Mitchell, A (Albert) Mr

u16172028

[email protected]

CMO 310 MASS TRANSFER 310

DESIGN PROJECT:
DISTILLATION COLUMN DESIGN FOR THE
SEPARATION OF β-PINENE AND LIMONENE

DEPARTMENT OF CHEMICAL ENGINEERING

UNIVERSITY OF PRETORIA

2019-04-13
Nomenclature

Symbol Description
𝑃 Pressure (atm)
𝑇 Temperature (°𝐶)
𝑇𝑏𝑝 Normal boiling point (°𝐶)
𝑀𝑊 Molecular weight (g/mol)
𝜆 Heat of evaporation (kJ/mol)
𝐶𝑝 Constant pressure heat capacity (𝑘𝐽/𝑚𝑜𝑙𝐾)
𝜌𝑣 Vapour density (kg/𝑚3 )
𝜌𝐿 Liquid density (kg/𝑚3 )
𝐷𝑐 Column diameter (m)
𝐻𝑐 Column height (m)
𝑙𝑡 Plate spacing (m)
𝑑𝑜 Tray hole diameter (mm)
𝐹 Feed rate (kmol/h)
𝐷 Distillate rate (kmol/h)
𝐵 Bottoms rate (kmol/h)
𝑞 Quality of feed ()
𝐻𝑣 Enthalpy of vapour stream (kJ/h)
𝐻𝐹 Enthalpy of feed stream (kJ/h)
𝑅𝑚𝑖𝑛 Minimum reflux ratio ()
𝑁 Number of equilibrium stages ()
𝑁𝐹 Feed stage number ()
𝐸𝑜 Efficiency correlation ()
𝛼 Relative volatility ()
𝜇 Viscosity (Ns/𝑚2 )
𝑢
̂𝑣 Maximum allowable vapour velocity (m/s)
̂
𝑉 𝑤 Vapour flow rate (kg/h)
𝑙𝑑𝑖𝑠𝑒𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 Height of vapour disengagement region (m)
𝑙ℎ𝑜𝑙𝑑𝑢𝑝 Height of liquid holdup region (m)
𝑙𝑝 Sieve pate hole pitch (mm)
𝑙𝑤 Weir length (m)
ℎ𝑤 Weir height (mm)
𝐹𝐿𝑉 Flow parameter ()
𝐿𝑤 Liquid flow rate (kg/h)
𝑢𝑓 Flooding velocity (m/s)

i
 𝑢𝑜𝑝 Operating velocity (m/s)
ℎ𝑜𝑤 Weir liquid crest height (mm)
𝑢ℎ Minimum vapour design velocity (m/s)
ℎ𝑑 Dry plate head (mm)
ℎ𝑟 Residual head (mm)
ℎ𝑡 Total plate head (mm)
Δ𝑃𝑡 Plate pressure drop (kPa)
𝑡𝑠 Column shell thickness (mm)
𝑓𝑎𝑠 Axial stress (Pa)
𝑓𝑐𝑠 Circumferential stress (Pa)
𝑊𝑠ℎ𝑒𝑙𝑙 Weight of shell (N)
𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑠ℎ𝑒𝑙𝑙 Dead weight of shell (Pa)
𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑖𝑛𝑠 Dead weight of insulation (Pa)
𝑓𝑑𝑒𝑎𝑑 𝑤𝑒𝑡,𝑙𝑖𝑞 Dead weight of liquid in column (Pa)
𝑓𝑡𝑜𝑡𝑎𝑙 Total dead weight of column (Pa)
𝑄𝐵 Boiler heat duty (kW)
𝑄𝐶 Condenser heat duty (kW)
𝐴𝐵 Area of heat exchanger (𝑚2 )
Δ𝑇𝑚𝑒𝑎𝑛(𝑐𝑜𝑐𝑢𝑟𝑟𝑒𝑛𝑡) Mean temperature of heat exchanger (°𝐶)

ii
Table of Contents

Nomenclature ......................................................................................................................... i

Table of Contents ................................................................................................................. iii

Table of Figures.................................................................................................................... iv

Table of Tables ..................................................................................................................... iv

1. Introduction............................................................................................................................. 1

2. Column Diagram ..................................................................................................................... 2

3. Distillation Column Design Summary ..................................................................................... 3

4. Assumptions and Vapour-Liquid Equilibrium Data ................................................................. 4

5. Column Conditions and Calculations ....................................................................................... 7

6. Design Optimisation .............................................................................................................. 12

7. Mechanical Design ................................................................................................................ 15

Tray spacing, 𝑙𝑡 .................................................................................................................... 15

Column diameter, Dc ............................................................................................................ 15

Column Height ..................................................................................................................... 16

Plate Type ............................................................................................................................ 16

Sieve plate design and calculations ....................................................................................... 18

Stress Analysis of Column .................................................................................................... 25

Column Support Structure .................................................................................................... 28

Feed point............................................................................................................................. 30

Packed column ..................................................................................................................... 30

8. Heat Exchangers ................................................................................................................... 31

Boiler ................................................................................................................................... 31

Condenser ............................................................................................................................ 33

9. References............................................................................................................................. 34

iii
Table of Figures

Figure 1: Column diagram..........................................................................................................2

Figure 2: x-y diagram for binary mixture of β-pinene(1) and limonene(2)...................................4
Figure 3: T-x-y diagram for binary mixture of β-pinene(1) and limonene(2). ..............................5
Figure 4: McCabe Thiele diagram of Rmin times 1.80. ................................................................9
Figure 5: McCabe Thiele diagram of Rmin times 2.85. .............................................................. 10
Figure 6: McCabe Thiele diagram of Rmin times 4.00. .............................................................. 10
Figure 7: Sieve tray design with supports (Sinnott & Towler, 1999). ........................................ 18
Figure 8: Cross flow on single pass sieve tray. .......................................................................... 18
Figure 9: Performance diagram for a sieve plate design. ........................................................... 19
Figure 10: Perforated area and column area ratio. ..................................................................... 20
Figure 11: Flow parameter vs the constant required to calculate flooding velocity .................... 21
Figure 12: Fractional entrainment related to the flow parameter. .............................................. 22
Figure 13: K2 vs the summation of the weir height and liquid crest height. ............................... 23
Figure 14: Orifice correlation. .................................................................................................. 24
Figure 15: Performance diagram obtained from Aspen. ............................................................ 25
Figure 16: Straight support skirt (Sinnott & Towler, 1999). ...................................................... 29
Figure 17: Support skirt dimensions (Sinnott & Towler, 1999). ................................................ 29
Figure 18: Heat exchanger initial cocurrent temperatures. ........................................................ 32
Figure 19: Kettle boiler internal design (Subramanian, 2011). .................................................. 32

Table of Tables

Table 1: Column specifications like pressure, efficiency and reflux ratio. ...................................3
Table 2: Component Data obtained from various sources. ..........................................................3
Table 3: The column dimensions such as type, height and tray whole diameter...........................4
Table 4: Assumptions made to obtain solution to McCabe Thiele diagram. ................................6
Table 5: Flowrates of streams. ....................................................................................................7

iv
Table 6: Relationship between the multiplication factor and the column stages. ........................ 11
Table 7: User specified Aspen parameters. ............................................................................... 13
Table 8: Sensitivity analysis results after varying reflux ratio. .................................................. 13
Table 9: Sensitivity analysis for plate number and feed plate. ................................................... 14
Table 10: Efficiency correlation at various points in column. .................................................... 14
Table 11: Plate type comparison. .............................................................................................. 17
Table 12: Insulation material selection. .................................................................................... 27
Table 13: Dimensions for support skirt. .................................................................................... 30
Table 14: Reboiler types. .......................................................................................................... 31

v
1. Introduction

Methods of separation are investigated in this report to develop an optimal design for the binary
system of β-pinene and limonene. Stage distillation and packed column distillation are the two
main methods being considered for this system.

Limonene and β-pinene are part of the terpene class of compounds. Terpenes are aromatic
hydrocarbons that are a major part of plant resins. Most terpenes have a very strong smell, which
helps in attracting pollinators and repelling predators from the plant. Terpenes have been found to
have a wide variety of medicinal benefits like limonene which has anticarcinogenic properties.
These compounds are also known as essential oils.

According to (WebMD, sa) limonene is used in weight loss promotion, cancer prevention and
treatment and as a treatment for bronchitis. It is naturally occurring in the rind of various fruits
including but not limited to lemons.

Pine oil also known as β-pinene is an essential oil that originated from the bark of the Pinus
sylvestris tree. Similarly to limonene, it also has medical benefits including anti-inflammatory
properties (Levy, 2017).

1
2. Column Diagram

Figure 1: Column diagram.

2
3. Distillation Column Design Summary

Table 1: Column specifications like pressure, efficiency and reflux ratio.

Specifications
Pressure (absolute) 1 atm
Minimum temperature (top of column) 165 ℃
Maximum temperature (bottom of column) 176 ℃
Overall efficiency 66.0 %
Reboiler efficiency 85 %
Number of actual stages 77 + PR
Optimal feed stage 35
Reflux ratio 15.23
Boilup ratio 17.35
Reboiler heat load 5300 kW
Condenser heat load -4200 kW
Feed temperature 20 ℃

Table 2: Component Data obtained from various sources.

Component BP (1) LM (2) Average

Boiling point, TBP (°C) 165 176
Molecular Weight, MW (g/mol) 136.23 136.24
Heat of vaporization, λ at (kJ/mol) 41.6 39.4 40.5
Specific heat capacity, Cp (kJ/molK 272.71 269.8 271.3
Liquid density (𝑘𝑔/𝑚3 ) 738.044
Vapour density (𝑘𝑔/𝑚3 ) 3.95
Surface tension (𝑁/𝑚) 0.0159

3
 Table 3: The column dimensions such as type, height and tray whole diameter.

Column Dimensions
Column type Stage distillation
Column diameter, Dc (m) 2.8
Column height, Hc (m) 49.7
Tray spacing (m) 0.6
Tray hole diameter, do (mm) 6.0

4. Assumptions and Vapour-Liquid Equilibrium Data

Vapour-Liquid Equilibrium data (VLE) was collected also detailing the interaction between a
binary mixture of limonene and β-pinene at atmospheric pressure. The data was experimentally
obtained in a miniature equilibrium still and correlated using the Antoine equations for the
respective components (Bernardo-Gil & Albertina Ribeiro, 1993).

The raw VLE can be seen in Figure 2 below. An azeotrope does not exist in this system thus the
assumption can be made that it mixes almost ideally as there is very slight deviation from ideal
mixing. This is due to negligible interactions between unlike molecules.

Figure 2: x-y diagram for binary mixture of β-pinene(1) and limonene(2).

4
In Figure 3 below the relationship between liquid- and vapour mole-fraction and temperature can
be seen. The boiling point of β-pinene and limonene is 165 ℃ and 176 ℃ respectively. This
indicates that our more volatile component (1) will be β-pinene.

Figure 3: T-x-y diagram for binary mixture of β-pinene(1) and limonene(2).

The assumptions summarized in Table 4 are made to ease in the preliminary design. Constant
molal overflow (CMO) is assumed because the column will be well insulated (adiabatic). The heat
liberated by 1 mole of vapour during condensation is approximately equal to the heat absorbed by
one mole of liquid upon vaporization. The flowrates in the rectifying- and stripping section
respectively, will be assumed to be constant.

Ideal mixing is assumed as there exists no azeotrope in our system. As a result, we know there will
be no forces of attraction/repulsion between unlike molecules that would cause the mixture to
deviate a significant amount from Raoult’s law.

5
 Table 4: Assumptions made to obtain solution to McCabe Thiele diagram.

Assumption Implication
Constant molal overflow (CMO)
1. Column is adiabatic
2. Specific heat changes are small
compared to latent heat changes.
3. Latent heat of vaporization per mole is
constant.
4. Saturated liquid and vapour lines on h-
x-y diagram are in parallel.

Ideal mixing No interactions between unlike molecules

Cp is constant The constant pressure heat capacity does not
vary with temperature

λ is constant The heat of vaporisation does not vary with

temperature
Relative volatility is constant over mole McCabe Thiele diagram can be plotted on
fraction intervals Excel by using relative volatility for steps.

6
5. Column Conditions and Calculations

For the design of a column an initial design needs to be completed that can be optimized in
computer software. The McCabe Thiele method uses the vapour-liquid equilibrium data to obtain
an integer number for the theoretical equilibrium stages/plates required to complete the specified
separation. This method requires numerous calculations before drawing the diagram.

A material balance needs to be completed over the entire column to obtain the distillate- and
bottoms flow rate. The total mole balance

𝐹 = 𝐷+𝐵 1.

and the component mole balance

𝑥𝐹 𝐹 = 𝑥𝐷 𝐷 + 𝑥𝐵 𝐵 2.

are solved for simultaneously using F=50 kmol/hr, xF = 0.5, xD= 0.995 and xB= 0.005 to yield the
flowrates summarised in Table 5 below.

Table 5: Flowrates of streams.

Stream Flowrate (kmol/hr)

F 50.00
D 25.00
B 25.00

The first obstacle in developing our initial design is calculating the quality of the feed which
describes in what phase the feed is, with

𝐻𝑣 − 𝐻𝐹 3.
𝑞=
𝜆𝑉𝑎𝑝

Cp and λVap data, reflected in Table 2, is obtained for each individual component in the binary
mixture. Due to the assumption of ideal mixing applying the enthalpy of the vapour and enthalpy
of the feed is obtained with

7
 𝐻𝑣 (𝑇, 𝑦) = 𝑦1 (𝜆1 + 𝐶𝑝1,𝑣 (𝑇𝑑𝑟𝑢𝑚 − 𝑇𝑟𝑒𝑓 )+ 𝑦2 (𝜆2 + 𝐶𝑝2,𝑣 (𝑇𝑑𝑟𝑢𝑚 − 4.
𝑇𝑟𝑒𝑓 )
and

ℎ𝐹 (𝑇𝐹 , 𝑦) = 𝑧1 𝐶𝑝1,𝑣 (𝑇𝑑𝑟𝑢𝑚 − 𝑇𝑟𝑒𝑓 ) + 𝑧2 𝐶𝑝2,𝑣 (𝑇𝑑𝑟𝑢𝑚 − 𝑇𝑟𝑒𝑓 ) 5.

Knowing that the feed is subcooled at T F = 20 °C because is it far less than the normal boiling
point of either component at P = 1 atm, we can calculate the quality of the feed to be

𝑞 = 1.84 6.

Next, we must get the slope of the feed line. We use the equation

−𝑞 7.
𝑠𝑙𝑜𝑝𝑒 𝑜𝑓 𝑓𝑒𝑒𝑑 =
1−𝑞

𝑠𝑙𝑜𝑝𝑒 𝑜𝑓 𝑓𝑒𝑒𝑑 = 2.19 8.

The slope of the feed line is plotted on the VLE graph. The minimum reflux ratio, Rmin, needs to
be obtained from the VLE graph so that Ractual can be calculated. Plotting the feed line and then
plotting the top operating line such that a pinchpoint is formed and the y-intercept can be read off,
Rmin is calculated with the Equation (Linninger, sa):

𝑥𝐷 9.
𝑅𝑚𝑖𝑛 = −1
𝑦 − 𝑖𝑛𝑡𝑒𝑟𝑐𝑒𝑝𝑡

Rmin is calculated to be 4.31. Ractual, under normal circumstances, abides by the following rule of
thumb:

1.2 ≤ 𝑅𝑎𝑐𝑡𝑢𝑎𝑙 ≤ 1.5 10.

After Ractual is calculated, the top operating line is given by:

𝑅𝑎𝑐𝑡𝑢𝑎𝑙 𝑥𝐷 11.
𝑦=( )𝑥 +
𝑅𝑎𝑐𝑡𝑢𝑎𝑙 + 1 𝑅𝑎𝑐𝑡𝑢𝑎𝑙 + 1

8
Knowing the top operating line equation as well as the feed line, between the point of intersection
of the two and xB the bottom operating line can be calculated with:

𝐿̅ 𝐵 12.
𝑦 = ̅ 𝑥 + 𝑥𝐵 ̅
𝑉 𝑉

After this is done, the stepping can be done on the McCabe-Thiele diagram and the number of
theoretical stages and the feed stage is known. In this design the rule of thumb was utilised for the
first estimate. It was immediately clear that this Ractual would be too low as there were too many
theoretical plates. After this different factors above the rule of thumb were used as can be seen in
Figure 4, Figure 5 and Figure 6 to get a more realistic number of equilibrium stages.

Figure 4: McCabe Thiele diagram of Rmin times 1.80.

9
 Figure 5: McCabe Thiele diagram of Rmin times 2.85.

Figure 6: McCabe Thiele diagram of Rmin times 4.00.

Table 6 summarises the results of all the McCabe-Thiele diagrams that were drawn. For the next
stage of the design N will be taken as 53 (including the partial reboiler) and the feed stage will be
21 counting from the top.

10
Table 6: Relationship between the multiplication factor and the column stages.

Rmin*Factor Stages, N Optimum Feed Stage,

NF-Opt (from bottom)
Rmin*1.50 83 30
Rmin*1.80 75 26
Rmin*2.85 60 22
Rmin*4.00 53 21

11
6. Design Optimisation

To begin an optimisation an appropriate choice of a thermodynamic model must be made in Aspen.

The selection of such a model greatly influences the optimisation that will occur later in the design
and thus this decision should be undertaken only after thorough research and investigation.

In this design the thermodynamic model selected was the SRK–EOS (equation of state) given by:

𝑅𝑇 𝑎 13.
𝑃= −
𝑣 − 𝑏 𝑣(𝑣 − 𝑏)

where 𝑎 represents the attraction parameter and 𝑏 represents the repulsion parameter of the
effective molecular volume.

This EOS was selected after consulting a research paper detailing the selection of a thermodynamic
model for calculating properties of terpenes. A set of EOS’s were used and in parallel with this
experimental data was gathered to compare to the calculated values for critical properties of
limonene as well as α-pinene. The experimental- and calculated values were compared through
calculating the %𝐴𝑅𝐷 (average relative deviation) for various properties like critical pressure as
well as temperature. It is given by:

𝑁 14.
1 𝑋 𝑒𝑥𝑝 − 𝑋 𝑐𝑎𝑙𝑐
%𝐴𝑅𝐷 = ∑ | |
𝑁 𝑋 𝑐𝑎𝑙𝑐
𝑖=1

The %𝐴𝑅𝐷 for estimating pressure was found to be 2.495 % indicating that this EOS is 97.5 %
accurate for terpenes for calculating properties (Martins et al, 2017). As a result of this accuracy
this thermodynamic model was selected because it will produce the most accurate results.

The discussed thermodynamic model is used, along with the following initial parameters:

12
 Table 7: User specified Aspen parameters.

Parameter Value
Stages, N 53
Feed Stage 21
Reflux ratio 17.23
Feed temperature 20 °C
Feed pressure 1 atm

A sensitivity analysis was the first optimisation step done in Aspen. The reflux ratio was varied
and the xD and xB recorded and summarised in Table 8. The stages were kept constant at 53 and
the feed plate was 21 (from the top) throughout the analysis.

Table 8: Sensitivity analysis results after varying reflux ratio.

Reflux ratio xD xB
13.72 0.9941 0.0059
14.22 0.9944 0.0056
14.73 0.9947 0.0053
15.23 0.9950 0.0050
15.73 0.9953 0.0047
16.23 0.9955 0.0045
16.73 0.9957 0.0043
17.23 0.9958 0.0042

Ractual is selected to be 15.23 as it gives an optimum separation with a lower reflux than originally
chosen. This optimisation step is done to decrease the energy cost of the column as it will be a
recurring cost, unlike adding an additional plate into the column which is a once-off cost during
construction and design.

In the following sensitivity analysis feed stage and total stages were varied to identify what number
of plates will be the most efficient for the separation needed. The results can be seen in Table 9
below:

13
 Table 9: Sensitivity analysis for plate number and feed plate.

N Nfeed xD xB
46 24 0.9906 0.0094
47 24 0.9915 0.0085
48 22 0.9921 0.0079
49 25 0.9931 0.0069
50 23 0.9936 0.0064
51 24 0.9943 0.0057
52 25 0.9949 0.0051
53 22 0.9947 0.0053
53 23 0.9950 0.0050
53 24 0.9952 0.0048

In conclusion after optimising the Ntheo is now 51 with a partial reboiler and a total condenser. This
is the least amount of plates and the lowest feed that will give the wanted separation of β-pinene
and limonene. The feed stream is to be fed 23 plates from the top of the column.

We determine the Eo (efficiency correlation) at 4 places in the column namely, top, bottom, above
the feed stage and below the feed stage with

𝐸𝑜 = 51 − 32.5 log(𝛼𝜇) 15.

where 𝛼 is the relative volatility and 𝜇 is the viscosity of the liquid flowing on the plate (Sinnott
& Towler, 1999). It is calculated at the 4 different places to ensure we take a conservative approach
to the design. In Table 10 the smallest Eo is 0.66 thus this will be used in the follow-up calculations.

Table 10: Efficiency correlation at various points in column.

Plate Eo
1 0.660
22 0.664
24 0.665
51 0.670

We calculate the real number of stages using the following equation adapted from Sinnott &
Towler (1999):

14
 𝑁𝑡ℎ𝑒𝑜 16.
𝐸𝑜 =
𝑁𝑟𝑒𝑎𝑙

Rearranging for Nreal

𝑁𝑡ℎ𝑒𝑜 17.
𝑁𝑟𝑒𝑎𝑙 =
𝐸𝑜

where Nreal is equal to 78 (Sinnott & Towler, 1999). The smallest Eo was chosen in Equation
because it will yield the highest number of plates as it is in the denominator. This amount of stages
will be used in our mechanical design as well as any further calculations.

7. Mechanical Design

Tray spacing, 𝒍𝒕

In mechanical design of distillation columns, the tray spacing is normally dependant on the column
diameter. According to Sinnott & Towler (1999) the tray spacing is between 0.1 to 1.0 m. For a
column with a diameter greater than 1 m, the spacing should be in the range of 0.3 to 0.6 m.
Another factor influencing the tray spacing is maintenance, as there should be enough space
available between plates for someone to be able to comfortably clean it.

For this design 𝑙𝑡 is selected to be 0.6 m as it will allows efficient crawlspace for maintenance and
fall within the range suggested by Sinnott & Towler (1999).

Column diameter, Dc

The primary factor influencing the column diameter is the vapour flow rate inside the column. If
the vapour velocity is too high there will be an uncontrolled amount of liquid entrainment on the
plate which greatly influences the separation’s efficiency.

The maximum allowable velocity of the vapour

15
 1
𝜌𝐿 − 𝜌𝑉 2 18.
𝑢̂𝑣 = (−0.17𝑙𝑡2 + 0.27𝑙𝑡 − 0.047) [ ]
𝜌𝑉

The maximum allowable velocity is calculated to be 0.733 m/s.

The column diameter is given by:

̂
4𝑉𝑤 19.
𝐷𝑐 = √
𝜋𝜌𝑣 𝑢̂𝑣

which yields Dc to be 2.73 m. This is rounded up to 2.8 m to allow for a conservative diameter and
to match the nearest standard head size for columns.

Column Height

Column height is calculated with:

𝐻𝑐 = 𝑁𝑟𝑒𝑎𝑙 × 𝑙𝑡 + 𝑙𝑑𝑖𝑠𝑒𝑛𝑔𝑎𝑛𝑔𝑒𝑚𝑒𝑛𝑡 + 𝑙ℎ𝑜𝑙𝑑𝑢𝑝 20.

Where 𝑙𝑑𝑖𝑠𝑒𝑛𝑔𝑎𝑛𝑔𝑒𝑚𝑒𝑛𝑡 is to allow enough space at the top for vapor disengagement and 𝑙ℎ𝑜𝑙𝑑𝑢𝑝
account for the liquid hold up that occurs at the bottom of the column. We calculate the height of
the column

𝐻𝑐 = 78(0.6) + 1.4 + 1.5 = 49.7 𝑚 21.

Plate Type

Selecting a plate type for a column is quite difficult because so many different options exist, each
with their own benefits and shortcomings. In Table 11 the 3 mainstream tray types are compared
by looking at a few design considerations that is relevant to this separation:

16
 Table 11: Plate type comparison.

Considerations Sieve Bubble Cap Valve

Cost Least expensive Most expensive Moderately
expensive
Capacity Highest capacity Second highest Lowest capacity
capacity
Efficiency All three have very similar Murphree efficiencies when they are
operating over flow rates they were designed for.
Pressure drop Lowest pressure Highest pressure Moderate pressure
drops drops drops

It is obvious that sieve plates are the logical choice as it is relatively inexpensive, has very similar
Murphree efficiencies than the other two options considered and has a very negligible pressure
drop. Sieve plates will be utilized in my design. In a sieve tray the vapour is passed through
perforated holes on the plate and the liquid flows across the plate into the downcomer and onto the
next plate. This setup allows the vapour to make enough contact with the liquid so that mass
transfer can occur.

A ring and beam support will be utilized so that the trays are fully supported to handle the liquid
that will be on the tray at any given time. The ring support is welded to the column and will be 50
mm wide. The beams are also 50 mm wide and will be spaced 0.6 m apart meaning there will be
4 beams supporting each tray. Each tray will have a manhole cut onto the perforated area that is
0.6 m by 0.6 m to allow a decent sized adult to fit through during maintenance as well as
construction.

17
 Figure 7: Sieve tray design with supports (Sinnott & Towler, 1999).

Sieve plate design and calculations

The flow for the column is chosen as a cross flow single pass. This is chosen to lessen the costs of
manufacturing the sieve plates and because it will provide efficient separation for our binary
mixture. The implication of this is that each plate only has a single downcomer as seen in Figure
8 (Sinnott & Towler, 1999).

Figure 8: Cross flow on single pass sieve tray.

The hole size for the perforated holes are normally selected to be between 2.5 to 19 mm but
according to Sinnott & Towler (1999) values closer to 5 mm are preferred to avoid fouling taking
place on the tray. In this design we select the material of the plate to be stainless steel type 410
18
which is a martensitic stainless steel that has a high tensile strength and is corrosion resistant even
at higher temperatures. The hole size will be 6 mm. The thickness of our plate will be 3 mm. If the
tray diameter is 2 times the tray thickness the plate holes can be punches instead of being drilled
which drastically reduces the manufacturing cost (Sinnott & Towler, 1999).

The layout on each individual sieve tray is selected to be an equilateral triangle arrangement where
the pitch, the distance between two holes,

𝑙𝑝 = 2.6 × 5 = 13 𝑚𝑚 22.

Figure 9 gives a brief overview of all the factors that need to be considered for the design of the
sieve plates. There needs to be a certain percentage of flooding and checks need to be done to
ensure that no weeping occurs . There is an area of operation on the diagram in which the separation
will be most efficient.

Figure 9: Performance diagram for a sieve plate design.

Weir length plays a crucial role as it dictates the area of the downcomer. A good rule of thumb
according to Sinnott & Towler (1999) is to select the weir length to be between a factor of 0.6 to
0.85 times the column diameter. For this design we select the factor as 0.71:

19
 𝑙𝑤 = 0.711 × 2.8 = 1.99 𝑚 23.

This is rounded up to 2 m to make construction easier.

The weir height directly affects the volume of all the liquid on the plate. This means it can greatly
influence the efficiency of the separation. If the weird length is also chosen too low or high, it
increases the pressure drop over each plate. In Sinnott & Towler (1999) it is suggested to pick the
height of the weir between 40 to 50 mm. In this design it is selected to be 42 mm.

The total column area as a ratio of the perforated are is read off from Figure 10 to be 8.7%

Figure 10: Perforated area and column area ratio.

The flow parameter, which is needed to calculate entrainment, is given by

𝐿𝑤 𝜌𝑣 24.
𝐹𝐿𝑉 = √
𝑉𝑤 𝜌𝐿

The flow parameter is calculated to be 0.0687. Reading the constant K1 is read off to be 0.12 from
Figure 11.

20
 Figure 11: Flow parameter vs the constant required to calculate flooding velocity

After obtaining the constant, the flooding velocity can now be obtained. The flooding on the plate
will fix the efficiency of the separation.

𝜌𝐿 − 𝜌𝑣 25.
𝑢𝑓 = 𝐾1 √
𝜌𝐿

𝑢𝑓 = 1.64 𝑚/𝑠 26.

For design purposes a percentage flooding should be chose to be between 80 to 85 % (Sinnott &
Towler, 1999). In this design we will assumed the flooding to be 82 % as it falls in this given
range. The operating velocity can be obtained from the following equation that is rearranged:

𝑢𝑜𝑝
% 𝐹𝑙𝑜𝑜𝑑𝑖𝑛𝑔 = 27.
𝑢𝑓

𝑢𝑜𝑝 = % 𝐹𝑙𝑜𝑜𝑑𝑖𝑛𝑔 × 𝑢𝑓 = 1.34 𝑚/𝑠 28.

The fractional entrainment can be read of from Figure 12 to be ψ = 0.051 kg/ kg gross liquid flow.
This is below 0.1 which according to Sinnott & Towler (1999) means that it will have a negligible
effect on the efficiency.

21
 Figure 12: Fractional entrainment related to the flow parameter.

The liquid weir crest is the height of the liquid flowing over the weir. This crest is calculated with
the following equation, where Lw represents the liquid flow rate:

2
𝐿𝑤 3 29.
ℎ𝑜𝑤 = 750 [ ]
𝜌𝐿 𝑙𝑤

ℎ𝑜𝑤 = 34.25 𝑚𝑚 liquid 30.

The constant K2 is read off Figure 13 to be equal to 30.61. This constant is unitless.

22
 Figure 13: K2 vs the summation of the weir height and liquid crest height.

This constant is used to calculate the minimum design vapour velocity, uh :

[𝐾2 − 0.9(25.4 − 𝑑ℎ )]
𝑢ℎ = 1
31.
(𝜌𝑣 )2

𝑢ℎ = 6.6 𝑚/𝑠 32.

The following calculation are all done to quantify the pressure drop over a plate, taking various
factors into consideration that would influence the pressure. First the dry plate pressure drop is
calculated by:

𝑢ℎ 2 𝜌𝑣
ℎ𝑑 = 51 [ ] 33.
𝐶0 𝜌𝑙

but first the orifice coefficient needs to be read of but first Ah/Ap needs to be calculated from the
following equation:

2
𝐴ℎ 𝑑ℎ 34.
= 0.9 [ ]
𝐴𝑝 𝑙𝑝

𝐴ℎ
= 0.192 35.
𝐴𝑝

23
It is read of from Figure 14 to be C0 = 0.8.

Figure 14: Orifice correlation.

ℎ𝑑 = 6.49 𝑚𝑚 𝑙𝑖𝑞𝑢𝑖𝑑 36.

Next the residual head is calculated with:

12.5 × 103
ℎ𝑟 = 37.
𝜌𝑙

ℎ𝑟 = 16.94 𝑚𝑚 liqui 38.

This brings the total head drop to:

ℎ𝑡 = ℎ𝑑 + (ℎ𝑤 + ℎ𝑜𝑤 ) + ℎ𝑟 39.

ℎ𝑡 = 99.68 𝑚𝑚 𝑙𝑖𝑞𝑢𝑖𝑑 40.

The pressure drop per plate is:

24
 Δ𝑃𝑡 = 9.81 × 10−3 ℎ𝑡 𝜌𝐿 41.

Δ𝑃𝑡 = 0.721 𝑘𝑃𝑎 42.

This is considered acceptable according to Sinnott & Towler (1999) and thus the pressure drop
will not greatly affect our separation.

As an additional resource, the following graph was obtained from Aspen. It shows that the column
complies with the requirements to not weep, even when operating at a minimum vapour flow rate.
The operating point falls well within the area where the column is operating efficiently.

Figure 15: Performance diagram obtained from Aspen.

Stress Analysis of Column

A crucial part of the mechanical design of a distillation column is the stress analysis of the column
as well as the structural design and support. Without the consideration of these factors, the design
will never be able to realise as it will be unsafe for operation. The following equations and
considerations were obtained from (Brownell & Young, 1959)

The first calculation given by the equation below, will tell us the minimum thickness of the column
shell ts:

25
 𝑝𝐷𝑐
𝑡𝑠 = 43.
2𝑓𝐽 − 𝑝

Where the p represents the internal column pressure which is chosen to be about 5 to 10 % higher
than the column’s operating pressure and the J represents the joint efficiency which is selected to
be 0.8 to allow for a conservative estimate. The f represents the allowable stress for the shell
material which is selected to be stainless steel SA-312 which has an allowable stress of 109 MPa.

(101325)(2.8)
𝑡𝑠 = 45.
2(109 × 106 )(0.8) − (101325)

𝑡𝑠 = 1.628 𝑚𝑚 ≈ 2 𝑚𝑚 46.

The shell thickness is calculated to be 2 mm. We use a thickness of 7 mm to allow for corrosion.
Next the axial stress, fas, can be calculated. This stress results from the internal pressure in the
vessel. This is given by:

𝑝𝐷𝑜
𝑓𝑎𝑠 = 47.
4𝐽𝑡𝑠

(101325)(2.804)
𝑓𝑎𝑠 = = 12.7 × 106 𝑃𝑎 48.
4(0.8)(7 × 10−3 )

This is related back to the circumferential pressure with:

𝑓𝑐𝑠 = 2 × 𝑓𝑎𝑠 = 25.4 × 106 𝑃𝑎 49.

Next, the compressive stress caused by the column, plates, insulation amongst other will be
considered and quantified. The first stress calculated is the stress caused by the dead weight of the
column, where 𝜌𝑠 is the density of the chosen stainless steel.

𝜋
𝑊𝑠ℎ𝑒𝑙𝑙 = 𝑚𝑠ℎ𝑒𝑙𝑙 × 𝑔 = [ (𝐷𝑜2 − 𝐷𝑐2 )𝜌𝑠 𝐻𝑐 ] 𝑔 50.
4

26
 𝜋
𝑊𝑠ℎ𝑒𝑙𝑙 = 𝑚𝑠ℎ𝑒𝑙𝑙 × 𝑔 = [ 4 ((2.804)2 − 51.

(2.8)2 )(8000)(49.7)] (9.81) = 68669.42 𝑁

The dead weight of the column is then calculated

𝑊𝑠ℎ𝑒𝑙𝑙
𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑠ℎ𝑒𝑙𝑙 = 𝜋 52.
(4 (𝐷𝑜2 − 𝐷𝑐2 )

𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑠ℎ𝑒𝑙𝑙 = 3.9 × 106 𝑃𝑎 53.

The column shell needs to be well insulated so that it can operate as close to possible to adiabatic.
A choice of insulation material needs to be made from the two materials presented in Table 12.

Table 12: Insulation material selection.

Calcium Silicate Fiber glass

Thermal conductivity 0.06 0.04
density 200 100

In this design fiber glass is chosen as it has the lower thermal conduction, meaning most of the
heat is maintained inside this column and enough heat transfer can take place between passing
streams. Another reason it is chosen is because it is also the safest option in terms of protecting
the people that will work and interact with the column as they are at risk of suffering severe burns
if they column is not well insulated. The compressive stress of the insulation is calculated by taking
the thickness as 0.08m (Sinnott & Towler, 1999). The equation is given by:

[𝜋𝐷𝑖𝑛𝑠 𝜌𝑖𝑛𝑠 𝐻𝑐 𝑡𝑖𝑛𝑠 ] × 𝑔

𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑖𝑛𝑠 = 54.
𝜋(𝐷𝑐 )(𝑡𝑠 − 𝑐 )

𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑖𝑛𝑠 = 1.95 × 106 𝑃𝑎 55.

The next compressive stress factor that contributes to the columns weight is the weight of the liquid
on each plate as well as in the liquid downcomer. This is calculated to be:

27
 𝜌𝐿 𝐴𝑝 (ℎ𝑤 + ℎ𝑜𝑤 )𝑔𝑁
𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑙𝑖𝑞 = 56.
𝜋𝐷𝑐 (𝑡𝑠 − 𝑐 )

𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑙𝑖𝑞 = 1.6 × 106 𝑃𝑎 57.

The final compressive stress that needs to be accounted for is the stress of all the attachments
like the ladder, bolts, support rings, beams etc. The weight of the attachments is considered to be
equal to the full weight of the column

𝜋
[ (𝐷𝑜2 − 𝐷𝑐2 )𝜌𝑠 𝐻𝑐 ] 𝑔 58.
𝑓𝑑𝑒𝑎𝑑 𝑤𝑡, 𝑎𝑡𝑡𝑎𝑐ℎ𝑚𝑒𝑛𝑡𝑠 = 4
𝜋𝐷𝑐 (𝑡𝑠 − 𝑐)

𝑓𝑑𝑒𝑎𝑑 𝑤𝑡, 𝑎𝑡𝑡𝑎𝑐ℎ𝑚𝑒𝑛𝑡𝑠 = 3.9 × 106 𝑃𝑎 59.

This bring the total compressive stresses on the column to :

𝑓𝑡𝑜𝑡𝑎𝑙 = 𝑓𝑑𝑒𝑎𝑑 𝑤𝑡, 𝑎𝑡𝑡𝑎𝑐ℎ𝑚𝑒𝑛𝑡𝑠 + 𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑙𝑖𝑞 + 𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑖𝑛𝑠

60.
+ 𝑓𝑑𝑒𝑎𝑑 𝑤𝑡,𝑠ℎ𝑒𝑙𝑙

The total compressive force of the column is 15.25 MPa which is well below the materials
allowable stress of 108 MPa.

Column Support Structure

It is an industry standard to use a skirt support system for the support of a distillation column. This
support needs to account for the total weight of the vessel as well as factors like wind loading
which could also cause the column to fail. A skirt support allows for the stress in the support to be
distributed around the skirt so there’s no concentrated stress on it. An example of a support skirt
can be seen in Figure 16.

28
 Figure 16: Straight support skirt (Sinnott & Towler, 1999).

According to Sinnott & Towler (1999), for a column of a diameter of 2.8 m the following
dimensions for the skirt support apply:

Figure 17: Support skirt dimensions (Sinnott & Towler, 1999).

The letters correspond to the dimensions given in Table 13.

29
 Table 13: Dimensions for support skirt.

Vessel Diameter 2.8

Maximum
(m) Weight 1350
V (m)
(kN) 1.58
Y (m) 0.25
C (m) 2.50
E (m) 1.10
J (m) 0.625
G (m) 0.150
T (mm) 16
T (mm) 12
Bolt diameter (mm) 27
Bolt holes (mm) 33

Feed point

The subcooled feed will be fed into the downcomer of below plate . According to Sinnott & Towler
(1999), it is important to feed a liquid feed into the downcomer else the plate will be excessively
flooded. The feed into the column is displayed in the column diagram and the feed is fed on the
35 plate, counting from the top. The feed comes in at 20 °C and 1 atm meaning it will not have a
very high velocity. As a result of the normal feed velocity, the feed does not need to have an elbow
or a plate to slow down the liquid.

Packed column

A packed column is not feasible for this design as the column diameter is 2.8 m which is not
considered a “small diameter”. This is one of the main criteria to consider according to Wankat
(2007).

30
8. Heat Exchangers

Boiler

From the Aspen optimisation the boiler duty was determined to be :

𝑄𝐵 = 4477.96 𝑘𝑊 61.

The assumption is made that the boiler will only be 85% efficient. This has the implication that
more heat will be required than specified by Aspen. The heat required will be 5300 kW. It is
important to select the right type of reboiler for the distillation column as to minimize the costs
and maximize the efficiency. Table 12 below summarises the boilers that were considered during
this design.

Table 14: Reboiler types.

Kettle Thermosyphon Forced Circulation

Price Expensive Most economical -
Pressure Operates at a wide High vacuum Low vacuum
variety of pressures operations operations
State of fluid Suitable for most Not suitable for Highly viscous fluid
liquids. highly viscous liquids
Disengagement Does not require Requires a Requires a pump as
vessel disengagement disengagement well as
vessel. vessel. disengagement
vessel.

In this design a submerged-bundle reboiler (kettle type) will be selected to reboil the liquid that is
fed back into the column. This is done because the reboiler can be accommodated underneath the
vessel itself which saves money because a shell does not have to be manufactured and it is suitable
for most liquids meaning it will be suitable for the oils we are distilling. This is an ideal boiler for
the distillation column as it can vaporize up to about 80% of the feed according to Sinnott &
Towler (1999). The area of the heat exchanger is calculated with (Hall, 2018):

31
 𝑄
𝐴𝐵 = 62.
𝑈Δ𝑇𝑚𝑒𝑎𝑛

𝑊
The overall heat transfer coefficient is taken as 80 . To calculate AB the mean temperature
𝑚 2 °𝐶

first must be determined with:

(𝑇𝑖𝑛 − 𝑡𝑖𝑛 ) − (𝑇𝑜𝑢𝑡 − 𝑡𝑜𝑢𝑡 )

Δ𝑇𝑚𝑒𝑎𝑛(𝑐𝑜𝑐𝑢𝑟𝑟𝑒𝑛𝑡) = 63.
𝑇 −𝑡
ln (𝑇 𝑖𝑛 − 𝑡𝑖𝑛 )
𝑜𝑢𝑡 𝑜𝑢𝑡

The temperatures are selected as follows:

Figure 18: Heat exchanger initial cocurrent temperatures.

This yields the mean temperature to be equal to 61.67 °C. In turn the reboiler heat exchanger area
is calculated to be in the range of 1000 𝑚2 .

Figure 19: Kettle boiler internal design (Subramanian, 2011).

32
Condenser

The assumption is made that the condenser, similarly to the boiler, will be 100 % efficient. The
heat that will need to be liberated will be 4200 kW. It is important to select the right type of
condenser so that the right heat exchange can take place. For this design, a total condenser was
selected. This total condenser will use water to facilitate heat transfer.

𝑄𝐶 = −4200 𝑘𝑊 64.

33
9. References

Bernardo-Gil, MG, and Albertina Ribeiro, M (1993), "Vapor-liquid equilibria of binary mixtures
of β-pinene with limonene and P-cymene at atmospheric pressure" Fluid Phase Equilibria, 85,
153-160.

Brownell, LE, and Young, EH (1959), Process Equipment Design. John Wiley & Sons.

Hall, SM (2018), Rules of Thumb for Chemical Engineers (6th Edition). Elsevier.

Levy, J (2017), "Powerful Pine Oil — Cleanse House, Skin & Liver", https://draxe.com/pine-oil/
[21 April].

Linninger, A (sa), "Senior Design CHE 396".

Martins, MAR, Carvalho, PJ, Palma, AM, Domańska, U, Coutinho, JAP, and Pinho, SP (2017),
"Selecting Critical Properties of Terpenes and Terpenoids through Group-Contribution Methods
and Equations of State" Industrial & Engineering Chemistry Research, 56(35), 9895-9905.

Sinnott, RK, and Towler, G (1999), Chemical Engineering Design - SI Edition (5th Edition).
Elsevier.

Subramanian, M (2011), "Reboiler Design", http://www.msubbu.in/ln/design/II/DesignII-

Lecture-02-ReboilerDesign.pdf [26 April].

Wankat, PC (2007), Separation process engineering, 2nd ed., Prentice Hall, Upper Saddle River,
NJ.

WebMD (sa), "LIMONENE", https://www.webmd.com/vitamins/ai/ingredientmono-

1105/limonene [21 April].

34