Sammendrag
A kick entering a drilling riser and expanding upwards uncontrolled can lead to severe consequences such as riser unloading and riser collapse, and in the worst case a blowout scenario may evolve. If the riser is filled with water-based mud, the kick will normally migrate on its own to surface, but it has been observed both in small scale experiments and in field tests that small amounts of gas are trapped by the mud during the kick migration. In some cases, the kick is not able to reach the surface without additional circulation. Hence, a certain kick size may or may not lead to an unloading scenario depending on the effect of gas suspension in the drilling fluid.
In this paper, two different modelling approaches for describing the unloading scenario will be compared and the differences will be highlighted. In the first approach, the single bubble model will be considered. Here the gas bubble is assumed to occupy the whole cross-sectional area, and it is fully separated from the mud regions. This will be solved by two different calculations methods, one that is taken from literature and one that is based on a shooting technique. The second and recommended approach is to use a transient drift flux model, which includes friction, acceleration terms, and gas slip. For the gas slippage model, different flow patterns will be accounted for, as will the suspension effect that causes small amounts of gas to be trapped by the mud. The drift flux model will be solved numerically using the explicit AUSMV scheme.
The impact of gas suspension will be studied by varying the onset for gas suspension to determine from simulations whether a riser will be unloaded or the kick become fully trapped in the riser. In addition, a sensitivity analysis will be presented where kick size, riser ID and riser length are varied to determine when the riser will be unloaded. The different simulations presented solves physical equations of the unloading scenario to calculate pressure at BOP, displaced mud volume (pit gain), liquid mass in well, surface rates, riser friction, and depth profiles of the gas distribution at a certain time. the tables provide a comprehensive overview of which combinations of parameters lead to a trapped gas scenario or and which lead to unloading the riser.
It is shown that a fully transient drift flux model can cover a vast range of different situations e.g. gas becomes fully trapped in the riser, the riser becomes fully unloaded, and situations where only a very small part of the kick reaches the surface. The simulations show how the dynamics of the scenarios are quite different. A single bubble model will not have this capability.
In this paper, two different modelling approaches for describing the unloading scenario will be compared and the differences will be highlighted. In the first approach, the single bubble model will be considered. Here the gas bubble is assumed to occupy the whole cross-sectional area, and it is fully separated from the mud regions. This will be solved by two different calculations methods, one that is taken from literature and one that is based on a shooting technique. The second and recommended approach is to use a transient drift flux model, which includes friction, acceleration terms, and gas slip. For the gas slippage model, different flow patterns will be accounted for, as will the suspension effect that causes small amounts of gas to be trapped by the mud. The drift flux model will be solved numerically using the explicit AUSMV scheme.
The impact of gas suspension will be studied by varying the onset for gas suspension to determine from simulations whether a riser will be unloaded or the kick become fully trapped in the riser. In addition, a sensitivity analysis will be presented where kick size, riser ID and riser length are varied to determine when the riser will be unloaded. The different simulations presented solves physical equations of the unloading scenario to calculate pressure at BOP, displaced mud volume (pit gain), liquid mass in well, surface rates, riser friction, and depth profiles of the gas distribution at a certain time. the tables provide a comprehensive overview of which combinations of parameters lead to a trapped gas scenario or and which lead to unloading the riser.
It is shown that a fully transient drift flux model can cover a vast range of different situations e.g. gas becomes fully trapped in the riser, the riser becomes fully unloaded, and situations where only a very small part of the kick reaches the surface. The simulations show how the dynamics of the scenarios are quite different. A single bubble model will not have this capability.