Bulk mineralogical characterisation of oilfield reservoir rocks and sandstones using DRIFTS and partial least squares analysis

CLEGG, Francis, BREEN, C., HERRON, M.M., HILD, G.P., HILLIER, S., HUGHES, T.L., JONES, T.G.J., MATTESON, A. and YARWOOD, J. (2008). Bulk mineralogical characterisation of oilfield reservoir rocks and sandstones using DRIFTS and partial least squares analysis. Journal of Petroleum Science and Engineering, 60, 1-17.

[img]
Preview
PDF
Clegg_bulk_mineralogical_oilfield(AM).pdf - Accepted Version
Creative Commons Attribution Non-commercial No Derivatives.

Download (387kB) | Preview
Official URL: https://www.sciencedirect.com/science/article/pii/...

Abstract

The feasibility of applying Partial Least Squares (PLS) to the Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) spectra of mineral mixtures, quarry sandstones and oilfield reservoir rocks has been investigated and shown considerable potential for accurate and precise mineralogical analysis. Rapid spectrum acquisition and data processing together with small sample size requirements are key advantages of the PLS–DRIFTS method. A PLS model was created from the DRIFTS spectra of mixtures of seven mineral standards chosen to represent the most frequently encountered minerals in sandstone-type rocks; quartz, dolomite, montmorillonite, illite, kaolinite, chlorite and albite. The PLS–DRIFTS model was able to quantify the mineral components of independent mixtures with an absolute error of 1 wt.% for all the minerals (concentration range 0–30 wt.%) with the exception of quartz which exhibited an absolute error of 3 wt.% (concentration range 50–90 wt.%). The results provided by applying this PLS–DRIFTS model to several sandstone-type quarry rocks and a suite of oilfield reservoir rocks were considerably better than anticipated even though the model did not describe all the mineral components present in the samples nor the entire variance of constituent mineral components (e.g. crystallinity). The model was not able to differentiate between montmorillonite and illite probably due to the similarity of the DRIFTS spectra of these minerals, but it was able to quantify the combined (montmorillonite + illite) concentrations to within 1 wt.%. The model over-predicted the concentration of albite in the quarry rocks due to the presence of K-feldspar, which has a similar DRIFTS spectrum and was not included in the model. However, the model accurately predicted the total (albite and K-feldspar) concentration to within 4 wt.%. A separate PLS–DRIFTS model constructed using the DRIFTS spectra of the oilfield reservoir rocks showed that the carbonate components, calcite and dolomite could be differentiated and quantified to within 5.0 and 3.6 wt.%, respectively. This feasibility study confirmed the strong potential of combining DRIFTS with a multivariate statistical approach such as PLS and it is clear that more sophisticated models, that incorporates and describes a higher percentage of the variance in unknowns, would further improve the predictions.

Item Type: Article
Uncontrolled Keywords: 0914 Resources Engineering And Extractive Metallurgy; Energy
SWORD Depositor: Symplectic Elements
Depositing User: Symplectic Elements
Date Deposited: 21 Jan 2019 14:06
Last Modified: 21 Jan 2019 14:15
URI: http://shura.shu.ac.uk/id/eprint/23750

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year

View more statistics