CFRP-patch sizing for strengthening perforated steel tubular structures

Abstract

The use of statistical and non-parameter-based modeling approaches to size a CFRP-patch thickness to restore the intact compressive capacity of perforated steel tubular structures is examined in this paper. Firstly, the effect of the perforation is assessed using non-linear FEM simulations of full-scale tubular structures with multiple levels of slenderness and cutout sizes. Secondly, the contribution of the CFRP-patch to the strengthening of damaged structures is evaluated using a full-factorial design-of-experiment of FEM-models that vary both the ultimate compressive strength and the elastic modulus of the CFRP properties in the longitudinal direction to properly populate a database with the models’ responses. Thirdly, a dataset with thicknesses that restore each damage case's observations to their intact capacity is created. This was accomplished using an ensemble learning method applied to the responses database to estimate the proper CFRP-patch thickness while considering a Hashin's damage criterion level. Finally, a multilinear regression methodology is adopted to describe the dataset, based on the sample and CFRP-patch explanatory variables as well as the Akaike information criterion to fit a model of response surface for the suggested thicknesses. The response predictions from simulated samples with suggested thicknesses are then compared with the ones retrieved from their respective intact models to assess the fitted model's level of effectiveness. Responses from strengthened samples are comparable to those obtained from intact samples, indicating that the fitted model for inferring the CFRP-patch thickness can suggest suitable values of thickness to recover the intact capacity of the samples with ±3 % deviation range.