Page 388 - Design of Simple and Robust Process Plants

9.5 Closed Loop Steady-state Optimization 375
The constraints set are, or are coming from:

. operation such as the capacity of the process plant or product specification or
receipt; these are communicated through the basic control system.
. from the control system, in case of operational constraints as experienced by
the constraint controllers; this might be the maximum capacity of the feed or
the maximum load of a refrigeration system.
. hard constraints already fixed in the program as design pressures, tempera-
tures, product specifications.

Optimization Based on a robust model, the optimization is just run time, which
limits the cycle time. The optimization always takes much more time than the pre-
vious steps if we exclude the time that the process takes to comply with the steady-
state criteria for plant stability.

Decider 4 The decisions to be made before the new optimized set-points are relea-
sed for implementation to the control system are:

. Is decider 1 true?
. Is decider 2 true?
. Is decider 3 true?
. Did the optimization converge?
. Are the calculated optimal set-points within the defined feasible range?
. Are the DOFs selected for implementation still available for implementation?
. Is the difference in projected performance versus last implementation opti-
mization meeting a certain criteria? The calculated simulated profit might be
within the noise range of the operation. The implementation of new set-
points introduces a disturbance into the process, which always has some dis-
advantage. Therefore, a criterion might be defined based on projected saving
versus the last implementation.

9.5.2.5 Implementation module
This downloads set-points that are made available from a latest optimization results.
In this module are defined the ramp-up rates and the schedule of implementation
at BC and MBC levels. In particular, the scheduling requires careful design as this
might lead to hazardous or unstable areas.

9.5.2.6 Overall sequence and operator interface
The overall sequence is operated by an executive, that operates according to a sche-
dule. The cycle as discussed above has the steady-state detection module always oper-
able as long as the process plant is in the run step. The reconciliation step is also
always operable, even if decider 1 is not true, but the process need to be in the run
step. This means that the performance meter that is coupled to the reconciled mass
balance operates, even if the plant is not steady enough from an optimization per-
spective. The mass balance reconciliation will have a correction for any variation of