Abstract
It is over a century ago that testing of reinforced concrete slabs by the pioneers of this form of construction such as Lord, Turner and Maillart showed that restrained slabs could carry significant loads. Since that time the interest in and knowledge of the internal arching, or compressive membrane action, that enhances the strength of reinforced concrete, has waxed and waned. The current generation of Eurocodes do not mention or specifically consider compressive membrane action. In this thesis a review of the key 20th century research, theory and testing is critically discussed, with particular emphasis on aspects related to bridge decks. The limitations and validity of the tests, particularly the use of small-scale tests, are noted. Current theories, which can be over complex, are also reviewed. It is also noted that all theory and most tests consider the ultimate limit state only; there is limited information on serviceability issues of compressive membrane action. The more recent advances in compressive membrane action and punching shear in the 21st century is also outlined. The two main codes of practice that specifically allow the use of compressive membrane action, the American AASHTO (2007) standard and the UK CD360 Standard (HE, 2017) are compared. The limitations of these codes are outlined Whilst there are over a thousand tests in which compressive membrane action may be present have been carried out over the last 100 years not all of the tests are useful towards assessing the contribution of compressive membrane action. As part of this research project a database of test data for structures in which compressive membrane action have influenced load capacity was collected and analysed rigorously. Based on recent work on interpretation of databases for code drafting, a simple set of criteria for evaluating the usefulness and consistency of the test data is proposed. An analysis of test data was carried out and compared with theoretical results and current AASHTO, CD360 and Eurocode requirements. The data was firstly presented graphically in ways seen previously in the literature; however, the number of data points is significantly increased from previous publications. Subsequently the data was refined and presented as part of a multi-dimensional interaction limit, to outline new insights. In the Eurocodes non-linear methods of analysis are accepted, finite element analysis is one way to carry out the non-linear analysis. Non-linear finite element analyses of partially restrained concrete slab strips under static loads were carried out to investigate behaviour in both bending and compressive membrane action. One of the drawbacks of non-linear finite element analyses is the number of parameters involved in obtaining a solution, and how to systematically define them. The material parameters of the Damaged Plasticity model within the ABAQUS program were defined systematically. A number of differing assumptions were made for the material parameters, and the effects of the variations assessed. The finite element method was used to fill in some of the gaps identified in the test database for unreinforced (plain) and reinforced concrete slab strips. The NL-FEA was then used to compare slab strips and one-way spanning slabs. The predictive capability and limitations of the calibrated model was outlined by simulating further slabs with different material properties and restraints under static line loads. The finite element models were validated against published test results shown to be of good quality by the database analysis. It was shown that the strains and rotations from the analysis can also be used to estimate shear capacity using the critical shear crack theory (Muttoni, 2008). The observations from the existing theoretical approaches (McDowell, et al., 1956; Christiansen, 1963; Park, 1964; Eyre, 1997; and Rankin and Long, 1997), the test database and finite element analyses were brought together towards the development of a novel 3-phase analysis that considers both the serviceability and ultimate limit states. The method considers an initial un-cracked phase; a cracked phase and a final reinforcement yielding phase. The first phase is a conventional elastic analysis. For the second phase, at the serviceability limit state, the use of an arching geometry method or a new effective strain method is proposed. A validation of the methods against the test data and verification with non-linear finite element analyses were carried out. For phase-3 some improvements or refinements to the Rankin and Long (1997) strut method as a strut and tie analysis based on the results of the non-linear finite element analyses and consideration of shear and rotational limits not defined by existing theories is proposed. The applicability of the proposed 3-phase method demonstrated for the analysis of slab strips, 1-way slabs and more complex beam and slab bridges. The majority of existing theory, tests and finite element analysis concentrate on local effects. In bridge structures both local and global effects need to be considered. The interaction of local and global effects was investigated using the 3-phase method, confirming their importance. The existing knowledge and additional findings about compressive membrane action from this thesis are brought together to consider the requirements for a tentative Annex to the Eurocodes, to allow a safe and systematic use of compressive membrane action.