1) Overview the top and bottom concentration(xD and xB)

at the desired set point

Whereas, we also have to guarantee the
stability in column operation, thus we have
to keep the level at the top(MD) and the
bottom(MB) in controllable range

To achieve those requirements, we will

take four variables into account as four
manipulated variables: xD, xB, MD, MB and 4
other variables as measured variable: L,V,
D, B. We will build up dynamic equations
Figure 1: Basic model of distillation column based on those variables. Besides, to
The distillation column is a typical describe 2 main physical process in the
process in process control in which a liquid column, we have to take into consideration
or vapour of two or more substances is the two following equation
seperated into its component fractions of α xi
desired purity by the applycation and VLE equation[1]: y i=
removal of heat.
1+ ( α −1 ) xi
The distillation column works based on (1)
thermodynamic principles. There are two Hydrodynamic equation[2]:
main physical process existing in the column M i −M 0i
Li=L0 i + + λ( V i−V 0 i )
which are the condensation of the heavy τ
components and the vaporation of the light (2)
components. The feed flow( in our situation
which is the inlet beer) is fed into the feed In which, Li and Vi are liquid and vapor flow
tray. The components in combination which from ith tray [kmol/min] respectively α is
have lower boiling temperature will volatile relative volatility between light and heavy
and move the top of the column and then go component . τ is time constant [min] for
through a condenser before being retrieved liquid flow dynamics on each stage. λ is
to a tank as the top compositions. The constant for effect of vapor flow on liquid
components which have higher boiling flow; Mi is liquid holdup on ith tray [kmol]
temperature will flow down to the bottom of
distillation column in the liquid formation Using those equation and emperical
and be retrieved to the tank as the bottom parameter, we achieve the following transfer
compositons. This kind of distillation function of the column
column is called binary column

( )
In this article, we are focusing on 0.008003 ❑
researching a control method to get the G P= 22.1442 s +1 22.3665 s +1
outlet beer with less than 0,5% alcohol. 0.00023623 −0.0009011
Therefore, controlling the concentration of 6.7812 s+1 6.5366 s+1
the bottom composition is our top priority. The following figure show the respone of
2) Modelling the column and control the open-loop system
configuration
The top priority is to detach alcohol from
beer to attain the alcoholic proportion in
outlet beer below 0,5%; therefore, the
crucial purpose in controlling is to maintain
 - For (xD, L) loop: KP=139.86; Ti=14,76;
TD=3,696
- For(xB, V) loop: KP=-1242,23;
Ti=4,538;TD=1,137

Unfortunately, the controller parameters

gained theoretically make the respone get
some undesirable feature: the large settling
time(50-60 minutes), remaining steady state
error
 Control configuration
Therefore, we changed the parameters
The choice of a proper control configuration empirically to get more appropriate respone
is very important when designing distillation
control systems. The steady state RGA is a - For (xD, L) loop: KP=500; Ti=10;
common method used for selecting the TD=1,1
appropriate structure. - For(xB, V) loop: KP=-15000;
Ti=0,7;TD=0,182
Based on a frequency-dependent RGA
analysis, each control configuration have 4) Simulation results
their own pros and cons in 3 different modes
of operation which are : manual operation,
1-point composition control, 2-point
composition control. We can find out details
in [3] that LV configuration is preferable
for1-point composition control. As we can
see, the operation we are dealing with is 1-
point composition control(because we focus
on the concentration of the outlet beer).
Thus, we strongly recommend using LV
configuration for designing.

3) Design the PID controller for distillation

model
We have 2 measured variable(L, V) and 2 Figure 2.Respone against disturbance
manipulated variables( xB, xD). This is a
multivariable system. To reduce effort in
designing, we used RGA [4] This allows us
to match up variables that have the biggest
effect on each other, without having
undesirable effects on the others. We
obtained 2 pairs variables (x D, L) and
(xB, L)
After decoupling, we used a practical tuning
rule for PID controllers: the T sum rule of
Kuhn [5] to design the controller for each
pair of variables
We got the parameters of the controller as
follow:
Figure 3.Respone against noise
[1] “Vapor-Liquid Equilibrium” Billy G.
Harper, John C. Moore Ind. Eng.
Chem., 1957, 49 (3), pp 411–414

[2] ”Hydrodynamic equations and correlation

functions” - Leo Pkadanoff

[3] ”Selecting the Best Distillation Control

Configuration” Sigurd Skogestad Petter

Lundstriim Elling W. Jacobsen Chemical

Engineering Norwegian Institute of Technology

(NTH) N-7034 Trondheim, Norway

[4] “Relative gain array analysis
for uncertain process models”D
Chen, DE Seborg -AIChE
journal, 2002 - Wiley Online
Library