Page 339 - Design of Simple and Robust Process Plants

8.4 Control Design 325
The number of MVs is not always equal to the number of CVs This can be solved
from two sides. Sometimes, an additional MV can be created; for example, when the
duty of a heat exchanger cannot be controlled, a bypass might be provided. Another
example is when a feed composition has a relevant impact on an controlled variable,
a separate feed system might be installed. Another option to be explored is to evalu-
ate the CVs carefully, as on occasion we simply do not have any capability to control
it ± in which case it needs to be removed or solved in a multivariable, model-based
environment.

8.4.6
Determine the Controlled, Manipulated, Measured and Disturbance Variables

Manipulated and controlled variables are respectively input and output variables
which can be selected by qualitative criteria as formulated by Newell and Lee (1988).
Controlled variable (CVs) selection might preferably follow the guidelines:
. Select variables which are not self-regulating; variables which result in a new
steady-state situation without a feed-back control should not be selected. An
example of a self-regulating system is an hydraulic overflow system. As is
often applied in decanter systems, this is termed a simple self-regulating sys-
tem.
. Select variables which exceed the constraint of the process. Safety constraints,
as well the tendency to operate close to its constraints under optimized condi-
tions, requires that controlled variables are selected which exceed the con-
strained values.
. Choose variables which directly measure the condition and quality of the pro-
cess streams, or that have a strong relation to it. As composition measure-
ments are expensive and often have long response times, the preference is to
select variables that are easy to measure with short response times, but
which have a direct relation to the property of interest.
. Selection of output variables which have a significant interaction with other
controlled variables are preferred. Closed loop performance is improved by
stabilizing control variables which interact significantly.
. Select output variables which have favorable static and dynamic responses
compared to other controlled variables.
. Variables selected for control should not correlate too closely.
The selection of manipulated variables (MVs) must follow certain guidelines:
. MVs must affect the CV strongly, which means that it requires a large steady-
state gain. This results in rapid responses and less variations in flow of the
manipulated stream, and less potential impact on other units.
. Operational range of the manipulated variable is a characteristic which need
to be respected. A bypass valve around a heat exchanger might give a limited
controllability of the exchanger duty.