Abstract
EN 1992-4:2018 does not verify concrete pry-out on its own terms. It sets the pry-out resistance of a fastener at a multiple k8 of its tension resistance: the concrete cone resistance for mechanical fasteners, and the smaller of the cone resistance and the combined pull-out resistance for bonded fasteners, with default values k8 = 1 or 2. This paper examines that factor. For the bond term, clause 7.2.2.4(1) names the mechanism itself: pull-out under a tension force induced by the shear. With k8=2 the code therefore assumes that half of the shear reaches the bond as tension. The mechanics of a laterally loaded anchor in concrete, with local crushing at the mouth of the hole, a pivot at 1.5d–2d and a tilted shank, give an induced tension of 14–27 % of the shear for plastic tilts of 8∘–15∘, so the factor between bond resistance and the shear it can resist is about 4 to 7. A credibility criterion shows that shear-induced pull-out can precede steel shear failure only below a bond-strength threshold that depends on hef/d alone; in 76 single-anchor configurations within the admitted embedment range the bond term governed 21 times under the code and never with a factor of 4, at the characteristic and at the design level; where the sustained-load factor reduces the whole bond resistance the bond term can return, which makes the treatment of ψsus inside the pry-out check a second question. For the concrete term, the largest recent test database shows the code to be conservative, with mean test-to-prediction ratios of 1.27 for single anchors and 1.63 for groups, corresponding to an effective factor of about 2.5 and 3.3, and product assessments already grant k8 up to 2.9. The group rules reduce the resistance with overlap factors derived for tension, although for anchors in line with the load the rear anchor compresses the concrete in which the front anchor’s crater must form. The provisions are also inconsistent: the same Hilti HUS4 screw is checked without its pull-out resistance as a mechanical fastener and with it as a bonded screw, and the mortar is credited for embedment depth but not for diameter. A separate factor kp≥4 on the bond term and a test-based recalibration of k8 for single anchors and groups are proposed, with a test programme to confirm them.