This paper describes the results of a research study to determine the effectiveness of the Troxler model 4640 thin-lift nuclear density gauge in estimating core densities. The study consisted of obtaining density measurements using cores and the nuclear gauge on seven construction projects and comparing the nuclear to core density readings. The projects were either newly constructed or under construction when the tests were performed. Correlation coefficients were determined to indicate the degree of correlation between core and nuclear densities. Linear regression was used to investigate how well the core densities could be predicted from nuclear densities. Using statistical analysis, the ranges of differences between core and nuclear measurements were established for specified confidence levels. Analysis of the data shows that the accuracy of the nuclear gauge is highly material-dependent: the gauge produced acceptable results with limestone mixtures, but it did not perform satisfactorily with mixtures containing siliceous aggregate. The data presented in this paper indicate that the gauge could be used as a quality control tool, provided calibration lines are developed for each project; calibration lines can be developed using simple linear regression.
Abstract