PSI - Issue 17

Maria Apostolopoulou et al. / Procedia Structural Integrity 17 (2019) 914–923 Maria Apostolopoulou et al. / Structural Integrity Procedia 00 (2019) 000 – 000

920

7

5. Conclusions

In the current study, the soft computing method namely ANN was developed for prediction of the compressive strength of NHL mortar mixes. A database which was compiled containing experimental datasets (a total of 253 datasets) as found in relevant published and related to natural hydraulic lime mortars produced with different types of natural hydraulic lime (NHL5, NHL3.5, NHL5) was used for training and validating the ANN model. Out of these, the mortar mix design parameters, selected in order to be used as input parameters in the ANN, and considered as the most crucial for the development of compressive strength, were maximum grain size of the sand used as aggregate (MDA), the age of the specimen(CT), the natural hydraulic lime category (5, 3.5, 2) (MEP), the ratio by weight of binder (natural hydraulic lime) to aggregate (BS) and the ratio by weight of water to binder (WB), while compressive strength was selected as the output parameter (CS). The optimum results were obtained when the three types of natural hydraulic lime were considered as three different input data parameters (assigning the number 1 to the type of NHL used in the dataset and the number 0 to the other two types respectively), revealing that the number of classification (2, 3.5 and 5) is not an adequate designation for this purpose. Furthermore, the ANN where the aggregates’ maximum size was incorporated as an input parameter p resented better results than the ANN where this information was excluded, highlighting the importance of this mortar mix parameter. Thus, the optimum ANN corresponded to a back-propagation neural network (BPNN) model with two hidden layers, incorporating all examined mortar mix parameters as input parameters and three encoding parameters for NHL type, resulting an 7 12-25-1 architecture, with the use of the normalization technique; its transfer functions are the Log-sigmoid transfer function (logsig) for the first hidden layer, and the hyperbolic tangent sigmoid transfer function (tansig) for the second hidden layer and the hyperbolic tangent sigmoid transfer function (tansig) for the output layer. The results of this study show that ANNs can predict the compressive strength of natural hydraulic lime mortars in a satisfactory manner, indicating that they can act as a tool for decision making when designing a natural hydraulic lime mortar. However, it is also worth noting that, despite the satisfactory derived results, the proposed neural network should be applied with caution. Despite the fact that the database used is the largest used in the relevant literature up to day, the authors consider that this database needs to be embellished with further experimental data. To this end, it is within the authors’ future plans to conduct further experiments related to these NHL mortars. In particular, the experiments lacking are related to the cases of NH2 and NH3.5. This embellishment will allow the ANN to reveal the combined influence of the different parameters on compressive strength to its full extent, thus revealing laws which govern the development of compressive strength of natural hydraulic lime mortars. The authors would like to thank Dr. Liborio Cavaleri, Prof. of Structural Engineering and Seismic Design at Dipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali, University of Palermo, Italy and Dr. Binh Thai Pham, Prof. at University of Transport Technology, Hanoi, Vietnam, for their valuable comments and discussions. The authors would also like to express his acknowledgement to graduate students Chrysoula Karamani, Athanasia Skentou and Ioanna Zoumpoulaki for their assistance on the computational implementation of the ANN models. References Abad, S.V.A.N.K., Yilmaz, M., Armaghani, D.J., Tugrul, A. (2018). Prediction of the durability of limestone aggregates using computational techniques. Neural Computing and Applications, 29(2), 423 - 433. Akkurt, S., Ozdemir, S., Tayfur, G., & Akyol, B. (2003). The use of GA–ANNs in the modelling of compressive strength of cement mortar. Cement and concrete research, 33(7), 973 - 979. Amenta, M., Karatasios, I., Maravelaki - Kalaitzaki, P., & Kilikoglou, V. (2017). The role of aggregate characteristics on the performance optimization of high hydraulicity restoration mortars. Construction and Building Materials, 153, 527 - 534. Acknowledgements