Basic assumptions for the analysis and design of steel frames include modelling the beam to column connections as either pinned or rigid. Neither accord with reality in the sense that all practical forms of steelwork beam to column connection possess some degree of rotational stiffness and have some ability to transmit moments. For bare steelwork, there will be arrangements for which both properties are negligibly small - leading to the very reasonable assumption of "simple construction" - or some for which almost complete rigidity coupled with moment capacities equal to that of the connected members are generated - leading to treatment as "continuous construction". Both approaches are, of course, simplifications, leading to convenient design procedures and associated calculations. When considering composite frames both assumptions become less satisfactory. Whilst "simple construction" leads to the minimum of calculations and a design approach that essentially works at the level of individual elements, it neglects several potentially beneficial features and actually requires adaptation of the resulting design to recognise the impossibility of actually providing pinned joints. Continuous construction is clearly irrational if conducted on an elastic basis due to the imbalance of the hogging and sagging moment capacities for composite sections, whilst moving to a plastic approach introduces severe limitations on the acceptable structural arrangements. All this leads naturally to an approach [1] that recognises the true characteristics of the joints and treats them as semi-rigid and partial strength (semi-continuous construction) - or in North American parlance "partially restrained (PR) construction". Research over several decades devoted to the subject of semicontinuous construction [2] has now advanced understanding for the bare steel case to a sophisticated level. Beginning with the development and representation of connection moment-rotation curves, then incorporating this in frame analysis, studying the effects on member stability and moving on to extensions to consider repeated and reversed loading of connections, the subject is now generally well understood. An important feature is that the situation is rarely affected by considerations of the available ductility (rotation capacity) of the connections. Having developed the concept for bare steelwork, it was natural to consider the possibility of extending the scope to cover composite construction. The immediate attraction was that designing as continuous was not sensible if done elastically - peak moments occur at supports for which the cross-section is in hogging bending and thus possesses a lower moment capacity than the sagging cross-section - and difficulties occur when treating as fully plastic due to the rotation capacity requirements in the hogging cross-sections imposing severe limits on the dimensions of the steel members. Locating the plastic rotation necessary for moment redistribution in the connections was an obviously attractive solution [1]. However, even the earliest studies of composite beam to column joints [3], [4] showed that enhanced moment capacity was accompanied by reduced ductility. Unlike tests on bare steelwork, for which it was frequently not possible to achieve a peak load due to exhausting the travel of the jacks, concern over the measuring equipment etc., several different failure modes were observed at relatively low levels of rotation. Although careful detailing of the connections could alleviate the situation somewhat, it was clearly not possible to adopt the convenient position for bare steelwork that "providing adequate rotation capacity in the connections was not normally an issue". In contrast, for semicontinuous composite design, ensuring an appropriate balance between the rotations required in the connections to generate the assumed degree of moment redistribution (supports to midspan) and that actually available from practical forms of joint was clearly an important problem parameter and something requiring explicit attention in any design procedure. The present paper focuses on this issue, using it to illustrate several features of modern structural steelwork research and showing how it concentrates attention on several fundamental properties of the behaviour of steel structures.
