Using Statistical Regression Analysis to Establish Construction Cost Models for Seismic Retrofit of Confined Masonry Buildings

Abstract

The presented research deals with the development of statistical models for predicting seismic retrofit net construction cost (RNCC) of confined masonry (CM) buildings. Real data from 183 CM school buildings in Iran were collected to establish the RNCC and its determinant variables, among which two were unprecedented in the literature: (1) mortar quality, and (2) the concrete quality of confinement elements. Parametric models were developed using the stepwise regression technique, and the leave-one-out cross-validation technique was utilized to examine the predictive performance of these models. Four variables were selected that collectively best predicted the RNCC variation: (1) total floor area, (2) seismic weight indicator, (3) floor and roof diaphragm type, and (4) mortar quality. On the basis of this selection, similarities and discrepancies between the RNCC prediction of framed and CM structures were described, with dissimilarities being postulated to arise from different seismic evaluation and strengthening approaches practiced for these structures. Finally, the importance of the total floor area for the RNCC prediction of CM buildings was highlighted, with the double-log cost-area model being recommended for making this prediction at an early design stage of seismic retrofitting.