Computer Simulation Helps Solve SO2 Scrubber Problem

in Primary Aluminum Smelter Plant

Chang-LI Hsieh, Koch-Glitsch, LP, Wilmington, Massachusetts
Frank Hartgers, Hoogovens Technical Services, IJmuiden, The Netherlands
Peter Mihalik, Hoogovens Technical Services, Burlington, Ontario
Greg Boles, Koch-Otto York, Inc., Wichita, Kansas

Computer simulation helped to solve a challenging problem in scrubbers used to remove sulphur
dioxide from gas in the production of aluminum at the Reynolds Metal Company, St. Lawrence
Reduction Plant, Massena, N.Y. Tests showed that under conditions requiring high gas flowrates,
the mist eliminators did not completely remove fine mist and liquor droplets in the gas stream.
This resulted in potentially acidic moisture droplets being emitted to the atmosphere. Engineers
simulated the operation of the scrubbers using computational fluid dynamics (CFD) and discovered
a significant gas maldistribution in the absorber. This was due to the geometry of the tower and inlet
ducting. In further CFD work, they evaluated several alternate tower and ducting designs, with
involvement of HTS and Koch engineers, and selected one that solved the maldistribution problem at
minimum expense.
Engineered and installed by Hoogovens Technical Services, the gas desulphurization scrubbers used
at the aluminum smelter consist of two 51 foot, 6 inch diameter stainless steel vertical absorber towers
that spray a sodium sulphate/sodium carbonate solution countercurrent to the gas flow. After the gas
has contacted the solution, it passes through a mist eliminator to remove any fine mist and liquor
droplets remaining in the gas stream. The FLEXICHEVRON Mist Eliminators used in this application,
produced by the Koch-Otto York division of Koch-Glitsch, consist of a series of turning vanes that force
the gas to change direction so that inertia causes droplets to impinge on the blades where they are
removed from the gas stream. These mist eliminators collect essentially 99+% of all particles between
8 and 40 microns in diameter, depending on design parameters.

Moisture Carryover Experienced

The normal gas flowrate specified through the mist eliminators is 1,099,600 actual cubic feet per
minute (ACFM) through each of the two absorbers. During periods when one of the absorbers is not in
operation, a single absorber is specified to handle 2,034,870 ACFM. During air/water trial operations,
moisture carryover was visually witnessed at a flowrate of 1,300,000 ACFM, well below the specified
design conditions for single unit operation. At the request of Hoogovens, the mist eliminator
manufacturer was brought in to help solve the problem. It is important to note that mist eliminators
are capable of effectively removing droplets only when velocity is maintained at or below their design
limit, which in this case was 17-18 feet per second.

While the scrubbers had been designed so that average velocity was well within these limits, both
Hoogovens and Koch-Glitsch engineers were concerned that geometrical irregularities might be
causing the FLEXICHEVRON mist eliminator velocity limits to be exceeded in certain areas of the
scrubber.
This problem would have been very difficult to resolve using conventional physical testing methods.
First of all, it would have been very difficult to equip the scrubber with enough sensors to determine
whether and where velocity limits for the mist eliminators were being exceeded. Secondly, assuming
that measurements showed that excessive velocity was the problem, the testing results would provide
little or no guidance to engineers in determining the cause. Finally, designing and testing a solution
would be very expensive and time-consuming using the physical testing approach. It would be
necessary to modify the scrubber and perform another series of tests for each hypothesised solution to
the problem. For all of these reasons, CFD simulation was adopted to investigate the problem and
examine proposed solutions.

CFD advances
In recent years, advances in CFD simulation have made it possible for engineers to generate a relatively
fast and inexpensive computer model of flow within a scrubber. CFD involves the solution of the
governing equations for fluid flow, heat transfer and chemistry at several thousand discrete points on a
computational grid in a defined flow domain. The use of CFD enables engineers to obtain solutions for
problems with complex geometry and boundary conditions. A CFD analysis yields values for fluid
velocity and temperature throughout the solution domain. Based on the analysis, a designer or
engineer is able to optimize fluid flow patterns or temperature distribution by adjusting either the
geometry of the system or the boundary conditions, such as inlet velocity/temperature and wall heat
flux.
Koch-Glitsch engineers selected FLUENT CFD software, from Fluent Incorporated, Lebanon, New
Hampshire to perform the analysis. FLUENT was selected because it has a powerful range of choices
for modeling the turbulent flows seen in scrubbers and other process equipment.

Simulation matches experiments

The simulation of the scrubber performed by Koch-Glitsch
engineers showed, as was suspected, that there is a significant
vapor maldistribution in the absorber due to the geometry of
the tower and inlet ducting. The countercurrent liquor spray
in itself did not provide enough pressure drops across the
tower to equalize vapor flow, and the subsequent vapor
velocity through the mist eliminator. The result was that
vapor velocities were above 20 feet per second just prior to
entering the mist eliminator. The analysis showed the high
velocity flow concentrated in the area directly opposite the
inlet duct. These results match the evidence that Hoogovens
and Koch-Glitsch engineers obtained, which showed moisture
coming out of the stack in that area.
Computer Simulation Helps Solve SO2
Scrubber Problem

Page 2

The CFD results further confirmed and quantified

various design features which the team suspected
to be causing the flow-maldistribution. In
particular, they examined
two-dimensional plot with velocity vector contours
on the symmetry plane of the scrubber and
ductwork. This plot provided an excellent birds-eye
view of the flow through the scrubber and helped
to almost immediately gain an intuitive
understanding of what was causing the problem.
The plot showed that a "rain hood" around the
scrubber inlet helped to focus the gas flowing
through the duct into a high-speed flow path that
shot across the width of the scrubber until it hit the opposing wall. Once it hit the wall, the flow was
forced upward directly into the mist eliminator. Understanding the flow within the scrubber and
both inlet and outlet ducting helped to focus the teams' attention on the inlet ductwork and inlet
transition to the scrubber. Hoogovens engineers realized that modifying the gas flow within the inlet
duct itself and at the junction of the inlet duct and scrubber would greatly impact the performance of
the Koch-Otto York mist eliminator.

Iterating to a solution
With the problem now well understood, iterating to a solution was a relatively straightforward process.
Since the rain hood was clearly identified as a major contributor to the velocity maldistribution, one of
the first things tried was removing the hood from the model and re-running the analysis. This change
provided a major improvement in velocity distribution and reduced the maximum velocity to just
below the limit for the mist eliminator. Hoogovens engineers also suggested smoothing the bottom of
the transition duct by dropping it to the maximum open position allowable from a construction point
of view. Engineers tried a variety of other changes in an effort to further improve the velocity
distribution and to gain a larger margin of safety. They tried a variety of different baffles in the inlet
ducting in order to diffuse the flow entering the scrubber. The most obvious position for the baffle was
right on the inlet plane, and the analysis showed this geometry to be very effective in reducing
maximum velocity. However, Hoogovens engineers ruled this out because it would be difficult to
install and support the baffle plates.
Koch-Glitsch engineers then tried a range of other
baffle positions based on Hoogovens engineers
suggestions with the goal of finding a low-cost,
low-maintenance solution. They finally found a
position in the ductwork, well upstream of the inlet
that provided a very even velocity distribution.
Hoogovens engineers agreed that it would be easy
to install a baffle in this position and that virtually
any type of maintenance work could be performed
Computer Simulation Helps Solve SO2
Scrubber Problem

Page 3

without removing it. The analysis of this new configuration showed that the velocity in the
Y direction was 16 feet per second or less at all areas of the cross-section of the scrubber at the inlet to
the mist eliminator. The results were accepted and Hoogovens proceeded with the necessary
modifications in a very timely manner.
What would have otherwise required a long and expensive build and test process was resolved
in about two weeks. Using the traditional approach, engineers would have wasted months
trying to solve the problem by changing the
geometry in the area where the
maldistribution occurred. Instead, using the
simulation results as a guide, Hoogovens
and Koch-Glitsch focused immediately on
the inlet duct area that the analysis results
indicated was the root cause of the problem.
In addition to obtaining a quick and
inexpensive solution, the customer gained a
new level of respect for the analytical
capabilities of the two suppliers involved in
the project.

Computer Simulation Helps Solve SO2

Scrubber Problem

Page 4