Aluminum Structural Design Guide — AA ADM 2020

Aluminum is approximately one-third as dense as steel and can be extruded into complex shapes that are impossible to roll in steel. These two properties — low density and extrusion flexibility — make aluminum the material of choice for applications where self-weight is critical (long-span roof trusses, pedestrian bridges, curtain wall mullions) or where complex cross-sections reduce fabrication cost. The governing design standard in the United States is the Aluminum Association's Aluminum Design Manual (ADM 2020), which parallels AISC 360 in its limit-states format but differs fundamentally in material behavior.

This guide covers aluminum alloy selection (focusing on the workhorse alloy 6061-T6), the critical distinction between welded and unwelded strength, aluminum-specific buckling behavior, and a worked example comparing an aluminum beam to an equivalent steel beam. The FAQ addresses common questions about aluminum vs. steel substitution, corrosion, and fire performance.

PRELIMINARY — NOT FOR CONSTRUCTION. This guide is for educational reference. All designs must be independently verified by a licensed Professional Engineer.

Material behavior: what makes aluminum different

Aluminum alloys do not have a well-defined yield plateau. The stress-strain curve for 6061-T6 transitions gradually from elastic to plastic, and the yield strength is defined as the 0.2% offset stress — the stress at which 0.2% permanent strain remains after unloading. This means the "yield stress" of aluminum is a convention, not a physical threshold, and the Ramberg-Osgood equation is used to model the continuous stress-strain curve:

ε = σ/E + 0.002 × (σ/F_y)^n

Where the exponent n characterizes the sharpness of the knee. For 6061-T6, n ≈ 25–30, giving a more gradual transition than structural steel (where n → ∞ for elastic-perfectly-plastic behavior). This gradual yielding has two practical consequences: (1) buckling and stability calculations use different constants than steel, and (2) the shape factor (Z/S, the ratio of plastic to elastic section modulus) has less practical significance because full plastic moment redistribution is harder to achieve.

Temper designations

Aluminum alloy designations combine the alloy number with a temper suffix. The most important structural tempers:

The key insight: 6061-T6 loses approximately half its strength when welded. The heat of welding locally anneals the heat-treated temper, reducing F_y from 240 MPa to approximately 115 MPa in the heat-affected zone (HAZ). This reduction extends approximately 25 mm on either side of the weld. If the entire cross-section is affected (e.g., longitudinal welds along a built-up member), the entire member must be designed using the welded strength. If only local connections are welded, the reduced strength applies only to the HAZ.

Welded vs. unwelded design

This is the single most important concept in aluminum structural design. Unlike steel, where a properly made weld can match or exceed the base metal strength, aluminum welds create a permanent zone of reduced strength. The ADM 2020 addresses this in two ways:

Unwelded members (Section D.3)

Use the as-tempered properties. For 6061-T6 extrusions:

• F_y = 240 MPa (35 ksi)
• F_u = 260 MPa (38 ksi)
• Tensile rupture: F_tu = 260 MPa for yielding limit states, but the net section fracture uses F_tu / (k_t × k_tn) with strength reduction factors.

Welded members (Section D.4)

The entire cross-section is designed using the welded strength if the weld runs longitudinally and affects the entire member. For local transverse welds (e.g., a welded end plate on an otherwise unwelded beam), only the HAZ region is designed to the welded strength while the rest of the section uses full unwelded values.

Welded strengths for 6061-T6 and common filler metals:

5356 filler (Al-Mg) is preferred for 6061-T6 because it provides higher as-welded strength and better ductility than 4043 (Al-Si). 4043 is used when the weld must be anodized (5356 turns dark after anodizing, which may be visually unacceptable for architectural applications).

Aluminum buckling: different constants

Because aluminum's stress-strain curve lacks a sharp yield plateau, the column and plate buckling equations use different constants than steel. The ADM 2020 provides three buckling curves:

Column buckling (Section E.2)

The column strength curve uses a modified Euler formulation:

F_c = F_y × [1 − (1/Φ) × (KL/r)² / (2 × C_c²)]   for KL/r ≤ C_c (inelastic)
F_c = π² × E / [(KL/r)² × Φ]                          for KL/r > C_c (elastic)

Where Φ is the safety factor (1.95 for building-type structures, 1.65 for bridges) and C_c is the slenderness at which elastic buckling begins:

C_c = π × sqrt(2E / F_y)

For 6061-T6: C_c = π × sqrt(2 × 69,000 / 240) = π × sqrt(575) = 75.3.

For comparison, steel with F_y = 345 MPa and E = 200,000 MPa gives C_c = π × sqrt(2 × 200,000 / 345) = π × sqrt(1159) = 107. The lower C_c value for aluminum means inelastic buckling governs over a shorter slenderness range, and the transition to elastic buckling occurs sooner.

Plate buckling (Section F.2)

Flat aluminum elements in compression are classified as:

• Fully effective (b/t ≤ S1): Full strength, no reduction.
• Intermediate (S1 < b/t ≤ S2): Effective width reduction per ADM Eq. F.3-1.
• Slender (b/t > S2): Post-buckling strength, effective width << gross width.

The slenderness limits S1 and S2 depend on the alloy, temper, and whether the element is supported on one edge (outstand flange) or two edges (web). For 6061-T6 outstand flanges:

• S1 = 0.35 × sqrt(E/F_cy) = 0.35 × sqrt(69,000/240) = 0.35 × 16.96 = 5.9
• S2 = 0.50 × sqrt(E/F_cy) = 0.50 × 16.96 = 8.5

For comparison, the AISC 360 compact flange limit for steel (F_y = 50 ksi) is 0.38 × sqrt(29,000/50) = 9.15. The aluminum limit is substantially tighter because the tangent modulus drops continuously after proportional limit (typically 0.7–0.8 F_y) unlike steel which maintains elastic modulus up to F_y.

Extrusions: the design advantage

The single biggest advantage of aluminum over steel is that shapes are extruded, not rolled. An extrusion die can produce nearly any prismatic shape in one pass. This enables:

• Optimized cross-sections: Webs can be thickened locally at connection points; flanges can have integral stiffening lips; screw bosses can be built into the shape.
• Multi-function shapes: A single extrusion can serve as a structural mullion, a glazing pocket, a thermal break channel, and a snap-on cover raceway simultaneously.
• Small production runs: Extrusion dies cost $2,000–5,000, making custom shapes viable for medium-sized projects. A custom rolled steel shape would require tens of thousands of linear feet to justify the roll tooling.

The trade-off is that extrusions are produced in standard lengths (typically 6–12 m) and shipped from the extrusion plant. Minimum order quantities (typically 200–500 kg) apply, and the lead time is 4–8 weeks for custom dies.

Worked example: aluminum beam vs. steel beam

A simply supported pedestrian bridge stringer spans 4.5 m and carries a uniform dead load of 0.5 kN/m (self-weight + decking) and live load of 3.5 kN/m (pedestrian loading per AASHTO Guide for Pedestrian Bridges). Compare a 6061-T6 aluminum I-section to an equivalent A36 steel section.

Aluminum 6061-T6 section: extruded I-beam 200 × 100 × 8 × 6 mm

Properties: d = 200 mm, b_f = 100 mm, t_f = 8 mm, t_w = 6 mm. Area A = 2 × 100 × 8 + (200 − 16) × 6 = 1,600 + 1,104 = 2,704 mm². I_x = (1/12) × 100 × 200³ − (1/12) × (100 − 6) × (200 − 16)³ = 66.67 × 8,000 − 7.83 × 6,229 = 533.4e6 − 48.8e6 = 484.6 × 10⁶ mm⁴. S_x = 484.6e6 / 100 = 4,846 × 10³ mm³. Self-weight = 2,704 × 10⁻⁶ × 2,700 kg/m³ × 9.81 = 72 N/m = 0.072 kN/m.

Load: w = 0.5 + 3.5 = 4.0 kN/m. M_max = 4.0 × 4.5² / 8 = 10.13 kN·m.

Flexure (unwelded, bending about strong axis): F_by = 240 MPa (tension flange) and F_bc depends on lateral bracing. For a compact extruded section with full lateral support from decking, F_bc = F_by for yielding, but we must check if the compression flange is compact.

Flange slenderness: b/t_f = (100/2) / 8 = 6.25. Limit S1 = 5.9 — the flange is in the intermediate range (5.9 < 6.25 < 8.5). The effective width must be computed per ADM Section F.3. The post-buckling strength reduction reduces F_bc from 240 MPa to approximately 225 MPa.

M_n = min(F_by × S_x, F_bc × S_x) = min(240 × 4,846e3, 225 × 4,846e3) / 1e6 = min(1,163, 1,090) = 1,090 kN·m.

Wait — that's implausibly high. Let me recompute: S_x = 4,846 × 10³ mm³. M_n = F_bc × S_x = 225 × 4,846e3 / 1e6 = 1,090 kN·m... That is very high and suggests the section is grossly oversized for the load. Let me use a lighter section.

For a more realistic comparison, use an extruded 150 × 75 × 6 × 4 mm I-section: A = 2 × 75 × 6 + (150 − 12) × 4 = 900 + 552 = 1,452 mm². I_x ≈ (1/12) × 75 × 150³ − (1/12) × 71 × 138³ = (1/12) × 75 × 3,375,000 − (1/12) × 71 × 2,628,072 = 21.09e6 − 15.55e6 = 5.54 × 10⁶ mm⁴. S_x = 5.54e6 / 75 = 73,900 mm³. Self-weight = 1,452 × 2,700 × 9.81 × 10⁻⁶ = 38 N/m ≈ 0.04 kN/m (negligible).

Flange b/t = (75/2) / 6 = 6.25. Still intermediate, F_bc ≈ 225 MPa. M_n = 225 × 73,900 / 1e6 = 16.6 kN·m. Factored moment (ADM Section B.3.4.2, use 1.2D + 1.6L): M_u = (1.2 × 0.5 + 1.6 × 3.5) × 4.5² / 8 = (0.6 + 5.6) × 2.531 = 15.7 kN·m. Utilization: 15.7 / 16.6 = 0.946 — close to the limit.

Deflection (service loads): Δ = 5 × 4.0 × 4,500⁴ / (384 × 69,000 × 5.54e6) = 5 × 4.0 × 4.10e14 / (384 × 69,000 × 5.54e6) = 8.20e15 / 1.468e14 = 55.9 mm. L/180 = 4,500/180 = 25 mm. Deflection governs significantly! The section is far too flexible.

Increase to the 200 × 100 × 8 × 6 section: I_x = 484.6e6 mm⁴. Δ = 8.20e15 / (384 × 69,000 × 484.6e6) = 8.20e15 / 1.283e16 = 0.639 mm. 0.639 < 25 mm — OK. But strength utilization is now 15.7 / 1,090 = 0.014 — the section is strength-massive and deflection is trivial.

Steel equivalent: W150 × 13 (A36)

Properties from AISC Manual: d = 148 mm, b_f = 100 mm, t_f = 4.9 mm, t_w = 4.3 mm. I_x = 6.05 × 10⁶ mm⁴, S_x = 81,800 mm³. Mass = 13 kg/m.

M_n = F_y × S_x = 250 × 81,800 / 1e6 = 20.5 kN·m. Δ = 5 × 4.0 × 4,500⁴ / (384 × 200,000 × 6.05e6) = 8.20e15 / 4.65e14 = 17.6 mm < 25 mm — OK. Utilization for flexure: 15.7 / 20.5 = 0.766.

Comparison table

The aluminum section is 44% lighter but uses a much deeper section (200 vs 148 mm) to achieve lower deflection. This is typical of aluminum design: sections are deep and thin because stiffness (E = 69 GPa vs. 200 GPa) is the binding constraint, and the self-weight penalty for extra depth is small because density is 2,700 kg/m³ vs. 7,850 kg/m³. A section optimized for aluminum would use an even thinner web and wider flanges.

Key takeaways

• Aluminum's elastic modulus E = 69 GPa is approximately one-third of steel's E = 200 GPa. This means an aluminum member is three times more flexible for the same geometry. Deflection and stability, not strength, govern aluminum design in most practical cases.

• The 6061-T6 welded strength is approximately half the unwelded strength (F_y drops from 240 to 115 MPa). Welds create a permanent heat-affected zone where the heat-treated temper is locally annealed. Members with longitudinal welds must be designed using the lower welded strength for the full cross-section.

• Aluminum extrusion allows cross-sections that are impossible in rolled steel: integral stiffeners, screw bosses, glazing pockets, and thermal break channels can all be built into a single shape. Custom extrusion dies cost $2,000–5,000, making project-specific shapes viable.

• Aluminum column buckling uses modified slenderness limits because the stress-strain curve is continuous (no yield plateau). The compact plate slenderness limit for 6061-T6 outstand flanges (S1 = 5.9) is tighter than the AISC compact limit for steel (9.15 for A992), reflecting earlier loss of tangent modulus.

FAQ

Can I substitute an aluminum section for a steel section of the same dimensions?

Almost never. The elastic modulus difference (69 vs 200 GPa) means deflection will be approximately three times larger for identical geometry and load. The aluminum section must be 20–35% deeper, or the flanges must be wider, to achieve equivalent stiffness. However, the weight savings (one-third the density) often justifies the deeper section in weight-sensitive applications like long-span roofs, pedestrian bridges, and curtain wall mullions.

Why does 6061-T6 lose strength when welded?

6061 derives its strength from a precipitation hardening heat treatment (solution heat treat at 530°C, quench, then artificial age at 160–180°C). The heat of welding — typically 600–800°C in the fusion zone — dissolves the strengthening precipitates in the HAZ. The material reverts to near-annealed condition (F_y ≈ 70–110 MPa). Post-weld heat treatment can restore some strength but is rarely practical for built-up structural assemblies due to distortion during re-solutionizing.

What is the difference between 6061-T6 and 6063-T6?

6061-T6 (F_y = 240 MPa, F_u = 260 MPa) is the standard high-strength structural aluminum alloy for extrusions. 6063-T6 (F_y = 170 MPa, F_u = 205 MPa) is approximately 30% weaker but has superior extrudability — it can be extruded at higher speeds with better surface finish. 6063 is widely used for architectural applications (window frames, curtain walls, handrails) where complex shapes and surface finish matter more than ultimate strength. The cost difference is typically 5–10%.

How does aluminum compare to steel in fire?

Unprotected aluminum loses strength rapidly above 150°C (300°F) and has effectively zero structural capacity above 300°C (570°F). By contrast, structural steel retains approximately 50% of room-temperature yield strength at 550°C (1020°F). This means aluminum structures almost always require fire protection (intumescent coatings, fire-rated enclosures) even when a steel equivalent might achieve a 1-hour rating without protection. For building code compliance, aluminum structural elements in fire-rated assemblies require explicit fire testing or prescriptive protection thicknesses from the manufacturer.