Truss vs solid modeling of tendons in prestressed concrete structures: Consequences on mechanical capacity of a Representative Structural Volume

Abstract

a b s t r a c tA comparative study is carried out on the mechanical behavior of a Representative Structural Volume of a prestressed concrete structure, using the state of the art of the numerical tools for engineers (and not the most recent developments of the academic research). The structural effect due to the presence of the tendons is carefully studied by comparing a classical truss finite element computation with a more sophisticated approach where the tendons are explicitly represented using a solid meshing. The classical approach based on truss elements appears to be poor to reliably represent the failure mode of concrete. Therefore, in some cases, it is essential to take into account both material (difference between steel and concrete stiffness) and geometrical heterogeneities through the effect of the inclusions. The well-known mesh dependency problem is also studied in this paper. For the truss simulations, contrary to the usual conclusions on bending beams for example, the mesh dependency does not only affect the position of the localized bands but also the global failure mode, questioning once again, for this particular application, the representativity of local simulations with truss elements. Finally, nonlocal calculations are carried out using the integral approach. Contrary to the well-documented simulations on bending beams, the value of the characteristic length does not only affect the width of the localized bands but also the global failure mode.

Mixed dimensional modeling of reinforced structures

“…Besides allowing to use coarser meshes around the reinforcements, using bar elements simplifies the meshing procedure since the volumic and lineic meshes can be independent. This method allows to get good global results (force-displacement curve for instance), but is obviously not as accurate as a volumic representation of the reinforcements as shown by [5]. Indeed, two drawbacks can be underlined:…”

“…The intersection of Γ with the edges of mesh can be easily computed using the linear interpolation on the triangular element (cf. equation (5)). This implies that the quality of the geometries discretized using level-set functions will depend on their curvature (the less the curvature, the better the approximation).…”

“…These approaches give good global results, but are not designed to get accurate local results around the reinforcements, as shown in [5] for instance. To sum up, there are two possibilities to model the reinforcements: using a 3D volumic mesh, which would give the most accurate results but is incompatible with industrial studies because of the too large computing costs, or using a smeared or a 1D representation of the reinforcements, decreasing the computing costs but leading to inaccurate local results.…”

Modelling Strategies of Prestressing Tendons and Reinforcement Bars in Concrete Structures

“…Therefore, these reinforcements are typically modeled in FE analyses using 1D elements, with beam or truss kinematics [4]. In some cases, where the three-dimensional effects of the tendon or rebar section have to be taken into account, full 3D models using conform meshes can be applied [8].…”

“…It consists in a numerical model of a representative structural volume (RSV) of a prestressed and reinforced concrete containment building, on which a pressure test is performed. This particular structure is strongly dependent on the prestressing tendons, which affect damage localization and failure mode [8]. In particular, previous studies have demonstrated an inclusion effect around the vertical tendon.…”