Abstract
Application of model-based estimation techniques to kick detection come with constraints imposed on the structure of the mathematical model by stochastic estimation algorithms as well as the limited number of flow-line measurements typically available in most drilling operations. This, along with high computational cost of numerical models call for low to medium order deterministic models that still capture the dominant effects of fluid flow during drilling. A built-for-purpose single phase deterministic model for conventional drilling is presented. The model development is the first step in a two-step process of real-time early kick detection using stochastic estimation techniques. The model is developed using lumped parameter modeling afforded by bond graph modeling technique. Aside a hydraulic model for the wellbore, well-formation interactions that give rise to kicks, lost circulation, and wellbore breathing are also modeled and coupled. This holistic modeling approach provides a compact set of low order equations which makes stochastic estimation feasible, and well monitoring easier so that, for example, wellbore breathing is not mistaken for kicks, leading to unnecessary non-productive time and possibly inducing kicks due to needless well control actions undertaken. Lumped well parameters are calibrated with an optimization tool, and the model is validated using historical data from a conventionally drilled, onshore well.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call