Core Research Projects: Capture, AC5 – Reduced Order Models (ROMs)

Why is this research needed?

Algorithms are being developed to simplify and speed up the computational models used in CCS research. This is done through creating metamodels (which are simplified models of more complex models), which once complete, will be used to analyse how various carbon capture technologies can impact the UK energy system as a whole. These metamodels will enable easier and quicker estimation of the cost and performance of CCS technologies. Policy makers could also use this as a tool to inform their strategic decisions about how CCS works at a system level.

What is this research investigating?

We are developing new and simpler models, which will be able to accurately predict the behaviour of post-combustion capture process. The complexity of process models accurately simulating post-combustion capture (PCC) systems prohibits their direct inclusion into a system wide assessment. Here, we will produce robust, accurate reduced order models (ROMs) for PCC technologies using solvents and solid adsorbents to bridge this gap and represent the dynamic operation and performance of the capture technologies under operating conditions within the larger energy system model.
ROMs will be developed in three stages:

• Consultation on the required quantities of interest (QoI) required as outputs, such as capture level/rate, specific energy consumption, CAPEX, OPEX and design parameters which can a priori be ignored;
• A Global Sensitivity Analysis to quantify the parameter effects on QoIs and rank their importance and decide which parameters can be removed. These methods have the benefit of being dimensionally nested, meaning that the remaining parameter sample sets can be re-used in the fitting of the ROMs even once variables are removed;
• The ROMs will be polynomial chaos expansions or Kriging/Gaussian processes that can easily be entered into other codes. They will be developed initially using the sample set at the previous step and then improved by adaptively enriching the sample set to include information where error exists in the parameter space, either due to sparsity of sample points or rapid changes in the underlying model.

What does the research hope to achieve

The University of Sheffield will benefit from detailed dynamic models for PCC based on chemical absorption using the benchmark solvent monoethanolamine (MEA) that have been developed in gPROMS, validated, scaled up and published. These detailed models will be reused for this project, with potential resimulation where necessary. The ROMs will feed forward into the systems research project CAB1 Cross-cutting value of CCS.

The outcomes from this research will contribute to the development of a more realistic assessment of different deployment scenarios of CCS systems.