Experimental and Numerical Study on Perforated Steel Plates Subjected to Blast Loading
MetadataVis full innførsel
The main objective of this thesis was to determine how thin, perforated Docol 600 DL steel plates behave under blast loading, and to validate to which extent the response may be predicted using different numerical techniques. Experiments were performed in the shock tube facility at Structural Impact Laboratory (SIMLab), Norwegian University of Science and Technology (NTNU), ensuring a controlled environment for the blast load. Plates with and without perforations were tested and the dimensions of the blast exposed area were 300 x 300 mm2 with thickness 0.8 mm. The perforated plates had four quadratic holes with dimensions 60 x 60 mm2. Both plate types were tested at three different firing pressures to vary the magnitude of the blast load. Three-dimensional digital image correlation (3D-DIC) was used to measure displacements and strains of the steel plates. The experiments showed clear signs of fluid-structure interaction (FSI) effects, especially for perforated plates. The perforations proved to weaken the structural capacity of the blast-loaded plates, and complete failure was only induced for the perforated plates. Numerical simulations in Abaqus/Explicit were performed with Lagrangian, Eulerian, uncoupled Eulerian-Lagrangian (UEL), and coupled Eulerian-Lagrangian (CEL) approaches. A preliminary numerical study was performed on the pure Lagrangian and Eulerian approaches to investigate parameters such as mesh size, boundary conditions, adiabatic heating, applied load, etc. The Lagrangian model proved to be easily established, and variations of most parameters did not affect the solution considerably. However, the preliminary Eulerian study did not result in an overall acceptable model, thus modifications had to be done. These modification were mainly employed to counteract errors from simplifications in the preliminary model. After the basic models were established, simulations with different computational methods were performed for both plate types. The Lagrangian approach was found to be the easiest to use and was the cheapest method with respect to computational cost. The UEL and CEL approaches required more effort to use and the computational costs were considerably higher than for the Lagrangian simulations. To be able to capture FSI effects the CEL method had to be employed, and even though the results of these simulations varied in quality, they did prove that FSI effects can be reproduced numerically in Abaqus/Explicit. All methods proved to have different pros and cons, but the overall accuracy was acceptable for all methods. The most challenging task proved to be generating accurate blast loads through Eulerian analyses and this was reflected in the results of both the UEL and CEL methods.