Geochemical analysis of rocks is fundamental to the understanding of geology and earth sciences. X-ray dispersive spectrometry and other automated techniques are increasingly being used to determined and quantify the abundances of the major, trace elements and other rock properties. This study utilized a combination of dispersive spectrometric techniques (MicroXRF) and impulse rebound hammer method to establish links between geochemical and mechanical properties of rocks through a non-destructive method. MicroXRF has high resolution and can detect trace elements within the parts per billion range. The micro-rebound hammer was used to generate a reduced Young's modulus (E*), which gives a measure of the rock strength with negligible impact on the rock itself.
In order to explore, visualize and understand the dataset generated, principal component analysis (PCA) was applied to emphasize variation and bring out strong patterns in the dataset. The first two dimensions of PCA express 57.09% of the total dataset inertia; that means that 57.09% total variability in the data is explained by the planes/dimensions. The first dimension, which showed a strong positive correlation to clay forming minerals and rock strength, was tentatively identified as the clay gradient. The second dimension describes diagenetic alteration processes responsible for the enrichment of elements such as Ni, Mo etc. Further, a positive correlation was established between E* and four elements Cobalt (Co), Strontium (Sr), Titanium (Ti), and Zircon (Zr). Remarkably, Silicon (Si) had a negative correlation with all elements but positive correlation with porosity and permeability. We therefore identified Co, Ti, Sr, and Zr as proxy for the determination of rock strength specific for studied samples and proposed a workflow based on our sequences of analysis and interpretation. Furthermore, we identified four chemo- mechanical facies through hierarchical clustering of the product of the PCA.
This presented methodology could be specifically useful for geomechanical characterization of rocks; a key requirement needed for in-situ stresses estimation, wellbore stability analysis, reservoir stimulation and compaction, pore pressure prediction, and more importantly for characterizing drill cuttings where size and time are limiting. Drilling operations require constantly evolving cost effective and time efficient techniques, the proposed workflow will serve these purposes i.e. rapid determination of elemental composition (microxrf) coupled with E*will give a reliable proxy for rock strength. The technique can be applied to, drill cuttings, slabs and whole core directly without prior sample preparation.
Geomechanical characterization of subsurface rocks is important for many applications throughout the asset life cycle such as borehole instability, pore pressure prediction, seal breach and fault reactivation, drill bits and drilling parameters selection, sand production, hydraulic fracturing, and reservoir compaction (Meyers et al., 2005; Klimentos, 2005; Germay et al., 2017). A key component of Geomechanical characterization is the model calibration with reliable core data. Typically, core data calibration is performed using triaxial tests data output, such as the uniaxial compressive strength (UCS) and elastic properties of rocks (Young's modulus, Poisson's ratio, etc.), that are empirically linked to wireline data. Sample availability, representativeness, time, and cost are problems associated with core-based rock measurements for mechanical properties [3; 4]. There is also the issue of uncertainty associated with upscaling laboratory generated data with wireline data. Core-based measurement output is usually very limited, hardly constitutes statistically representative data as compare to large data from wireline logs, making it difficult to generate a reliable empirical correlation. Another issue is the inherent heterogeneity in rocks, which varies from nano to field scales. This makes establishment of empirical relationship between scattered core data and wireline data a subjective task. Even rocks that appear identically twin in bulk properties can vary widely in microstructure. Characterizing such reservoir-scale heterogeneities requires statistically representative data and the problems associated with core-based measurement make such substantial number of data points requirement an abominable.