There is no single capacity — there is a governing mode
A fastener can fail in several distinct ways and they have nothing to do with each other. The STEEL can yield or fracture. The base material can break out as a cone of concrete or masonry. The fastener can PULL OUT through the material while the material stays intact. A shear load can lever a shallow anchor out backwards — pryout — or split the edge it sits near.
Each mode has its own expression, its own sensitivity and its own resistance factor, and the design capacity is the SMALLEST of them. A page reporting a breakout capacity is not reporting the anchor's strength; it is reporting one candidate.
Which mode governs is therefore the first question, and it is not a detail. Steel failure is comparatively ductile — the anchor stretches and gives warning. Concrete breakout is brittle and sudden. Codes explicitly prefer designs where the steel governs, and achieving that is a geometry decision rather than a product decision.
- N_sa
- steel capacity of the anchor itself — ductile, and the one to aim for
- N_cb
- concrete breakout: a cone of base material lifting out
- N_pn
- pullout: the anchor sliding out with the material intact
- N_sb
- side-face blowout, where an anchor sits close to a free edge
Breakout grows with embedment to the power of one and a half
The concrete breakout model treats the failure as a cone spreading from the anchor's embedded end at roughly thirty-five degrees to the surface. The cone's projected AREA therefore grows with the square of the embedment, while the tensile stress it can carry falls as the crack lengthens — and the two combine to a capacity that goes with embedment to the power of one and a half.
That exponent is worth internalising. Doubling the embedment does not double the breakout capacity: it multiplies it by about **2.8**. Going half as deep again — fifty per cent more embedment — buys about eighty per cent more capacity. Depth is the cheapest variable in anchor design and it is usually the one constrained by the thickness of the thing being drilled.
It also means the mode switches as depth increases. A shallow anchor is governed by breakout; drive the same anchor deeper and breakout capacity overtakes the steel's, at which point the steel governs and the connection becomes ductile. Designing deliberately past that crossover is what the codes mean by a ductile anchorage.
- h_ef
- effective embedment depth — the anchor's working depth, not its overall length
- f_c
- concrete compressive strength; note it enters as a SQUARE ROOT, so doubling it buys about 41%
Groups overlap, and edges cut the cone in half
Two anchors close together do not carry twice one anchor's load, because their breakout cones OVERLAP and the shared concrete cannot fail twice. The standard treatment computes the projected area of the combined group against the area a single unconstrained anchor would have, and reduces accordingly. Interaction begins once spacing falls below about three times the embedment.
A free edge does the same thing more brutally: it truncates the cone. An anchor at an edge distance less than about one and a half times its embedment has part of its cone missing before it starts, and the capacity falls in proportion to the area lost. Two edges — an anchor near a corner — remove two pieces.
This is why an anchor layout cannot be optimised by geometry alone. Pulling bolts inboard to gain edge distance pushes them together and triggers group interaction; spreading them for group reasons pushes them toward the edges. The two constraints work against each other and the plate usually has to grow, which is the conclusion the baseplate pages reach from the other direction.
Where it fails: cracked concrete, and the hole nobody cleaned
Every published anchor capacity is qualified by the state of the concrete it sits in. Concrete in a tension zone is CRACKED under service loads — that is what reinforced concrete does — and an anchor in a crack has substantially less capacity than the same anchor in uncracked concrete. Evaluation reports give both figures and they differ enough that using the wrong one is not a margin, it is the design.
For adhesive anchors the largest field variable is not the product at all: it is **hole cleaning**. The adhesive bonds to the hole wall, and drilling dust left in the hole is a bond breaker. Qualification testing is performed on holes cleaned by a stated procedure — brushing and blowing a stated number of times — and capacity on a hole that was not cleaned that way is not the tested capacity. It is the single most consequential thing that happens on site and the one nobody can see afterwards.
Torque is the other invisible variable. Tightening an anchor to a torque is a proxy for developing preload, and the relationship between them depends on thread condition and lubrication — which is why torque-controlled installation is the least reliable of the accepted methods, and why proof testing exists for cases where it matters.
Timber and masonry: the same logic, different modes
Timber fasteners divide into WITHDRAWAL — a screw or nail pulled straight out along its axis — and LATERAL, where the fastener is loaded across its shank and both the fastener and the wood around it yield. The two are calculated completely differently and a fastener good in one may be poor in the other.
Withdrawal from END GRAIN is the case to know: driving a screw into the end of a member puts its threads between fibres running parallel to it, and the holding power collapses. Design values for end-grain withdrawal are drastically reduced or not permitted at all, which is why a joist hanger exists rather than simply screwing through a header into the end of a joist.
Masonry adds its own geometry. An anchor in a hollow unit may be in a face shell, in a grouted cell or in a bed joint, and those are three different base materials with three different capacities from three different rows of an evaluation report. The mortar joint in particular is the weakest of them and is frequently where a drill happens to land.
The alternative: proof test what matters
Computation is the right tool at design stage and a poor one for confirming what was actually installed. PROOF TESTING loads a sample of the installed anchors to a stated proportion of their design capacity and confirms they hold — it tests the product, the base material, the hole and the installer together, which is the combination no calculation can see.
It is a proof test rather than a capacity test: the load is deliberately below the anchor's design ultimate so that a passing anchor is undamaged and stays in service. Testing to failure tells you more and costs you the anchor, which is why it is done on sacrificial trial installations rather than on the real ones.
Where the consequence is significant — facade restraint, safety anchorage, a connection with no redundancy — the honest programme is both: compute the mode that governs, design so the steel does, and then test enough of the installed population to show the assumptions survived contact with the site.
Calculators that use this method
Basis
- ACI 318, Chapter 17, Anchoring to Concrete. The modal framework above, the Concrete Capacity Design breakout expressions, group-area reduction, edge-distance effects and the cracked-versus-uncracked distinction.
- Fuchs, Eligehausen and Breen (1995), Concrete Capacity Design (CCD) Approach for Fastening to Concrete, ACI Structural Journal. The origin of the embedment-to-the-1.5 relationship.
- ACI 355.2 and ACI 355.4 — qualification of post-installed mechanical and adhesive anchors, including the hole-cleaning procedures that the published capacities are conditional on.
- ICC-ES evaluation reports for the specific product, which are the only source for a real anchor's values and are row-specific to base material and condition.
- ASTM E488 for static anchor testing and ASTM E1512 for proof testing of post-installed anchors in concrete.
- National Design Specification for Wood Construction (NDS), Chapter 12 — withdrawal and lateral design values, and the end-grain withdrawal restriction.
- TMS 402 / ACI 530, anchor bolts in masonry — capacity by embedment, edge distance and whether the anchor is in a face shell, a grouted cell or a bed joint.