DISTILLATION COLUMN COMPOSITION CONTROL
TEAMS: This project can be completed in teams of up to 4 students.
DISTILLATION COLUMN COMPOSITION CONTROL
Over 40,000 distillation columns operate today in the U.S. in petrochemical processing facilities and
cumulatively consume 40‐60% of all of the total energy used in the petrochemical industry and more
than 6% of the TOTAL U.S. energy usage. Reducing distillation energy usage can have a major impact on
energy independence, greenhouse gas emissions, and give a competitive advantage to any
manufacturer that is better able to implement improvements.
Task
Design a control system for a binary (cyclohexane / n‐heptane) distillation column and
maintain the distillate composition above 0.95 mole fraction cyclohexane (C6H12) and the bottoms
composition below 0.05 mole fraction in cyclohexane so that both streams can be sold commercially. If
the distillate composition is above 0.95, it indicates that energy was wasted to exceed the target
specification limit. If the distillate composition is too low, energy is also wasted because the product
must be recycled through the distillation column. If the bottoms composition is below 0.05
cyclohexane, then energy has been wasted to boil‐up more material than was needed to achieve the
desired purity specification. In order to meet the target specifications, two key column variables can be
manipulated: (1) the reflux rate (in [mol/min] flowrate) that is fed back onto the top tray in the column
can be modulated via a control valve and (2) the reboiler boil‐up rate (in [mol/min] units) can be
adjusted by means of a control valve feeding steam to the reboiler. A distillation column is a highly
coupled non‐linear system that inherently leads to interacting controllers. Your overall task is to
develop a control scheme that can provide good compositional control of both the overhead and
bottoms product streams in the face of disturbances of both feed flow, composition, and quality.
A fully non‐linear model of the cyclohexane/n‐heptane distillation column has been developed and is
supplied in the associated MATLAB file. You can use this model to both (1) empirically develop linear
system models for the column around a desired steady‐state operating point and (2) test your proposed
control schemes on the fully non‐linear process to determine how well the control scheme will work in
practice in the plant. The model provided is based on a steady‐state operating point of: (1) a feed
flowrate of 1 [mol/min] to the column that is 50 mol% cyclohexane and 50 mol% n‐heptane, (2) a reflux
flowrate from the total condenser back into the top of the column is 2.706 [mol/min], (3) a bottoms
boil‐up vapor flowrate in the reboiler of 3.206 [mol/min], (4) a column containing 41 stages including
the reboiler and total condenser (which produces saturated liquid distillate), (5) feed being introduced
on the 21 stage in the column where stage #1 is at the top of the column, (6) a relative volatility
between cyclohexane and n‐heptane of 1.5. You can assume that the control valve on the reflux return
line to the column is fast and can modulate the reflux rate between 0 [mol/min] and 5 [mol/min].
Likewise, you can assume the steam line control valve to the reboiler is sized so that it can deliver boilup
vapor flowrates from 0 [mol/min] to 5 [mol/min].
You should at a minimum complete the following tasks as a part of your control system design and
verification:
1. Empirically determine a linear system model that describes the distillation column and relates
changes in feed flowrate, composition, or quality to the distillate composition and bottoms
composition around the desired operating point of feed flowrate of 1 [mol/min], feed
composition of 50 mol% cyclohexane, 95 mol% cyclohexane in the distillate and 5 mol%
cyclohexane in the bottoms product. Implement that linear system model in SIMULINK.
2. Perform an analysis to determine what manipulated variables (i.e. reflux flowrate or reboiler
boil‐up vapor flowrate) should be paired with each regulated output variable (i.e. distillate
composition and bottoms composition).
3. If one assumes that the composition analyzers used on the column are on‐line gas
chromatographs that are properly calibrated but act as a time delay element in the
measurement of 10 minutes (i.e. due to the time required to elute material through the GC
column), use at least two different methods including IMC to design and tune controllers for the
system that do not utilize decouplers. Test your control system on the nonlinear column model
and discus how effective the control system performs in the face of +/‐ 10% disturbances to
feed flowrate, composition, and quality.
4. Redesign/retune controllers for the system in a case where static decouplers are used to allow
for better control of the system. Test your control system on the nonlinear column model and
discus how effective the control system performs in the face of +/‐ 10% disturbances to feed
flowrate, composition, and quality.
5. Redesign/retune controllers for the system in a case where dynamic decouplers are used to
allow for better control of the system. Test your control system on the nonlinear column model
and discus how effective the control system performs in the face of +/‐ 10% disturbances to
feed flowrate, composition, and quality.
6. Design a control system which implements feedforward control as a means to mitigate the
effect of feed flowrate changes on column performance. Test your control system on the nonlinear
column model for +/‐ 10% changes in feed flowrate.
7. Design a control system which implements a Smith Predictor structure to better deal with time
delays in the system. Test your control system on the nonlinear column model and discus how
effective the control system performs in the face of +/‐ 10% disturbances to feed flowrate,
composition, and quality.
8. BONUS: Show use of any other methods and approaches in the course to develop and effective
control system for the column.
DELIVERABLES:
Your report must be submitted in two parts: (1) a single PDF file containing all of the report addressing
the required tasks and any optional tasks undertaken and (2) a SIMULINK model file containing what you
deem as the best control system configuration implemented on the non‐linear column model supplied
with the project documentation.
Your design report should focus on what methods you apply to developing the response to each task,
what the outcome of those efforts are, and to what degree effective control is achieved for the column.
Effective use of tables, figures, and diagrams (e.g. SIMULINK diagrams, time domain response plots, etc.)
in addition to text are expected in the final report.