Second-Order Analysis Guide — P-Delta, B1/B2, Direct Analysis Method, and Notional Loads
Complete second-order analysis reference for steel structures: P-Delta and P-Delta effects, B1 and B2 amplification factors, the Direct Analysis Method (DAM) per AISC 360 Appendix 7, notional load requirements, and a worked example for a multi-storey moment frame. Based on AISC 360-22, Eurocode 3, and AS 4100.
PRELIMINARY — NOT FOR CONSTRUCTION. All results are for educational and reference use only. Must be independently verified by a licensed Professional Engineer (PE) or Structural Engineer (SE) before use in any project.
Why Second-Order Analysis Matters
First-order analysis assumes that equilibrium is satisfied on the undeformed structure — that member forces can be calculated assuming the structure does not displace under load. For most structures, this assumption is adequate. But steel structures, particularly unbraced moment frames, are flexible enough that the deformations under load are not negligible. When gravity loads act through these deformations, they produce additional moments and forces not captured by first-order analysis.
The difference between first-order and second-order results can be dramatic. A 10-storey unbraced steel moment frame with a drift of h/400 (within code limits) will have P-Delta moments that are 15–25% larger than first-order moments. A frame with excessive flexibility (drift approaching h/200) can see P-Delta amplification of 50% or more. Neglecting these effects is unsafe.
Second-order analysis accounts for two distinct phenomena:
P-Delta (P-δ): The additional moment within a member caused by axial load acting through the member's own curvature. This is a member-level effect — the classic beam-column moment amplification from Euler buckling theory.
P-Delta (P-Δ): The additional moments caused by gravity loads acting through the lateral displacement (drift) of the frame. This is a system-level effect — if the frame sways laterally, the gravity load on the columns produces an overturning moment that must be resisted by the lateral force-resisting system.
Together, these effects are called "second-order effects" (because they are proportional to the product of load and displacement). A rigorous second-order analysis (often called a "geometrically nonlinear" or "P-Delta" analysis) iteratively solves for equilibrium on the deformed structure, accounting for both P-δ and P-Δ.
P-Delta vs P-Delta: Member-Level and System-Level Effects
The distinction between P-δ and P-Δ is important because codes provide different methods for handling each.
P-δ (P-small-delta): Occurs within an individual member, independent of the frame's lateral displacement. Consider a column with pinned ends carrying an axial load P. Under a transverse load, the column bends. The axial load P, acting through the curvature, produces an additional moment M_add = P × δ(x), where δ(x) is the deflection at point x. This additional moment further increases the curvature, leading to a magnification of the first-order moments that can be captured by the classic amplification factor:
B1 = C_m / (1 - P_u / P_e1) ≥ 1.0
Where:
- C_m is the equivalent moment factor (0.6 - 0.4 × (M1/M2) for braced frames, where M1/M2 is the ratio of smaller to larger end moments, positive for reverse curvature)
- P_u is the required axial strength
- P_e1 = π² × EI* / (K1 × L)² is the Euler buckling load of the member in the plane of bending, using the effective length factor K1 = 1.0 for the Direct Analysis Method
P-Δ (P-big-Delta): Occurs at the system level when the entire frame displaces laterally. The gravity load on the columns, P, acts through the storey drift Δ, producing an additional overturning moment P × Δ. This moment must be resisted by the lateral force-resisting system, effectively reducing the frame's lateral stiffness. The amplification factor for P-Δ is:
B2 = 1 / (1 - ΣP_u × Δ_H / (ΣH × L)) ≥ 1.0
Where:
- ΣP_u is the total vertical load in the storey
- Δ_H is the first-order interstorey drift
- ΣH is the storey shear
- L is the storey height
The P-Δ effect is inherently destabilising — as the frame drifts farther, the P × Δ moment increases, which increases the drift further. If the frame is too flexible, this feedback loop diverges and the frame collapses. The stability index θ = ΣP_u × Δ_H / (ΣH × L) must be less than 0.5 (a factor of safety of 2 against sidesway buckling). If θ > 0.5, the frame is too flexible and must be stiffened.
The B1/B2 Amplification Method
AISC 360 Appendix 8 presents the B1/B2 method as a practical alternative to a full second-order analysis. The method computes first-order member forces and then amplifies them using the B1 and B2 factors:
Design Moment (including second-order effects): M_r = B1 × M_nt + B2 × M_lt
Where:
- M_nt: First-order moment from loads that do not cause sidesway (no-translation moment)
- M_lt: First-order moment from loads that cause sidesway (lateral-translation moment)
- B1: P-δ amplification factor for the no-translation moment
- B2: P-Δ amplification factor for the lateral-translation moment
For a braced frame, the lateral-translation moment M_lt = 0 and only B1 applies. For an unbraced frame, both B1 and B2 apply.
Worked B1 Calculation:
Consider a W310 × 97 column in a braced frame, 4 m long, carrying P_u = 800 kN, with end moments M1 = -30 kNm and M2 = 60 kNm (reverse curvature).
- EI* = 0.8 × τ_b × E × I (per DAM, τ_b = 1.0 for P_u / P_y ≤ 0.5)
- For W310 × 97, I_x = 275 × 10⁶ mm⁴, τ_b = 1.0
- EI* = 0.8 × 1.0 × 200000 × 275 × 10⁶ = 44.0 × 10¹² N-mm²
- P_e1 = π² × 44.0 × 10¹² / (1.0 × 4000)² = 27150 kN
- C_m = 0.6 - 0.4 × (-30/60) = 0.6 - 0.4 × (-0.5) = 0.6 + 0.2 = 0.8
- B1 = 0.8 / (1 - 800 / 27150) = 0.8 / 0.9705 = 0.824
Since B1 ≥ 1.0 per code, B1 = 1.0. The moment amplification is negligible because P_e1 >> P_u. For a column near its buckling load (P_u / P_e1 > 0.5), B1 can exceed 2.0.
Worked B2 Calculation:
Consider a 6-storey unbraced moment frame, 4 m storey height. For a typical storey:
- ΣP_u = 12,000 kN (total gravity load on the storey from all columns)
- ΣH = 600 kN (storey shear from wind/seismic)
- Δ_H = 20 mm (first-order interstorey drift)
- L = 4000 mm
B2 = 1 / (1 - (12000 × 20) / (600 × 4000)) = 1 / (1 - 240,000 / 2,400,000) = 1 / (1 - 0.10) = 1 / 0.90 = 1.111
The P-Δ effect increases the storey moments by 11.1%. For a more flexible frame (Δ_H = 40 mm): B2 = 1 / (1 - 0.20) = 1.25. A 25% increase in moments is significant and must be accounted for.
Stability Index Check: θ = ΣP_u × Δ_H / (ΣH × L) = 12,000 × 20 / (600 × 4000) = 0.10. Since θ < 0.5, the frame is stable.
The Direct Analysis Method (DAM) — AISC 360 Appendix 7
The Direct Analysis Method (DAM), introduced in AISC 360-10 Appendix 7 and mandatory in AISC 360-22 Chapter C, is the required analysis method for all steel structures designed to AISC 360. DAM replaces the effective length method (K-factors) by directly modelling the stability effects in the analysis.
DAM has three core requirements:
1. Direct Consideration of Second-Order Effects
The analysis must account for P-Δ and P-δ effects directly in the analysis model, either through a rigorous second-order analysis or through amplified first-order analysis (B1/B2 method). A rigorous second-order analysis is preferred because it captures interaction effects between P-Δ and P-δ that the simplified B1/B2 approach cannot.
2. Reduced Stiffness
To account for the softening effects of residual stresses and member out-of-straightness (initial geometric imperfections), DAM requires a reduction in member stiffness:
- Flexural stiffness: EI* = 0.8 × τ_b × E × I
- Axial stiffness: EA* = 0.8 × E × A
Where τ_b is the stiffness reduction factor:
- τ_b = 1.0 when P_u / P_y ≤ 0.5 (member is mostly elastic)
- τ_b = 4 × (P_u / P_y) × (1 - P_u / P_y) when P_u / P_y > 0.5 (member has significant inelasticity)
This 0.8 factor accounts for the fact that residual stresses in hot-rolled sections can cause partial yielding at stresses well below F_y, reducing the effective stiffness. The τ_b factor captures the additional stiffness loss when the member is heavily loaded.
3. Notional Loads
To simulate the destabilising effect of initial out-of-plumbness, member out-of-straightness, and incidental eccentricities that exist in every real structure, DAM requires notional loads — small lateral loads applied at each floor level in addition to the actual lateral loads.
Per AISC 360 C2.2b, notional loads are:
- N_i = 0.002 × Y_i (2% of the gravity load applied at level i)
- Applied in the direction that adds to the destabilising effect
- Can be applied in the same analysis as the other loads (no separate notional load combination required for the DAM)
Where Y_i is the total gravity load applied at level i from the LRFD load combination or 1.6 times the ASD load combination.
In Eurocode 3, the equivalent initial sway imperfection is φ = φ_0 × α_h × α_m, where φ_0 = 1/200, α_h = 2/√h (but ≤ 1.0), and α_m = √(0.5 × (1 + 1/m)). For a typical multi-storey frame, the equivalent notional load is 0.3–0.5% of the gravity load — somewhat smaller than the AISC 2% but applied as an initial imperfection shape rather than as applied loads.
Why DAM Replaces the Effective Length Method
The traditional effective length method (K-factors from alignment charts) attempted to capture frame stability by assigning effective length factors to each column based on the relative stiffness of the beams framing into it. This method had two fundamental problems:
K-factors are system-level, not member-level properties. A column's K-factor depends on the stiffness of every other column and beam in the frame. If one column approaches its buckling load, it loses stiffness and sheds load to the other columns — an interaction the K-factor method cannot capture.
K-factors cannot account for inelasticity. The alignment chart K-factors assume all members are elastic. In reality, heavily loaded columns have reduced stiffness due to partial yielding, which increases the effective length of adjacent columns.
DAM solves both problems by modelling the stability effects directly in the analysis, using reduced stiffness to capture inelasticity, and using notional loads to trigger the buckling modes. The effective length factor for all members is taken as K = 1.0 in DAM.
Notional Loads in Detail
Notional loads are the most frequently misunderstood requirement of DAM. They are not real loads — they are a mathematical device to trigger the lateral buckling modes in the analysis model.
Purpose: A perfectly straight, perfectly plumb frame analysed with P-Δ effects will not buckle until the theoretical buckling load is reached, even if the analysis is geometrically nonlinear. This is because the P-Δ effect is zero when Δ = 0, regardless of P. But real frames have imperfections — columns are not perfectly straight, frames are not perfectly plumb, and connections are not perfectly rigid. Notional loads simulate these imperfections by providing a small initial lateral perturbation.
Magnitude: The 0.002 × Y_i notional load corresponds to an initial out-of-plumbness of approximately h/500 (the AISC Code of Standard Practice erection tolerance) amplified by the gravity-to-lateral load ratio. Recent research and the AISC 360-22 Commentary acknowledge that 0.002 may be conservative for many frames, but the simplicity of a single value outweighs the complexity of a more refined approach.
Application: Notional loads are applied at each floor level, in the direction that increases the destabilising effect. For wind load combinations, the wind and notional loads are applied in the same direction. For gravity-only combinations, notional loads are applied in both orthogonal directions (X and Y) to trigger buckling modes in both axes.
Notional loads in braced frames: Even braced frames require notional loads because the bracing system has its own imperfections (brace out-of-straightness, connection slip, gusset plate flexibility). However, for braced frames, the notional loads are resisted almost entirely by the bracing, with negligible effect on the columns, and can often be checked by hand rather than requiring a full second-order analysis.
Eurocode 3 and AS 4100 Approaches
Eurocode 3 (EN 1993-1-1 Section 5.3.2): Eurocode 3 uses the "equivalent imperfection" method. Initial sway imperfections (frame out-of-plumbness) are modelled as an equivalent horizontal force at each storey: H_eq = φ × V_Ed, where φ is the sway imperfection and V_Ed is the total vertical load. Member imperfections (out-of-straightness) are modelled as an equivalent bow imperfection e_0, typically L/350 for flexural buckling curves a, b, and c. For frames where α_cr (the elastic critical buckling factor) ≥ 10, second-order effects may be neglected (first-order analysis is sufficient). For 3 ≤ α_cr < 10, amplified first-order analysis is permitted. For α_cr < 3, a full second-order analysis is required.
AS 4100 (Section 4.7): AS 4100 requires second-order analysis for all frames unless the moment amplification factor δ_s ≤ 1.1 (less than 10% amplification). The analysis can use the moment amplification method (equivalent to B1/B2), an elastic buckling analysis (using the effective length method, which AS 4100 retains alongside second-order methods), or a rigorous second-order analysis. The AS 4100 notional load recommendation is 0.002 × (design gravity load), consistent with AISC.
Worked Example: Second-Order Analysis of a 3-Storey Moment Frame
Problem Statement: Analyse a 3-storey unbraced steel moment frame, 4 m storey height, 6 m bay width. The frame resists wind and gravity loads. Perform a second-order analysis using the B1/B2 amplification method per AISC 360 DAM.
Step 1 — Frame geometry and loads:
- Columns: W310 × 97 (A572 Gr 50, I_x = 275 × 10⁶ mm⁴)
- Beams: W460 × 52 (I_x = 147 × 10⁶ mm⁴)
- Storey height = 4.0 m, bay width = 6.0 m
Gravity loads (LRFD combination 1.2D + 1.6L):
- Roof: w_u,roof = 24 kN/m, P_u,roof per column = 24 × 6 / 2 = 72 kN per internal column
- Floor: w_u,floor = 36 kN/m, P_u,floor per column = 36 × 6 / 2 = 108 kN
Column axial loads (accumulated):
- Roof level column: P_u = 72 kN
- Level 2 column: P_u = 72 + 108 = 180 kN
- Level 1 column: P_u = 72 + 108 + 108 = 288 kN
Wind loads (LRFD): Storey shears from left to right: V_roof = 30 kN, V_2 = 65 kN, V_1 = 100 kN.
Step 2 — First-order analysis for lateral translation moments:
The first-order interstorey drift for a moment frame under the applied storey shears can be estimated using the portal method or a structural analysis program. Using a stiffness analysis:
- Roof drift (Δ_H): 2.8 mm
- Level 2 drift: 5.1 mm
- Level 1 drift: 6.3 mm
Step 3 — Compute B2 for each storey:
Storey 3 (Roof):
- ΣP_u = 2 columns × 72 = 144 kN
- ΣH = 30 kN
- Δ_H = 2.8 mm
- L = 4000 mm
- B2_roof = 1 / (1 - (144 × 2.8) / (30 × 4000)) = 1 / (1 - 403.2 / 120,000) = 1 / (1 - 0.00336) = 1.003
Storey 2:
- ΣP_u = 2 × 180 = 360 kN (total gravity load on columns framing into storey 2 from above)
- ΣH = 65 kN (total storey shear at level 2)
- Δ_H = 5.1 mm
- B2_2 = 1 / (1 - (360 × 5.1) / (65 × 4000)) = 1 / (1 - 1836 / 260,000) = 1 / (1 - 0.00706) = 1.007
Storey 1:
- ΣP_u = 2 × 288 = 576 kN
- ΣH = 100 kN
- Δ_H = 6.3 mm
- B2_1 = 1 / (1 - (576 × 6.3) / (100 × 4000)) = 1 / (1 - 3629 / 400,000) = 1 / (1 - 0.00907) = 1.009
Step 4 — Check stability index:
θ = (576 × 6.3) / (100 × 4000) = 0.0091. Since θ < 0.10, second-order effects are modest.
Step 5 — Compute B1 for the governing column (Level 1 interior):
P_u = 288 kN, EI* = 0.8 × τ_b × 200000 × 275 × 10⁶. For P_u / P_y: P_y = 345 × 12,100 × 10⁻³ = 4175 kN. P_u / P_y = 0.069 < 0.5, so τ_b = 1.0. EI* = 0.8 × 1.0 × 200000 × 275 × 10⁶ = 44.0 × 10¹² N-mm². P_e1 = π² × 44.0 × 10¹² / (1.0 × 4000)² = 27150 kN. C_m = 0.85 (assumed for sway-permitted frame). B1 = 0.85 / (1 - 288 / 27150) = 0.85 / 0.989 = 0.859 → B1 = 1.0 (minimum).
Step 6 — Apply notional loads per DAM:
At each floor, N_i = 0.002 × Y_i.
- Roof: Y_roof = 24 × 6 = 144 kN per bay; notional load = 0.002 × 144 = 0.288 kN per bay. Total N_roof = 0.58 kN for the frame.
- Level 2: Y_2 = 36 × 6 = 216 kN; N_2 = 0.002 × 216 = 0.43 kN per bay. Total = 0.86 kN.
- Level 1: Same as Level 2.
The notional loads are very small (less than 2% of the wind storey shears) and can be added to the wind loads without appreciably changing the results. For gravity-only combinations (no wind), the notional loads must be applied in both X and Y directions. For this frame, the notional loads would produce moments approximately 0.5–1% of the wind moments — negligible in practice because the wind load dominates.
Step 7 — Final design moments:
For the Level 1 interior column under wind + gravity:
- First-order moment (from analysis): M_lt = 85 kNm (lateral), M_nt = 15 kNm (gravity, negligible for unbraced frame)
- Amplified moment: M_r = 1.0 × 15 + 1.009 × 85 = 15 + 85.8 = 100.8 kNm
- The P-Δ amplification adds 0.8 kNm (0.9%) — negligible for this stiff frame.
Analysis Summary: For this 3-storey frame, second-order effects are less than 1%. The frame is stiff, the stability index θ = 0.009, and the B2 factors are essentially 1.0. For a more flexible frame (say a 10-storey frame with the same column sections), B2 would be in the range 1.15–1.30 and second-order effects would be significant. The DAM notional loads are negligible because wind loads dominate.
Engineering Best Practices
- Always check the stability index θ = ΣP_u × Δ_H / (ΣH × L). If θ < 0.10, second-order effects are modest (< 11%). If 0.10 ≤ θ < 0.20, second-order analysis is mandatory. If θ ≥ 0.5, the frame is unstable and must be redesigned.
- For frames with θ < 0.05, the B2 factor is less than 1.05 and second-order effects can often be neglected for preliminary sizing — but the AISC DAM is still required for the final design even if θ is small.
- The 0.8 stiffness reduction in DAM applies to all members in the lateral force-resisting system. Do not apply additional stiffness reductions beyond those specified in Appendix 7.
- For frames with significant axial load in beams (transfer girders, tie beams, truss chords), include the beam axial load in the P-Δ calculation because the beam axial load also contributes to the P-Δ moment when the frame sways.
- For composite steel-concrete frames, the effective stiffness of composite members must account for concrete cracking. AISC 360 I2.1b provides the effective stiffness for composite columns.
- Many structural analysis programs (SAP2000, ETABS, RISA-3D) have built-in P-Delta analysis capabilities. When using these, verify that the program is using the reduced stiffness (0.8τ_b) and that notional loads are included in the analysis model.
FAQ
Q: When can I ignore second-order effects? A: Per AISC 360, never — the DAM is mandatory for all steel structures. However, for frames where B2 < 1.10 (θ < 0.09), a first-order analysis is permitted for the preliminary design, with second-order effects checked but negligible. Per Eurocode 3, second-order effects may be neglected when α_cr ≥ 10, which corresponds to B2 < 1.11. Per AS 4100, second-order effects may be neglected when δ_s ≤ 1.1.
Q: What's the difference between the Direct Analysis Method and the Effective Length Method? A: The Effective Length Method (ELM) uses K-factors from alignment charts to account for frame buckling, with member design using the K-factored slenderness. DAM models second-order effects directly (P-Δ, P-δ), applies reduced stiffness (0.8τ_b), and uses notional loads to trigger buckling modes. In DAM, K = 1.0 for all members. DAM is mandatory in AISC 360-22; the ELM is retained only for braced frames and frames with small θ.
Q: How are notional loads applied in a 3D analysis? A: Notional loads are applied at each floor level in both orthogonal directions (X and Y), added to the actual lateral loads. The notional load for each direction is 0.002 × Y_i, where Y_i is the gravity load on that floor. For wind load combinations, notional loads are applied in the same direction as the wind. For gravity-only combinations, notional loads are applied in the X and Y directions separately (two analysis cases). The notional loads are typically negligible compared to wind or seismic loads but are always applied.
Q: What happens if the B2 factor exceeds 2.0? A: A B2 factor exceeding 2.0 means that second-order effects more than double the first-order moments. This corresponds to a stability index θ > 0.50, which exceeds the AISC 360 limit. The frame is unacceptably flexible and will likely fail by sidesway buckling before reaching its design load. The frame must be redesigned — increase column and/or beam stiffness, add bracing, or switch to a braced frame system. A B2 factor between 1.5 and 2.0 indicates marginal stability (0.33 < θ < 0.50) and requires careful review.
References
- AISC 360-22 — Specification for Structural Steel Buildings, Chapter C and Appendix 7
- AISC 360-22 Commentary — Appendix 7 (extensive guidance on DAM implementation)
- Eurocode 3 (EN 1993-1-1) Section 5 — Structural Analysis
- AS 4100:2020 Section 4 — Methods of Structural Analysis
- AISC Design Guide 28 — Stability Design of Steel Buildings
- Chen, W.F. and Lui, E.M. — Structural Stability: Theory and Implementation (1987)
- Galambos, T.V. and Surovek, A.E. — Structural Stability of Steel: Concepts and Applications (2008)
Try Our Free Calculator
Try our free steel design calculator to perform second-order analysis, compute B1/B2 amplification factors, and verify the Direct Analysis Method requirements for your steel frames per AISC 360, EN 1993, and AS 4100. No signup required. Get full step-by-step calculation reports with code clause references.
Try it now: Check your second-order effects with our free calculator →