Page 174 - Adsorption Technology & Design, Elsevier (1998)

162 Design procedures

of being able to invoke this assumption together with the isothermal
assumption are evident from earlier considerations in this chapter. Strictly,
the equilibrium assumption cannot be invoked for any adsorption process
since mass transfer resistances will inevitably be present in the system.
However, the assumption is often invoked for PSA air separation to produce
oxygen by adsorption on zeolites. This is because the separation in this case
is determined by the preferential equilibrium in favour of one of the
components (nitrogen). In contrast, the assumption must not be invoked for
the production of nitrogen using carbon molecular sieve since the separation
in this case depends on the differences in rates of diffusion of nitrogen and
oxygen. Nor should the assumption be invoked for rapid pressure swing
adsorption processes (Sircar and Hanley 1995). The assumption of equili-
brium control has limited applicability for gas phase separations and it is
doubtful whether it can be used safely for the design of any liquid phase
separation.
Axial dispersion was often ignored in early modelling of PSA separations
not only because the material balance equations, e.g. equation (6.37), could
be simplified substantially, but also because the diffusional fluxes were
generally, but not always, small compared with the convective fluxes. This,
however, may not be the case for vapour phase TSA separations and for
liquid phase separations.
Despite the wide choice of mass transfer models which are available, the
simplest and most popular adsorption rate expression is the linear driving
force model because it represents actual processes reasonably well and
reduces the computational effort required. An example of how this and
various other simplifications and empirical correlations can be incorporated
into the design and analysis of pressure swing adsorption processes is
provided by White and Barkley (1989). The example used is the drying of
air. Examples of how simplifying assumptions can aid the modelling of PSA
air separation processes is provided by Knaebel and Hill (1985) and by
Kayser and Knaebel (1989). Further information on cycle models can be
found in Ruthven (1984), Yang (1987), Ruthven (1990) and Ruthven et al.
(1994).

6.6 CONSTANT PATTERN BEHAVIOUR
It is fortunate that for many fixed bed adsorption processes of commercial
interest the shape of the mass transfer zone remains unaltered as it
progresses through the majority of the bed because this leads to substantial
simplifications in design. For a favourable isotherm, particularly one of
Type I, the mass transfer wave spreads from a shock front as it progresses
through the initial region of the bed. As explained earlier in this chapter the