Residual Stresses in Structural Steel — Sources, Patterns, and Design Effects
Residual stresses are internal stresses that exist in steel members without any applied external load. They arise from non-uniform cooling during manufacturing (hot rolling, welding, flame cutting) and from cold working (bending, punching). Residual stresses do not affect the ultimate plastic capacity of a cross-section (they self-equilibrate), but they significantly reduce the inelastic buckling resistance of columns and beams and affect fatigue life. All design code column curves implicitly account for typical residual stress patterns.
Sources of residual stress
Hot rolling
When a W-shape exits the rolling mill, the thin flange tips and web cool faster than the thick web-to-flange junction. The parts that cool first contract first and develop residual compression. The junction region, constrained by the already-cooled extremities, cools last under restraint and develops residual tension.
Typical pattern for hot-rolled W-shapes:
- Flange tips: compressive, approximately 0.3*Fy (10-15 ksi for A992)
- Web-flange junction: tensile, approximately 0.3*Fy
- Web center: mildly compressive or near zero
- The stress distribution is roughly parabolic across the flange width
This pattern means the flange tips, which are farthest from the neutral axis and most critical for buckling resistance, start with compressive residual stress. Under applied compression, these fibers reach yield first, reducing the effective stiffness of the cross-section and initiating inelastic buckling at loads below the Euler load.
Welding
Weld metal shrinks as it cools, producing very high tensile residual stresses (up to Fy) in the heat-affected zone (HAZ). Balancing compressive stresses develop in the surrounding parent metal.
Typical pattern for welded I-sections:
- HAZ near flange-to-web weld: tensile, up to Fy (50 ksi for A992)
- Flange tips: compressive, 0.2-0.5*Fy depending on flange width
- Web: variable, depends on weld sequence
Welded sections have higher residual stresses than hot-rolled sections, which is why Eurocode 3 assigns them less favorable buckling curves (curves c and d vs. curves a and b for hot-rolled shapes). AISC uses a single column curve calibrated primarily to hot-rolled shapes.
Flame cutting
Flame-cut edges develop tensile residual stresses (approaching Fy) along the cut edge, with compensating compression in the interior. Universal mill plates (sheared edges) have lower residual stresses than flame-cut plates. For plate girder flanges, the edge condition affects the residual stress pattern and thus the LTB resistance.
Cold working
Cold bending, press-braking (for HSS), and bolt-hole punching introduce residual stresses through plastic deformation. Cold-formed HSS sections have through-thickness bending residual stresses at the corners, which are different from the membrane-type stresses in hot-rolled sections. This is why AS 4100 and Eurocode 3 use different buckling curves for hot-formed vs. cold-formed hollow sections.
Effect on column buckling
Residual stresses reduce column capacity in the inelastic buckling range (intermediate slenderness). The mechanism:
- Applied compression adds to residual compression at flange tips
- Flange tips yield first (at applied stress = Fy - sigma_rc, where sigma_rc is the compressive residual stress)
- The yielded zones no longer contribute to bending stiffness
- The effective moment of inertia decreases, lowering the tangent modulus
- Buckling occurs when the tangent-modulus load equals the applied load
For a typical hot-rolled W-shape with sigma_rc = 0.3Fy, early yielding begins at an applied stress of 0.7Fy. This is why the AISC column curve begins to deviate from the squash load at relatively low slenderness ratios.
Quantified effect: At KL/r = 80 (moderate slenderness), residual stresses reduce column capacity by approximately 15-25% compared to the ideal Euler curve. At KL/r < 30 (stocky) and KL/r > 150 (slender), the effect is minimal because yielding or elastic buckling dominates respectively.
Effect on lateral-torsional buckling
Residual stresses affect beam LTB similarly to column buckling. The compression flange tips yield early, reducing the torsional and warping stiffness of the beam. This effect is captured in the AISC Chapter F LTB equations through the inelastic zone (Lp < Lb < Lr), where the capacity transitions linearly from Mp to 0.7FySx. The 0.7 factor directly reflects the expected onset of yielding due to residual stresses (Fy - 0.3Fy = 0.7Fy).
Post-weld heat treatment (PWHT) and stress relief
Thermal stress relief (PWHT)
Heating the fabricated member to 600-650 degC (1100-1200 degF) and holding for 1 hour per inch of thickness relaxes residual stresses by allowing creep deformation. Controlled cooling (typically in the furnace or under insulation) prevents re-introduction of thermal stresses.
- Reduces residual stresses by 70-90%
- Required for: thick welded joints (AWS D1.1 for t > 2"), pressure vessels (ASME Section VIII), fatigue-critical structures
- Not required for: most structural steel building connections (residual stresses are accounted for in design codes)
- Cost: significant -- requires furnace large enough for the member and adds days to fabrication schedule
Vibratory stress relief (VSR)
An alternative to PWHT that uses mechanical vibration to redistribute residual stresses. Less effective than PWHT (30-50% reduction) and not accepted by all codes. Sometimes used for large fabrications that cannot fit in a furnace.
Effect on fatigue
Tensile residual stresses from welding are particularly harmful for fatigue because they keep the weld detail in tension even when the applied load is compressive. This means the full applied stress range contributes to fatigue damage, regardless of the minimum stress in the cycle. AISC 360 Appendix 3 fatigue provisions account for this by basing the fatigue check on the full stress range (f_SR), not the stress amplitude.
Important: Improving the weld detail category (e.g., grinding weld toes, using TIG-dressed welds) improves fatigue life more than stress relief because it reduces the stress concentration factor at the weld toe.
Practical tip: when to worry about residual stresses
For most building structures, residual stresses are already accounted for in the design code column and beam curves. No additional calculation is needed. However, explicitly consider residual stresses when:
- Designing fatigue-critical structures (crane runways, bridges, vibrating equipment supports)
- Using thick welded plate (t > 2") where AWS D1.1 may require PWHT
- Performing advanced nonlinear analysis (FEA) where initial stresses must be modeled explicitly
- Evaluating serviceability of precision members where residual-stress-induced distortion affects fit-up
Common mistakes
- Ignoring residual stresses in FEA. Linear elastic analysis inherently misses the early-yielding effect. Nonlinear FEA of columns and beams must include initial residual stress patterns to match experimental behavior.
- Assuming PWHT eliminates all residual stresses. PWHT reduces stresses by 70-90%, not to zero. Some residual stress always remains.
- Specifying unnecessary PWHT. PWHT for standard building connections adds cost and schedule without structural benefit -- the design codes already account for residual stresses through the column and beam curves.
- Neglecting welding sequence effects. Welding sequence affects the residual stress pattern and distortion. For large fabrications, a welding sequence plan (balanced welding, back-step technique) reduces distortion and peak residual stresses.
- Confusing residual stress effects with material defects. Residual stresses reduce buckling capacity but do not reduce ductility or toughness (unless they cause cold cracking, which is a separate issue related to hydrogen, restraint, and preheat).
Residual stress patterns in rolled shapes
The magnitude and distribution of residual stresses vary significantly by shape type, manufacturing process, and member size. The following table summarizes typical patterns for common North American structural shapes:
| Shape type | Manufacturing | Flange tips | Web-flange junction | Web center | Magnitude (ksi, A992) |
|---|---|---|---|---|---|
| W-shape (hot-rolled, moderate) | Hot rolling | Compression | Tension | Near zero | 10-15 (0.2-0.3 Fy) |
| W-shape (hot-rolled, heavy) | Hot rolling | Compression | Tension | Compression | 12-18 (0.25-0.35 Fy) |
| HSS (hot-formed) | Hot forming | Compression (outside) | Tension (corners) | Variable | 5-12 (0.1-0.25 Fy) |
| HSS (cold-formed) | Cold forming | Tension (corners) | Compression (flats) | Variable | 15-25 (0.3-0.5 Fy) |
| Built-up I (welded, fillet) | Welding | Compression | Tension (near weld) | Variable | 15-25 (0.3-0.5 Fy) |
| Built-up plate girder | Welding + flame cutting | Tension (cut edge) | Tension (near weld) | Compression | 20-30 (0.4-0.6 Fy) |
| Double angle | Hot rolling | Compression (toe) | Tension (heel) | N/A | 8-12 |
The key insight: hot-rolled shapes have lower residual stresses than welded built-up sections because the entire cross-section cools more uniformly. This is why Eurocode 3 assigns more favorable buckling curves (curves a, b) to hot-rolled shapes and less favorable curves (curves c, d) to welded sections.
Residual stress distribution in a typical W12x65
For a W12x65 (A992), the residual stress distribution across the flange width is approximately parabolic:
Flange tip (compression): -12 ksi (0.24 Fy)
Quarter-point: -6 ksi (0.12 Fy)
Flange center (web junction): +12 ksi (0.24 Fy, tension)
The sign convention is: negative = compression, positive = tension. Under applied axial compression of 30 ksi, the flange tips reach yield at 30 + 12 = 42 ksi (below Fy = 50 ksi), while the web-flange junction remains elastic at 30 - 12 = 18 ksi. This non-uniform yielding pattern reduces the effective flexural stiffness of the cross-section.
Welding residual stress distribution
Welding introduces the highest magnitude residual stresses in structural steel members. The distribution depends on weld type, sequence, and joint geometry:
Longitudinal residual stresses (parallel to weld)
| Region | Stress state | Approximate magnitude | Width of affected zone |
|---|---|---|---|
| Weld centerline | Tension, up to Fy | 40-50 ksi (A992) | Equal to weld width |
| Heat-affected zone (HAZ) | Tension, 0.5-0.8 Fy | 25-40 ksi | 2-3x weld width |
| Transition zone | Tension to compression gradient | 0-20 ksi | 3-5x weld width |
| Far field (base metal) | Compression (balancing) | 5-15 ksi | Remainder of cross-section |
Transverse residual stresses (perpendicular to weld)
Transverse residual stresses are typically 30-50% of longitudinal values. At the weld start and stop points, transverse stresses can approach Fy due to the triaxial restraint condition. This is why weld defects (cracks, porosity) are most common at weld terminations.
Through-thickness residual stresses
In thick welded joints (> 1 in.), residual stresses vary through the thickness of the weld. The last-deposited pass has the highest tensile residual stress at the surface, while the root pass may be in compression. This through-thickness variation is why UT inspection of thick CJP welds requires scanning from both surfaces.
Thermal cutting effects
Flame cutting and plasma cutting introduce residual stresses different from those caused by rolling or welding:
| Cutting method | Cut edge stress | Affected depth | Typical magnitude | Mitigation |
|---|---|---|---|---|
| Oxy-fuel (flame cutting) | Tension | 3-6 mm (1/8-1/4 in.) | 30-50 ksi (up to Fy) | Grinding cut edge, stress relief |
| Plasma cutting | Tension | 2-4 mm | 25-40 ksi | Grinding, or specify universal mill plate |
| Laser cutting | Tension | 1-3 mm | 15-30 ksi | Generally acceptable as-is |
| Mechanical shearing | Compression + tension | 1-2 mm | 10-20 ksi | Edge condition varies by machine |
For plate girders, the flange edges produced by flame cutting have tensile residual stresses that reduce the local buckling resistance of the compression flange. AISC 360-22 Section F4 accounts for this indirectly through the plate buckling coefficient. For heavy plate girders (flanges over 2 in. thick), some specifications require grinding the flame-cut edges to remove the heat-affected zone.
Stress relief methods
Thermal stress relief (PWHT)
| Parameter | Typical range | Notes |
|---|---|---|
| Temperature | 1100-1200F (595-650C) | Must not exceed tempering temperature for Q&T steels |
| Soak time | 1 hour per inch of thickness | Minimum 1 hour regardless of thickness |
| Heating rate | 400F/hr maximum (200C/hr) | Faster rates cause thermal gradients and new stresses |
| Cooling rate | 400F/hr maximum to 600F, then air cool | Controlled cooling in furnace or under insulation |
| Resulting reduction | 70-90% of original residual stress | Never eliminates 100% of residual stress |
| Applicable steels | A36, A992, A572 (all grades) | A514 (Q&T) requires special temperature limits (max 1150F) |
Mechanical stress relief
| Method | Reduction achieved | Cost | Applicability | Code acceptance |
|---|---|---|---|---|
| Vibratory stress relief (VSR) | 30-50% | Low ($500-2000 per member) | Large fabrications, simple shapes | Not universally accepted; AWS D1.1 does not recognize VSR as equivalent to PWHT |
| Shot peening | 40-60% (surface only) | Moderate | Small components, fatigue-critical surfaces | Used in aerospace and mechanical components, not typical for structural steel |
| Mechanical stretching | 50-70% | High (requires specialized equipment) | Plates, simple tension members | Not typical for building construction |
| Thermal cycling (multiple PWHT cycles) | 80-95% | High (multiple furnace cycles) | Critical applications | Required for some nuclear and pressure vessel applications |
Practical guidance on stress relief selection
| Situation | Recommended approach | Rationale |
|---|---|---|
| Standard building connections (A992) | No stress relief needed | Design codes account for residual stresses |
| Thick welded joints (t > 2 in.) | PWHT per AWS D1.1 | Reduces hydrogen cracking risk and improves toughness |
| Fatigue-critical structures (crane runways) | PWHT for critical joints | Reduces mean stress and improves fatigue life |
| Large built-up members (no furnace available) | VSR as supplementary measure | Better than nothing, but not equivalent to PWHT |
| Seismic moment frame connections | No stress relief (not practical) | Design detailing (RBS, etc.) manages the issue |
Impact on column buckling: AISC Commentary perspective
AISC 360-22 Commentary Chapter E provides the theoretical background for the single column curve used in US practice. The commentary explains that the AISC column curve (Equation E3-1/E3-2) was calibrated against approximately 100 column tests covering a range of shapes, residual stress patterns, and initial out-of-straightness values.
Comparison with multiple column curve approaches
| Code | Number of curves | Basis for curve selection | AISC equivalent (approximate) |
|---|---|---|---|
| AISC 360-22 | 1 curve | Calibrated average of all shapes | Single curve for all |
| Eurocode 3 EN 1993-1-1 | 5 curves (a0, a, b, c, d) | Based on shape type, manufacturing, and residual stress level | Curve b for hot-rolled W-shapes |
| AS 4100-2020 | 5 curves (1-5) | Based on section type and residual stress category | Curve 2-3 for hot-rolled sections |
| CSA S16-19 | 3 curves (1, 2, 3) | Based on shape type and manufacturing | Curve 2 for W-shapes |
The AISC single-curve approach was deliberately chosen for simplicity. The curve is positioned to be safe for all common structural shapes, meaning it is slightly conservative for hot-rolled W-shapes (which have lower residual stresses) and slightly unconservative for heavy welded built-up sections (which have higher residual stresses). The overall safety is maintained by the resistance factor phi = 0.90.
Measurement methods for residual stress
| Method | Principle | Accuracy | Destructive? | Cost | Typical application |
|---|---|---|---|---|---|
| Sectioning (hole drilling) | Release strain measured by strain gauges when material is removed | +/- 5-10% of Fy | Yes (small hole) | Moderate | Research, quality control of critical members |
| X-ray diffraction | Measures lattice strain at the surface | +/- 3-5% of Fy | No (surface only) | High | Aerospace, research |
| Neutron diffraction | Measures lattice strain through the thickness | +/- 3-5% of Fy | No | Very high (requires neutron source) | Research, nuclear applications |
| Ultrasonic (acoustoelastic) | Correlation between wave velocity and stress state | +/- 10-15% of Fy | No | Low-Moderate | Field screening, large structures |
| Barkhausen noise | Magnetic domain wall motion correlates with stress | Qualitative to semi-quantitative | No | Low | Surface stress mapping on ferromagnetic steel |
For structural engineering practice, residual stress measurement is rarely required. The most common application is in research validating finite element models of column and beam behavior. However, forensic investigations of unexpected failures may employ hole-drilling or ultrasonic methods to determine whether residual stresses contributed to the failure.
Mitigation strategies table
| Strategy | Description | Effectiveness | Cost impact | When to apply |
|---|---|---|---|---|
| Preheat before welding | Reduces cooling rate, lowers residual stress magnitude | Moderate (10-20% reduction) | Low (labor and fuel) | Always per AWS D1.1 minimums |
| Controlled welding sequence | Balanced welding, backstep technique | Significant (20-40% reduction in peak stress) | Low (planning and supervision) | All multi-pass welds and large fabrications |
| Interpass temperature control | Maintains consistent thermal state | Moderate | Low (monitoring) | All multi-pass welds |
| PWHT (furnace) | Heats to 1100-1200F, holds, controlled cool | High (70-90% reduction) | High (furnace, schedule) | Thick joints, fatigue-critical, per specification |
| VSR (vibratory) | Mechanical vibration during and after welding | Low-Moderate (30-50% reduction) | Low | Large fabrications where PWHT not practical |
| Weld toe grinding | Removes stress concentration at weld toe | High for fatigue life (not for residual stress) | Moderate (labor) | Fatigue-critical details |
| Peening (hammer or shot) | Plastic deformation at surface introduces compression | Moderate (surface only) | Low-Moderate | Fatigue-critical surfaces, not a substitute for PWHT |
| Use hot-rolled shapes instead of welded built-up | Lower residual stresses in mill-produced shapes | Significant | May reduce cost (less fabrication) | Always prefer standard rolled shapes when available |
Run this calculation
Related references
- Column Buckling Equations
- Column Curve
- Lateral-Torsional Buckling
- Fatigue Design
- Stress-Strain Relationship
- How to Verify Calculations
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. The site operator disclaims liability for any loss arising from the use of this information.
[object Object]
[object Object]
Frequently Asked Questions
What is the recommended design procedure for this structural element?
The standard design procedure follows: (1) establish design criteria including applicable code, material grade, and loading; (2) determine loads and applicable load combinations; (3) analyze the structure for internal forces; (4) check member strength for all applicable limit states; (5) verify serviceability requirements; and (6) detail connections. Computer analysis is recommended for complex structures, but hand calculations should be used for verification of critical elements.
How do different design codes compare for this calculation?
AISC 360 (US), EN 1993 (Eurocode), AS 4100 (Australia), and CSA S16 (Canada) follow similar limit states design philosophy but differ in specific resistance factors, slenderness limits, and partial safety factors. Generally, EN 1993 uses partial factors on both load and resistance sides (γM0 = 1.0, γM1 = 1.0, γM2 = 1.25), while AISC 360 uses a single resistance factor (φ). Engineers should verify which code is adopted in their jurisdiction.