Numerical modelling of the CHEMREC black liquor gasification
process: conceptual design study of the burner in a pilot
gasification reactor

Abstract: The work presented in this report is done in order to develop a simplified
CFD model for Chemrec´s pressurised black liquor gasification process. This
process is presently under development and will have a number of advantages
compared to conventional processes for black liquor recovery. The main goal
with this work has been to get qualitative information on influence of
burner design for the gas flow in the gasification reactor.

Gasification of black liquor is a very complex process. The liquor is
composed of a number of different substances and the composition may vary
considerably between liquors originating from different mills and even for
black liquor from a single process. When a black liquor droplet is gasified
it loses its organic material to produce combustible gases by three stages
of conversion: Drying, pyrolysis and char gasification. In the end of the
conversion only an inorganic smelt remains (ideally). The aim is to get this
smelt to form a protective layer, against corrosion and heat, on the reactor
walls.

Due to the complexity of gasification of black liquor some simplifications
had to be made in order to develop a CFD model for the preliminary design of
the gasification reactor. Instead of modelling droplets in detail,
generating gas by gasification, sources were placed in a prescribed volume
where gasification (mainly drying and pyrolysis) of the black liquor
droplets was assumed to occur. Source terms for the energy and momentum
equations, consistent with the mass source distribution, were derived from
the corresponding control volume equations by assuming a symmetric outflow
of gas from the droplets and a uniform degree of conversion of reactive
components in the droplets. A particle transport model was also used in
order to study trajectories from droplets entering the reactor. The
resulting model has been implemented in a commercial finite volume code (AEA-
CFX) through customised Fortran subroutines. The advantages with this simple
model are that no detailed information about the kinetic behaviour of the
fuel is necessary and that it is numerically well behaved.

The magnitude of the degree of conversion in the model was optimised during
the solution by minimising the difference between the computed and a target
value for the volume average temperature in the upper half of the reactor.
The optimisation was done iteratively with the aid of a modified version of
Brent’s algorithm that was implemented in the Fortran code.

The primary use of the model was to study the influence of spray angle,
hollow or full spray pattern, and degree of swirl for the burner. The
results were post processed to yield, in addition to the velocity, pressure
and temperature field, information about the magnitude of the degree of
conversion, the residence time of particles of different sizes, heat flux to
the walls and average temperatures. The optimum spray pattern according to
the approximate model was found to be the wide angle, full cone pattern with
moderate swirl.