No abstract

The leaping impact of increased population and commercialization on global energy demand, has prompted more concern for sustainable development and strength evaluation of concrete structures. This study was carried out to improve the strength of concrete by adopting nanosized cassava peel ash (NCPA) as partial replacement of cement and to model its strength with artificial neural network (ANN). Data used for the model were obtained experimentally. At any percentage not exceeding 20 % NCPA replacement, the concrete is a suitable structural material. The neural network was adequately trained to capture the relationship between the compressive strength values of NCPA-concrete and their corresponding mix ratios at 7 days, 14 days, 28 days, 56 days, 90 days and 150days curing. A 6-10-1 network architecture was created. A total of four hundred (400) training data set were presented to the network. Two hundred and forty (240) of these were used for training the network, sixty (60) were used for validation, and another sixty (60) were used for testing the network's performance. After training the network, the output and targets had an R - value of 0.99909 which is very close to 1. This shows that the data used for training the network, have a good fit. The results obtained from the network are approximately the same as that obtained experimentally. The adequacy of the network was further tested using the Student’s T test. The calculated T-value (-0.11) for the compressive strength of NCPA-concrete was less than that from the T-table (2.04) at 95% confidence level, proving that the network predictions are reliable. This model is reliable, time-effective and accurate for strength prediction of nanosized concrete.