CHEM ENG 3P04: PROCESS CONTROL TERM PROJECT - 2018
Binary Distillation Simulation-Based Multi-Input
Multi-Output Control Project
Notes and Instructions
1. Students have a choice of doing either a simulation-based binary distillation
project, or a temperature control project involving an Arduino lab device that
can be signed out. This handout describes and sets out the requirements for
the distillation project.
2. Projects are to be tackled in groups of 4 (or 3), with one report per group. No
groups larger than four, please.
3. Each student must submit at the end of the project a con dential assessment
of the relative contributions of the members of the group. A form. for this
purpose will be provided.
4. Group membership must be declared by Monday March 26 (a list for this
purpose will be set up in Avenue).
5. DUE DATE: 4:00pm Monday April 9, 2018. Reports should be de-
posited in the dropbox outside JHE-374.
6. Project reports are to be written in a professional and academic manner with
proper technical writing and language. The reports should include section
headings, with all necessary gures and tables included in the main body
of the report and cross-referenced with appropriate labels/captions. Reports
should be a maximum of 25 pages.
7. All plots should be formatted in a professional manner and must be readily
comprehensible with captions, axis labels and legends (if necessary). The plots
must not have a black background.
8. All published information referenced or consulted should be properly cited in
an appropriate referencing style. of your choice, with a list of references at the
end of the report (before the appendices).
9. Detailed calculations must be put in the Appendix, and all Matlab les and
Simulink models used to generate the presented results must be included in
your submission. Also, Simulink block diagrams of your model and/or control
con gurations must be included as part of the report.
1 Project Learning Objectives
The primary objective of the project is to design a control system for a nonlinear
binary distillation column. The steps to be followed mimic those that would be
applied in an actual industrial setting. In particular, this will include
1. Identi cation of a dynamic model through open-loop step testing.
2. Use of the model to con gure a multi-loop control system and determine con-
troller tuning parameters.
3. Evaluation of the control performance on the nonlinear plant, ne-tuning the
controllers as necessary.
2 Process Description
The term project involves dual composition control of a binary distillation column,
as illustrated in the gure below.
LC
PC
LC
The level (distillate receiver and column bottom) and pressure control loops may
be assumed to be in place and active. Your task is to design a control system to
control the distillate and bottoms compositions by manipulating the re ux rate and
bottoms product rate. Disturbances are the feed composition and ow rate.
A distillation simulation model is provided as a Matlab Simulink model le. This
may be used to generate open-loop step-test response data, from which an approx-
imate transfer function model can be derived. The transfer function model should
be used to design a multi-loop control system for the distillate and bottoms compo-
sitions. Note that composition is given as the mole fraction of the light component.
Your task is to design a control system to maintain the distillate and bottoms
compositions at their end point speci cations in response to set-point changes, and
uctuations in the feed composition.
You are to perform. the following tasks, showing clearly your calculations/analyses,
and comment on you ndings. You are expected to make use of Matlab/Simulink
and appropriate toolboxes.
3 Project Tasks
3.1 Part I
1. Apply appropriate step tests to the Matlab dynamic simulation model of the
distillation column, and use the input and response data to determine an
approximate rst-order-plus-dead-time model of the form.:
with each transfer function element of the form
The process variables are de ned in the table below.
Variable De nition Units
XD Distillate composition mole fraction of light component
XB Bottoms composition mole fraction of light component
L Re ux ow rate kmol/min
B Bottoms ow rate kmol/min
Z0F Feed composition mole fraction of light component
Be sure to use a reasonable size for input step changes, with consideration
the e ects of both measurement noise (which e ectively places a lower bound
on the step size), and process nonlinearity and process input and/or output
constraints (which e ectively places an upper bound on the step size). Brie y
discuss your choices.
Plot the open-loop response of the derived model and nonlinear process on
the same axes (one plot for each input-output pair), and brie y discuss your
observations.
2. Use the Relative Gain Array (RGA) to determine an appropriate loop pairing.
Are signi cant interactions expected for this system? Discuss.
3. Draw a clearly labelled block diagram of the control system, being sure to
include the disturbance. All variables should be labeled.
4
3.2 Part II
1. Discuss the measurement instrumentation and control valves that would be
required for the control task at hand.
2. Design a multiloop PI control system for this process, and implement it on the
Simulink model. You are expected to rst apply a systematic design procedure,
followed by ne-tuning of the controller parameters. The strategy you follow
in both instances should be clearly de ned/explained. The control system
performance should be tested for
(i) a step change in the range [0.02, 0.05] in each of the composition set-
points (performed separately), and
(ii) a 0.1 step change in the feed composition disturbance.
Note that the input response is important and should be included in the report.
Discuss the performance of your control system, in terms of the responses
of both the controlled outputs and manipulated inputs. The extent of loop
interaction observed should also be discussed.
3. Design a decoupler for your control system (steady-state or dynamic or both).
Repeat the simulations in 5, re-tuning the PI controllers if necessary. Discuss
your results.
Extra Credit: One or more of the following may be explored.
1. Robustness refers to the tolerance of the control system (in terms of stability
and closed-loop performance) to changes in the plant conditions. Evaluate
the robustness of the controllers designed in parts 2 and 3 by applying the
controllers to the plant that is modi ed to include some dead time in the
composition analyzers (in reality, this could be a few minutes). Discuss your
ndings.
2. Evaluate the control performance under changes in column feed ow rate.
3. Try an alternative control con guration, such as the LV con guration in which
the distillate and bottoms compositions are controlled using the re ux rate (L)
and vapor boil-up rate (V), with the liquid levels at the top and bottom of the
column controlled by manipulating the distillate rate (D) and bottoms rate
(B).
Note: To do extra credit questions 2 and 3 select the \Process Model" and use
\Ctrl+Shift+G" to expand the model. Once the model is in a expanded form,
necessary changes can be done.
4 Instructions: Simulink distillation column model
1. First, download and unzip the distillation column MATLAB model les
(cola Project.zip) from Avenue to Learn.
2. Start Matlab and change the le directory shown in MATLAB to the project
le location (similar to the set-up steps for the Arduino lab). For example,
if your lab les are placed on the desktop, the le location should be some-
thing like: C:n:::n:::nDesktopntclab simulink mimo. Copy this le route
and paste to the Matlab directory. You can check if you are in the correct
directory by using the command pwd.
3. To run the distillation column model, execute colas lb nonlin.mdl (by double-
clicking on the le name). You will be presented with the Simulink model as:
4. The values of the inputs and initial values of the states correspond to a nominal
steady-state with compositions xD = 0:93 and xB = 0:07. If you run the
simulation as it stands (without changing any value), the plots should indicate
this (except for numerical noise).
5. To apply a step change in a disturbance variable or an input variable, you may
simply change the value in the box corresponding to zF (Inp 1), L (Inp 2) or
B (Inp 3) and see the response in xD (Out1) and xB(Out2).
6. The input and output variables of ‘Process Model’ are not in deviation form.
Note the following about \process model" block in particular:
The \Process Model" block can be double clicked to see the inside details
(connections and blocks). See gure 1 below.
Figure 1: Expanded process model.
The column input variables enter the \Mux" box, and are (in order): L (Inp
2), V, D, B (Inp 3), F, zF (Inp 1) and qF. qF is the feed condition, with
qF = 1 corresponding to liquid feed and qF = 0 corresponding to vapor feed.
For this project, leave this as qF = 1.
The variables leaving the \Demux" box are (in order:) xD (distillate com-
position, Out1), xB (bottoms composition, Out2), MD (condenser liquid, or
condenser holdup), MB (bottoms liquid holdup) and the tray compositions
(Comp).
Proportional controllers are in place to control the top and bottom liquid levels
(via the holdups) by adjusting the distillate and vapor rates.
Since the controller output is in deviation form, the nominal steady-state val-
ues of the variables are added to the controller output signals (using the ‘Sum2’
and ‘Sum3’ blocks) before being applied to the column.