Seismic assessment and retrofit of concrete encased riveted steel buildings in New Zealand

Abstract

In the early 20th century, steel frame buildings were built to different standards from those used in modern construction. Riveted, built-up members and connections were often used, with joints and members encased in concrete for fire protection. These early steel buildings were designed based upon observations of past building performance rather than through detailed calculations and predictions of structural behaviour.

The walls were infill masonry or concrete, either unreinforced or very lightly reinforced and floors were typically cast in place reinforced concrete. The strength and stiffness of the semi-rigid connections and masonry infill as well as the effect of floor slabs integral with their supporting beams were not well documented and probably not well understood. Examples of these structures can be found throughout the cities and towns of New Zealand and many are quite prestigious, in full service and often enjoying heritage status. Developing methods of assessing their seismic strength and serviceability is a major objective of the research described herein. Assuming that their seismic strength is inadequate leads to the further objective of devising a retrofitting technique that will raise their strength to acceptable levels (without violating the heritage constraints). Seismic strength results from the resistance of the infill walls and steel frames, coupled together by the floor diaphragms. Assessment of the building strength was subdivided into two investigations, one focussing on the contribution of the walls (the subject of a separate PhD programme) and the present work which examines the contribution of the concrete encased steel frames.

A search for candidate buildings due for demolition was undertaken in the hope of obtaining full scale, as-built beam, column and joint assemblies that could be removed and tested. When this was unsuccessful it was decided to adopt the archetypal, 1928, 9 storey Hope Gibbons building in downtown Wellington as a case study. It offered typical construction details, a near complete set of drawings and limited access to sample material properties. Three experimental tests were conducted on half scale replicas of typical internal and external beam-column joints (including secondary beam stubs, integral composite floor slab and concrete iii encasing) to determine moment-rotation characteristics, failure mode, effect of concrete slab, crack development and opportunities for retrofit methods to enhance seismic performance. It was found that although riveted beam to column connections were typically not designed to resist moments, they could develop a modest moment capacity, contributing to the moment-rotation response and the horizontal strength and stiffness of a frame. Despite the early failure of the bolted beam flange to column face connections a moment-resisting action developed with the beam bottom flange bearing against the column face and the beam top flange tension transferred into the floor slab. This overloaded the floor slab and its low reinforcement ratio but suggested a retrofit technique in which carbon fibre reinforced polymer strips (CFRP) were embedded in the floor slab surface using the near surface method (NSM) to increase the tension capacity. Further tests on half scale joint replicas were carried out to test the effectiveness of CFRP NSM retrofit showing that increases of moment strength of the order of 20% were readily obtainable at rotations of around .04 radians (depending on initial beam flange to column face gap).

The experimental work was complemented by a comprehensive finite element (FE) analysis programme using Abaqus software. Half and full scale FE models were calibrated by the experimental data and the FE output in turn was used to construct simplified moment-rotation joint characteristics. Overall frame analysis was then undertaken using Ruaumoko software with the previously mentioned joint characteristics. This enabled seismic time-history and push-over analyses to be completed to assess the as-built and retrofitted strength and stiffness. It was found that the retrofitting increased the frame strength by about 5% and reduced drift by a similar percentage. However, the strength, even with retrofitting did not reach the benchmark of 0.3 of new structure strength.