As-designed, as-detailed, as-built: where connections diverge

Connections fail more often than members do, and the reason is largely organisational rather than mechanical. A member is designed, specified and fabricated as a single item. A connection is designed by one party, detailed by another, fabricated by a third and erected by a fourth, with revisions and field adjustments at each handover. Every transfer is an opportunity for the joint that gets built to differ from the joint that was analysed — and in a failure investigation, comparing those two is frequently more productive than any amount of examination of the broken steel.

Three versions of the same connection

The design version exists in the engineer's calculations and the contract drawings, often as a required capacity rather than a complete detail. The detailed version appears in shop drawings prepared by the fabricator or a detailer, converting that requirement into specific plates, welds and bolts. The built version is what erection actually produced, including any field modification. These three should agree and frequently do not, and establishing all three is the first analytical task.

Delegated design and where it goes wrong

Connection design is commonly delegated to the fabricator, with the engineer of record specifying forces and reviewing submittals. That arrangement works when the forces given are complete. It fails when they are not — when a required force is omitted, when the engineer specifies only a shear reaction for a joint that also carries axial load or moment, or when a stiffness assumption in the frame analysis is not communicated. The connection is then correctly designed for the wrong demand.

The shop drawing review record

Submittal review generates a documented exchange: what was submitted, what was marked up, what was resubmitted and what was finally approved. That record establishes what each party knew and when. A connection built exactly as an approved shop drawing shows sits differently from one built contrary to it, and a detail that was questioned in review and then approved unchanged sits differently again. This file is the single most useful documentary source in most connection disputes.

Requests for information mark the known problems

An RFI is a record that somebody encountered something they could not resolve, and its response is a record of what they were told to do. RFIs concerning fit-up, interference, member size changes or connection geometry at or near the failure location are directly relevant, and the absence of an RFI where a field change plainly occurred is itself informative — it indicates the change was made without engineering review.

Field modifications and the ones nobody recorded

Fit-up problems get resolved on site under time pressure, and the resolutions are not always documented. Holes are flame-cut or reamed oversize, bolts are omitted where they cannot be reached, plates are trimmed to clear an obstruction, stiffeners are left out, welds are added or substituted for bolts. Each changes capacity, sometimes substantially. Physical evidence of these modifications is usually visible on the recovered connection, and comparing it against the approved detail is what identifies them.

What the recovered hardware shows

The joint itself carries direct evidence: bolt grade markings against what was specified, hole sizes and condition against the detail, weld sizes and profiles against the drawing, and whether the specified number of fasteners is present. Flame-cut holes, mismatched bolt grades within a single joint, undersized welds and missing elements are all findings that need no analysis to state, and they are commonly the substance of the case.

Erection-stage conditions are a distinct question

A connection may be complete in its final condition and inadequate during erection, when only some bolts are installed, welds are not yet made, and the bracing that gives the frame stability is absent. Collapses during erection frequently trace to this temporary condition rather than to the permanent design, and analysing the joint only in its finished state misses it entirely. The erection sequence and the temporary bracing plan are the relevant documents.

Inspection records establish what was verified

Shop and field inspection reports, weld inspection records and their nondestructive testing results, bolt installation verification, and any nonconformance reports document what was checked and what was found. A defect in a joint that was inspected and accepted raises a question about the inspection; the same defect in a joint never inspected raises a different one. Both are relevant, and the distinction usually matters to the outcome.

What to preserve

The connection intact, cut out well away from the joint itself rather than through it, with all fasteners, plates and fragments retained and their positions recorded. Adjacent unfailed connections of the same type, which establish what the population actually looked like. And the complete documentary set: contract drawings, shop drawings with review markings, RFIs and responses, change orders, erection procedures and all inspection records.

This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.