Drilling mud property estimator
MetadataShow full item record
In 1969 the first petroleum discovery was made on the Norwegian continental shelf. This founding marks the beginning of what has become the most important contribution to the Norwegian economy. Advanced research within the field and development of techniques and equipment has been necessary in the ongoing work, and the technology has come a long way since 1969. However, when it comes to one of the most important parts of the drilling operation, namely the drilling fluid, the techniques used today are largely based on the same techniques used over 20 year ago. Today, the properties of the drilling fluid are manually measured in interval of 15 minutes. By obtaining an automatic update of the drilling fluid measurements, errors caused by human inaccuracy or uncertainties in old-fashioned measurement techniques, can be avoided. It will also be highly significant for the field of automated drilling operations. In this thesis, the functionality of the drilling fluid and the possibility of using unscented Kalman filters to estimate drilling fluid properties have been studied. The implementations were divided into two cases; Case 1 and Case 2.In Case 1 the theory behind the Instrumented stand pipe was applied on pressure measurements made in the annulus. Two systems were used in this case to generate test-data (system inputs and available measurements). A simple model simulating two differential pressures; one over a vertical annulus section, and the other one over a nearly horizontal annulus section, was first implemented in Simulink®. Based on test-data from this model the unscented Kalman filter was used to estimate the drilling fluid properties density and plastic viscosity. Secondly, the unscented Kalman filter was executed with test-data from the advanced drilling simulator WeMod. In Case 2, the unscented Kalman filter from Case 1 was combined with an unscented Kalman filter estimating the mud pump pressure, choke pressure, flow through the bit, and a geometry parameter. Also in this case, the unscented Kalman filters were tested in two scenarios. First by using test-data from a simple model of a drilling system implemented in Simulink®, and secondly by using test-data from WeMod. In this thesis it was found that in the scenarios where test-data were generated by the Simulink® models, the unscented Kalman filter estimating density and plastic viscosity had an acceptable performance. In the scenarios were test-data from WeMod was used, the unscented Kalman filters estimating density and plastic viscosity did not perform as desired. It found that the frictional pressure loss in the vertical and horizontal annulus sections were different in the WeMod simulation, which may partly be the reason for way the filters did not managed to estimate the right density and plastic viscosity.