Engineering Mechanics Institute Conference 2015

Full Program »

Experimental and analytical investigations of structural effects of concrete creep under varying load histories

Creep under a sustained load is a well-studied phenomenon. However, many structures are subjected to load changes at young age. In common practice the actual consideration of the load history, e.g. due to the construction process, is often omitted in the design process. Instead a constant load is assumed to predict the long-term behaviour of concrete elements. Furthermore, the time-dependent response in case of unloading differs from the response caused by a load increment, as only part of the creep deformation can be recovered, depending on the loading time. In order to investigate the creep deflection under load changes (including load decrements) and the long-term deflection of concrete beams subjected to a load history, four T-shaped reinforced concrete beams have been casted in the Magnel Laboratory for Concrete Research. In addition to these experiments, other data sets of previous experiments are incorporated in this investigation.
Time-dependent behaviour of concrete structures can be modeled by viscoelastic integral equations, which in most cases cannot be solved in a closed form. Hence, numerical step-by-step methods, in which the viscoelastic integrals are approximated with finite sums, are suggested in literature. These methods can be combined with finite element analysis by integrating such viscoelastic equations in the stiffness matrix. However, this still requires the storage of the stress history in each element. Even with current computer capacities, such methods are inefficient. The use of Dirichlet series to approximate the viscoelastic creep function can provide a solution to overcome this problem. This has the advantage that the integral equations can be solved in a closed form and that the solution is recurrent, meaning that only the previous solution step in needed in the analysis. A numerical procedure, as described, is developed to predict the available test results.

Author(s):

Pieterjan Criel    
Ghent University
Belgium

Robby Caspeele    
Ghent University
Belgium

Stijn Matthys    
Ghent University
Belgium

Luc Taerwe    
Ghent University
Belgium

 

Powered by OpenConf®
Copyright ©2002-2014 Zakon Group LLC