USE OF RICE HUSK ASH AS PARTIAL SAND REPLACEMENT IN LIGHTWEIGHT CONCRETE

Abstract

This paper presents a study of lightweight concrete using rice husk ash as partial sand replacement which applies from foam concrete and lightweight aggregate. Lightweight concrete made from Portland cement and sand at a 1:1 ratio by weight and 50% by volume of foam. Rice husk ash was used to replace sand at 0, 10, 20, 30 and 40 by volume of sand. Flows of all mixtures were controlled between 45 ± 5%. Compressive strength density and water absorption of lightweight concrete were determined at 7, 14, and 28 days. After that, the result were gathered for a mathematical model by using linear and nonlinear multiple regression techniques. Coefficient of determination (R²), root mean square (RMS) and mean absolute percentage error (MAPE) were analyzed to verify the accuracy of models. Furthermore, the scanning electron microscope (SEM) and energy dispersive x-ray spectrometry EDS were analyzed the particle characteristics of lightweight concrete. The results revealed that the use of rice husk ash as a fine aggregate in lightweight concrete caused to the decreasing of the weight. The optimum mixed ratio of rice husk ash as sand was 20% by volume which gave compressive strength density and water absorption at 28 days of 22 kg/cm², 954 kg/m³, and 28% respectively. In addition, the morphology particles and the chemical compositions of lightweight concrete shows that the used rice husk ash to replace sand did not have effect on artificial air pores and it does not have change the chemical composition of lightweight concrete. It can be concluded that the rice husk ash could be used as sand with optimum content for lightweight concrete.

  The non-linear regression models had more accurate and precise prediction than linear technique. The coefficient of determination (R²) of compressive strength, water absorption and density of lightweight concrete containing rice husk ash were provided at 0.978, 0.966 และ 0.987 respectively. In addition, the variation between database sets, the root mean square (RMS) and mean absolute percentage error (MAPE) were lower as compared to linear regression technique.