| dc.description.abstract | Drilling operations in oil and gas industry are getting more and more advanced and complicated. One of the main reasons is that the real time monitoring and control of the processes are getting complex in their executions. As the wells constructed are getting more complex, it leads to new type of drilling methods with an increase in need of various tools and equipment. Therefore now a day’s more and more research work is carried out and as a part of these research work, the flow rate estimations using level measurements in open channel venturi rigs are evaluated with mud flow applications.
This study mainly reveals the different approaches that can be used with respect to system identification, in order to achieve better flow rate estimator models. Deterministic and Stochastic system identification and Realization (DSR) method, Prediction Error Methods (PEM), N4SID with PEM method, State Space PEM (SSPEM) method and Neural Network approach were mainly discussed and evaluated in this report. The achieved models from these approaches were validated with two different sets of experimental data so that the reliability of these models were assured. However the obtained models estimated the mass flow rate with the flow depth level measurements as the input variables.
The best estimator model was obtained with N4SID algorithm together with the Prediction Error Method. This model only needs the level measurements of the flow depth in the open channel venturi. The model validation results provided the flow rate estimations with a mean percentage error of 2% and with a Root mean Square error of 8kg/min. Therefore further discussions were made with the use of this model instead of the Coriolis meter as a real-time flow rate estimator in offshore drilling industry. | nb_NO |
