Why an Anchor Fails and Which Failure You Get

Why this matters

When an anchored bracket comes down, the default repair is a bigger anchor. That is the right fix for exactly one of the failure modes below and the wrong fix for the rest, and the wrong ones come back. The failure leaves a readable record: the shape of what came out of the wall, the state of the hole, and what is still stuck to the anchor together name which element in the load path was weakest. Read that first and the repair picks itself.

Lead with this: a support that partly failed is still loaded

Almost every anchor failure you get called to is partial. One anchor of four let go and the other three are now carrying what four carried, which means they are closer to failure than they have ever been, and you are about to stand under them with a flashlight.

• Support or remove the load before inspecting anything. Independent support under the assembly, or lower it, before the first close look. Do not put a ladder under a suspended assembly with a failed anchor in it.
• If the assembly carries piping or duct, the contents are part of the hazard. Isolate, relieve pressure through a vent or drain, and confirm zero at a gauge downstream of the isolation, under 29 CFR 1910.147 for general-industry mechanical and stored energy.
• If anything electrical is attached or nearby, de-energize and lock out under 29 CFR 1910.333(b)(2) and prove dead with the live-dead-live sequence in NFPA 70E-2021, 120.5.
• Do not pull on a partially failed anchor to see how loose it is. That is a load test with your body in the load path.
• Re-drilling and hole cleaning are a silica exposure. Construction work falls under 29 CFR 1926.1153, whose Table 1 pairs handheld drilling in concrete or masonry with water delivery or a HEPA dust collector; general industry is 29 CFR 1910.1053, with respirators under a written program per 29 CFR 1910.134.

The failure names the weakest element

An anchored connection is a chain: the fixture, the anchor's steel, the interface between the anchor and the hole, and the base material around the hole. Whichever is weakest is what breaks, and the failure mode is a direct statement of which one it was.

That has a consequence people find uncomfortable. A steel failure of the anchor is the good outcome, because it means the concrete and the geometry were adequate and the only thing that needed changing was the anchor's own section. Every other mode says something about the installation or the base material, and upsizing the anchor does not touch any of them.

The field key

Two pairs that look alike, and the cut between them

Pullout versus breakout. Both end with the anchor out of the wall. The cut is whether concrete came with it. Breakout brings a cone, so there is a crater and the anchor is still buried in the piece that came out. Pullout leaves a hole you could re-use, which is exactly why it gets misread as breakout by anyone who only looks at the anchor. Pullout says the anchor never developed its mechanism: an expansion element that never set, a hole drilled with the wrong bit so the diameter is oversize, a screw anchor whose cut thread stripped, or an adhesive that never bonded.

Which side the adhesive failed on. Both look like a rod pulled out of a hole. Look at where the cured adhesive stayed. Adhesive left keyed into the hole with a smooth bore where the rod was means the bond to the rod failed, and the causes are a rod that was oiled, plated or too smooth, an annulus too large because the hole was oversize, or a rod pushed straight in without the rotation that wets it. Adhesive that came out on the rod, leaving a dusty hole wall, means the bond to the concrete failed, and that is a hole-cleaning finding almost every time.

The repairs that do not follow from any of these modes

Three field habits that survive because they feel like doing something:

• Upsizing the anchor after a breakout. The cone is a property of embedment and geometry. A wider anchor at the same depth in the same spot pulls a cone of nearly the same size.
• Adding adhesive around a mechanical anchor that pulled out. The mechanical anchor's hole was drilled for a mechanical fit, and adhesive injected around a set expansion element is not a qualified assembly. It has no published value because nobody tested it.
• Re-drilling in the same hole, or immediately beside it. The concrete around a failed anchor is damaged out to the extent of the cone, so a new anchor placed inside that zone is anchoring into cracked and disturbed material. Move outside the damaged region, or repair the concrete under a specification before re-anchoring.

Worked example: reading one failed pipe support

A pipe support drops at one end. Two adhesive anchors held that end. The load is supported and the line is isolated and drained before anyone gets close.

Anchor A brought a cone of concrete with it. Measured across the crater at the surface, the cone is about 6 in wide, so it reached the surface at roughly a 3 in radius from the anchor. A tension breakout cone projects out at about one and a half times the effective embedment, so a 3 in radius implies an effective embedment of about 2 in. The support's specification called for 4 in.

That single measurement is the finding. Breakout capacity scales with embedment to the 1.5 power, so an anchor set at 2 in where 4 in was specified develops (2/4) to the 1.5 power, which is 0.35: about 35 percent of the intended breakout capacity. At the specified 4 in the cone would have reached the surface at a 6 in radius and been about 12 in across, which is twice the crater that is actually there. The crater is the evidence, and it is not ambiguous.

Anchor B came out of a clean hole with cured adhesive still keyed into the hole wall and a smooth bore where the rod had been. No concrete came with it. That is bond failure at the rod, a different mode and a different weakest element: the rod's surface or the annulus around it, not the concrete and not the embedment.

What the pair tells you that neither tells you alone. Two anchors installed by the same person on the same day failed by two different mechanisms. If the design were wrong, both would have failed the same way. Two different modes on one bracket says the installation was uncontrolled: no depth stop on the drill, so the holes came out at whatever depth the operator felt, and no consistent rod preparation or hole-diameter check.

So the repair is not a bigger anchor. It is a controlled reinstall: a depth stop set to the specified embedment and verified with a depth gauge before injection, the manufacturer's hole-cleaning cycle run to its stated count with a HEPA vacuum, rod surface prepared to the manufacturer's requirement, hole diameter checked against the specified bit, and both new positions moved clear of the damaged concrete around the old crater. The anchor size on the drawing was never the problem.

What happens if only anchor A is read. The crater is obvious and the clean hole is not, so the natural conclusion is "breakout, we need more embedment," and the repair drills deeper at both positions with the same uncontrolled process. Anchor A's position improves. Anchor B's position gets a deeper hole with the same untreated rod and the same annulus, and it fails the same way it failed the first time, on a longer timeline because there is more bond area. The shop concludes the adhesive is bad.

Checking you got this right

• Photograph the crater with a scale in frame before anything is cleaned up. The cone's surface diameter is the measurement that gives you embedment, and it is gone the moment somebody sweeps.
• Keep the failed anchor. What is stuck to it is half the diagnosis, and it goes in the bin within the hour if nobody says otherwise.
• Check the other anchors on the same assembly, including the ones that held. If the installation was uncontrolled, the survivors are the same installation, and they are now carrying more than they were designed for.
• On the reinstall, verify the embedment before the fixture goes on, with a depth gauge in the hole, and write the measured value on the record. This is the one number that would have made the original failure impossible to argue about.
• Where a proof test is specified, keep the load path clear: nobody in line with the loaded rod, no hand on the loaded element, and the reaction frame bearing outside the anchor's own failure cone so you are not testing the frame's cone instead.

References

• ACI 318 Chapter 17, Anchoring to Concrete, in the edition incorporated by the building code your authority having jurisdiction has adopted, for the failure modes a design must check and the geometry that governs them
• Anchor manufacturer installation instructions and evaluation reports for hole diameter, cleaning procedure, rod preparation and embedment, which own the specific values a reinstall must meet
• 29 CFR 1910.147 for mechanical and stored-energy isolation; 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5, for electrical isolation and proving dead
• 29 CFR 1926.1153 and 29 CFR 1910.1053 for respirable crystalline silica; 29 CFR 1910.134 for respiratory protection programs
• See related: How an Anchor Transfers Load Into Concrete; The Anchor Types and What Each Needs From the Base Material