What your FDS case must provide

Read this before pointing --fds-dir at a case for the first time.

pyFDS-Evac never runs FDS. It reads the output of a finished FDS run. A deck written for some other purpose usually will not work as-is: it has to be told to dump the specific slices this tool samples.

The slices

&SLCF PBZ=1.6, QUANTITY='EXTINCTION COEFFICIENT' /
&SLCF PBZ=1.6, QUANTITY='VOLUME FRACTION', SPEC_ID='CARBON MONOXIDE' /
&SLCF PBZ=1.6, QUANTITY='VOLUME FRACTION', SPEC_ID='CARBON DIOXIDE' /
&SLCF PBZ=1.6, QUANTITY='VOLUME FRACTION', SPEC_ID='OXYGEN' /
&SLCF PBZ=1.6, QUANTITY='TEMPERATURE' /
&DUMP DT_SLCF=1.0 /

Nothing here is dumped by default. FDS writes only what the deck asks for.

For the default species (soot), FDS records the extinction-coefficient slice internally as SOOT EXTINCTION COEFFICIENT, and that is the exact quantity name load_slice_sampler looks up (fds-sampling.md).

It is all three gases or none of them; there is no partial FED. TEMPERATURE is independent of that gate — it needs neither CO/CO2/O2 nor any &REAC yield to work (see below), so a case can have heat FED without toxic FED or vice versa.

EXTINCTION and EXTINCTION COEFFICIENT are two unrelated gas-phase output quantities (FDS User Guide Table 22.4) — don’t confuse them. EXTINCTION (Sec. 22.10.29) is a combustion-suppression flag with no units: 0 if the extinction routine did not prevent combustion, 1 if it did, -1 if there was no fuel or oxidizer. EXTINCTION COEFFICIENT (Sec. 22.10.5) is the light extinction coefficient K [1/m] that the smoke-speed model actually needs. If your deck has QUANTITY='EXTINCTION' where you meant the extinction coefficient, fix it to QUANTITY='EXTINCTION COEFFICIENT' and rerun FDS.

Declaring a slice is not enough

The species also has to exist in the run. Under simple chemistry, OXYGEN, NITROGEN, WATER VAPOR and CARBON DIOXIDE are always present, but CARBON MONOXIDE only exists if CO_YIELD is set on &REAC, and SOOT only if SOOT_YIELD is set. A plain &REAC FUEL='PROPANE' / cannot produce FED no matter what &SLCF lines you add. Fix it at the source:

&REAC FUEL='CABLE_FUEL', C=3, H=5, O=0, N=1,
      SOOT_YIELD=0.172, CO_YIELD=0.063, HCN_YIELD=0.006,
      HEAT_OF_COMBUSTION=16400 /

That is the t_junction reaction. Yields are fuel properties: take them from your material’s data rather than copying these.

TEMPERATURE is the one exception: it is a core solved gas-phase variable in every FDS run, not a species yield, so it needs no &REAC setup at all — the &SLCF line above is sufficient on its own.

Silent failure modes

FED disabled. If CO, CO2 or O2 is missing, pyFDS-Evac carries on with smoke speed only. If the heat FED is off too, the result has no FED at all: fed_history is None, metrics has no fed_max, and --output-fed-history writes no file. If the heat FED runs but the gas FED does not, the FED history exists and its fed_cumulative column is all zero, which looks exactly like a survivable fire. Check metrics["fed_max"], or the warning, which is logged every time (#137):

FED is disabled for <dir>: it has no CO slice, and all three of CO, CO2 and
O2 are needed. ...

Heat FED disabled. The heat FED track is off unless --enable-heat-fed is given. With it, the same applies as for the gases: if there is no TEMPERATURE slice, heat FED is off, metrics has no heat_fed_max, and any heat FED column in the history reads zero:

Heat FED is disabled for <dir>: it has no TEMPERATURE slice. ...

Wrong slice height. --smoke-slice-height (default 1.6 m, FDS+Evac’s HUMAN_SMOKE_HEIGHT) is a preference, not a filter: a case with one slice of a quantity uses it whatever its height, and a mismatch never fails the run. A case you inherit often carries a single slice at whatever height its author chose, so you can end up sampling floor-level CO for standing agents. A warning fires only past 0.5 m. The common case, a request of 1.6 m on a deck with a single slice at 2.0 m, is 0.4 m away and gives no warning (#165). Beyond 0.5 m you see:

Requested a 'SOOT EXTINCTION COEFFICIENT' slice at z=0.50 m but the nearest
available in <dir> is at z=2.00 m ...

Do not ignore these warnings. Nothing else will tell you.

Failure modes

Errors from the command line itself (wrong --fds-dir, conflicting flags) are listed on Troubleshooting.