| dc.description.abstract | Drill strings are long slim rods used in well drilling to transmit torque and weight to the drill bit from the surface. As a consequence, these narrow structures experience various vibrations that could damage the well system. Since 1960, there have been an increased interest in modeling and analysis of drill string dynamics. Common practice had previously focused on eliminating vibrations by adjusting structural material rather than inspecting the underlying cause for the vibrations.
Today there exists a considerable amount of models for the drill string. Because of the complexity involved with modeling, two models are rarely alike. The methods and assumptions derived can deviate greatly from one another, subsequently presenting considerable different results for a seemingly similar starting point.
Vibrations are divided into axial, torsional and lateral vibrations. This thesis considers the effects of axial vibrations only, and aims to model them with a computational efficient model. If the model is fast and accurate enough, it will be suitable for real-time estimation, and could potentially be a step towards fully automated drilling rigs.
Two model approaches are presented, one semi-analytic and one numerical. The semi-analytic model assumes Newtonian drilling fluid, and includes basset force and added mass. The numerical model is developed as an attempt to incorporate non-Newtonian drilling fluids, as well as an alternative to the semi-analytical.
The numerical model uses the Finite Element Method to calculate the flow profile for Newtonian and non-Newtonian fluid in an annulus, but missing features that are present in the semi-analytical model include pressure and amplitude estimation at arbitrary points in the drill string.
A physical experiment was supposed to take place during the thesis. This experiment would compare measurements to simulation estimations, but was unfortunately delayed to the Summer of 2017. Due to the unforeseen delay, changes were made to the original problem formulation which were more time-consuming than first anticipated. The results of the numerical model is therefore incomplete, but provides a basis for further research. | en |
