ARTICLE DOWNLOAD

Predicting the compressive strength of unreinforced brick masonry using machine learning techniques validated on a case study of a museum through nondestructive testing

10$
ARTICLE DOWNLOAD

Predicting the compressive strength of unreinforced brick masonry using machine learning techniques validated on a case study of a museum through nondestructive testing

10$

Mayank Mishra, Amanjeet Singh Bhatia & Damodar Maity 

Abstract

Historical buildings, such as museums, are an important class of buildings because ancient historical artefacts are collected and preserved in them. These buildings must be maintained to prolong their service life and monitored continuously for possible signs of building damages. Determining the compressive strength is essential for predicting the remaining service life of museums because monumental museum buildings usually comprise load-bearing masonry walls. This study analysed the applicability of three machine learning techniques as alternative methods for predicting the compressive strength of masonry constructions. A database was established by performing experimental testing on 44 masonry specimens. In the nondestructive techniques used, the ultrasonic pulse velocity and the rebound number were inputs and the compressive strength of masonry unit (fmufmu) was the desired output. The remaining compressive strength of a museum was determined using commonly available nondestructive testing apparatus. The model predictions obtained through nondestructive measurements were compared with the experimental results obtained through the destructive testing of masonry units in the laboratory. The proposed approach was compared with traditional empirical models and commonly used machine learning techniques. The analyses revealed that machine learning techniques are robust, can successfully be used for the prediction of the remaining compressive strength of historical constructions, and thus can provide decision support for inspection professionals.

Only units of this product remain
Year 2020
Language English
Format PDF
DOI 10.1007/s13349-020-00391-7