Development of a multiphase flow simulator for drilling applications

Drilling, or gas kick, simulators are becoming prevalent in industry due to their ability to replicate wellbore conditions that are not feasible in a laboratory setting. This is becoming more desirable as deeper wells are being explored. One of the biggest dangers that could happen during drilling operations is the onset of a gas kick. This occurs when a zone in the formation whose pressure is higher than that of the wellbore is breached. This allows for the undesired influx of formation fluids into the wellbore. If left uncontrolled, it could develop into a blowout.

Gas kick simulators allow for testing of procedures that could be used to contain kicks at such depths. Furthermore, the use of drilling simulators could provide more insight into other phenomena. These include wellbore breathing and fracture ballooning, that cause similar kick symptoms at the surface and lead to expensive misdiagnosis, and the dissolution of gas into oil based mud, which could delay the identification of a kick. This thesis investigates the development of the initial integration of a drilling simulator into UTWELL, the wellbore simulator program developed at The University of Texas at Austin, by implementing a gas kick module.

The transport equations of mass and momentum conservation were discretized using a Semi-Implicit Homogeneous Method over a one dimensional staggered grid. The multiphase phenomena were modelled using a Drift Flux approach as opposed to a mechanistic, Two Fluid approach. This was due to increased stability of the solution and faster computation time, despite the risk of loosing accuracy.

The simulator was successful at simulating single phase flows for fluids with distinct rheology models, and with wellbores with discontinuities in the geometry. When attempting to simulate the well control of a gas kick in water based mud, the results were mixed. Attempt at simulating a `Floating Mud Cap' method failed due to the simulator's inability to perform drainage functions that allow for the raising and lowering of the mud level in the wellbore. However, the simulator was successful at capturing the behaviour of the gas kick as it entered and migrated through the wellbore, matching literature results.

The simulator was compared to experimental data gathered from a test well. Three different scenarios were tested: No Drillstring, Semi-Submerged Drillstring and Drillstring at the Bottom. In all three cases, there was a good match between the experimental and simulation results for the bottomhole and choke pressures. The pit gain was severely overestimated in the 'No Drillstring' and 'Semi-Submerged Drillstring Case', however this was due to a higher influx of simulated gas having entered the wellbore during simulations. The 'Drillstring at the Bottom' simulation matched well with all data and with other simulators.

Recommendations included full integration and testing of a compositional model to simulate oil based mud cases, implementation of automatic choke control and special flux splitting techniques in the discretization in order to better handle pressure waves caused by discontinuities.