Steady-state yield map - L/day vs relative humidity

Current machine at five ambient temperatures. Excludes defrost downtime, tank stops and the controller envelope (15-40 °C / 35 % RH cutoff). Hollow markers: the manufacturer's own 28 °C capacity claims - unreachable below ~60 % RH at nameplate power. The cliff at low RH is the dominant design constraint.

Refrigerant cycle - P-h diagram (R134a)

Saturation dome from R134a property tables; cycle ①→②→③→④ at the current operating point. Physical compressor model shows the true computed cycle; empirical model shows the equivalent cycle at the same temperatures.

Psychrometric process

Moist-air path across the evaporator: inlet → outlet with apparatus dew point (ADP) on the saturation curve. Condensate = vertical drop in humidity ratio.

Specific energy - Wh per litre vs relative humidity

Energy cost per litre at steady state. AW3 test implied ≈ 750–850 Wh/L.

Sensitivity - 24 h yield response to ±10 % parameter change

Tornado for the current scenario & configuration (noise off). Longest bars = the parameters that matter most for this design point. Recomputed on demand.

Design candidates

Save the current configuration as a named candidate, reload it later, overlay it on the live charts, and batch-compare all candidates on the anchor scenarios. Candidates persist in this browser; use Export for a portable copy.

Comparison - all candidates on anchor scenarios

Each candidate is run headless (24 h rated · 25 h AW3 test day · 24 h current scenario, noise off, drained tank).

Overlay on live charts

Pick a candidate to overlay (dashed) on the production-rate and cumulative charts, run against the same scenario as the live simulation.

Validation against AW3 reference data

Headless anchor runs, auto-recomputed on every parameter change (drained tank, noise off). Applies to the active compressor model.

Inferred test conditions - important. The AW3 report did not log numeric humidity. Fitting the measured 15.4 L (split 8.3 / 7.1) requires a dew point near 24 °C - a humid coastal profile. At inland-Riyadh July dryness (dew point ≈ 13 °C - "dry stress case" preset) the model predicts near-zero yield: the coil cannot reach below the dew point. Confirm the actual test location; log temp/RH in all future tests.

Tuning guide. Calibration knobs: collection efficiency + (empirical: Carnot efficiency · physical: isentropic/volumetric efficiency). Keep hardware parameters physical; tune knobs until the header chips are green; that parameter set is the calibrated digital twin. Explore Hayat candidates from that baseline and save them in Candidates.

Manufacturer capacity curve - 28 °C (manual p.17)

The AW3 manual's own production-capacity table vs this model (active compressor model, steady 28 °C, 24 h headless runs). The models match the manufacturer above ~80 % RH; below ~60 % RH the claimed figures physically exceed what ~470 W supports - both independent models agree on this. Treat low-RH vendor ratings as unattainable.

Physics self-test suite

Built-in assertions against textbook reference values and internal-consistency invariants. Runs at load and on demand - a red header dot means the physics is compromised and results must not be trusted.

Model assumptions & scope

• Psychrometrics - Arden Buck saturation-pressure correlation; humidity ratio, enthalpy, density from ASHRAE-form relations at 101.325 kPa; dew point by numerical inversion (consistent with the same correlation).
• Airflow (fluid dynamics) - quadratic system-resistance curve (ΔP = R·V̇²) intersected with a quadratic fan curve; filter fouling and coil frost increase R and derate airflow; fan electrical power = V̇·ΔP/η_fan.
• Evaporator (heat transfer) - apparatus-dew-point / bypass-factor formulation (enthalpy-potential wet coil): BF = e^−NTU, NTU = UA/(ṁ_ac_p); frost degrades effective UA.
• Refrigerant cycle (thermodynamics) - two selectable fidelities. Physical: R134a saturation-property table (−40…70 °C), positive-displacement compressor (V_disp, rpm, η_vol derated with pressure ratio), isentropic compression via η_is, superheat/subcool, isenthalpic expansion - capacity, power and COP all computed from state points. Empirical: Carnot-fraction capacity model at fixed electrical power.
• Condenser - condensing temp = condenser-air inlet + approach ΔT; air path selectable: parallel (fresh ambient) or series (evaporator outlet air - cooler → lower T_cond, typical AWG layout).
• Operating point - bisection each timestep on the coil surface temperature where air-side heat extraction = refrigeration capacity; solver residual is asserted < 2 % in the self-tests. Low-humidity collapse is emergent, not scripted.
• Frost & defrost - sub-threshold condensate accumulates as ice, degrading UA and airflow; timed defrost at trigger mass with configurable heater power and melt recovery.
• Controller - AW3 logic: tank-full stop, low-RH cutoff (+5 % hysteresis), ambient envelope, controls/UV standby load.
• Environment - steady, sinusoidal diurnal (constant absolute humidity), Saudi city presets, 24 h hourly CSV import, optional seeded (reproducible) weather noise. Noise is always off for validation runs.
• Scope - system-level model for design exploration; not CFD, not coil-geometry design, no chemical water-quality model. Validate final designs against prototype tests (Testing & Analysis protocol).

Full derivations and calibration record: Simulation/Model_Documentation.md.

Data - current run

Hourly summary (accessible alternative to the charts). Export full minute-resolution data as CSV.

Shortcuts

Space run / pause · R reset · ↑/↓ speed up / down · → skip +1 h · 1–4 switch tabs