Steel Frame Analysis — Direct Analysis Method, P-Delta Effects & Stability

Frame analysis determines the internal forces and displacements in a steel structure by accounting for member stiffness, connection behavior, and second-order (P-delta) effects. AISC 360-22 Chapter C establishes the Direct Analysis Method (DM) as the primary approach for stability design of steel frames. The DM replaced the older Effective Length Method (ELM) as the preferred approach because it directly models the effects of residual stresses, initial imperfections, and geometric nonlinearity.

Direct Analysis Method (AISC 360-22 Section C2)

The DM requires three modifications to the structural analysis model:

1. Reduced stiffness (Section C2.3):

EI* = 0.80 × tau_b × E × I    (flexural stiffness)
EA* = 0.80 × E × A              (axial stiffness)

Where tau_b = 1.0 when alpha × Pr/Pns ≤ 0.5, and tau_b = 4 × (alpha × Pr/Pns) × (1 - alpha × Pr/Pns) when alpha × Pr/Pns > 0.5. Alpha = 1.0 for LRFD.

The 0.80 factor accounts for the combined effect of residual stresses (which reduce inelastic stiffness) and initial geometric imperfections (which create additional moments in the deflected shape).

2. Notional loads (Section C2.2b):

Ni = 0.002 × Yi    (applied as lateral load at each story level)

Where Yi = the gravity load at story level i from the applicable LRFD load combination. The 0.002 factor represents an initial out-of-plumb of L/500 (1/500 = 0.002), matching the AISC Code of Standard Practice erection tolerance.

Notional loads are applied in the direction that adds to the destabilizing effect. For frames with significant lateral loads (second-order drift ratio B2 ≥ 1.7), notional loads must be applied as a minimum in all combinations. For B2 < 1.7, notional loads may be applied only in gravity-only combinations.

3. Second-order analysis: The analysis must account for both P-Delta (story sway, frame-level) and P-delta (member curvature, member-level) effects. A rigorous second-order analysis (geometric nonlinear analysis) captures both simultaneously. Alternatively, the B1-B2 amplification method separates them.

B1-B2 amplification method

For engineers using linear analysis software, the B1-B2 method (AISC 360-22 Appendix 8) approximates second-order effects:

Mr = B1 × Mnt + B2 × Mlt

Where Mnt = moment from the no-translation (gravity) analysis, Mlt = moment from the lateral translation analysis.

B2 factor (story amplifier):

B2 = 1 / (1 - alpha × Pstory / Pe,story)

Where Pstory = total factored gravity load on the story, and Pe,story = elastic critical buckling load of the story. Pe,story can be estimated from:

Pe,story = RM × (sum H × L) / delta_H

Where H = story shear, L = story height, delta_H = first-order inter-story drift, and RM = 1 - 0.15 × (Pmf/Pstory) for moment frames.

B1 factor (member amplifier):

B1 = Cm / (1 - alpha × Pr / Pe1) ≥ 1.0

Where Cm = equivalent uniform moment factor (0.6 - 0.4 × M1/M2 for members with end moments, no transverse load) and Pe1 = pi² × EI / (K1 × L)² with K1 = 1.0 (no sway).

Worked example — B2 factor for a moment frame story

Given: 4-story moment frame, story 2: Pstory = 2,400 kips (total gravity load), story height L = 13 ft. Wind story shear H = 45 kips produces first-order drift delta_H = 0.32 in. Moment frame carries 60% of gravity (Pmf/Pstory = 0.60).

Step 1 — RM: RM = 1 - 0.15 × 0.60 = 0.91

Step 2 — Pe,story: Pe,story = 0.91 × (45 × 13 × 12) / 0.32 = 0.91 × 7020 / 0.32 = 0.91 × 21,938 = 19,964 kips

Step 3 — B2: B2 = 1 / (1 - 1.0 × 2400/19964) = 1 / (1 - 0.120) = 1 / 0.880 = 1.14

This means second-order effects amplify the lateral moments by 14%. Since B2 < 1.5, the structure is not overly sensitive to P-Delta, and K = 1.0 can be used for column design (a key advantage of the DM).

Drift limits

Drift calculations should use the reduced stiffness from the DM for consistency with the strength analysis. If drift governs the design, the DM's 0.80 reduction factor tends to increase member sizes compared to the old ELM approach.

Code comparison

AISC 360-22 Chapter C (USA): Direct Analysis Method is the primary method. Notional loads of 0.002Yi. Stiffness reduction of 0.80 × tau_b. K = 1.0 for all members when using the DM. The ELM is still permitted as an alternative (Appendix 7) but requires K > 1.0 for unbraced frames, which is a disadvantage.

AS 4100-2020 Section 4.4 (Australia): Uses a similar approach to the DM but applies notional horizontal forces of 0.002 × sum(Nf) at each story level (Section 4.3.6). Member effective lengths use the braced/sway frame classification. AS 4100 does not prescribe a universal stiffness reduction factor — instead, the column capacity factor alpha_c implicitly accounts for residual stresses through the column strength curve.

EN 1993-1-1 Section 5.2 (Eurocode 3): Uses an initial sway imperfection phi = 1/200 × alpha_h × alpha_m (where alpha_h and alpha_m depend on frame height and number of columns). Second-order effects must be included when alpha_cr < 10 (alpha_cr = ratio of elastic critical load to design load). If alpha_cr ≥ 10, the frame is "non-sway" and first-order analysis suffices. Eurocode permits the equivalent column method (buckling lengths) as an alternative.

CSA S16-19 Clause 8.4 (Canada): Requires notional loads of 0.005 × gravity load (larger than AISC's 0.002 to account for combined imperfection and inelastic effects). Second-order analysis is mandatory. The Canadian U2 factor is equivalent to AISC's B2 factor.

Common mistakes engineers make

1. Running first-order analysis without amplification. Software default is often linear (first-order) analysis. Without enabling P-Delta or applying B1-B2 amplification, column moments are underestimated by 10–30% in typical frames. Verify that the analysis includes second-order effects.

2. Applying notional loads in only one direction. Notional loads must be applied in the direction that produces the worst effect. For symmetric frames, this means checking both ±X and ±Y directions. Many engineers apply them in only one direction, missing the critical combination.

3. Forgetting the 0.80 stiffness reduction in drift calculations. If strength design uses the DM with 0.80 × EI, the drift calculations should use the same model for consistency. Using unreduced stiffness for drift but reduced stiffness for strength produces inconsistent results and underestimates actual drift.

4. Using K > 1.0 with the Direct Analysis Method. The entire point of the DM is that notional loads and reduced stiffness capture the same effects as K > 1.0. When using the DM, K = 1.0 for all members. Applying K > 1.0 on top of the DM double-counts the stability effects and produces overly conservative column designs.

AISC Direct Analysis Method — step-by-step procedure (Appendix 7 / Chapter C)

The Direct Analysis Method (DAM) is AISC 360-22's primary method for stability design. It was introduced to address the shortcomings of the Effective Length Method (ELM), which required engineers to calculate K-factors that were often inaccurate for real structures with partial fixity, leaning columns, and variable stiffness. The DAM directly models the physical phenomena that cause instability — residual stresses, geometric imperfections, and inelastic stiffness degradation — through three calibrated modifications to the analysis model.

Step 1: Apply notional loads

Ni = 0.002 x Yi  (at each story level, in each lateral direction)

Notional loads represent the effect of initial out-of-plumbness in the erected steel frame. The AISC Code of Standard Practice permits columns to be erected with a tolerance of L/500, which corresponds to a lean of 0.002 radians. This lean creates an additional overturning moment (P x Delta) that must be captured in the analysis.

Application rules:

• Notional loads must be applied in all load combinations that include gravity loads
• The direction of Ni must be chosen to produce the most destabilizing effect
• For two-way frames, apply Ni independently in each orthogonal direction
• If the ratio of maximum second-order drift to first-order drift (B2) is less than 1.7 for all stories, notional loads may be applied only in gravity-only combinations (ASCE 7 load combinations without lateral loads)
• If B2 >= 1.7 for any story, notional loads must be applied in ALL load combinations

Step 2: Reduce member stiffness

Apply the stiffness reduction factors to all members that contribute to the stability of the structure:

EI* = 0.80 x tau_b x EI    (flexural stiffness)
EA* = 0.80 x EA             (axial stiffness)

The tau_b factor accounts for inelastic stiffness degradation due to residual stresses:

Where alpha = 1.0 for LRFD and 1.6 for ASD. Pr is the required axial strength, and Pns is the nominal axial strength (cross-section squash load for tension, or critical buckling load for compression). For most beams and lightly loaded columns, tau_b = 1.0 and the full 0.80 reduction applies.

The combined effect of the 0.80 factor and tau_b is calibrated so that the analysis correctly predicts the strength of a frame with realistic imperfections and residual stresses. The 0.80 factor alone accounts for approximately 20% stiffness loss from the combined effects of geometric imperfections and residual stresses.

Step 3: Perform second-order analysis

The analysis must capture both types of P-delta effects:

• P-Delta (P-big-delta): Story-level sway effect. When a story displaces laterally by Delta, the gravity loads on that story create an additional overturning moment P x Delta. This is captured by a geometric nonlinear (P-Delta) analysis in most structural analysis software.

• P-delta (P-little-delta): Member-level curvature effect. When a column bows laterally between floor levels, the axial load creates an additional moment within the member. This effect is captured automatically in a rigorous second-order analysis with sufficient element subdivision (typically 2-4 elements per member), or by applying the B1 amplification factor.

Most commercial software packages (ETABS, SAP2000, RAM Structural System, RISA-3D) include P-Delta analysis as a standard feature. The engineer must verify that both effects are captured — some programs only capture P-Delta at the story level and require member subdivision to capture P-delta.

Step 4: Design members using K = 1.0

The key benefit of the DAM is that all member designs use an effective length factor K = 1.0. This eliminates the need to calculate K-factors using the alignment charts (Nomographs), which are based on idealized boundary conditions that rarely match reality. With K = 1.0:

• Column buckling length = actual unbraced length
• Beam-column interaction uses the actual member length
• No iteration between K-factor calculation and member design

Comparison: DAM vs Effective Length Method vs First-Order with B1/B2

Three methods are available for stability design in AISC 360-22. The DAM is the primary (preferred) method; the Effective Length Method (ELM) and the First-Order Analysis Method are permitted alternatives with specific limitations.

Method 1: Direct Analysis Method (DAM) — Chapter C

Procedure:

1. Apply notional loads Ni = 0.002 x Yi at each story
2. Reduce stiffness: EI* = 0.80 x tau_b x EI, EA* = 0.80 x EA
3. Perform second-order analysis (P-Delta and P-delta)
4. Design members using K = 1.0

When to use: This is the default method for all steel frames. It should be used for:

• All moment frames (braced and unbraced)
• Frames with leaning columns
• Frames with partial fixity at foundations
• Any frame where K-factors are difficult to determine accurately
• Frames in high-seismic regions where accurate force prediction is critical

Method 2: Effective Length Method (ELM) — Appendix 7

Procedure:

1. Apply notional loads Ni = 0.002 x Yi (in gravity-only combinations)
2. Use unreduced stiffness (no 0.80 factor, tau_b = 1.0)
3. Perform second-order analysis
4. Calculate K-factors using alignment charts or equations
5. Design members using the calculated K-factors

When to use: Only when ALL of the following conditions are met:

• The structure supports gravity loads through columns (no arch or cable structures)
• The ratio of maximum second-order drift to first-order drift (B2) does not exceed 1.5 at any story
• All members in the structure meet the strength requirements using their actual unbraced lengths (no member has a required strength that exceeds its available strength by more than a small margin)

The ELM is simpler conceptually (no stiffness reduction) but requires K-factors that are often difficult to calculate correctly. The alignment chart assumes idealized boundary conditions (fixed, pinned, or rigid) that do not account for partial fixity, foundation flexibility, or leaning columns. Errors in K-factor estimation of 20-50% are common.

Method 3: First-Order Analysis with B1/B2 Amplification — Appendix 8

Procedure:

1. Apply notional loads Ni = 0.002 x Yi
2. Run a first-order (linear) analysis
3. Calculate B1 (member amplifier) and B2 (story amplifier)
4. Amplify the first-order moments: Mr = B1 x Mnt + B2 x Mlt
5. Use K = 1.0 for member design

When to use: Only when B2 <= 1.5 at every story. This method is a simplification of the DAM that avoids running a true second-order analysis. It is useful for:

• Preliminary design and hand calculations
• Frames where second-order effects are small (stiff braced frames)
• Verification of software results

Comparison table

Advantages and limitations summary

DAM advantages:

• No K-factor calculation required (K = 1.0 for all members)
• Directly accounts for the three sources of instability: residual stresses, geometric imperfections, and geometric nonlinearity
• Works correctly for frames with leaning columns without manual adjustment
• Produces more accurate results for flexible frames and unusual geometries
• Required by AISC 360-22 for frames where B2 > 1.5

DAM limitations:

• The 0.80 stiffness reduction increases member sizes compared to an unmodified analysis
• Software must support second-order (P-Delta) analysis
• Iteration may be needed if tau_b changes significantly between analysis cycles (because tau_b depends on the axial load, which depends on the analysis results)

ELM advantages:

• Familiar to experienced engineers (the alignment chart method has been used since the 1960s)
• No stiffness reduction — members are designed using their full elastic stiffness
• Simpler for hand calculations on simple frames

ELM limitations:

• K-factors from alignment charts are often inaccurate for real structures
• Leaning columns require manual treatment (they do not contribute to lateral stiffness but add P-Delta effects)
• Foundation flexibility is difficult to incorporate into the alignment chart
• Limited to frames where B2 <= 1.5
• Does not account for residual stresses explicitly

First-Order + B1/B2 advantages:

• Can be performed with any structural analysis software (even linear-only programs)
• Conceptually straightforward — separate gravity and lateral analyses
• Useful for preliminary sizing and hand-checking software output

First-Order + B1/B2 limitations:

• B2 calculation requires an estimate of the story buckling load (Pe,story), which introduces approximation
• The B1 factor uses the member buckling load with K = 1.0, which may underestimate P-delta effects for non-prismatic members
• Limited to B2 <= 1.5 (stiff frames only)
• Does not capture interaction between P-Delta and P-delta effects (treats them independently)

Run this calculation

• Portal Frame Calculator
• Section Properties Database

Related references

• How to Verify Calculations
• Structural System Selection
• Effective Length Factors
• Column Design Guide
• steel beam capacity calculator
• High Rise Steel

Disclaimer

This page is for educational and reference use only. It does not constitute professional engineering advice. All design values must be verified against the applicable standard and project specification before use. The site operator disclaims liability for any loss arising from the use of this information.