Diagnostic measured-HRR replay, not validation of the house or its graphics. No accuracy acceptance threshold has been set.
Loading comparison…
On a narrow screen, scroll the chart horizontally to read the full time range.
Solid: CFAST zone · Dashed: temperature-profile-derived observation
The vertical axis is fitted to the data and does not start at zero, so differences look larger than they are. Read the error values listed below.
No paired observations are available for this quantity. Source files and exclusions remain available below.
- Bias
- Mean absolute error
- Root mean square error
Errors are model minus observation. Counts are paired time points, not independent samples.
Download zone pairs and all summary metrics · Local-probe pairs
Source files and calculation
3.66 × 3.66 × 2.44 m chamber; 0.9 × 2 m doorway open throughout. The measured fire power is prescribed input, not predicted growth.
File hashes · FSRI MIT license · Full method record
Project diagnostic window: 60–599 seconds after ignition. Source and solver clocks are paired within 1 ms with no fitted shift or temporal extrapolation. Missing observations are excluded, never filled.
The dashed series is derived from temperature probes, not a directly measured interface or zone mean. Each of four quadrant profiles is reduced separately using Kelvin integrals, then averaged. This implementation assumes linear interpolation between sensors and constant spatial extension over 0–0.34 m and 2.415–2.44 m. Celsius is used for display.
These CFAST 7.7.7 runs use measured HRR compressed from 601 to 200 knots and 1-second output. The FSRI report used CFAST 7.5.0, a nominal 10-second ramp and 15-second output. This is a source-informed adaptation, not reproduction or endorsement of that evaluation.
The separate local-probe download contains qualified probe comparisons. Probe assignment uses the model interface, with equality assigned to the lower zone. Dry oxygen is calculated from wet CFAST oxygen and water-vapor concentrations before comparison. These are not headline zone accuracy results.
No confidence interval, smoke-optics, tactical, tenability, furniture-fire or real-house validity is established. Original solver executable identity is retained from its receipt, not independently rehashed here.