Investigation of interface modelling techniques using finite element analysis with ATENA

Abstract

Study of interface behavior has been the primary area of focus for researchers in the field of concrete science and technology. This interface between two concrete elements becomes important for prefabricated systems, combination of precast and in-situ concrete elements, combinations of concrete cast at different times, repairs of existing concrete structures, strengthening of a structural element such as beam or slab and so on. In this research an attempt is made to check if the interface behavior is reliably predicted using finite element analysis in ATENA with the help of standard elements available in the software. Validation of the numerical results is performed by simulating an experimental bond test and structural test and conclusions are drawn based on the response of these numerical models.

Strength of the interface is mainly governed by the roughness of the adjacent surfaces. In the literature study, a chronological overview of the development of shear models is presented that make use of these roughness parameters in calculating shear capacity of the interface. A finite element analysis is performed using standard elements available in ATENA. ATENA interface material model (ATENA IMM), present in the software uses 2D line interface elements having zero thickness to model an interface. Roughness parameters assigned to this model are based on the guidelines proposed in different design codes. Many parameters need to be defined in this interface that are difficult to measure accurately from experiments. For this, a 5 mm thick artificial interface layer is created by using 2D linear quadrilateral material elements. This technique is named as artificial interface model (ArtIM) that uses physical material properties to define the interface. Since, cohesion and friction coefficient parameters cannot be specifically defined in ArtIM unlike in ATENA IMM, an explicit roughness is incorporated by designing the interface layer in a wave pattern with certain wavelength and amplitude depending on the different surface roughness classes as defined in the Model Code 2010.

A check is performed for the assigned input parameters for the two interface modelling techniques, by simulating bond tests (direct tension test and shear load test) on a small scale composite concrete specimen. Furthermore, validation of the two interface modelling techniques is carried out by comparing the numerical results to experimental findings for a bond test performed by T. Paulay, R. Park and M. H. Phillips [1] and for a structural test case of the Eindhoven airport car park garage failure [2], [3]. ATENA IMM rightly predicts the initial response of the interface but does not comply with the experimental results once the interface fails. With the use of ArtIM, an overestimation of the shear strength is exhibited and the conventional ratio of 2 between the shear and tensile strength [4], [5] cannot be obtained. Moreover, once the material in the interface layer fails, a brittle behavior is exhibited pertaining to the concrete properties assigned to the interface material layer.

Different reinforcement modelling techniques are studied using ATENA IMM with a very low bond strength (no bond condition). The default 1-D reinforcement (RF) bar element when modelled perpendicular to the interface does not reliably predict the response of loading. Hence, other RF modelling techniques are discussed among which, 1-D RF bar elements modelled in cross pattern reliably predict the initial stiffness of the RF bars. The inclination of the RF bars was tested for different angles to comply with the experimental results. However, further research is required to calculate appropriate inclination of the RF bars. In this case the optimum angle between the cross RF bars and interface is around 80°.

In the structural test, interface behavior is investigated for a specimen with (fully reinforced) and without (partially reinforced) shear reinforcement in the flexural span. ATENA IMM safely predicts the bond strength in case of both, partially reinforced and fully reinforced models. However, by using ArtIM, an overestimation of the interface strength is observed. The important aspect in the structural model is the shear strength of the interface and since, while using an artificial material element for the interface the shear strength obtained is almost 40% higher than the experimental results, the ultimate load carrying capacity of the whole model increases manifold.

After investigating the two interface modelling techniques, on bond level and structural level, it can be concluded that, the ATENA interface material model (IMM) as well as the artificial interface model (ArtIM) do not predict the interface behavior reliably. However, by using ATENA IMM a conservative response is obtained as opposed to the ArtIM which overestimates the shear capacity of the interface.