1.

1 Methodology
Distillation column models are used to provide a preliminary design basis for
the distillation of a mixture of components, based on a set of specifications.
Distillation column models can be made using the following methods:

Fenske-Underwood-Gilliland (FUG) Method an approximate method used

for making preliminary designs and optimization of simple distillation.

Rigorous Method a model used for the final design of distillation towers
involving precise determination of temperatures, pressures, stream flow
rates, stream compositions, and heat-transfer rates at each stage using
material balances, energy balances and equilibrium relations (MESH
equations) for each stage.

1.1.1 FUG Method

In using the FUG method to determine the preliminary design operating
specifications for the column, bubble point and dew point calculations were
required. These calculations were performed using Microsoft excel, according
to the algorithm presented in Figure 1.0. From these calculations, the
operating pressure of the column was determined, along with the condenser
type.
The feed for this project was comprised of essentially non-ideal components,
and as such, initial calculations of equilibrium constants for the feed,
distillate and bottoms product required the use of Van Laar activity
coefficients. These activity coefficients were calculated using Equation 3,
then substituted into Equation 1 to determine the values of the equilibrium
constants, using the modified Raoults law. Also, after calculating the value of
the partial pressure for each component using Equation 2.0, the bubble point

pressure was found using the values of the activity coefficient for each
species.

Using the bubble point pressure calculation, the bubble point

temperature of the feed, distillate and bottoms was determined through the
use of Microsoft excels solver.
According to the algorithm for the FUG method, which is presented in Figure
2.0, an estimation of the split of the two key components is required before
calculations can be made. For this project, methanol was selected as the
light key, while ethanol was the heavy key. The calculated splits for these
components were based on the specification given in the problem statement
which required 95% methanol in the distillate, and 90% ethanol in the
bottoms. These values, along with the estimation for the non key component,
acetone, are presented in table 1.0 below.
Table 1 displays the calculated and estimated splits of the components.
Component
s

Feed , F
(kmol/hr)

Distillate, D
(kmol/hr)

Bottoms ,B
(kmol/hr)

Ethanol

30

27

Acetone

10

Methanol

60

57

Total

1000

68

32

According to the algorithm for FUG, the next step requires an adiabatic flash
calculation on the feed. This step was unnecessary for our case since the
temperature of the feed, 75C , was less than the calculated bubble point
temperature of the feed 96.35 C, thus indicating the feed exists in the liquid
phase region and is subcooled.
In order to use the Fenske equation to determine the minimum number of
stages, Nmin, the relative volatilities of the components were required. The
relative volatility for each component was calculated in Microsoft Excel using

the ratios of the equilibrium constants of each component, relative to the

heavy key component in both the distillate and bottoms. Once N min was
determined, Equation 6.0 was used to determine the molar flow rates, d and
b, for the non key component, acetone, at total reflux . The total distillate and
bottoms flow rates were then recalculated using the acetone compositions
obtained from the Fenske equation. To determine the minimum reflux ratio,
the Underwood equations, Equation 7.0 were utilised.
Although analysis of the distributing components indicated that the system
was class 2, since most of the lighter than light key component, acetone,
distributed mostly to the distillate, calculations were made on a class 1 basis
for design considerations. This concept may be used since this project is
meant to give a preliminary design basis for the separation of the mixture
(Seader and Henley 2006).
The Gilliland correlation, Equation 8.0, was used to determine the actual
reflux ratio and the number of theoretical stages. According to Seader and
Henley, the general rule of thumb used to determine the actual reflux ratio
was the ratio of R/Rmin should equal 1.30 (Seader and Henley 2006). The
reflux ratio was used in the correlation, along with Microsoft excels solver to
determine the actual number of stages, N. The Kirkbride equation, Equation
9.0 was used to determine the feed stage location. Using Excel, the use of
the solver was also required in this stage.

1.1.2 Rigorous Method

The Rigorous method involves a series of flash calculations, being carried out
on each stage of a distillation column. This method is used to calculate the
unknown variables such as temperature, liquid flow rate, vapour flow rate
and the liquid and vapour compositions for each component of a distillation
column. These calculations are made using a mathematical model which
includes material, equilibrium, summation and energy balances (M.E.S.H)
which may be solved using Microsoft Excel, or Matlab.
For this project, since the FUG method approximates that 24 stages were
required to achieve the desired separation, Matlab was used to perform
rigorous since Excel is accurate for the calculation of up to approximately
100 equations.
The mfile used to determine the unknown variables using the rigorous
method is presented in the Appendix below. Initially, the conditions and data
required to solve the problem were stated. Some specifications, such as the
number of stages, feed stage location and reflux ratio were obtained from
the FUG method. The unknown variables for each stage, which include the
temperature, the liquid and vapour flow rates and the liquid and vapour mole
fractions were specified, in that order. The equilibrium constants were then
calculated using Van Laar activity coefficients. Once these preliminary
calculations were performed, material, equilibrium, summation and energy
balances were performed from the reboiler to the feed, and the feed to the
condenser. By using this mfile in the command window of Matlab, with an
appropriate initial set of guess values governed by the results obtained in
FUG, the unknown variables for each stage were solved. The rigorous
method was also used to determine the condenser and reboiler duty, as seen

in the mfile in the appendix. These were calculated by taking overall energy
balances over the condenser and reboiler respectively.

1.1 Discussion
In the configuration of the distillation column, the FUG method was utilized to
determine preliminary design estimates such as the condenser type,
pressure of the column, the number of trays, feed stage and the reflux ratio.
The obtained data was then used in the Rigorous Method to obtain estimates
of unknown variables such as temperatures, stream flow rates and stream
compositions for each stage.
On analysis of the results obtained from the FUG method, since the bubble
point pressure of the distillate was found to be 8.56 psia, according to the
algorithm in Figure 1.0, a total condenser is recommended. The use of a total
condenser indicates that all the vapour leaving the top stage is condensed,
hence the composition of the vapour leaving the top tray, is the same as that
of the liquid distillate product and reflux. The column pressure was
determined using Figure 1.0 which allowed the distillate pressure to be reset
to 30 psia. By assuming no pressure drop in the column, 30 psia was chosen
as the column pressure since it was greater than the feed pressure, and
above atmospheric pressure.
From the FUG method, the calculated bubble point temperature of the feed
was found to be 96.35 C, which is greater than the feed temperature of
75C.

Since the feed temperature was less than the bubble point

temperature, the feed would enter the column in a complete liquid phase,
and is thus classified as a subcooled feed so there is no need to adiabatically
flash the feed.
In order to determine the number of stages required, the minimum number
of equilibrium stages, Nmin, was determined. This corresponds to calculations
under total reflux conditions, and is thus only achieved in theory. From the
Fenske equation, the minimum number of stages required was found to be
11.746 stages. The minimum number of equilibrium stages is influenced by
the nonkey components and their effect on the relative volatility between the
key components. Using this value of N min, the actual number of stages which
was required to achieve the separation was found to be 24 stages according
to Gillilands equation and the feed stage was found to be the 12 th stage
according to Kirkbride.
The reflux ratio affects the number of plates required and thus the
separation efficiency of the column. For a given separation, using a higher
reflux ratio results is less plates being required and vice versa. The minimum
reflux ratio, Rmin, was determined based on a Class 1 separation and was
found to be 2.24. The reflux ratio was found using this value of R min to be
2.907948 as seen in Table 6.0. Operating costs are tied to the reflux ratio
and thus to fuel costs for providing heat to the reboiler. Less reflux is needed
if more stages are added, so operating costs decreases as capital costs
increase (Seader and Henley 2006).
Figure 3.0 displays the results obtained for the temperature variation in the
column for each stage. Since the rigorous method was operated on the 12 th
and 13th feed stage, both set of results were included. The temperatures
obtained using the rigorous method are seen to increase as the number of
stages increase. Since the stages are number from the top to bottom, this
indicates that the temperature at the top stage is lower than the
temperature at the bottom stage. This is due to the presence of the reboiler

at the bottom stage, which provides heat to the column. As such, the reboiler
will have the highest temperature, and as we go up the column, the
temperature decreases. On analysis of the data obtained, the temperature of
the distillate and bottoms product obtained from the FUG method (81.23C
and 92.93C respectively) were less than the temperatures for these same
products using rigorous. These values were 82.28C and 97.05C for the feed
on the 12th stage, and 82.05C and 96.45C for the feed on the 13th stage.
These results are acceptable since for the last stage or the reboiler stage,
the temperature must be above the bubble point of the bottoms in order to
vaporize the component.
On analysis of Figure 4.0, it is observed that the liquid flow rate in the
column fluctuates, where there is a sudden increase and decrease in its
behaviour. The trend for the results using both feed stages is displayed in
this figure. Initially, the liquid flow rate down the column from the top stage
decreases slightly. This decrease in the flow rate is due to vaporization of the
liquid as it flows down the column. As such, with the increasing temperatures
down the column, the liquid flow rate decreases. The sudden increase seen
in the figure is due to the introduction of the components at the feed stage.
From this point the flow continues to decrease slightly until it reaches the last
stage where there is a sharp decrease. This decrease is due to the presence
of the reboiler. Throughout the column, the vapour flow rate steadily
decreases, with a maximum value at the top of the column. This may be due
to the fact that the most volatile components exist at the top of the column
while the least volatile are at the bottom so the vapour flow at the top is
greater.
Figure 5.0 displays the distribution of the components using the FUG method.
These values are in accordance with the specification requirements of 95%
methanol in the distillate and 90% ethanol in the bottoms. The graph
indicated that the largest mole fraction of the light key, methanol exists in
the top stage, while the lowest mole fraction is in the bottoms. This is

expected since methanol is much more volatile than ethanol. Further

analysis shows that ethanol is present in small amounts in the top stage, and
has a large mole fraction in the bottom stage. The nonkey component exists
at a fairly constant mole fraction in both the top and bottom stages.
Using the Rigorous method, the liquid and vapour mole fractions were
calculated as seen in Figures 6.0 and 7.0. In order to compare the mole
fractions obtained from the rigorous method to FUG, only the compositions at
the top and bottom stages were considered. For the top stage in rigorous,
the vapour composition (y) was considered, and for the bottom stage, only
the liquid composition was considered (x). In both graphs, the mole fraction
of the light key component, methanol, decreases from the top to the bottom
of the column. This trend is similar to that of Figure 5.0 for methanol, and
indicates that the more volatile component is present in larger amounts at
the top of the column, as compared to the bottom. The trend for ethanol is
also similar to that of the FUG method, in that the less volatile component is
present in larger amounts in the bottoms. It is also observed that both
graphs of vapour and liquid mole fractions displayed the same general trend
for each component. Analysis of tables 7.0 and 8.0, indicated that the
general trend vapour and liquid mole fractions were similar for each stage.
As you go up the distillation column from the bottom stage, the non key and
light key compositions generally increase as the less volatile heavy key
component condenses and moves own the column. During contact, some of
the light component in the entering liquid is vapourized and leaves with the
vapour, while some of the heavy component in the entering liquid
condenses, and leaves in the liquid.
From the FUG method, the mole fraction of methanol required in the distillate
was calculated to be 0.838, while the mole fraction of ethanol in the bottoms
was found to be 0.844. The rigorous method yielded a value of 0.845 for
methanol in the tops (Table 7.0) and 0.972 for ethanol at the bottoms (Table
7.0). Although the value from rigorous contained an error of 0.84% for the

distillate product, the error for the bottoms product was 15.16%. In order to
reduce this error, and thus bring the value of the ethanol in the bottoms
product to the specification of 90%, several variables were manipulated in
the rigorous calculation. These factors include the reflux ration, number of
stages and the feed location.
According to theory, an increase in the reflux ratio would decrease the
number of stages required to achieve the separation. In this case however,
the operating costs will increase due to fuel costs for the reboiler, while
capital costs will decrease due to the requirement of fewer stages (Seader
and Henley 2006). An increase in the number of stages also relates to better
separation; however more capital is required to build more stages. For this
project, unreasonable solutions were obtained when both the reflux ratio and
number of stages in the column were varied. This may be due to the fact
that although the separation, as seen in table XX is a Class 2 separation, the
reflux ratio was calculated on a Class 1 basis. This was done after seeking
professional advice from the lecturer who advised that a Class 1 calculation
be performed since it would yield a result which is appropriate for
preliminary design considerations.
The best results were obtained when the feed stage was varied to the 13 th
stage instead of the 12th. This is not surprising since the Kirkbride equation
postulates an approximation to the feed stage location. It was determined
that as the feed stage was moved lower down the column, the top
composition becomes less rich in the more volatile component while the
bottoms contains more of the more volatile component. This was seen in the
results as the value of methanol in the distillate was found to be 0.834 (Table
8.0) while the mole fraction of ethanol in the bottoms was 0.946(Table 8.0).
The error calculation for the distillate was reduced to 0.48% for the distillate
and 12.08% for the bottoms.

When the feed stage was set on the 12 th stage, the reboiler and condenser
heat duties required for the operation of the column was found to be
1.2156x107 J/hr and 1.2046x107 J/hr respectively as seen in table XX. These
values were compared to the heat duties for the reboiler and condenser
found for a feed stage set on the 13 th stage which was 1.2182x107 J/hr and
1.2074x107 J/hr respectively. This indicates that the location of the feed stage
in a distillation column affects the heat duties for the column. Operation on
the 13th stage as the feed stage requires an additional 26,000 J/hr for the
reboiler, and 28,000 J/hr for the condenser. A trade off is thus met, where an
additional expense is required to provide energy to the column in order to
obtain the results closer to the desired composition separation. This trade off
is one which must be considered in the final design of the column since lower
heat duties would translate to lower operating costs.

Start

Distillate and
Bottoms
compositions
known or
estimated

Calculate bubble
point pressure(Pd)
of distillate at
120F (49C)
Pd>215
psia

Pd<215 psia

Use Total
condenser (reset
Pd to 30psia if
Pd<30psia

Tb> bottoms
decomposition or
critical temperature

Pd>365 psia

Calculate dew
point pressure(Pd)
of the distillate at
Pd> 365
120F (49C)
psia

Choose a
refrigerant so as to
operate a partial
condenser at 415
psia(2.86 MPa)

Use
Partial
Condens

Estimate
bottoms
pressure
(Pb)

Calculate the
bubble point
temperature(Tb) of
Tb> bottoms
bottoms at Pb
decomposition or
critical temperature

Lower pressure Pb
appropriately and
recompute Pd and
Td

Figure 1.0: Algorithm for establishing distillation-column pressure and

condenser type

Start
Specified Feed

Specify splits of two

components

key

Estimate splits of nonkey

components

Determine column
Bubble point/
dew point calculations
pressure and type
of
condenser

Repeat
only if
Flash the feed at column
estimated
pressure Adiabatic flash procedure
and
calculated
splits of
Calculate
minmum
Fenske equation
theoretical
stages
nonkey
component
s differ
Calculate splits of nonkey
Fenske equation
components

Calculate minimum
Underwood equations
reflux ratio

Calculate actual
theoretical stages
for
Gilliland correlation
specified reflux ratio >
minimum value

Calculate feed Kirkbride

stage
equation
location

Calculate condenser
Energy-balance equations
and reboiler duties

Exit

Figure 2.0: Algorithm for multicomponent distillation by FUG method.

Equations
Fug Method

Modified Raoults Law:

K i=

iLP
P

Equation 1.0

Antoine Equation:

log 10 P si = A

B
T +C

Equation 2.0

Van Laars Equations for (ternary mixtures):

Equation 3.0 5.0

Bubble Point Pressure

Eqn 4

Fenske Equation( for minimum number of stages):

Eqn 5

Fenske Equation( for split of non-key components):

Eqn 6

Underwood Equation for minimum reflux ratio

Eqn 7

Gilliland Correlation Equation for theoretical number of stages

Eqn8

Kirkbride Equation for feed stage location

Eqn9