Abstract
EN 1992-4 verifies concrete edge failure of a multi-row anchor group loaded in shear towards a free edge by assuming that only the row closest to the edge is effective and that it carries the whole shear load. The consequence is paradoxical: the verified resistance does not grow when rows are added behind the front row, it falls when a row is added in front of existing anchors, and it rises when the holes of the front row are deliberately slotted so that those anchors carry nothing. For a 3 × 3 group of M12 wedge anchors (s = 150 mm, hef = 70 mm, C25/30, h = 250 mm) at c1 = 75, 100 and 150 mm, the code gives 31, 39 and 56 kN (uncracked), while the same anchors with the front row slotted give 73, 78 and 87 kN. The paper examines this rule with three tools: the EN 1992-4 calculation, the equal-share method implemented in the AAS anchor-design software (every anchor carries the same share and the front row is verified with its share only), and a three-dimensional nonlinear finite-element analysis in Code_Aster with a regularised damage model for concrete, elastic steel shanks and a rigid fixture. The analyses show that, without hole clearance, every row is active up to failure: at peak load the front row carries practically its own stand-alone resistance while the two rows behind it carry 2.1 to 4.4 times as much, and the group resists 5.4, 3.9 and 2.9 times the load of the front row acting alone — where EN 1992-4 assumes 1.0. Removing the front row altogether does not increase the computed resistance, so adding anchors near the edge does not weaken the group, and with a hole clearance that lets the front row fail first the group still reaches 92 to 94 % of its resistance, at a larger displacement. Normalised to the characteristic level, the finite-element group resistance is about 147, 146 and 161 kN, i.e. 4.7, 3.8 and 2.9 times the EN value. The equal-share (AAS) method is close to this behaviour and conservative for c1 = 75 and 100 mm, but at c1 = 150 mm it reaches the resistance of the global break-out body that starts behind the back row, which the equal-share rule does not see. A verification rule is therefore proposed that keeps the equal share and adds that global check: the group resistance is the lesser of (n/n1) times the front-row resistance and the resistance of the back row carrying the whole load. For the cases studied it is conservative by a nearly constant factor of 1.4 to 1.6 with respect to the normalised finite-element results, against 2.9 to 4.7 for the present rule. Evaluated with the ACI 318-19 break-out equations, the same two checks — which correspond to cases illustrated in the ACI commentary — give 94, 124 and 147 kN, and a worked design example in cracked concrete shows the practical consequence: a fixture that fails the EN check by a factor of two passes the proposed one with a utilisation of 0.8.