In BalbiNov2011, rate of spread increases with live fuel moisture above Ml ≈ 0.8. The shipped fuels.csv uses Ml = 1.0, so this affects the default setup.
Cause: xsi = (Ml - Md)·(Sd/Sl)·(Deltah/DeltaH) goes above 1, so (1 - xsi) < 0 and flame temperature drops below ambient. R00 ∝ T⁴ then turns that back into a large positive value.
Fuel 1, no wind, flat ground, Md = 0.1:
Ml |
0.1 |
0.8 |
1.0 |
2.0 |
| ROS (m/s) |
0.0227 |
2.3e-8 |
8.2e-5 |
0.16 |
Balbi2015, BalbiNov2011Curv and BalbiUnsteady all decrease monotonically and agree to 4 significant figures. The only difference: BalbiNov2011 uses Sd/Sl where Balbi2015 uses Sigmal/Sigmad. Worth checking against Santoni et al. (2011).
Options: clamp xsi at 1, fix the coefficient, or reject Ml ≥ ~0.8 for this model. Each changes results, so it is a physics decision. #158 only clamps at 0.
Also: BalbiNov2011TMdMl does not respond to Ml at all. Related: #35, #99.
Drafted with Claude Opus 5, reviewed by a maintainer.
In
BalbiNov2011, rate of spread increases with live fuel moisture aboveMl ≈ 0.8. The shippedfuels.csvusesMl = 1.0, so this affects the default setup.Cause:
xsi = (Ml - Md)·(Sd/Sl)·(Deltah/DeltaH)goes above 1, so(1 - xsi) < 0and flame temperature drops below ambient.R00 ∝ T⁴then turns that back into a large positive value.Fuel 1, no wind, flat ground,
Md = 0.1:MlBalbi2015,BalbiNov2011CurvandBalbiUnsteadyall decrease monotonically and agree to 4 significant figures. The only difference:BalbiNov2011usesSd/SlwhereBalbi2015usesSigmal/Sigmad. Worth checking against Santoni et al. (2011).Options: clamp
xsiat 1, fix the coefficient, or rejectMl ≥ ~0.8for this model. Each changes results, so it is a physics decision. #158 only clamps at 0.Also:
BalbiNov2011TMdMldoes not respond toMlat all. Related: #35, #99.Drafted with Claude Opus 5, reviewed by a maintainer.