Experimental set up of the laboratory scale distillation column
Researchers: B. Buttinger
Ir. M. Springer
Dr. R. Baur
Prof. R. Krishna

 1. Description of the hardware:
The experiments were carried out with a laboratory scale distillation column with an inner diameter of  50 mm and a total height of app. 2160 mm, which is shown in the picture on the right and in figure 1 and 2. The plant consists of a total condenser at the top, a double wall cylinder with vacuum between the outer and the inner shell and a partial reboiler at the bottom. The column itself is subdivided into 2 parts. Each part contains a set of 5 bubble cap trays, the set up and the main dimensions of such a tray you can see in figure 3. Between those two parts there is a flanged intermediate piece at which a continuous feed can be introduced to the column (point 6A in fig. 2). Product streams can be tapped automatically from the condenser and manually from the reboiler.
The glass column has several small openings (app. 10 mm in diameter), which allow us to measure the temperatures inside of the column aswell as taking samples of the vapor and the liquid phase.
The reboiler is placed in a heating coat, which is controlled by a standard PC provided with the following software (Honeywell):

   - operating system: WinNT-workstation 4.0
   - FIX MMI V 6.15/75-I/O-points runtime
   - OPTO CONTROL rel.2.2a

By means of the PC, a set point for the heating temperature can be defined and will be kept constant during the experiments. In our case a set temperature of 180 oC was chosen. Furthermore, the software provides a feed- and product-flow control as well as an automatically (alarm) shut down, in case the column-reboiler tends to dry up. 
The condenser is connected with a water-tap, which keeps the cooling-water flowing through the thin glass-tubes inside the condenser. The rate of the cooling-water is a dimension for the condenser capacity.
  Flowsheet with computer controlled temperature measurement

To be conformable with the nomination used in the ChemSep simulation, the condenser is named the theoretical stage 1, so the first real tray is called stage number 2, since we count from the top to the bottom. The last tray of the column is thus stage number 11, the reboiler is named stage 12. (figure 2)

Figure 1, Column with dimensions
Figure 2: stage numbering

Bubble cap tray:
Fig. 3: Schematic drawing of a bubble cap tray
Tray close up
Bubble cap tray when column is filled Bubble cap tray when column is operating
Bubble cap tray with downcomer Bubble cap tray with downcomer

The complete design of a stage is listed in table 1:
Column diameter  0.0500 m Hole pitch 0.0142 m
Tray spacing 0.0462 m Cap diameter 0.0281 m
Number of flow passes 1 Skirt clearance 0.0030 m
Liquid flow path length 0.0308 m Slot height 0.0050 m
Downcomer clearance 0.0039 m Active area (of total area) 97.30  %
Deck thickness 0.0030 m Total hole area (of total area) 8.27    %
Hole diameter 0.0142 m Downcomer area (of total area) 1.35    %
Weir type Circular Slot area 0.000221 m2
Weir length 0.0182 m Riser area 0.000158 m2
Weir height 0.0092 m Annular area 0.000462 m2
Weir diameter 0.0058 m
Table 1, Stage design of the laboratory scale distillation column


2.  Description of the experiment:

One experimental run consisted of measuring the temperature and the composition profiles of the system 
Water (1) – Ethanol (2) – Acetone (3) in the distillation column.
As a first step the column was filled with a certain mixture of these components (total filling volume was 2 liters). For each  run the composition of the feed was changed by adding the appropriate volume of the desired component(s) before heating up the column. 
The distillation column operated at total reflux and atmospheric pressure. No product streams were taken from the column.

Temperature measurement:

The temperatures were continuously enregistered by a PC who gets its information from temperature sensors, which measure the vapor temperature in the reboiler, at 6 points along the column and at the top (figure 2). The temperatures used in the simulations were taken when the column operated at steady state, what was indicated by constant temperatures. The temperature trends can be followed directly on the PC- screen.
After the temperature measurements were done, the column was cooled down in order to reduce the vapor loss while replacing the sensors by teflon coated septums. It is important to minimize the loss to match the measured temperatures with the composition profiles of the vapor and the liquid phase. After sealing the openings with the septums the column was reheated again.

Composition profile measurement:

When the column reached steady state again, samples of the vapor and the liquid were taken at several points along the column using a syringe.
The withdrawal positions of the vapor are identical with the points from the temperature measurement plus 1 point supplementary situated in the middle of the intermediate piece (V1, V2, V4, V6, V6A, V7, V9, V11, V12). The liquid was taken from L2, L6, L7 and L11, see figure 1.
The sample volume is 100 ml, in order to be able to calculate the exact value of the taken volume, the samples were also weighted. (the accuracy of the balance is 0.1 mg)
It is important to take such a small sample volume to avoid a significant change of the compositions in the column.
Thus we obtained 4 samples of the liquid and 9 samples of the vapor phase per run which were analyzed subsequently.

  3.  GC- analysis:
For the analysis of the samples, a gas chromatograph of type GC8000-top with pressure/flow control was used. The autosampler type is AS800, the channel Hayesep-Q 80- 100 is made of stainless steel with a length of 1 m and a diameter of 0,125 inch. The analysis time was fixed with 30 minutes per sample. The oven heating rate was chosen with 70 K/ min, the initial oven temperature was 110 ° C. The carrier gas used was Helium.

The GC mainly consists of a thermal conductivity detector 
(TCD or HWD) that is sensitive to any compound having thermal conductivity other than that of the carrier gas used. The TCD essentially is a stainless steel block containing two pairs of filaments (generally tungsten/ rhenium filaments) having the
same electrical resistance. The block is housed in an 
aluminium case that accomodates the heating elements and the temperature sensor. The two pairs of filaments are electrically connected according to the Wheatstone bridge diagram (see figure 4). 
Two gas flows, a reference and a measuring one, enter the TCD cell and pass through the two pairs of filaments. When the bridge is properly powered, the filaments are heated at a temperature 
(resistance) that is a function of the thermal conductivity of the gas flowing through the filaments.


Figure 4: Wheatstone bridge

The reference element is exposed only to the pure carrier gas flow, whereas the measuring one is exposed to 
the effluents of the gas chromatographic column (carrier gas and sample). When pure carrier gas crosses both the reference and the measuring channels, a constant temperature gradient is established between the elements and the detector walls, so the Wheatstone bridge is in equilibrium, in other words there is no output signal.
While a chromatographic component is eluted, a change takes place in the heat transfer followed by a variation of the filaments temperature. Since the electrical resistance is a function of the temperature, the bridge becomes unbalanced and the detector generates a signal (peak) that is proportional to the difference in thermal conductivity between the eluted component and the carrier gas. The signal is sent to a recorder, here the Chrom- Card software.

Helium is the recommended gas because of ist high thermal conductivity and chemical inertness.

To start the GC- analysis, 2 more solutions have to be produced.

First, we need a reference solution consisting of a solvent (in our case n- propanol) and a component that
isn’t present in the sample to be analyzed (for instance cyclohexane).
Furthermore we need a calibration solution, which contains a certain amount of each component in the sample and again the solvent n- propanol.

The exact composition of the reference and the calibration solution is shown in the table below:
n- propanol [ml]
Reference Solution       247,5 2,5
Calibration Solution 0,25 0,25 0,25 5,0  
Table 2: Reference and calibration solution

In order to be able to calculate the calibration curve, we need at least 2 points on this curve, CS 1 and CS 2. Both contain the same amount of reference solution, what is different is the amount of calibration solution in the mixture.

Composition of CS 1 and CS 2:
Reference solution
Calibration solution
CS 1 1 0,1
CS 2 1 0,2
Table 3: Composition of CS 1, CS2

As a result of the GC- analysis we obtain a time dependent area for each component. From this area the mass 
of the component can be calculated, wherefore we need a concentration factor, the so called K- value.

Calculation of the K- value:
The mass mi, CS1 of a component i in CS 1 is given by:
whereas mi the mass of component i in the calibration solution is, mcal1 the mass of calibration solution in CS 1 and mcal is the total mass of the calibration solution.

Analog we can write for mi in CS 2:


The mass mcyclo, CS1 of cyclohexane in CS 1 is equal to:


mcyclo is the total mass of cyclohexane in the reference solution, mref,CS1 the mass of reference solution in 
CS 1 and mref the total mass of reference solution.

Same procedure for the mass of cyclohexane in CS 2:


The area of the component i in CS 1 in the chromatogram is given by:


Vinjection is the volume taken by the GC to be analyzed, mi, CS1 is the mass of the component i in CS 1 and
ri is the response factor of the component i.

For the area of cyclohexane the same procedure leads to :


from equation (5) and (6) follows:


whereas Ki is called the concentration factor for component i.
Analog we find for CS 2:


Plotting the area ratios in equation (7) and (8) as a function of the mass of the component i , we obtain a straight line though the origin, whereas the Ki – value is got from the slope.

The concentration factor K has to be determined for each component in the calibration solution.

Analog we can write for the sample solution (sample from the column and reference solution):


Since we know now Ki we are able to calculate the mass of component i in the sample:


Last update: Nov 22, 2000