Process Control Project

Document preparation and some tips

May 18, 2015

The intention of this document is to provide with some guidance to develop the project for
the Process Control course (TKP4140), which is a 4t h year course in the Chemical Engineer-
ing program at NTNU (first year of the master program). The main description of the project
and the problem statements are found in It’s learning.

This document was first developed for the 2015 Fall semester. If you find any mistakes or
have any comments, please let us know.

This project is relevant for the course because you will apply many of the concepts you have
learned or will learn. It is also thought to be developed in teams. The workload is for a team
of up to 3 students and every team member should participate equally. Start working on your
project with time so that you have enough time in case you have any unforeseen problems.

If you have any further questions or specific questions regarding your project, do not hesitate
to come during "drop-in hours" to K4, 2nd floor. Outside these hours you can get help using
the forum in It’s learning.

1 General Tips
In this section you will find some general tips that might be useful for each part of the project.
There are more tips in section 2.

1.1 Part 1

• You may use the example files as a starting point.

• When writing your nominal values in the code, try to not truncate the values. It is better
to use the exact values or, when applicable, the equations. This can avoid some round-
up errors. Be careful with the units.

1
 • You may declare more symbolic variables, and then simplify your equations.

• Remember that the example files are for a 1 ×1 system and yours is not; so at some
points you will have to use vectors instead of scalars 1 .

• When simulating, you should do one step change at a time.

• After applying a step change, you should wait for steady state; except for integrating
plants (e.g. level).

• Check that when you are not disturbing your process nor changing the set point, your
outputs are actually zero when you simulate your process. This can help you to detect
issues with your model or with your nominal values.

1.2 Part 2

• Remember that after you linearize you are working with deviation variables. This is
important when setting up the Simulink block diagram; you should be careful when
adding the nominal values.

• By this point you should already know how to use the half-rule approximation. How-
ever, this could be a good reference for some special cases:
http://www.nt.ntnu.no/users/skoge/publications/2012/skogestad-improved-simc-pid/

You can also find information about the SIMC rule in the same reference; you will use
this tuning rule in part 3.

1.3 Part 3

• As for the previous parts, you can use the example files as starting point.

• For the Bode diagram:

– Remember to change the unit for magnitude from dB to absolute, also in loga-
rithmic scale. You can do this in the Property Editor of the Bode figure. You can
open the Property Editor by double clicking on the diagram or by right clicking
and then selecting "properties". You can also change the units directly from the
code (see the documentation for bodeplot).
– You can draw the asymptotes using Insert → line.
– If you right click on the lines in the figure, a small window with the frequency and
the magnitude (or phase) at that exact point will open. You can use this feature to
read all the required points in the plot.

1 Remember that in Matlab [2,2] is a horizontal vector and [2;2] is a vertical vector. You can transpose vectors

with an apostrophe (’), so that [2;2]’=[2,2]

2
2 Using Matlab/Simulink
The project is done in Matlab. All NTNU students can install Matlab on their own computer;
you should already have done it to solve some of the assignments. If you still do not have Mat-
lab installed, refer to: https://innsida.ntnu.no/wiki/-/wiki/English/Matlab+for+students.

Remember to include Simulink and the Control Toolbox in the installation. If you have any
problems with your computer, the computers in the computer room in K5 also have Matlab
installed.

Previous knowledge is of course useful, but it is not essential; you can learn along the way.
In this section you can find some tips for each part of the project. These might be useful es-
pecially for those who have not used Matlab previously. These tips are not absolute, and you
might find better ways to solve your problem or to code.

If you require specific information about functions or would like to have code examples, it is
highly recommended that you visit the Matlab Documentation website: http://se.mathworks.
com/help/matlab/.

2.1 Simulink

• You can start Simulink by typing simulink in the command window in Matlab.

• Remember to set an appropriate simulation time. The default 10 seconds may not be
appropriate for all systems. You can change this directly in Simulink or in your script:
http://se.mathworks.com/help/simevents/ug/regulating-the-simulation-length.html

• Use the To Workspace block to send variables to the workspace for plotting, for exam-
ple. The default save format is Timeseries. If you prefer to treat your variables as arrays
in the workspace, change the format to Array; this way you can plot as usual.

• In "Sinks", you can also find the block scope, which can help you to see your results
while simulating.

• You can combine several signals into one vector using mux, and then separate them
using demux; both are blocks in Simulink. You find these blocks in "Signal routing".

• You can find the S-function under "user-defined functions".

• When you save your Simulink file, check whether you saved it as *.slx or *.mdl; it has to
be consistent if you use the extension when running the model using sim(’model’) in
the code.

2.2 Figures and plots

Probably the most important recommendation is to include axis titles and legends.

3
 • In Matlab preferences, set the figure copy options to transparent background.

• You can format the figures directly from the code.

• You can save your figures directly from the code, using saveas.

• You can use subplot to plot together trends that are related.

• When defining the axis for your plots, it is good to have relevant axis limits. If you are
comparing two comparable plots, use the same limits for both plots.

• In your report, remember to describe your figures with a caption.

• For the axis titles and legends you can use a caret (ˆ) for superscript and underscore (_)
for subscripts. For example: ’y_s’ for y s , ’mˆ3’ for m 3 . You can also use \ to display Greek
letters: for example, ’\tau’ to display τ. See the documentation: http://se.mathworks.
com/help/matlab/creating_plots/greek-letters-and-special-characters-in-graph-text.
html

Matlab help for plotting:

http://se.mathworks.com/help/matlab/ref/plot.html

2.3 Matlab script

Not all the students have the same experience with Matlab, but everybody should be capable
of doing the project. Especially if you are not very experienced, try to code as simple as pos-
sible and include all the comments you need 2 , so that you can understand your own code
later; remember that the project is cumulative.

If you have more experience with Matlab, you can use everything you know when coding.

2.3.1 Matlab functions

As mentioned previously, Matlab documentation is very complete; so, if you are not sure of
how to use a function, you should visit the website.

An alternative option to get documentation is to type doc name + enter, in the command
window. Then you will get the documentation for that functionality. For example, if you type:
doc minreal, you will get the documentation for minreal.

Other ways to get documentation are explained here:

http://se.mathworks.com/help/matlab/matlab_env/help-resources.html

Some functions that you might need or want to use are listed below. The list is not exhaustive.
2 Remember that everything written after a percentage sign (%) is a comment.

4
 • saveas: saves figures in your preferred format.

• sim: simulate dynamic (Simulink) system.

• syms: creates symbolic variables that you can use for your equations.

• subs: symbolic substitution, replacing the symbolic variables with their values.

• jacobian: calculates the Jacobian of a matrix.

• inv: inverse of a matrix

• minreal: removes unobservable states for transfer functions.

• set: in general, it sets properties; for example, you can set plot properties.

Another use of set is for transfer functions. By default, transfer functions are not dis-
played in a time constant format. To change this, you need to specify the display for-
mat. For example, if you have a transfer function called g 11 :

g11 = set(g11,'DisplayFormat','time constant');

This feature can help you to get the transfer function in the format you are used to,
so that you see the time constants easily. However, remember that for the exam you
should be able to do it by hand.

3 Document preparation
Remember that the second and third reports are cumulative. This means that the second re-
port should contain what was done in the first part and that the third delivery is a final report
covering the whole project.

Your report should be "readable", meaning that there should be a sequence and it should be
simple to follow and to understand what you did.

Simulink diagrams should be included in the report or in the appendix, where you find it bet-
ter. Please include the code that you wrote or modified as an appendix; you can also include
code snippets in the main report.

Even if you do not include a front page, please include the names of every team member in
the first page of the document.

3.1 LaTex or Word

You may use either LaTex or Word. In any case, please upload only the *.pdf file.

5
LaTex is a free document preparation software that is widely used in academia. It is conve-
nient when typing mathematical formulas or including code in your document (listings
package).

If you choose to use Word, you may use an equation editor such as Math Type to type your
equations. You can get it from progdist (Innsida).

3.2 Other useful software

If you need to draw some additional diagrams or generate some images you may use:

• Dia Diagram Editor: open source software to draw diagrams (such as PFDs).

• Microsoft Visio: Microsoft software to draw diagrams (such as PFDs). It is included in

the Microsoft Office package in Sofware Farm.

• Ipe: open source drawing editor to create vector graphics that can also to be included
in LaTex documents.

• Inkscape: open source software to create vector graphics.

These are just some examples. You may use any software that you find most convenient for
your purposes, if you require any.

3.3 Report contents

Each report should cover what is required in the description. This is just a summary.

The first report should include:

• Introduction: description of the process and statement of the control objectives.

• Derivation of the nonlinear model and all the assumptions that you made.

• Simulation results: step change (10% increase) all inputs and disturbances, as in the
example.

• S-function script (.m file), as an appendix; so we can check your equations.

• Discussion: are your simulation results reasonable? Why?

The second report should include:

• First report. If there were corrections, these should be included.

• Linearization

• Laplace transform of linearized models and transfer function matrix for u’s and d ’s.

6
 • Comparison of original non-linear models with linearized models (plots).

• Zero, pole, gain of transfer functions.

• Half-rule approximation of transfer functions to obtain a first-order with time delay of

the transfer functions relating MVs to the CVs.

• Discussion of the zeros, poles, gain. Is it consistent with the simulations?

The third and final report should include:

• Part 1 and 2. If there were corrections, these should be included.

• Pairing: selection of the best control structure (pairing), using your engineering in-
sight. Include a block diagram.
– For single MVs and single CVs projects, instead of pairing, you should implement
either a cascade controller or a feedforward (from the disturbances) controller.

• Tuning of PI controllers using the SIMC rules.

• Bode plot:
– Choose one CV and the corresponding controller.
– Do the bode plot for that loop: L(s) = C 1 (s) ∗ g 11 (s)
– Compute phase and gain margin.
– Identify poles and zeros in the plot.
– Draw the asymptotes.

• Closed loop behavior

– Add the controller(s) to your Simulink simulation.
– Apply step changes of of 10% from nominal for set points.
– Apply step changes of of 10% from nominal for disturbances.

• Performance: calculate IAE (integral absolute error) to evaluate the controllers

• Discussion: discuss the pairing, quality of control and performance.

• Conclusion of the project: overall conclusion of the project, summarizing what has
been done.
– The conclusion may include a feedback about the project itself - how to improve
it for the next year.

7
4 Uploading your report
Only one person per team should upload the files to It’s Learning. Remember to add every
team member in It’s Learning. Please upload:

• Report (*.pdf)
Only the *.pdf file is required.

• Matlab files (zipped)

This is not mandatory but could be helpful when reviewing your project, specially if
there are any mistakes or areas of improvement in parts 1 and 2.

5 Final Comments
Part 1 and part 2 will not receive a grade and will be either "approved" or "not approved". If
your report is approved you might receive some comments that should be included/corrected
in the next delivery. If it is not approved, you should correct your report and upload it again
to have it reviewed again and continue with your project.

The final report (third delivery) is graded and will count for your final grade, as stated in the
course description.