Conductivity and Artificial Neural Networks applied to the evaluation of the apparent mass diffusion coefficient in concrete

摘要

The aim of this paper is the determination of the routine of chemical species in porous modeling material in transient state by means of conductivity mass balance. The experimental results and the inherent characteristics of concrete which from have given an apparent diffusion coefficient in the range of 10-11, gathered into a data base enable to us develop a neural model capable of predicting the apparent diffusion coefficient in concrete with error not exceeding 0.09%.