Combining non-invasive techniques for reliable prediction of soft stone strength in historic masonries. (15th August 2017)

• Record Type: Journal Article
• Authors: Vasanelli, Emilia; Colangiuli, Donato; Calia, Angela; Sbartaï, Zoubir-Mehdi; Breysse, Denys
• Is Part Of: Construction & building materials. Volume 146(2017)
• Page Start: 744
• Page End: 754
• Publication Date: 2017-08-15
• Subjects: Limestone masonry -- Uniaxial compressive strength assessment -- Non-destructive tests -- Regression analysis -- Combination techniques -- Artificial Neural Networks -- Cross-validation procedure; Building materials -- Periodicals; 624.18
• DOI: 10.1016/j.conbuildmat.2017.04.146
• Languages: English
• ISSNs: 0950-0618

Abstract

Highlights: UPV, Rebound and compressive microcore tests for stone masonry strength estimation. Statistical evaluation of accuracy of each test measurements is performed. Multivariable methods improves correlation models between direct and indirect tests. Iterative cross-validation is effective in evaluating model predictive performances. Combined UPV and microcore strength tests gives the best estimation of UCS. Abstract: In this study, some NDTs (Ultrasonic Pulse Velocity UPV and Rebound Hammer) and uniaxial compressive test on microcores (UCSm ) as a moderately destructive test, were investigated as tools for assessing the uniaxial compressive strength (UCS) of a soft limestone. Correlations between UCS and results of each above-mentioned tests were determined by a univariable regression analysis. Artificial Neural Network and the Multiple Regression Analyses were considered to search correlations between UCS and combined results of the non-invasive tests. An iterative cross-validation procedure was implemented to validate the predictive performances of the models. It was found that combining UPV and UCSm results gives the best reliability in the indirect estimation of UCS, with a notably reduced predictive error.