SYNTHETIC VERIFICATION BENCHMARK
Aerosol House Lab vs CONTAM
Two matched kitchen scenarios. Genuine NIST ContamX 3.4.0.3 results alongside the Aerosol House Lab solver and an exact mass-balance reference.
Choose a stage above. The comparison changes in place—no long-page scrolling required.
01 / Design the scenario
Matched inputs
A single well-mixed kitchen, 43.2 m³ at 20 °C, with zero initial and outdoor PM₂.₅. A 20-minute source releases 30,000 µg/h into the first size bin and 15,000 µg/h into the second. Both models use deposition losses of 0.12 and 0.40 h⁻¹ multiplied by the Aerosol House Lab humidity factor at 50% RH. Inter-room exchange and filtration are disabled. Each model uses a 15-second integration step with results compared every minute for 120 minutes.
The actual CONTAM ventilation rates were checked against its ACH export. CONTAM output mass fractions were converted using the air density at 20 °C and 101,325 Pa. This is a controlled solver verification, with no SPHERE measurements or multi-room airflow validation.
Source: NIST CONTAM download and release information. The project file format was adapted from the NIST tutorial project.
02 / Follow the air
Compare concentrations
━ Aerosol House Lab━ NIST CONTAM┄ Exact solution
The curves nearly overlap. The vertical marker shows when the synthetic emission stops.
Concentrations and errors are in µg/m³. The reference is an analytical solution for the specified inputs, with no added sensor noise.
Calibration and held-out comparison
Both physics models fit one cooking-source multiplier on the 0.30 ACH training run, with ventilation and deposition fixed. The AI correction learns Aerosol House Lab residuals using ridge regression on elapsed time and model output. All parameters are frozen for the 1.50 ACH test run.
Concentration errors are in µg/m³. Peak errors are prediction minus synthetic observation. Timing is in minutes. Synthetic observations contain a short additional source pulse and deterministic noise. This test checks transfer to one synthetic ventilation condition.
04 / Run your own case
Upload and compare your own PM₂.₅ case
Run a single well-mixed room with a constant emission that switches off at a specified time. Start with the JSON template and edit the room volume, outdoor air exchange, source and particle loss rates. Optionally add PM₂.₅ results from an external model and a separate set of measurements.
The browser runs an analytical mass-balance model. For CONTAM, run your project externally and export the time series as CSV. Files are processed locally in your browser.