Pore scale imaging has developed over the last 20 years from a primarily academic technique used to visualize pore structures for fundamental research into an increasingly crucial industrial tool. It is now possible to routinely image, both in 2D and 3D, structures and processes occurring from the whole core down to the sub-nanometer scale. In this webinar we will review how pore scale imaging can be used within the Oil and Gas community, focusing on three major applications; digital core analysis, multiscale rock characterization and automated optical petrography.
Digital core analysis focusses on how multiscale non-invasive imaging can be used to characterize pore structures. This can be used as the input for computational fluid dynamics simulations for the prediction of effective fluid properties (also known as digital rock physics). It can also be integrated with miniaturized core analysis equipment to allow for fluids to be imaged in situ under representative subsurface pressures and temperatures.
Rock characterization uses correlated light, electron and x-ray datasets to characterize rock, pore and grain structures from the cm down to the nm. This allows for the investigation of petrophysically relevant properties, such as pore connectivity or mineralogy. The challenge of scale can be addressed by the spatial correlation of high-resolution structures and microfluidics with macro-scale heterogeneity. Local insights can then be upscaled back to the core-plug scale using macro-scale rock type maps, generating using machine learning based computer vision tools.
Finally we will show how optical petrography can be used to bridge the gap between pore, plug and core scales by automating the acquisition, classification and analysis of thin section images. Multi-polarized birefringent light data can be fed to automated segmentation tools to classify pore, grain and mineralogy information directly. The segmented information can then be fed into advanced algorithms allowing for petrophysical properties (e.g. permeability) to be inferred, directly creating porosity-permeability upscaling functions. Trained models can be then applied to newly acquired datasets, allowing for high throughput analysis at scale, informing interpretation across an entire reservoir.