How an Anchor Transfers Load Into Concrete

Why this matters

Anchor catalogs publish a capacity per anchor, and that number is what most people size against. It is the wrong number to size against on its own, because it was established remote from any edge, in a stated concrete condition, at a stated embedment. Move the same anchor 3 in from a slab edge and a large share of that capacity is gone before the wrench touches it. The failures that follow do not look like anchor failures. A cone of concrete comes out with the anchor still bolted to the bracket, perfectly intact, and everyone stands there wondering how a good anchor pulled out.

Before the bit turns

Drilling concrete creates two hazards that the job did not have before, and one of them is invisible.

• Respirable crystalline silica. Drilling concrete puts silica in the air, and it is a lung carcinogen with no useful contact-route control. A glove and safety glasses do nothing about it. Construction work is covered by 29 CFR 1926.1153, whose Table 1 pairs handheld drilling in concrete with either water delivery to the bit or a shroud and dust collector with HEPA filtration; general industry falls under 29 CFR 1910.1053. Where a respirator is part of the control, it belongs to a written program under 29 CFR 1910.134, not to a box of dust masks in the van.
• What is inside the concrete. Scan before drilling. Cutting reinforcement weakens the member you are anchoring to. Cutting a tendon in a post-tensioned slab releases a large stored energy in a member that spans the whole bay, and it is not recoverable in the field: do not drill a post-tensioned slab without locating tendons and getting the structural engineer's sign-off. Embedded conduit is an electrical hazard, so where the scan is ambiguous, isolate the affected circuits under 29 CFR 1910.333(b)(2) and prove dead with the live-dead-live sequence in NFPA 70E-2021, 120.5.
• Overhead work adds falling spoil and a bit that binds while your arms are extended. Face protection per 29 CFR 1910.133, and a position where a bound bit twists the tool away from you rather than into your chest.

An anchor does not grip concrete, it loads a volume of it

The mental model that causes the failures above is that an anchor holds by friction or by biting into the hole, so a bigger, harder, better anchor holds more. What actually happens is that the anchor delivers your load into the concrete at some depth, and the concrete carries it out to the surface through a roughly conical volume of material working in tension.

Concrete has very little tensile strength. That cone is the weak element in almost every anchor load path, and its size is what you are buying.

Two consequences follow immediately, and both are counterintuitive until you hold the cone in your head:

• Steel grade is rarely the lever. A stronger anchor body only helps if steel is the governing failure mode, which it usually is not in unreinforced concrete near the surface.
• Geometry is the lever. How deep the load is delivered, and how much intact concrete surrounds the cone, sets the number.

The two numbers that set the cone

Effective embedment is the depth at which load actually transfers into the concrete, not the length of the hole and not the length of the anchor. For a wedge anchor it is measured to the expansion element, not to the tip. Get this wrong and every number downstream is wrong.

In the concrete capacity method that ACI 318 Chapter 17 uses, breakout strength rises with embedment to the 1.5 power, and the projected failure area for a single anchor remote from edges is a square of side three times the embedment, or nine times the embedment squared. That geometry, a projection of one and a half embedments in every direction, is the whole reason edges and neighbours matter so much.

Edge distance and spacing truncate that projection. An edge closer than one and a half embedments cuts the square off, and there is a second, separate reduction on top of the area loss to account for the way the cone behaves against a free surface. Two anchors closer together than three embedments share one cone rather than each getting a full one.

ACI 318 binds through the building code your authority having jurisdiction has adopted, in the edition that code incorporates, and it reaches you through the permit, the engineer of record, and any special inspection the permit requires. It is not something you apply on your own authority to a job someone else designed.

Concrete strength is the weak lever

Breakout capacity rises with the square root of concrete compressive strength. That means a mix at double the strength buys about 41 percent more breakout, not double. Compare that to embedment, which at the 1.5 power gives about 84 percent more for a 50 percent deeper anchor before any edge effects.

The practical version: you cannot specify your way out of a bad geometry with better concrete, and testing the slab to find out it is stronger than assumed will not rescue an anchor that is too shallow or too close to an edge.

Worked example: one bracket, one slab edge

A bracket needs anchoring to a concrete slab. The chosen anchor's published capacity assumes an effective embedment of 4 in, remote from edges. The only place the bracket can land puts the nearest anchor 3 in from a free edge of the slab.

Step one, the full projection. At an effective embedment of 4 in, the projection extends 1.5 x 4 in, which is 6 in in each direction. The full projected area is a 12 in square, so 144 square inches. That is the area the catalog number was established against.

Step two, truncate at the edge. The edge is 3 in away, less than the 6 in the projection wants, so on that side the projection stops at 3 in. The available area is 3 in plus 6 in wide by 6 in plus 6 in long, so 9 in by 12 in, which is 108 square inches. Against 144 square inches that is 75 percent.

Step three, the second edge reduction. The area ratio is not the whole loss. There is a separate edge factor because a cone against a free surface does not behave like a truncated version of the full cone. Under the ACI 318 form, at an edge distance of 3 in against a 6 in projection, that factor works out to 0.85.

Step four, combine. 0.75 multiplied by 0.85 is 0.64. The anchor at that position develops roughly 64 percent of its catalog breakout number, and it does so with no defect, no bad installation and no weak concrete. The geometry took 36 percent before anyone did anything wrong.

What deepening actually bought

The obvious fix is a deeper anchor, and it does work, but not by as much as the 1.5 power suggests once the edge is in the picture. Run the same three steps at an effective embedment of 6 in:

• Raw breakout scales as (6/4) to the 1.5 power, which is 1.84, so 84 percent more before geometry.
• The projection now wants 1.5 x 6 in, which is 9 in in each direction. Full area is an 18 in square, 324 square inches. Truncated at the 3 in edge it is 12 in by 18 in, 216 square inches, a ratio of 0.67.
• The edge factor at 3 in against a 9 in projection works out to 0.80.
• Combined geometry factor: 0.67 x 0.80, which is 0.53.

Net against the original: 1.84 multiplied by 0.53 divided by the original 0.64 gives 1.53. Deepening from 4 in to 6 in at that location gained about 53 percent, not 84 percent, because the deeper anchor's cone wants more room and the edge is still 3 in away. The edge takes a bigger bite the deeper you go.

One check the deeper anchor does not fail here, and it is worth running anyway: for headed anchors, deep embedment close to an edge can trigger side-face blowout, where the concrete fails sideways at depth rather than coning to the surface. The screening condition is embedment greater than 2.5 times the edge distance. At a 3 in edge distance that threshold is 7.5 in, and 6 in is below it, so this joint is not in that regime. Push the same anchor to 8 in at that edge and it would be.

What the arithmetic tells you to do. The gain from deepening is real but expensive per inch of drilling, and it does not touch the underlying problem. Moving the bracket so the nearest anchor sits at least 6 in from the edge, at the original 4 in embedment, recovers the full 144 square inches and the full catalog number, which is a 56 percent gain over the as-drawn position for the cost of relocating a bracket. Geometry is cheaper than depth every time it is available.

The failure mode if nobody runs this. The bracket goes on at 3 in from the edge with anchors sized against the catalog number, the installation is clean, the torque is correct, and the joint holds. It holds until the first load case that approaches the design value, at which point a wedge of concrete lifts out of the slab edge with the anchor still torqued into it. The post-mortem finds a good anchor, good concrete and a correct installation, so the shop concludes it was a bad batch of anchors and buys a different brand for the repair, at the same 3 in from the edge.

The condition the number was established under

Two conditions come attached to every published anchor capacity and both get dropped in the field:

Cracked or uncracked concrete. Concrete in a tension zone is assumed cracked, and cracked concrete carries substantially less anchor capacity than uncracked. An anchor must be qualified for cracked concrete to be used there. Under ACI 318 the design assumption is cracked unless analysis shows the concrete will remain uncracked at service loads, which is not a call a tech makes on site.

Sustained tension versus short-term load. Adhesive anchors in particular behave differently under a load that never comes off, because the adhesive creeps. ACI 318 Chapter 17 treats adhesive anchors installed horizontally or upwardly inclined to resist sustained tension as a special case requiring a certified installer and continuous inspection. If you are hanging something permanently overhead, that is exactly the case being described.

Checking you got this right

• Measure the effective embedment, not the hole depth, and write it on the job record with the edge distance and spacing you actually achieved. Those three numbers are what any later review needs and none of them are visible once the bracket is on.
• Clean the hole to the manufacturer's instruction and count the strokes. For adhesive anchors, hole cleaning is not a finishing touch, it is a load-path element: dust at the bond line is the interface the anchor is relying on. If you use compressed air, 29 CFR 1910.242(b) limits it to under 30 psi for cleaning with effective chip guarding and eye protection, and a HEPA vacuum is a better answer because it captures the silica rather than launching it.
• If you proof test, keep the load path clear of people. A proof-test rig stores energy in a loaded rod. Nobody stands in line with the rod, nobody puts a hand on the loaded element, and the reaction frame bears on sound concrete outside the anchor's own cone or you are testing the frame's cone instead.
• When the geometry does not close, hand it up. An anchor position that cannot make its edge distance is a design question, not a field improvisation. The engineer of record can add reinforcement, change the anchor family, or move the connection. A tech cannot make a cone out of concrete that is not there.

References

• ACI 318 Chapter 17, Anchoring to Concrete, in the edition incorporated by the building code your authority having jurisdiction has adopted, which binds the design and reaches the field through the permit and the engineer of record
• 29 CFR 1926.1153 (construction, including its Table 1 controls for handheld concrete drilling) and 29 CFR 1910.1053 (general industry) for respirable crystalline silica; 29 CFR 1910.134 for respirator programs
• 29 CFR 1910.242(b) for compressed-air cleaning limits; 29 CFR 1910.133 for eye and face protection; 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5, where embedded conduit may be present
• Anchor manufacturer evaluation reports for capacity, embedment, hole-cleaning procedure and cracked-concrete qualification, which own every specific value
• See related: Why an Anchor Fails and Which Failure You Get; The Anchor Types and What Each Needs From the Base Material