Modelling Phase Behaviour of Impure Carbon Dioxide: An update from Richard Graham’s Call 1 Project

By Dr Richard Graham, Lecturer in Applied Mathematics, University of Nottingham

Principal Investigator on Call 1 Project – Tractable equations of state for CO₂ mixtures in CCS: Algorithms for automated generation and optimisation, tailored to end-users

Carbon capture and storage is a crucial technology in the international efforts to meet carbon dioxide emission targets. Capturing carbon dioxide from industrial sources can lead to a 90% reduction in emissions. However, no gas separation process is 100% efficient, and as a result the carbon dioxide generated contains a number of different impurities, depending on its source. These impurities can, depending on their composition and concentration, greatly influence the physical properties of the fluid compared to pure CO₂. They have important design, safety and cost implications for the compression and transport of carbon dioxide and its storage location, for example geological sequestration.

Our research is designed to tackle one of the key technical challenges facing the development of commercially viable CO₂ transport networks: modelling phase behaviour of impure carbon dioxide, under the conditions typically found in carbon capture from power stations, and in high-pressure (liquid phase) and low-pressure (gas phase) pipelines. Accurate modelling of the physical properties of CO₂ mixtures is essential for the design and operation of compression and transport systems for CO₂.

Models for phase behaviour are known as equations of state (EoS). EoS vary in their mathematical form, accuracy, region of validity and computational complexity. Because different applications have different requirements there is no single EoS that is ideal for all applications. To optimise their accuracy, EoS need to be calibrated by fitting their parameters to experimental measurements on carbon dioxide mixtures. However, new measurements become available very frequently, offering the opportunity to improve the models. Thus, there is an ongoing need to regularly rederive, refine and reparameterise EoS. Unfortunately, parameterising an EoS is a laborious and time-consuming task by hand and is not easy to automate on a computer. This delays or prevents knowledge gained from experiments from being applied in CCS modelling.

In this project we are using cutting-edge computer algorithms to automatically reparameterise EoS for CCS modelling. This flexible technique will allow a user to specify their requirements and rederive model parameters matched to their needs. The attached figure shows and example of an EoS that was parameterised by our methods, compared to experimental data for pure CO₂. We are also evolving the technique so that our algorithms go beyond parameter fitting and begin, instead, to directly produce functional forms for EoS from experimental data, thus fully automating the derivation of EoS. These new EoS will rigorously account for uncertainties in both the modelling and the experiments. This will allow effective control and minimisation of risk in CCS processes, which will improve CCS regulation and safety. Our algorithms will enable users to rapidly produce bespoke EoS, tailored to their particular application. It will also enable these models to continually evolve as new measurements become available, ensuring that experimental advances are rapidly converted into improved CCS modelling and, ultimately, better performance and efficiency of real CCS processes.