Control with discrete output: Step choke control
MetadataVis full innførsel
The need and requirement for obtaining good and reliable models of the oil and gas reservoirs are increasingly important. A trustworthy and high-quality model will facilitate the optimisation of the productivity, which will increase the profit and ensure improved utilisation of natural petroleum resources. The potential to provide better and more accurate models of the wells and reservoirs are prominent by improving the quality of well-tests, and by automate the control of the step chokes during the tests.In this master thesis, some techniques are proposed and developed as different methods to obtain less variation in both the step choke?s position and the process variable when differential well-tests are performed. The objective is to limit the step choke usage during differential well-tests, in combination with achieving as small deviation in the well flow as possible within this limited use of the step choke. Additionally, reduce the need for manual operations during these tests.More precisely; two different main techniques is presented, with some additional approaches that controls the step choke with a limited usage by guarantying the choke movements within a certain time period. The first of these two is based on optimisation. This technique chooses the choke position that leads to an output sequence that is closest, in average over a period backwards in time, to the output signal of a conventional PI controller. The second technique to control the step choke is based on a process model. This technique replaces the PI controller and implements a discrete controller output transferred to the step choke. It uses the model and linearisation to predict the time instant for position change and deciding the necessary choke position that achieves the desired well pressure at the end of a specified prediction horizon.The techniques are structured as modules, where the algorithms are implemented in Dymola and tested with simulations of a differential well-test performed on a reservoir model with common pipeline structure. Their performance is assessed by counting the number of choke position changes and measuring the control error obtained when simulating a well-test performed on a corresponding well in the common pipeline network.It was found by the simulation results that including a restricted time period between position changes so that the choke usage is limited, has a great potential to reduce the wear and tear on the step choke at the same time as maintaining some accuracy. The overall performance assessment of the developed techniques showed also that using a method that dealing with both the control of the process and takes account of the discrete step choke, has advantages compared to other techniques studied in this thesis and the techniques in use today.