Modeling the influence of composition and pore structure on mechanical properties of autoclaved cellular concrete.

摘要

This dissertation presents a comprehensive investigation of the influence of composition and pore structure on the mechanical properties of Autoclaved Cellular Concrete (ACC). The effect of the mix proportions on the mechanical properties (i.e. compressive strength, density) of ACC is analyzed using the Orthogonal Array Experimental Design. The effect of the pore structure on the mechanical properties of ACC is analyzed using a combination of image Analysis, Fractal Theory, and Fracture Mechanics Theory. From the image analysis a broad category of pore structure parameters are obtained. The pore roughness and pore-size distribution are quantified using factal theory. The mode of failure of ACC when subjected to compressive loads is interpreted by the principles of fracture mechanics theory. Using the results of laboratory tests, a compressive strength-porosity relationship is formulated which in turn is a function of the raw material properties, mix proportions, and density of ACC. Also, artificial neural network-based ACC models are developed for reproducing experimental results and predicting the results of other experiments as well as for optimization of the manufacturing process of ACC.; This investigation revealed that the orthogonal array experimental design was practical for the design of the ACC. Using this method, the number of mix proportions and the number of experiments needed were minimized, while at the same time, the maximum amount of information such as the impact of raw materials of ACC on its mechanical properties and the optimum mix proportions for optimizing test results was obtained. Image analysis was a very practical tool to investigate the pore structure of ACC. By using an image analyzer, a broad category of pore parameters such as perimeter, area, and diameter of the pores was obtained. It was also discovered that using fractal theory, a reliable evaluation of the pore roughness and the pore-size distribution in ACC can be obtained. The laboratory results indicated that, in general, the higher the compressive strength, the smaller were the porosities and the pore sizes of the ACC. The compressive strength increased with an increase in value of the fractal dimension for the pore-size distribution. High values of this fractal dimension represents ACC with a large number of small, uniform pores. Thus, ACC with large percentage of small, uniform pores have high compressive strength values. Additionally, a relationship was derived between the compressive strength of ACC and its porosity. This porosity, in turn, was found to be related to the water-cement ratio, the saturated density, the mix proportions (by weight) of the raw materials, and the specific gravity of each raw material. The compressive strength-porosity relationship can be used as a guideline for the mixture selection to develop a particular ACC design that meets some particular requirement. Finally, artificial neural network-based ACC models were formulated in order to reproduce experimental results and approximate the results of other experiments using the same materials. The potential of using neural network techniques in the manufacturing of ACC is also discussed.