The main objective of this research is to understand the development of arch action in a single T-beam acting as part of the bridge system, ignoring the distribution of the load in transverse direction. As the beam is loaded, several mechanisms work simultaneously in the bridge system, resulting in enhanced ultimate load bearing capacity of the bridges, like compressive membrane action (CMA) in deck slabs and arch action in concrete beams. For this research, an approximate analytical model for quantification of arch action in underwater concrete slabs loaded with uniformly distributed loads (suggested in CUR-077) is verified using non-linear finite element analysis for varying span-to-depth ratios, stiffness of lateral restraint and initial prestressing in the system. The adopted analytical model seems to be able to conservatively predict the arching capacity (within 12%), horizontal stretch (within 10%) and membrane forces (within 10%) in concrete members when compared to the numerical models, provided that the slenderness is less than 15 and the stiffness of end-restraint is at least equal to the stiffness of restrained member. The verified analytical model is then extended to beams loaded with concentrated loads and within the central half of the span, the adapted model is able to conservatively predict the arching capacity with an accuracy of at least 15%. The analytical model is then further extended to beams with T-shaped cross-sections for uniform and concentrated loads. These models are able to predict the arching capacity in T-beams with an accuracy of 7.5% when the numerical failure is due to crushing of concrete. In T-beams with thin webs, the strut failure is observed and the adapted models are not able to predict the arching capacities. As a case-study the Vechtbrug beam is modeled using 2D, 2.5D and 3D approaches in DIANA and the models are validated using the experimental work done by Ensink as part of his PhD studies. All the models show comparable load-deformation behavior and peak loads (within 7%) but only the 3D model is able to simulate the crack pattern observed during experiments. The validated numerical model of the Vechtbrug beam is then adapted as though it is connected to the bridge through cross-beams by applying full restraint at the edge faces of the cross-beams in longitudinal direction. The arching behavior of the adapted Vechtbrug beam model is compared with a disjointed bridge model developed by Ensink in which the distribution of the load is prevented in transverse direction by disconnecting the slab of the loaded sub-span with neighboring beams. Applying full restraints at the edge faces of cross-beams is found to overestimate the influence of arch action in the loaded sub-span when compared to the disjointed bridge model. The adapted analytical model for arching in T-beams with point loads is applied on the loaded sub-span but is found unable to conservatively predict the arching capacity owing to the thin web of the Vechtbrug beam causing strut failure, which is not taken into account by the analytical model.