A suitable approach to achieve a fit-for-purpose model for field application is upscaling of reservoir properties. However, upscaling of saturation functions like relative permeability is still a source of debate and is frequently omitted due to its inconvenient complexity. Upscaling of these functions, whether by steady-state methods or other techniques, typically generates a complex myriad of relative permeability curves that is challenging to handle pragmatically. This study investigates whether a threeparameter correlation of relative permeability is able to represent the saturation function for fine grid simulation models with the most common heterogeneities, coarse grid simulation models with upscaled properties in addition to verify two-phase core-flow experiments.Implementation of the proposed three-parameter correlations of the relative permeability in the algorithm of full field simulators will overcome the representation challenges of present published upscaling methods.Upscaling of each coarse grid block individually can then be handled pragmatically, and hence provide fit-forpurpose models for field application. Statoil is acknowledged for giving permission to publish the experimental results. Vilgeir Dalen is acknowledged for valuable contributions and advices during the work of this paper. Egil Boye Petersen Jr. and Eimund Gilje are acknowledged for valuable discussions. Weatherford Petroleum Consultancy, Norway is acknowledged for including the new correlation in the core flow simulator Sendra.

Summary We present and test a new method for selection of pseudocomponents and averaging equation of state (EOS) parameters with excellent results. The new method is especially useful for compositional reservoir simulation with or without extended downstream process modeling. Introduction The main results from a compositional reservoir simulation are the production rates. Normally, these are stable oil and dry gas streams, but other streams may also be reported. The detailed description of the fluid compositions can therefore be reduced to some extent if the overall description of the fluid properties is acceptable. This is achieved by lumping the components into component groups called pseudocomponents. This reduces the computation time and the needed memory. The traditional way to prepare a fluid model with pseudocomponents is to group them based on personal experience and parameter regression. This requires a lot of experience and continuous practice, and probably represents one of the greatest obstacles for a reservoir engineer confronted with compositional reservoir simulation for the first time. This paper presents an automatic method of selecting pseudocomponents. We will compare the new method with other existing methods that automatically group components into pseudocomponents. The existing methods used in the comparison follow.The method of mole-weighted average with components grouped to have approximately equal mole fraction.The method of mass-weighted average with components grouped to have approximately equal mass fraction.The method of Newly et al. The method of Danesh et al. is not included in the comparison because it does not use the common basic EOS parameters, and it is difficult to port the method of Danesh et al. to other computer programs.

The paper presents the results of an integrated full field reservoir description study that has been carried out on the Statfjord reservoir in the Statfjord field. Two major issues addressed are reservoir modeling and history matching of the simulation model. Geostatistical modeling was used as a main tool to integrate all available quantitative and qualitative data into a high resolution geological model of the reservoir. Stratigraphic framework for the modeling was set up by the high resolution sequence stratigraphic interpretation in 34 development and 11 exploration wells. Based on this interpretation, the new reservoir zonation was established and major facies types were identified. The reservoir modeling created a solid base for building a reservoir simulation model. Advanced grid builder allowed for accurate representation of such important features as heavily slanted faults and typical heterogeneity pattern. Nonlinear regression based technique was consistently applied to match reservoir performance data for 16 years of production. Introduction Statfjord field, discovered in 1974, is one of the largest oil fields in the North Sea, with an original oil in place of over a billion Sm3 and estimated recovery factor exceeding 60%. The field is located in the prolific northern part of the Viking graben on the U.K./Norway boundary. The Statfjord field which is 24 km long and averages 4 km in width is located in a fault-block structure which is tilted at about 70 to the west (Fig. 1). On the east, the field is bounded by a major boundary fault system. Between the structural crest and the boundary fault, the reservoirs are cut by rotational faulting and truncated by erosion events. The two most important reservoir intervals are the Middle Jurassic Brent Group and the Upper Triassic to Lower Jurassic Statfjord Formation. The Statfjord reservoir which is the focus of this paper, contains around 20% of the total original oil in place. The Brent reservoir is developed under line water drive, whereas the Statfjord reservoir is under high-pressure miscible gas flood. The development scheme utilizes 3 gravity-based Condeep platforms - A, B, and C which became operational in 1979, 1982, and 1985 respectively. Continuous gas flood led to an effect that the upflank area covered by the first line of the oil producers became gas flooded and a "wedge zone" was formed. The wedge zone, located between initial drainage line and OWC, contains the bulk of the oil to be produced. At this stage, the primary oil producers were either sidetracked or converted to the Brent reservoir. The two first horizontal oil producers targeting the wedge zone, were drilled in 1990. This marked a change of the management strategy towards production from the wedge zone (Figs. 1 and 2). The current status of the reservoir is shown in Tables 1 and 2. With production from Upper Statfjord declining rapidly, the Raude formation will play a more important role in efforts to sustain the total production from the reservoir. The next phase in the development of the Statfjord reservoir will most likely be updip water injection in Upper Statfjord and downdip WAG injection in Lower Statfjord. An existing full-field reservoir simulation model that has been adequate for reservoir management purposes in the plateau production period is no longer capable to deliver sound and reliable basis for crucial management decisions. To meet the new challenges a Statfjord field reservoir description project was initiated in 1993. The main goal of the project was to generate an improved reservoir description by developing new technologies in a multidisciplinary environment in order to:–Better optimize and predict future production.–Identify remaining reserves.–Optimize well placement and productivity. P. 915^

Mathematical correlation has been widely used in oil and gas industry to model relative permeability and capillary pressure from water saturation. The application of mathematical correlation is essential especially in the absence of laboratory data. Additionally, the correlation is also applied to generate a refined relative permeability and capillary pressure table as the input to reservoir simulation.There are several correlations being used in the industry such as Corey, Skjaeveland and LET correlation. The focus in this paper is the LET correlation. The correlation offers more flexibility as well as accuracy in matching the responses from laboratory experiments.Having a representative correlation is the basic, but the curve-fitting to the experimental data is also indispensable. In a problem which involves a non-linear correlation, the attempt to find a solution which fits the experimental data becomes more complex. To overcome this problem, it is fundamental to have a search method which can fit the experiment data with the lowest possible residual errors.In this paper, different search methods of curve-fitting are investigated. In the last part of the paper we will compare the performance of each method. The main evaluation parameters are the residual error and the computational time. The methods studied in this paper are the Levenberg -Marquardt method, particle swarm optimization and mesh pattern search.

No abstract