| Summary: | This thesis focuses on exploring applications and optimizing transient numerical models for simulating well control situations. The main scope of the research was to find opportunity for improving existing numerical models and to improve the models accordingly. Relevant cases were constructed, simulated with different mathematical models and numerical methods, and the results were compared. The cases constructed were to a large degree motivated from challenges associated with kick handling in subsea back pressure MPD systems and gas in riser unloading events. The models that have been used for evaluating the transient scenarios are the single bubble model and Drift-Flux model. A static analytical model was also developed for kick tolerance evaluations. The first topic studied was kick tolerance evaluation from a probabilistic perspective, using Monte Carlo simulations. By adopting this approach, one can get a probability for whether a certain kick volume can lead to fracturing the formation in the weakest spot. It is also shown how this approach can be useful for analyzing how the uncertainties in each input parameters change the results. The Monte Carlo simulations has, to our knowledge, not been used so far for kick tolerance evaluation. An important matter explored throughout the research was the effect of the numerical diffusion in the results when simulating well control situations and the importance of restricting this effect. We demonstrated how to use different techniques for restricting the numerical diffusion and compared the results between them. This thesis also studies kick behavior when using subsea backpressure MPD systems with oil based mud. In this system, one need to evaluate what will be the maximum surface rates and surface pressure compared to equipment limitations when trying to circulate a certain kick volume directly through the MPD system. The transient flow model for simulating a kick in oil based mud was provided by SINTEF Industry. This model uses the Drift-Flux formulation solved ...
|
|---|