Clause 6.3.1 — Flexural Buckling Resistance

The fundamental design verification for a compression member:

N_Ed / Nb,Rd  <=  1.0

Where Nb,Rd is the design buckling resistance:

Nb,Rd = chi x A x f_y / gamma_M1     (Class 1, 2, 3 cross-sections)
Nb,Rd = chi x A_eff x f_y / gamma_M1  (Class 4 cross-sections)

The reduction factor chi captures the combined effects of initial out-of-straightness (L/1000), residual stresses from rolling or welding, and material inelasticity. It is a function of the non-dimensional slenderness lambda_bar and the imperfection factor alpha, calibrated against thousands of full-scale European column buckling tests conducted through the European Convention for Constructional Steelwork (ECCS).

Non-Dimensional Slenderness lambda_bar (Clause 6.3.1.3)

The slenderness parameter is the ratio of the member slenderness to the squash load slenderness:

For Class 1, 2, 3:   lambda_bar = sqrt(A x f_y / N_cr)  =  (L_cr / i) / lambda_1
For Class 4:          lambda_bar = sqrt(A_eff x f_y / N_cr)  =  (L_cr / i) x sqrt(A_eff/A) / lambda_1

Where:

• N_cr = Euler elastic critical buckling load = pi^2 x E x I / L_cr^2
• L_cr = buckling length (effective length) about the relevant axis
• i = radius of gyration about the relevant axis = sqrt(I/A)
• lambda_1 = pi x sqrt(E / f_y) = 93.9 x epsilon — the reference slenderness
• epsilon = sqrt(235 / f_y) — the material factor

lambda_1 Values for Common Structural Steels

Using higher-strength steel: while f_y increases, lambda_1 decreases — so lambda_bar INCREASES for the same geometry. This is why the net gain in buckling resistance from moving from S235 to S355 is only 35-40%, not the full 51% f_y ratio would suggest. The increased penalty from the buckling curve partially offsets the yield strength gain.

Buckling Curves — Table 6.2 (Five Curves)

EN 1993-1-1 defines five distinct buckling curves (a0, a, b, c, d) corresponding to different cross-section types, manufacturing methods, and buckling axes:

Buckling Curve Selection Matrix for I/H Sections

For doubly-symmetric hot-rolled HEB columns with h/b = 1.0 (stocky H-section): major-axis buckling uses curve b, minor-axis uses curve b (both with alpha = 0.34). This is more favourable than the standard H-section assignment because h/b <= 1.2 triggers the curve b for y-y buckling instead of curve a.

Chi Reduction Factor — Perry-Robertson Formula (Clause 6.3.1.2)

The chi factor is the heart of the EN 1993 column design method:

chi = 1 / [phi + sqrt(phi^2 - lambda_bar^2)]     but chi <= 1.0

Where:  phi = 0.5 x [1 + alpha x (lambda_bar - 0.2) + lambda_bar^2]

At lambda_bar <= 0.2, chi = 1.0 (column achieves full squash load — local buckling or yield governs). The 0.2 plateau length reflects the test observation that very stocky columns are insensitive to initial imperfections because the squash load is reached before buckling deformations develop.

chi Values for All Five Curves (lambda_bar = 0.2 to 3.0)

At the critical lambda_bar = 1.0 region (typical for building columns of moderate slenderness): chi_a0 = 0.688 vs chi_d = 0.428, a 61% difference in column capacity between the best and worst buckling curves. Selecting hot-rolled sections that qualify for curves a0 or a is a significant economic driver for column design.

Effective Buckling Length L_cr

The buckling length L_cr is the distance between points of contraflexure (zero moment) in the buckled shape. It depends on end restraint conditions:

Braced Frames (Non-Sway, Clause 6.3.1.3)

The most common assumption in multi-storey braced frames is L_cr = 1.0 x L (storey height), reflecting nominally pinned beam-to-column simple connections that provide negligible rotational restraint. This is conservative but justified unless a rigorous Annex BB global buckling analysis demonstrates partial rotational restraint.

Unbraced Frames (Sway Mode, Annex BB.1)

For sway frames (moment frames, portal frames), the buckling length exceeds the storey height because the frame translates laterally. Annex BB.1 provides:

L_cr / L = max( [1 - 0.2(eta_1 + eta_2) - 0.12 x eta_1 x eta_2] / [1 - 0.8(eta_1 + eta_2) + 0.6 x eta_1 x eta_2],   1.0 )

Where eta_1, eta_2 = K_c / (K_c + K_b) at each column end. For portal frame columns with pinned bases: eta_base = 1.0 (K_b = 0 for pinned), eta_top approx 0.3-0.6 (rafter partial restraint), giving L_cr/L approx 2.0-2.5. This is why portal frame columns use heavy HEA/HEB sections.

Torsional and Flexural-Torsional Buckling (Clause 6.3.1.4)

While flexural buckling governs most doubly-symmetric I/H columns, torsional and flexural-torsional buckling must be checked for:

1. Open thin-walled sections (channels, tees, angles, cruciform sections) — where the torsional rigidity G x I_t is low
2. Point-symmetric and asymmetric sections — where buckling is inherently coupled
3. Columns with closely spaced lateral restraints — reducing L_cr,z but not L_cr,T, making torsional buckling the governing mode

Torsional Buckling Elastic Critical Load

N_cr,T = (1 / i_0^2) x (G x I_t  +  pi^2 x E x I_w / L_cr,T^2)

Where i_0^2 = i_y^2 + i_z^2 + y_0^2 + z_0^2 (polar radius of gyration), I_t = St Venant torsion constant, I_w = warping constant, and L_cr,T = buckling length for torsion (simply supported: L_cr,T = L).

For typical hot-rolled HEB/HEA columns, I_t is large due to the thick flanges and the warping rigidity I_w is significant. Torsional buckling rarely governs for doubly-symmetric hot-rolled H-sections but should be checked for all designs.

Flexural-Torsional Buckling

For mono-symmetric sections (channels, tees, unequal angles), flexural buckling about the symmetry axis couples with torsion. EN 1993-1-1 Clause 6.3.1.4 provides the elastic critical load as the smallest root of:

(N_cr - N_cr,y) x (N_cr - N_cr,z) x (N_cr - N_cr,T) x i_0^2  -  N_cr^2 x z_0^2 x (N_cr - N_cr,y)  -  N_cr^2 x y_0^2 x (N_cr - N_cr,z)  =  0

For doubly-symmetric sections (y_0 = z_0 = 0), this degenerates to the minimum of N_cr,y, N_cr,z, and N_cr,T — the three modes are uncoupled. The chi value for flexural-torsional buckling uses the buckling curve for the z-z axis (Table 6.2) with lambda_bar derived from the governing N_cr.

Combined Compression and Bending — Clause 6.3.3

For beam-columns (columns subject to both axial load and bending moment), EN 1993-1-1 provides two interaction methods:

Annex B (Method 2 — Simplified, Recommended for Manual Design)

For doubly-symmetric sections not susceptible to torsional deformations:

Flexural buckling about y-y axis:

N_Ed / (chi_y x N_Rk / gamma_M1)  +  k_yy x M_y,Ed / (M_y,Rk / gamma_M1)  +  k_yz x M_z,Ed / (M_z,Rk / gamma_M1)  <=  1.0

Flexural buckling about z-z axis:

N_Ed / (chi_z x N_Rk / gamma_M1)  +  k_zy x M_y,Ed / (M_y,Rk / gamma_M1)  +  k_zz x M_z,Ed / (M_z,Rk / gamma_M1)  <=  1.0

The interaction factors k_yy, k_yz, k_zy, k_zz are given in Annex B Tables B.1-B.3. For columns with only major-axis bending (M_z,Ed = 0, the common case):

k_yy = C_my x [1 + (lambda_bar_y - 0.2) x N_Ed / (chi_y x N_Rk / gamma_M1)]  <=  C_my x [1 + 0.8 x N_Ed / (chi_y x N_Rk / gamma_M1)]
k_zy = 0.6 x k_yy   (for lambda_bar_z > 0.4)

The equivalent uniform moment factor C_my accounts for the moment distribution shape:

• Uniform moment (psi = 1.0): C_my = 1.0
• Triangular moment (psi = 0): C_my = 0.6
• Double curvature (psi = -1.0): C_my = 0.4

Annex A (Method 1 — General, for Design Software)

Annex A provides more refined interaction factors accounting for cross-section slenderness, moment distribution shape parameter alpha_h, and the relative contributions of the two buckling modes. Method 1 is recommended for: asymmetric sections, biaxial bending with significant M_z,Ed, columns susceptible to lateral-torsional buckling, and cases where Annex B produces utilisation > 0.95 (to find additional capacity).

Worked Example — HEB200 Column, S355, L = 5.0 m

Design problem: Verify a HEB200 column in S355JR steel, pin-ended both axes, forming part of a braced multi-storey frame. The column is at ground floor level with a nominal pinned base and is in an internal heated environment (C1 corrosion, no temperature reduction).

Design data:

• Section: HEB200, S355JR, hot-rolled
• System length L = 5,000 mm; L_cr,y = L_cr,z = 5,000 mm (pinned both ends, conservative)
• N_Ed = 600 kN (axial compression, ULS), M_y,Ed = 25 kNm (minor eccentricity from beam reaction)
• gamma_M0 = 1.00, gamma_M1 = 1.00 (UK NA)

Step 1 — HEB200 Section Properties

Material: f_y = 355 MPa (t_f = 15 mm <= 40 mm), E = 210,000 MPa.

Step 2 — Cross-Section Classification (Clause 5.5)

epsilon = sqrt(235/355) = 0.814

Flange outstand: c = (b - t_w - 2r) / 2 = (200 - 9.0 - 36.0) / 2 = 77.5 mm c/t_f = 77.5 / 15.0 = 5.17 Class 1 limit: 9 x epsilon = 9 x 0.814 = 7.33. 5.17 <= 7.33 — Class 1 flange.

Web: c_w = h - 2t_f - 2r = 200 - 30 - 36 = 134 mm c_w/t_w = 134 / 9.0 = 14.89 Class 1 limit (pure compression): 33 x epsilon = 33 x 0.814 = 26.85. 14.89 <= 26.85 — Class 1 web.

HEB200 in S355: Class 1 cross-section. Full area effective.

Step 3 — Squash Load N_c,Rd (Clause 6.2.4)

N_c,Rd = A x f_y / gamma_M0 = 7,810 x 355 / 1.00 = 2,772,550 N = 2,773 kN

Cross-section utilisation: N_Ed / N_c,Rd = 600 / 2,773 = 0.216 — well below squash load.

Step 4 — Non-Dimensional Slenderness

lambda_1 = 93.9 x epsilon = 93.9 x 0.814 = 76.4

Strong axis (y-y):

lambda_bar_y = (L_cr / i_y) / lambda_1 = (5,000 / 85.4) / 76.4 = 58.55 / 76.4 = 0.767

Weak axis (z-z):

lambda_bar_z = (L_cr / i_z) / lambda_1 = (5,000 / 50.6) / 76.4 = 98.81 / 76.4 = 1.293

Weak-axis slenderness governs. The ratio lambda_bar_z / lambda_bar_y = 1.293 / 0.767 = 1.69, reflecting the significantly lower radius of gyration about the weak axis.

Step 5 — Buckling Curve Selection

For HEB200, h/b = 200/200 = 1.0 <= 1.2:

• Y-Y axis: hot-rolled I-section, h/b <= 1.2, tf = 15 mm <= 40 mm -> Curve b (alpha = 0.34)
• Z-Z axis: hot-rolled I-section, any h/b, tf = 15 mm <= 40 mm -> Curve b (alpha = 0.34)

For the HEB family with h/b = 1.0, both axes use curve b — this is more favourable than sections with h/b > 1.2 where z-z uses curve c (alpha = 0.49).

Step 6 — Chi Reduction Factors

Z-Z axis (governs):

phi_z = 0.5 x [1 + 0.34 x (1.293 - 0.2) + 1.293^2]
      = 0.5 x [1 + 0.34 x 1.093 + 1.672]
      = 0.5 x [1 + 0.372 + 1.672]
      = 0.5 x 3.044
      = 1.522

chi_z = 1 / (1.522 + sqrt(1.522^2 - 1.293^2))
      = 1 / (1.522 + sqrt(2.316 - 1.672))
      = 1 / (1.522 + sqrt(0.644))
      = 1 / (1.522 + 0.802)
      = 1 / 2.324
      = 0.430

Y-Y axis (for comparison):

phi_y = 0.5 x [1 + 0.34 x (0.767 - 0.2) + 0.767^2]
      = 0.5 x [1 + 0.193 + 0.588]
      = 0.5 x 1.781
      = 0.891

chi_y = 1 / (0.891 + sqrt(0.891^2 - 0.767^2))
      = 1 / (0.891 + sqrt(0.794 - 0.588))
      = 1 / (0.891 + sqrt(0.206))
      = 1 / (0.891 + 0.454)
      = 1 / 1.345
      = 0.743

Step 7 — Buckling Resistance

Nb,Rd,z = chi_z x A x f_y / gamma_M1 = 0.430 x 7,810 x 355 / 1.00 = 0.430 x 2,773 = 1,192 kN
Nb,Rd,y = chi_y x A x f_y / gamma_M1 = 0.743 x 7,810 x 355 / 1.00 = 0.743 x 2,773 = 2,061 kN

Axial utilisation: N_Ed / Nb,Rd,z = 600 / 1,192 = 0.503 — weak-axis buckling governs, OK.

Step 8 — Combined Check (Annex B)

Bending resistance: M_pl,y,Rd = W_pl,y x f_y / gamma_M0 = 642 x 10^3 x 355 / 1.00 = 228.0 kNm

C_my = 0.6 (triangular moment distribution, psi = 0):

k_yy = 0.6 x [1 + (0.767 - 0.2) x (600 / 2,061)] = 0.6 x [1 + 0.567 x 0.291] = 0.6 x 1.165 = 0.699
k_zy = 0.6 x k_yy = 0.6 x 0.699 = 0.419

Y-Y plane interaction:

600 / 2,061 + 0.699 x 25 / 228.0 = 0.291 + 0.077 = 0.368 <= 1.0 — OK

Z-Z plane interaction:

600 / 1,192 + 0.419 x 25 / 228.0 = 0.503 + 0.046 = 0.549 <= 1.0 — OK

Step 9 — Alternative Section Comparison

HEB160 fails (utilisation 1.157 > 1.0 — buckling resistance inadequate). HEB180 is adequate (0.812) but leaves less reserve for incidental moments or future load increases. HEB200 is the practical choice (0.503 utilisation, reasonable reserve). HEB220 and above are overly conservative unless the column also serves as a transfer element or carries heavy moments.

UK National Annex vs EU Recommended Values

EN 1993-1-1 allows National Annexes to adjust certain parameters. The key differences for column design:

*DIN NA may use gamma_M1 = 1.10 for some buckling cases — this is the most significant deviation. For the worked HEB200 example with gamma_M1 = 1.10 instead of 1.00: Nb,Rd,z = 1,192 / 1.10 x 1.00 = 1,084 kN, utilisation rises to 600/1,084 = 0.553 (still OK but with less margin). UK NA allows designers to use BS 5950-1 effective length guidance for braced frames as an alternative to Annex BB, though this is rarely used for new designs.

Critical guidance: Always check the applicable National Annex for the country where the structure will be built. While gamma_M0 and gamma_M2 are universally 1.00 and 1.25 respectively across all major EU economies, gamma_M1 may differ. Most National Annexes (UK, France, Netherlands, Italy, Spain, Poland, Czech Republic) adopt gamma_M1 = 1.00. Germany is the main exception with gamma_M1 = 1.10. For cross-border projects, the worst-case National Annex should be adopted or the specific jurisdiction confirmed by the client.

Practical Column Design Workflow

A systematic approach to column sizing per EN 1993-1-1:

1. Establish effective lengths: L_cr,y and L_cr,z based on frame type (braced/unbraced), end restraints, and Annex BB if applicable. Document assumptions — the most common error is overestimating restraint.
2. Select trial section: Based on experience, approximate load, or quick lambda_bar = 1.0 trial. For hot-rolled columns, start with HEB or UC sections.
3. Classify cross-section: Determine Class (1-4) per Clause 5.5. Use effective area A_eff for Class 4.
4. Compute lambda_bar: About both axes. The larger value governs.
5. Select buckling curve: From Table 6.2. Check both axes — they may differ.
6. Compute chi: Using the Perry-Robertson formula. The lowest chi governs.
7. Verify Nb,Rd: N_Ed / Nb,Rd <= 1.0.
8. Check combined loading: If moments present, use Annex B (Method 2) for doubly-symmetric sections.
9. Check torsional buckling: Clause 6.3.1.4 for open sections, asymmetric shapes, or closely spaced lateral restraints.
10. Verify serviceability: Axial shortening, storey drift, and vibration if relevant.

Frequently Asked Questions

Why does EN 1993 use five buckling curves when AISC uses a single curve?

The five-curve system (a0/a/b/c/d) emerged from the ECCS column test programme in the 1960s-70s, which tested thousands of full-scale columns across European steel mills. The data showed that different cross-section types (rolled I-sections vs welded box sections vs cold-formed hollow sections) have systematically different imperfection sensitivities. Residual stress patterns from hot-rolling are more favourable than from welding; hot-finished hollow sections have lower residual stresses than cold-formed ones. The five curves capture these systematic differences, giving hot-rolled sections an economic advantage and penalising welded and cold-formed sections that have higher residual stresses. AISC 360 uses a single curve (Eq. E3-2) with a simplified 0.877 elastic buckling transition because US practice predominantly uses hot-rolled W-shapes with a narrow range of residual stress patterns.

How do I select the correct buckling curve for a welded I-section?

For welded I-sections (plate girders), the buckling curve depends on axis, flange thickness, and steel grade per Table 6.2: y-y axis with tf <= 40 mm uses curve b; y-y axis with tf > 40 mm uses curve c; z-z axis with tf <= 40 mm uses curve c; z-z axis with tf > 40 mm uses curve d. The higher imperfection factors for welded sections reflect the longitudinal fillet welds that introduce higher residual stresses (up to yield magnitude near the web-to-flange junction) compared to hot-rolled sections where residual stresses are typically 0.3-0.5 x f_y. For S460 welded sections, check if the steel is thermomechanically rolled (qualifies as hot-rolled) or quenched and tempered (treated as welded for classification purposes).

What is the lambda_bar = 0.2 plateau in the Perry-Robertson formula?

At lambda_bar <= 0.2, the column is sufficiently stocky that initial imperfections do not reduce the capacity below the squash load — the column yields in compression before buckling deformations develop. The 0.2 limit corresponds to a physical slenderness L_cr/i of approximately 0.2 x 93.9 = 18.8 for S235 steel, or 0.2 x 76.4 = 15.3 for S355 steel. Columns shorter than this threshold are classified as "short columns" and designed for the full cross-section compression resistance N_c,Rd. In practice, most building columns have lambda_bar between 0.4 and 1.2, well into the buckling-sensitive range.

When does torsional buckling govern instead of flexural buckling?

Torsional buckling governs when N_cr,T < N_cr,z (for I-sections, torsional buckling about the longitudinal axis is the critical mode when the column is laterally restrained at closely spaced intervals). This occurs when: (1) the minor-axis unbraced length L_cr,z is short (closely spaced lateral restraints), making N_cr,z high; (2) the cross-section has low torsional rigidity (thin open sections like channels, tees, and angles); (3) the warping rigidity E x I_w is small relative to the torsional rigidity G x I_t. For standard hot-rolled HEB/HEA/UC sections at typical storey heights, N_cr,T is approximately 2-3 times N_cr,z, so flexural buckling about the weak axis governs. However, for crane columns with intermediate lateral restraints, torsional buckling should be checked explicitly.

What is the difference between the Annex A and Annex B interaction methods?

Annex A (Method 1) was developed by the French-German research group and provides a comprehensive interaction formulation based on second-order inelastic analysis calibration. It requires more input parameters (section shape factors alpha_y, alpha_z; equivalent moment factors for each axis and mode; detailed C_m factors from moment diagrams). Annex B (Method 2) was developed by the Austro-Swiss group as a simplified method suitable for manual calculation. It uses fewer parameters and tabulated interaction factors k_ij in Tables B.1-B.3. For doubly-symmetric I/H sections with < 5% minor-axis bending and no lateral-torsional buckling susceptibility, both methods give similar results (within 5%). For more complex cases, Annex A can yield 10-15% higher capacity by accounting for moment distribution shape more accurately. Design software typically implements Annex A, while hand calculations use Annex B.

Related Pages

• EN 1993 Column Buckling — Curves a0-d Complete Reference →
• EN 1993 Column Design Example — HEB240 Axial Compression →
• EN 1993 Beam Design — Flexure, Shear & LTB →
• EN 1993 Steel Grades — S235, S275, S355, S460 →
• EN 1993 Connection Design — Bolts & Welds Overview →
• EN 1990 Load Combinations — ULS & SLS →
• European Beam Sizes — IPE, HEA, HEB, UB, UC →
• AS 4100 Column Buckling (Australia) →
• CSA S16 Column Buckling (Canada) →
• Column Capacity Calculator — Free EN 1993 Tool →

Educational reference only. All column buckling resistances are per EN 1993-1-1:2005 with the recommended gamma_M1 = 1.00 unless stated otherwise. Verify all values against the applicable National Annex for your jurisdiction. Effective length assumptions must be justified by frame analysis and restraint conditions. Section properties should be taken from the manufacturer's current published data. All column designs must be independently verified by a licensed Professional Engineer (PE) or Chartered Structural Engineer (CEng). Results are PRELIMINARY — NOT FOR CONSTRUCTION without professional structural engineering review.

Design Resources

Calculator tools

• Column Capacity Calculator — Free Online Tool
• Steel Buckling Summary Calculator
• Steel Column Base Design Calculator
• Steel Beam Calculator — Multiple Codes

Design guides

• Column Capacity Worked Example — Step by Step
• Column Buckling Workflow Guide
• Column Buckling Calculator Guide
• EN 1993-1-1 Column Buckling Worked Example