A METHOD FOR ANALYZING GEOTHERMAL GRADIENT HISTORIES USING THE STATISTICAL ASSESSMENT OF UNCERTAINTIES IN MATURITY MODELS
O. Huvaz1*, R.O. Thomsen2, and S. Noeth3
1 Turkish Petroleum Corporation (TPAO), Exploration Group, Basin Modeling Dep., Mustafa Kemal Mah. 2. Cad. No:86 06520, Ankara, Turkey.
2 Maersk Oil and Gas AS, Exploration Department, Esplanaden 50, DK-1263, Copenhagen, Denmark.
3 Schlumberger Data & Consulting Services, 1325 South Dairy Ashford Road, Houston, TX 77077, USA,
*corresponding author, email:firstname.lastname@example.org
A major factor contributing to uncertainty in basin modelling is the determination of the parameters necessary to reconstruct the basinís thermal history. Thermal maturity modelling is widely used in basin modelling for assessing the exploration risk. Of the available models, the chemical kinetic model Easy%Ro has gained wide acceptance.
In this study, the thermal gradient at five wells in the Danish North Sea is calibrated against vitrinite reflectance using the Easy%Ro model coupled with an inverse scheme in order to perform sensitivity analysis and to assess the uncertainty. The mean squared residual (MSR) is used as a quantitative measure of mismatch between the modelled and measured reflectance values. A 90% confidence interval is constructed for the determined mean of the squared residuals to assess the uncertainty for the given level of confidence. The sensitivity of the Easy%Ro model to variations in the thermal gradient is investigated using the uncertainty associated with scatter in the calibration data. The best thermal gradient (minimum MSR) is obtained from the MSR curve for each well. The aim is to show how the reconstruction of the thermal gradient is related to the control data and the applied model.
The applied method helps not only to determine the average thermal gradient history of a basin, but also helps to investigate the quality of the calibration data and provides a quick assessment of the uncertainty and sensitivity of any parameter in a forward deterministic model.