both duct losses and the fact that ducted conditions the whole house while a split conditions a room.

Modelling those as two separate multipliers double-counts. An earlier version of this calculator did exactly that and came out 25% under the benchmark on every multi-split row while matching ducted perfectly, which is how the error surfaced. The 1.29 was derived independently from the heating table and the cooling table and came out the same both times.

System efficiency

Coefficient of performance for heating and energy efficiency ratio for cooling, from the star rating on the label: 2.30 at one star rising to 5.30 at six. A resistive electric heater is exactly 1.0 and cannot be otherwise. Ducted gas is modelled at 78% delivered, not the 86% on the compliance plate — the plate figure is measured at the burner and 78% is what the benchmark implies once the ducts are counted.

What this does not model

• Zoning. A zoned ducted system used properly behaves partway between the ducted and multi-split rows. Nothing here models how disciplined a household is about closing vents.
• Orientation, shading and glazing. These move cooling load substantially and are not captured by floor area and star rating alone.
• Humidity, except through the cooling exponent. Latent load is approximated, not calculated.

Heat required

Raising water from the inlet temperature to the stored temperature takes a fixed amount of energy, and one litre of water is taken as one kilogram:

Q (kJ) = litres × 4.186 kJ/kg·°C × (60 °C − inlet temperature)
Q (kWh) = Q (kJ) ÷ 3,600

This is integrated month by month rather than annually, because the water entering a house is several degrees colder in July than in January. Averaging that away costs a few per cent of accuracy in the southern states.

Why 60 degrees

Below about 60 °C, Legionella bacteria can multiply in a storage tank. Australian plumbing standards require storage at 60 °C with a tempering valve limiting the temperature at the tap to 50 °C. It is a health requirement rather than a comfort setting, which is why the calculator does not offer it as an adjustable input on storage systems.

Daily hot water demand

50 litres a day for one person, 90 for two, 120 for three, 150 for four and 180 for five or more, at 60 °C. These are Sustainability Victoria's figures, used here so that our output stays comparable with theirs. The calculator lets you override the litres directly.

System efficiencies

These are fitted to the benchmark above rather than taken from manufacturer literature.

Heat pump coefficient of performance

This is where the model differs most from the rest of the market, so it is worth setting out fully.

The nameplate COP is multiplied by two factors. The first is a temperature adjustment applied month by month against each capital's mean air temperature, running from 0.65 at 0 °C to 1.12 at 35 °C. The second is a flat in-service derating of 0.74, covering compressor cycling, standing losses in the pipework, control deadbands and the difference between a steady laboratory draw and a household taking four showers in an hour.

That derating is not an assumption we chose. It is the value that makes the model agree with Sustainability Victoria's measured Melbourne figure, and it produces an effective annual COP of about 2.2 in Melbourne against a nameplate of 3.2. A calculator that uses the nameplate figure understates heat pump running costs by roughly a third.

Solar thermal fraction

The share of the annual load a correctly sized collector supplies before boosting: Darwin 82%, Brisbane 75%, Perth 72%, Sydney 70%, Adelaide 68%, Melbourne 62%, Canberra 60%, Hobart 55%. Derived from each city's STC zone and annual mean temperature and cross-checked against the 45–80% range in the literature.

The gas daily supply charge

Hot water carries a share of the gas connection charge depending on what else the household uses gas for: 100% where hot water is the only gas appliance, 50% where gas is also used for cooking or for heating, and 34% where it is used for both.

This single term moves the answer by up to $300 a year and reverses the ranking between a heat pump and gas instantaneous in most states. It is the most commonly omitted term in every competing calculator we examined.

Solar self-consumption

Where a household has rooftop solar, a heat pump is assumed to be timed so that 70% of its run falls in daylight. That electricity is not free and is not charged at the retail rate: it costs the export credit given up, taken as 4c per kilowatt hour. Feed-in tariffs are deregulated across most of Australia in 2026 and sit near 3 to 5 cents.

Total cost of ownership

The ten and twenty year views add the installed cost, subtract rebates at the bottom of their published range, and add a replacement where the system will not survive the horizon. No energy price inflation is applied by default: compounding a guess about future prices makes a calculator look precise rather than making it accurate.

AS/NZS 4234 is the Australian standard for water heater modelling and carries the authoritative inlet temperature series. It is paywalled, so it cannot be reproduced here. Rather than quote figures we cannot show you the derivation of, the site uses a stated physical assumption you can check:

Buried mains equilibrate toward soil temperature at depth, which tracks the annual mean air temperature. Inlet temperature is therefore set to each city's long-term annual mean from Bureau of Meteorology climate averages, with monthly variation damped to 40% of the air temperature swing and lagged one month.

That puts Melbourne's inlet between 13.0 °C and 17.2 °C through the year, with an annual mean of 15.1 °C, and Darwin's between 26.5 °C and 28.3 °C. Inlet temperature is a user-overridable input in the calculator's advanced settings, and the fact that the model reproduces Sustainability Victoria's Melbourne figures is itself evidence that the assumption is close enough.

Government reference rates effective 1 July 2026. These are what a household on a standing or default offer pays; market offers from retailers are frequently below them, which is why every rate on this site can be overwritten with your own.

The determinations behind each row are the AER Default Market Offer 2026–27 for New South Wales, south-east Queensland and South Australia; the ESC Victorian Default Offer 2026–27; the WA Government Standard Electricity Prices 2026–27; the OTTER-approved Aurora Energy standing offer for Tasmania; the ICRC retail price recalibration for the ACT; and the NT Electricity Pricing Order.

Gas

Residential gas retail is deregulated in New South Wales, Victoria, Queensland and South Australia, so there is no determination to quote. Those figures are mid-2026 retail market averages, cross-checked against the published rates of the three largest retailers. Western Australia's is the regulated maximum set by Energy Policy WA. Tasmania's and the ACT's are proxies, and are labelled as such in the table.

LPG is taken at $167 per 45 kg bottle holding 2,172 MJ, which is about 7.6 cents a megajoule — the dearest common fuel for water heating in Australia by a wide margin.

Scope 2 electricity emission factors and fuel combustion factors are taken from the National Greenhouse Accounts Factors 2025, published by the Department of Climate Change, Energy, the Environment and Water, which apply to the 2025–26 reporting year.

Natural gas distributed in a pipeline is 51.53 kg CO₂e per gigajoule (0.05153 kg/MJ); LPG is 60.6 kg CO₂e per gigajoule.

One consequence is worth stating because it is uncomfortable and because a site that hid it would not be worth reading: on today's grid, a gas instantaneous water heater produces fewer emissions than a heat pump everywhere in Australia except South Australia and Tasmania. That reverses as the grid decarbonises, and it does not apply at all to a heat pump running on the household's own solar — but it is true now, and the gas versus electric guide sets out the numbers.

Every rebate is shown as a separate line item with the administering body named and linked, and never folded into a headline price. In the total cost of ownership figures they are counted at the bottom of their published range.

Rebate values are indicative. The figure that counts is the one on an accredited installer's quote, and the schemes change without much notice — Victoria's income test arrived on 1 July 2026 and the New South Wales co-payment rule changed in September 2025. Every entry carries its own verification date.

Duty-cycled appliances — refrigeration, spas — carry measured annual consumption figures from the Energy Rating (GEMS) scheme, and the calculator does not offer an hours-per-day input for them, because substituting a guess for a measurement makes the answer worse rather than more personal.

Directly switched appliances are modelled as rated power × hours a day × days a year. The days-a-year term matters: a fan heater at five hours a day over a 120-day heating season costs about a third of what the same figure spread across 365 days would suggest, and published appliance figures routinely make that error.

Appliance figures are typical values for a mid-range current model and are adjustable throughout. The daily supply charge is excluded from individual appliance costs, because it does not change with what you plug in.

Pumping: the strongest source on this site

Since 2022 every swimming pool pump sold in Australia has carried a GEMS energy label, and the register behind it is published as open data by the Commonwealth. It gives, for all 662 registered products, the labelled annual consumption, the star rating, the weighted energy factor and the daily run time the label assumes.

The labelled figure is exact arithmetic on the weighted energy factor — 50,000 litres divided by litres moved per watt hour, times 365 — and every row on the register reproduces to the first decimal place. This site's pump model is the same arithmetic, so at a pump's own labelled operating point it returns the labelled figure exactly rather than approximately. The build re-runs that check against every pump class and fails on a residual above half a per cent.

Each class here takes its labelled consumption and its labelled run time from the same registered product — the median of the class by consumption — rather than from two independently taken medians, so the pair always describes a pump that exists. The number of products behind each class is published alongside it, because a class fitted from four products is a much weaker guide than one fitted from 229.

Heating: the one fitted constant on this site

This is the weakest number the site publishes and it is worth saying so clearly.

Heat leaves a pool by evaporation, convection, radiation and conduction into the ground, and evaporation dominates. Modelling that properly needs humidity, wind and solar irradiance for every capital, which this site does not carry. So pool heating is modelled as proportional to surface area and to the gap between the water temperature you asked for and the mean air temperature, with a single heat loss coefficient of 45 W per square metre per degree.

That coefficient is fitted, not measured. It was set against two things — the evaporation rates Australian pool suppliers report for summer, and the running costs the pool heating industry publishes — and here is where the model sits against both today, computed rather than remembered:

What follows for how far to trust it. The ranking of heating types is robust, because it rests on efficiency ratios that are not in dispute: a burner cannot exceed 100%, an element is exactly 100%, and a heat pump at a small temperature lift is several hundred per cent. The build asserts that ranking and fails if it ever inverts. The absolute figures carry real uncertainty — an exposed, windy site will do worse than these numbers and a sheltered one better. Every other model on this site is fitted to a published determination or an official register. This one is fitted to industry figures, and it is the only one that is.

What the pool model does not cover

Chemicals, salt, filter media, repairs and the water itself. Water is reported in litres rather than dollars because it is billed by utility rather than by state and this site carries no water tariffs. The evaporation figure the calculator reports is the extra water a heated pool loses over an unheated one, not a pool's total evaporation — an unheated pool evaporates too, driven by humidity and wind, which this model has no inputs for.

Generation: a Commonwealth table

Daily generation per kilowatt of panels comes from the Australian Government's own figures at energy.gov.au, produced with the Australian PV Institute and the UNSW School of Photovoltaic and Renewable Energy Engineering. It is the only per-city table published by the Commonwealth rather than by someone selling panels, and it is used unchanged.

The last two columns are a second Commonwealth figure that disagrees with the first. The Clean Energy Regulator's zone rating is what your rebate is calculated on, and it is deliberately conservative because it is the basis for issuing tradeable certificates rather than a forecast. Both numbers are correct; they answer different questions, and this site keeps them apart.

Self-consumption: the assumption that decides the answer

This is the weakest link in any solar payback, including this one, and it is worth being precise about what is assumed.

Panels generate usefully from roughly seven in the morning to five in the afternoon. Whatever share of your annual consumption falls inside that window is the ceiling on what solar can displace without a battery. The model takes that share from a question — whether anyone is home during the day — at 28%, 40% or 52%, adds the part of any load you say you could shift that is not already inside the window, and treats the result as the ceiling. Generation above it is exported.

The words "not already inside the window" are doing real work there. The day pattern applies to your whole bill, hot water and pool pump included — so if you are home some days, the model already has 40% of your hot water running in daylight. Crediting the whole of that load again when you tick the box would count the same kilowatt hours twice. An earlier version of this model did exactly that, and reported the shifting benefit about 65% higher than it should have been. The calculator now shows you how much of what you ticked is actually being moved. That is still a simplification of an hour-by-hour problem. What it gets right is the shape: self-consumption falls as the system grows relative to the household, and rises with anything moved into the middle of the day. Both of those drive the decision, and both are checked on every build. What it cannot capture is the detail of a particular household's load curve — nor, being an annual-total model, whether the sun is actually shining at the moment the load runs, which will make the real figure lower again on a cloudy winter week. Treat the percentage as an indication and the ranking as solid.

It is also deliberately conservative. The model puts a typical 6.6 kW system on an average household at roughly a quarter of generation self-consumed. Figures of 30% to 40% are commonly quoted; most of them come from people selling systems.

What the payback does not do

It does not escalate electricity prices, and it does not discount future savings. Both adjustments are defensible and both would shorten the payback, which is exactly why neither is here. Panels are degraded at half a per cent a year over a twenty year life.

Installed prices are market observation rather than a published determination. The systems modelled here run from about 73 cents to $1.10 a watt installed after the certificate discount, with the midpoints clustered between 86 and 97 cents. Every surface that shows a payback shows that range and what it does to the payback, because a single figure to one decimal quoted off a midpoint reads like a finding when it is an observation. Certificate values use a flat $38 per certificate against a $40 clearing house ceiling.

Feed-in tariffs

Three are regulated and published: Tasmania's minimum from the Tasmanian Economic Regulator, the Northern Territory's Jacana rate, and Western Australia's Distributed Energy Buyback Scheme. The rest are retailer-set, and the figures used are observed market rates with the basis stated on the page beside every one of them. Where a state's rate is a market observation rather than a determination, the site says so rather than presenting it as settled.

Stating the limits is part of the method.

• Time-of-use tariffs in detail. The calculator handles a flat rate and a controlled load rate. A household on a three-band time-of-use plan can enter their own effective average rate, but the tool does not shift load between bands.
• The interaction between heating appliances and the house. A fridge dumps heat into your kitchen and a heater's output partly offsets other heating. Both effects are real and small; modelling them precisely would be false precision.
• Water usage cost. The figures are for energy only. The water itself is charged separately and does not vary with which system heats it.
• Installation edge cases. Installed cost ranges assume a like-for-like replacement in an accessible location. Running a new circuit, relocating a unit or upgrading a switchboard is extra and is not in the ranges.
• Your actual behaviour. Every figure assumes typical usage. The single largest source of error in any of these numbers is the household, not the model.

Energy prices are reviewed quarterly and after every regulatory reset — the next significant one is the 1 July determination cycle. Rebate schemes are reviewed quarterly. System specifications, climate data and emission factors are reviewed annually, or immediately if a new National Greenhouse Accounts Factors edition is published.

The current figures were verified on 5 September 2026. If you are reading this well after that date and a number matters to your decision, follow the source link and check it — that is what the links are there for.

Found something wrong? Tell us. Corrections are the most useful thing anyone sends us.