4.1 Buckling and column curves
After this lesson you should be able to:
- State Euler buckling as the elastic reference load and explain why real steel columns need Perry-Robertson / IS buckling curves instead of pure Euler.
- Write the IS 800:2007 design compressive strength Pd = Ae fcd and the closed-form expressions for λ, φ, χ and fcd (cl. 7.1.2 and cl. 7.1.2.1), with γm0 = 1.10.
- Select the buckling class (a/b/c/d) from Table 10 for a rolled I-section, take α from Table 7, take K from Table 11, and compute fcd and Pd for a given length and axis.
Theory
A perfectly straight, pin-ended, elastic column under concentric compression remains straight until the Euler load Pcr = π² EI/(KL)². At that load a neighbouring bent shape is in equilibrium. The corresponding Euler stress is fcr = π² E/(KL/r)², with r the radius of gyration about the buckling axis and K the effective-length factor for the end restraints. For a steel strut this ideal load is only a reference: mill residual stresses, an initial bow, and accidental eccentricity make the member bend and yield progressively before Pcr is reached. Perry and Robertson expressed that behaviour with an initial-imperfection amplitude; modern codes fold the same physics into non-dimensional column curves.
IS 800:2007 Section 7 designs common hot-rolled and built-up members for axial compression by flexural buckling. Cl. 7.1.1 assigns each section a buckling class a, b, c or d that reflects residual-stress pattern and shape. Cl. 7.1.2 requires the factored compression P to satisfy P ≤ Pd, with design strength Pd = Ae fcd. The effective area Ae is the gross area for a semi-compact or better section without unfitted holes (cl. 7.3.2); Class 4 (slender) sections take a reduced effective area per cl. 3.7.2 and are outside the worked examples here.
Cl. 7.1.2.1 gives the design compressive stress as fcd = χ fy/γm0, not exceeding fy/γm0. The stress reduction factor is χ = 1/[φ + √(φ² − λ²)] ≤ 1.0, with φ = 0.5[1 + α(λ − 0.2) + λ²]. The non-dimensional effective slenderness is λ = √(fy/fcr) = (KL/r) √(fy/(π² E)) = (KL/r)/π × √(fy/E). The imperfection factor α is taken from Table 7: 0.21 (class a), 0.34 (b), 0.49 (c), 0.76 (d). When λ ≤ 0.2 the curve gives χ = 1.0 (stocky stub). Tables 8 and 9 tabulate χ and fcd for convenience; the closed form is used in this lesson and matches those tables when both are checked.
Cl. 7.1.2.2 and Table 10 classify cross-sections into buckling classes by type and thickness. For a rolled I-section with h/bf > 1.2 and flange thickness tf ≤ 40 mm, buckling about the major (z–z) axis is class a and about the minor (y–y) axis is class b. IS 800:2007 names z–z the major principal axis and y–y the minor axis (see cl. 9.3.2.2 and Table 10), which is the opposite of some other texts, so read the axis labels with care. Thicker flanges step down to classes b/c or d; hot-rolled hollow sections are class a and cold-formed hollow sections class b on any axis. The width-to-thickness limits that decide plastic, compact, semi-compact or slender plate elements sit in Table 2 and are used only by number in this lesson.
Effective length KL comes from cl. 7.2.1–7.2.2. The actual length L is centre-to-centre of intersections (or free-standing length for a cantilever). Table 11 then supplies K for idealised end restraints: both ends restrained against translation and rotation → 0.65L; one end so restrained and the other restrained in translation but free to rotate → 0.8L; both ends hinged (translation restrained, rotation free) → 1.0L; one end fixed and the other free → 2.0L; and a sway case with 1.2L. Frame cases that need deformed-geometry effective lengths are pointed to Annex D-1 and D-2 by the clause; they are not worked here. Truss compression members take KL between 0.7L and 1.0L in the plane of the truss (cl. 7.2.4).
EN 1993-1-1 uses the same family of buckling curves with imperfection factors, and AISC 360 uses a different closed-form column curve on a different non-dimensional slenderness. Both are named here for comparison only; they are not ranked, and their numerical factor tables are not quoted. Background commentary matching the IS approach is in N. Subramanian (OUP, 2008) and S. K. Duggal (3rd ed., 2019).
Derivation
- Euler reference: Pcr = π² EI/(KL)² (N); fcr = π² E/(KL/r)² (N/mm²); E = 2.0 × 10⁵ N/mm² for structural steel.
- Non-dimensional slenderness (cl. 7.1.2.1): λ = √(fy/fcr) = (KL/r)/π × √(fy/E). Stocky limit: λ ≤ 0.2 ⇒ χ = 1.0.
- Imperfection / curve (cl. 7.1.2.1): φ = 0.5[1 + α(λ − 0.2) + λ²]; χ = 1/[φ + √(φ² − λ²)] ≤ 1.0; α from Table 7.
- Design stress and strength: fcd = χ fy/γm0 ≤ fy/γm0 with γm0 = 1.10 (Table 5); Pd = Ae fcd. Check P ≤ Pd (cl. 7.1.2).
- Axis check: compute Pd about both principal axes with the matching buckling class (Table 10) and the matching r; the smaller Pd governs.
Euler: Pcr = π² EI/(KL)² (N); fcr = π² E/(KL/r)² (N/mm²); E = 2.0 × 10⁵ N/mm²λ = (KL/r)/π × √(fy/E) = √(fy/fcr) (dimensionless)φ = 0.5[1 + α(λ − 0.2) + λ²]; χ = 1/[φ + √(φ² − λ²)] ≤ 1.0α (Table 7): class a 0.21; b 0.34; c 0.49; d 0.76fcd = χ fy/γm0 ≤ fy/γm0 (N/mm²); γm0 = 1.10; Pd = Ae fcd (N or kN)Table 11 (examples): K = 0.65 (both ends fixed); 0.80; 1.0 (both hinged); 1.2; 2.0 (fixed–free)
- Static axial compression; fatigue and seismic ductile detailing are outside this lesson.
- E = 2.0 × 10⁵ N/mm² and γm0 = 1.10 as in IS 800 for ordinary structural steel.
- Examples use semi-compact or better sections so Ae = Ag (cl. 7.3.2); no unfitted holes are deducted.
- Steel grade E250 (fy = 250 N/mm²) unless stated; changing the grade changes fy (and Table 8/9 lookup) only.
- Section areas and radii of gyration in the examples are given design data (handbook values), not re-derived from plate dimensions.
- Local buckling limits are taken as satisfied by pointing to Table 2; specific b/t numbers are not quoted.
- Effective lengths use the idealised Table 11 factors; second-order frame K-factors (Annex D) are not computed here.
Codes and standards
Indian Standards are cited by number and clause only; no clause text, tables or figures are reproduced. Other standards are listed for comparison, not ranked.
- IS 800:2007 cl. 7.1.1 (buckling classes a/b/c/d), cl. 7.1.2 (P ≤ Pd; Pd = Ae fcd), cl. 7.1.2.1 (fcd, χ, φ, λ, fcr), Table 7 (imperfection factor α), Table 8 (χ), Table 9 (fcd), cl. 7.1.2.2 and Table 10 (classification of cross-sections), cl. 7.2.1–7.2.2 and Table 11 (effective length), cl. 7.2.4 (truss members), cl. 7.3.2 (effective sectional area), Annex D-1/D-2 (named only for frame and stepped columns), Table 5 (γm0 = 1.10), Table 2 (plate classification — by number only).
- Background: N. Subramanian, Design of Steel Structures, Oxford University Press, 2008; S. K. Duggal, Limit State Design of Steel Structures, McGraw Hill Education (India), 3rd edition, 2019.
- EN 1993-1-1 buckling curves and AISC 360 column curves are named for comparison only; they are not ranked and their factor tables are not quoted.
Worked examples
Problem. A rolled I-section column, E250 (fy = 250 N/mm²), has Ag = 5 400 mm², ry = 35.0 mm (minor axis y–y) and rz = 120 mm (major axis z–z) as given design data. Flange thickness tf = 12 mm and h/bf > 1.2, so Table 10 gives buckling class a about z–z (major, α = 0.21) and class b about y–y (minor, α = 0.34). The storey height centre-to-centre is L = 5.0 m and both ends are hinged (Table 11: K = 1.0). Find the design compressive strength Pd. Take E = 2.0 × 10⁵ N/mm² and γm0 = 1.10.
- Minor axis y–y (likely to govern; class b, α = 0.34): KL/ry = 1.0 × 5 000/35.0 = 142.86.
- λ = (142.86)/π × √(250/2.0×10⁵) = 1.6077.
- φ = 0.5[1 + 0.34(1.6077 − 0.2) + 1.6077²] = 2.0317.
- χ = 1/[2.0317 + √(2.0317² − 1.6077²)] = 0.3055.
- fcd = 0.3055 × 250/1.10 = 69.42 N/mm² (≤ 250/1.10 = 227.3 N/mm²).
- Pd,y = Ag fcd = 5 400 × 69.42 / 1 000 = 374.9 kN.
- Major axis z–z check (class a, α = 0.21): KL/rz = 5 000/120 = 41.67; λ = 0.4689; φ = 0.6382; χ = 0.9337; fcd = 212.2 N/mm²; Pd,z = 1 145.8 kN.
- Governing Pd = min(374.9, 1 145.8) = 374.9 kN (minor axis y–y). Euler reference on the minor axis: I = Ag ry² = 6.615 × 10⁶ mm⁴; Pcr = π² EI/L² = 522.3 kN, above Pd as expected with imperfections.
Answer. Pd = 374.9 kN (minor axis y–y governs)
Problem. The same E250 section and length as Example 1 is detailed with both ends restrained against translation and rotation (rigid end-plate base and beam continuity). Table 11 gives K = 0.65. Recompute Pd about the minor axis y–y (class b, α = 0.34).
- KL/ry = 0.65 × 5 000/35.0 = 92.86.
- λ = 92.86/π × √(250/2.0×10⁵) = 1.0450.
- φ = 0.5[1 + 0.34(1.0450 − 0.2) + 1.0450²] = 1.1897.
- χ = 1/[1.1897 + √(1.1897² − 1.0450²)] = 0.5687.
- fcd = 0.5687 × 250/1.10 = 129.3 N/mm².
- Pd = 5 400 × 129.26 / 1 000 = 698.0 kN (the major axis z–z, already 1 145.8 kN at K = 1.0, does not govern).
- Compared with the pinned case (374.9 kN), fixing both ends raises Pd by a factor 698.0/374.9 ≈ 1.86. On site, that gain is real only if the end connections and the adjoining members actually deliver the assumed restraints.
Answer. Pd = 698.0 kN (K = 0.65, minor axis y–y)
Problem. A stockier E250 strut has Ag = 7 730 mm² and ry = 40.0 mm (given). Buckling about the governing axis is class b (α = 0.34). Unsupported length L = 3.6 m; one end is restrained in translation and rotation, the other is restrained in translation but free to rotate (Table 11: K = 0.80). The factored load is P = 900 kN. Check the member.
- KL/ry = 0.80 × 3 600/40.0 = 72.00.
- λ = 72.00/π × √(250/2.0×10⁵) = 0.8103.
- φ = 0.5[1 + 0.34(0.8103 − 0.2) + 0.8103²] = 0.9320.
- χ = 1/[0.9320 + √(0.9320² − 0.8103²)] = 0.7181.
- fcd = 0.7181 × 250/1.10 = 163.2 N/mm².
- Pd = 7 730 × 163.2 / 1 000 = 1 261.5 kN ≥ 900 kN, so the strut is adequate for flexural buckling about this axis. Still check the other axis, local section classification (Table 2) and the connections.
Answer. Pd = 1 261.5 kN ≥ 900 kN — OK
Step-by-step procedure
- Identify the steel grade (fy), the governing axis (smaller r, matching buckling class from Table 10) and Ae (= Ag when cl. 7.3.2 applies).
- Take the actual length L and the end-restraint factor K from Table 11 (or Annex D when frame analysis requires it); form KL/r.
- Compute λ = (KL/r)/π × √(fy/E). If λ ≤ 0.2, set χ = 1.0; otherwise take α from Table 7 and compute φ and χ from cl. 7.1.2.1.
- Compute fcd = χ fy/γm0 ≤ fy/γm0 and Pd = Ae fcd. Repeat for the other principal axis; take the smaller Pd.
- Check P ≤ Pd. Separately confirm plate classification (Table 2), shear, and connection design.
Site notes
- A pinned base with a flexible cap beam is closer to K = 1.0 than to 0.65; claiming K = 0.65 without a moment connection overstates Pd.
- The minor axis of a rolled I usually governs even when it sits in a different buckling class: for a rolled I with h/bf > 1.2 and tf ≤ 40 mm the minor axis y–y is class b and the major axis z–z class a, and ry is much smaller than rz, so both effects point to y–y.
- Shop drawings should state the assumed K and the bracing that delivers it; missing intermediate restraints change KL overnight.
- For very stocky stubs (λ ≤ 0.2) the curve returns χ = 1 and Pd = Ae fy/γm0 — still check local crushing and the base plate.
- Tables 8 and 9 are handy on site for a quick fcd; the closed form is preferred in exams and for odd KL/r values between table steps.
- Using pure Euler Pcr as the design strength (ignores residual stress and bow; IS uses χ < 1 for λ > 0.2).
- Taking α = 0.21 for every section instead of reading Table 10 then Table 7.
- Computing λ with L instead of KL, or mixing ry into a major-axis class.
- Swapping the IS 800 axis labels: in IS 800:2007 z–z is the major axis and y–y the minor axis; using the major-axis class a on the minor axis overstates Pd.
- Forgetting the cap fcd ≤ fy/γm0, or using γm1 = 1.25 by mistake in place of γm0 = 1.10.
- Assuming Ae < Ag for a compact rolled section with no unfitted holes (cl. 7.3.2 uses Ag).
- Claiming K = 0.65 for a nominally pinned base, or K = 0.5 (theoretical fixed–fixed) instead of the Table 11 value 0.65.
For a quick minor-axis check on an E250 rolled I with K = 1: λ ≈ (L/ry)/88.9 because π √(E/fy) = π √(800) ≈ 88.9. Example 1: 5 000/35 = 142.9 and 142.9/88.9 ≈ 1.61, matching λ = 1.6077. Then read χ from the class-b curve for the minor axis y–y (or compute φ): χ ≈ 0.31, so Pd ≈ 0.31 Ag fy/γm0 ≈ 0.28 Ag fy for that λ.
Summary
- Real columns follow IS buckling curves, not pure Euler; Euler supplies fcr and λ only.
- Pd = Ae fcd with fcd = χ fy/γm0; χ from φ = 0.5[1+α(λ−0.2)+λ²] and χ = 1/[φ+√(φ²−λ²)] (cl. 7.1.2.1).
- α comes from Table 7 after the section is classed in Table 10; K comes from Table 11 (or Annex D).
- Check both axes; the smaller Pd governs. γm0 = 1.10. In IS 800:2007, z–z is the major axis and y–y the minor axis.
- EN 1993-1-1 and AISC 360 use related but different column curves — named here, not ranked.
Practice questions
Tap a question to see the answer.
1Compute λ for KL/r = 100, fy = 250 N/mm², E = 2.0 × 10⁵ N/mm².
λ = 100/π × √(250/2.0×10⁵) = 1.125.
2For class c (α = 0.49) and λ = 1.125, find φ, χ and fcd (fy = 250, γm0 = 1.10).
φ = 0.5[1 + 0.49(1.125 − 0.2) + 1.125²] = 1.360; χ = 0.471; fcd = 107.0 N/mm².
3Ae = 4 000 mm² and fcd = 120 N/mm². What is Pd?
Pd = 4 000 × 120 / 1 000 = 480 kN.
4A cantilever strut has L = 3.0 m and r = 40 mm. Using Table 11 K = 2.0, find KL/r.
KL/r = 2.0 × 3 000/40 = 150.
5What is the upper bound on fcd for E250 with γm0 = 1.10?
fy/γm0 = 250/1.10 = 227.3 N/mm².
6A rolled I with h/bf > 1.2 and tf ≤ 40 mm: which buckling classes apply about z–z and y–y (Table 10)?
z–z (major) class a (α = 0.21); y–y (minor) class b (α = 0.34).
References
Standard texts, listed in no order of merit. Our explanations, examples and diagrams are original; nothing is copied from these books.
- N. Subramanian, Design of Steel Structures, Oxford University Press, New Delhi, 1st edition, 2008.
- S. K. Duggal, Limit State Design of Steel Structures, McGraw Hill Education (India), 3rd edition, 2019.
Correction note. Corrected 10 November 2026: the axis labels in the theory text and Examples 1 and 2 were swapped. IS 800:2007 uses z–z for the major axis and y–y for the minor axis. The minor axis y–y of the rolled I in the examples is buckling class b (α = 0.34), not class a, so Pd in Example 1 is now 374.9 kN (was 405.5 kN) and in Example 2 698.0 kN (was 778.2 kN). The earlier values were on the unsafe side. The quick-check rule was updated to match. See our corrections policy.