(107f) Control Structure Design for Optimal Operation of Thermally Coupled Columns

Control Structure design for optimal operation of thermally coupled
columns

Deeptanshu Dwivedi¹,
Ivar J. Halvorsen²
and Sigurd Skogestad¹

¹ Department of
Chemical Engineering, Norwegian
University of
Science and Technology,
N-7491 Trondheim, Norway, Email: dwivedi@nt.ntnu.no, skoge@nt.ntnu.no

² SINTEF ICT, Applied Cybernetics,
N-7465
Trondheim, Norway, Email:
Ivar.J.Halvorsen@sintef

Keywords:
Thermally coupled Distillation Columns, Optimal Operation, Control Structure Design,
Divided Wall Column

Abstract:

The dividing wall column
(DWC) has gained increased industrial attention due to its energy and
capital-saving properties. To obtain the energy-savings in practice, it is
required to apply a control strategy that keeps the operation close to the
optimal operation in presence of unknown disturbances and model uncertainties.
Otherwise, the potential energy savings may easily be lost or lower purity may
result, especially in the side stream product. The three-product DWC has been
used in more than hundred industrial applications.  As we increase the complexity by going to the
4-product Kaibel column and even the 4-product DWC with two partition walls,
and use the column to separate feeds with a large number of components which
shall be grouped into suitable product streams, the question of control design
and control performance will become even more important. 

Minimum energy operation
and tight control are closely related. In a conventional two product column,
the most common control strategy is a simple one-point temperature control.
With a large the energy input, in event of usual disturbances during operation,
this very simple strategy can ensure purity in both products.  This however, usually results in either both
or one of the products is over purified. 
To obtain minimum energy operation, it is well known that both products
should be controlled to their purity specifications. This is actually a more
difficult mode of operation.

This work aims to
demonstrate the control structure design and performance by operation of a
four- product lab scale Kaibel column using the experimental runs followed by
modelling and identification. A separation of mixture of four alcohols:
methanol, ethanol, propanol and butanol are studied using the lab column.  The lab column is a made of standard vacuum
glass sections with 50 mm internal diameter and filled with 6 mm Rachig rings.
The height of the set-up is about 8 m. There are 24 temperature sensors inside
the column and the column is operated using a Labview interface. The column has
a magnetic funnel that divides the liquid between the pre-fractionator and the
main column. The vapour split is affected using two motor operated butterfly
valves, one each in the pre-fractionator and in the main column. The reboiler
is a 2 KW kettle type electrical heater.

A control structure
using four decentralised PI temperature controllers is proposed. One
temperature in the pre-fractionator and three temperatures in the main column
were controlled using four manipulated variables namely liquid split ratio and
3 product flow rates: distillate, side product 1 and side product 2. The column
was successfully stabilized and all the four temperature were controlled within
±0.5 C of the set-point. The product samples were collected and analyzed. The
preliminary experiments yielded significantly pure top and bottom product while
the two side products were not very pure. A suitable temperature profile should
be maintained using the 4 PI decentralized loops to improve the purities. This
is being further studied using a rigorous model of the same and lab
experiments.

An effective vapour
split is critical during minimum energy operation of the column. The vapour
split valves were examined and an experiment was designed to test their
efficacy. The column was run under total reflux conditions using only two
components: methanol and ethanol. The vapour split valves are operated under
split range control to set a constant temperature difference between a sensor
in the pre-fractionator and one in the main column.  This control objective was successful tested
for servo and regulatory performance.

The model of the lab
setup is a rigorous first principle equilibrium stage based model. There is no
vapour hold up and model accounts for non-ideality in liquid phase using
Wilson equation. The
number ideal stages in the real column is determined experimentally by doing a
total reflux experiment and using only two components, methanol & and
ethanol. The compositions of the product sample were then analyzed to determine
the number of ideal stages using Fenske Equation.  The next step in the identification was to
manipulate the inputs in the model to match the experimental conditions and
data reconciliation. Here we fit the steady state temperature profiles, and the
compositions of the product samples collected during the steady conditions of
operations of the lab column. An optimization problem was formulated for the
same [2]. A good fit of temperature snap shot from the steady operation
condition of the experiment was obtained. The dynamic response of the
experiments can be further fit using other parameters of the model like stage
hold up and sizing.

This model is then used to
test the efficacy of control structure for its regulatory response. The Kaibel
column, owing to its highly interactive nature has some unique control problems
and demand novel control solutions [3]. The key is to stabilize each internal
sub-column profile to ensure that the separation task in each sub-column is
carried out properly in an event of disturbances like feed property variations,
tray/packing performance variations, measurement uncertainties and
uncertainties in setting the internal flow splits. This can be done by use of
feedback control, based on knowledge of how the internal sub-columns should
behave at and around the minimum energy operating region. 

Here we will show a
procedure based on characterizing the optimal operating region.  The V-min diagram [4, 5] is a simple tool
that can be used to assess both design and operation issues.  It can be point out the overall minimum
energy requirement and flow distribution in fully thermally coupled arrangements.
Further, we also get information on allowable variation margins in operation of
the internal sections or sub-columns of the arrangement without compromising
the overall minimum energy operation. This is very important for practical
implementation. Due to inevitable design uncertainties and unknown disturbances
an industrial column control system has to be designed to allow for certain
slacks and the information about how and where operation can be relaxed is
important in order to design the control strategy. For instance, an allowable
range for vapour splits, allowable variation in feed properties for a selected
control strategy, or if the focus should be on controlling the upper or lower
part of the pre-fractionator profile.  By
identifying the key operation parameters that need high attention and thereby
accurate control, we obtain a system that can give high purity in all products
with a simple feedback control strategy.

 References:

1. Kaibel, G. (1987). Distillation Columns with
vertical partitions, Chem. Eng, Tech. 10
(1987) 92-98

2. Lid, T. Skogestad, S (2008). Scaled steady
state models for effective on-line applications, Comp. & Chem. Engg.Vol. 32, Issues 4-5, April 2008, Pages 990-999

3. Strandberg, J., Skogestad, S.,
HalvorsenI.
(2010), Practical control of dividing wall column, Distillation & Absorption (Eindhoven, Netherlands)

4. Halvorsen,
I.,
Skogestad, S. (2006), Minimum Energy for the four-product Kaibel-column, AICHE Annual meeting, 216d

5. Halvorsen,
I., Skogestad, S. (1999),
Optimal Operation of Petlyuk Distillation: Steady State behaviour, Journal of Proces. Control, vol 9, 1999,
407-424