Explore the data →

An 80-litre tank with a 1 kW compressor and a small top element passes medium-load STC, VEU and ESS

We simulated almost 7,000 small-HPWH designs to ask whether a cheap, small integral unit could qualify for residential incentives at medium load, the load size VEU and ESS both require. It can. An 80 L tank, a compressor no bigger than the smallest already sold, and a 0.8 kW top element running on its own thermostat pass every scheme's medium-load test in every zone we checked — at lower build cost, lower running cost and more STC value than a conventional 250 L unit, with no efficiency penalty for going small.

15-year net spend, zone 3, medium load, central cost estimate

Small hybrid
Conventional reference
Retail + install, paid on day one Running cost over 15 years, net of STC
cheaper upfrontcheaper net over 15 yearsmore STC value, not lessplug-inno electrician needed

What we found

  1. A small integral HPWH can pass medium load. 80 L, a 1.0 kW compressor and a 0.8 kW top element 28 L above it at 55 °C pass MDT and the 60% savings gate in zones 3, 4 and 5 — eligible for STC, VEU and ESS.
  2. Going small costs nothing on efficiency or incentives. The headline design uses less electricity and earns more STC value than the 250 L reference it is compared against. Compressor capacity does not predict COP in the fleet (R2 = 0.01), so there was never an efficiency penalty to trade away.
  3. The mechanism is the element, not a bigger heat pump. The same tank and compressor with no element at all fails medium-load MDT by more than 10 K. The element buys that back for about 3 points of savings, in a design that is already a plug-in.
  4. It is cheaper, not just smaller. About 29% less upfront and 21% less over 15 years at the central cost estimate, though the cost bands are wide enough that this needs a real OEM quote to be a claim rather than an estimate.
  5. Real use is where the small tank gives something up. Under a realistic random draw pattern it delivers about 5% of its annual hot water below 45 °C, against near zero for the 250 L reference. Savings and STC barely move. This is reported, never used to fail a design.
  6. TRNSYS confirms the headline design, and is why the study now floors compressor capacity at 1.0 kW. The 80 L / 1.0 kW headline and the 150 L / 1.2 kW no-element design both pass in every zone on the production V15 template. An earlier, smaller 0.6 kW variant did not: TRNSYS failed it in zones 4 and 5, exactly the compressor class the component-cost research could never find a public listing for. Direct checks against the fleet confirm why: the smallest lab-tested compressor in the 298-model fleet is 1.06 kW, and none exists below it. Every sweep in this study, and the interactive explorer's own sliders, now stop at 1.0 kW — just inside real lab-tested experience, never past it.

The question

Can a manufacturer build a small, low-cost integral HPWH that still earns STC, VEEC and ESC at medium load, by pairing a small compressor with a top-mounted element that covers the draw events a heat pump alone cannot deliver fast enough?

Small HPWHs have historically been ruled out at medium load on delivery temperature or savings. The premise going in, carried over from a commercial lifecycle-cost finding (P68), was that a smaller heat pump plus an element would trade a few STCs for much lower build cost. What we measured is better than that premise: there was no STC trade to make.

Engine
Python `hpwh-simulator`, integral TYPE 104 path, AS/NZS 4234:2021 basis for all three schemes (STC included — TYPE 104 only, by decision)
Swept
Tank 80 to 315 L across four design families; compressor 1.0 to 3.5 kW rated heating (floored at 1.0 kW, see "What we found"); element 0 to 3.6 kW at three heights and three setpoints; zones 3, 4, 5; small and medium load
5125 coefficients
Synthetic, generated from the 298-model fleet's own regression (no capacity-COP trend found) plus a condenser-area COP penalty measured from the fleet's own lift sensitivity
A design passes if
MDT ≥ 45 °C in every claimed zone; medium load also needs savings ≥ 60% in zone 3; AS 3498 sanitation holds
Cost
Public component prices and public retail benchmarks (19 Oct 2026), banded low/central/high; no EnergyAE client or internal cost data

The answer

Small hybrid

net over 15 years

Conventional reference

net over 15 years
Where the 15-year money goes. Retail and install are paid on day one; the running-cost bar is net of the STC value both designs earn.

This is a central cost estimate, not a quote: the cost bands are wide enough (see "What can change the answer") that an unlucky quote on the small design against a lucky one on the reference could close the gap. The ordering is a finding worth an OEM RFQ, not yet a guarantee.

Why it works

R2 = 0.01COP vs rated capacity, fleet-wide

Regressing the fleet's own rated COP against rated capacity gives almost no relationship. The premise that small compressors are less efficient does not hold, so there is no COP price for going small to begin with.

+11 KMDT gain from an 0.8 kW element

The same 80 L / 1.0 kW design with no element at all fails medium-load MDT at 40.4 °C. Adding the element lifts it to 51.4 °C for 3.4 points of savings — a small, bounded cost for turning a delivery failure into a pass.

28 L, 55 °Cbeats low and cool every time

Across every tank size swept, an element placed high with a hot setpoint passes MDT at a lower savings cost than the same element lower in the tank or at a cooler setpoint.

No template change, no new CER question

The production TYPE 104 template already feeds a second element slot unconditionally (Aux_2pw = AUX2), so it switches on its own thermostat whatever the heat pump is doing. A top element that runs alongside the compressor is parameter values in the existing, lodged template — nothing to register differently. The Python engine itself needed a one-line fix to match that deck behaviour; it previously zeroed that slot unless a demand-response controller was attached, which happened to be inert for every one of the 298 fleet models already in the engine.

A day in the life of the 80 L design

Simulated minute-by-minute, not just as an annual total: a worst winter day and a normal summer day in zone 3, for the headline design against the 250 L reference.

These two days are picked by mean ambient temperature, not by the single coldest moment the year's minimum delivery temperature comes from — they show the mechanism at work, not the worst case already reported in "What we found". On both days the small design's tank-top temperature never gets close to the 45 °C line; the element fires in short, frequent bursts timed to draw events, while the compressor carries the bulk of the recovery in long, steady runs.

Worst winter day (zone 3, mean ambient  °C). Top: tank-top (delivered) temperature. Bottom: electrical draw, compressor and element stacked, same kW scale for both designs so the smaller compressor's smaller draw is visible directly.
Normal summer day (zone 3, mean ambient  °C) — the Dec-Feb day closest to that season's median, not the warmest.
AmbientSmall hybrid, delivered temp / compressorReference, delivered temp / compressorElement

Starts, run-hours, and what that means for longevity

Annual totals from the same simulation, not the two days above:

The small hybrid's compressor runs about twice the hours per day of the reference (a 1.0 kW compressor simply takes longer to move the same daily energy than a 2.5 kW one does), but it does not start much more often — about 14% more starts per day, and each of its cycles is longer, not shorter. Long, steady runs are generally easier on a compressor than frequent short ones: the wear mechanisms usually pointed to (inrush current, oil migration during the off period, thermal stress at startup) concentrate in the start, not in the hour that follows it. The element is a resistive heater with no moving parts, so its much higher duty-matched cycling rate is not a comparable wear concern.

Cycling behaviour is measured here; cycling life is not. Whether a 1.0 kW compressor class is rated and built for about 1,300 starts and 3,000 run-hours a year at the same reliability as a larger unit is a question for the OEM's own compressor datasheet during prototyping, not something this engine can answer. TBC.

The real objection to a small tank: maximum delivery volume

The strongest objection to an 80 L tank is not efficiency or STC — it is that a small tank simply holds less hot water for a single continuous draw, whatever heat source sits inside it. MEPS (AS/NZS 5125.1 Appendix H) measures exactly this: a continuous full-flow draw at 9.5 L/min (570 L/hr) until the outlet falls through 50 °C, giving V50, the maximum delivery volume the product can declare and the figure GEMS uses to cap the declarable load size. We ran the full Appendix H maximum draw-off on the headline design and on the cheapest-passing design at every swept tank volume, element fitted and removed, at all three climatic conditions. The honest number: the 80 L headline delivers V50 of about 71–76 L, against 214–293 L for the 250–315 L reference family. That gap is real, it is the main downside of this design, and no amount of clever control changes it much.

The top element does not close that gap. It adds about 1 L, under 2%, in every case we ran — and nothing at all for the reference family, whose backup element never gets a chance to contribute before the floor is reached. The reason: a 570 L/hr draw removes heat around 50 times faster than a 0.8–1.0 kW heat source can replace it, so almost none of V50 comes from production during the draw itself — it comes from the water already stored in the tank when the draw starts. That matches the standard's own closed-form identity for V50 (engine notes, H.7.2): set by stored volume and charge efficiency, not by auxiliary heating rate. None of the deltas we measured were large enough to move a design across a Table 15 (GEMS) 50 L declarable-size band.

This does not contradict the annual result above, it sits next to it. The element earns its keep on STC savings and medium-load MDT because the AS/NZS 4234 scheme day draws in short, intermittent events it can keep up with between them. A single continuous 570 L/hr draw is a different, harder regime it was never sized for, and MEPS is what exposes that. A buyer who needs one long, uninterrupted hot-water draw bigger than about 75 L — a large bath, back-to-back showers with no gap — is not well served by this design at this tank size; one who draws in the pattern AS/NZS 4234 models (and most households do) is. Full per-volume, per-condition numbers and the draw-off trace are in the interactive explorer's MEPS V50 tab.

The viability frontier

For each tank size, the cheapest design (retail + install) that passes medium load in zones 3, 4 and 5. "No element" needs a much bigger compressor at every tank size; the element is a cheaper way to add delivery margin than more steel is.

Every point here uses a compressor of 1.0 kW rated heating or larger — the floor this study now applies throughout, matching the smallest real lab-tested unit in the fleet (1.06 kW) and the smallest public compressor listing found in this study's own component-cost research.

This is one slice through the design space at one tank size and one cost estimate. To try your own combination of volume, compressor, element capacity, element position and element setpoint — and see the price/efficiency trade-off across all 720 buildable combinations at once — use the interactive explorer's design configurator.

What can change the answer

Every finalist was re-run under the coefficient bands, the condenser-area bands, the electricity price range and the cost bands.

FactorRange testedWhat movedCan it reverse the result?
Coefficient bandoptimistic to pessimistic COP level and lift-slopeSavings 69.0 to 74.0%, MDT 50.9 to 51.8 °CNo passes at every band
Condenser coverage bandoptimistic to pessimistic contact areaApproach penalty 4.8 to 8.0 K; MDT barely moves, 51.4 to 51.6 °CNo
Electricity price25 to 45 c/kWhRunning-cost gap widens from $499 to $897 over 15 yearsNo hybrid stays cheaper to run
Cost bandslow to high, both designsSmall design $944-8,643 upfront; reference $2,862-9,778Yes bands overlap — needs a real quote

Real draws are not the scheme day

The scheme test uses one repeating day. We ran both designs under a seeded, realistic stochastic draw pattern (energy-matched to the same annual load), three seeds each, and report the share of the year's hot water delivered below 45 °C — never used to fail a design, by standing decision.

Share of annual hot-water energy delivered below 45 °C, three seeds per design. Savings and STC barely move for either design between the deterministic and stochastic runs.
This is the real trade-off the deterministic pass test does not show. The small tank passes the engineered 4234 test exactly as well as the conventional one, but an occupant drawing water at a random moment is measurably more likely to get a cooler draw from it. State the number; do not hide it behind the deterministic pass.

Does TRNSYS agree?

Three synthetic finalists were rendered as parameter-only overwrites of a real 82 L registered unit's own TRNSYS template (TYPE 104, unchanged -- a full line diff confirms no control-logic or card-count change, only equation-block values) and run via the production harness, stepping down the timestep from the house standard 0.02 h only where a run hit a recorded convergence fault.

    What we corrected on the way

    What we first thoughtWhat the tests showed
    A co-running top element would need a new CER question.The production TYPE 104 template already supports it in its second element slot, with no template change. Only the Python engine needed a one-line fix to match the deck.
    Smaller compressors carry a COP penalty, so STCs trade off against build cost.Fleet-wide R2 between rated capacity and rated COP is 0.01. The real size-driven cost is condenser area, not compressor size, and it is small enough to barely move the result.
    Power coefficients scale with a fixed slope borrowed from the smallest real fleet units.A fixed absolute slope sent 0.6 kW designs' power negative at the physicality screen's cold corner. Scaling by elasticity (a fraction of the design's own rated power) fixed it; 143 of 144 swept designs then pass.
    A short condenser makes a design infeasible.The registered 82 L unit itself implies an 11 K condenser approach at its real contact area — real and sold, just a COP cost, not a wall. The geometry module reports the approach, it does not gate on it.
    A design within about 2 K of the 45 °C line (an earlier measured Python-vs-TRNSYS band) is settled once it clears that margin.This study's own TRNSYS check found real gaps down to -4.0 K on new designs, large enough on one finalist to flip a verdict. That earlier band does not generalise; anything within about 4 K still needs its own TRNSYS run, and even then is not automatically safe.
    The engine's condenser-area substitution (contact area as a single number) is a safe way to build any TYPE 104 Mode 4 deck.TRNSYS enforces a real geometric limit the engine does not: extrusion length cannot exceed the tank's circumference (ERROR #1043). A synthetic condenser design needs a realistic strip layout, not just the right total area.

    What is not settled, and next steps

    Not yet established
    • Real OEM component quotes. Every BOM figure is a public-listing estimate with a wide band.
    • VEEC and ESC per-certificate traded prices (left TBC, not invented).
    • Whether CER accepts 2021 (TYPE 104) modelling for STCs from the next round.
    • AS/NZS 5125.1 Appendix H (MEPS V50) is now simulated on the finalists ("A day in the life" above) — but no MEPS minimum level exists yet for this product class, so there is nothing to pass or fail against.
    Next steps
    1. Get OEM quotes for the 1.0 kW compressor class and an 80 L tooled tank.
    2. Confirm VEU/ESS scheme rules against the primary Specifications and the IPART HEER guide.
    3. Price VEEC and ESC at current traded value.
    4. Build the manufacturer-facing blueprint (coefficient targets, BOM ranges, element and condenser rules) around this design, not a registered product.

    Basis and method

    Engine. `hpwh-simulator` (Python), integral TYPE 104 path (microchannel Mode 4), pinned commit, AS/NZS 4234:2021 loads and weather for all three schemes. STC on the 2021 basis is a plan decision, pending CER confirming 2021 modelling is accepted for STCs from the next round.

    Geometry. Tank diameter bounded by the fleet's own aspect-ratio range; condenser contact area banded from the fleet's microchannel coverage fraction; standing loss banded on a fleet U-value (regressing UA directly on area gives R2 = 0.02, so a band is used, not a line).

    5125 coefficients. COP level banded by condenser mode from the fleet (no capacity trend found); condenser-area COP penalty from the fleet's own measured dCOP/ddT1 on the small-unit cohort (cap_rated < 2.0 kW), since the P34 physics simulator has no small-tank validated model and inventing one would be unvalidated.

    Cost. Public component prices and public AU/NZ retail benchmarks, researched 19 Oct 2026, each a central estimate with a low-high band. Not yet signed off internally; fixed before results were read, per the rule that costs cannot be set after the optimum is known.

    Confidentiality. Registered fleet models (used only to infer coefficients, geometry and the compressor/condenser statistics above) are never named. Every design in this report is synthetic.

    Explore every one of the 4,758 simulated designs, the viability frontier, and the MEPS V50 study →