Scope and limitations

Capabilities, assumptions, and known limitations of the Advanced Concrete Section tool.

Scope

The Advanced Concrete Section tool is designed for cross-section level analysis and design of reinforced, prestressed, and post-tensioned concrete members. It computes section capacities and checks them against applied actions provided by the user.

ACS supports:

• Arbitrary polygonal cross-sections with voids
• Reinforced concrete with multiple bar sizes and grades
• Prestressed concrete with bonded and unbonded tendons
• Combined axial force, biaxial bending, shear, and torsion actions
• Ultimate and serviceability limit state checks
• Fire resistance assessment with 2D heat transfer
• Nonlinear moment-curvature analysis
• Time-dependent effects (creep and shrinkage)
• Design to AS 3600:2018 Amendment 2:2021 (ACI 318-19 and EN 1992-1-1 are Coming Soon and cannot yet be selected)

Assumptions

The following assumptions apply to all analyses:

Known limitations

Section-level analysis only

ACS analyses the cross-section in isolation. It does not account for:

• Member-level effects: slenderness, moment magnification ($\delta$ factors), P-$\Delta$ effects. You must compute magnified moments externally and input them as the design actions.
• System-level effects: load redistribution, continuity moments, lateral stability. ACS assumes you have determined the design actions from a separate structural analysis.
• Tendon curvature sense: whether a tendon sags or hogs at this section, which is what decides where its strand bears inside the duct. ACS sees one cross-section, not the tendon’s profile, and the same section is checked against sagging and hogging combinations alike — so there is no single answer it could derive. Declare it per tendon with the Strand position in duct control — Top, Centreline or Bottom, with the offset derived from your duct and strand (see Prestressing). Left at Not declared, ACS models the strand on the duct centreline and says so on the PT tab.
• Detailing: anchorage, lap splices, development length, bar curtailment. ACS checks section capacity but not reinforcement detailing.

Shear and torsion

• Shear capacity is computed at a single critical section using the AS 3600 simplified MCFT (Cl. 8.2.4.3) or general method (Cl. 8.2.4.2), as selected automatically by Cl. 8.2.4.1. Strut-and-tie models for disturbed regions (D-regions) are not supported.
• Torsion capacity is accepted as an input but the torsion design check is not yet implemented. The shear check does not account for torsion-shear interaction.
• A mixed-grade fitment cage is designed on its weakest grade, except for $\phi V_{us}$. Each fitment is designed on the reinforcement material it is placed under. Where the fitments do not all share one grade, $\phi V_{us}$ credits each fitment’s legs at its own grade ($\sum A_{sv,i} f_{sy,i}$). Every clause that takes a single fitment $f_{sy}$ uses the lowest grade present: the Cl. 8.2.1.7 $A_{sv,\min}$, the $k_v$ branch and its $\phi$, $\phi T_{us}$ and the Table 10.7.4.3 minimum fitment diameter. A cage with a cell-anchored tie keeps $\phi V_{us}$ on the lowest grade too. No design code addresses a cage of several grades. The summation assumes every leg a shear crack crosses reaches yield (ductility class N). The report states the basis each clause used. See Transverse reinforcement limits.

Fire design

• No spalling modelling. Explosive spalling of high-strength concrete cover is not captured. For $f'_c > 55$ MPa, the fire analysis may be unconservative if spalling occurs.
• Siliceous and calcareous aggregate types are supported (selectable when using EN 1992-1-2). Lightweight aggregate thermal properties are not currently available.
• Fire exposure is assumed uniform along the member length (2D section analysis).

Prestressing

• Friction losses assume a simplified linear model. Complex tendon profiles with reverse curvature are not supported.
• Unbonded tendon stress increase at ultimate uses the simplified code formula, not a full member-level analysis.
• Post-tensioning anchorage zone design (bursting and spalling reinforcement) is not included.
• The tendon strain check applies to bonded tendons only. An unbonded tendon carries the member-average Cl. 8.1.8 $\sigma_{pu}$, which no local strain compatibility produced, so there is no local strain to bound. See Tendon strain at ultimate.
• Constraining $\varepsilon_{cu}$ under a rectangular stress block retains $\alpha_2$ and $\gamma$, which the codes calibrate at their nominal $\varepsilon_{cu}$ — the block depth is tied to $d_n$, not to the extreme-fibre strain, so the concrete contribution does not fall with $\varepsilon_{cu}$ the way a real stress-strain law’s does. This matches RAPT’s behaviour, and it is why the reduced $\varepsilon_{cu}$ is reported rather than applied quietly: re-run on a fibre model (Hognestad, parabola-rectangle) where the reduction is carried by the law itself.
• In fire, the tendon strain limit is the ambient $A_{gt}$. The temperature-reduced strand curve scales its strain-hardening ratio and leaves $A_{gt}$ at its ambient value, so the limit applied at elevated temperature is less strict than a temperature-consistent one would be. Still bounded where it previously was not; tracked separately.

Geometry

• Self-intersecting polygons are not supported. The outline must be a simple (non-crossing) polygon.
• Circular sections are approximated as polygons (typically 36 or more sides). This introduces negligible error for practical sizes.

Reinforcement above 500 MPa

AS 3600 Cl. 1.1.2(d) admits reinforcing steel grades above the Table 3.2.1 listed 500 MPa, up to 800 MPa, where the grade meets the Table 3.2.1 Class N requirements. ACS supports them, with a 600 MPa ceiling in ultimate-limit-state design models (see Reinforcement yield strength in ULS design models). Three boundaries are refusals, not reduced-accuracy results:

• Unqualified grades. Above 500 MPa the grade must declare ductility class N or E, uniform elongation $\varepsilon_{su} \geq 0.05$ and a tensile-to-yield ratio $R_m/R_e \geq 1.08$. The evidence belongs to the grade — a yield strength typed above the value the grade itself declares carries no evidence for the higher number. Without it, ACS refuses the analysis instead of designing on assumed ductility.
• Above 800 MPa. Outside Cl. 1.1.2(d) entirely. Refused, never quietly designed at the 600 MPa ceiling.
• Section 14 seismic and fire. Both are withheld above 500 MPa — see Seismic design and Fire design.

The limits apply to the AS 3600 path. ACI 318 and EN 1992-1-1 carry their own grade rules and are unaffected.

Material models

• Concrete tension stiffening is not available for all analysis types.
• The Mander confined concrete model requires the user to select it explicitly; ACS does not automatically detect confinement from stirrup configuration.
• Time-dependent effects (AEMM) assume a single loading age. Multiple loading events at different ages are not supported.
• ULS capacity on a curvilinear stress-strain model is evaluated at that model’s fixed ultimate compressive strain. AS 3600 Cl. 8.1.2 Note 1(b) permits the extreme-compression-fibre strain to be adjusted to obtain the maximum bending strength; ACS does not apply that optimisation, so AS 3600 flexural capacities are conservative with respect to it. The conservatism does not extend to a result used as a lower bound on overstrength (capacity design), nor to $k_{uo}$. The report’s Assumptions section states this whenever it applies.

Solver convergence

Several quantities are produced by iterative solvers that do not converge for every section and action combination — the ULS fibre stress distribution and the cracked-section serviceability results among them. On non-convergence the solver returns a fallback state: numbers at their defaults, or an achieved action that is not the one requested.

The fibre stress distribution has two distinct non-convergence causes, and the refusal detail names which one it hit. The requested $M^*$ can exceed what the section carries; or the solver can reach the requested moment and still find no strain state that carries $N^*$ at the converged curvature. The second reads very differently from the first — the moment is met exactly — so do not read a stress-distribution refusal as a flexural capacity verdict without checking the message.

The curvature search spans both signs, so a prestressed section is not a special case: where the applied $M^*$ sits below the section’s decompression moment, the equilibrium curvature is negative and the solver finds it there. A refusal on a post-tensioned section means the same thing it means on a reinforced one.

ACS handles that fallback state in two different ways, and which one you get depends on the quantity. Check the table before relying on a returned value.

For the returned rows, a returned value is not necessarily a converged one — an API caller must read the converged flag, and the reason is carried in the analysis warnings. Do not treat the presence of a crack width or a service stress as evidence the solve succeeded.

One check has no fallback answer at all. The linear-creep validity flag (linearCreepStressExceeded, reported beside $\varphi_{cc}$ and $E_{c,\text{eff}}$ on every long-term SLS combination) compares the service compressive stress against $0.45 f'_c$ — the threshold above which $E_{c,\text{eff}} = E_c / (1 + \varphi_{cc})$ stops being valid. It is derived from the SLS stress check, so when that check does not run or does not converge there is no stress to compare and the flag is null, not false. Read null as the check was not performed, never as the assumption held — the long-term result is unconfirmed, not confirmed valid. The PDF report says so in words at the same place it would otherwise print the exceedance warning.

For the withheld rows, the by-id API refuses with analysis_did_not_converge and carries the solver’s own explanation in the refusal detail — see API error codes.

Wherever it arises, non-convergence is a statement about the section rather than about the request: reduce the applied action or increase the section’s capacity.

Other results carry a converged flag too — flexure and moment-curvature among them. Wherever the field exists, read it. On flexure it is not decoration: a combination whose $N^*$ sits at or beyond the section’s axial capacity returns converged: false with every moment field at its default, and that $\phi M_u = 0$ means no valid stress state exists here, not this section has no capacity.

Valid input ranges

Accuracy and validation

ACS has been validated against hand calculations and published benchmark problems:

Differences of less than 1% are typical and arise from iteration convergence tolerances and the finite number of integration points.

Features not yet implemented

Related pages

• Section analysis — full analysis documentation
• Design standards — code comparison