Improved Calibration of the Absolute Thermal Maturity of Coal-Sourced Oils and Gas Condensates Using PLS Regression
Sykes, Richard; Zink, Klaus-Gerhard
Dept of Petroleum Geoscience, GNS Science, Lower Hutt, New Zealand.
Knowing the absolute maturity of petroleum fluids can help to constrain the depth and timing of expulsion from source rocks and to clarify genetic relationships. For example: do suspected oil shows represent indigenous or migrated oil; are various oils from the same or sequential charge events; and did gas condensates form via phase separation from oils or directly from source rocks at higher maturity?
Absolute maturities are usually established by calibrating the fluids' molecular maturity parameters against equivalent parameters determined for bitumens extracted from source rocks of known maturity. Use of individual parameters can be problematic if the selected parameters reach equilibrium part-way through the oil window or are adversely affected by variation in source organofacies. In this study, we have developed a new method for predicting absolute maturities of coal-sourced oils and gas condensates based on Partial Least Squares (PLS) regression. The use of factor-based regression allows multiple maturity parameters to be used collectively, thus minimising potential errors associated with individual parameters.
Initially, 28 molecular maturity parameters were determined for 129 New Zealand coaly rocks of known Suggate coal rank [Rank(Sr)] and Tmax values from throughout the oil window. Then, using bivariate correlations, Principal Component Analysis and PLS regression, two PLS regression models were developed for prediction of Rank(Sr) and Tmax. The first model, optimised for Rank(Sr), uses a training set of 58 source rocks and 5 methylnaphthalene maturity parameters. The second model, optimised for Tmax, uses 52 source rocks, the same 5 methylnaphthalene parameters and 3 methylphenanthrene parameters. Internal cross-validation shows that the models predict known Rank(Sr) and Tmax values with standard errors of 0.40 rank units and 2.7°C, respectively, which are comparable to those of measured values.
Application of the PLS models to a set of 132 New Zealand coal-sourced oils and gas condensates predicts maturity values that are largely consistent with Rank(Sr) and Tmax limits for the oil and gas windows determined independently from source rock pyrolysis studies. The gas condensates are overall more mature than the oils, but with significant overlap in their maturity ranges. The calibrated maturities of petroleum fluids in Taranaki Basin provide useful insight into the migration and charge histories of various fields, prospects and leads.
AAPG Search and Discovery Article #90155©2012 AAPG International Conference & Exhibition, Singapore, 16-19 September 2012