Prediction and identification of capillary water absorption capacity of travertine dimension stone

Abstract

The capillary transport mechanism has an important role on natural building stones that could be affected by moisture due to atmospheric conditions. Travertine with its natural porous structure is more sensitive to water which has a significant impact on durability. One of the important indicators of stone–water interaction is color variations, while natural stones with high capillary water absorption potential are in contact with water due to atmospheric conditions. As far as water absorption is concerned, water absorption by weight (WAW), water absorption by volume (n, apparent porosity), and capillary water absorption (CWA) parameters are in close relation with each other. To develop the relationships between capillary water absorption and the other water absorption parameters, regression and artificial neural network (ANN) analyses were performed. Within the scope of this study, the ANN models with three different input structures were established using various input variables in order to determine the relationship between CWA and the other parameters. The results of the different ANN structures of these models have been calculated and evaluated. For this purpose, travertine samples quarried from Denizli region in Turkey were studied. In order to evaluate the model results of regression and ANN methods, statistical performance evaluation parameters, i.e., the coefficient of correlation (CORR), efficiency (E), and root mean square error (RMSE), were calculated. In both regression and ANN models, reasonable coefficients of correlation were calculated. The capillary water absorption test is time consuming. The necessary periodical measurements make the process even more impractical. It is known that capillary water absorption is directly correlated with similar water absorption parameters such as water absorption by weight and apparent porosity. In this study, the estimation of capillary water absorption by means of other water absorption parameters is proposed. Moreover, there is no proposed classification in literature for natural stones based on their capillary water absorption. In the final stage of this study, by using the relationships between porosity and capillary water absorption, a capillary water absorption classification was proposed for travertine. Proposed classification consists of five categories and capillary water absorption coefficient which is ranging between very high (>68 g/m2 s0.5) and very low (<1 g/m2 s0.5). It is thought that this study could provide significant advantages in engineering practice and also make contribution to the related literature. © 2015, Saudi Society for Geosciences.