Modeling of Production Processes
Model identification and process analysis for the optimized control of conveyor processes in the petroleum industry.
For a large part of today‘s known oil reserves, only about 35 percent of the available amount can be exploited due to economic or technical reasons. A major factor affecting the overall yield potential is the production strategy used for a reservoir.
Among other things, this is to prevent certain parts of a reservoir from being drained prematurely, because this can make further use considerably more difficult or even impossible. Likewise, proactive support strategies should ensure that no premature water leaks or poor mixing ratios between gas, water and oil occur, as these situations can also adversely affect the total amount of recoverable oil. Today, very sophisticated technology and processes for the development and exploitation of petroleum resources are used in many cases, meaning that the production processes are becoming increasingly complex to manage.
In order to make optimal decisions regarding the production processes, accurate knowledge of the geological structures and physical interactions in the reservoirs is necessary. Since this information is not directly available, simulation models for their description and prediction are used. Traditional methods can often be applied only under the condition of well-defined model assumptions that frequently cannot be guaranteed in real situations.
The objectives of the Algebraic Oil project were to develop mathematical foundations and to implement a prototype software application for improving model identifications of production systems. The Algebraic Oil Project was initiated several years ago by the Exploratory Research Division of the Shell International Exploration and Production Group in Rijswijk / Netherlands and finished at the end of 2012 with the completion of a prototype application.
In the framework of the project, mathematical foundations in the field of symbolic and numerical computer algebra were developed further so that improved simulation models can be calculated. These models allow the behavior of a production system to be described and predicted over an extended period of time without restrictive model assumptions more reliable than was the case with previous approaches. In this context the Buchberger Theory of Groebner Bases plays a central role, since it allows the structural capturing of non-linear relations in models.
Measurements on the oil fields possibly include large and random errors, so that the exact algebraic computation methods of the classical theory of Groebner Bases need to be adapted or extended for such applications. This work is carried out by Shell in cooperation with the Institute for Symbolic Computation at the University of Passau.
A second objective was to demonstrate the feasibility and efficiency of the method under realistic conditions. Therefore, the RISC Software GmbH implemented a prototypical software system that serves as a tool for the improvement of modeling methods, as well as a basis for a software component to be subsequently used to monitor and control the production in real-life operations for a better use of the existing reserves in the deposits.