Green environment needs green concrete. To fulfill this requirement high-performance
concrete can be a good choice. High-performance concrete has better durability and longterm service life than ordinary concrete. Based on this fact we can decrease the cement consumption in construction industry and so does its production. Practically it leads to a lower carbon dioxide emission into the atmosphere. Since concrete corrosion is the main problem of durability all over the world, especially in Persian Gulf region, in this paper it is tried to model two important concrete properties related to its durability; i.e. chloride diffusion coefficient (D) and electrical resistivity (+). These properties are measured and determined numerically for different types of concretes with natural and semilightweight relatively high strength aggregates. The latter aggregate can be used in structural elements. In order to model D and + different concrete mixes are designed and tested in laboratory. Based on
the experimental findings an artificial neural network model has been developed with mix proportion values as inputs and D and + as outputs. The trained neural network has been verified with the checking data. Experimental and predicted results are compared with the results of other research works. It is found that the artificial neural network is capable of predicting D and + well enough for durability design requirements. This model can be improved by updated and/or new data whenever become available.