Model based Real–Time Flow Rate Estimation in Open Channels with Application to Conventional Drilling

Improvements in kick/loss detection are a key research interest in the drilling industry. Cost-effective, accurate and advanced online return flow sensors play a vital role in this regard. We have proposed the use of a Venturi channel in the return flow line with the possibility of developing it into an online soft sensor. A mechanistic model based online estimator would possibly estimate the flow rate with a reasonable accuracy. A reduced order mathematical model for this purpose has been developed by the authors [4]. In this paper, we study the possibility of estimation of flow rate using different estimators based on this reduced order model. A reduced order linear observer has been designed and tested in simulations. Further, a linear Kalman filter, an extended Kalman filter and an unscented Kalman filter are tested and compared. The extended and unscented Kalman filters are further tested with experimental results. The estimations are accurate enough with a mean absolute error of 2% and 1.9%, respectively. The proposed model based flow estimation idea has a promising potential of developing into an online soft sensor in kick-loss detection algorithms.