Non-Linear Finite Element Analyses of Reinforced Concrete with Large Scale Elements: Including a Case Study of a Structural Wall

Abstract

The accuracy of large scaled elements in nonlinear finite element analyses (NLFEA) ofreinforced concrete is investigated. A much used finite element analysis design procedureused to design large offshore concrete structures is presented and suggestions forthe utilization of NLFEA in the process is given. Means to obtain effective use of NLFEAare discussed and the importance of large elements to minimize the computational costis stressed. The use of large elements is investigated through a case study of a structuralwall. The wall is analyzed using medium scaled elements that should be able to predictthe behavior well, and by use of large elements. Both analysis results are compared withexperimental results. The finite element models are created as they would in a designsituation and the analyses are conducted without tweaking of the material parameters.State-of-the-art material models are chosen that accurately describe the most importantmaterial characteristics of reinforced concrete. A short presentation of the Smearedcrack approach for finite element modeling of concrete is given. Both a fixed and arotating crack model is used.The results of the analyses reveal a poorly predicted ultimate load carrying capacityof the structural wall and it is found that the missing inclusion of the volumetric expansioneffect of concrete is the main cause. No significant difference is found betweenthe fixed and the rotating crack models, a finding that is attributed to the fact that norotation of the stress field in the structural wall is seen before the peak load. A higherload capacity is found when using large scale elements. It is believed that the highercapacity is caused by the inability of the large elements to pick up the most extremecompressive stresses in the compressive zone. Thus the failure of the compressive zoneis delayed and a higher load capacity is found. In the first phase of the deformationalresponse the predicted behavior of the wall in the large scale analyses is found to bestiffer due to non-localized cracks, as compared to the medium scale analyses where alocalized crack pattern is observed. When the crack pattern is progressed the differenceis negligible. It is found that the all-over prediction of the wall behavior in the largescale analyses is good. The stiffness for high load levels is very accurate and the failuremode is correct. If part of a larger structure the predicted behavior of the wall wouldprobably allow for finding of the correct distribution of forces in the structure and thecause and progression of failure. Thus the use of large elements is found to be possiblein a design situation. However, a lot of work remains to be done before such use canbe done confidently. Suggestions for further work are given.