Page 390 - Design of Simple and Robust Process Plants

9.5 Closed Loop Steady-state Optimization 377
. The optimization cycle applied for the implementation of OO is discussed for
the individual modules: data analysis, data reconciliation, simultaneous data
reconciliation and parameter estimation, optimization and implementation
of set-points. Each module is provided with an input block and a decision
block after execution of its task.
. Data analysis primary tasks are described as: removal of corrupt measured
data; steady-state detection based on defined criteria; and decision on pro-
gress to the next sequential module.
. Date reconciliation is based on the minimization of the deviation of the
square of the differences between the reconciled values of measurements
and the measured values. The model is based on the reconciliation of mass
and heat balances. The calculated values are used for the performance meas-
urement as described before, and to define the feed rates and composition of
the process. The minimization can be done using the Lagrange multipliers
method. During reconciliation, measurement errors can be both detected
and eliminated.
. Simultaneous data reconciliation and parameter estimation is done on the
entire process model to determine the value of those variables (parameters)
that are subject to change over time or capacity. For this step an accurate vali-
dated model is essential, including constraint descriptions. The parameter
estimation will be done in concert with data reconciliation of the full model,
despite the determination of the feed rates and composition in the previous
step. The result of this step are: detection and elimination of measurements
with gross errors, estimated parameters (some of these parameters may just
be calculated due to a lack of redundant measurements); simulated perfor-
mance in comparison to the measured performance.
. Optimization of the updated model (adjusted parameters)must be done for
DOFs which are accessible. Verification of the results and submittal for imple-
mentation to the implementation module must pass the final decision block.
. Implementation of set-points that are made available by the optimization is
done through the implementation module. The module has the possibility to
define the ramp-up rate of the individual set-point and the schedule for
implementation
. Overall sequence and operator interface. An executive controls all the individ-
ual steps of the implementation, and requires a carefully designed interface
up to operation that in one picture has a complete overview of the system
status and its actions.