Implementing GIS and linear regression models to investigate partial building failures

Abstract

One of the most dangerous field problems in the civil engineering discipline is the suddenly developed cracks in the building, which could be caused by the swelling of the subsurface soil. Thus, this work has focused on employing a procedure in the geographic information system known as the inverse distance weighted (IDW) technique, to analyze the extent of cracks in a residential complex in the city of Kirkuk in Iraq using the physical and chemical soil data for seven boreholes from the field of the study. Physical soil parameters such as liquid limit (LL), gravel, sand, silt and clay percentages were characterized first, followed by chemical properties such as gypsum content (GYP), total suspended solids (TSS), potential of hydrogen (pH), and organic content (ORG). Furthermore, statistical studies such as plasticity index (PI) and soil characteristics association, linear single, and various linear multi-regression models were used. The data analysis shows that there are significantly positive and negative relationships between PI as a swelling indicator and the physical and chemical soil properties, although weak to moderate correlations were observed between PI and these variables. The PI values were accurately predicted by the proposed linear multi-regression models of the physical and integrated physical and chemical soil characteristics, with multiple R values of 0.92 for both models. As a result, the suggested statistical models can provide complete geographic and mechanical explanations for the crack sources in the investigated residential complex.