DHW Heat Pump Validation Report · Test Matrix

DHW Heat Pump Test Matrix

Every parameter we varied, every gate we scored, every result we published — including the ones that fell outside band. Two stages, 28 in-scope scenarios, 19 PASS against EnergyPlus 26.1 on annual and monthly energy. All four exception classes are named and mechanism-characterised in the report.

A note on the downloadable case codes. The stages here are numbered sequentially, but the case codes inside the workbooks keep the prefixes they were run under — Stage 2’s files begin STAGE5_. Those strings are the identifiers of the EnergyPlus run directories the results came from, and renaming them would break the link between a published workbook and the run that produced it. We kept the traceable name over the tidy one.

Stage 1 — Steady-state single-tank shakedown

12 test cases, pass: 4/12

4/12

Reference engine: EnergyPlus 26.1.0 — WaterHeater:HeatPump:WrappedCondenser and :PumpedCondenser coupled to a 12-node WaterHeater:Stratified tank. Tank geometry, condenser span and control-sensor heights are taken verbatim from EnergyPlus's own shipped example files rather than fitted. EnergyPlus defines the thermal load; the two engines are then compared on heat-pump compressor electricity. Scored against ASHRAE Guideline 14-2023.

A 30-day constant-draw run has neither an annual nor a monthly dimension, so only the total-energy gate applies. This stage is a model shakedown, not the headline validation — its value was diagnostic. All six pumped-condenser failures share one documented cause (small-volume sump temperature), and both wrapped-condenser failures sit at the band edge under that stage's most demanding combination.

Parameters varied

ParameterLevelsValues
Condenser architecture2Wrapped-condenser residential, pumped-condenser stratified
Evaporator ambient temperature310 °C, 20 °C, 30 °C (constant)
Continuous draw rate20.003 L/s, 0.007 L/s — sized inside sustainable capacity so the resistance backup never fires
Storage volume10.189 m³ (50 gal), residential scale
Simulation period1720 hours (30 days), constant ambient and constant draw

Gates scored on every case

GateToleranceResult
EnergyPlus reference integrity0 Severe, 0 Fatal; energy balance ≤ 1% of load12/12 — balance −0.05% to −0.30%
Total compressor electricity±5% (ASHRAE G14-2023)4/12 — wrapped 4/6, pumped 0/6
Backup element quiescent≈ 0 kWh (proves draw is within capacity)12/12 — 0.00 kWh in every case
COP physical plausibilityCOP 2–6, rising with ambient12/12 — wrapped 3.90→6.30, pumped 2.97→4.73
Download sample workbook (.xlsx)7 sheets · inputs, both engines' hour-by-hour output, live formula-driven annual and monthly deviation you can re-tolerance yourself, the gate table including the statistics that were reported but not gated, and the 16-scenario timestep control test.

Stage 2 — Annual realistic building load, four climates

16 test cases, pass: 15/16

15/16

Reference engine: EnergyPlus 26.1.0 — WaterHeater:HeatPump:* + WaterHeater:Stratified, driven by the DOE / ASHRAE 90.1-2019 Large Hotel prototype's own SWHSys1 use-side heat delivery. The prototype is run as shipped, so the DHW draw is an authority's rather than an assumption. Wet-bulb evaporator inlet cross-checked against EnergyPlus's own System Node Wetbulb Temperature to within 0.024 K over 8,760 hours. Scored against ASHRAE Guideline 14-2023.

This is the headline validation. Hourly and sub-hourly agreement are out of scope: the hourly disagreement was measured to be intra-hour phase difference between two tank solvers, not an energy error — hourly CV(RMSE) of 43–104% collapses to 1.8–9.5% when the identical arrays are binned monthly, while NMBE passes 16/16 at both resolutions. Peak hourly demand is reported but not gated for the same reason.

Parameters varied

ParameterLevelsValues
Condenser architecture2Wrapped-condenser, pumped-condenser
Climate4Chicago 5A (cold), San Francisco 3C (mild marine), Tampa 2A (hot humid), Golden CO (semi-arid, −24.6 °C winter)
Evaporator ambient source2Indoor mechanical room (constant ~20 °C), outdoor (weather-file driven)
Equipment bank14 × 20 kW modules with 3.0 m³ storage, sized from the measured load against the element-trip threshold
Annual DHW load4161–219 MWh/yr, load factor ≈ 0.43 (≈ 55 L per room per day across ~200 rooms)
Simulation period18,760 hours (full year), real weather

Gates scored on every case

GateToleranceResult
Annual compressor electricity±5% (ASHRAE G14-2023 annual sum)15/16 — spanning −2.14% to +3.79%
Monthly NMBE±5% (ASHRAE G14-2023 monthly)16/16
Monthly CV(RMSE)≤ 15% (ASHRAE G14-2023 monthly)16/16 — spanning 1.75% to 9.49%
EnergyPlus reference integrity0 Severe, 0 Fatal; balance ≤ 1% of load16/16 — balance −0.02% to −0.04%
Cold-ambient lockoutreported vs EnergyPlus, not gatedLockout hours match exactly (598 h, Golden outdoor)
Peak hourly demand±10% — REPORTED, NOT GATED14/16 — 2 mild-climate indoor misses
Timestep-sensitivity controlflat part-load curve must show ≈0% shiftConfirmed 16/16 — wrapped +0.19%, pumped −6.21%
Download sample workbook (.xlsx)7 sheets · inputs, both engines' hour-by-hour output, live formula-driven annual and monthly deviation you can re-tolerance yourself, the gate table including the statistics that were reported but not gated, and the 16-scenario timestep control test.

Timestep-sensitivity control test

A falsifiable prediction, tested on all 16 Stage-2 scenarios. For a single-speed compressor the part-load ratio is its fractional on-time, and the part-load efficiency curve penalises low part-load. That penalty is evaluated at the simulation timestep, so a coarser timestep averages the on-time into a lower part-load ratio and charges a larger penalty. The wrapped-condenser archetype carries a flatpart-load curve and therefore cannot be timestep-sensitive; the pumped-condenser archetype carries a sloped one and must shift. Every scenario was re-run at EnergyPlus’s own 15-minute timestep to test it.

ArchetypePart-load curveMean shift 60 → 15 minRangen
Wrapped condenserflat (1.0, 0, 0)+0.19%+0.10% … +0.28%8
Pumped condensersloped (0.75, 0.25, 0)−6.21%−8.13% … −3.90%8

The two families do not overlap at any point, and there is no counter-example. One consequence is reported because it bears on how the result should be read: the finer timestep does not produce a clean sweep. At 15 minutes the score is also 15 of 16, but a differentscenario falls outside band. Moving to the finer timestep is therefore a re-calibration of the part-load treatment rather than a correction, which is why it has not been adopted as a fix. The practical figure for a user is that pumped-condenser annual energy carries a timestep sensitivity of roughly 6–8%, and wrapped-condenser carries none.