Comparison of strength durability models for GRC

Cementitious materials are inherently brittle, usually exhibiting a modest tensile strength. This behaviour can however be modified by the addition of fibres: depending on the amount and type of fibres, an important increase in toughness can be obtained as well as a post cracking strain hardening behaviour in tension. Glass fibres are often used for this purpose, either in the form of chopped fibre strands (premix or sprayed), or as textile reinforcement. However, a main topic of consideration for glass fibre reinforced cementitious composites remains the durability. Glass fibre reinforced concrete (GRC) is prone to ageing, especially in a wet environment. The most important manifestation of this ageing is the loss of composite tensile strength and/or toughness, depending on the intended mode of reinforcement. Durability of GRC remains thus a widely studied topic, despite significant material optimization over recent decades. Evaluation of strength durability normally employs accelerated ageing tests involving immersion of specimens in water at a range of temperatures for various durations. These results are then used to calibrate a strength durability model.

Strength durability models can be based on several basic assumptions. In this presentation, we present an overview of four models available in the literature, all based on the fundamental idea that the loss of strength of a GRC component results from the growth of flaws on or in the glass fibre surfaces, based on the classical Griffith relationship. The difference between the models is the assumption which is made concerning the flaw size rate: the linear model (constant rate), the square-root model (rate gradually decreasing with time), the non-linear model (non-linear decrease or increase of the rate with time), and the combined model (a combination of the linear and non-linear models).

Two sets of experimental results from literature are used to calibrate the strength durability models, by achieving a best fit of the model curves with the experimental results at different temperatures. Secondly the influence of quantity and quality of accelerated ageing test results on the determination of model parameters is discussed, in order to examine the robustness of the models. In order to achieve these goals several theoretical/practical investigations are carried out on the different models: (1) the solution space is inspected for multiple minima, which could cause ambiguous application of a given model and (2) parameter studies are performed on the models in order to establish whether some models are prone to inherent calibration or convergence problems. This work therefore contributes to more acceptable/reliable service life predictions and more reliable activation energies and therefore better understanding of degradation mechanisms.