Page 372 - Design of Simple and Robust Process Plants

9.4 Performance (Profit) Meter 359
If the results are integrated over a longer time period, the variations in hold up with-
in the process will have a reduced effect. If storage is included in the flowsheet, the
variations in storage are considered to be captured in the model.
Hold-up variations in the process might be included in the flowsheet model by
defining a vessel with hold-up measurements (level).
Process stability has a direct impact on performance meters with a relative short
integration time, as used for CLOs. The only way to achieve this is through
improved control. This would make it possible to narrow down the criteria for
steady-state detection and so would impact on the calculated value over smaller time
intervals. The noise of the signal (performance)would be reduced and as such make
it a more accurate measurement for tracking of CLOs.
The model is mostly written in the same simulator (equation-based with open
equations)as the process model and the design of the performance meter is com-
bined with the data reconciliation step. A stand-alone performance meter is de-
signed to be simpler in which case the optimization routine is written explicitly to
avoid installing an extended simulator/optimization package.

Summary
In general, operation optimization of continuous processes include:

. The objectives for process plants are to operate continuously at the most eco-
nomical conditions in a continuously changing environment of: feed product
and energy prices, capacity, feed compositions, product qualities and distribu-
tion, aging/fouling factors of process equipment and catalyst, meteorological
conditions, and transient operations.
. OO might be executed off-line versus closed loop; off-line optimization is pre-
ferred when the conditions for the plant do not change frequently (daily
basis), while the number of DOF is limited.
. CLO follows a sequential operational cycle to be discussed in detail in the
next section.
. OO savings to be realized are in the order of percentages of the variable opera-
tional margin, while capacity increments may fall in the 5±10% category.
. The level of accuracy required for OO models is rather high, in the order of tenths
of a percentage of the operational cost. The following statement supports this:

Any inaccuracy of an optimization activity results in a missed opportunity

The validation of OO models to achieve the required accuracy next to model building
is a very conscious activity.
. Performance measurement is a detection method to measure the variable
operating margin of an on-line process based on reconciled mass and heat
balances and prices. The measurement is designed continuously to track the
performance of a process, and can be used to detect mismatch between actual
and simulated performance; thus it is an ultimate validation of the simula-
tion. The design of an accurate performance meter is discussed.