Welding Across an I-Beam Flange — What D1.1, D1.5, and AASHTO Actually Say
Welding across the tension flange of an I-beam is not prohibited by AWS D1.1 — it is fatigue-categorized. Section 4.17 with Table 4.5 places transverse welds in Category C, E, or E*, dropping the allowable stress range. D1.5 Clause 12 layers FCM provisions on top for bridge work.
The shop-floor question: “don’t weld across the flange” is treated as gospel without a code citation. The actual answer lives across two standards. AWS D1.1:2025 handles non-bridge structural work via the Section 4.17 + Table 4.5 fatigue framework. AASHTO/AWS D1.5:2025 handles bridge work via the Clause 12 Fracture Control Plan. The two layers stack, and conflating them is what causes the “everything is anecdote” problem.
The Rule of Thumb in One Paragraph
Welding transversely across an I-beam tension flange is not prohibited by AWS D1.1:2025. It is fatigue-categorized: Section 4.17 with Table 4.5 places transverse welds on tension-loaded plate elements into Category C, E, or E*, which produces a lower allowable stress range than longitudinal welds. The section has to grow to carry the same cyclic load — so on a real project, the design just doesn’t use the detail. On bridge work governed by AWS D1.5:2025, all welds to tension zones of fracture-critical members trigger Clause 12 Fracture Control Plan provisions on top of the D1.1 fatigue penalty — mandatory CVN testing, hydrogen-controlled electrodes, drying schedules, and qualified inspectors. Static loading and EOR-stamped drawings can override the rule of thumb in either case. Live-load buckling during welding is a separate concern.
Plain D1.1 (Non-Bridge) — Fatigue-Categorized, Not Prohibited
For structural steel work covered by AWS D1.1:2025 — buildings, industrial frames, transmission towers, anything that is not a highway bridge — the design rule for cyclically loaded transverse welds lives in Part C of Clause 4 (Sections 4.15 through 4.17). The framework is fatigue-design, not prohibition.
Per Section 4.15.1: “No evaluation of fatigue resistance shall be required if the live load stress range is less than the threshold stress range, F_TH (see Table 4.5).” Below F_TH the detail has infinite life; above F_TH the allowable stress range comes from Formulas (4-12) through (4-19) and the per-category constants C_f and F_TH in Table 4.5. Figure 4.16 plots the curves graphically for stress categories A, B, B*, C, D, E, E*, and F.
Where do transverse flange welds land? Section 4.17.2 states it directly: “For tension-loaded plate elements at cruciform, T and corner joint details with CJP welds, PJP welds, fillet welds or combinations of the preceding, transverse to the direction of stress… Category C” (Formulas 4-16 and 4-17). Specific configurations land per Table 4.5 Section 5:
- Line 5.4 — Weld metal and base metal in or adjacent to CJP groove welds in T- or corner joints, weld reinforcement not removed: Category C (C_f = 44 × 10⁸; F_TH = 10 ksi [69 MPa]).
- Line 5.7 — Pair of fillet welds on opposite sides of a tension-loaded plate element, transverse: Category C at the weld toe; root cracks use a separate formula.
- Line 5.8 — Base metal of tension-loaded plate elements and on built-up shapes and rolled beam webs or flanges at toe of transverse fillet welds, adjacent to welded transverse stiffeners: Category C.
Compare longitudinal welds parallel to stress (Section 3 of Table 4.5):
- Line 3.1 — Continuous longitudinal CJP groove welds in built-up members: Category B (C_f = 120 × 10⁸; F_TH = 16 ksi [110 MPa]).
- Line 3.2 — Continuous longitudinal CJP with backing left in place, or continuous PJP: Category B' (C_f = 61 × 10⁸; F_TH = 12 ksi [83 MPa]).
Same physical weld, different geometry. The transverse case loses roughly a factor of 2-4 on the threshold stress range and a similar margin on cycle life. To carry the same cyclic load with a transverse fillet, the section grows. On most non-bridge projects the engineer either reroutes the load path or uses a different connection — the rule of thumb is the field shorthand for “the category penalty makes this impractical.”
One important framing change in current D1.1: Section 4.15.4 states “This code no longer recognizes a distinction between redundant and nonredundant members.” The fracture-critical concept was moved out of D1.1 into D1.5. So if your project is governed by D1.1 alone, FCM is not in your binding code. If your project is governed by D1.5, FCM is.
Bridge Work — D1.5 Clause 12 FCM Provisions
For highway bridges governed by AASHTO/AWS D1.5:2025, transverse welds on tension flanges trigger an additional layer beyond the D1.1 fatigue category penalty: the Clause 12 Fracture Control Plan (FCP). Per Section 12.1: “This clause shall apply to fracture-critical nonredundant members. All steel bridge members and member components specified on the contract drawings or elsewhere in the contract documents as fracture critical shall be subject to the additional provisions of this clause.”
Section 12.2.2 defines the member: “AASHTO LRFD Bridge Design Specifications define an FCM as a steel primary member or portion thereof subject to tension whose failure would probably cause a portion of or the entire bridge to collapse.”
The attachment rule in D1.5 Section 12.2.2.1 is the one that catches engineers off guard: “Any attachment welded to a tension zone of an FCM, except for bearing sole plates, shall be considered an FCM when any dimension of the attachment exceeds 100 mm [4 in] in the direction parallel to the calculated tensile stress in the FCM.” A 4-inch-long stiffener welded transverse to the tension flange of an FCM girder is itself an FCM. All welds attaching it inherit the FCP.
Per D1.5 Section 12.2.2.2: “All welds to FCMs, except for those to bearing sole plates, shall be considered fracture critical and shall conform to the requirements of this FCP. Welds to compression members or compression areas of bending members shall not be defined as fracture critical.”
What does the FCP add? Mandatory CVN toughness testing per Section 12.6.3. Hydrogen-controlled electrodes per the H-designator system (H4, H8, H16). Strict electrode drying and storage schedules per Sections 12.6.4 through 12.6.6. Lead Inspector qualification requirements (minimum three years of steel bridge fabrication inspection per D1.5 Section 12.16.1.1). Separate preheat tables (Tables 12.4 through 12.8) that add hydrogen and heat input as lookup axes beyond steel grade and thickness.
The combination is what makes “weld across the flange” expensive on bridge work: D1.1’s fatigue category penalty drives the section size, then D1.5’s FCP adds NDE, electrode controls, and inspector cost on top.
Static vs Cyclic — Why the Rule Exists
Part C of D1.1 Clause 4 (Sections 4.15 through 4.18) governs cyclically loaded connections. It does not govern static loading. A 1/2-inch-thick flange with a 1-inch transverse weld carrying static gravity load is not constrained by Table 4.5 at all — the fatigue framework only fires when the live load produces a stress range above F_TH.
The shop-floor reality is that most cases that get welded across a flange in the field end up cyclic over the structure’s life. A trailer frame sees road oscillation. A bridge sees truck loading. A crane runway sees lift cycles. Even an industrial building can pick up cyclic load if someone bolts an HVAC unit, a conveyor, or vibrating equipment to the structure five years after construction. The rule of thumb exists because the category of the detail follows the structure, not the welder’s original intent.
Per Section 4.16.1: “Calculated stresses and stress ranges shall be nominal, based upon elastic stress analysis at the member level. Stresses need not be amplified by stress concentration factors for local geometrical discontinuities.” The category constants in Table 4.5 already bake in the stress concentration of the joint geometry. The Commentary C-4.17.2 is explicit: “The stress range cycle life curves criteria provided by Formulas (4-12) through (4-23) and plotted graphically in Figure 4.16 were developed through research sponsored by National Cooperative Highway Research Program (NCHRP) on actual details which incorporated realistic geometrical discontinuities making it inappropriate to amplify calculated stresses to account for notch effect.”
That answers the “why don’t we just amplify the stress” question. The category itself is the amplification — the NCHRP test program ran real specimens with real geometric discontinuities to derive the curves.
Welding While the Beam Is Under Load — Live-Load Buckling (Separate Issue)
A common shop comment is “you can’t weld across the flange while the beam is under load — it could buckle.” That is correct, but it is a different concern from fatigue category. The live-load-buckling issue is a localized loss of section stiffness during the welding operation: the heat-affected zone temporarily loses yield strength as it passes through the critical temperature range, and if the member is carrying live gravity load at that moment, the heated cross-section can deflect, twist, or buckle laterally.
This is a fabrication-sequence and shoring concern, not a fatigue category concern. Welds made during fabrication on members not yet under load (the typical shop-fab case) do not raise this concern. Welds made on existing structures under load (retrofits, repairs, attachment installation on operating equipment) require either temporary shoring to remove the load or a calculated check that the residual section can carry the load while a strip is at elevated temperature.
For repair work specifically, D1.1 Clause 10 governs welded modification or repair of existing structures, including the heat-effect-on-loaded-member analysis the Engineer must perform.
AASHTO Detail E' = D1.1 E* — The Most Penalized Category
AASHTO LRFD Bridge Design Specifications use the symbol E' (with prime/apostrophe) for the most-penalized fatigue detail category. AWS D1.1:2025 Table 4.5 uses E* (with asterisk) for the equivalent — constant C_f = 3.9 × 10⁸, threshold F_TH = 2.6 ksi [18 MPa]. The two notations refer to the same allowable stress range curve.
Where does E* show up? Per Table 4.5 Section 3:
- Line 3.6 — Base metal at ends of partial-length welded cover plates wider than the flange, with welds across the ends, flange thickness > 0.8 in [20 mm]: E*.
- Line 3.7 — Cover plates wider than the flange without welds across the ends: E* for thin flanges; not permitted for flange thickness > 0.8 in [20 mm].
- Line 4.1 — Longitudinal fillet welded end connections, plate thickness t > 0.5 in [12 mm]: E*.
- Line 3.3 — Base metal at ends of longitudinal welds terminating at weld access holes, R ≥ 3/8 in [10 mm] without grinding: E*.
The pattern: E* fires on details that combine geometric discontinuity, transverse weld termination, and tension-flange location. On a CWI Part C exam or a structural engineer’s detail review, recognizing the E* / E' configurations is the difference between a detail that lasts the design life and one that fails by fatigue cracking from the weld toe in service.
For a per-category walkthrough of all eight stress range categories (A through F including B* and E*), see the D1.1 Fatigue Stress Categories Explained reference page.
The 1/4-Inch Contouring Fillet at Reentrant Corners
One related provision sits in Section 4.18.3: “In transverse corner and T-joints subject to tension or tension due to bending, a single pass contouring fillet weld, not less than 1/4 in [6 mm] in size shall be added at reentrant corners.” This is a cyclically loaded rule (Part C governs) that smooths the geometric discontinuity at the weld toe in reentrant-corner CJP joints — the dashed contouring fillet shown on TC-U4a in Figure 5.1, for example. See the bevel-groove weld symbol page for the contouring-vs-reinforcing fillet distinction in detail.
“On non-bridge work, the rule of thumb ‘don’t weld across the flange’ isn’t a code prohibition — it’s the fatigue category penalty in Table 4.5 making the section uneconomic. On bridge work, D1.5 Clause 12 layers FCM treatment on top — that’s a different conversation. Conflating the two is what makes the rule sound mysterious.”
— Field observation, structural fabrication practice
Related Standards Guides
Frequently Asked Questions
Is it against AWS code to weld across an I-beam flange?
For non-bridge structural work governed by AWS D1.1, no — it is not prohibited. It is fatigue-categorized. Section 4.17 with Table 4.5 places transverse welds on tension-loaded plate elements at cruciform, T, and corner joints into Category C, E, or E* depending on the detail. The result is a lower allowable stress range than longitudinal welds, which forces a larger section to carry the same cyclic load. Under static loading the constraint disappears. For bridge work governed by AWS D1.5, fracture-critical members are subject to additional Clause 12 provisions. The Engineer of Record's stamped drawings and the WPS govern in either case.
Why is welding longitudinally on the flange OK but transversely is penalized?
Longitudinal welds parallel to the line of stress fall into Section 3 of D1.1 Table 4.5 — typically Category B or B' (continuous CJP or PJP groove welds joining built-up members). The crack initiation point is internal weld discontinuities, which research has shown produce a higher allowable stress range. Transverse welds across a tension flange fall into Section 5 — Category C for CJP T or corner joints with reinforcement not removed (line 5.4), Category C for transverse fillets adjacent to stiffeners on rolled beam flanges (line 5.8), and Category E or E* for cover plates wider than the flange (lines 3.6, 3.7). The crack initiates at the weld toe under cyclic tension, which is a much shorter fatigue life. Same physical weld, different geometry, different category.
What is AASHTO Detail E' and why does it match D1.1 E*?
AASHTO LRFD Bridge Design Specifications use the symbol E' (with prime/apostrophe) for the most-penalized fatigue detail category. AWS D1.1:2025 Table 4.5 uses E* (with asterisk) for the equivalent — a constant of 3.9 × 10⁸ and a threshold stress range of 2.6 ksi [18 MPa]. The two notations refer to the same allowable stress range curve. Both apply to details such as cover plates wider than the flange, cover plates without welds across the ends, and longitudinal fillet end connections on plates thicker than 0.5 in [12 mm]. AASHTO and AWS publish parallel documents — when a comment cites E' (AASHTO), the D1.1 equivalent is E*.
Does AWS D1.1 still recognize fracture-critical members?
No. AWS D1.1:2025 Section 4.15.4 states: "This code no longer recognizes a distinction between redundant and nonredundant members." The fracture-critical member (FCM) concept lives entirely in AASHTO/AWS D1.5 Clause 12 for bridge work. The D1.1 Commentary explains that the redundant/nonredundant distinction was based on consequence-of-failure judgment, not on actual fatigue performance differences. AASHTO continued to need the concept for bridges (where collapse risk is concrete), so it migrated to D1.5. If you are working from D1.1 alone, FCM is not in your code. If you are working from D1.5, all welds to tension zones of FCMs (with the bearing-sole-plate exception) are fracture-critical and trigger Clause 12 provisions.
What is the 1/4 inch contouring fillet rule for reentrant corners?
AWS D1.1:2025 Section 4.18.3 requires a single-pass contouring fillet weld of not less than 1/4 in [6 mm] in size at reentrant corners of transverse corner and T-joints subject to tension or tension due to bending. This is a cyclically loaded fatigue rule, not a static rule — Part C of Clause 4 governs cyclically loaded connections. A contouring fillet smooths the geometric discontinuity at the weld toe, moving the crack initiation point and improving the fatigue category for that detail. It applies on top of the prequalified joint detail (Figures 5.1 through 5.10) — for example, the dashed contouring fillet shown on TC-U4a in Figure 5.1. The Engineer specifies it on contract drawings; if not shown, the 1/4 in minimum still applies for any reentrant-corner joint in the cyclically loaded regime.
CWI Exam Tip: Table 4.5 (fatigue design) and Table 8.1 (visual inspection acceptance) test together on the CWI Part C exam — practitioners commonly confuse them. Table 4.5 governs allowable stress range BEFORE fabrication. Table 8.1 governs accept/reject AFTER fabrication. Two different decisions on the same physical weld, two different code layers. See CWI Exam Prep for the open-book navigation discipline.