Effects of micro pore characteristics on strength of cement mortar using artificial neural network
No Thumbnail Available
Date
2011
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Cementitious materials comprise a great part in construction process of structures such as buildings, bridges, roads and dams. The most expected properties of structural members prepared with cement mortar or concrete, are strength and durability. These structural members are supposed to have strength values determined in the structural analysis and to be durable against aggressive media in their service life. These characteristics are the most effective criteria in civil and material engineering. Therefore, these two parameters depend mainly on the pore structure and its characteristics of structural members. Nowadays, scientists and engineers are using new computer technologies, simulations and experimental techniques try to perform to characterize the inner structure of structural materials in order to define microstructural formations and the effects of microstructural phases such as pores on macro properties. New image capturing tools and their improved magnification capacity induced researchers to have an expanded view on investigation of microstructures. In addition, the results of these studies are simply not enough to realize the simulation of effects of inner structure. Some numerical and statistical methods performed by computers are needed at this stage. Artificial neural network (ANN) is one of these methods. In last decades, artificial neural network applications have become more considerable issue in engineering applications. In the scope of this chapter, pore area ratio values represent total pore area amount in a polished section of cement mortars were determined. Also, some pore characteristics representing the probability of channels between pores are investigated. The pore amounts and these pore characteristics are related to compressive strength values of cement mortars in order to establish a microstructure - macro property relationship. Thus, nondestructive methodologies and artificial neural network have been used in the prediction of a macro property, which only be determined by destructive testing techniques. © 2011 by Nova Science Publishers, Inc. All rights reserved.