Bayesian Lithology and Fluid Prediction on the Mikkel Field using a geologically constrained prior model

Master's thesis in Petroleum Geosciences engineering The main objective of this project was to predict lithology and fluid classes from prestack seismic data at the Mikkel field in the Norwegian Sea. Segments 5 and 7 on the flank of the main field are of special interest since these segments ha...

Full description

Bibliographic Details
Main Author: Castillo, Isaias
Other Authors: Buland, Arild
Format: Master Thesis
Language:English
Published: University of Stavanger, Norway 2017
Subjects:
Online Access:http://hdl.handle.net/11250/2464531
Description
Summary:Master's thesis in Petroleum Geosciences engineering The main objective of this project was to predict lithology and fluid classes from prestack seismic data at the Mikkel field in the Norwegian Sea. Segments 5 and 7 on the flank of the main field are of special interest since these segments have not been drilled yet, and Statoil is considering to drill a well which might improve the hydrocarbon production. The Mikkel field is a gas condensate field and the reservoirs are in the Garn and Ile Formations. The area is structurally complex with faulted segments and the seismic interpretation of the reservoirs is challenging. Bayesian prestack seismic inversion to lithology and fluid prediction (LFP) has been applied in an attempt to reduce the uncertainty in the geological knowledge of the field. A vertical coupling algorithm which applies transition probabilities has been tested in the Bayesian LFP by using a geologically constrained prior model. Two inversions were performed: the first analysis was done using only one single zone defined by 2 horizons for the inversion window, in the second analysis the inversion window was divided into multiple zones defined by 5 horizons. The results were compared with extended elastic impedance (EEI) and a pointwise LFP algorithm which does not use transition probabilities. The results with the vertical coupling algorithm in general were in agreement with the pointwise algorithm and EEI, especially where the gas sand was predicted. The gas sand predicted in the Ile Formation follows the geological structure and stops at the gas/light oil water contact. In segment 5, the gas sand in the Ile Formation was not detected because this Formation is below the gas/light oil water contact. In segment 5, the gas sand predicted in the Garn Formation has probabilities above 90% and it follows the geological structure. However, it does not stop at the gas/light oil water contact, indicating possible residual gas below this contact. The techniques do not separate between high and low gas ...