Creating a simplified model of the Alkali-Silica Reaction in concrete by utilising finite element modelling techniques

The Alkali-Silica Reaction (ASR) in concrete was first discovered in the 1940s and has since become a well documented problem in structures in Australia and around the world. The result of ASR can result in unexpected and premature deterioration of large and expensive buildings and pieces of infrastructure such as bridges and dams. The nature of ASR means that not only does it reduce the capacity of the concrete but also expose it to further deterioration. It is for this reason that it has become a serious concern for the owners of these structures and engineers alike. There have been many studies into the mechanisms that drive ASR, many of which have involved complex numerical and mathematical modelling as part of the study. Due to the complex nature of the reaction there are no models that are simple but effective enough to use on a practical basis. This thesis was designed to produce a macroscopic model of the alkali silica reaction using Finite Element Analysis (F.E.A.) software in order for engineers to use on a practical basis. A simplified model operates by idealizing a scenario and focusing on producing fairly accurate results for a given set of parameters rather than try to model the process in its entirety. These models can be very useful when conducting preliminary assessments because they can model the behaviour of a system on a macro scale. Having a simplified model would potentially save a large number of man- hours and material costs when trying to repair these structures by giving the engineers a greater understanding of the current structural state and the extent to which they need to either repair or demolish and rebuild. Last year, Tracy Knight simulated the growth of ASR in physical specimens, demonstrated the reduction in compressive strength of a sample and then showed that the compressive strength could be restored by using composite materials to create a confining pressure around the column. I intended to construct a simplified model of the ASR by using the Finite Element Modelling (FEM) software Strand7 and then check my model by replicating her compression test results. A test such as this where the model has to replicate certain known characteristics is a good method of validation, otherwise the model might predict erroneous behaviour. Strand7 does not have the capability to model the crack initiation and propagation which is integral in the ASR process therefore an alternate way of replicating the reduction in compressive strength needed to be found. The tensile forces associated with the alkalai silica reaction will be replicated by way of elevated temperatures at the nodes. By encouraging expansion in the localised as a function of temperature it recreates the expansion characteristic of the ASR gel along with the tensile forces within the surrounding concrete matrix. The development of this model started by reproducing the results of a simple compression test on a concrete cylinder to achieve a baseline for my results to be reflected against. Next the nature of the ASR would be replicated by way of a temperature being applied at various nodes to induce thermal expansion in the concrete and the patterns of cracking investigated. This is then refined in the next stage which involved applying both the temperature and compressive loading simultaneously to confirm that this process replicated the reduction in strength associated with ASR and this was verified using a three dimensional model. The modelling process had been planned to include a stage at the end where the three dimensional model would be confined with fibre composites in order to replicate Tracy Knight’s thesis and restore the original strength to the concrete but the work load and time constraints did not allow for this to happen. Many opportunities exist to extend the work I have begun. Not enough time has been devoted to refining the method of determining how many temperature nodes are required and what type of loading conditions various patterns will give. The ultimate goal of being able to replicate the results in Tracy Knights thesis is yet to be achieved. The results of the compressive testing will continue to be used to validate the models created