August 25, 2026 · 15 min read

14 CFR 25.571 / 25.1529 App. H14 CFR 26 Subpart E / 21.137AD 2026-15-11 / RB 737-53A1408

When the Inspection Interval Branched on the Repair History: The 737 MAX Bear Strap AD (2026-15-11) Through a 14 CFR 25.571 Damage-Tolerance and Part 26 Repair-Assessment Lens

The most interesting exchange in AD 2026-15-11 is not in the rule text. It is in the comment disposition, where United Airlines asks whether a blend-out repair performed inside the structural repair manual's allowable damage limits counts as a "repair found," and the FAA answers yes — and adds that depending on findings, "it may be necessary to reduce the inspection interval." That is a regulator saying, out loud, that the published interval was computed against a part that no longer exists on some airframes.

This is a story about an inspection program whose first decision gate is the maintenance record rather than the structure, and about a damage-tolerance evaluation whose initial-flaw assumption the assembly line falsified three airplane generations ago. Nobody has found a crack on a MAX. That is the point.

1. The public record

On August 6, 2026, the FAA published final rule AD 2026-15-11 (Amendment 39-23423, Docket FAA-2025-3985, Project Identifier AD-2025-00493-T, 91 FR 50698), effective September 10, 2026. It applies to certain Boeing Model 737-8, 737-9, and 737-8200 airplanes and covers 471 airplanes of U.S. registry. The unsafe condition, verbatim from paragraph (e):

"This AD was prompted by reports of cracks in the bear strap at the forward upper corner of the forward galley door cutout. The FAA is issuing this AD to address cracks in the fuselage skin and bear strap, which may lead to the inability of the principal structural element to sustain limit loads and adversely affect the structural integrity of the airplane." (91 FR 50698)

A bear strap is a doubler that wraps a fuselage cutout and carries the hoop and shear load the opening interrupts. Every pressurization cycle drives the skin around the door corner, and the corner radius is where the stress concentration lives. This one sits at the forward upper corner of the forward galley door.

The chain runs across three product generations:

| Date | Event | Applicability | |---|---|---| | 2014 | FAA mandates bear strap crack inspections | 737 Classic | | 2019 | Boeing issues inspection instructions to operators | 737 Next Generation | | 2021 | FAA mandates the NG inspections | 737 Next Generation | | Dec 20, 2024 | Boeing Alert Requirements Bulletin 737-53A1408 RB issued | 737 MAX | | Nov 25, 2025 | FAA NPRM published (90 FR 53245) | 737 MAX | | Aug 6, 2026 | Final rule AD 2026-15-11 published (91 FR 50698) | 737 MAX | | Sep 10, 2026 | AD effective; IBR of RB 737-53A1408 approved | 737 MAX |

The cracks were found on 737NG airframes. No cracks have been confirmed on any 737 MAX. The FAA proceeded anyway, on the basis that the MAX has "a similar design and build process" to the older models, "making those airplanes susceptible to the same crack condition." (AP, August 10, 2026) Boeing's own statement is that it identified and reported the issue and "has been working with operators on it over the past six years."

The required actions, from the incorporated bulletin: an external general visual inspection of the fuselage skin for any repair, then applicable on-condition actions — repetitive detailed inspection of the fuselage skin for cracking; repetitive external surface high frequency eddy current (HFEC) inspection at certain fastener locations and along the edge of the door corner radius; repetitive external subsurface low frequency eddy current (LFEC) inspection of the bear strap at certain fastener locations; and, for anything outside the envelope, obtaining instructions from Boeing.

Costs, verbatim: the general visual inspection is 1 work-hour × $85 = $85 per airplane, $40,035 fleet-wide. The detailed and eddy current inspections run up to 4 work-hours × $85 = up to $340 per inspection cycle, up to $160,140 per cycle fleet-wide. The FAA states it "has received no definitive data on which to base the cost estimates for the on-condition repairs."

Two comments matter.

The Foundation for Aviation Safety asked what is being done about root cause, and why the inspection is not accomplished sooner if a detected defect is serious enough to ground an airplane. The FAA's answer contains the sentence this whole post turns on:

"In addition, Boeing is introducing changes to the manufacturing process that address the root cause of the unsafe condition on in-production airplanes."

United Airlines noted that Table 1, Condition 1 of the bulletin directs operators to contact Boeing for alternative inspection instructions "for any repair found," but that the alert service bulletin gives no guidance for blend-out repairs accomplished within the allowable damage limits of 737-8 and 737-9 SRM 53-00-01. Should a blend-out inside ADL count as a repair, or as an unrepaired area? The FAA:

"Condition 1 applies to any repair, which includes blend out repairs accomplished per the SRM within the allowable damage limits. Such blend out repairs need to be evaluated to determine whether the repetitive inspection interval specified in the service information provides an acceptable level of safety for those repairs. Depending on findings, it may be necessary to reduce the inspection interval."

2. The standards lens

This is metallic primary structure on a transport category airplane, so the governing code is 14 CFR Part 25, and specifically the damage-tolerance regime. There are no software levels here, no DAL, no ASIL. There are four clauses, and each of them names an artifact.

14 CFR 25.571(a) and (b) — Damage-tolerance and fatigue evaluation of structure. The rule requires that catastrophic failure due to fatigue, corrosion, manufacturing defects, or accidental damage be avoided throughout the operational life of the airplane. For each principal structural element, the damage-tolerance evaluation must determine the extent of damage the structure can sustain at the required residual strength level, and must establish inspections that will detect that damage before it grows to critical size. The AD's own words — "inability of the principal structural element to sustain limit loads" — are the residual-strength criterion of §25.571(b) restated as an unsafe condition. The rule was satisfied at type certification. What the rule cannot do is protect an assumption.

The assumption at issue is the initial flaw size. Guidance in AC 25-571-1D frames the damage-tolerance evaluation around assumed initial damage consistent with the manufacturing and inspection quality of the part — a fastener-hole flaw of assumed depth, a surface discontinuity, a rogue flaw where the process warrants one. That assumption is not a physical measurement. It is a claim about what the factory produces. Change the factory and you have changed an input to a certification analysis without touching a drawing.

14 CFR 21.137 — Quality system. Production approval holders must establish and maintain a quality system with procedures covering, among other things, manufacturing process controls and the identification and correction of nonconformances. When the FAA writes that Boeing "is introducing changes to the manufacturing process that address the root cause," it is reporting a §21.137 corrective action on a process that had been delivering airplanes for three generations. The design was fine. The process delivered a part whose crack-initiation behavior did not match the population the §25.571 analysis assumed — and it did so consistently enough that the FAA is willing to mandate inspections on a variant with zero findings.

14 CFR 25.1529 and Appendix H (H25.4) — Instructions for Continued Airworthiness / Airworthiness Limitations Section. The inspections that fall out of the §25.571 evaluation live in the ALS as mandatory structural inspection items, with thresholds and repeat intervals in flight cycles. Those numbers are outputs of a crack-growth calculation. Their validity is exactly as good as the initial-flaw and local-stress inputs that produced them.

14 CFR Part 26, Subpart E — Damage tolerance data for repairs and alterations to the fuselage pressure boundary. This is the clause the United comment lands on and it is the most underrated rule in the aging-airplane package. Subpart E requires design approval holders to develop repair evaluation guidelines and requires operators to incorporate them into their maintenance programs, so that repairs to the pressure boundary get their own damage-tolerance-based inspections rather than inheriting the baseline structure's. A blend-out inside SRM allowable damage limits is the boundary case: it is by definition not a repair in the operator's paperwork sense, because ADL exists precisely so that minor damage can be dressed out and returned to service without engineering disposition. And yet it removes material from a fatigue-critical radius. The FAA just ruled that for the purposes of this AD, it counts.

Put plainly: the AD's first gate is a records question that the records were never designed to answer. "External general visual inspection of the fuselage skin for any repair" assumes repairs are visible and traceable. Blend-outs within ADL are often neither.

3. A worked snippet — how a legal blend-out eats the interval

Below is a damage-tolerance interval derivation for a door-corner bear strap in three configurations. The numbers are illustrative — Boeing's actual crack-growth curves, stress spectra, and the compliance times in RB 737-53A1408 are proprietary and not in the public docket. The structure of the calculation is not illustrative; it is how every ALS structural interval is produced.

Assume 2024-T3 clad skin over a 7075 bear strap, a fatigue-critical fastener row along the door corner radius, one full pressurization cycle per flight cycle, and a repeat interval set at half the calculated crack-growth life so the crack gets at least two detection opportunities before reaching critical length.

| # | Configuration | Local peak stress, 1P (ksi) | Assumed initial flaw a₍i₎ (in) | a₍crit₎ at limit-load residual strength (in) | Crack growth life a₍i₎ → a₍crit₎ (FC) | Repeat interval = N/2 (FC) | |---|---|---|---|---|---|---| | 1 | Design basis — as-drawn, as-assumed process | 18.0 | 0.050 | 2.00 | 62,000 | 31,000 | | 2 | As-built — process-induced surface discontinuity at the radius | 18.0 | 0.125 | 2.00 | 34,000 | 17,000 | | 3 | Row 2 + SRM blend-out, 10% thickness removed inside ADL | 20.0 | 0.125 | 1.70 | 19,000 | 9,500 |

Row 1 is what the certification analysis computed. Row 2 is what the factory shipped. Row 3 is what a legally compliant maintenance action produced on top of it — and row 3 is flying against the row 1 interval, because ADL blend-outs do not trigger a repair-specific damage-tolerance evaluation unless somebody decides they do.

The interval collapse from 31,000 to 9,500 flight cycles is a factor of 3.3. That is the entire safety margin of the inspection program, spent by two events neither of which produced a paperwork trigger.

Now the AD's decision logic, as an operator actually executes it:

                RB 737-53A1408 RB — entry gate
                            |
        External General Visual Inspection of fuselage skin
                            |
              +-------------+-------------+
              |                           |
      "No repair found"            "Repair found"  <-- Condition 1
              |                           |
    Baseline repetitive DET /      Contact Boeing for
    HFEC / LFEC per published      ALTERNATIVE inspection
    thresholds and intervals       instructions; interval
              |                    may be REDUCED
              |                           |
              |                           +--> repair-specific
              |                                damage tolerance
              |                                evaluation (Part 26
              |                                Subpart E territory)
              |
              +--> [SILENT PATH]
                   SRM 53-00-01 blend-out inside ADL:
                   material removed, K(t) raised,
                   a(crit) reduced — but historically
                   dispositioned as "no repair"
                   ==> baseline interval retained
                   ==> FAA now says this is WRONG

The silent path is the finding. The FAA did not change the AD in response to United's comment, because the AD text already said "any repair." It clarified that the industry's working definition of "repair" was narrower than the rule's.

4. Derived requirements (excerpt)

Five requirements a structures organization should be able to point at after reading this file. IDs are stable; thresholds are the kind you commit to, not the kind you negotiate.

| ID | Requirement | Verification | Traces to | |---|---|---|---| | BS-DT-001 | Every damage-tolerance evaluation of a principal structural element shall record its assumed initial flaw size a₍i₎, the manufacturing process control that substantiates it, and the process characteristic (surface finish, residual stress state, fastener hole preparation method) on which that substantiation depends. The record shall be a controlled configuration item, not an appendix to a stress report. | Design review; ALS traceability audit sampling ≥ 20% of PSEs | 14 CFR 25.571(a), (b); AC 25-571-1D | | BS-DT-002 | Any change to a manufacturing process that affects a characteristic named in a BS-DT-001 record shall trigger re-run of the affected crack-growth analysis and re-issue of the corresponding ALS threshold and repeat interval, prior to first delivery of the changed configuration. Process changes shall not be dispositioned as "no effect on type design" without that re-run. | Change board record; §21.137 quality system audit finding closure | 14 CFR 21.137; 21.93; 25.1529 App. H25.4 | | BS-DT-003 | Structural repair manual allowable damage limits for any fatigue-critical location shall state the maximum material removal that preserves the published inspection interval, expressed as a percentage of nominal thickness. Where blend-out exceeds that value, the SRM shall require a repair-specific damage-tolerance evaluation before return to service. Target: no ADL entry at a PSE radius without an associated interval-validity statement. | SRM review against PSE list; 100% coverage of door and window cutout radii | 14 CFR part 26 subpart E; 25.571(b) | | BS-DT-004 | The maintenance record system shall capture blend-out repairs performed within ADL at PSE locations, with location, depth, and post-blend thickness, and shall make that record retrievable by structural location for the life of the airframe. Absence of a record shall not be interpreted as absence of material removal. | Records system data model review; retrieval test on ≥ 50 sampled airframes | 14 CFR part 26 subpart E; 25.1529 | | BS-DT-005 | Where a derivative model inherits a structural detail and its build process from a predecessor, the derivative's ALS intervals shall be re-validated against the predecessor's accumulated in-service findings at intervals not exceeding 24 months, and shall not rely solely on the predecessor's original certification analysis. Escalation trigger: any predecessor finding at the same feature. | Continued operational safety review; fleet findings correlation report | 14 CFR 21.101; 21.3; 25.571(b) |

BS-DT-005 is the one this AD actually vindicates. The FAA mandated MAX inspections on the strength of NG findings, before a single MAX crack. That is the correct call and it is also an admission that the derivative's intervals had been riding on inherited analysis.

5. What the headline really tells us

The headline version is FAA orders crack inspections on 471 Boeing 737 MAX jets. It reads like a Boeing story, and every outlet that carried it reached for the 2018-19 crashes and the 2024 door plug within two paragraphs. Those are real, and they are not what this file is about.

This file is about a number. Somewhere in the 737's damage-tolerance substantiation there is an assumed initial flaw at a door corner radius, and behind it an assumption about what the assembly process leaves in the metal. That number produced an inspection threshold, and the inspection threshold produced a line in the Airworthiness Limitations Section, and the line in the ALS produced a task card. Three generations of airplane were maintained against that chain. The chain was correct in the sense that every link was properly derived from the one above it. It was wrong in the sense that its top link was a claim about a factory, and the factory changed — or never matched the claim in the first place, which is why Boeing is only now "introducing changes to the manufacturing process that address the root cause."

The missing artifact is not an inspection. Boeing wrote the inspection in December 2024 and the FAA has now mandated it. The missing artifact is a controlled record binding each damage-tolerance initial-flaw assumption to the specific process control that substantiates it, so that a process change anywhere in the plant raises a flag in the structures group instead of quietly shortening a fleet's crack-growth life. And the second missing artifact is a repair record that captures material you removed legally — because the AD's entire gating logic assumes that "no repair found" means the structure is as-drawn, and the SRM has spent decades authorizing engineers to make that statement false without writing anything down.

The generalizable version has nothing to do with airplanes. Take your safety-critical part and answer two questions. First: what does your reliability or life calculation assume about how the part is made, and is that assumption written down anywhere a manufacturing engineer would encounter it before changing a process? Second: what maintenance actions are your people allowed to perform without generating a record, and would any of them invalidate that calculation?

If the answers are "it's in a stress report from the original program" and "quite a few, that's what the allowable-damage section is for," you are running the bear strap file. Twelve years, three product generations, and 471 airplanes into it, and still no crack — which is either a very good inspection program or a very lucky one, and the difference between those two is exactly the artifact nobody wrote.

Sources

— Jherrod Thomas, The Lion of Functional Safety™