New, model-based approaches to optimise the production of hydrocarbons from a field use mathematical control theory concepts that enable injection and production strategy, for example, to maximise NPV over the lifetime of the field. TNO has co-developed and implemented a robust optimisation tool that account for uncertainty in reservoir characteristics and can easily be incorporated in a closed-loop reservoir management framework.
Advances in well technology enable more and more information on well and reservoir conditions to be extracted and well inflow to be controlled. A major challenge is to exploit the full potential of this information and capability, and this is where automated methods come into play. First, the amount of data from smart wells or soft-sensing techniques may quickly become too overwhelming too handle manually. Secondly, it is now widely recognised that a single reservoir model does not fully describe the range of uncertainty inherent in our knowledge of the reservoir. To properly deal with this uncertainty, a whole range of plausible reservoir models consistent with available prior information should be generated, in itself a computational challenge. Optimisation methods that provide solutions for both issues are currently available and may provide a guide to day-to-day operations as well as help to find an optimum long-term development strategy.
In an EU-funded production forecasting with uncertainty quantification project, several methods for the automated generation of models were explored, some of which now appear in commercial packages for history matching and uncertainty analysis. The recent Integrated Systems Approach for Petroleum Production (ISAPP) project, a collaboration between TNO, Shell and Delft University of Technology, has gone further and explored new methods for the automated updating of models and production strategies, comparing their effectiveness in an SPE workshop on closed-loop reservoir management. Subsequently, TNO developed a simulator-independent optimisation tool that allows for uncertainty in the reservoir properties as well as help to find an optimal production strategy for the remaining field lifetime by computing time-dependent optimal injection and production rates, or the optimal settings for smart ICV's.
A recent project TNO was involved with concerns an industrial partner interested in finding an optimal strategy for production from a thin oil rim reservoir, which encountered problems with gas coning and wax deposition. Given the lack of downhole sensors, an optimal strategy was devised on the basis of surface temperature measurements, providing an improved guide to daily operations. Recently TNO also constructed the set of reservoir models for the SPE workshop on closed-loop reservoir management. A synthetic, but realistic, reservoir was built, equipped with smart wells, and workshop participants from industry and academia were able to test their methods for automated reservoir management and to compare their benefits and drawbacks. TNO has extensive experience of history matching of models of many types of reservoirs as well as reservoir engineers and geologists able to provide advice on field development planning from both a geological and economic perspective.