Maximizing oil production

Dynamic simulation of fluid flow in oil reservoirs is an important tool in reservoir management.

Such simulations are used for estimating reserves, generating production profiles, planning of new wells, evaluation of enhanced recovery measures and well interventions, etc. A main challenge for the oil companies is to maximize the information from all data, thus being able to build models with a minimum of uncertainty. Also, while the amount of data to be integrated in the models and the complexity increase, the companies put increasingly larger demands on efficiency, meaning that methods and tools for fast model updating will be more and more important for the industry.

The geometric complexity of the oil bearing formation models and the non-linear coupling in the flow models represent extremely challenging tasks. A main goal for Centre for Integrated Petroleum Research (CIPR) is to integrate the geological, physical and chemical modelling with the mathematical tool development, enabling us to handle and study such complex systems. Our information about the reservoir is typically very sparse. It is also to a large degree provided on scales which are not the same as the model scale. To reduce the uncertainty in the model predictions, it is important that the models are conditioned to all available data: seismic data, well logs, geological knowledge, pressure measurements, observed production, etc. Time dependent measurements are called dynamic data. Typical dynamic data are pressure development and observed production from wells. Recently, also repeated seismic surveys, also called 4D seismic indicating measurements both in space and time, have become an important source of dynamic information. Conditioning reservoir simulation models to dynamic data is called history matching, and since it typically requires a large number of simulation runs, this is a very complicated and time consuming process.

An important factor in flow modelling is the model resolution. On the one hand, high resolution gives good accuracy, but on the other hand, high resolution increases computational time. The degree of resolution, which is necessary, also to a large degree depends on what the models are to be used for. While high resolution is needed for detailed well planning, estimating uncertainty in oil reserves requires running a large number of predictions, but does not require very accurate results for individual wells. Multiscale modelling is currently an active area of research within computational science and applied mathematics, and at the Department of Mathematics, UiB, one has worked on methods where the model resolution varies continuously both in time and space.

In one of the projects at CIPR we have developed techniques for multiscale history matching, and the results so far indicate that by using these techniques one may both speed up the process and at the same time generate models with different resolution, all being conditioned to the dynamic data. This means that not only do these techniques make the history matching process more efficient, but one also obtains a suite of models suitable for different purposes. We also work with methods where the resolution in reservoir properties is gradually refined in a “clever” way, and the objective is to find the finest resolution warranted by the data (Adaptive Multiscale Estimation). Using these techniques also requires efficient methods for minimizing cost functions. As a part of a PhD project at CIPR, such a technique is implemented in an in-house reservoir simulator, and as far as we know, this is the first implementation of such methods in a compositional simulator, that is, a reservoir simulator where the oil and the gas may be split into their individual chemical components. In addition, new methods for the minimization of cost functions are developed in a separate CIPR project, which has direct support from the Research Council of Norway through the Petromaks programme.

UiB wins competition for petroleum projects

The University of Bergen is involved in more petroleum research projects than any other institution in Norway . The Research Council of Norway has gathered all petroleum research into a single major programme called Petromaks, whose overarching aim is to help keep Norwegian oil production going for another fifty years. The University of Bergen has been given funding for eight projects in the programme, and the University is also participating in five projects in which another institution is the lead partner with the Research Council. “The projects will have a duration of three to four years, a time-scale which will enable us to engage a good number of new research fellows,” says Arne Skauge, research director at the Centre for Integrated Petroleum Research (CIPR).

www.cipr.uib.no

Back to front page

To order a free example of the paper edition of Features, contact the Communication and Media Centre either by e.mail or telephone +47 55 58 69 00