| dc.description.abstract | This thesis presents both experimental and numerical investigations into some
of the plethora of parameters influencing pipeline impact behaviour and the
potential fracture arising thereof. The seamless pipes studied are made from
an X65 offshore steel widely used by the industry. A succinct review of today’s
valid design codes (and analytical or empirical methods) for problems similar
to this one is presented.
Impact tests against empty and water-filled pipes at different velocities were
carried out to simulate a collision by trawl gear or an anchor. Subsequent
straightening of the pipe representing a rebound after the impact revealed that
fracture always presented itself given sufficient stretching (especially if the initial
impact velocity had been at the high end of the spectrum). In some cases fracture
(not necessarily visible on the surface) was present even after impact only,
where in a metallurgical examination cleavage fracture surfaces were observed in
the initially ductile material. Equivalent quasi-static three-point bending tests
showed no such signs of fracture initiation, meaning that the problem being
dynamic is an important factor and the cracking most likely initiates during the
rebound after the impact.
Through quasi-static uniaxial tensile tests the X65 steel was characterised as
isotropic and homogeneous across the cross-section, with kinematic hardening
being present in the material. Testing at elevated strain rates showed that viscous
effects caused the flow stress to increase, while the fracture strain remained
as for the quasi-static tests. As very high compressive strains were observed in
the component tests prior to fracture, a material test using notched specimens
was contrived to investigate this further. Specimens were compressed to various
levels of plastic strain before being stretched to failure in tension. The tests
showed that when the preceding compression increased, the strain to fracture
in the following tensile step decreased. Metallurgical studies showed more shallow
pores in the compressed specimens and cracked particles were prominent
sights, both being indications of earlier onset of fracture. Cleavage fracture was
observed in both the material and component tests where large compression
preceded tension.
Quasi-static stretch-bending experiments indicated that adding an axial tensile
load to the pipe increased its resistance to bending. The same tests were repeated
with the addition of internal pressure, which provided further resistance
to bending while changing the local deformation significantly.
Finite element simulations of the impact were in general very accurate, whereas
in the stretch phase the force was typically overestimated. This was mainly
caused by fracture being inadequately described by the numerical model, thereby
offering more resistance to straightening compared with the experiments. Fully
coupled fluid-structure interaction simulations of the impact against water-filled
pipes were also completed with satisfying accuracy, employing a variety of different
techniques. The effect of submerging the pipe in water was investigated
numerically.
Unit cell simulations with constant triaxiality were used to investigate the fracture
mechanisms related to a load cycle of large compression before tension.
Results indicated that increasing compression led to an accelerated void growth
during tension, but the onset of coalescence appeared to be delayed, contrary
to the experimental data. Very high local stresses after compression and load
reversal indicate what might initiate cleavage fracture, so making use of a stress
based fracture criterion is a natural progression from this thesis. In tension only,
the unit cell simulations gave good predictions of the fracture strain given that
the triaxiality remained fairly constant in the tests (i.e. notched tests).
In summary the global response of the numerical simulations was very accurate,
whereas a small scale phenomenon as initiation of fracture occurs on a
scale much smaller than the element size in these global models and is as such
not represented with sufficient accuracy and other approaches are needed. Experiments
in general, and the advent of the technologies like scanning electron
microscopy, are of paramount importance for understanding the physical processes
at hand. | nb_NO |
