Looking for new ways of capturing and storing CO 2 , the IBM Research team decided to use high performance computing (HPC) and hybrid cloud technologies .
In an article recently published in Scientific Reports in the journal Nature, IBM scientists presented an optimized algorithm for modeling microcavities - capillary networks - in porous rocks naturally occurring in geological formations. These microcavities can safely store, in liquid or solid form, carbon dioxide captured from exhaust gases and other point emissions that occur during energy production.
Together with scientists from the Institute of Physics of São Carlos, part of the University of São Paulo (Brazil), and representatives of the international petroleum corporation Petrobras, the IBM Research team showed that using the proposed algorithms, basic analysis of rocks can be performed faster and more efficiently than with using only laboratory tests. The full analysis cycle can be shortened from a few months to a few days and, accordingly, reduce costs and potentially increase efficiency, as well as mitigate some of the risks associated with storing carbon in geological formations.
Rock as a storehouse of carbon dioxide
The idea is to first compress CO 2 to a liquid form and then pour it into porous rock. After that, the gas is mineralized, that is, it becomes solid. In this form, it can be stored safely for decades or even centuries.
While this is one of the most promising solutions to the carbon storage problem, there are some complexities. How effective will this process be given the current state of the repositories? What are the physical and chemical characteristics to keep in mind when choosing a breed? How to accelerate CO 2 mineralization and ensure the stability of the geological repository?
To answer these questions, IBM Research has developed algorithms to analyze high-resolution images of rock samples. Using X-ray microtomography, the researchers took a series of images and then created a digital 3D model of the rock sample. The results were then extrapolated to a rock mass more than 3,000 times the size of the sample taken to see what would happen at that scale up. It is this scale that is commonly used in laboratory assessment and verification of rock capabilities. The high accuracy of the algorithms makes it possible to more correctly estimate how much CO 2 the rock micropores can accommodate - their diameter can range from a few nanometers to a few millimeters.
Cloud and HPC Applications
Having figured out the spatial distribution of interconnected cavities within a rock sample, the scientists apply a flow analysis model and predict capacity based on the geometric boundaries of the capillary network. The ability to accurately estimate the geometry of the cavities is essential for simulating the flow of CO 2 as it fills the rock.
This is where cloud technologies and high-performance computing systems are used.
IBM Research simulators use advanced image processing algorithms to create 3D models. These algorithms walk the sample segment by segment, analyzing layer-by-layer images and detecting the presence of cavities and other features of different layers of the rock mass. After this segmentation, the algorithm creates an extremely accurate model of the capillary network. It analyzes the identified cavities and estimates the volume of available space of associated cavities in the rock.
In the published article, the team of scientists showed that the improved forecasting accuracy achieved by the use of the new algorithm allows identifying patterns and calculating rock capacity without additional adjustment or calibration. In addition, she demonstrated that it is possible to calculate rock capacity at the same scale as in laboratory studies, taking as a basis the characteristics of microflows within a capillary system. In this case, the forecast will be distinguished by increased accuracy, and it can be made much faster than in the laboratory.
New algorithm for digitizing and modeling rocks worksbased on the cloud system IBM FlowDiscovery Simulator. This tool, presented at the March 2021 meeting of the American Physical Society, is helping to assess CO 2 capture capabilities and, in the long term, porous storage scenarios.
Potential Outcomes of Carbon Capture Techniques
While IBM scientists have significantly improved the accuracy of porous media flow modeling, much remains to be done. The main task now is to improve the digital models, add information about the chemical processes and the use of substances that will contribute to the mineralization of CO 2 . The ultimate goal of the study is to find highly efficient, low-cost materials and scalable methods for the safe long-term storage of carbon dioxide.
The results of the work carried out in collaboration with researchers from the University of San Carlos and the Petrobras corporation showed that the proposed model can be effective. IBM Research has taken microtomographic images of rock samples and key algorithms for detecting capillary networks are publicly available so that other researchers can test and apply them in practice.
This work is part of a broader IBM Future of Climate initiative launched in 2020 to advance materials selection and knowledge building technologies through an international network of IBM research laboratories. This initiative is exploring and developing strategies to reduce the carbon footprint of cloud systems and supply chains, and to model the impacts of climate change.
Link to original material in English