The complexity of the present-day automation in a process
plant can be characterized by the large number of recording controllers
interconnected by the flow diagram on the centralized control panel board in a
process control room. On the control board, the process is divided into several
interconnected unit operations, such as reactor, heat exchanger, etc. Associated
with each unit operation, there are a number of measured variables, such as temperatures,
pressures, and flows. Some of these variables are also controlled by individual
controllers and can be controlled within a certain range by adjusting the controller
set points on the control panel. Although the number of variables being
controlled, this way is relatively large, there are still process disturbances
which cannot be controlled, such as ambient temperature, variation of feed quality,
and aging of catalyst. The optimum way to operate the process is generally a
function of these disturbances. It is the operator's job to sense these
disturbances, estimate the new optimum operating conditions, and readjust the
set points of the controllers accordingly. This describes the function of the
present-day controllers in a process plant and their relation to the operator.
A. Optimization. The fundamental objective of a
process operation is to make a profit. The optimum operating condition is
defined as a particular combination of process variables which causes the
process to yield maximum profit. Because if the uncontrolled disturbances, the
optimum operating condition may change from time to time. The conventional
regulating control philosophy is based on the assumption that either the
optimum operating condition does not change much or the effect on profit resulting
from the change of the optimum is small. B.Multivariable System. The basic control phylosophy in applying a process
control computer is to control the process as a whole. Since the performance of
a process is generally dependent on a number of important process variables
which are not independent of each other, the control problem falls into the
category of multivariable systems. The multivariable problem distinguishes
itself from the single-loop control problem by the fact that each output
variable is a function of all the input variables, and a change in any input
variable causes changes in all the output variables. One of the control
problems the process computer must solve is how to handle this multivariable
operation in a logical and optimal manner. C. Constraints on Process Variables.
Constraints are defined here as the upper or lower limits set for each process
variable under normal operating conditions. There are two basic types of
constraints to be considered. One kind is the constraint due to physical or
equipment limitations and the other is that due to product specifications. For
example, the maximum wall temperature of a furnace is a physical constraint and
the minimum octane number of gasoline is a product specification constraint.
During normal operation, a basic requirement of the process control system is
that neither type of constraint be violated.
D. Estimation of Process Variables. One of the basic requirements
of a good feedback control system is accurate measurement. The measurement
problem encountered in process control has been a continuing challenge to the
instrument industry for a long time. To date, there are still serious limitations
or the scope of process control due to lack or inadequacy of measurements. E.
Disturbances.The need for any feedback
control system
arises mainly because there are unpredictable and uncontrollable
disturbances affecting the operation of the process. Therefore, the
characteristics of the disturbance have a very important effect on the nature
of the computer control problem.