Alagappan Ponnalagu

Most constitutive relations that are used to capture the mechanical response of concrete are based on the assumption that concrete is homogeneous. However, it is a well-known fact that concrete has three distinct regions, namely - aggregate, mortar and Interface Transition Zone (ITZ). One of the reasons for assuming that concrete is a homogeneous material is to simplify the problem for finite element analysis (FEA). The damage behaviour of concrete depends upon the interaction of these three regions. Hence it is necessary to have a clear demarcation of the three regions for discretization and further analysis. It is well known that damage is initiated in the ITZ, which is a weak, porous and heterogeneous region of cement paste around aggregates whose thickness ranges from 9–50 µm. Due to the vast scale difference in the ITZ with aggregate, the ITZ of concrete cross-section is either computer-generated or manually inserted. While generated images are very limited by the algorithms/procedures employed, manual insertion of ITZs on real images takes time and is prone to certain uncertainties. An algorithm that processes the images of concrete for more reliable identification of the location of the ITZ and provides control on its thickness, colour and the minimum size of the aggregates to be included can drastically reduce human-induced error and enable faster and more reliable processing of existing images. In this work, an image processing algorithm is developed using OpenCV and NumPy, which are open source libraries in Python. The concrete image is processed by multiple means like log transformation, erosion, dilation, bilateral filtering and adaptive gaussian thresholding, which significantly improve the identification of different regions in concrete which further enhances appropriate FEM meshing. A contour feature extraction tool called canny edge detection is used to identify the aggregate and to draw the ITZ. The damage predicted through the FEM analysis of the problem domain that is processed by the proposed algorithms is validated by comparing it with the experimentally obtained damage patterns. The proposed algorithm performs better on computer-generated images than the images of actual concrete cross-sections. The accuracy of this algorithm on computer-generated images is over 75%, and it achieves over 90% accuracy on real images. The resulting image is also comparable to images that are computer programmed to have ITZs. Our algorithm enhances the accuracy of FEM analysis of images through the inclusion of ITZ and enhancement of the features of the image.

Bridges are key components of transportation network, especially in strategic border areas in a country, and consequently are susceptible to subversive blast attacks. Hence in this study, dynamic response of a reinforced concrete (RC) bridge (single span) consisting of a deck slab supported on longitudinal girders along with transverse ones placed symmetrically has been numerically investigated when subjected to blast loading using ABAQUS/CAE 2020. The effects of an explosive charge of 226.8 kg (TNT) at 1 m standoff distance have been analyzed using the CONWEP algorithm. Three different locations of the bursting charge along the cross section at mid span of the bridge above the deck, such as on the central girder, between two adjacent longitudinal girders, and on the cantilever part, have been considered. Concrete damage distribution in terms of concrete spalling and cracking has been studied with concrete damage plasticity (CDP) model. Also, the response in terms of damage dissipation energy, maximum displacements, and stresses has been compared for the blast scenarios. Furthermore, AASHTO: LRFD Bridge Design Specifications (2017) provisions have been used to compare obtained maximum displacement values.