Ambitions in exploration of oil and gas fields at deeper water depth require continuous investigation and maintenance. The transportation pipelines laid in deep waters are both subjected to corrosion and buckling due to environmental phenomena. They may also often undergo branching (namely hot tapping) to redirect (or add to) the transportation paths. Mechanical joints and welding are both considered as available alternatives when sectioning and replacement of the pipes at shallow waters is necessary, yet, welding is more promising for deep waters where remote operation is central.
Fusion welding on the other hand comprises several technological detractions for sound operations under high ambient pressures disregarding its low cost and flexibility. The foremost detracting phenomenon in the arc welding is called ‘arc root constriction’, which is defined as arc geometry shrinkage under the increased pressure. Consequently, the power delivery to the weld pool at different pressure levels is a major worry.
Effects of ionization and dissociation energies of different gases and mixtures, partial pressure of environmental gases including hydrogen and oxygen, gasification and degasification of the weld metal, inclusions that affect the phase transformation, absorption and desorption kinetics, oxidation and deoxidation reactions and many more are the phenomena that can possibly be altered by the gas type and ambient pressure level.
Spattering and fume generation is a problematic issue since the arc is rather unstable under high pressure. Thus, seeking the effect of different chamber gas mixtures on welding parameters, final microstructure and mechanical properties is the main objective of this work.
Statistical analysis of the collected voltage and current waveforms is carried out to identify the source of arc misbehavior and instability (discussed in Paper I). The stochastic parameters is related to the electrical stability and resolved into a number of varying welding parameters. The datasets are representing the effects of using pure argon under 14 incrementally increased pressure levels. Fast Fourier Transformation (FFT) is used to characterize the frequency domain of the waveforms. Auto-correlation Function (AF) and Power Spectral Density (PSD) were calculated assuming the Wiener-Khinchin theorem. Considering the AF, it is possible to visualize the deteriorating stability of the arc. The rate of stability degradation is quite gentle after 20 bar, though, huge differences were observed from 1 to 20 bar. The characteristic frequencies of 100-150 Hz and 350-400 Hz were observed. The first range can be associated with the mass transfer or molten droplet launch frequency and the latter range is representative of the rectified mains. The spread of large low-frequency peaks at higher pressures is illustrating the mass transfer deterioration. The aforementioned peaks were found above 125 bar where the range of the characteristic frequency peaks in voltage and current waveforms started to deviate. The calculated arc power is higher at high-pressure range while the weld bead geometry was barely varied. It implies that the arc efficiency factor decreases at high pressures.
The heat source dimensions and heat efficiency factor are two major inputs for finite element (FE) simulations of the weld. However, a systematic classification of these factors was hardly available prior to this work. Additionally, to the best of author’s knowledge, the direct high-speed observation of the arc inside the hyperbaric welding chamber has not been investigated in detail by far due to several technological issues. The varying bead-on-plate welds including the end crater appearance can possibly be good candidates to categorize the FE heat source dimensions. Double-ellipsoidal heat source (Goldak’s Model) was implemented in WeldSimS® FE code that is used in this study. Since the model incorporates two superimposed reference heat sources, the amount of dissipated heat from each source should be differentiated. An intermediate heat source model was employed for this purpose. The latter model is after Myhr and Grong that is called distributed point heat sources. This model can be accurately fit to the weld cross section geometry if calibrated accurately. The calibrated parameters were found to be very close to the ones required by Double-ellipsoidal heat source model. By using this approach, not only the effect of welding parameters on weld bead geometry can be categorized, but also the spent time for double-ellipsoidal heat source adjustments will be cut by 90%.
A Gaussian heat source was also employed for welding thermal cycle simulations. Accompanying experiments suggested that the thermal gradients hardly change as pressure elevates. However, it was found that the increased pressure level might not necessarily result in higher or lower cooling rates despite the geometrical changes.
In a parallel investigation, the metallurgical effect of different shielding environments on phase transformation and mechanical properties of the bead-onplate weld samples was studies. Electron backscattered diffraction (EBSD) and orientation imaging microscopy (OIM) techniques were used to identify the effect of five different shielding environments on the phase transformation. Argon and Helium chamber gases offer the conditions that facilitate the highest amount of acicular ferrite transformation, yet, they show some differences in a number of crystallographic details. CO2 gas provided conditions for a lot of porosity in addition to the dominant polygonal ferrite/bainite transformation. He+½CO2 mixture resulted in bainite transformation that was found to follow the maximum heat flow direction in terms of crystallographic orientations.
Very small sized tensile and single-edged notch bending (SENB) samples were machined from the weld metal material. The tests revealed that the best mechanical properties are associated with the He chamber gas and the poorest properties were presented by the samples welded in He+½CO2 shielding environment. It was also observed that there is a good correlation between the acquired acoustic signals and the fracture properties of the weld samples