Abstract

NGANDU, Cornelius Ngunjiri. Prediction of Compressive Strengths for Rice Husks Ash incorporated concrete, Using Neural Network and Reviews. Iteckne [online]. 2021, vol.18, n.2, pp.99-107.  Epub May 16, 2022. ISSN 1692-1798.  https://doi.org/10.15332/iteckne.v18i1.2532.

Modelling of concrete that incorporates agricultural wastes such as rice husk ash (RHA) could potentially enhance utilization of green concrete and application of sustainable construction materials. This paper evaluations compressive strength prediction for rice husk ash (RHA) cementitious material incorporated concrete using artificial neural networks (ANNs) one of the various prediction methods. The research is based on various previous experimental studies.

Literature reviews of 72 datasets for RHA incorporated concrete from 15 previous researches, were used and subjected to ANNs models, having learning rate of 0.06 with tanh activation functions. Four (4) input variables were considered, namely:superplasticizer or water reducers variation from control (%), water to binder ratio, percentage of RHA and control compressive strengths. Output variable was compressive strength of RHA cementitious material incorporated concrete. The ANN with 15 neurons in the hidden layer was selected and indicated overall values of 5.10MPa, 0.99, 3.81MPa and 9.73% for the root mean square error (RMSE), absolute factor of variance (R2), mean absolute error (MAE) and mean absolute percentage error (MAPE) respectively and for individual training, validation/checking and testing datasets, the RMSE, R2, MAE and MAPE ranging between 3.98MPa-6.56MPa, 0.98-0.99, 3.44MPa-4.94MPa and 9.19%-12.41% respectively. Generally, both predicted and original dataset, indicated higher and lower strength values for 5-10% and 15-30% RHA incorporated cementitious material concrete respectively compared to the control strengths.

Considering that the study utilized data from different sources and with a wide range of concrete strengths the selected ANN showed relatively good performance. The study provides an indicator that machine learning techniques could accurately predict green concrete strength. Based on model performance the percentage RHA cementitious materials in concrete and the other 3 input variable had a significant impact on concrete strengths. Future research should be conducted to predict green concrete focused on particular concrete class.

Keywords : Prediction methods; concrete; artificial neural networks; Literature reviews.

        · abstract in Spanish     · text in English     · English ( pdf )