A Finite Element Simulation of a Single Shear Lap Joint Made of Carbon Fiber Composites
Publish date: 2014-01-27
Report number: FOI-R--3771--SE
Pages: 25
Written in: English
Keywords:
- Joint
- single shear
- carbon fiber
- bolt
- tension loading
- secondary bending
- contact
- conditions
- finite element calculation
- mesh generation
Abstract
In order to see whether it is possible to simulate an experimental study on the failure load of a single shear lap joint of carbon reinforced plastic finite element (FE) calculations have been performed. The calculations do not follow the experiment completely since they only considers the elastic part of the experiment. The calculations show that it is possible to simulate the elastic process, but further refinement is required to get the right stiffness, and to get the fastener deformation to conform better with the experiment. The calculations provide information on an estimate of the time an analysis of the entire process may take (ie until the fastener creates a hole bearing pressure such that the carbon fibers are crushed). To take the carbon fiber damage into account in the FE calculation requires that some type of damage model is applied. The finite element program used is ABACUS. Unfortunately, there are no similar problems in ABACUS collection of examples and hence it becomes difficult to get a good input file. It is likely that the problem can be solved with ABACUS, but it requires detailed knowledge of the software. The problem to be solved contains many contact conditions with friction, preload of fasteners and a damage model. The first two calculations were stopped, by the ABACUS program, before they reached the final goal due to lack of convergence. The second last computation converged, but required almost 1500 time steps. The last calculation only needed 11 time steps, and although the calculation contained almost 1.5 million more degrees of freedom, the overall execution time was significantly faster. This means that it is probably possible to solve the complete problem (with a damage model and more accurate grading in the thickness direction) on fairly large CPU clusters within a reasonable time.