A regression model predicting the compressive strength of concrete by means of nondestructive, acoustic measurements

Abstract

A sample size of 81, 4" diameter concrete standard cylinders were tested using nondestructive means. These cylinders were collected from three different ready-mix plants across Texas located in Houston, San Antonio and Victoria. The sound produced by the bounces of a steel ball on the test cylinder were digitally recorded. The independent variable, compressive strength, in pounds per square inch was then measured by means of a standard ASTM test using a hydraulic press. Certain acoustic measurements were made on the recordings using a software package called Cool Edit 2000. The study found that the compressive strength of a concrete cylinder could be predicted by a combination of plant location and acoustic measurements. A multiple linear regression model was found to be significant at p-value < 0.001 and had an adjusted R² of about 0.85. The model was: COMPRESSIVE STRENGTH = 6447.202709 - 1378.58823* Houston - 2602.62171* San Antonio + 33177.54748* Elapsed Time from Peak-two-to-Peak-three. Elapsed time is the time between two bounces measured in seconds. A peak is produced every time a steel ball bounces on the surface of the concrete cylinder. The elapsed time here is the time between the second and third bounce.