Page 403 - Design of Simple and Robust Process Plants

390 Chapter 9 Operation Optimization
meter will initially be based on the average data over a longer time period to smooth
the dynamic effects.
The next step will be the selection and implementation of the Steady-state Repre-
senting Values (SRVs)(Krist et al., 1994). The selection of SRVs is based on process
insight, and should be representative if the overall process is to be considered at
steady state. Some guidelines for selection include:

. Major process flows
. Pressure is a good indication of stability; preferred pressures are inlet and
outlet pressure of units, top pressure of distillation columns and pressure
difference representing the process condition of the unit operation (not the
head of a pump or liquid levels).
. Suction pressure and revolutions in time of compressors (including refrigera-
tion machines). In the case of a floating suction pressure, the discharge pres-
sure might be selected.
. Temperature measurements representing the overall unit performance, like
inlet and outlet temperature of an adiabatic reactor and a peak temperature
(not necessarily fixed to a location), the temperature of a CSTR reactor. Do
not take a temperature measurement which might be impacted by inertia or
heavy streams of which the composition might change (and thus its tempera-
ture). Temperatures that are influenced by pressure should have a pressure
compensation before use as SRVs. Specifically, vapor liquid equilibrium
units which are subject to floating pressure operation.
. Quality measurements on essential process streams having an impact on
product distribution and quality.
. Units with a very large response time; for example, liquid±liquid extraction
columns, distillation columns with large numbers of trays and high reflux
ratios should be judged on the stability of product quality and less on recycle
quality. This to avoid too much time being lost before the next optimization
cycle can be implemented.
. Levels in accumulators are purposely varied, and therefore should not be se-
lected as SRVs (as for pump pressures).
Implementation of the data analysis and development of the filtering and stability
criteria can only take place after the control design improvements have been imple-
mented. Control design as required for steps 5C, 6C, 7C is discussed in Chapter 8,
while the design of model based controllers step 8C is considered outside the scope
of this work. .

9.6.6
Development and Validation ofReactor Model(s): Step 5M

The development of the reactor model is, for most process optimizations, one of the
crucial steps. The reactor in most processes has a major impact on the overall opti-
mization. Several different ways exist to deal with reactor models, including: