PPT

Report
Forget Laplace Transforms…

Industrial process control involves a lot more
than just Laplace transforms and loop tuning

Combination of both theory and practice

Understanding of core engineering principles
is key (thermodynamics, mass transfer, etc)


Control design requires collaboration with
others to understand objectives and provide
process design guidance
Importance of both “big picture” and details

Maintain the process at the desired state or
set of conditions – “keep it out of the ditch”
◦ Safety
 Ensure the process conditions minimize risk
◦ Optimal operation
 Running at the appropriate operating conditions
improves quality, yield, plant capacity, energy
consumption, etc
◦ Recover from upsets or disturbances

It’s not just about optimization; it’s about
successful operation of the entire plant

A primary objective of the process control system is
to keep the process running at the desired operating
conditions
◦ Presumably these conditions have been chosen
appropriately from a safety standpoint (hint, hint, design
engineer  )

“Cruise control”
◦ The basic process control system should be able to handle
many disturbances, but not all
◦ Cruise control on your car can handle hills and curves, but
if there’s an accident ahead, you’ll have to stop the car
yourself
 Safety Instrumented Systems (interlocks)


A good process control system will keep the process
running stably, even when hit with disturbances or upsets
This results in better efficiency, higher capacity, etc.
Improvements
to this temp
control strategy
resulted in a
steam savings
of $260K/yr,
or $1.1M NPV


Running at the optimal operating conditions
can maximize production rate and yield,
improve energy consumption, and is crucial
for product quality
However, these objectives often compete
◦ Best product quality may be attained at the cost of
additional energy consumption

Advanced Control techniques can help with
balancing this tradeoff

Advanced control applications provide an
additional layer of control, to meet a variety of
control objectives
◦ Feed-back composition control based on lab data
◦ Feed-forward to other unit operations or plant areas
◦ Perform complicated online calculations and close the
loop to manipulated variables
◦ Plant-wide supervisory control strategies can balance
rates, maximize throughput, minimize conversion costs
or energy consumption…
◦ Model Predictive Control (MPC) incorporates a process
model to optimize operation when there are multiple
input, output, and disturbance variables
“You’re a chemical engineer
first and foremost!”

If you truly understand the chemical principles at
work in the process, then controlling it is easy!
◦ Or easier, at least…

You have to understand the fundamental stuff
that’s going on in order to determine:
◦ What the control objectives are in the first place, and
which variables should be controlled
◦ What your “control knobs” are and how they will affect
the process as a whole – how it all fits together
 If you increase the steam flow to a distillation column’s
reboiler, what will happen to the composition on tray 15?
What about the distillate? What about the pressure profile?

Another way to think about it: the goal is to move
variability to some place where you don’t care
about it
◦ If the temperature in a reactor cycles or varies, that’s bad
◦ We can control this temperature (keep it stable) by
implementing a control loop which manipulates steam
flow to the reactor jacket
 Who cares if the steam flow moves around? The reactor
temperature is constant, and that’s what we want.

Comes back to fundamental process understanding
◦ Must understand where variability is acceptable, and
where it’s not
◦ Must understand how everything fits together
Distillation Control

Need to understand manipulated variables
(“control knobs”) available to us

Chemical Engineering knowledge tells us…
◦ Increasing the reflux will help purify the distillate
◦ The hotter the base, the more material will boil
overhead  the entire composition profile will shift
◦ The dynamics of liquid effects vs. vapor effects are
very different
◦ The temperature on each tray is a function of the
tray’s composition and pressure


In order to maintain the desired top and
bottom compositions, it is important to
prevent the composition profile from moving
The temperature profile of a column is
indicative of the composition profile
◦ By selecting the right temperature to control, we
can actually peg the entire temperature profile
◦ The appropriate temperature control strategy (tray
location, manipulated variable, etc) is highly
dependent on the individual column design

Manage inventory
◦ Need to ensure there is always reflux “available”
◦ Likewise, need sufficient holdup in the column base

Maintain desired product compositions
◦ What are acceptable impurity ranges?
◦ Is one product stream more important?

Other objectives
◦ Pressure control, column loading, minimize steam

Respond to certain upsets
◦ What process upsets is this column likely to see?

First, obtain or develop a steady-state model
◦ Need to know target compositions, normal flows,
pressures, the column’s temperature profile, etc.
◦ This gives you a snapshot of the desired operation
◦ A steady-state model also yields insight on the “control
knobs”

Next, pair controlled variables with manipulated
variables
◦ Based on “Chemical Engineering” knowledge
◦ Utilizing information regarding key control objectives
and predicted disturbances
LC
FFC
Tray 8
LC
TC
Steam
VACUUM LINE
TO
HEADER
XC
PC
PC
LC
CONDENSATE
FC
HOT
CONDENSER
FC
FC
FC
IX
REFLUX DRUM
REFLUX
RATIO
TARGET
LC
TO REACTORS
SGI
TI
FY
COMPOSITION
FI
FI
And more…
PRODUCT
• Plant-wide
supervisory
control
• Feed-forward to
other unit ops or
plant areas
• Model predictive
control (MPC)
• And so on…
FEED
LC
HC
FC
600 PSIG
STEAM
LC
LC
CONDENSATE
PC

Beneficial to create a dynamic
simulation of the column
using this control strategy
LC
◦ Allows for testing of the strategy
under various disturbance
scenarios
◦ Gives valuable information
regarding dynamic behavior of
the column
◦ Provides initial tuning data
FFC
Tray 8
LC
TC
Steam
“Tray 8 – to – Steam” Control Strategy
“Tray 42 – to – Reflux” Control Strategy
Double-Ended Temperature Control Strategy

Once the control strategy framework has
been laid out, then you get into the “nuts and
bolts” of configuration
◦ Algorithm type
◦ Controller action
◦ Tuning (gain, time constants, etc)
Capital Project Involvement

For each unit operation, work closely with design
engineer and other project/operations
representatives to…
◦ Understand design intent, including steady-state flows,
desired recoveries, conversions, etc.
◦ Gain insight on potential process disturbances
◦ Define key control objectives
◦ Provide guidance on the actual process design
 Determine residence times required for stable operation
 Specify instrumentation placement
 Other recommendations based on dynamic simulation and
other analysis (is desired steady-state operation feasible?)

Provide guidance on plant-wide control
◦ Decouple interactions as much as possible
◦ Control valve placement, piping layouts
◦ Inventory management

Instrumentation selection

Safety considerations, interlocks

“Control Narrative”
◦ Detailed document describing control objectives
and strategies for each unit operation, the plan for
managing inventory plant-wide, etc.


Remember: always think about process
control from the perspective of Chemical
Engineering fundamentals
Understand your process, as well as your
control objectives
◦ What needs to be controlled? Which variables effect
each other (and how)? Where does variability hurt
you most? Etc.

Remember there’s a dynamic component

Think about control early in design phase

similar documents