ASET-RSET maps: the Schröder room

• The criteria are not incapacitation limits for a person. K ≥ 0.23 1/m and 45 °C are guideline acceptance values of vfdb TB 04-01 (2020), Table 8.3, for a stay of up to 30 minutes. The paper uses both (0.23 1/m p. 3; 45 °C p. 4, Fig. 4) and cites vfdb, but does not name the column; 45 °C occurs only in the < 30 min column. K = 0.23 1/m is D_L = 0.1 1/m converted with vfdb Eq. 8.1 (K = D_L · ln 10). ISO 13571:2012 §4.6 f treats the early effects of obscuration as behavioural and leaves them out of its incapacitation model. How these values compare with other published criteria is under Tenability criteria in the literature.
• Each set is shown whole. Taking CO from one source and temperature from another would mix two sets of assumptions. Which set is the headline and why is stated at the top of this section.
• The time column is a sensitivity. vfdb defines the long class as a stay of about 15–30 min (p. 324), so for evacuations under 5 minutes it is the more conservative choice. The < 5 min column gives 50 °C, not 45 °C, and allows D_L = 0.2 1/m where the area is clearly laid out or people know it (note 4).
• vfdb assumes a mixed fire load. The guide values assume typical mixed fire loads and a CO : HCN ratio of 12.5 : 1 (pp. 323, 325); vfdb calls for a dose analysis for sensitive groups, very short or very long exposures, or unusual smoke composition (p. 324). Our polyurethane burner is a single fuel, not a mixed load. vfdb also calls toxicity criteria “not conservative” and says they should not replace layer or smoke criteria (p. 312).
• Temperature is not a criterion on its own. vfdb Table 8.3, note 2: gas temperature is not to be assessed in isolation from smoke density. T ≥ 45 °C is never the first criterion in a cell here; it ties with smoke in 4 cells at the burner (0.2 m, one door; 2 on the 0.1 m grid, 3 with two doors).
• EA gives no visibility constant. K ≥ 0.3 1/m for 10 m visibility uses C = 3, our assumption. EA §5.2 also allows 5 m (K ≥ 0.6 with C = 3) in enclosures of about 10 m, and reduced visibility where occupants may be queuing next to exits, which is where DIFF is negative here, subject to assessment of CO and HCN levels (§5.2, p. 15). That condition is met here for CO; HCN is not tracked.
• Radiant flux is estimated as 0.25 · (U − 4σT_amb⁴) from the FDS INTEGRATED INTENSITY U, with T_amb = 20 °C (assumed): a small body in an isotropic field receives about U/4.
• Gas FED ≥ 0.3 is the dose of someone standing still from ignition, with the pyFDS-Evac form of the FDS+Evac FED (Fractional effective dose). ISO 13571:2012 §5.4 requires a reduced threshold for more conservative objectives; A.5.2 gives 0.3 as an example, at which about 11.4 % of a population would still be susceptible (see Incapacitation thresholds). No HCN is set, so its term is zero.
• Heat FED ≥ 0.3 uses the clothed convective form of ISO 13571:2012 §8.3.1, Eq. (9) (Heat). ISO gives no numeric threshold; §8.5 applies the chosen one as for the gases. The 0.3 is a pyFDS-Evac choice, by analogy with A.5.2.
• Total-flux heat, as a sensitivity. The total-flux dose (SFPE Handbook Ch. 63, Eq. 63.43) with the radiant part f · (U − 4σT_skin⁴): at f = 0.25 the radiant part never reaches the 2.5 kW/m² from which ISO counts it, so the radiant term is zero and the exceedance comes from the convective part of Eq. 63.43; the tolerance dose reaches 0.3 in 3 cells next to the burner, from 455 s (0.2 m) or 255 s (0.1 m). At f = 1, the upper bound the engine accepts, it passes 0.3 only in the same 3 cells and is not grid-stable (maximum 13.5 on the 0.2 m grid against 0.97 on the 0.1 m grid).

Effect. The fractional effective dose (FED) comes from rat data: the dose is summed until it reaches the one that produces a chosen effect, such as incapacitation or death (Hartzell and Emmons 1988, p. 356). ISO 13571:2012 sets FED = 1 at the median of “compromised tenability”, the loss of acceptable cognitive and motor performance (§3.1, §5.4). The Engineers Australia (EA) fixed limits are incapacitation tolerances for 10 min; CO 2,700 ppm matches about 27,000 ppm·min (§5.2, p. 15). The vfdb values are one level lower: guide values (Anhaltswerte) for protection goals, which at light activity reach FED ≈ 0.3 at the end of each stay class (Table 8.3 p. 325; Fig. 8.4, pp. 326–327). Purser and McAllister list escape impairment, incapacitation and death as separate columns per irritant (SFPE Handbook Ch. 63, Table 63.6, p. 2344).

Limit or dose. A fixed limit is checked at every place and instant. A dose must be summed along each person’s path, which couples the fire and evacuation models (Węgrzyński et al. 2026, §1–2). ISO 13571 uses doses for asphyxiant gases and heat, and the current concentration for irritants (§4.2, §4.4). vfdb and DIN 18009-2 base every assessment on fixed limits; a person-based dose, with a justified FED threshold, may be added for special cases (DIN §7.2.2–7.2.3, pp. 22–24). Our reading: the three vfdb CO values are almost the same dose, 2,500–3,000 ppm·min, about one tenth of the incapacitating dose EA cites.

Time and height. vfdb has three stay classes, up to about 5 min, about 5–15 min and about 15–30 min; the goal is met if no value is exceeded during the stay (p. 324). EA’s fixed limits assume up to 10 min and use FED for up to 30 min (§5.2–5.3, pp. 15–17). DIN adds the AEGL-2 for CO as a check, 420 ppm for 10 min and 150 ppm for 30 min, without interpolation (§7.2.3). EA and DIN evaluate at 2.0 m (EA §5.2; DIN §7.2.1); zero-exposure criteria use a clear layer of about 2.5 m (vfdb §8.1, p. 312; Purser and McAllister, p. 2320). Surveyed practitioners use 1.5–2.5 m, most often 2.0 m (52.1 %) (Węgrzyński et al. 2026, §4.3).

Population. ISO requires a threshold below 1 for more conservative objectives; its informative annex gives 0.3 as an example, under which 11.4 % of the population would still be affected, with no assurance that these percentages are valid (§5.4, A.5.2). Purser and McAllister suggest a design FED of 0.3, or 0.1 for sensitive groups (p. 2343); vfdb gives 0.1–0.3 (§8.4, p. 319). EA considers 1.0 suitable for the vast majority of occupants and leaves the margin to RSET (§5, p. 13). In the survey, 7 of the 13 continental-European respondents who stated a value use 0.1; all from the UK and Australia/New Zealand use 0.3 (§4.3).

Smoke. ISO 13571 leaves the behavioural effects of smoke out of its model (§4.6 f). Its obscuration endpoint, sight of about arm’s length at 0.8 g/m³ (clause 9, Note 1), is K ≈ 8 1/m with σ = 10 m²/g (our conversion), about 35 times vfdb’s 0.23 1/m. The lower limits come from behaviour: about 30 % of people turn back rather than enter smoke at about 3 m visibility, and the suggested limits are 0.08 OD/m (K ≈ 0.18) for large enclosures and 0.2 OD/m (K ≈ 0.46) for small ones (Purser and McAllister, Table 63.5, p. 2339). Where D_L ≤ 0.1 1/m, the toxic and temperature values are usually met as well (vfdb p. 325; DIN §7.2.2). In the survey (254 respondents, a convenience sample), visibility is the criterion most often decisive (91.4 %); FED is used by 22.2 % (§4.3–4.4).

Criticism. Babrauskas et al. call such criteria “highly arbitrary” (2010, p. 347); the NIST sublethal-effects study puts a factor of two on its generic values (Gann et al. 2001, pp. 82–86).

The values. vfdb gives three per quantity, one per stay class: < 30 / ≈ 15 / < 5 min.

Sources are in the Sources list at the end of the page.

Two doors, 0.2 m FDS grid. This is a separate FDS run: the second door changes the ventilation. The cells at the west door D2 exceed K ≥ 0.23 1/m at 21–32 s, the cells at the east door D1 at 73–76 s. A pocket of 8 cells beside the plume stays clear until up to 189 s.

The same ASET map on two FDS grids. Mean |Δ| 9 s; 6 % of cells differ by more than 30 s. In the door region (x ≥ 24 m, y ≥ 6 m) |Δ| is at most 16 s (95th percentile 7 s). The 0.1 m grid has a late pocket beside the plume, mostly at x ≈ 1–5 m and y ≈ 1.5–5 m, with a few cells at x ≈ 10–13.5 m (latest cell 203 s against 88 s); it is not grid-converged.

• ∃ against ∀. The paper’s text (§2.2.3) counts a cell as exceeded when the criterion holds at any data point in it; Eq. 2 as printed reads “for all”. We follow the text (∃); the ∀ rows show the difference. They count a cell as exceeded from the first 1 s step at which every FDS node in it exceeds at the same time.
• The 120 s fill. The paper gives cells that never exceed within 120 s the value 120 s (p. 4). We keep them open (“not by 600 s”). One door, 0.2 m grid: every cell exceeds by 88 s, so nothing changes. One door, 0.1 m grid: 23 cells exceed only after 120 s, but none of them is occupied after 120 s in any version, so no DIFF measure changes. Two doors (0.2 m): 8 cells beside the plume exceed only after 120 s (latest 189 s); none is occupied after 120 s in any version, so no measure changes.

The paper pools the seeds by the maximum RSET per cell (p. 5), and so do the tables above. Two other views, one door, 0.2 m grid:

With 10 seeds the 95th percentile interpolates between the 9th and 10th values, so it lies close to the maximum. For two doors the views differ more (capped, pre-movement 0): maximum −39 s, 95th percentile −37 s, per-seed mean −29.1 [−32.8, −24.9] s, because the binomial split changes how many agents use D2.

Tags: [P] stated in the paper, [F] read off a paper figure, [A] assumed because the paper is silent.

Deviations from the paper

• Our own FDS and model. None of the authors’ data is used.
• Censoring instead of the 120 s fill, the ∃ rule, and 1 s steps instead of 10 s. The sensitivity table above shows each change.
• Door jambs, 0.8 m deep, in the walkable area only. The engine removes an agent within its radius + 0.5 m of a random point in the exit polygon, which makes a door drawn flush with the wall act about 1.3 m wider (#349). The jambs force every agent into the 1.2 m passage. They cost about 1.1 m² (about 3 cells) at the north-east corner. FDS has no jambs.
• One-door and two-door rooms are separate FDS runs, so the ventilation differs as well as the egress.
• FDS device trees sit on mesh interfaces. They are not used for the maps.

Grid. The characteristic fire diameter at 60 kW is D* ≈ 0.31 m, so D*/δx = 1.56 at 0.2 m and 3.1 at 0.1 m, and H/D* = 9.6. The FDS User’s Guide (§6.3.6) warns against taking any tabulated D*/δx as an acceptable minimum. The 2.0 m slice sits in a weak, smeared layer interface, about 1–5 K above ambient, so first crossings there are ill-conditioned. Neither grid is shown to be converged: min DIFF and the door-region ASET agree within 5 s and 16 s; the area and C do not, so both grids are given.

• Schröder, B., Arnold, L., & Seyfried, A. (2020). A map representation of the ASET-RSET concept. Fire Safety Journal, 115, 103154. doi:10.1016/j.firesaf.2020.103154. Version-of-record pages: Eqs. 1–3, 0.23 1/m, 2 m and 0.6 m p. 3; the 120 s fill, Eq. 4 and 45 °C (Fig. 4) p. 4; Eq. 5 and pooling p. 5; Eq. 7 and the note on individual exposure p. 6; the two-exit thought experiment (§3.1), Fig. 5 and Eq. 8 p. 7; Figs. 6–8 p. 8.
• vfdb (2020). Leitfaden Ingenieurmethoden des Brandschutzes, TB 04-01, §8.1 (p. 312); §8.2 Eq. 8.1 (p. 313); §8.4 (p. 319); section “8.6 Anhaltswerte zur Beurteilung der Personensicherheit” (pp. 323–325) with Table 8.3 and notes 2–6 (p. 325), and Fig. 8.4 (pp. 326–327). The Leitfaden prints two sections numbered 8.6; its table of contents lists only “8.6 Rauchausbeuten”. Paraphrased.
• DIN 18009-2:2022-08, §7.2.1–7.2.3, Table 1 (pp. 22–24). Paraphrased. Table 1 heads the long column “(> 30 min)”, where vfdb Table 8.3 has “(< 30 min)”; we have not resolved which is intended.
• Engineers Australia Society of Fire Safety (2014). Practice note for tenability criteria in building fires, version 2.0, §5 (p. 13), §5.2 and Fig. 8 “Short Exposure” (p. 15; a second Fig. 8, “No Exposure”, is on p. 13), §5.3 (p. 17). Full reference on ASET and RSET.
• ISO 13571:2012, §3.1, §4.2, §4.4, §4.6 f, §5.4, §8.2–8.5 (§8.3.1 Eq. (9)), clause 9 and A.5.2; ISO/TR 16738:2009, §5.7 Eq. (2). Paraphrased; see Incapacitation thresholds.
• Purser, D. A., & McAllister, J. L. (2016). Assessment of hazards to occupants from smoke, toxic gases, and heat. SFPE Handbook of Fire Protection Engineering, 5th ed., Ch. 63, Eq. 63.43, pp. 2320, 2338–2339, 2343–2344, 2383. doi:10.1007/978-1-4939-2565-0_63
• Hartzell, G. E., & Emmons, H. W. (1988). The fractional effective dose model for assessment of toxic hazards in fires. J. Fire Sci., 6, 356–362. doi:10.1177/073490418800600504
• Gann, R. G., et al. (2001). International study of the sublethal effects of fire smoke on survivability and health (SEFS): Phase I final report. NIST TN 1439. doi:10.6028/NIST.TN.1439
• Babrauskas, V., Fleming, J. M., & Russell, B. D. (2010). RSET/ASET, a flawed concept for fire safety assessment. Fire Mater., 34, 341–355. doi:10.1002/fam.1025
• Węgrzyński, W., Spodyniuk, N., Zimny, M., Jahn, W., Vigne, G., & Arnold, L. (2026). Tenability criteria in performance-based fire safety engineering: practitioner’s perspectives from a global survey. Fire Saf. J., 165, 104938. doi:10.1016/j.firesaf.2026.104938
• SFPE Handbook of Fire Protection Engineering, 5th ed., App. 3, Tables A.38 and A.39 (fuel yields).
• McGrattan, K. et al. Fire Dynamics Simulator User’s Guide, sixth edition, NIST SP 1019, §6.3.6.
• The original FDS+Evac, for PRE_MEAN.
• Code: pyfds_evac/core/fed.py (gas and heat FED rates), pyfds_evac/core/scenario.py (exit removal and throttling), pyfds_evac/core/run_config.py (--smoke-blind), pyfds_evac/core/simulation_init.py (_apply_default_premovement).