Magazine of Concrete Research · 2012 · 32 citations · 15 references
EngineeringMaterials EngineeringMaterials ScienceBuilding MaterialsComputational Materials ScienceMachine LearningCementitious MaterialsConcrete ConstructionConcrete MaterialsConcrete TechnologyCorrosion Of Construction MaterialsNeural NetworksCarbonationCarbonation EffectsDiffusion CoefficientsDiffusivityIon PermeabilityIonic Diffusion CoefficientsMultilayer PerceptronPerceptron
In this paper the effect of carbonation on chloride diffusivity in concrete is investigated. An artificial neural network model was used to determine the relation between chloride diffusion coefficients and concrete mix design in carbonated and non-carbonated concretes. The models were trained by results of chloride profile experiments. Input parameters were water-to-binder ratios, the amount of silica fume, rapid chloride ion permeability test and capillary absorption coefficient. The output parameter was chloride diffusion coefficient. The neural network models are multi-layer perceptron models and they differ in the number of hidden layers and neurons.
15