Abstract
EN 1992-4:2018 is the European standard for the design of fastenings in concrete. Its resistance models rest on an extensive experimental basis, but several of its load-distribution rules, interaction rules and modification factors are not derived from that basis, or apply it beyond the conditions in which it was obtained. This paper reviews seven such provisions. Three have been examined in detail in earlier studies by the author and are summarised here: the default assumption of cracked concrete and the undefined depth of its 0.3 mm reference crack; the pry-out factor k8, which links pry-out to tension resistance by a factor of 1 or 2 where mechanics and test databases indicate about 2.5 to 4; and the concrete edge verification of multi-row groups, in which only the row closest to the edge is effective, so that adding anchors can lower the verified resistance by 36 to 58 %. Four further provisions are examined here for the first time, with worked examples computed with the formulae of the standard: (i) the tension–shear interaction of Table 7.3 , which combines the most utilised tension mode and the most utilised shear mode with one exponent whatever the mechanisms involved; (ii) the splitting provisions of §7.2.1.7, where the only exemption for uncracked concrete requires designing for cracked concrete, so that for an M12 wedge-anchor group the verified resistance falls from 63.1 kN to 32.2 kN with the splitting check, or to 40.0 kN plus 138 mm² of reinforcement without it; (iii) the constant factor ψ90°,V = 2.0 for shear parallel to an edge, which a resistance-dependent expression of the fib type places between 2.4 and 1.1 for the same anchor as c1 grows from 60 to 200 mm, so that the constant is unconservative by up to 87 % at larger edge distances; and (iv) the shell-spalling factor ψre,N, which reduces cone, bond and, through k8, pry-out resistance by up to 30 % because reinforcement is present, and has no counterpart in ACI 318. The provisions are classified by the nature of the departure — contradiction of tests, inconsistency with a parallel code, internal logical inconsistency — and a research programme is proposed for each. The review refers to the first-generation text; the second-generation Eurocodes were not available to the author.