ESTIMATING THE PERMEABILITY FOR THE TROLL FIELD USING STATISTICAL METHODS QUERYING A FIELDWIDE DATABASE.
R. Nicolaysen
Saga Petroleum MS
T. Svendsen
Atlas Wireline Services
ABSTRACT
A technique is described for enhancing the accuracy of estimated permeability curves for the Sognefjord formation of the Troll field. Statistical methods are used, querying a field-wide database built from core data, its geological descriptions, and wireline logs.
The Troll field is located in the Norwegian sector of the North Sea, and has been explored by a total of 27 wells. Large reserves of gas and oil have been encountered. The reservoir has been described by an extensive logging and coring program. The field is still being developed and reservoir models are being refined.
Due to the wide permeability distribution observed within the field’s high porosity range, it has previously been difficult to establish realistic permeability curves, which reflect these permeability contrasts. Predicting permeabilities by means of standard log-core regression relationships has proven to yield unsatisfactory results on a field basis. Standard regression applications often introduce significant uncertainties, which limit their usefulness for accurate permeability determinations.
The present permeability study is based on building a database by integrating selected log measurements with core permeabilities over intervals that meet established quality criteria. The normalized gamma ray, the density-neutron porosity difference, and density logs were selected to best reflect the lithological, porosity, and diagenetic effects upon permeability.
The fieldwide database establishes the implicit relationship in the data set and no apriori knowledge of equations or relationships between variables or their distribution are assumed.
Over depth intervals where core permeability data are absent, or in wells without core measurements, the log data are used to query the fieldwide database for prediction of permeability and its associated statistical uncertainty.
This technique will be demonstrated to have given significant improvement in the estimation of continuous permeability curves, when compared with traditional methods. The method can easily be extended to other fields.