EN 1993 Column Buckling — Flexural Buckling per Clause 6.3
Complete reference for flexural buckling design of steel columns per EN 1993-1-1 Clause 6.3 (Eurocode 3: Design of Steel Structures). Buckling curves a0, a, b, c, and d with imperfection factors α and χ reduction factor derivation, effective length Lcr for braced and unbraced frames, non-dimensional slenderness λbar calculation for Class 1-3 and Class 4 sections, column buckling resistance Nb,Rd, combined axial compression and bending interaction per Clause 6.3.3 (Annex A or Annex B methods), and a fully worked example using HEA240 in S355 buckling about both axes.
Quick access: EN 1993 Steel Design → | EN 1993 Steel Grades → | European Beam Sizes → | Section Properties → | Beam Calculator →
EN 1993-1-1 Clause 6.3 Buckling Framework
EN 1993-1-1 Clause 6.3 covers the buckling resistance of uniform members in compression. The fundamental design check:
NEd / Nb,Rd ≤ 1.0
Where Nb,Rd is the design buckling resistance of the compression member:
Nb,Rd = χ × A × fy / γM1 (Class 1, 2, 3 cross-sections)
Nb,Rd = χ × Aeff × fy / γM1 (Class 4 cross-sections)
The reduction factor χ accounts for buckling and is a function of the non-dimensional slenderness λbar and the imperfection factor α:
χ = 1 / [Φ + √(Φ² − λ̄²)] but χ ≤ 1.0
where Φ = 0.5 × [1 + α × (λ̄ − 0.2) + λ̄²]
For the UK National Annex: γM1 = 1.00 (same as recommended value).
The χ-λbar relationship captures the physical behaviour of imperfect columns: at very low slenderness (λbar < 0.2), the column achieves full squash load (χ = 1.0). As slenderness increases, geometric imperfections and residual stresses reduce the capacity below the Euler buckling load. The α parameter calibrates the severity of imperfection sensitivity for different section types and manufacturing methods.
Non-Dimensional Slenderness λbar
The non-dimensional slenderness is the ratio of the member slenderness to the squash load slenderness. It is the critical intermediate parameter for buckling design:
For Class 1, 2, 3: λ̄ = √(A × fy / Ncr) = Lcr / i × [1 / λ1]
For Class 4: λ̄ = √(Aeff × fy / Ncr) = Lcr / i × √(Aeff/A) × [1 / λ1]
Where:
- Lcr = buckling length in the buckling plane considered (effective length)
- i = radius of gyration about the relevant axis (i = √(I/A))
- λ1 = π × √(E / fy) = 93.9 × ε — the slenderness at which the Euler buckling stress equals the yield stress
- ε = √(235 / fy) — the material factor
λ1 Values for Common Steel Grades
| Steel Grade | fy (MPa) | ε = √(235/fy) | λ1 = 93.9 × ε |
|---|---|---|---|
| S235 | 235 | 1.000 | 93.9 |
| S275 | 275 | 0.924 | 86.8 |
| S355 | 355 | 0.814 | 76.4 |
| S420 | 420 | 0.748 | 70.2 |
| S460 | 460 | 0.715 | 67.1 |
Higher-strength steel has a lower λ1, meaning for the same Lcr/i, the non-dimensional slenderness λbar is higher for higher-strength steel. This is why the net buckling resistance gain from moving to a higher steel grade is less than the fy ratio suggests. Example: changing from S235 to S355 (fy increases 51%) but λ1 drops 19%, so λbar increases 23% for the same geometry, increasing the χ penalty. The net Nb,Rd gain is typically 35-40%, not 51%.
Buckling Curves and Imperfection Factors α
EN 1993-1-1 Table 6.2 defines five buckling curves (a0, a, b, c, d) with corresponding imperfection factors:
| Buckling Curve | α | Typical Sections | Table Ref |
|---|---|---|---|
| a0 | 0.13 | Hot-rolled S460 hollow sections, hot-finished S355-S460 RHS | 6.2 |
| a | 0.21 | Hot-rolled S235-S460 I/H sections (tf ≤ 40 mm, y-y axis), hot-finished S235-S355 RHS | 6.2 |
| b | 0.34 | Hot-rolled S235-S460 I/H sections (z-z axis), hot-finished S420-S460 RHS | 6.2 |
| c | 0.49 | Cold-formed RHS (S275-S355), hot-rolled I/H (tf > 40 mm about z-z), channels, angles, tees, welded box sections (tf ≤ 40 mm) | 6.2 |
| d | 0.76 | Cold-formed RHS (S420-S460), hot-rolled sections with tf > 100 mm about z-z, welded box sections (tf > 40 mm) | 6.2 |
The buckling curve depends on four factors:
- Section type: I/H sections follow different curves than hollow sections
- Manufacturing method: Hot-rolled/finished sections use more favourable curves than cold-formed or welded sections
- Axis of buckling: The major axis (y-y) uses a more favourable curve than the minor axis (z-z) for I-sections
- Steel grade and flange thickness: Thicker flanges (> 40 mm) shift to a less favourable curve due to higher residual stresses
Buckling Curve Selection Quick-Reference
| Section | Axis | Steel Grade | tf ≤ 40 mm | tf > 40 mm |
|---|---|---|---|---|
| Hot-rolled I/H (UB, UC, HEA, HEB, IPE) | y-y | S235-S460 | a | a |
| Hot-rolled I/H | z-z | S235-S460 | b | c |
| Hot-finished RHS (SHS) | both | S235-S355 | a | — |
| Hot-finished RHS (SHS) | both | S420-S460 | b | — |
| Hot-finished CHS | any | S235-S460 | a | — |
| Cold-formed RHS (SHS) | both | S235-S355 | c | — |
| Cold-formed RHS (SHS) | both | S420-S460 | d | — |
| Welded box (plate) | both | S235-S460 | c | d |
| Hot-rolled channels, angles, tees | z-z | S235-S460 | c | d |
Buckling Reduction Factor χ — Worked Tables
The χ factor varies with λbar and buckling curve. Here is a comprehensive tabulation:
| λbar | χ (a0) | χ (a) | χ (b) | χ (c) | χ (d) |
|---|---|---|---|---|---|
| 0.2 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| 0.4 | 0.975 | 0.960 | 0.938 | 0.906 | 0.848 |
| 0.6 | 0.903 | 0.870 | 0.823 | 0.769 | 0.692 |
| 0.8 | 0.799 | 0.749 | 0.688 | 0.626 | 0.546 |
| 1.0 | 0.688 | 0.631 | 0.564 | 0.502 | 0.428 |
| 1.2 | 0.583 | 0.525 | 0.461 | 0.403 | 0.337 |
| 1.5 | 0.444 | 0.393 | 0.339 | 0.292 | 0.240 |
| 2.0 | 0.285 | 0.249 | 0.212 | 0.180 | 0.146 |
| 2.5 | 0.194 | 0.169 | 0.144 | 0.121 | 0.098 |
| 3.0 | 0.139 | 0.121 | 0.103 | 0.087 | 0.070 |
At λbar = 1.0, the difference between the best curve (a0, χ = 0.688) and the worst (d, χ = 0.428) is a 60% higher resistance for columns classified under a0. This underscores the economic importance of selecting sections that qualify for favourable buckling curves.
Effective Buckling Length Lcr
The buckling length Lcr is the distance between points of inflection (zero moment) in the buckled shape. It depends on end restraint conditions and frame sway sensitivity.
Braced Frames (Non-Sway Buckling)
Columns laterally restrained at floor levels. Buckling between floor diaphragms:
| End Condition | Lcr / L | Typical Application |
|---|---|---|
| Both ends rigidly connected (full rotational restraint) | 0.7 | Continuous column with stiff, deep beams providing partial fixity |
| One end fixed, one end pinned | 0.85 | Ground floor column with rigid base, pinned top |
| Both ends nominally pinned | 1.0 | Default for standard braced frame columns — conservative |
| Cantilever (one end free) | 2.0 | Flagpole column — rare in building frames |
For most braced multi-storey columns with typical beam-to-column simple connections, use Lcr = 1.0 × L (storey height). This is the conservative default unless Annex BB analysis demonstrates partial rotational restraint reduces the effective length.
Unbraced Frames (Sway Mode)
For unbraced frames (moment frames, portal frames), the buckling length exceeds the storey height because the frame sways laterally. EN 1993-1-1 Annex BB.1 provides:
Lcr / L = max( [1 − 0.2 × (η1 + η2) − 0.12 × η1 × η2] / [1 − 0.8 × (η1 + η2) + 0.6 × η1 × η2], 1.0 )
Where η1, η2 = Kc / (Kc + Kb) at each column end, with Kc = column stiffness (Ic / Lc) and Kb = beam stiffness (Ib / Lb, modified by far-end condition factor).
For portal frame columns with pinned bases (η = 1.0 at base) and rafter providing partial restraint (η ≈ 0.5-0.7 at top): Lcr / L ≈ 2.0-2.5. This is the primary reason portal frame columns use heavy HEA/HEB sections.
Combined Axial Compression and Bending — EN 1993-1-1 Clause 6.3.3
For columns subject to combined axial load and bending moment, EN 1993-1-1 provides two methods:
Annex B (Method 2 — Simplified)
For doubly-symmetric sections not susceptible to torsional deformations:
y-y buckling plane:
NEd / (χy × NRk / γM1) + kyy × My,Ed / (My,Rk / γM1) + kyz × Mz,Ed / (Mz,Rk / γM1) ≤ 1.0
z-z buckling plane:
NEd / (χz × NRk / γM1) + kzy × My,Ed / (My,Rk / γM1) + kzz × Mz,Ed / (Mz,Rk / γM1) ≤ 1.0
The interaction factors kyy, kyz, kzy, kzz are given in Annex B Tables B.1-B.3. For common building columns (major-axis bending only, Mz,Ed = 0):
kyy = Cmy × [1 + (λ̄y − 0.2) × NEd / (χy × NRk / γM1)] ≤ Cmy × [1 + 0.8 × NEd / (χy × NRk / γM1)]
kzy = 0.6 × kyy (for λ̄z > 0.4, Annex B Table B.3, simplified)
Where Cmy is the equivalent uniform moment factor:
- Uniform moment (ψ = 1.0): Cmy = 1.0
- Triangular moment (ψ = 0): Cmy = 0.6
- Equal end moments causing double curvature (ψ = -1.0): Cmy = 0.4
Annex A (Method 1 — General)
Design software typically implements Annex A, which provides more refined interaction factors accounting for cross-section shape, moment distribution shape, and the relative contribution of each buckling mode. Method 1 is recommended for asymmetric sections, combined biaxial bending with high minor-axis moment ratios, and columns susceptible to lateral-torsional buckling.
Worked Example — HEA240 Column, S355
Column details:
- Section: HEA240, S355J2, hot-rolled
- Length L = 4,500 mm, braced frame — Lcr,y = Lcr,z = 4,500 mm (pinned ends, conservative)
- NEd = 850 kN, My,Ed = 45 kNm (uniform major-axis moment from eccentric beam reaction)
- Class 1 cross-section (c/t checks satisfied for ε = 0.814)
Step 1 — Section Properties (HEA240)
| Property | Value |
|---|---|
| h × b | 230 × 240 mm |
| tw, tf | 7.5, 12.0 mm |
| A | 76.8 cm² = 7,680 mm² |
| Iy, iy | 7,763 cm⁴, 10.05 cm = 100.5 mm |
| Iz, iz | 2,770 cm⁴, 6.00 cm = 60.0 mm |
| Wpl,y | 745 cm³ = 745 × 10³ mm³ |
Step 2 — Non-Dimensional Slenderness
λ1 = 93.9 × √(235/355) = 93.9 × 0.814 = 76.4
y-y axis (strong): λ̄y = (4,500 / 100.5) / 76.4 = 44.78 / 76.4 = 0.586 z-z axis (weak): λ̄z = (4,500 / 60.0) / 76.4 = 75.00 / 76.4 = 0.982
Step 3 — Buckling Reduction Factors
y-y (Curve a, α = 0.21): Φy = 0.5 × [1 + 0.21 × (0.586 − 0.2) + 0.586²] = 0.5 × [1 + 0.081 + 0.343] = 0.712 χy = 1 / [0.712 + √(0.712² − 0.586²)] = 1 / [0.712 + 0.405] = 0.895
z-z (Curve b, α = 0.34): Φz = 0.5 × [1 + 0.34 × (0.982 − 0.2) + 0.982²] = 0.5 × [1 + 0.266 + 0.964] = 1.115 χz = 1 / [1.115 + √(1.115² − 0.982²)] = 1 / [1.115 + 0.528] = 0.609
Weak-axis buckling (z-z) governs. The column is 60.9% efficient about the weak axis vs 89.5% about the strong axis.
Step 4 — Resistance Checks
Squash load: Npl,Rd = 7,680 × 355 / 1.0 = 2,726 kN Buckling resistance (governs): Nb,Rd,z = 0.609 × 2,726 = 1,660 kN Axial ratio: NEd / Nb,Rd,z = 850 / 1,660 = 0.512
Bending resistance: Mpl,y,Rd = 745 × 10³ × 355 / 1.00 = 264.5 kNm Moment ratio: My,Ed / Mpl,y,Rd = 45 / 264.5 = 0.170
Step 5 — Combined Interaction (Annex B)
Cmy = 0.6 (uniform moment, ψ = 1.0 conservative) kyy = 0.6 × [1 + (0.586 − 0.2) × 0.512 / 0.895] = 0.6 × [1 + 0.386 × 0.572] = 0.6 × 1.221 = 0.733 kzy = 0.6 × kyy = 0.440
y-y plane: 0.512 + 0.733 × 0.170 = 0.512 + 0.125 = 0.637 ≤ 1.0 — OK z-z plane: 0.512 + 0.440 × 0.170 = 0.512 + 0.075 = 0.587 ≤ 1.0 — OK
The HEA240 in S355 is adequate (59-64% utilisation). A lighter HEA220 could potentially work subject to re-check, or a heavier section for longer spans.
Practical Buckling Design Guidance
Restraint Assumptions
The single most common error in column buckling design is assuming a shorter effective length than is justified:
- Floor diaphragms provide lateral restraint — columns in braced frames buckle between floors. But this only holds if the diaphragm is adequately braced (concrete slab, or steel deck with adequate horizontal bracing).
- Pinned bases are conservative for Lcr — a nominally pinned base plate provides negligible rotational restraint. Use Lcr = 1.0 × L for the lowest storey.
- Intermediate purlins alone are not restraint points — a single purlin connected to a column flange restrains lateral-torsional buckling of the beam but does not prevent minor-axis column buckling unless the purlin line is itself tied back to a braced bay or diaphragm.
- Torsional restraint is critical for thin-walled open sections — for slender I-sections and channels, check EN 1993-1-1 Cl. 6.3.1.4 for flexural-torsional buckling.
Frequently Asked Questions
Why does a higher-strength steel not proportionally increase column buckling resistance?
Because λ1 = 93.9 × √(235/fy) decreases with increasing fy. Changing from S235 to S355: fy increases 51% but λ1 drops from 93.9 to 76.4 (19% drop). For the same geometry, λbar increases 23%, which increases the χ penalty. The net buckling resistance gain is typically 35-40%, not the full 51%. This effect is most pronounced for slender columns (λbar > 1.0). For stocky columns (λbar < 0.4), the fy increase translates almost directly to higher resistance.
When should I use buckling curve c instead of curve b for an I-section column?
Use curve c for hot-rolled I/H sections buckling about the z-z axis when the flange thickness exceeds 40 mm (Table 6.2). Also use curve c for: welded I-sections with tf ≤ 40 mm about z-z; cold-formed RHS in S235-S355; channels, angles, and tees; and any I-section with tf > 100 mm about z-z. At λbar = 1.0, curve c gives χ = 0.502 vs curve b χ = 0.564 — a 12% lower resistance.
How do I determine Lcr for a column in an unbraced steel portal frame?
For unbraced portal frames, EN 1993-1-1 Annex BB.1 provides the sway buckling length. For a portal frame column with a nominally pinned base (η = 1.0) and a rafter providing partial rotational restraint (η ≈ 0.5-0.7 at top), Lcr/L ≈ 2.0-2.5. Many designers conservatively use Lcr = 2.5 × L for single-span portal columns with pinned bases. With a rigid (moment-resisting) base, η drops to 0.0-0.2 and Lcr/L reduces to approximately 1.5-1.8.
What is the difference between torsional buckling and flexural-torsional buckling?
Flexural buckling (Cl. 6.3.1) is buckling by bending about a principal axis — the standard column buckling check. Torsional buckling is buckling by pure twisting about the longitudinal axis, relevant for cruciform sections and thin-walled open sections. Flexural-torsional buckling combines bending and twisting — it governs single-symmetric sections (channels, tees, unequal angles) and slender I-sections. EN 1993-1-1 Cl. 6.3.1.4 provides the check: NEd ≤ Ncr,TF / γM1. For doubly-symmetric hot-rolled I/H sections in typical building frames, flexural-torsional buckling is rarely critical but must be checked.
Can I use the Annex B interaction method for all column designs?
Annex B (Method 2) applies to doubly-symmetric sections with no susceptibility to lateral-torsional buckling in the column segment. For single-symmetric sections (channels, tees) or columns subject to lateral-torsional buckling, Method 1 (Annex A) should be used. Annex B is conservative for most building column applications and is the default for manual design. Design software typically implements Annex A (Method 1) for generalised cases.
Related Pages
- EN 1993 Steel Design Overview →
- EN 1993 Steel Grades →
- European Beam Sizes — IPE, HEA, HEB →
- EN 1990 Load Combinations →
- EN 1998 Seismic Design →
- European Steel Fire Protection →
- AS 4100 Column Buckling (Australia) →
- CSA S16 Column Buckling (Canada) →
- Section Properties →
Educational reference only. Verify all buckling parameters against current EN 1993-1-1 and the relevant National Annex. Effective length assumptions must be justified by the structural framing and restraint conditions. All column designs must be independently verified by a qualified structural engineer. Results are PRELIMINARY — NOT FOR CONSTRUCTION without professional structural engineering review.