Samenvatting

Finite element analysis on large structures in the mining industry is mainly set up with beamelements.
During the modelling of the structures a couple of assumptions are made to ensure that the structure is not too complex. The connections between the beams are usually released in a couple of direction to simulate the bolted connection used.
However, questions remain regarding the accuracy of the assumption to end-release the beams. The purpose of this study is to compare the accuracy of the Finite Element Analysis (FEA) models against measurements on a fabricated structure. For the fabricated structure a simple column was used with a beam connected to the top of the column with an endplate and four bolts. The FEA models were set up in beam- and plate-elements and calculated the same natural frequencies.
The measurements on the fabricated structure returned lower frequencies. Adjusting the FEA-models improved the accuracy of the calculated natural frequency. The difference between the measured and the calculated natural frequencies is due to the way the structure was anchored to a concrete slab and the gusset used on the column.
This concludes the assumption that fixed nodes have a high impact on the natural frequency and makes the simple structure used too complex. Plate-element models give higher compared to beamelement models and can be used to determine the correct natural frequency. However, plate-element models take significantly more time to model and calculate. Furthermore it is now known that the FEA calculations are stiffer than structure.
Advice given to PBA Projects is to continue the research on a simply supported beam. The beam should not be too long to keep the influence of the vibrations of the baseplates on the beam to minimum