This thesis considers the development of predictive thermodynamic models for
amine-based carbon capture processes, motivated by the imminent requirement
for the reduction in anthropogenically produced carbon dioxide emissions.
In the introduction, we show how the use of molecular-based equations of state,
such as SAFT (Statistical Associating Fluid Theory), can be highly effective in this
context. Due to the level of molecular detail captured in their theoretical development,
one can reduce the reliance on experimental data by transferring their parameters
based on sound physical arguments. In particular, the inherent chemical
reactions in amine-based carbon dioxide absorption processes can be modelled by
a physical association scheme, offering a vast simplification over the conventional
treatments.
In the following chapter a rate-based absorber model is presented to investigate
the reactive capture of carbon dioxide CO2 using aqueous monoethanolamine
(MEA) as a solvent. The SAFT-VR SW equation is used as the thermodynamic
model. Due to the physical treatment of the reactions, the process model equations
only needs to consider the apparent concentrations of the molecular species,
while the reactions are implicit in the SAFT equation. With the assumption that
the species diffuse as non-associated species, the rate of CO2 absorption is overpredicted,
providing an upper bound on the solvent performance. A single parameter
is adjusted to the pilot plant data, reflecting the reduction in mass transfer
rate in the apparent CO2 in its aggregated form, which is found to be transferable
over all of the pilot plant runs.
The development of new models for the SAFT-
Mie equation is then considered
for improvement of the thermodynamic model. This is because the thermodynamic
models developed for the SAFT-VR SW (used in the absorber) provide
an inaccurate description of the liquid heat capacity and the heat of absorption
of CO2. We consider a novel approach to the parameter estimation problem. It
is shown that posing the parameter estimation as a multi-objective optimization
problem offers numerous advantages over conventional (single-objective optimization)
techniques. A robust and efficient algorithm that deals with multiple objective
functions is tailored for this purpose. We consider objective functions that
characterise the deviation between the SAFT model and experimental measurements
for different thermodynamic property types. Using the multi-objective optimization
technique we develop SAFT-
Mie models for water where saturated
liquid denisty, vapour pressure and isobaric heat capacity are treated as competing
objectives. A single (non-spherical) model for water is chosen from the Pareto
fronts obtained.
Next, we develop SAFT-
Mie (or SAFT-VR Mie) models for the CO2 + MEA
+ H2O mixture, with focus on developing models that provide a simultaneous
accurate description of the vapour-liquid equilibria and the caloric properties. In
comparison with the previous models developed for the SAFT-VR SW equation of
state, the new models provide a better description of key thermodynamic properties
in the chemisorption process, in particular the vapour pressure of CO2 above
v
the solvent mixture, the isobaric liquid heat capacity and the heat of absorption.
We show that incorporating the new thermodynamic models in our process model
for the absorber, we obtain a slightly better prediction of the column temperature
profile.
In the last chapter we derive a classical density functional theory (DFT) that
incorporates the SAFT-VR Mie equation of state (SAFT-VR Mie MF DFT). The proposed
method is applicable for a wide variety of fluids, including fluids/ fluid
mixtures that consist of associating molecules and molecules of varying chain
length. We derive a theory that is numerically tractable and show a novel implementation
of the DFT equations in gPROMS. We show that the theory can be used
to accurately predict the experimental interfacial tension for the SAFT models developed
in this thesis, and the predicted density profiles in the intefacial region
compare favourably with molecular simulations. The SAFT-VR Mie MF DFT approach
developed in this chapter is used throughout the thesis for validation of
the thermodynamic models.