Smart system application for capacity estimation of reinforced concrete elements reinforced with fibrous polymer

Reinforced concrete, which has been used as a composite material by using concrete and reinforced concrete reinforcement together for a century, is now an indispensable type of building material in contemporary civil engineering approaches with its superior properties. It is essential for a building design that the reinforced concrete has sufficient strength, ductility and rigidity for all section forces and deformation demands that arise due to time-dependent reasons under the influence of internal and external loads in the load-bearing system elements. This obligation becomes even more important for structural engineers, especially for a country like Turkey, which is in an active earthquake zone. In reinforced concrete structures produced today; They are damaged or collapse altogether under their loads and under the cross-sectional effects that occur in the event of an earthquake (shearing, bending, torsion and their combinations) due to architectural concerns, mistakes made in calculations, errors occurring during application, non-standard materials, user errors and time-related deformation. . When such structures are examined, it is seen that there is a major problem in terms of capacity in the critical areas of the elements in terms of shear (areas close to the column-beam connection points). Due to insufficient sections resulting from incomplete engineering services at these points, damages under earthquakes and similar loads negatively affect the entire structure and such structures often need to be strengthened. Today, among the many strengthening methods for the insufficient capacities of reinforced concrete structures and their components, the "Strengthening with Fibrous Polymers" method has been widely preferred by technical staff in recent years. While many assumptions are made regarding the behavior of concrete and reinforced concrete, which already have a complex structure, the contribution of fibrous polymers, which are generally bonded to reinforced concrete with epoxy and in some cases fixed with fibrous polymers or mechanical anchors, to the cross-section capacity remains unknown today. Based on this motivation, in this study, a data pool was created by scanning all current articles published in the last twenty years for an intelligent system-based software application that can largely predict the capacity values of all kinds of reinforced concrete elements reinforced with fibrous polymers. In this context, a large number of experimental data made for different purposes in the literature were compiled and divided into groups, and an idea was obtained about the estimation of element capacities with different algorithms for each data. In this data pool; beams under the influence of shear, bending and torsion; Columns under the influence of column moment and axial load; The bearing capacities of slabs under the influence of bending and punching and of walls under the influence of axial load were estimated using MLR, Decision Tree, Bagging-Dt., Random Forest, SVR, XGBoost, AdaBoost, GradientBoost machine learning methods. The XG-Boost algorithm gave the best prediction results in beam shear, slab bending, and slab punching strength studies. While the Bagging-DT algorithm gave the best prediction result in the study of beam torsional strength, column moment carrying capacity and column load carrying capacity, the AdaBoost algorithm was more successful than the others in estimating the wall load carrying capacity and the DVR algorithm was more successful than the others in estimating the beam bending capacity. The results obtained also show that there is variability in the success of the algorithms depending on the type of element studied and the strength feature desired to be obtained, and that the performance of the algorithm varies depending on the output data examined, the parameters considered in the data group, the number of data and the data range.