Estimating Strain Changes in Concrete during Curing Using Regression and Artificial Neural Network

Abstract

Due to the cement hydration heat, concrete deforms during curing. These deformations may lead to cracks in the concrete. Therefore, a method which estimates the strain during curing is very valuable. In this research, two methods of multivariable regression and neural network were studied with the aim of estimating strain changes in concrete. For this purpose, laboratory cylindrical specimens were prepared under controlled situation at first and then vibration wire strain gauges equipped with thermistors were placed inside each sample to measure the deformations. Two different groups of input data were used in which variables included time, environment temperature, concrete temperature, water-to-cement ratio, aggregate content, height, and specimen diameter. CEM I, 42.5 R was utilized in set (I) and strain changes have been measured in six concrete specimens. In set (II) CEM II, 52.5 R was employed and strain changes were measured in three different specimens in which the diameter was held constant. The best multivariate regression equations calculated the determined coefficients at 0.804 and 0.82 for sets (I) and (II), whereas the artificial neural networks predicted the strain with higher of 1 and 0.996. Results show that the neural network method can be utilized as an efficient tool for estimating concrete strain during curing.