Process Control
Process Control
boundaries. The main purpose of a process control is to maintain a process at desired operating
conditions while taking process constraints into account.
Process control strategies can be classified and organized in a hierarchy of process control
activities.
The fourth step involves specific decisions on control structure, algorithms, and tuning. Here, if
single-loop control technology is used, the single-loop controlled and manipulated variables are
paired, and the modes of the PID controllers are specified. In addition, special requirements for
the tuning are made in conjunction with the pairing. For example, level controllers are specified
as tight or averaging. Also, tight and loose tuning of interacting loops is specified, to reduce the
effects of unfavorable interaction while retaining the beneficial effects of favorable interaction, as
required. Hierarchy and decomposition are used in managing the design decisions.
Step 5: Optimization
The fifth step determines whether optimization opportunities are available after consistently high
product quality has been achieved and, if so, whether additional manipulated variables, not used
for control at previous steps, exist. It may be necessary to add sensors to provide information for
optimization and to automate additional manipulated variables for optimization. If opportunities
exist, an analysis is performed to determine the economic benefits which can be realized through
optimization. If significant benefits are available and can be realized through real-time control, the
strategy is designed at this step.
The sixth and final step evaluates monitoring and diagnostics. At this step, the major analysis is
the sensors required for this function. In addition, any calculations required for the monitoring
are defined.
The sequence of steps is selected to maximize information gathering and understanding at the
early steps and to reduce the need for iterations.
Design of controllers up to and including product quality is done in the fourth step which is control
structure to give the best performance for the important variables. Safety controls for safety should
be designed at this stage. In the fifth stage, optimization, the remaining degrees of freedom which
are not used at the previous steps are being used for profit maximization. Finally, monitoring and
diagnosis is designed.
In this section the activities in the fourth step in the sequence, addressing control structure, are
presented in greater detail. Proper design relies on an integrated analysis of the entire process or
plant under consideration; however, the integrated design may involve too many variables and
processes to be analyzed by currently available methods. Therefore, the engineer temporarily
separates the design problem into smaller segments, and if the interactions among the segments
are small, each can be analyzed individually to develop provisional control designs. It is important
to recognize that these methods are used only when required by the large scope of the problem and
that the methods employ approximations to simplify the analysis. It is essential that each decision
contribute to the good performance when considering all factors in the integrated process.
A common approach for decomposing the design decisions is based on a temporal hierarchy, as
originally suggested by Buckley (1964) and expanded here:
The hierarchy is commonly used because it is difficult to design controllers for product quality
without first defining how feed and product flows and process environments are controlled. Thus,
Here, the flows and inventories considered are for the "process" materials, which are used
to make the product. The flows of utility streams, such as fuel, cooling water, and steam, are not
specified here, because they are manipulated to achieve other control objectives. The structures
that control the process flows determine how the feed and production rates are specified and
whether flow rates are nearly constant or are likely to vary significantly.
The goal is to provide a design in which the overall material and component compositions
are stable without further control. Naturally, this does not imply that satisfactory performance is
achieved with only these controls, only that all material entering the process leaves the process at
steady state, which is a reasonable basis for further analysis. One controller should influence the
production rate; this is usually a flow controller at the beginning (feed) or end (product) of the
plant, although other designs are possible. Then, the liquid levels and gas pressures are controlled
in a manner to achieve a self-regulatory process.
One common technique to improve dynamic behavior is to provide a small purge to allow
inerts to behave in a self-regulatory manner; this design is common in spite of the economic losses
due to valuable materials also leaving in the purge. Control designs should ensure that feed
components are self-regulating, so that they do not accumulate in the process.
2. Process Environment
The second level addresses the process environment variables: pressure, temperature, feed
ratios, catalyst addition, and so forth. These variables have a great influence on the product quality
and are often manipulated, in a cascade structure, by the product quality controllers. Thus, this
level provides tight control of the environment by compensating for many disturbances, and it can
be adjusted by cascade feedback from higher levels.
Partial Control
Partial control is not a separate level in the control hierarchy, but concepts related to partial control
influence decisions in levels 2 to 4 of the hierarchy. Recall that partial control involves selecting
of a subset of variables that can be measured and controlled, so that all key variables remain within
an acceptable range as disturbances occur. To achieve partial control, the engineer seeks dominant
variables that strongly influence the process behavior, and when regulated, yield good process
performance.
The fifth level involves monitoring and diagnosis of process and control performance. This
includes rapid monitoring and reporting to plant operating personnel, as well as longer-term
monitoring for periodic analysis. Plant operations are influenced by decisions made at this level
through actions of plant personnel, usually after detailed analysis of likely causes of unusual
process performance. These decisions may not be implemented through the control strategies,
because they may involve variables, such as feed purchases and reactor regeneration scheduling,
that are outside of the purview of the continuous control system.
Optimization Step
There are no optimization objectives in the control design form. The control design to this
point has allocated all manipulated variables, except for v3 and v6, which were found to be
redundant for the previous control objectives. These valves provide no additional process
flexibility, except that of controlling some intermediate pressures in liquid flow lines. There seems
to be no reason to control these pressures, and ordinarily, these valves would be eliminated to save
equipment and pumping costs. In this case, the valves will simply be retained at their base-case
percent opening.
The reactor environment control options are evaluated to determine the best control design.
Each is discussed briefly as follows:
1. Design I, shown below, controls T5. The reactor temperature is affected by several
disturbances. These disturbances influence other measured variables before the reactor
temperature measurement responds; thus, the potentialfor enhancements exists. For
example, the measured fresh feed temperature T1 could be a feedforward variable, and the
feed temperature T3 could be a secondary cascade variable. As a preliminary decision, the
single-loop design T5 v2 is chosen with a PI algorithm. The resulting control of F1, T5,
and A2 is controllable.
2. Design II, shown below, controls the reactor composition A1. A more direct measure of
the reactor operation is the concentration of A, which can be controlled by adjusting valve
v2, although with slow dynamics. Therefore, the cascade design A1 T4 v2 is selected,
which gives good responsesto temperature disturbances. The resulting control of F1, A1,
and A2 is controllable.
Design I
The response of Design I is shown in figure below. The product composition (A2) and the
product flow rate (F5) experience only small deviations and return quickly to their set points. In
spite of the good behavior of these key variables, other variables experience large variations; notice
that the recycle flow rate changes dramatically. For this case, the reaction rate disturbance of only
10% requires recycle flow changes of about 75% to achieve a new steady state.
The reason for this large change can be understood by analyzing the dynamic behavior of
the total amount of reactant in this recycle system. The amount of reactant A entering the plant is
constant because the fresh feed flow is unchanged; also the percent of reactant in the product
leaving the plant is controlled. To achieve a new steady-state, the rate of reaction of A must return
Design II
The response of Design II is shown in figure below. Again, the product composition (A2)
and the product flow rate (F5) experience only small deviations and return quickly to their set
points. As discussed above for Design I, the accumulation of reactant A must reach zero for the
plant to achieve a new steady state. Also, the flows of A in and out are identical for both the original
and final steady states. Therefore, the reaction rates for the original and final operations must be
the same. In Design II, the analyzer controller Ai senses a change in concentration and adjusts the
feed preheat (effectively changing the reactor temperature) to control the concentration.
After a transient, the process returns to nearly the same flow rates, with the reactor
concentration and volume at their initial values. To return the concentration to its set point, the A1
controller increased the reactor temperature, thus maintaining the production rate of B constant.
This response returns to steady state faster, satisfies all performance objectives for F5 and A2, and
does not require excessive equipment capacity. Based on this analysis, Design II provides better
performance for the feed impurity disturbance.
Control Design II should be evaluated for all disturbances in the CDF; these others are
discussed briefly here but not plotted. Because of the T4 temperature controller, it performs well
for the +20°C disturbance in T2, with only very small deviations in the compositions and product
flow. The system experiences a rather large, but brief, disturbance when F8 increases in a step of
20°C. The maximum allowable short-term variations in the product flow F5 and the product
composition A2 are reached or slightly exceeded. If plant experience indicated that this disturbance
occurred frequently, a feedforward compensation for changes in T8, adjusting v7 could be added
to Design II. Finally, the response of a change in desired production rate, F5, is rather sluggish,
because the feed flow rate is manipulated manually, and the product increases slowly as the recycle
system responds, finally attaining steady state. This is a direct result of the problem definition,
because short-term variation in the product rate was stated to have negligible influence on the
process performance in the CDF.
The IAE for the product quality variable (A2) is 7.11 for Design I and 6.62 for Design II
for the feed impurity disturbance.
Since Design II has good performance for the key quality variable, has well-behaved
dynamics for all variables, satisfies the control objectives, and requires equipment with smaller
capacities, it is selected as the better control design for this process.
References:
Downs, J., "Distillation Control in a Plantwide Control Environment," Chapter
20 in Luyben, W (ed.), Practical Distillation Control, Van Nostrand Reinhold, 1992.
Downs, J., and E. Vogel, "A Plant-wide Industrial Control Problem," Comp.
Chem. Eng., 1993.
Fogler, S. , Element s of Chemical Reaction Engineering, 3rd ed., Prentice-
Hall, Englewood Cliffs, 1997.
Luyben, W., B. Tyreus, and M. Luyben, Plantwide Process Control, McGraw-
Hill, New York, 1998.
Skogestad, S., P. Lundstrom, and M. Morari, "Selecting the Best Distillation
Control Configuration," AIChE J., 1990.