Page 357 - Design of Simple and Robust Process Plants

8.4 Control Design 343
flow reactor which contains the finely suspended catalyst, while the heat is removed
by evaporation of the product. The reaction (which proceeds to extension) takes place
at the liquid±solid (catalyst) interface. Due to the large heat of reaction, a solvent is
introduced at the feed point to flatten the temperature profile, and the solvent is also
removed in the vapor phase. The solvent is split and supplied as quench stream at
two places in the column. The control of the temperature over the reactor is critical
as it determines the conversion which must go to completion while it is constrained
by a run-away reaction which will develop at higher temperatures. The temperature
profile goes to a peak value at one third of the height, where it is quenched with the
solvent.
The system is pressure-controlled at a fixed setting. The control system as origin-
ally designed with flow rate controllers for the reactants which required adjustment
during start-up and shut-down. Apart from the criticality of the control of tempera-
ture profile, the liquid hold-up was critical. Any deviation in the heat balance of the
system results in accumulation of liquid in the reactor, with potential boil-over or
drying out of the top part of the system. This would lead to too low a conversion and
interruption of the catalyst recycle. The physical phenomenon which impacts the
reaction is the gas/liquid flows in the reactor. The gas and liquid flows determine
the hold-up on the trays (which is very sensitive to these flows), while the flow
regimes also change upward in the column due to the volumetric changes of the gas
and liquid flows. The hold-up impacts the conversion of reaction.
The whole system was in hands of the operators, who had six basic loops avail-
able: feed and solvent flow rates; ratio controller; pressure and level controllers.
There was no feed-back controller for conversion/temperature profile control, heat
balance compensation to keep hold-up in the system. Neither was there any type of
coupling for capacity ramping. The operator had multiple set points of basic loops
available for manipulation, all of which had a high level of interaction. A trip system
was available to cope with temperature excursions. Thus, the operator had a difficult
task to operate the system at design conditions, as any upset would cause an off-
spec product as a result of too low conversion.
The challenge was to develop a control design to unload the operator with capacity
control to ramp the process up, and provide feed-back controllers for conversion
through temperature profile control and energy balance control by liquid hold-up
regulation. The control design was planned to be executed at basic control level, and
did not include any predictive or constraint controller. The design started with the
development of a dynamic reactor model. The kinetics and overall conversion rates
were collected from the literature, while physical properties were obtained from
public data banks. The reaction kinetics had to be implemented in a tray reactor
model, and were described in stages. The tray model had to include hold-up calcula-
tions which were subject to gas and liquid flows and heat and mass transfer terms.
The model was written in gPROMS (Process Systems Enterprises).
The model was verified at basic reactor conditions. This was also accepted as the
nominal operation point with regard to conversion and peak temperature level and loca-
tion. These last conditions were constrained by safety aspects and the physical system.