Fire design

Assess fire resistance of concrete sections using heat transfer analysis, the AS 3600 Section 5 tabulated design method, or AS 3600 Section 5 tabular cover requirements.

Overview

Fire design verifies that a concrete member retains adequate structural capacity after exposure to fire for a specified duration. Concrete has inherently good fire resistance, but elevated temperatures reduce both concrete compressive strength and steel yield strength. The depth of temperature penetration and the resulting capacity reduction depend on the member geometry, exposure conditions, and fire duration.

ACS performs three complementary fire assessments:

1. Heat transfer + fire capacity — a 2D finite element heat transfer analysis followed by a fire-reduced biaxial interaction surface and utilisation check (available for every design code, so it follows ACI 318-19 and EN 1992-1-1 when those become selectable)
2. AS 3600 Section 5 tabulated design method — a deemed-to-satisfy design gate that looks up minimum section width and axis distance from the Section 5 tables using your actual support condition and computed fire load level μ_fi; outputs a PASS, FAIL, or NOT ASSESSED verdict with demand/supply ratios (AS 3600 only)
3. AS 3600 Section 5 tabular cover — reads axis distances from the code tables using conservative worst-case structural assumptions and converts the axis distance to a minimum clear cover for use in the Cl 4.10.1 cover calculation (AS 3600 only; driven by the Member Type setting in the General tab)

All three fire assessments run independently of your ULS load combinations. The Fire tab is available and produces results even when no ULS combinations have been defined. Fire limit-state demands ($N^*_f$, $M^*_{x,f}$, $M^*_{y,f}$) are entered separately in the Fire Demands panel and are not derived from the ULS actions grid.

When to use fire design

Check fire resistance when:

• The member must achieve a specified Fire Resistance Level (FRL), such as 60/60/60 or 120/120/120 (structural adequacy / integrity / insulation)
• Building regulations require fire-rated construction for the occupancy class
• The member is in a fire compartment boundary or supports fire-rated construction above

Member type (General tab)

The Member Type selector in the General tab drives both the AS 3600 Section 5 tabular cover calculation and the AS 3600 Section 5 tabulated design method. ACS currently supports four member types:

Beams and slabs use the support condition you declare. Set it under Fire tab → Fire method → AS 3600 Section 5 (tabulated) → Support Condition. Declaring Continuous selects the continuous table, which permits smaller axis distances — and therefore smaller cover — than the simply-supported one. If you declare nothing, the simply-supported table applies: that is the conservative default, not an assumption ACS makes on your behalf against a stated choice.

Both the tabular cover and the AS 3600 Section 5 tabulated design method read that same declaration, so the axis distance shown on the Cover panel and the one shown in the fire results are the same number for the same section.

:::note[Support condition default is now persisted — September 2026 (#5287)] The Support Condition selector (Fire tab → Fire method → AS 3600 Section 5 tabulated) now writes its displayed default — Simply-supported — to the section record on save, rather than leaving the field null and resolving the default at analysis time. Before this fix, saving the section without explicitly choosing a support condition could produce a different fire verdict from the one displayed (§874). If you have fire sections where you accepted the simply-supported default without explicitly setting it, re-open and save them to record the value. :::

Columns and walls differ. Their tables key on the fire load level μ_fi rather than a support condition, and μ_fi is computed from your loads by the tabulated design method. The cover resolver runs no fire analysis, so it has no computed μ_fi to read and uses the conservative tabulated level instead — meaning the tabulated design method may report a smaller axis distance for a column or wall than the cover resolver assumed. The Cover panel states this where it applies.

The member type does not affect the heat transfer analysis or the fire-reduced interaction surface — those depend only on the section geometry, fire duration, fire curve, and exposed edges.

Axis distance and clear cover

AS 3600 Cl 5.2.1 defines the axis distance $a_s$ as the distance from the nearest exposed concrete surface to the centre of the reinforcing bar. The minimum clear cover for fire is the axis distance less however far ACS actually places that bar’s centre in from the face — and that depends on where the bar sits.

A bar on a face — a mid-edge bar — sits one fitment plus one bar radius in:

A corner bar does not. It nestles inside the fitment’s bend, so its centre is pushed further in, and the fire cover is:

Where:

• $a_s$ = axis distance from Table 5.4.1(A), 5.5.2(B), 5.6.3, or 5.7.2 (mm)
• $d_{fitment}$ = fitment (stirrup) bar diameter (mm); zero when no fitments are present
• $d_b$ = governing fire-exposed longitudinal bar diameter (mm)
• $r_{eff}$ = the fitment’s effective bend radius, $\max(r_{internal}, d_{b,corner,max}/2)$ (mm) — the largest corner bar on the tie opens the bend for every corner bar on it
• $\theta$ = the interior angle of the corner the bar sits in
• $c_{fire}$ = minimum clear cover for fire (mm)

Both terms exist because ACS derives the cover by inverting its own bar placement: a bar sitting outside a stirrup has its centre that much further from the surface than the cover alone, and a bar seated in a bend further still. Ignoring the fitment overstates the fire cover; using the mid-edge expression for a corner bar understates it, burying the steel deeper than the standard asks. The two expressions coincide when there is no fitment ($r_{eff} = 0$) and when the corner bar is fat enough to open the bend itself ($r_{eff} = d_b/2$).

The distinction matters most for Table 5.4.1 Note 1, whose $a_{sd}$ is a corner-bar requirement by definition — the corner expression always applies there. The cover panel’s Section 5 disclosure names whichever term was subtracted, so the printed arithmetic always describes the placement the canvas draws.

The governing cover for the section is the largest of the fire cover, the durability cover (from the exposure class and concrete grade), and the placement cover (minimum 20 mm, or 30 mm adjacent to formed surfaces per AS 3600 Cl 4.10.1).

Beam axis distances (Table 5.4.1(A) and 5.4.1(B))

For beams the required axis distance depends on the FRL, the minimum web width $b$, and the support condition you declare. The table below is Table 5.4.1(A), simply-supported — the default when you declare nothing. Declaring Continuous selects Table 5.4.1(B), whose axis distances are smaller: a 300 mm beam at FRL 120 requires 55 mm simply-supported and 35 mm continuous.

ACS reads the section’s actual width and selects the governing row:

Axis distances in mm. A dash means the section is too narrow for that FRL — ACS reports Resolved = false and shows a failure reason rather than returning a silently-substituted value.

Column axis distances (Table 5.6.3)

For columns at μ_fi = 0.7, the required axis distance depends on the smaller cross-section dimension and the number of exposed sides:

For a multi-side-exposed column at FRL 240, AS 3600 Table 5.6.3 has no entry — ACS refuses and reports a failure reason rather than extrapolating.

The number of exposed sides is determined by the Exposed Elements configuration in the Fire tab.

Slab axis distances (Table 5.5.2(B))

For slabs the required axis distance depends only on the FRL and the support condition you declare. The table below is the simply-supported one-way column of Table 5.5.2(B) — the default when you declare nothing; declaring Continuous reads that table’s continuous column instead. Table 5.5.2(A) is deliberately not used: its smaller values would under-state the requirement.

Wall axis distances (Table 5.7.2)

For load-bearing walls at N*/φNu = 0.7:

AS 3600 Section 5 tabulated design method

The AS 3600 Section 5 tabulated design method is a deemed-to-satisfy design gate and is an alternative to the EN 1992-1-2 heat transfer and fire capacity calculation methods — the two approaches are never cumulative (AS 3600 §4.1). It is available for AS 3600 designs only.

Unlike the tabular cover resolver (which uses conservative worst-case inputs and outputs a cover number), the tabulated design method uses your actual inputs and outputs a PASS, FAIL, or NOT ASSESSED verdict alongside demand/supply ratios.

:::note[Tabulated route reads all inputs from the persisted design option — August 2026 (#5182)] The AS 3600 Section 5 tabulated route (GET .../fire-section5) reads all its inputs — member type, support condition, FRL, exposed-face configuration, and aggregate type — from the persisted design option. No additional parameters are required in the request body. This contrasts with the heat-transfer and fire capacity routes, which additionally require fire demand inputs (N*f, Mx*f, My*f). For columns and walls, the fire load level $\mu_{fi}$ is computed from the design option’s saved fire load combinations if present; when none are stored, ACS falls back to $\mu_{fi} = 0.7$ and flags this in the response (see below). The tabulated route is therefore fully operable from a bare design option ID, making it the simplest fire check to call via the API or MCP tools. :::

Verdicts

Demand/supply ratios

The assessment expresses each check as a demand/supply ratio so they slot into the governing-entry result machinery:

• $b$ — minimum section width: the minimum web width for beams; the smaller cross-sectional dimension (thickness or lesser side) for all other member types
• $a_{s,\text{required}}$ — the axis distance read from the applicable Section 5 table
• $a_m$ — the Cl 5.2.1 average axis distance measured from the section geometry (see below)
• $a_{sd,\text{required}}$ — the Table 5.4.1 Note 1 corner-bar side axis distance, $a_s + 10$ mm (beams only, see Corner-bar side axis distance); $a_{sd,\text{min}}$ is the shallowest corner bar’s measured distance to the side face

The governing ratio is the larger of the applicable ratios. A governing ratio ≤ 1.0 is a PASS; > 1.0 is a FAIL. Slabs have no tabulated minimum width, so ratio_b does not apply to them, and ratio_asd applies only to beams that meet Note 1’s conditions.

:::caution[ratio_asd can fail while ratio_a passes] $a_m$ is an area-weighted average over all the longitudinal steel, so a single shallow corner bar can be masked by deeper or larger bars elsewhere in the section. Note 1 is a per-bar requirement on the corner bars alone. The two are reported as separate ratios for exactly this reason — read ratio_asd on its own, not as a stricter version of ratio_a. :::

:::note[Compliance margin, not capacity margin] The ratios are compliance margins, not capacity margins. The Section 5 tables are a step function calibrated at discrete fire-resistance periods, so ratio_a = 0.9 does not mean 10% spare structural capacity — it means the section is 10% below the next discrete table entry. Small changes in axis distance near a table boundary can shift the verdict without a corresponding change in structural capacity. :::

Average axis distance $a_m$ (Cl 5.2.1)

The average axis distance is an area-weighted mean over the longitudinal steel that has at least one fire-exposed surface within measuring distance:

where $a_i$ is the perpendicular distance from bar $i$ to the nearest fire-exposed surface of the section and $A_{si}$ is the bar’s cross-sectional area. Bars located on unexposed edges are excluded from the sum per Cl 5.2.1.

:::note[Beams average the bottom reinforcement only] Cl 5.2.1 gives the formula; Cl 5.4.1(b)(iii)(B) decides which steel a beam sums it over — “the average axis distance to the longitudinal bottom reinforcement”. Table 5.4.1’s legend scopes the companion quantity the same way (“b = width of the beam at the centroid of the bottom reinforcement”). So for a beam, ACS averages the steel below the section’s mid-height and leaves the top steel out. Columns (Cl 5.6.3), slabs (Cl 5.5.2) and walls (Cl 5.7.2) carry no such scoping and are averaged over all their longitudinal steel.

The difference is not cosmetic. Top bars are usually larger and further from the exposed surfaces — worst of all where the top face is shielded by a slab, the very case Cl 5.4.1(b)(i) addresses, because a top bar’s nearest exposed face is then a side face. On a 300 × 600 beam at FRL 90 with three N20 bottom bars 25 mm from the soffit and two N32 top bars, including the top steel lifts $a_m$ from 25 mm to 47 mm and turns a genuine 1.60 FAIL into a 0.85 PASS.

Excluded steel is still measured and listed in the Cl 5.2.1 derivation table, marked as not entering $a_m$ (the report’s In a_m column, and ”— not in a_m” in the panel), so an item left out of the average stays distinguishable from one you never placed. A tendon is partitioned the same way: one draped low at the section under assessment is bottom reinforcement, one draped high is not. :::

:::note[Reinforcement and tendons are averaged separately] Cl 5.2.1 permits this average where the steel “consists of either reinforcement or prestressing tendons with the same characteristic strength”. A section holding both therefore gets two averages, each weighed against its own requirement: the reinforcing bars against the tabulated $a_s$, and the tendons against $a_s$ plus the Cl 5.3.3 increase. The panel shows them as separate rows — $a_m$ for the bars, $a_{m,p}$ for the tendons.

This matters because a single blended average lets one family compensate for the other: a deep bar can average a shallow tendon back into apparent compliance even though the tendon carries the higher requirement of the two. :::

If the section has no longitudinal steel with a positive area, or has no fire-exposed surface to measure from, the average axis distance cannot be computed and the assessment returns NOT ASSESSED. A beam whose steel all sits above mid-height returns NOT ASSESSED for the same reason: there is no bottom reinforcement for Cl 5.4.1(b)(iii)(B) to average, and the Section 5 beam tables address the bottom reinforcement of a simply-supported or continuous beam.

Corner-bar side axis distance (Table 5.4.1 Note 1)

Note 1 under Table 5.4.1(A) and Table 5.4.1(B) imposes an extra requirement on beams:

In beams with only one layer of bottom reinforcement, the axis distance to the side of the beam for the corner bars including tendons or wires, shall be increased by 10 mm, except, where the value of b is greater than that given in Combination 4, no increase is required.

The reason is geometric: a corner bar is heated from two faces at once, so it runs hotter than a mid-face bar at the same axis distance, and an area-weighted average cannot see that. Where the beam has a second layer of bottom reinforcement the inner layer shields the corner, and the note does not apply.

ACS applies the increase when all three conditions hold:

The third condition is a strict comparison: a beam exactly as wide as Combination 4’s $b$ still owes the increase. Where a table row has no Combination 4 at all — Table 5.4.1(B) tabulates only two combinations below FRL 120 — there is no width to exceed, so the increase always applies.

:::note[Per-edge cover] The corner test compares each face’s distance past that face’s own cover, not the raw distances. Under a single cover the two are the same question; when you set a different cover on the soffit than on the sides, they are not. A bar seated in the tie’s bend sits the same amount past each of its two covers, so it is recognised as a corner bar whatever the two covers are.

Because the test is measured against cover, it cannot be answered without one. Inside ACS the cover always comes from your section, so this never arises; a caller of the fire_capacity API that omits cover gets the conservative reading — corner bars assumed present — and a disclosure saying so, rather than a silent “this beam has no corner bars”.

When cover is supplied with per-edge values — a different cover on the soffit than on the sides, for instance — those values are now propagated through to the full fire capacity calculation unchanged. Prior to 2026-07, the stateless capacity route collapsed per-edge cover to a single value before the fire check; results for sections with non-uniform cover from that route will differ.

The result states which of the three it reached, as cornerClassification: corner-bars-found, none (a determined absence, measured — the note is switched off and this beam owes no increase), or undetermined (the question could not be asked, so the note was applied on the conservative assumption). It is absent where the corner set is not what decided the note — a non-beam, $b$ greater than Combination 4’s, or more than one layer of bottom reinforcement — so an absent value never means none. Where no $a_{sd}$ ratio was produced, the results panel and the report state the classification in words rather than leaving the check silently absent. :::

The requirement is measured to the side face specifically — not to whichever of the corner’s two faces happens to be nearest, which under unequal cover is the soffit. It raises the required cover for the corner bars, and the Cover panel discloses it as its own line so the +10 mm is traceable back to the note:

Axis distance a_s 65 mm (AS 3600:2018 Table 5.4.1(A), b = 200 mm, FRP 120 min)
Corner-bar side axis distance a_sd 75 mm = a_s 65 + 10 (Note 1 — one layer of
  bottom reinforcement, b not greater than Combination 4)
Fire cover 55 mm = a_sd 75 − fitment Ø10 − ½ × Ø20

Before reinforcement is placed, ACS cannot yet establish whether the section will have corner bars in a single layer, so it applies Note 1 on the conservative assumption that it will and shows a disclosure saying so. The advised cover tightens once the bars exist and the conditions can be checked.

Fire load level $\mu_{fi}$ (columns and walls)

For columns and walls, the Section 5 tables are subdivided by the fire load level $\mu_{fi}$, which is the ratio of the fire combination axial demand to the section’s ambient axial capacity at the design eccentricity:

where $N^*_f$ is the axial force from the fire load combination and $\phi N_u$ is the axial capacity from the ambient interaction surface at the eccentricity of the fire combination — not the squash load, because the squash load overstates capacity whenever a moment is present.

When no fire load combination is defined: ACS falls back to $\mu_{fi} = 0.7$, the most onerous tabulated level. This is conservative and may require a larger axis distance than the actual design needs. A disclosure is shown in the result to flag this. Defining a fire load combination with a compressive axial design action will allow ACS to compute $\mu_{fi}$ from your actual loads.

$\mu_{fi}$ is a table selector for columns and walls only. For beams and slabs the table is keyed on the support condition, not the load level — a missing fire combination does not affect the beam or slab assessment and does not raise the disclosure.

Tendons (Cl 5.3.3)

When the section contains prestressing tendons, AS 3600 Cl 5.3.3 requires the axis distance read from the Section 5 table to be increased for those tendons:

ACS checks this as a separate ratio, alongside the reinforcing bars’ own check:

where $a_{m,p}$ is the Cl 5.2.1 average taken over the tendons alone. It can govern the verdict while the bar ratio sits comfortably below 1.0, so read it in its own right.

:::caution[The increase applies to the tendons, not to the section] Cl 5.3.3 raises the requirement for prestressing tendons. It is not a section-wide uplift of the tabulated value, and it does not enter the code-based cover — a tendon is positioned by its own profile, never off the cover + fitment + ½·d_b chain, so its requirement cannot be inverted into a cover for the reinforcing bars.

A practical consequence: adding or removing a tendon does not change the cover the Cover panel advises. If you are checking a PT section against an earlier ACS release, note that versions before this behaviour landed did raise the advised cover when a tendon was present, and attributed the result to fire when durability governed. :::

Which increase applies

ACS reads the arm from the product form on the tendon’s material — strand, wire, or bar. Strand and wire share arm (a); only bar tendons take arm (b). The applied clause and its arithmetic are named in the result disclosures, so a result always states which of the two it used and how it got there.

A section may hold both products. Cl 5.3.3 reads per tendon, and Cl 5.2.1 will not average across steels that do not share a characteristic strength, so each arm is checked separately — its own $a_{m,p}$ against its own requirement. The governing one is reported as $\text{ratio}_{ap}$, and both carry their own disclosure.

:::caution[An undeclared product form takes the more onerous arm] A tendon with no catalogue material selected — or one saved before ACS recorded the product form — has nothing to read. ACS then applies Cl 5.3.3(a)‘s +15 mm, the more onerous of the two, and adds a disclosure saying the type was assumed. Select a catalogue material on the tendon to have the requirement come from your actual product; a bar-tendon design assessed this way is being asked for 5 mm more axis distance than AS 3600 requires of it. :::

The beam tables assume a protected upper surface (Cl 5.4.1)

Table 5.4.1 as reached through Cl 5.4.1 is conditional. The clause applies to a beam that:

has the upper surface integral with or protected by a slab conforming with Clause 5.5

Three-side exposure is therefore the premise of the beam table, not an incidental configuration. A beam that can be exposed on all four sides is governed by the separate Cl 5.4.2 route, which uses the same tables but adds three further requirements: the total depth $D$ must be no less than the least $b$ tabulated for the fire-resistance period, the concrete area must satisfy $A_c \ge 2b^2$, and $a_m$ must be determined using the beam’s minimum dimension for $b$ and taken over all longitudinal reinforcement and tendons — not the bottom steel alone.

Because a protected upper surface is the ordinary configuration for a beam in a floor system, selecting the AS 3600 Section 5 method for a beam or slab marks the section’s upward-facing outline edges as not exposed, once, at the moment you select it. The change is visible — the canvas repaints and the exposed-element count drops — and a single undo reverses it. It is applied on selection only, never re-imposed afterwards, so a deliberate decision to expose the top face is never silently overwritten.

ACS identifies the upper surface from each edge’s outward normal, not from its height. On a T-section that selects the flange top while leaving the flange soffits and the flange tips exposed, which is exactly the configuration Cl 5.4.1 describes — the bounding-box top alone would also catch the tips.

Beams exposed on all four sides (Cl 5.4.2)

If you re-expose the upper surface, Cl 5.4.1(b)(i) is no longer satisfied, and ACS moves the assessment to Cl 5.4.2 rather than continuing to report the three-side reading. Three things change, all in the conservative direction:

:::note[Results change in 2026-07] Prior to 2026-07, a four-side-exposed beam was assessed on the three-side Cl 5.4.1 route rather than being moved to Cl 5.4.2. Results for four-side-exposed beam configurations will differ from earlier versions — specifically, the table entry point (minimum section dimension rather than minimum web width), the averaging scope for $a_m$ (all longitudinal steel, not bottom steel only), and the two added proportioning ratios (Cl 5.4.2(i) and (ii)). :::

• The table is entered on the beam’s minimum dimension, not its minimum web width. For a beam that is deeper than it is wide these are the same number and nothing moves. For a wide, shallow beam they differ by the aspect ratio, and the web width is the less onerous of the two: a 900 × 250 beam at FRL 90 reads $a_s = 35$ mm on $b = 900$ and $a_s = 45$ mm on $b = 250$. The advised cover follows, so such a beam is now detailed deeper.
• Two proportioning requirements are added and reported as their own rows: $D \ge$ the least $b$ tabulated for the period (Cl 5.4.2(i)), and $A_c \ge 2b^2$ (Cl 5.4.2(ii)). The area requirement genuinely binds — a near-square beam sitting at its least tabulated $b$ fails it.
• The axis-distance requirement extends to all longitudinal steel. Cl 5.4.2(iii) applies it to all longitudinal reinforcement and tendons, because a beam heated on four sides heats its top steel too. ACS keeps the bottom reinforcement’s $a_m$ as the governing average and weighs the remaining longitudinal steel as its own Cl 5.2.1 average against the same requirement, reported as $a_{m,\text{other}}$. The two are kept apart deliberately: one blended average would let deep bottom bars raise $a_m$ and report shallow top steel as compliant.

:::note[Cl 5.4.2 result disclosures — changed in 2026-08] The Cl 5.4.2 “gate-two reading” — the disclosure that identifies which value of b ACS used to check the $A_c \ge 2b^2$ area requirement (Cl 5.4.2(ii)) — appears only when that gate produced a result: specifically on a BelowTabulatedMinimum Fail and on the final Pass or Fail. A NOT ASSESSED verdict does not carry the gate-two reading.

NOT ASSESSED results carry a RouteDisclosure instead, which names which route was taken and why assessment stopped. Prior to 2026-08, the gate-two reading appeared alongside NOT ASSESSED results — beside null values for the area ratio — which has been corrected. :::

The clause substitution is stated on the compliance result, in the report, and beside the advised cover — $b$ in the results table is the minimum dimension on this route, so it will not match a hand lookup on the beam’s width.

:::caution[Two configurations have no tabulated route at all] Cl 5.4.2 covers “a beam of approximately rectangular cross-section”, and it is a beam clause. Two cases therefore fall outside Section 5 entirely, and ACS refuses them rather than borrowing a clause: no compliance ratio, and no code-based cover.

• A non-rectangular beam exposed on all four sides — a T, I or L section, or a thin-walled box. Cl 5.4.1 needs its upper surface protected and Cl 5.4.2 needs it approximately rectangular, so neither reaches it. ACS treats a section filling less than 95 % of its bounding box as not approximately rectangular, and states the measured figure in the refusal.
• A slab with its upper surface exposed. Cl 5.5.2 tabulates a slab heated from beneath, and Cl 5.4.2 does not extend to slabs.

In both cases, mark the upper surface unexposed if it is in fact protected, or assess the member by choosing a different analysis method. :::

Shielded faces and the advised cover

Section 5 tabulates an axis distance to a fire-exposed surface (Cl 5.2.1), so a face you have marked as not exposed carries no fire requirement. The code-based cover follows that: the fire term applies only to fire-exposed faces, while the durability (Cl 4.10.3) and placement (Cl 4.10.2) terms are section-wide and apply to every face.

A section with a shielded face therefore resolves to a per-edge cover — the exposed faces on the fire-governed value, the shielded ones on max(durability, placement). The Cover panel states which faces the fire term reached and the governing term for each edge, and locks the cover mode to Per-edge while that derivation is live so a switch to Uniform cannot discard it. Switching to Manual cover, choosing another fire method, or exposing every face releases the lock.

The Cl 4.10.1 fire term is attributed face-by-face: each edge’s fire cover is derived from the axis distance applicable to that specific exposed face. Results for sections with mixed exposure — some faces fire-exposed, some shielded — may differ from versions prior to 2026-07, where the term was not attributed to the specific face of the section.

With no face exposed, Section 5 asks nothing of the member: no fire term is applied, and the cover resolves on durability and placement alone.

Fire exposure configuration

Open the Fire tab in the right panel to configure the heat transfer and fire capacity analysis:

:::note[Empty fire scenario prompt — September 2026 (#5438)] When the Fire tab has no fire scenario defined — no fire exposure has been added yet — the panel now displays an informative prompt explaining what the fire analysis does and how to get started. Previously the panel showed a blank state with no guidance. The prompt disappears once a fire scenario is added and the configuration inputs appear. :::

Fire duration

Set the required fire exposure duration in minutes. Quick buttons provide common values (30, 60, 90, 120, 180, 240 min); the picker opens at 60 minutes — the first standard FRL value and a valid entry for both the tabular cover and heat transfer + capacity analysis paths. The maximum supported duration for the heat transfer + fire capacity analysis is 360 minutes.

AS 3600 Section 5 tabular cover is limited to FRLs up to 240 minutes. When the fire duration exceeds 240 minutes, the Cover panel shows Resolved = false together with a failure reason identifying which required FRL falls outside the tabulated range. The tabular method cannot produce a compliant value for an out-of-range FRL — ACS refuses rather than silently using the 240-minute axis distance, which would understate the actual cover requirement. The heat transfer and fire capacity analysis is unaffected: it continues to run for any duration from 1 to 360 minutes regardless of whether the tabular cover resolves. For members requiring fire resistance beyond 240 minutes, use the heat transfer + fire capacity analysis to demonstrate structural adequacy by calculation.

Fire curve

Select the time-temperature relationship:

:::note[500°C isotherm method and fire curve scope — EN 1992-1-2 B.1.1(1)] EN 1992-1-2 Annex B.1.1(1) qualifies the 500°C isotherm method for standard fire exposure (ISO 834). When you select ASTM E119 or the hydrocarbon curve with the isotherm method, ACS computes the temperature field using your chosen curve but surfaces this scope note: the method’s calibration was performed under ISO 834, and its conservatism is not validated for non-standard fire curves. The advanced-fibre method (EN 1992-1-2 §4.3) integrates the full temperature field and carries no equivalent curve restriction. :::

ASTM E119 is evaluated as a closed-form fit, not from ASTM E119-20 Table 1. It runs 3.4 °C cooler than the tabulated curve at 60 min, 2.5 °C cooler at 120 min and 17.4 °C hotter at 240 min. The R240 difference is in the conservative direction and is too small to change a design outcome — see the fire curves section of the theory page for the equation, the full comparison and the measured effect on bar temperature, 500 °C isotherm depth and effective section.

Exposed elements

By default, all external edges of the section are exposed to fire. Toggle individual edges on or off to model:

• Three-sided exposure: typical for beams with a slab on top (top edge unexposed)
• One-sided exposure: typical for walls or slabs exposed on one face only

Internal void edges are controlled separately. The thermal boundary condition for each cavity is set in the Void Cavities section of the Fire panel:

• Sealed (default) — adiabatic, no heat loss. A sealed cavity has nowhere to shed heat, so treating it as an ambient sink would compute a cooler wall than the physics supports. Adiabatic is the conservative choice of the two boundary conditions — it is not an upper bound on the section, because heat transfer across a cavity is not modelled at all.
• Ventilated — ambient convection at $h_c = 9$ W/m²·K, 20 °C, the same boundary as an unexposed outline face. Use this only for a cavity with real, sustained airflow during the fire. Ventilated is the unconservative direction — the cavity wall runs cooler, leaving a larger effective section after the 500 °C carve — and rests on your assertion that airflow persists for the full exposure duration. A cavity whose edges are already declared fire-exposed cannot be set to ventilated.

Heat transfer across a cavity is not modelled, under either boundary kind. Both are conditions on the cavity wall: neither carries heat from the fire-side wall to the shielded wall, and surface-to-surface radiation across the cavity is absent. So the shielded wall of a cavity computes colder than reality while the fire-side wall computes marginally hotter, and which way the capacity moves depends on where the governing bar or tendon sits — a bottom-flange strand under a void is on the safe side of this; a bar in the shielded wall of a box column, or a compression zone above a hollow core, is not. Thin-walled box and hollow-core sections are where it bites. There is no setting for it: declaring the void ventilated does not address it. The editor and the report both disclose it whenever the section has a void.

:::note[Cavity radiation ceiling — September 2026 (#5463)] A cavity radiation ceiling parameter is now available for void cavities. When set, it caps the radiation contribution to heat transfer at the cavity boundary, independently of the convection setting. This is useful for sealed cavities where the radiation exchange between opposed walls is limited by geometry or emissivity. The ceiling applies to the cavity boundary only; it does not affect heat transfer at external exposed faces. Leave the field empty to use the uncapped default behaviour (equivalent to pre-September 2026 releases). :::

See Boundary conditions for the full rules and the remaining limitations of the model.

The exposed-edge configuration affects both the heat transfer analysis and — for columns and walls — the number of exposed sides used to look up the Section 5 axis distance.

Aggregate type

Select the aggregate type used in the concrete mix. This input is active when the design code is AS 3600 or EN 1992-1-2; it has no effect under ACI 318-19. AS 3600 Cl 5.3.1(b) Note delegates elevated-temperature material properties to EN 1992-1-2 Tables 3.1, 3.2a and 3.3, so AS 3600 fire calculations use the same siliceous/calcareous distinction as EN 1992-1-2.

The aggregate type affects the concrete strength reduction factor $k_c(\theta)$ used in the fire capacity calculation. Calcareous aggregate concrete generally provides better fire performance.

Refused when unrecognised. The endpoint maps aggregateType to its EN 1992-1-2 table column by exact token match. A value outside siliceous / calcareous — including a saved value that no longer parses — returns aggregate_type_invalid (400) rather than falling through to the siliceous table. If the design option was last saved with an unrecognised aggregate type, open the Fire panel, reselect the aggregate type, and re-save before retrying the capacity route.

:::note[Aggregate type is persisted on the design option — August 2026 (#5181, #5180)] The aggregate type selection is persisted as part of the design option and read back by all fire capacity routes. Prior to the August 2026 release, the selection was held in local session state only and was lost when the panel was closed or the page was navigated away from; the route then resolved to the siliceous default regardless of what was displayed. The fix landed in two stages: #5180 added the persisted column, and #5181 wired the Fire panel to save and reload the aggregate type on every design-option write. Agents and scripts that previously supplied aggregateType on each capacity request for reliability may continue to do so — the passed value takes precedence over the persisted one — but the persisted value is now a correct fallback. :::

Steel class

Select the steel class for the reinforcement. This input is active when the design code is AS 3600 or EN 1992-1-2; it has no effect under ACI 318-19. AS 3600 Cl 5.3.1(b) Note adopts the EN 1992-1-2 material tables for elevated-temperature steel properties, including the distinction between hot-rolled and cold-worked reinforcing bar.

For most reinforced concrete sections, use hot-rolled. Select cold-worked only when the reinforcement was specified as cold-worked deformed bar. Cold-worked steel generally results in a lower (more conservative) fire capacity.

Refused when unrecognised. The endpoint maps steelClass to EN 1992-1-2 Table 3.2a (hot-rolled) or Table 3.2b (cold-worked) by exact token match. A value outside hot_rolled / cold_worked — such as cold-worked (hyphen), COLD_WORKED (uppercase), or coldworked (fused) — returns steel_class_invalid (400) rather than falling through to the hot-rolled table. The substitution blocked here is ~30 % unconservative in reinforcement yield strength at 500 °C ($k_y = 0.78$ hot-rolled vs $0.60$ cold-worked from Table 3.2a vs 3.2b), and would be undetectable because the response echoes the caller’s token verbatim in its disclosure.

The same refusal applies to supportCondition: values outside simply-supported / continuous are refused rather than defaulted to simply-supported. See API error codes for the full code listing.

Refused above 500 MPa reinforcement (AS 3600). Where the reinforcement declares $f_{sy} > 500$ MPa on the AS 3600 path, ACS refuses the fire check with fire_reinforcement_production_route_undeclared (400). The Table 3.2a factors are not calibrated to a grade — they are normalised ratios selected by the bar’s production route — but nothing on a higher grade records which route it is, and the analysis default (hot-rolled) is the more favourable column: at 700 °C the two retain 23 % and 12 % of ambient yield respectively. Rather than publish a fire capacity that may assume nearly double the residual strength the bar actually has, the check is withheld. Record the manufacturer’s production route for the grade, or run the fire check with 500 MPa reinforcement. ACI 318 and EN 1992-1-1 sections are unaffected.

:::note[Aggregate type and steel class disclosed in the fire capacity panel — September 2026 (#5291)] The fire capacity panel now shows which EN 1992-1-2 table columns (aggregate type and steel class) governed the tabulated strength-reduction values — the same disclosure the PDF report has included since launch. Select Siliceous or Calcareous aggregate and Hot-rolled or Cold-worked steel class in the Fire panel; the panel now confirms which of the four table column combinations was read for the displayed result. :::

Analysis method

Select the method for computing fire-reduced capacity:

:::note The 500°C isotherm method is the default because its conservatism is calibrated: the “full ambient strength below 500°C” overestimate is deliberately paired against the “zero strength above 500°C” underestimate, and EN validated the pairing. The advanced-fibre method is more rigorous but returns the higher capacity of the two and is substantially slower, so it is a deliberate opt-in rather than the default.

Annex B.1 carries an applicability limit on minimum section width, and ACS enforces it — see Minimum section width below. :::

Minimum section width (EN 1992-1-2 Table B1)

EN 1992-1-2:2004 B.1.1(2) states that the 500°C isotherm method “is valid for minimum width of cross-section given in table B1”. That is a hard limit on where the method may be used, not a recommendation, so ACS checks it before running the analysis and refuses rather than returning a number from outside the method’s declared scope.

How the width is measured. ACS measures the narrowest limb of the section across which fire can attack from two opposing faces — the thickness that has to survive being heated from both sides at once. That is the quantity Table B1’s minimum widths were calibrated against, and it is not the same as the section’s overall size. A 600 mm-wide T-beam on a 150 mm web is assessed on the 150 mm, and a hollow box on its wall rather than on its envelope.

Exposure is part of the measurement. A face you have marked unexposed is not a fire front, so a T-beam whose flange is covered by a slab is assessed on its web, not on the flange — the flange is only heated from below. Mark that same flange exposed and it is measured too. A section with no exposed faces at all has nothing to measure and is not gated.

:::note[Continuum width measurement — changed in 2026-08] The Table B1 minimum-width check now uses a continuum estimate of the maximum exposure distance — a compass-search from the raster argmax — rather than pixel-centre sampling. This eliminates raster-dependent verdict flips for sections near a tabulated threshold: a section whose narrowest confronted limb was incorrectly admitted at the production raster resolution due to pixel placement is now refused correctly.

This change invalidated all cached fire-capacity results; ACS recomputes them on next access. Results for sections near Table B1 thresholds computed before 2026-08-09 may differ from those computed after. :::

:::note[Continuum cool-core determination for wall-like sections — changed in 2026-08] For wall-like sections (those with two opposing fire-exposed faces), the existence of a cool core — concrete remaining below 500°C — is now determined via the same continuum compass-search approach as the Table B1 width check, rather than pixel-centre grid sampling. A near-threshold wall whose pixel-centre grid produced a marginal cool-core verdict is now checked continuously against the isotherm field. Results for near-threshold wall geometries computed before 2026-08-09 may differ from those computed after. :::

:::caution[No opposing faces means “not evaluated”, which is not “passed”] A member with no pair of opposing exposed faces — a slab exposed on its soffit alone is the ordinary case — has no confronted limb, so there is no thickness for Table B1 to bind on and the limit is not applied. ACS says so on the result rather than letting the absent check read as a passed one.

This depends on exposure alone, not on the section’s size or shape: a 300 × 200, a 500 × 200 and a 2000 × 200 soffit-exposed member all reach the same outcome at R120.

Table B1’s limit bounds the error where two fronts meet and burn a limb through. The check is inapplicable here — not satisfied — and the method still takes full strength over the whole cool zone, with concrete a little below 500°C retaining nearer 60% of its ambient strength. A section the 500°C carve leaves with no cool concrete at all is refused outright, which is what happens to a soffit-exposed slab shallower than the 500°C depth for its period. :::

:::note[A tapering limb is measured, not exempted] What has to survive being heated from two sides at once is the thickness of the largest circle that fits inside the limb, and for a taper that is not the width you would read off the drawing: for a triangle it is confined to between two thirds of the minimum width and the full width, and equals exactly two thirds for an equilateral section. ACS scores that inscribed thickness against the tabulated row, so a wedge or tapered web whose overall width clears the row can still be refused — and should be. An equilateral pier 165 mm across at R120 is scored at 110 mm against the 160 mm row and refused; at that lower bound the member is entirely above 500°C while its width still clears the table, at every period the table covers. :::

A void is treated as a fire front even when its own surface is unexposed. EN 1992-1-2 says nothing about cavities, the study behind Table B1 contains no cavity section, and a sealed cavity is not obviously a heat sink — so ACS takes the cautious reading and measures the wall. A hollow section may therefore be refused where a solid one of the same wall thickness would not.

:::note[Void geometry in the 500°C isotherm carve] The holes in the carved effective section come from the 500°C contour, not from the void you drew. The carve traces every boundary of the cool region and keeps the enclosed ones as holes, so a hollow section is analysed as one region with a hole in it — the void’s own boundary is never counted as a second, disconnected piece.

On the fire-design path every cavity is sealed and adiabatic, so no fire front runs inside it and the traced hole follows the drawn void closely; the difference is the contour’s own resolution rather than a thermal effect. A hole materially larger than the drawn void arises only where a void surface is genuinely driven as a fire front — the concrete ringing it has then itself passed 500°C. That case is no longer reachable from any surface: it required exposedVoidEdges on the raw POST /api/v1/fire/heat-transfer endpoint, which was retired in August 2026 (#5040). A void can still be declared sealed or ventilated in the Fire panel; individual void edges can no longer be declared fire-exposed. Where it applies, the effective cool area is smaller than a gross-area-minus-drawn-void calculation would suggest, and ACS uses the carved holes rather than the authored ones so fire-damaged concrete is never counted at full $f_{ck}$. :::

The measurement does not depend on how the section is drawn: rotating a member gives the same answer, and a circle measures its diameter whether you drew it with arcs or as a many-sided polygon. A rounded or chamfered corner is not read as a thin spot.

Because the width is measured from your geometry rather than declared, the check carries a small tolerance and applies it in one direction only: it admits a section up to about 1% of the tabulated width narrower than the row — around 1.6 mm at R120 — and never refuses one that meets it. That is well inside the ±5–10 mm a section is actually built to, and inside the 25–40% steps between Table B1’s own rows.

Fire durations between the tabulated periods round up to the next period in the table — a 110-minute design is checked against R120’s 160 mm, not R90’s 120 mm, and values are never interpolated. A duration below 60 minutes is checked against R60’s 90 mm, the narrowest row the table carries. Above R240 the table states no minimum width at all, so the method cannot be applied and ACS refuses on that basis instead.

Where there is nothing to measure, ACS says so — and does not refuse. A section no limb of which is attacked from two opposing faces — a slab exposed on its soffit alone is the ordinary case — has no width for Table B1 to bind on at any thickness. The analysis runs, and the result states that the width limit was not evaluated: inapplicable is not the same as satisfied, and a silent pass would leave you believing a check succeeded that never ran. It is not a claim that the answer is exact — the 500°C isotherm method still takes the full $f_{ck}$ over the whole cool zone, and concrete a little below 500°C retains nearer 60% of it — only that this section is not exposed to the cliff below.

:::caution[Why this is a refusal and not a warning] At and above the tabulated width the 500°C isotherm method is accurate to about ±2%. Below it the accuracy does not taper off — it collapses, and the result becomes a step function of the geometry, where a 1 mm change to a web thickness can move the computed capacity by a large factor. The error is in the conservative direction, but a number that jumps like that is not one to design against, so ACS declines to produce it.

Switch to the advanced-fibre method (EN 1992-1-2 §4.3), which integrates the temperature field over the section directly and carries no Annex B width limit. :::

:::note[Parametric fire exposure] Table B1 has a second part keyed on fire load density, for parametric fire exposure with an opening factor $O \ge 0{,}14\ \mathrm{m}^{1/2}$. ACS does not offer a parametric fire curve — the available curves are ISO 834, ASTM E119 and the hydrocarbon curve — so that part of the table is not applicable to any analysis you can set up here. :::

:::caution[When the isotherm method refuses] Several things can stop a 500°C isotherm run before it produces a capacity.

The section is outside Annex B.1’s field of application — see Minimum section width above. This is the common one. The refusal names your section’s own width against the width the table requires, so you can see how far short it falls.

The 500°C contour separated the remaining concrete into disconnected pieces. ACS analyses what survives the carve as a single effective cross-section, and Annex B.1 describes no rule for one in several pieces — so there is nothing to analyse, and no dimension to blame: the section already met the width limit. The refusal states the number of disconnected regions and the net area of each (area of the region less any interior holes), so you can see which piece is dominant and decide whether to simplify the geometry. This is uncommon; a cool region normally separates only once a limb has burned right through, and such a limb is usually already below the Table B1 minimum, so the width check catches it first. An authored void is not a separate piece — a hollow section is analysed as one region with a hole in it, and the void’s own boundary is never counted as a second region (see the note on void geometry above).

Nothing is left below 500°C. At long durations a thin or tapered section can be heated right through, and Annex B.1 then reduces it to nothing: the method discards everything above 500°C and takes full strength below it, so with no concrete below 500°C there is no effective section to analyse. ACS does not fall back to the capacity of the bare reinforcement. Bare bars with no surrounding concrete are a different structure rather than a reduced one — there is no bond left to develop the bar forces, and nothing restraining the longitudinal bars against buckling between the ties — so a surface computed from the reinforcement alone would assume confinement the section no longer has. The refusal names the duration and the gross concrete area.

The convex-corner allowance exhausts the cool core. At exposed convex corners — where two fire faces meet — the concrete heats from two directions simultaneously and the 500°C isotherm sits closer to the corner than it does along a straight face. ACS applies the EN 1992-1-2 convex-corner axis distance allowance only when the geometric wedge between the two fire-exposed faces is physically present in the section — that is, only where a real concrete wedge sits at the junction of those faces. If a chamfer, arc or section boundary has removed the corner material, no wedge exists and the allowance is not applied. On a thin section or at a long fire duration, the corner correction can remove the last remaining cool concrete even where the section otherwise clears the Table B1 minimum width. When it does, ACS refuses on the same basis as the “nothing below 500°C” case — no effective section remains to analyse — rather than ignoring the corner effect and returning an overstated capacity.

The 500°C contour produced a self-intersecting outline or a self-intersecting hole, or could not be closed into an outline at all. This can arise from complex re-entrant geometry or from opposing thermal fronts converging to a near-zero residual. There is no dimension at fault here either — the contour simply cannot be expressed as an effective section.

A section with no two opposing exposed faces does not hit the width refusal — the width limit is not evaluated for it at all, and it analyses with the disclosure described above. It can still hit the others. A slab exposed on its soffit alone is the ordinary case, and a thin one at a long period — say 100 mm at R240 — is heated right through, so it passes the width check and is then refused for having no concrete left below 500°C.

In every case, switch to the advanced-fibre method (EN 1992-1-2 §4.3), which integrates the temperature field over the section directly and needs no carve. Every isotherm refusal carries a Switch to the advanced fibre method button: pressing it changes the design’s stored fire analysis method, so the selection in the Settings panel changes with it and the change is saved. ACS never makes that switch for you — the method is your input, and substituting one behind the scenes would leave you reading results from a method you did not choose. :::

:::caution[Removed in 2026-07] A third mean-temperature tier was previously offered and was the default. It applied a single area-weighted mean $k_c(\bar{\theta})$ over the full, unclipped section — it was not the EN 1992-1-2 Annex B.2 zone method and did not divide the section into zones, despite once being documented that way. Because it retained fire-damaged perimeter concrete at mean strength its conservatism was not guaranteed: measured on a 400 × 600 mm column at 90 min four-sided ISO 834 exposure ($f'_c$ = 32 MPa) it overstated the isotherm squash load by 12.5 %. It has been removed, and designs that used it now resolve to the 500°C isotherm. :::

Heat transfer analysis

When the Fire Heatmap tab is selected on the canvas, ACS runs a 2D finite element heat transfer analysis to compute the temperature field through the cross-section at the specified fire duration.

The heatmap shows temperature contours using a colour gradient from ambient (blue) to the fire temperature (red). You can observe:

• Temperature penetration depth from exposed surfaces
• Corner effects (corners heat faster due to two-sided exposure)
• The temperature at each reinforcement bar location

Hovering anywhere on the heatmap shows a tooltip with the temperature at that point and its engineer Y-up coordinate — the $y$ value is measured upward from the section’s geometric reference, matching the section geometry inputs, not from the top of the canvas viewport.

:::tip[Fire panels refresh automatically after saving — August 2026 (#5165)] The heat transfer heatmap and the fire capacity results now update automatically whenever you save the section. Previously, fire results could reflect the state of the section before the last save; navigating away and back or manually re-running the analysis was required to see results for the saved geometry. No manual re-run is needed after saving. :::

The heat transfer model uses:

• Thermal conductivity, specific heat, and density of concrete as temperature-dependent properties (per EN 1992-1-2 Annex A)
• Convective and radiative boundary conditions on exposed surfaces — the convection coefficient $h_c$ is derived from the selected fire curve per EN 1991-1-2: $h_c = 25$ W/m²·K for the standard fire (ISO 834 / ASTM E119, §3.2.1) and $h_c = 50$ W/m²·K for the hydrocarbon curve (§3.2.3)
• Ambient convection on unexposed outline surfaces, where continuous construction beyond the face genuinely acts as a sink at roughly ambient
• Adiabatic (zero-flux) boundaries on unexposed void surfaces — a sealed interior cavity has nowhere to shed heat, so treating it as an ambient sink would compute a cooler wall and an unconservatively large effective section
• The section boundary you drew, curves included — a curved soffit, a circular column or a circular duct is meshed on its arc, using the same discretisation the fire capacity check and the section-property engines use, so one section has one meshed geometry on every surface

:::caution[Heat transfer results changed for curved sections — August 2026 (#4872)] Before this release the thermal mesh was built on the straight-line chord between each pair of arc endpoints, not on the arc itself. A section with any curved edge was therefore analysed on a slightly different shape from the one drawn — and from the one the fire capacity check displayed beside it was already using. A fully circular section could not produce a heatmap at all.

Curved sections now mesh on the curve. Re-run any heat transfer or fire capacity analysis on a section with a curved edge; a section made only of straight edges is unaffected and will return the same numbers. Circular sections now produce a heatmap where the tab previously reported that it could not represent them.

Fire-exposure selections continue to name the edges you drew: marking a curved soffit as exposed exposes the whole curve. :::

:::caution[Heat transfer results changed for the hydrocarbon fire curve — August 2026 (#4899)] Before this release, ACS used a single convection coefficient for all fire curves. The hydrocarbon curve now uses $h_c = 50$ W/m²·K (EN 1991-1-2 §3.2.3) instead of the previously-applied standard-fire value of 25 W/m²·K — a significant change in the thermal boundary that will increase temperature penetration depth and bar temperatures for hydrocarbon-curve analyses. Re-run any heat transfer or fire capacity analysis that used the hydrocarbon fire curve. Analyses using ISO 834 or ASTM E119 are unaffected — their convection coefficient is unchanged at 25 W/m²·K.

Additionally, a constant in the heat-transfer disclosure output was corrected in August 2026 (#4907). Engineers who saved heat-transfer results before this update will see the corrected value when they re-run the analysis. :::

:::tip[New: the Fire Heatmap can now be zoomed] The Fire Heatmap previously rendered at a fixed fit-to-window scale with no way to get closer. It is now a full viewport: scroll to zoom, middle-drag or Ctrl (⌘) + left-drag to pan, and the house icon in the bottom-left corner to reframe the whole section.

This matters for reading a fire result rather than just looking at it — the 500 °C isotherm at a re-entrant corner, or the temperature at one individual bar, is often only legible zoomed in. Isotherms, bar markers and labels are drawn as vectors, so they stay sharp at any zoom; the temperature field itself is redrawn at full resolution a moment after you stop zooming. :::

The full control set, which is shared with every other canvas in the platform, is in Canvas navigation.

Mesh resolution

The finite element mesh is generated automatically from the section geometry. Near fire-exposed boundaries the elements are sized at 8 mm; in the cool interior they grow to a 25 mm cap, with a 1.3 grading ratio to keep the transition smooth. Corner refinement is always enabled.

The base element sizing is not user-configurable. Earlier versions of this page described a Max element area field in the Fire settings panel, with a mm²-to-element-size conversion table and cache-sharing semantics between the web UI and the API. No such field exists, and none of that behaviour was ever implemented — see #5155. The maxElementArea parameter it referred to was accepted only on the raw POST /api/v1/fire/heat-transfer endpoint and the CalculateHeatTransfer MCP tool, both of which were retired in August 2026 (#5040).

What does control the mesh, and is the engineer’s to set, is the design option’s arc discretisation tolerance — the sagitta tolerance used when a curved face is faceted. Both fire analyses read it from the saved design option, so the heatmap and the capacity check always mesh one section the same way.

Adaptive mesh refinement

The second control that is yours to set is the Adaptive Mesh Refinement section of the Fire Analysis panel. Enabling it re-solves iteratively, refining wherever the estimated thermal-gradient error is highest (a Zienkiewicz–Zhu error estimator with Dörfler marking), until the Error tolerance you set is met. The tolerance is entered as a percentage and is clamped to 1–20%; the default is 5%. It is off by default, and it is slow — budget up to five times the solve time of a single pass.

The section is not shown on the tabulated method, which looks up a table and builds no mesh at all (#3740).

:::note[The refinement setting now persists — August 2026 (#5162)] The adaptive refinement toggle and its error tolerance are saved with the design option and restored when you reopen it. Previously they were live inputs that were never written to the section, so both reset on every page load and any run you had tuned had to be re-tuned from scratch. Design options saved before this release carry no stored value and open with refinement off; set it once and save to pin it. :::

:::note[Adaptive refinement convergence corrected — September 2026 (#5366)] The Zienkiewicz–Zhu error estimator now computes a proper relative error — the estimated error divided by the reference field norm — rather than the raw absolute error it used previously. With an absolute error, sections whose thermal gradient norms are large relative to the tolerance threshold could never satisfy the stopping criterion regardless of mesh density, so refinement always ran to the maximum depth without converging. The relative error converges correctly: analyses that previously exhausted the refinement depth without meeting the tolerance will now converge at a coarser mesh. Results may differ slightly on sections that were affected — the converged result is more accurate than the max-depth result that preceded it. :::

Fire capacity check

The fire capacity analysis uses the temperature at each concrete fibre and reinforcement bar to compute reduced material strengths.

:::note[Scope] The fire capacity check covers axial-force and biaxial-moment (N–M) interaction only. Shear capacity, torsional resistance, and anchorage at the fire limit state (AS 3600 Cl 5.3.1(b)) are outside the scope of this analysis and must be verified separately using the applicable code provisions. :::

:::note[Degenerate thermal field — HTTP 500, not 400] If the heat transfer solver produces a malformed result — no mesh nodes, no elements, or a temperature array whose length does not match the node count — the fire capacity endpoint returns HTTP 500, not 400, and the metered API credit is refunded. This is an internal-fault classification: no input geometry you supply can produce this state (the mesher throws on a zero-element mesh before any temperature is computed). It is a platform fault, not a user error. If you see a 500 from a fire capacity call, retry; if it persists, file an issue with your section geometry. :::

Partial factors at fire limit state

EN 1992-1-2 §2.4.2 sets the material partial factors for the fire limit state to unity:

ACS therefore applies characteristic (nominal) strengths — $f'_c$ and $f_{yk}$ — as the base values before temperature reduction. The resulting fire interaction surface is a born-nominal surface: fire load combination demands are compared against it at the fire combination level (typically $\phi = 1.0$ per EN 1990).

Concrete strength reduction

$k_c(\theta)$ follows EN 1992-1-2 Table 3.1 for the selected aggregate type. The same curve is applied for both analysis methods. For $f'_c > 50$ MPa, EN 1992-1-2 §6 notes that the standard $k_c(\theta)$ curve is non-conservative for high-strength concrete — ACS surfaces a material warning for sections above this threshold regardless of the chosen method.

Steel reduction factors

For the 500°C isotherm method:

Both $k_y$ (yield strength) and $k_{Es}$ (elastic modulus) are applied and tracked; for these simplified methods the stress-strain model is elastic-perfectly-plastic so $k_y$ governs the ultimate-strength result.

For the advanced-fibre method, a third factor is additionally applied:

where $k_{sp}(\theta)$ is the proportional-limit reduction factor — the stress below which the stress-strain curve remains linear. All three factors follow EN 1992-1-2 Table 3.2a.

Advanced-fibre steel law (EN 1992-1-2 §3.2.3 / Figure 3.3)

The advanced-fibre method evaluates a five-region steel stress-strain law at each bar’s local temperature:

Figure 3.3 strain limits (temperature-independent): $\varepsilon_{sy,\theta} = 0.02$, $\varepsilon_{st,\theta} = 0.15$, $\varepsilon_{su,\theta} = 0.20$ (Class B/C ductility). The law applies with odd symmetry in compression (§3.2.3(4)). At 20 °C, $f_{sp} = f_{sy}$ collapses the elliptic region and the law degenerates exactly to the ambient elastic-perfectly-plastic model.

Prestressed concrete (PT)

All three fire analysis tiers carry post-tensioned and pre-tensioned tendons through the fire capacity calculation. The fundamental approach is to hold the ambient-basis locked-in prestrain fixed and reduce only the tendon’s constitutive law with temperature:

• The locked-in tensile strain ($\varepsilon_{pe} + \varepsilon_{decomp}$) is computed on the ambient (20 °C) basis and carried unchanged into the fire state.
• $k_p(\theta)$ and $k_{Ep}(\theta)$ (EN 1992-1-2 Table 3.3) scale the characteristic yield and ultimate strengths and the elastic modulus — they do not inflate the locked-in strain.

Which Table 3.3 column applies depends on the tendon’s prestressing product form, as Cl 3.2.4(2) words it — cold worked (wires and strands) against quenched and tempered (bars). ACS reads it from the tendon’s material, so a bar tendon is reduced on its own curve rather than a strand one:

Cl 3.2.4(2) makes the Class A / Class B choice for cold-worked steel a National Annex parameter. ACS applies Class B, the more conservative column over 400–600 °C — the range a tendon at a realistic axis distance occupies under a standard fire. Annex selection is not yet a user choice; when it becomes one, the class will follow it.

Where a tendon’s product form is not recorded — sections saved before ACS captured it — the cold-worked columns are applied.

The strain-hardening modulus $E' = Q E_p$ reduces with $k_p$ rather than $k_{Ep}$, so the ratio handed to the reduced curve is $Q\,k_p/k_{Ep}$. §3.2.4 names only the strength and the modulus, so this is a reading — but it is the one the model can carry: $E'$ is a stress-dimensioned quantity like every other entry in Table 3.3, and holding $Q$ fixed instead drives the derived knee factor below 1 by 500 °C, which is not a curve at all. It leaves the knee factor almost unchanged, and exactly unchanged at $k_p = k_{Ep} = 1$, which is what preserves the ambient degeneracy above.

Thermal relaxation of the prestressing force is not modelled as a separate time-dependent loss term — it is embedded in the $k_p(\theta)$/$k_{Ep}(\theta)$ reduction factors per EN 1992-1-2 Table 3.3.

When prestress is omitted: if the effective prestress $\sigma_{p.ef}$ is zero (or cannot be resolved), ACS runs the fire capacity check for the reinforcement only and shows a warning. This is not a refusal — the capacity result is valid for the non-prestressed state.

Axis distance (AS 3600 Cl 5.3.3): For tendons, AS 3600 Cl 5.3.3 requires an additional axis distance on top of the base value from Table 5.4.1(A) or 5.5.2(B):

Under the AS 3600 Section 5 tabulated method this is checked for you, as its own ratio against the tendons’ own average axis distance. The Cover panel reports the tabulated axis distance for reinforcement bars only, and the tendon increment never enters the advised cover — a tendon is placed by its profile, not off the cover chain.

Under the calculation methods (500 °C isotherm, advanced-fibre) there is no tabulated axis distance to increase: the heat-transfer solve models each tendon’s temperature directly, so Cl 5.3.3 does not apply.

Bar detail factors

The Bar Details table in the Fire results panel shows the temperature-driven reduction factors applied to each bar:

Concrete reduction display

The section capacity is recalculated using these reduced properties and compared against the fire load combination demands.

Fire interaction diagram

ACS generates a fire-rated interaction diagram overlay on the ambient diagram. The fire curve sits inside the ambient curve, showing the reduced capacity envelope. Your fire load combination point must fall inside the fire curve for adequacy.

Interpreting results

Cover panel (AS 3600)

The Cover panel in the Materials section shows the governing cover term and its source.

Exposure Class requirement: the Cover panel results are disabled until an Exposure Class is selected. Without an Exposure Class ACS cannot compute the durability cover, which is required to determine the governing cover term. Select an Exposure Class in the General tab to activate the panel.

Conservative structural assumptions: the cover resolver always uses the most conservative branch of each Section 5 table (simply-supported beams, μ_fi = 0.7 columns, one-way simply-supported slabs, N*/φNu = 0.7 walls). The tabular cover is therefore valid for any structural configuration within the member type without requiring you to specify end conditions or load ratios.

These pinned worst cases apply to the cover number only. The AS 3600 Section 5 tabulated design method selects its table from your real inputs instead — the support condition you set for beams and slabs, and the Cl 5.6.3(1) load level μ_fi computed from N*_f/φN_u for columns and walls — so a lightly-loaded or continuous member is not failed by an assumption it does not carry. If you know your actual load level is significantly lower than the conservative assumption, either the tabulated design method or the heat transfer + fire capacity analysis will give a less conservative result than the cover figure.

Section width for beams: the width $b$ used to look up Table 5.4.1(A) is the minimum horizontal solid width of the web — not the bounding-box width. For T-sections, I-sections, and L-sections the flange is not part of the beam web; ACS scans the actual section geometry to find the narrowest solid horizontal slice. This means a T-beam with a 600 mm flange and a 200 mm web is looked up at $b = 200$ mm, which gives a higher required axis distance than a 600 mm rectangular section at the same FRL. For rectangular sections the web width and the bounding box are identical.

Narrow-section refusal: when the section’s controlling dimension falls below the narrowest tabulated minimum for a particular FRL at the selected member type, ACS sets Resolved = false and reports a failure reason rather than returning the most-onerous axis distance. For walls that controlling dimension is the tabulated wall thickness $t_w$ of Table 5.7.2, which earlier releases did not check — a wall thinner than the table permits was previously issued the axis distance for its FRL despite satisfying no tabulated combination. This is consistent with the §874 principle — a silently-substituted value would understate the actual cover requirement for an undersized section. Widen the section or reduce the required FRL to obtain a resolved result.

When the fire term governs, increasing the section width (for beams), using a heavier bar schedule with larger diameter bars, or reducing the required FRL will all change the required cover.

Fire capacity results

Enter the fire limit-state action effects in the Fire Demands panel — N*f, Mx*f, and My*f: the applied axial force and biaxial moments at the fire combination level. These are the accidental combination demands per AS/NZS 1170.0 §4.2.4 (typically G + ψ_l Q, no concurrent wind or earthquake). For EN 1990, use the accidental fire combination; for ACI 216.1, use the gravity combination per §3.4.

PDF report output

The fire section of the PDF report includes the following disclosure blocks:

The Bar Details table in the report also includes Aggregate Type and Steel Class rows alongside the temperature and reduction factors for each bar.

Tips and best practices

• Set the Member Type in the General tab before configuring the fire duration — the tabular cover depends on both
• Larger sections have better fire resistance due to the thermal mass effect — the core remains cool even after extended exposure
• Increasing cover improves fire resistance by insulating the reinforcement from heat; the Section 5 tables give the minimum axis distance, not the minimum cover — the clear cover is the axis distance minus however far in ACS places the bar’s centre ($c_{fire} = a_s - d_{fitment} - d_b/2$ for a mid-edge bar; less the deeper tie-bend seat for a corner bar); when fitments are present their diameter adds to the required cover
• The code-based cover method automatically accounts for fire rating requirements; for a given FRL the fire term may or may not govern over the durability and placement covers
• For FRLs above 240 minutes, Section 5 tables do not apply — verify fire resistance by the heat transfer + fire capacity analysis instead
• Check both the slab-supported case (three-sided exposure) and the free-standing case (four-sided exposure) for beams

Related pages

• Fire design theory — heat transfer and material degradation background
• Section analysis — overview of all analysis types
• Design standards — fire design code references