.5
... r:

435%»!

in. L
”rm;
hawk—C

 

 

 

 

 

. _. ., . J I133... 5.1. 115:? ‘4 «1.51.31. RT:

 

V ,I::..rJn.Aw\..¢..nx~. «0.35.5513!» :6 .. 2...: fi$;.w§...§u.ln:§wfi 3.: , «an. .9.
I ‘

 

2 LIBRARY
(3 “-1. Michigan State
5.: ,'- ’0? {3’7 University

This is to certify that the
dissertation entitled

MEASUREMENT AND PREDICTION OF THE COMPOSITION OF
FRUIT DISTILLATES

presented by

Michael Joseph Claus

has been accepted towards fulfillment
of the requirements for the

PhD degree in Biosystems Engineering

 

 

%W QLLLWWQ

Major Professor’s Signature

1.! JULY 2003

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

| DATE DUE

DATE DUE

DATE DUE

 

JIJLW m a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIFlC/DateDue.p65-p.15

MEASUREMENT AND PREDICTION OF THE COMPOSITION OF FRUIT
DISTILLATES

By

Michael Joseph Claus

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Agricultural Engineering

2003

ABSTRACT

MEASUREMENT AND PREDICTION OF THE COMPOSITION OF FRUIT
DISTILLATES

By

Michael Joseph Claus

The production of high quality fruit spirits has always been and continues to be
heavily dependent upon the sensory evaluation of the distillate by the distiller. Sensory
fatigue can be a problem for distillers during production. A simple method for prediction
of the distillation characteristics of the important flavor compounds present in these
spirits can increase the yield and quality of these spirits and reduce the dependence of the
distillation process on the distiller's senses. This distillation process is difficult to model
because of the number of components present in the spirits as well as the constantly
changing thermodynamic interactions present on each tray due to its batch nature. The
Chemstations CI-IEMCADTM modeling software program with a batch distillation model
was utilized to predict the concentrations of the important flavor components present in
distilled fruit spirits as a function of distillate volume. This modeling process utilized the
analysis of the composition of the fermentation mash as a starting point. This approach
permits a distiller to predict which volume fractions of the distillate should be retained as
product, increasing yield and quality. Additionally this approach can be used to analyze
the effects of flaws in the fermentation process that might lead to distilled products not fit

for consumption.

ACKNOWLEDGMENTS

I would first like to thank Dr. Kris Berglund, who has been my advisor for this
project. His guidance and assistance throughout this project was essential to completing
this project. I would also like to thank the members of my committee, Dr. Steffe, Dr,
Ofoli, and Dr. Guyer, for their support, and timely advice. The support of the Biosystems
Engineering program also has been influential in allowing for the completion of this
project. I also wish to acknowledge Dr. Oriel and Dr. Bakker-Arkema for their
continuing support.

I would like to thank all the members of my research group past and present, with
a special thanks to Deirdre Lindemann who assisted greatly in completing this project.
Many of the distillations were performed by Deirdre and I greatly appreciate all that she
has done. I wish her luck in all her future endeavors. I also Wish to acknowledge the
efforts of those who worked on the distillation project in one form or another; Klaus,
Anna, Eric, Dave, Heather, Johnny, Matt, and Deirdre.

The support of my family and friends has been influential in finishing this project.
I am blessed with a long list of friends who I can mention for their support and kind
words, however, I especially would like to thank Pat, and Mike. I feel our time at MSU
has allowed us to become very close friends and I hope our friendships continue forever.
I also would like to thank Gary, Jenny, Holly, Rob, Janet, John and the rest of my friends
who fall into the realm of family.

Without the support of Joann I would not be completing this degree at this time.

Your love and support have been influential in allowing me to grow as a scientist and an

iii

person. I am very lucky to have had you become a part of my life. Thank you for all
your love, support, and time.

Two friends and mentors encouraged me from the moment I met them,
unfortunately they are not still with us. Mark Krueger and Joel Hansen were good friends
who taught me to set goals that may be out of reach, and the determination to reach those
goals. I owe a debt of gratitude to the two of them and I know they are looking down on
me thrilled at this accomplishment.

I would also like to recognize the wineries and distilleries for whom this project is
designed. My time on this project has seen the industry grow which has given me
motivation to continue working. The people in this industry have been very supportive of
this project. Finally I would like to thank you for reading this, I hope you find it useful

and enjoyable reading.

iv

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

KEY TO SYMBOLS

INTRODUCTION

1.1 Michigan Brandy Industry

1.2 Fruit Brandy Production

1.3 Congener Formation and the Need for Control
1.4 Chemstations CHEMCADTM Program

1.5 Objective

1.6 Literature Cited

DISTILLATION PRINCIPLES

2.1

2.2

2.3

2.4

2.5

2.6

2.7

Degrees of Freedom

Batch Distillation of Fruit Spirits
Rayleigh Distillation

Multistage Batch Distillation
Multicomponent Batch Distillation
Trays and Usage

Literature Cited

BRANDY FLAVOR COMPONENTS

3.1

3.2

3.3

Overview

Ethanol

Methanol

ix

xi

XV

13

16

20

22

25

26

26

27

27

3.4 Fusel Alcohols
3.5 Carbonyl Compounds
3.5.1 Aldehydes
3.5.2 Ketones
3.5.3 Esters
3.6 Literature Cited
MATERIALS AND METHODS
4.1 Overview
4.2 Fermentation
4.3 Standard Solutions
4.4 Rayleigh Distillation
4.5 Computer Modeling
4.5.1 Using the CHEMCAD 7’” Computer Program
4.5.2 CHEMCAD TMas a Predictive Model
4.5.3 CHEMCAD 7” Calibration
4.6 Experimental Multistage Batch Distillation
4.7 Comparison
4.7.1 Predictive Comparison of CHEM CAD 77"
4. 7.2 Synthetic and Fruit Batch Distillations
4.8 Gas Chromatography Methods
4.9 Retention Time and Reproducibility Validation
4.10 Literature Cited

vi

32

32

32

33

33

35

37

37

39

4O

43

45

45

49

51

52

57

57

57

57

61

63

RESULTS

5.1

5.2

5.3

5.4

5.5

5.6

5.7

Ethanol Modeling

Simple Rayleigh Distillations

5.2.1 Standard Batches

5.2.2 Fruit Mash Batches

Batch Distillation with Reflux

5.3.1 Standard Mash Multistage Batch Distillation with
Reflux

5.3.2 Fruit Mash Multistage Batch Distillation with
Reflux

CHEMCADTM Predictive Modeling

Compound Comparison

5.5.1 Ethanol

5.5.2 Methanol

5.5.3 F usel Alcohols

5.5.4 Aldehydes

5.5.5 Other Carbonyl Compounds

Summary

Literature Cited

SUMMARY AND CONCLUSIONS

6.1

6.2

6.3

Process
Fruit Distillation Modeling

Improving Cut Determination for Methanol

vii

64

73

73

77

78

78

79

79

80

80

87

93

102

109

112

114

115

115

115

116

6.4 Still Size Comparison
6.5 Still Characteristics
6.6 Conclusions
FUTURE WORK

APPENDICIES

viii

117

117

118

120

121

Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table 4.5

Table 4.6

Table 5.1

Table 5.2

Table 5.3

Table 5.4

Table 5.5

LIST OF TABLES

Standard batch composition for the 10L distillations.

Standard batch composition for the 150 L distillations.

Manufacturer and Lot Numbers for compounds used
in synthetic batches prepared.

Units used in CHEMCADTM program.

Compound name, formula and internal
CHEMCADTM number for identification.

Mean retention time, and error associated with the
compounds analyzed in these experiments.

The ethanol concentration, time, and calculated
distillate flow rate for the distillation of an 8%
ethanol/water mixture in the 150 L still.

The flow rates, analysis times, and reflux ratios
entered into CHEMCAD“.

The distillate flow rate and reflux ratios entered into
the CHEMCADTM program for each size of still.
The composition of the standard batches made for
these studies.

The composition of the first standard batch made
and the average composition extrapolated from the

simple distillation procedure.

ix

41

42

42

46

50

62

67

68

69

74

75

Table 5.6

Table 5.7

Table 5.8

Table 5.9

Table 5.10

The average composition of the congener
components of the second standard batch used for
the 10 L scale distillations.

Mean congener concentration and standard deviation
from analysis of the Rayleigh distillate from apple
fermentation.

Average ethanol concentration of a blend of all cuts
taken during distillation.

Methanol concentration of blended cuts for each
distillation type and the predicted values from
CHEMCADTM.

The best fit thermodynamic models within
CHEMCADTM for each of the compounds studied in

these experiments.

76

78

82

88

113

Figure 1.1
Figure 1.2
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4

Figure 2.5

Figure 2.6
Figure 2.7

Figure 3.1

Figure 3.2

Figure 3.3

Figure 4.1

Figure 4.2
Figure 4.3
Figure 4.4

Figure 4.5

LIST OF FIGURES

The process involved in making distilled fruit spirits.
Volatiles in distilled spirits.

Alambic style still.

Multistage batch still.

Simple batch distillation schematic.

Schematic of a multistage batch distillation apparatus
McCabe-Theile diagram for multistage batch distillation
with variable reflux

Christian Carl style sieve trays.

Bubble cap tray.

The Embden-Meyerhoff-Pamas pathway for the
fermentation of glucose by yeast.

The structure of pectin.

The reaction of the pectinesterase enzyme with
pectin.

Procedure of comparing CHEMCADTM predicted
concentrations against the experimental concentrations
during distillation.

Bench top distillation apparatus.

Schematic of 10L Holstein Still.

Schematic diagram of 150L Christian-Carl Still.

Sample chromatograph.

xi

11

12

15

18

19

23

24

28

3O

31

38

55

56

59

Figure 4.6

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Sample chromatograph.
Concentration of ethanol (% v/v) against cumulative

distillate volume (mL) for actual distillation and two

simulated constant reflux distillations in CHEMCADTM.

Experimental and predicted values of the ethanol
concentration at volume intervals.

The procedure used to determine the reflux ratio of
each step of the distillation.

Change in the reflux ratio over distillate volume

The ethanol concentration of a 10 L fruit distillation
compared to the values predicted by the CHEMCADTM
computer model using three thermodynamic models.
The ethanol concentration of a 10 L standard batch
distillation compared to the values predicted by the
CHEMCADTM computer model using three
thermodynamic models.

The ethanol concentration of a 150 L fruit distillation
compared to the values predicted by the CHEMCADTM
computer model using three thermodynamic models.
The ethanol concentration of a 150 L standard batch
distillation compared to the values predicted by the
CHEMCADTM computer model using three

thermodynamic models.

xii

6O

65

70

71

72

83

84

85

86

Figure 5.9

Figure 5.10

Figure 5.11

Figure 5.11

Figure 5.13

Figure 5.14

Figure 5.15

Figure 5.16

Figure 5.17

Methanol concentration of a 10 L fruit distillation
normalized to a 40% ethanol solution.

Methanol concentration of a 10 L standard batch
distillation normalized to a 40% ethanol solution.
Methanol concentration of a 150 L fruit distillation
normalized to a 40% ethanol solution.

Methanol concentration of a 150 L standard batch
distillation normalized to a 40% ethanol solution.
Standard 10 L batch distillation compared with predicted

values from CHEMCADTM for t-butanol.

Comparison of predicted values from CHEMCADTM
compared with the t-butanol concentration in fruit
spirit distillate in 150 L distillation.

The concentration of l-propanol present in 10 L fruit
distillation compared to predicted values.

The predicted values of l-propanol concentration
and the experimental distillate concentration of
l-propanol in 150 L batch distillations.

The 10 L distillate concentration of isoamyl alcohol
and the values predicted by CHEMCADTM using the

UNIFAC thermodynamic models.

xiii

89

9O

91

92

95

96

97

98

99

Figure 5.18

Figure 5.19

Figure 5.20

Figure 5.21

Figure 5.22

Figure 5.23

Figure 5.24

Figure 5.25

Figure 5.26

The concentration of isoamyl alcohol present in
distilled spirits compared with the predicted
CHEMCADTM values.

The additive values of 1-propanol and isoamyl alcohol
which represent the bulk of the fuse] alcohol
concentration present in distilled spirits.
Acetaldehyde concentration of 10L fruit batch
distillation.

Acetaldehyde concentration of 10L standard batch
distillation.

The CHEMCADTM predicted values are greater
than the experimental concentration of the acetaldehyde
in the second, third and fourth 750 mL fraction of
the 150L distillation of a synthetic fruit mash.
Benzaldehyde in 10 L standard batch distillation.
The benzaldehyde concentration maxima is
predicted by the CHEMCADTM thermodynamic
models at earlier distillate volumes than the actual
distillation data shows.

Predicted and experimental concentration of
acetone in a 10 L standard batch distillation.
Predicted and experimental concentration of

acetone in a 150 L standard batch distillation.

xiv

100

101

104

105

106

107

108

110

111

xDav
XF

xj+l
xw

Vj+l
y}
Yj+l

KEY TO SYMBOLS

Number of components (Gibb’s Equation)

Total distillate

Total feed

Number of degrees of freedom (Gibb‘s Equation)

Vapor enthalpy

Enthalpy of liquid in the distillate

Enthalpy of the liquid stream leaving plate j

Liquid stream leaving plate j

Liquid stream entering plate j

Number of Phases (Gibb's Equation)

Total heat supply to bottoms (reboiler)

Heat load on condenser

Reflux Ratio

Total batch time

User defined time

Distillate composition of component 1 in mixture

Average distillate composition of component 1 in mixture
Feed composition of component 1 in mixture

Liquid composition of component 1 in mixture leaving plate j
Liquid composition of component 1 in mixture entering plate j
Liquid composition of component 1 in bottoms

Vapor molal flow rate

Vapor stream leaving plate j

Vapor stream entering plate j

Vapor composition of component 1 in mixture leaving plate j
Vapor composition of component 1 in mixture entering plate j

XV

1. INTRODUCTION

1.1 Michigan Brandy Industry

The growing distillation industry in Michigan is the result of recent changes to the
state laws regarding the licensing of fruit distilleries. The number of stills has increased
from zero in 1996 to nine in 2003. Michigan is ideal for the production of fruit spirits
due to the variety of fruit that grow in the state. Michigan's brandy industry primarily
follows the traditional German process for fruit spirits]. This batch distillation process
with reflux utilizes a multistage still to preserve the flavors and essences in the distillate.
Research at Michigan State University has been focused on enhancing the efficiency of
the traditional European style of fruit spirits production, and improving the quality of the

distilled spirits.

1.2 Fruit Brandy Production

Distillation is one of the oldest separation processes and is the most widely used
unit operationz. There are two competing styles for producing fruit spirits in Europe.
Alambic distillation is used for producing spirits similar to cognac, and batch distillation
produces eau-de-vie or schnapps. The alambic style involves distillation through a
simple pot still, collecting the distillate and repeating the distillation multiple times to
achieve high proofs3. The German style of batch distillation involves a single distillation
utilizing a column with reflux on the batch still to obtain high proof spirits. These spirits

are traditionally stored in glass and served as water clear brandies‘.

Fruit brandies are produced by fermentation and distillation of the whole fruit to
enhance the flavor and quality of the spirits. The fruit utilized for the production of fruit
spirits should be unblemished; however, overiipe fruit is preferred for fruit spirits as there
is more sugar to ferment. The fruit is mashed and yeast is added for fermentation. Upon
completion of the fermentation the fruit mash is distilled resulting in high proof water
clear spirits. Finally after storage at high proof (aging) for one month to several years,
the distillate is diluted to drinking strength (>40% Alcohol By Volume (A.B.V)) and
bottled4. Figure 1.1 shows the process involved in the production of fruit spirits, from the

fresh fruit to the bottled product.

1.3 Congener Formation and the Need for Control

The number and concentrations of compounds present in the fruit spirits are much
greater than other types of distilled spirits due to the use of whole fruit mashes. Figure
1.2 compares the congeners present in distilled fruit spirits with other types of spirits such
as gin, vodka, and whiskeys. The skins and flesh of the fruit contribute to the increased
number of flavor compounds present in these spirits. The distillation process also
attempts to maximize the flavor compounds for fruit spirits, where vodka and gin types of

spirits attempt to reduce the concentration of the flavor constituentss.

A~M~

Fresh Fruit Crushed or
Mashed Fruit

 

Fermentation
1-2 Weeks

 

 

 

Distillation

 
  
   
 

Storage
1 Month to

Dilute Several Years

with water
and bottle

Consumption

Figure 1.1 The process involved in making distilled fruit spirits. Fresh fruit is
mashed, fermented and then distilled. The distiflate collected is aged
and then diluted to drinking strength and bottled.

3

 

2200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

iTwBinESu 7 7 T 7
2000 . '
CI Higher alcohols I
'8 1800 i ICarbonyl corrpoundsI
3:: E 1600 EJEthyl acetate I
"3 8 ”misses- L}
g :5 1400 ~
"’ o
r:
a .g 1200 I
o
g 8 1000 ~
E”:
8 a 800i
0 3
r:: :5 600 - .
E 5; a
o - \\ I
> 400 § .. a .
§ . . N
200 * I g I .5,
(900066} ”OQQ‘)&">\° “gs “690%”:er {g 0'6 as gigs 0‘0 08gb
a 00 \S (1" :6‘
9° s3‘ 0°18, 69¢ \ e
«0 o 0 c
Q as?
9

Figure 1.2 Volatiles in distilled spirits. The concentration of volatile congeners in

the distilled spirits (mg/100 mL of alcohol). Fruit spirits have higher

concentrations of volatiles than other distilled beveragess.

After water and ethanol, methanol is present in the next highest concentration of
the flavor components found in these distilled spirits. Methanol is regulated by the
Alcohol and Tobacco Tax and Trade Bureau (TTB) and the Food and Drug
Administration (FDA), to prevent distillers from producing a product which can be a
health hazard6. Methanol is present in these spirits because naturally occurring enzymes
present in fruit induce the breakdown of pectin, which in turn releases methanol into the
mash. This increased methanol in the mash is directly related to the concentration of
methanol present in the final distilled product.

Fusel alcohols such as l-propanol, t—butanol, and isoamyl alcohol, and carbonyl
compounds such as acetaldehyde, acetone, and benzaldehyde are present in the mash of
fruit fermentations at higher concentrations than in wines or grain mashes7’ 8' 9. The
concentration of the fusel alcohols in the final distilled spirits must be carefully
controlled by the distiller because of the negative aromatic characteristic associated with
the higher concentration of these higher chain alcohols. Acetone and acetaldehyde also
have negative aroma characteristics.

Fruit fermentations vary widely in chemical makeup. Each fruit and variety in
each fruit will have unique characteristics. Furthermore, within each variety of fruit
exists variations caused by growing conditions (e. g. soil), seasonal variation, and regional
climate. A batch of cherries harvested on one day can have a very different makeup than
a batch of cherries harvested a week later. Additionally, the yeast involved in the
fermentation process can affect the flavor profile of the fruit mash. In general, the effects

of wild strains of yeasts and molds are mitigated by over-inoculation of the mash with the

desired yeast; however, it is possible for some wild microbes to be present in the mash
and alter the composition of the congener compounds.

Variation in fruit mash composition directly leads to changes in distillate
composition. Predictive control of the distillation based upon the composition of the fruit

mash will increase the overall quality of the spirits.

1.4 Chemstations CHEMCADTM Program

Chemstations Inc. CHEMCADTM process simulation software possesses a batch
distillation module capable of modeling the distillation process involved in the
manufacture of fruit spirits. The CHEMCADTM program can provide simulation of a
number of industrial functions/unit operations including: batch distillation (with reflux as
required by the present work), reactions, extraction, continuous distillation, electrolytic
processes, vapor/liquid/liquid equilibrium calculations, equipment sizing, environmental
calculations, cost estimates, heat exchanger networks, and safety analyseslo‘ ”.

The CHEMCAD” program is designed to be user friendly, which is one of the
primary reasons for its choice for the current work. Use of the program involves entering
the flow rates involved in the distillation, analysis step size (time), and composition of the
pot charge. The CHEMCADTM program is designed to produce data that can be easily
interfaced with other computer programs such as Microsoft Office programs”). The

ability to transfer the data to a spreadsheet program makes analysis of the data easier than

other programs which do not interface well with other programs”).

1.5 Objective

This project involves predicting the concentration of congener compounds
common to fruit spirits by utilizing simple distillation methods of the mash, and the
CHEMCADTM computer program. If the quality of the resulting fruit spirits could be
predicted based on the makeup of the mash content, the distillation process could be
controlled to increase the quality and yield of the final product. Use of CHEMCADTM as
a predictive model with actual fruit spirit distillations is aimed at determination of the

usefulness of this approach.

1.6

10.

11.

Literature Cited

Tanner, H., and Brunner, H. R., Fruit Distillation Today 3“ ed., (English
translation), Heller Chemical and Administration Society, Germany, 1982.

Diwekar, U., Batch Distillation. Taylor and Francis, Washington, DC. 1995.
Leaute, R., "Distillation in Alambic." Am. J. Errol. Vitic., 41, (1) 1990.

Claus, Michael J ., "An Investigation of Relationship Between Tray Usage of the
Still and Congener Compound Concentration in Distilled Fruit Spirits." MS.

Thesis. Michigan State University. East Lansing, MI. 2001.

Piggott, J. R., and Patterson, A. ed., Distilled Beverage Flavour: Recent
Developments. Ellis Horwood Ltd., 1989.

Bureau of Alcohol, Tobacco, and Firearms. Alcohol and Tobacco Newsletter (1)
1998. Available at web site <http://www.ttb.gov/pub/alctob_pub/cm98-1.pdf>.

Guymon, J.F., "Higher Alcohols in Beverage Brandy: Feasibility of Control of
Levels." Wines and Vines, 53, (1) 1972.

Stinson, ENE, et al., "Composition of Montmorency Cherry Essence. 1. Low-
boiling Components." Journal of Food Science, 34, 1969.

Stinson, E.E., et al., "Composition of Montmorency Cherry Essence. 1. High-
boiling Components." Journal of Food Science, 34, 1969.

Chemstations CHEMCADTM Process Simulation Software. Informational
Material. Available at website <www.chemstations.net> 2002.

Chemstations CHEMCADTM Process Simulation Software. Internal Help
Function and Tutorial. 2002.

2. DISTILLATION PRINCIPLES

2.1 Degrees of Freedom

Gibb's phase rule states that the number of degrees of freedom associated with a

mixture is based on the equation below,
F=C—P+2

where F is the number of degrees of freedom, C is the number of components, and P is
the number of phases]. A two component mixture would then have F = 4 - P degrees of
freedom. If temperature is kept constant the remaining degrees of freedom F ' = 3 - P,
which has a maximum value of two. These two degrees of freedom are pressure and
composition of one component (mole fraction XA). When increasing to a three
component mixture there are F = 5 - P degrees of freedom. As the number of
components present in the mixture increases, the degrees of freedom increase, which in

turn makes modeling the process more difficultl' 2.

2.2 Batch Distillation of Fruit Spirits

There are two competing styles for producing fruit spirits in Europe. Simple pot
style, alambic, distillation is used for producing spirits similar to cognac and column still
distillation is used to produce eau-de-vie (French) or schnapps (German). The alambic
style involves distillation through a simple pot still, collecting the distillate, and repeating
the procedure multiple times to obtain high proof distillate products. This style utilizes

high temperatures at the heat transfer surface and the resulting spirits are often harsh. To

overcome the dour flavor of these spirits, they are often stored in oak casks to extract the
flavors from the wood. Figure 2.1 is an illustration of an alambic style still.

The German style of batch distillation involves a single pass through a reflux
column on a batch still to obtain high proof spirits. This approach utilizes lower
temperatures and the resulting spirits are traditionally stored in glass and served as water
clear brandiesB. Figure 2.2 is an illustration of the multistage batch still used for this type
of distillation.

Fruit brandy distillation is different from traditional brandies which are distilled
from grape wines. Fruit spirits are produced by the crushing, fermentation, and
distillation of the mash of the whole fruit. Using the whole fruit in the process increases
the essences and improves the sensory experience of consuming distilled spirits3‘ 4. The
fruit flesh, skins and seeds presence in the fermentation mash and distillation process
increase the concentrations of con gener compounds (compounds other than ethanol and
water) present in the fruit spirits. Fruit spirits have higher concentrations of nearly all
types of congener compounds when compared to other types of distilled spirits4.

Distillation is the process of concentrating or separating compounds based on
their volatility. Traditionally three fractions of fruit spirits are collected. The first
fraction, the heads, will have higher concentrations of the lower boiling point
compounds. The middle fraction, the hearts, is the fraction which will eventually become
the potable product. The final fraction, the tails, will have a higher concentration of the

higher boiling point compounds.

10

 

Alambic Still

One stage distillation

E 0Multiple distillations
required to increase
separation

OHigh amount of product
loss

 

 

 

A. Fire or Steam Heat
Supply

 

 

Figure 2.1

B. Copper Pot
C. Alambic I-Iat
D. Swans Neck
E. Condenser

F. Distillate Spout

Alambic style still. This still utilizes only one stage requiring the
distillate to be redistilled to obtain a good product. The alambic still
is recognized from the alambic (onion shaped) hat and swans neck

transfer tube.

11

 

 

 

C fl Batch Still

Multiple stage distillation

Each stage acts as one
alambic distillation

N
is,
\\\k\

 

 

 

 

 

 

 

 

 

1 TI" °Higher separation with
\h ..... :5 multiple trays
°Higher purity of product
D °Greater yield
B
A ——“
E
Tray Operation

A. Steam Heat Supply

B. Copper Pot /

C. Distillation Column 31:1

1. Tray
2. Downcomer

 

 

 

Liquid

 

D. Condenser

E. Distillate Spout

Figure 2.2 Multistage batch still. This apparatus used multiple stages to
increase the separation of the components. The trays force the vapor

phase to pass through the condensed liquid allowing reflux and better

separation of the components.

12

2.3 Rayleigh Distillation

The most basic type of distillation is a simple binary distillation also known as a
Rayleigh distillation. Two compounds with different boiling points can be distilled under
simple distillation conditions. A simple distillation involves no trays, the pot for boiling
the mixture is connected to the condenser. When the liquid in the pot is boiled, the vapor
is removed in each time interval and condensed in the condenser. The vapor becomes
richer in the more volatile component than the liquid remaining in the pot, thus reducing
the concentration of the more volatile component present in the pot. The vapor
condensed in the condenser increases in the concentration of the more volatile
component. As time increases the two components are separated into two separate
vessels, the more volatile in the distillate and the less volatile in the potz’ 5. Figure 2.3
shows a schematic diagram of the Rayleigh distillation apparatusz.

The mass balance around the entire system for the entire operating time is:

F =Wfim, +D

total
FxF = xW,fina1meal + DtotalxD.avg

where F represents the feed, D represents the distillate W represents the bottoms, and Xf
represents the mole fraction of the feedz.

The variables F, XI: and the desired value of either xwfinal or 2(1),an are specified requireing
an additional equation to solve for the three unknown variables Dtoml, Wfinal, and the
unspecified variable above. The Rayleigh equation is derived from a differential mass
balance. The assumption is made that the holdup in the accumulator and column are
negligiblez. The differential amount of material -dW of concentration xD is removed

from the system, resulting in the differential mass balance:

13

— deW = —d(wa ) = —dew — doW
Rearranging and integrating gives:

Wn x ,m
fialdW Wfal de l
—: n =

_ 01'
W21, W xF xD xW F

 

 

Wfinal _ x] de

xo "xw

xW . final

The vapor product is in equilibrium in simple batch distillations. Because a total

condenser is used, the substitution of y = XI) can be madez. Then:

 

 

 

Wm, __ xF dx _ xF dx
In fit - I I f(X)-x

x w , final x w , final
where x and y are in equilibrium which can be expressed as y = f(x,p)
By integrating the above equations it is now possible to find a solution for the
unknown variables Dtotala Wfinal, and ij—maj or xuavg. Time is implicitly present in these

. . 2
equations because W, xw, and X]; are time dependant .

l4

 

 

 

W, XW
W Bottoms
D Distillate
XD Mole Fraction of Distillate
XW Mole Fraction of Bottoms
QR Reboiler Heat Load

Figure 2.3 Simple batch distillation schematicz.

2.4 Multistage Batch Distillation

For multistage systems xp and xw are no longer in equilibrium, and the Rayleigh
equation can not be integrated until a relationship between xD and xw are foundz. Stage
by stage calculations must be made to obtain the relationship between X]; and xw. By
assuming that the holdup is negligible at each tray, the condenser, and the accumulator,
mass and energy balances around any stage j and the top of the column can be performed

as shown in figure 2.4. At any time 1 these balances become:

WH=L +D

V'+lyj+l : ijj + on

1

Q. +Vj+lHjH = Ljhj + DhD

j+l

The molal flow rates are expressed in these equations by V, L, and D. The energy
balance is not needed if constant molal overflow is assumed, because the vapor and liquid

flow rates will be constant. The equation for constant molal overflow then becomeszz

_L L
yj+l-ij+1_-‘7 3‘0

This represents a straight line on a y-x diagram. The slope will be UV and the intercept
with the y = x line will be xD. Either xD or UV will need to vary during the batch
distillation, and the operating line will be constantly changingz. The 150 L Christian Carl
still is based on a varying reflux ratio, attempting to keep the concentration of XI) at a
maximum.

For variable reflux ratio operation of a batch distillation, the equation for constant

molal flow rate holds, with the slope varying, and the intersection with the y = x line at a

16

constant xD. Figure 2.5 shows the McCabe-Thiele diagram for multistage batch
distillation with variable reflux. The McCabe-Thiele diagram relates xw and x0 allowing
integration of the Rayleigh equationz.

The feed concentration x1: is found by identifying the initial value of UV is found
by trial and error, and specifying the number of stages and the distillate composition XI).
The final value Xmeal occurs when UV is in total reflux, or UV = (UV)max. Once xwfmal

is found, Wfinal is determined from integration by the equationzz

7 dx
Wfina, = F exp — I

xw , final

 

—x

17

 

 

 

 

_-q————_———_——_

 

 

 

 

 

 

 

 

X]
L]
tJIIv
_ P
I _4 n.
._ V
a
_
_
_

 

QR

 

Schematic of a multistage batch distillation apparatusz.

Figure 2.4

18

 

 

 

 

 

 

 

 

 

 

04 ~ " sz
0.2
0
0 02 0.4 06 08 1
X

Figure 2.5 McCabe-Theile diagram for multistage batch distillation with

variable refluxz.

l9

2.5 Multicomponent Batch Distillation

The additional degrees of freedom associated with the addition of more
compounds to the mash make the modeling of multicomponent batch distillations
difficult. By increasing the number of components from two to three the number of
degrees of freedom increases to account for the composition of the feed. Each additional
compound included in the feed will in turn increase the complexity of the interactions of
the components. Each interaction between each component must be taken into account
for modeling these types of distillations. The variety in compound interactions (e. g. polar
and non-polar, size, hydrogen bonding, etc.) require many different types of models to
explain the behavior of the components present in the multicomponent distillations.
These models rely upon large sets of coupled, nonlinear ordinary differential equations.
The models available in CHEMCADTM, and the formulas used for this work are included
in the appendix of this work.

In multicomponent distillation, neither the distillate composition nor the bottoms
composition is completely specified because there are not enough variables to allow
complete specificationz. The calculation procedure is greatly affected by the inability to
completely specify the distillate and bottoms composition. It is possible to identify
components in four classifications. Those components for which the fractional recoveries
in the bottoms or distillate are known as key components, the most volatile of which is
known as the light key (LK) and the least volatile of which is known as the heavy key
(HK). The other compounds are known as the non-key components. Those non-key

components that are more volatile than the LK component are known as light non-key

20

(LN K) and the compounds less volatile than the LK compound are the heavy non-key
compounds (HN K).
The overall balance equation is:
F=B+D
The component balance equations become:
F z, = Bx

+ Dx,

1.1m! r.disl

and the mole fractions must sum to 1.

C
in.dist :1'0
(1‘
wam = 1.0

For a three component mixture the component balance equation is written three times,
and must then sum to meet the overall balance equation.

The problem of solving for the external mass balances arise. The unknowns are
B, D, x14“... x3,at;a, x11...“ and xibw. This leaves six unknowns with five independent
equations. The additional equations of energy balances or equilibrium expressions
always add additional variablesz. Internal stage-by-stage calculations for tertiary systems
rely on the compositions of the components at one end of the column, and these as
mentioned earlier are unknown. By assuming one of them is known, the problem
becomes a trial-and—error exercisez.

Many formulae for these trial-and—error calculations have been developed for
mixtures of three or more components. Additional variability to the molal flow rate of
each component and chemical interactions further complicate the problem resulting in the

need for a software suite like CHEMCADTM to perform these calculations using a variety

21

of these distillation models". A description of the models used in this work can be found

in the appendix.

2.6 Trays and Usage

Distillation trays rely on the principle of reflux for increasing the separation of the
compounds present. Reflux is the partial condensation of vapor and the return of the
liquid down the column. At every interface between the liquid layer and the condensed
layer, contact is occurring causing greater separation of the compounds present1‘2‘3’7.

As the name implies, sieve trays have numerous small holes in the plate of the
tray which, for the Christian Carl still used in this experiment, are capped by a plate as
illustrated in Figure 2.6. These holes are small enough that the pressure of the vapor
from the tray below causes only the upward flow of vapor and liquid does not flow
downward. The liquid phase flows across the top of the tray until it reaches the
downcomer (the opening for the liquid to flow downward) and the vapor from the tray
below is forced to pass through the holes and the liquid, and then onto the next tray. This
configuration leads to excellent contacting between the two phases and makes the
composition approach vapor-liquid equilibrium.

The 10 L Holstein still uses bubble cap trays. Bubble cap trays use the same idea
of passing the vapor from the tray below through the condensed layer on the tray above.
As seen in figure 2.7 a bubble cap tray has a cap over a tube from the lower tray. The

vapor must pass through the condensed liquid in order to get through to the next tray.

Both types of trays can be modeled by the CHEMCADTM software program.

22

Condensed
Liquid

    

o%%%%%%%°

 

 

 

 

 

 

 

Condensate Return

Top View Side View

Figure 2.6 Christian Carl style sieve trays. The sieve tray has a number of holes
along the tray through which the vapor from the tray below must
pass. This action forces the vapor to pass through the condensed
liquid on the tray, increasing the separation of the components

present.

23

Figure 2.7

Condensed
Liquid

 

 

 

 

 

Condensate Return

Bubble cap tray. Used by Holstein stills, this tray design traps the

condensed liquid on the surface of the tray without allowing it to flow
through to the tray below. The design also forces the vapor from the
tray below to pass through the condensed liquid layer, increasing the

separation of the components present.

24

2.7

Literature Cited

Atkins, P. Physical Chemistry 5’" ed., W. H. Freeman and Co. New York, 1994.
Wankat, P.C., Equilibrium Staged Separations. Prentice Hall, New Jersey, 1998.
Tanner, H., and Brunner, H. R., Fruit Distillation Today 3'd ed., (English

translation), Heller Chemical and Administration Society, Germany, 1982.

Piggott, J. R., and Patterson, A. ed., Distilled Beverage Flavour: Recent
Developments. Ellis Horwood Ltd., 1989.

Diwekar, U., Batch Distillation. Taylor and Francis, Washington, DC. 1995.

Chemstations CHEMCADTM Process Simulation Software. Internal Help
Function and Tutorial. 2002.

Claus, Michael J ., "An Investigation of Relationship Between Tray Usage of the

Still and Congener Compound Concentration in Distilled Fruit Spirits." MS.
Thesis. Michigan State University. East Lansing, MI. 2001.

25

3. BRANDY FLAVOR COMPONENTS

3.1 Overview

Organic acids, esters, and fusel alcohols form the main body of the congener
compounds (all compounds other than ethanol and water) in the distilled spirits]. Some
other compounds are also present in the distillate, but the above organic compounds
account for the majority of con geners present. The formation of most of the congeners in
the distillation occurs in the fermentation of the mash in the presence of yeast]. The
fermentation process is controlled to prevent an excess of undesired compounds, and
increase the yield of ethanol. Control of the fermentation includes temperature control,
fermentation duration, and mixing. High temperature fermentations have reduced ethanol
yield and increased congener concentrations. If the temperature of the fermentation is too
low, the yeast activity decreases, requiring a longer duration of fermentation. Common
fermentation conditions include a temperature range of 13°- 18°C and a two week
fermentation time. Stirring the fermentation can decrease the time required for complete
fermentation, and decrease the mash viscosityz‘ 3.

Any wild strains of yeast, molds, or other microbes may cause increased
concentrations of undesired compounds. Over-inoculation of the mash with the desired
strain of yeast is the best method for preventing other microbes from producing undesired
compounds. Over-inoculation requires adding an amount of yeast greater than the
minimum amount required to ferment the mash. The desired yeast will out-compete the
wild microbes for the nutrients needed for growth and reproduction. This method

reduces the production of undesired flavors at a cost of more yeast.

26

3.2 Ethanol

The distillation of fruit spirits relies on the conversion of fruit sugars to ethanol by
yeast. The Embden-Meyerhoff-Parnas Pathway (EMP) is the well know process for the
conversion of sugars to ethanol by yeast. This pathway proceeds by degrading the sugar
to acetaldehyde where it is then reduced to ethanol. Figure 3.1 shows the EMP pathway
common to yeast fermentation of ethanol". The yield of ethanol is dependant upon the
initial concentration of the total sugar present in the fruit which is measured as total
dissolved sugar present in the liquid mash. The total dissolved solids (brix), however,
also includes unfermentable compounds such as sorbitol, and must only be used as a
guide to determine the approximate concentration of the sugars present in the mash. The
EMP process yields two moles of ethanol for every one mole of glucose present in the

fruit. Other sugars present in the fruit include fructose, maltose and sucrosez.

3.3 Methanol

Methanol is a very important compound in the production of fruit brandies. The
United States Food and Drug Administration (FDA) and Alcohol and Tobacco Tax, and
Trade Bureau (TTB) regulate the methanol concentration in distilled spirits at 0.35% v/v
(2.765 g/L)5. Methanol is a positive flavor component of brandies; therefore, its
complete elimination from brandies is not the aim of the regulations. Methanol is similar
to ethanol in taste and smell; however, it is toxic and potentially dangerous if present in
high concentrations. These regulations are primarily a consequence of the use of
methanol by unlicensed distillers to adulterate beverages by addition of methanol to

increase the alcohol concentration.

27

2ADP
+2Pi 2ATP

 

Glucose 2 Pyruvate
SA ’ 2C0
2 N AD+ 2 NADH 2
2 Ethanol 2 Acetaldehyde

 

Figure 3.1 The Embden-Meyerhoff-Parnas pathway for the fermentation of
glucose by yeast. One mole of glucose produces 2 moles of C02 and 2
moles of ethanol. This pathway also produces energy for the yeast

cells‘.

28

The regulation of methanol is based on associated health hazards. Methanol is a
poison that interrupts nerve impulses. Methanol causes headache, nausea, and can attack
the optic nerve blurring vision or even causing blindness6’ 7. Chronic exposure to
methanol can cause kidney and liver dysfunction. Methanol is metabolized in the body to
formaldehyde, which is also poisonous to humans6. Interestingly, the antidote for acute
methanol poisoning is administration of ethanol; therefore, the low regulated levels of
methanol in high proof fruit brandies pose little or no health hazard.

Methanol is a side product of the fermentation process along the EMP process.
However, a larger concentration of the methanol comes from an enzymatic interaction of
pectinesterase with the pectin of the fruit3. The structure of pectin and the enzymatic
reaction of pectinesterase on pectin can be seen in figures 3.2 and 3.3 respectively3. This
is part of the natural decomposition process of the fruit, which is designed by nature to
prevent animals from removing the seed from the nutrients of the fruit because of the
poisonous methanol that is present. The whole fruit is used in the fermentation of fruit
spirits which increases the amount of pectin and pectinesterase in the mash and
consequently increases the concentration of methanol in fruit spirits when compared with

other distilled spiritsz’ 3.

29

\OCH3

C=O

£3 OCH3 \OCH3

'\ /I

Pectinesterase Activity Sites

Figure 3.2 The structure of pectin. The arrows point to sites where the
pectinesterase enzyme cleaves methanol from the pectin in the

reaction seen in figure 4.33.

30

Pectinesterase Mode of Action

0 0‘ Hz0
E | KI
nzyme C —- O — CH3 —> Enzyme — C — O — CH3
V l I \
R R
CH3OH
O O: O
l I t)

C — OH ‘— Enzyme— C — OH ‘— Enzyme— C

l l |

R R R x
OH'

+ Enzyme

Figure 3.3 The reaction of the pectinesterase enzyme with pectin. Only the active
site on pectin for the enzyme is shown. This reaction is the primary

source of methanol in distilled fruit spirits3.

31

3.4 Fusel Alcohols

Fusel alcohols are defined as those alcohols larger than ethanol (e. g. C>2) and
compose the largest group of aroma compounds in alcoholic beveragesl’ 8. The most
common fusel alcohols in distilled spirits include 1-propanol (n-propanol), 2-methyl-2-
propanol (isobutyl alcohol), and 3-methyl-l-butanol (isoamyl alcohol). Isoamyl alcohol
is the main fusel alcohol synthesized during fermentation by yeast accounting for 40 to
70% of the total fusel alcohol concentration in distilled spirits depending upon the type of
mash]. Formation of these fusel alcohols is thought to be independent of the raw
materials used in the mash in that the formation of these longer chain alcohols can occur
in whiskeys as easily as in tequila, gin, or fruit spiritsl’ 8. N-propanol, and branched C4
and C6 alcohols are formed from valine, leucine, and isoleucine in the presence of yeast.
OL-keto acids are thought to act as key intermediates in the formation of higher alcohols.
The a-keto acids are first decarboxylated to aldehydes and then reduced to the
corresponding alcohol. Fusel alcohols are thought to form in fermentation under both

anaerobic conditions from amino acids and aerobic conditions from sugarsg.

3.5 Carbonyl Compounds

3.5.1 Aldehydes

Aldehydes are the intermediates in the production of alcohols by yeast. The
aldehyde concentration in the distilled spirits is due to the inefficiency of the yeast in

reducing the aldehydes to their corresponding alcoholg. The yeasts are making the

32

aldehydes as well as reducing them to alcohols; however, the reduction of the aldehyde to
alcohol is not as efficient as the production of the aldehydesg.

The most common aldehyde present in the distilled fruit spirits is acetaldehyde.
Acetaldehyde is an intermediary in the EMP pathway and is present in all distilled spirits.
Acetaldehyde has a low boiling point and is soluble in both water and ethanol, and is at
its highest concentration in the early (heads) portion of the distillation”).

Benzaldehyde, sometimes referred to as bitter almond oil, is another important
aldehyde present in stone fruits. Benzaldehyde comes from the amygdalin present in the
pit of stone fruit. The hydrolysis of one mole of amygdalin yields two moles of glucose

one mole of cyanide and one mole of benzaldehydelo’ ”

. Benzaldehyde is considered a
positive aroma characteristic in stone fruit brandiesz’ 10‘ 12. It is present in the late hearts

and tails of the distillate due to the relatively high boiling point.

3.5.2 Ketones

Ketones are produced in the yeast cells as an oxidation product of alcohols and
excreted as an unwanted side product. The most common keytone present in the distilled
spirits is acetone, which has a negative aroma associated with it in fruit spiritsz’ 10.
Acetone may be produced from oxidation of 2-propanol as well as other microbes present
in the fermentation media. Clostridium acetobutylicum for example, is used in the

industrial fermentation of acetone, butanol, and ethanol”.

3.5.3 Esters

Esters are formed during the distillation and storage of the spirits and generally

add positive flavor aromatic qualities to the distilled fruit spirits. The highest

33

concentration esters present are ethyl formate and ethyl acetate. Ester formation is due to
the esterification of alcohols with organic acids. The formation of ethyl acetate and ethyl
formate involve the reaction of acetic acid with ethanol and methanol respectfully”.
These two esters are present in the highest concentration in the distilled spirits because

they are derived from the two highest concentration alcohols.

34

3.6

10.

ll.

12.

13.

Literature Cited

Suomalainen, H., and Lehtonen, M., "The Production of Aroma Compounds by
Yeast." J. Inst. Brew., 85, 1979.

Tanner, H., and Brunner, H.R., Fruit Distillation Today 3rd ed., (English
Translation) Heller Chemical and Administration Society, Germany, 1982.

Andraous, J ., "The Effect of Temperature and the Use of Liquefaction Enzymes
on the Congener Concentration in Distilled Fruit Brandies." MS. Thesis.
Michigan State University. East Lansing, MI. 2002.

Adapted from Boggan, B. Alcohol, Chemistry and You. Available on website
<www.chemcases.com/alcohol/alc—03.htm>.

United States Code of Federal Regulations, Title 27—Alcohol, Tobacco Products
and Firearms, Chapter I, Part 5, Subpart C—Standards of Identity for Distilled
Spirits. <www.access.gpo.gov/nara.cfr/cfr-table-search.html>.

The Merck Index, 11th ed. Rathway, New Jersey, 1989.

Pharmco Products Inc. Material Safety Data Sheet, Methanol. Available from
website <www.pharmco-prod.com/pages/METHANOLMethylAlcohol.pdf>.

Guymon, J.F., "Higher Alcohols in Beverage Brandy: Feasibility of Control of
Levels." Wines and Vines, 53, (1) 1972.

Stinson, E.E., et al., "Composition of Montmorency Cherry Essence. 1. Low-
boiling Components". Journal of Food Science, 34, 1969.

Claus, Michael J ., "An Investigation of Relationship Between Tray Usage of the
Still and Congener Compound Concentration in Distilled Fruit Spirits." MS.
Thesis. Michigan State University. East Lansing, MI. 2001.

Zimmerli, B., and Schlatter, J., "Ethyl Carbamate: Analytical Methodology,
Occurrence, Formation, Biologic Activity and Risk Assesment." Mutation
Research, 259, 1991.

Piggott, J.R., and Patterson, A. ed., Distilled Beverage Flavour: Recent
Developments. Ellis Horwood Ltd., 1989.

Grube, M., et al., "Application of Quantitative IR Spectral Analysis of Bacterial

Cells to Acetone-Butanol-Ethanol Fermentation Monitoring." Anal. Chem. Acta.,
471, (1) 2002.

35

14. Morrison, R., and Boyd, R., Organic Chemistry 6th ed. Pretence Hall,
Englewood Cliffs, New Jersey. 1992.

36

4. MATERIALS AND METHODS

4.1 Overview

Description of batch column distillation by computer modeling involves many
steps. First, a distillation medium must be designed to test the methodology. Two types
of batches were used and compared for testing the methodology of this procedure.
Standard batches based on the congener components, and fruit fermentation batches were
the feed for the distillations. The distillation of standard batches allowed for material
balances on the process, where fermentation of fruit allowed for a better understanding of
the process as implemented in the fruit spirits production process. Analysis of the
concentration of the compounds studied was completed using gas chromatography (GC).

A simple bench-top distillation was used to determine the concentrations of the
compounds in the mash to input as the feed to the distillation modeling program
CHEMCAD“. This distillation was run to completion to achieve complete recovery of
the con gener compounds. Analysis of the distillate provided values to enter for the feed
for the simulation program. Comparison of the modeling program with actual distillation
can allow the distiller to predict where to make cuts and therefore improve the quality
and yield of spirits produced. Figure 4.1 illustrates the procedure used in comparison of

the computer modeling prediction approach with distillation results.

37

t

\ j ’ )
L. “ “,4 15:;

Simple Distillation

Fermentation

     

Analysis

 

/
\-

 

 

 

 

CHEMCADTM

/

Figure 4.1 Procedure of comparing CHEMCADTM predicted concentrations

 

 

 

Comparison

 

against the experimental concentrations during distillation. The
simple distillation of the mash provides the concentrations of each

component for input into the CHEMCADTM program.

38

4.2 Fermentation

The fermentation of fruit mash followed procedures common to industrial
practices. Fruit used in this process was crushed using either a rolling mill or a mortar
and pestle. The whole fruit was used in the fermentation mash to provide increased
flavor components. Care was taken to prevent seeds from being crushed in the mashing
process, as this can lead to undesired products in the distillate. After mashing the fruit,
the mash was fermented in temperature controlled ferrnenters equipped with stirrers.
These fermentation conditions increase the ethanol yield when compared to uncontrolled
fermentationsl. Fermentation is a controlled form of the natural rotting process
experienced by all fruit. By controlling the fruit, temperature, atmosphere, and stirring
the mash a higher quality is expected than an uncontrolled fermentation.

Pries de mousse yeast (Lalvin, EC-1118, Lallemand Inc.) was used for fermenting
the mash. This strain of saccharomyces cerevisiae is used in whole fruit fermentations
because of its robust nature and ability to survive in a wide range of conditions. The mash
was held between 13 and 15°C allowing the yeast to be active, while also producing
ethanol rather than other compounds. Ferrnentations were allowed to continue for
between ten days and three weeks based on the extent of fermentation that was monitored
using refractive index measurements of the liquid mash. A portable refractometer
available from Fisher (Catalog Number 1394621) was used to measure the dissolved
solids (Brix) present in the liquid portion of the mash. This analysis allowed for the
monitoring of the sugar in the fermentation, which is consumed by the yeast to produce

ethanoll‘ 2.

39

Two scales of fermentations were used in this study; a 200 gallon (750 L)
ferrnenter and 30 gallon (115 L) ferrnenter. Each batch fermentation produced multiple
distillations to attempt to eliminate error associated with different fermentations. In
previous work it was determined that the fermentation mash composition varied from one
fermentation to the next due to differences in sugar content, yeast activity, and
temperature variations. By fermenting large batches and distilling multiple samples of

the fermentation batch the error associated with fermentation variation is lessened.

4.3 Standard Solutions

A solution was prepared to simulate the composition of the liquid phase of a fruit
mash after fermentation. An aqueous solution of the ten highest composition components
present in the solution was prepared. These solutions did nothave solid particles from
the fruit flesh, yeast, unferrnented sugar or other components common to fermentations of
fruits. The concentration of each component was based upon simple distillations of fruit
mash to attempt to mimic the concentrations of these mashes as best as possible. The
standard batches were used to reduce the dependence of this research on fruit, and
attempt to reduce errors caused by working with fermentation mixtures with unknown
concentrations. The standard batches allowed for material balances to be performed on
the process as well.

Two sizes of batches were created, one for use with the 150 L still and one for use
with the 10 L still, respectively. A 30 L standard batch allowed three distillations in the
10 L still and a 750 L batch allowed five distillations on the larger scale still. The

composition of the standard batches can be seen in tables 4.1 and 4.2. Each batch had an

40

ethanol concentration of 8%, similar to that found in fruit fermentations before
distillation. Each batch was well mixed before use.

Standard batches were treated in a similar manner as the fruit mashes. Three
samples from each batch were taken for simple distillation and analysis via gas
chromatography. These results were first checked to verify that the concentration of each
component was similar to that of the standard batch, and these values were entered into
the computer simulation program. Distillations were run in the appropriate still (10 and
150 L) and the distillate was analyzed using GC. A comparison was made between the
distilled samples and the computer programs predictions. Table 4.3 list the manufacturer
and lot numbers for the chemicals used in making these standard batches.

Table 4.1 Standard batch composition for the 10 L distillations. The

composition of the 30 L mixture used for diStillation in the 10 L still.

 

 

 

 

Compound Volume (mL) Concentration (% WV)

1 -Propanol 24 0.08
Acetaldehyde 1 5 0.05
Acetone 1 5 0.05
Benzaldehyde 2 0.0067
Ethnaol 2400 8.00
Ethyl Acetate 6 0.02
Ethyl Formats 6 0.02
Isoamyl Alcohol 24 0.08
Methanol 24 0.08
t-Butyl Alcohol 24 0.08
Water 27460 91 .53
Total 30000

 

41

 

Table 4.2

Standard batch composition for the 150 L distillations. The

composition of the 750 L mixture used for distillation in the 150 L

 

 

 

still.

Compound Volume (L) Concentration (% v/v)
1-Propanol 0.90 0.12
Acetaldehyde 0.45 0.06
Acetone 0.30 0.04
Benzaldehyde 0.1 5 0.02
Ethnaol 60.00 8.0
Ethyl Acetate 0.15 0.02
Ethyl Formate 0.20 0.0267
Isoamyl Alcohol 0.75 0.10
Methanol 1 .00 0.133
t-Butyl Alcohol 0.50 0.067
Water 685.60 91.41
Total 750.00

 

 

 

 

 

 

Table 4.3 Manufacturer and Lot Numbers for compounds used in standard
batches prepared.
Compound Manufacturer Lot Number Notes
t-Propanol Spectrum OA0142 and Kl389
Acetaldehyde Sigma HO 00550 HO 99.5% +
Acetone J. T. Baker T 38821
Benzaldehyde Aldrich 09323LA 99% +
Ethanol Pharmco 209184 Absolute Anhydous 200 proof
Ethyl Acetate Columbus Chemical Industries. 200019824 ACS Grade
Ethyl Formate Sigma 1051 IDA 97%
Isoamyl Alcohol Sigma 012K1320 98%
Methanol Spectrum QR0243 Spectroscopic Grade
t-Butyl Alcohol Spectrum PL0883 2-methyl-2-propanol
Water Distilled and Deionized on Campus

 

42

 

4.4 Rayleigh Distillation

A bench-top Rayleigh distillation was used to determine the concentration of the
congener components in the fermentation mash. A 250 to 300 mL sample of the
fermentation mash was distilled in a simple distillation apparatus similar to the one
shown in figure 4.2. The mash was heated using a silicon oil bath and cold tap water was
used to cool the condenser. The mash was distilled five minutes after the temperature
reached 98°C to achieve complete recovery of the congener components present in the
mash. This process required an average of 65 minutes. Longer distillations where the
mash was heated to 100°C for ten minutes, 110 minutes total runtime, produced results
where congeners present in smaller concentrations were reduced below their detection
level by GC due to the increased water content of the distillate. In addition, these longer
distillations produced congener concentrations similar to those of the 65 minute run.

These Rayleigh distillations were run in triplicate and the distillate of each was
analyzed using gas chromatography. Triplicate GC analyses were performed on each
distillate yielding nine total data points for each congener for each fermentation mash.
The concentrations of the compounds present in the distillate were determined from a
standard curve for each compound using ethanol as an internal standard. The compound
concentrations were then averaged and normalized to the concentration of ethanol present
in the fermentation mash. These values were used in subsequent modeling attempts as

the starting concentration for each compound present in the fermentation mash.

43

 

 

 

 

 

 

 

 

a
~———Thermometer
i
n ' Condenser
, In : _ _
1.. i 13‘
iii \_Water
_. - Outlet Water .1;
Inlet
31:313th Collection
‘ Flask l ‘
Silicon Oil Bath

Figure 4.2 Batch bench top distillation apparatus. The procedure was carried
out using 250-300 mL of mash. The distillation was held at 98° C for
5 minutes to achieve complete collection of the congeners. The

distillate ranged between 25 and 55 mL recovered3.

4.5 Computer Modeling

4.5.1 Using the CHEMCAD 7” Computer Program

The Chemstations Inc. CHEMCADTM program using batch distillation mode is
used for modeling the distillation process. The computer program can predict the
composition of the compounds present in distillate and bottoms at time intervals set by
the user. The program requires many steps to reach the final solution; however, the steps
for producing results are not overly complicated“.

Using CHEMCADTM involves first beginning a job which includes choosing a file
name and the operating units included in the flowsheet. The flowsheet comprises the
input and output streams and the application equipment that is to be used. The default
units are based on the English system of units; however, a wide range of units are
available. For the present work, the units chosen were based on the metric system. Table

4.4 lists the units for these experiments4.

45

Table 4.4

Units used in CHEMCADTM program. These units allow for easier

comparison of the CHEMCADm predicted values with the values

acquired from actual distillationss.

 

 

CHEM CAD TM Unit Selections

Total Flow StdL Uh
Component Flow Liquid Volume Fraction
Stream Edit Automatic Conversion
Time h Viscosity
Mass/Mole kg Surface Tension
Temperature C Solubility Par.
Pressure atm Dipole Moment
Enthalpy kJ Cake Resistance
Work kW'h Packing DP
Liquid Volume Liter Currency
Liq. Vol. Rate Liter Currency Factor
Crude Flow Rate BPSD
Vapor Volume in3
Vapor Vol. Rate m3/h
Liquid Density kg/L
Vapor Density kg/m3
Thickness m
Diameter m
Length in
Velocity m/sec.
Area m2
Heat Capacity kJ/kmole'K
Specific Heat kJ/kmole
Heat Trans. Coeff. KW/mz'K
Therm. Conduct. W/m'K

Pa-sec

N/m
(J/m3)“0.5
C'm

m/kg

mm Water/m
$

1

 

46

 

After selecting the appropriate units, the chemical components are chosen. The
list of chemical components in CHEMCADTM allow for identification via chemical
formula, chemical name, or numerical identifier. Ethanol, for example can be identified
by name (ethyl alcohol, ethanol) chemical formula (C2H60) and numerical identifier
internal to the computer program (134). The list of chemical compounds includes both
organic and inorganic compounds and numbers about 1,8000 components‘. Those
compounds not present in the list can be added if some chemical information is included
as well. For the purposes of these experiments, all the compounds of interest were
included in the list in CHEMCADTM.

After entering the chemical compounds of interest, the thermodynamic model and
K-values are chosen. CHEMCADTM will suggest the best thermodynamic and K—value
options for the compounds that were entered in the previous step using the Thermo
Wizard option; however, the user can modify the thermodynamic and physical properties
at any time in the simulation process. The concentration and volume of the components
in the pot of the still (input stream) are then input into the program“.

The next step in setting up the simulation program involves specifying the details
on the unit operations. For example, in this work the number of trays on the still, the
number of operations, the reflux ratio, the volumetric flow rate of the distillate, the step
size, record frequency and stop criteria were all specified. The number of trays refers to
the number of physical trays used by the still; however, the size and type of tray are
included later in the process. The number of operations refers to the different operating
conditions used in the process. For example if the distillate flow rate, or reflux ratio were

changed, a new operation would need to be included. These operations run sequentially,

47

and should not be confused with trays or analysis steps. The time step size refers to the
timeframe in between analysis, for example, a setting of 0.05 hours will allow for the
numerical integration once every three minutes4. The record frequency refers to the
number of time steps where the data is recorded. The default is once every third time
steps. Stopping the operation can be completed via many methods. Time, volume of the
distillate, composition of the distillate and temperature are all options for stopping the
distillation. For these experiments a stop time was utilized, to correspond with the
average distillate collection time associated with the distillation process“.

After specifying the unit operations, the physical properties of the equipment can
be included in the simulation process. Tray type (bubble cap or sieve) and the size of the
components can be included. The help function of the CHEMCADTM program clearly
describes the individual measurements needed for this portion of the process4' 5 .

Finally running the simulation and plotting the results complete the simulation
process. The software calculates and plots the results in these experiments as liquid
volume fractions (concentration) against time. The time frame can be converted to
cumulative distillate volume using the distillate flow rate. The raw data at each point can

be transferred to Microsoft ExcelTM in using the plot batch column history function of the

. 4,
software and processed as desrred 5.

48

4.5.2 CHEM CAD W as a Predictive Model

The compounds used in the modeling of the fruit spirits by the CHEMCADTM
program are given in table 4.5 with the chemical formula, name and internal
CHEMCADTM numberss. The concentration of these compounds were entered based on
the results of the GC analysis of the simple batch distillation. The thermodynamic model
chosen by the Theme Wizard was the NRTL method for its ability to model non-ideal
solutions. The six thermodynamic models selected as best for the alcohol distillations
included; NRTL, ESD, UNIFAC, Modified UNIFAC, UNIFAC LLE, and UNIQUAC.
All six of these models were used for these experiments

The appendices give an overview of the thermodynamic models and the
methodology for determining the best thermodynamic model to use for analysis. The
timeframe, distillate flow rate, and analysis time were all based on a distillate volume
consistent with the volumes collected during the actual distillation of these fruit spirits
which were 750 ml for large still distillations and 75 ml for 10 L still distillations. The
data were analyzed in Microsoft ExcelTM to plot concentration against cumulative-

distillate-volume.

49

Table 4.5 Compound name, formula and internal CHEMCADTM number for
identification. These compounds were analyzed in the distillate. The
chemical name common to CHEMCADTM were given as well as

alternative namess.

 

Compound/Alt. Name Formula CHEMCADTM Number

Ethanol C2H60 134
Ethyl alcohol

Methanol C1140 1 l7
Methyl alcohol

Acetone C3HgO 140

Acetaldehyde CanO 128

Tert-butyl alcohol C4H100 161
2-Methyl-2-propanol

l-Propanol C3H30 146
N-propyl alcohol

3-Methyl-1-butanol C5H120 315
Iso amyl alcohol
Isopentyl alcohol

Ethyl formate C3H602 141

Ethyl acetate C4H802 155

Benzaldehyde C71160 342

Water H20 62

50

4.5.3 CHEMCAD 7’" Calibration

Designing the operation of the CHEMCADTM program around a variable reflux
still required a calibration method to produce approximations for the reflux ratio at the
different operation steps of the distillation. As the flow rate of the cooling water into the
condenser varied, the reflux ratio and distillate flow rate also varied. This is not a set of
conditions that is directly possible with the simulation software which uses a single reflux
ratio throughout the distillation. To mimic the results of a commercial batch distillation,
it was necessary to divide the distillation into a number of smaller steps each of which
possessed its own reflux ratio and distillate flow rate to produce a matching distillation
profile by CHEMCADTM. To approximate the reflux ratio and distillate flow rate for
each operation step, a distillation of 8% ethanol in water was used. Volume fractions
were collected in the same manner as the actual batch distillation, however, the time for
each fraction was also recorded, and each fraction was analyzed using the GC. The time
for each volume fraction was calculated and the total distillation was broken up into the
appropriate number of operation steps.

The flow rate data were entered into each of these operation steps and the data

were plotted using CHEMCADTM. By comparing the ethanol concentration against

distillate volume for the CHEMCADTM results and the actual concentration data obtained
from the GC analysis of the distillate, the reflux ratio for each operation step was varied
until the CHEMCADTM prediction matched that of the distillation. These operation steps
which included parameters (e. g. reflux ratio, distillate flow rate, analysis time, end
conditions) at each time interval were used as the basis for the CHEMCAD predictive

modeling of these fruit spirits distillations. It is assumed that by modeling the distillation

51

reflux ratio on the experimental results of an 8% ethanol/water binary mixture, the error
in reflux ratio and flow rate will be minimized as the bulk concentration of the fruit mash

is an 8% ethanol in water solution.

4.6 Experimental Multistage Batch Distillation

The distillation of fruit spirits was performed at two scales. A 10 L still and a 150
L still were used in this study. Both stills are of German manufacture and made of
copper. The 10 L Holstein still uses electrically heated water to produce steam, and the
150 L Christian Carl still uses direct injected plant steam for the heating of the pot. The
10 L still uses bubble cap trays where the 150 L still uses sieve trays. Other than these
few exceptions, the stills use the same principles for operation. Figures 4.3 and 4.4 are
schematic diagrams of the 10 L and 150 L stills used in these experiments, respectively.

The distillation of fruit spirits uses the variable reflux mode of operation to
attempt to keep the concentration of alcohol constant throughout the distillation. The
temperature at the top of the condenser was kept at 72°C for the duration of the
distillation2‘6. The sensor at the top of the condenser regulated the flow of the cooling
water into the system which in turn varied the reflux ratio of the distillation process. This
process improves ethanol yield when compared to fixed reflux ratio systems].

The distillation process was begun by filling the pot with fermented mash and
filling the condenser and partial condensers with water. A silicon based antifoam agent
(Dow Corning, AF Emulsion, Food Grade, Lot # III-1077895) was added to the mash to
reduce the frothing that occurs in these types of distillations. Next the stirrer was turned

on for mashes that are heavy in solid material to prevent baking and also to prevent

52

foaming. The steam was then injected into the steam jacket and heating begins. Slow
heating works best and increases the yield of ethanol produced. As the heating of the
mash continues, the evolution of lower boiling compounds increases and the vapor passes
onto the trays. As the vapor comes in contact with the cool metal surface, the vapor
condenses. Vapor then has to pass through this condensed layer increasing the separation
of the compounds. Finally after passing through all the trays the vapor was condensed in
the condenser. The liquid distillate was water clear and came out from the distillate tube
at the end of the still.

Three collection fractions are commonly recovered in fruit spirit production. The
first fraction is the heads which contains the highest concentration of lower boiling point
(relative to ethanol) compounds. The concentration of ethanol is at its greatest in this
fraction, but the heads are not potable and blending with other fractions is not practiced
due to the high concentrations of acetaldehyde, acetone, and methanol. The next fraction
of distillate is the hearts which will become the potable product. The hearts has an
average ethanol concentration of approximately 72-77% alcohol by volume. Also, the
hearts, has relatively low concentrations of low and high boiling compounds and
methanol. Often more than one heart cut is taken and blended together after aging.
Finally the tails cut is collected. The ethanol concentration is usually below 55% A.B.V.,
but the presence of relatively high concentrations of fusel alcohols and terpenes reduce
the clarity and overall aroma of the distillate. Tails cuts are often collected and redistilled
to increase the overall yield of ethanol from the process; however, the resulting distillate
from this re-distillation is inferior when compared to a hearts out from a distilled fruit

spirit. After collecting the tails, the steam is turned off and the cooling water is turned on

53

to maximum. The pot is emptied and cooled and cool water is used to clean the trays.

The distillate is aged in glass before dilution and bottling.

54

To Drain Temperature
1 Sensor

  
  
 

 

 
    

/////// if}:

     

uni-In.
"a
I
.
r5. =‘ir=.'=:?:‘ix'- 214$“?
. < ‘

Partial i’
—* r327e572.EEE;::_E:;>;:-Lj.g;:-*
Condenser

 

      
       
 

/

7f

/

   

      
    

/

     

7/

 
 

Cooling Water
Regulator

 

 
  

L Steam
j Release

 

 

 

 

   
 

Distillation

 

 

Cooling
Distillate Water
Collection Inlet
of“ Electric

I Heater

 

Steam Jacket

Figure 4.3 Schematic of 10L Holstein Still. The electric heater is used to

generate steam from an internal water source.

55

 

 

 

 

 

 

 

 

 

 

 

    
     
       
 
  

 

 

 

 

 

 

 

 

Steam Jacket Distillation Vapor Path
Partial Condensers Cooling Water Path
Total Condenser ”may?" i
_ ,._. .Il 4 ...__._.. —~-~~— I atalytic
[3 Copper Packing I : Converter
I Fermented Mash :1, 2. 1
41°— _ _ -
Drain
Steam Cooling Water
Inlet Control

Distillate
g Discharge

Water

   
 
 

Bottoms
Removal
Port

 

 

Figure 4.4 Schematic diagram of 150L Christian-Carl Stillz.

56

4.7 Comparison

4.7.1 Predictive Comparison of CHEMCAD 7’”

The concentrations of the components in the distillate were compared with the
predicted CHEMCADTM values. The cumulative distillate volume was used as the
independent variable instead of time because in the laboratory it is easier to plan an
experiment around a constant volume fractionation procedure than a constant time
fractionation procedure. The volume fraction concentration used in the dependent
variable is in terms of % v/v, a common concentration unit in the distilled beverage

industry.

4.7.2 Standard and Fruit Batch Distillations

The standard batch distillations were compared with fruit spirits distillations by
comparing congener concentration (% v/v) against cumulative distillate volume (mL).
These comparisons were made to increase the reproducibility of the results as fruit spirits
distillations are often not reproducible on a batch basis due to varying concentration of

the fermentation fed.

4.8 Gas Chromatography Methods

A Shimadzu GC-17A gas chromatograph with flame ionization detector (FID),
and an AOC-20i autosampler (12 vial capacity) was used for analysis of the distillates.
The column used for the analysis is a 30 meter long 0.25 mm id. Stabilwax 30 column
obtained from Restek. A 0.2 ILL injection volume was used for all samples to avoid

overloading the column. The injector and detector temperature were set at 240°C and

57

255°C respectively. Separation of the compounds was achieved using a ramp program
for the column. The initial temperature of the oven was set at 38°C and a heating rate of
1.2°C/min was used until 80°C was reached. A 30°C/min heating rate was used from
80°C to 170°C and the final temperature was held for 2 minutes. The total cycle time of
the analysis was 42 minutes which includes the sample preparation, injection of the
sample, and cooling of the GC oven. Each sample was run in triplicate and average
values were used in all results. Figures 4.5 and 4.6 show sample chromatographs taken

from the distillate of one fruit batch.

58

 

1000,
Ethanol

/

500t

mfi—o<3

 

 

 

 

l-Propanol Isoamyl Alcohol Benzaldehyde

. f L 1L . - - . r
0 10 20 30 40

 

Minutes

Figure 4.5 Sample chromatograph. This sample chromatograph is from a heart
cut of an apple spirits distillation. The peaks of the compounds with

longer retention times are identified.

59

50

 

Methanol
\ Ethanol

Acetone /
25. \

Acetaldehyde

\ l-Propanol

0. _LILAI\1L l/

l

O . . l - - . - - - -
Ethyl Acetate 5 10

Minutes

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.6 Sample chromatograph. Enlarged view of the chromatograph shown
in figure 4.5. The peaks identified represent the compounds that elute

in the first ten minutes of the chromatography run.

60

4.9 Retention Time and Reproducibility Validation

The Shimadzu GC-17A equipped with an AOC-i Autosampler and the
temperature program outlined above was able to produce reliable, reproducible results.
Each distillation fraction was analyzed in triplicate via this chromatographic method. In
addition, because there were multiple distillations from each fermentation batch or
standard fermentation batch, the results from each set of conditions were able to be
analyzed for instrumental errors. The use of the autosampler system allowed for
reproducible retention times throughout the duration of these experiments. Table 4.6 lists
the compounds analyzed, their average retention time, and the error found in the retention
time. These results are complied from two sets of procedures. One large still batch
fermentation, with the analysis of the simple distillate (n = 9), three distillations with 16
fractions (n = 144) and one small still batch fermentation, with simple distillation
analysis (n = 9) and three batch distillations (n = 81). The retention times were constant
throughout these experiments as shown by the small percent error over the large sample
size (n = 243). While this is just a fraction of the total data collected, this reliability in
reproducing the retention time of these components was present throughout the whole

experimental procedure.

61

Table 4.6 Mean retention time, and error associated with the compounds
analyzed in these experiments. The small error is due to reproducible

injection conditions because of the autosampler.

 

 

 

Compound I Mean Retention Time [Standard Deviation] % Error
Acetaldehyde 1 .450 0.006 0.42
Acetone 1 .996 0.010 0.52
Ethyl formate 2.058 0.017 0.81
Ethyl acetate 2.196 0.016 0.75
2-Methyl-2-propanol 2.657 0.01 5 0.55
Methanol 2.814 0.015 0.54
Ethanol 3.507 0.023 0.64
1 -Propanol 5.972 0.025 0.42
Iso amyl alcohol 14.690 0.036 0.24
BenzaldehLde 37.1 36 0.012 0.03

 

62

 

4.10

Literature Cited

Tanner, H., and Brunner, H.R., Fruit Distillation Today 3rd ed., (English
Translation) Heller Chemical and Administration Society, Germany, 1982.

Claus, Michael J., "An Investigation of Relationship Between Tray Usage of the
Still and Congener Compound Concentration in Distilled Fruit Spirits." MS.
Thesis, Michigan State University. East Lansing, MI. 2001.

Distillation apparatus set-up. Adapted from
(http://www.ComCapeCanaveral/lab/l4444/distillation.html>

Chemstations CHEMCADTM Process Simulation Software. Informational
Material. Available at website <www.chemstations.net> 2002.

Chemstations CHEMCADTM Process Simulation Software. Internal Help
Function and Tutorial. 2002.

Christian-Carl, Alexandr. Correspondence with Author. 1998-2001.

Diwekar, U., Batch Distillation. Taylor and Francis, Washington, DC. 1995.

63

5. RESULTS

5.1 Ethanol Modeling

The first objective was the modeling of the largest concentration component,
ethanol. Ethanol is the most important component present in the distillation of fruit
spirits, because the ethanol concentration in the fruit spirits related directly to the total
yield of the available product.

Preliminary modeling of the ethanol concentration in the fruit spirits with
CHEMCADTM was inconsistent with the experimentally obtained concentration profile
from distillations. Both the 150 L Christian-Carl and 10 L Holstein stills used vary the
flow rate of the cooling water introduced into the condensing column. As more cooling
water flows through the condenser the reflux ratio increases and the distillate flow rate
decreases. Because the flow rate of the cooling water increases and decreases throughout
the distillation, the reflux ratio is also constantly changing. The reflux ratio is also
changing due to changes in the distillate composition.

Figure 5.1 shows the difference between the experimentally determined ethanol
concentrations obtained fruit spirits distillations against the concentration of ethanol
predicted by the CHEMCADTM program with constant reflux ratios. These two results do

not agree well because of the varying reflux ratio in the experimental distillations.

64

Ethanol Water Prediction with Constant Reflux Ratio

90

 

70

 

Ethanol Concentration (% vlv)

 

 

50 . . . . I
0 2000 4000 6000 8000 10000 12000

Cumulative Distillate Volume (mL)

 

.-°-_ __ Emafiol/WatéDi—Ltilafism :57: 99119926 Etmgljfifiui: NJBLell

Figure 5.1 Concentration of ethanol (% v/v) against cumulative distillate volume
(mL) for an experimental distillation and two simulated constant

reflux distillations in CHEMCAD”. The reflux ratios used for the

CHEMCAD“ simulations were 2 and 5 and are given in parentheses.

65

Clearly the lack of agreement between experimental and simulated results is
largely due to the variation of the reflux during the experimental batch distillations. A
method was developed to introduce variable reflux ratio into the distillation modeling by
CHEMCADTM. The first step in this procedure involved distillation of an ethanol/water
mixture (8% ethanol v/v) through the 150 L and 10 L stills. The flow rate of each
fraction (750 mL and 75 mL cuts respectively) was measured and recorded. Each
distillation was performed in triplicate and the average flow rate of the distillate were
recorded. For the larger still, sixteen fractions of 750 mL were collected, and nine
fractions of 75 mL were collected for the smaller still.

The sixteen fractional cuts in the larger still were reduced to nine distillation steps
to be entered into the CHEMCADTM program. The first and last 750 L cuts were used as
steps 1 and 9. The other fourteen fractions were paired to obtain the data for steps 2
through 7. Cut 2 and 3 were averaged for the data for step 2, cut 4 and 5 were averaged
for step 3 and so on. These nine steps were then entered into the CHEMCADTM program
in terms of Uh for the distillate flow rate. The time of each step in terms of hours were
calculated so that CHEMCADTM would identify the data points at 750 mL increments for
the distillations.

The smaller still, with nine fractional cuts, did not require the averaging of flow
rates together to obtain the nine steps entered into the CHEMCADTM program. The same
methodology is followed for the small still as above, with the exception of the averaging
of the cuts, and the step size of each data point in the program were entered to result in 75

mL fractions.

66

With a fixed distillate flow rate data and the step size for analysis by the
CHEMCAD program set to yield the appropriate distillate volumes, the reflux ratio was
approximated for each step by trial and error such that the resulting ethanol concentration
at the end of each step met the concentration of ethanol determined by the experimental

distillation. Table 5.1 illustrates the data collected from the experimental distillations and
the resulting data input into CHEMCAD“. Table 5.2 shows the reflux ratio entered into
the CHEMCADTM program at each step and the resulting ethanol concentration predicted
by CHEMCADTM using these values. Figure 5.2 shows the experimental distillation
concentrations of the ethanol water mixture, and the predicted values from the
CHEMCADTM computer program. Figure 5.3 illustrates the procedure used to acquire

the reflux ratio for each step of the distillation model.

Table 5.1 The ethanol concentration, time, and calculated distillate flow rate for

the distillation of an 8% ethanol/water mixture in the 150 L still.

 

 

 

Cumulative Ethanol
Cut Volume Time Difference Flow Rate Concentration

Number (mL) (min) (s) (L/h) (% v/v)
1 750 1 .38 83 32.53 83.7
2 1 500 2.53 69 39.13 85.4
3 2250 3.67 68 39.71 85.9
4 3000 4.78 67 40.30 87.8
5 3750 5.88 66 40.91 84.5
6 4500 6.97 65 41 .54 83.7
7 5250 8.10 68 39.71 83.1
8 6000 9.28 71 38.03 81.5
9 6750 1 0.52 74 36.49 80.3
10 7500 11.92 84 32.14 78.0
11 8250 13.33 85 31.76 75.1
12 9000 14.92 95 28.42 72.6
13 9750 16.50 95 28.42 67.4
14 10500 18.15 99 27.27 62.7
15 11250 19.87 103 26.21 59.7
16 12000 21 .67 108 25.00 59.0

 

 

67

Table 5.2 The flow rates, analysis times, and reflux ratios entered into
CHEMCAD”. Also shown are the concentration of the ethanol
concentration predicted by CHEMCAD” when an 8% ethanol/water

distillation is simulated.

 

 

Step Ethanol
CH EMCAD Step Flow Analysis Number of Concentration
Step Rate Time Analysis Total Time Total Time Reflux Predicted
Number (Uh) (h) Points (h) (min) Ratio (% WV)
1 32.53 0.0231 1 0.0231 1 .38 2.05 83.70
2 39.42 0.0190 2 0.0421 2.52 3.55 86.16
0.061 1 3.67 85.85
3 40.60 0.0185 2 0.0796 4.77 3.42 86.56
0.0981 5.88 84.52
4 40.62 0.0185 2 0.1165 6.99 3.82 84.88
0.1350 8.10 83.12
5 37.26 0.0201 2 0.1551 9.31 3.67 81.94
0.1752 10.51 80.31
6 31.95 0.0235 2 0.1987 1 1.92 3.88 77.84
0.2222 13.33 75.13
7 28.42 0.0264 2 0.2486 14.91 4.22 72.80
0.2750 16.50 67.44
8 26.74 0.0280 2 0.3030 18.18 4.1 1 63.14
0.3310 19.86 59.74
9 25.00 0.0300 1 0.3610 21.66 4.29 59.02

 

 

 

The reflux ratios and distillate flow rates from the steps determined
experimentally from a binary ethanol/water distillation are used for the modeling of the
fruit fermentations. The ethanol concentration behavior in the distillation of a fruit mash
is assumed to behave like that of an ethanol/water binary mixture of the same ethanol
concentration present in the mash. It is assumed that the bulk of the fermentation mash
behaves as a binary ethanol/water mixture when distilled.

Table 5.3 gives the reflux ratio and flow rate for both sizes of stills used in these
experiments. Figure 5.4 shows the plot of the reflux ratio against distillate volume for the

150L still. The initial data point is low at the beginning of the distillation as the cooling

68

water is not flowing through the condenser. As the cooling water flow rate increases, the
liquid reflux will increase while the distillate flow rate remains constant as shown in the
second data point. As the cooling water reaches a maximum flow rate, the liquid reflux
into the column will decrease, resulting in a lower reflux ratio. Finally as the distillate
flow rate begins to decrease, the liquid reflux also decreases which results in the more
constant reflux ratio seen on the final three data points. The operating line (UV) is
constantly changing during the distillation to maximize the ethanol concentration of the
distillate. The reflux ratio (UD) begins at a lower value, reaches a maximum, and then
decreases through the rest of the distillation

Table 5.3 The distillate flow rate and reflux ratios entered into the

CHEMCAD"M program for each size of still.

 

 

 

 

 

10 1. stilt“ . f
Step Reflux Flow Rate Cumulative Distillate
Number Ratio (L/hL Volume (mL)

1 3.94 4.50 75

2 16.8 2.70 150

3 7.72 2.65 225

4 6.17 2.60 300

5 4.85 2.21 375

6 4.17 2.35 450

7 3.73 2.05 525

8 3.62 2.13 600

9 4.18 2.13 675
f w . 150LjStill . ’ _ ‘ ‘ ‘
Step Reflux Flow Rate Cumulative Distillate

Number Ratio (L/h) Volume (mL)

1 2.2 32.53 750

2 3.16 39.42 2250

3 2.87 40.62 3750

4 2.59 40.62 5250

5 2.23 37.26 6750

6 2.03 31.95 8250

7 1.84 28.42 9750

8 1.77 26.74 1 1250

9 1 .77 25.00 12000

 

69

 

90.0

 

 

 

 

 

l-g-Xctgaleisrméiiéfi :‘L‘c‘tjegegu_p}§aaeg virtue t
85.0
’2; V\ .
,2 80.0 » '\"
e \
.2
§ 75.0 . i
E \.
r \
a 70.0
2
g 65 0 I
g . \
1.1.1 \
60.0 ' xx. .,,
55.0 . . . .
0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)
Figure 5.2 Experimental and predicted values of the ethanol concentration at

volume intervals. The experimental distillation values are from a 8%

ethanol and water binary mixture.

70

 

Determine

Distillate Flow Rate

| —’ Ethanol Concentration

 

a 0000

.. Timeframe
'
Volume Units (750 mL
. . . d 75 mL
Distillation an )
Ethanol/Water Mix

CHEMCADTM Input
Number of Steps (9)

Step Flow Rate

Timeframe of Analysis

Reflux Ratio (trial and
error) CHEMCADTM Output

Ethanol Concentration at
Volume Units

Reflux Ratio Determined

When Ethanol Concentration (Experimental)
Equals Ethanol Concentration (CHEMCADTM)
to Three Significant Figures

Figure 5.3 The procedure used to determine the reflux ratio of each step of the

distillation.

71

 

3.5

2.5

Reflux Ratio

 

 

 

1.5 e . . .
0 2000 4000 6000 8000 10000 12000 14000

Cumulative Distillate Volume (mL)

Figure 5.4 Change in the reflux ratio over distillate volume. The line of best fit

for the reflux ratio is shown.

72

5.2 Simple Rayleigh Distillations

The procedure for identifying the concentration of the compounds present in the
fermentation mash, prior to batch distillation, was a simple Rayleigh distillation of the
mash and analysis of the distillate. The simple Rayleigh distillation apparatus is shown
in figure 4.2. The composition of the distillate from the Rayleigh distillation was
normalized from the chromatographic analysis to the equivalent concentrations in an 8%

ethanol mixture.

5 .2. 1 Standard Batches

The concentration profile of the standard batches are shown in table 5.4. Two
different batches were used to test the ability of CHEMCADTM to simulate the
concentration profiles of each of the compounds present in the distilled spirits. The
variation in the concentration of the congener compounds in these standard batches
allows for classifying the ability to simulate the behavior of each compound by the

CHEMCADTM program.

The first step in the process of using the CHEMCADTM program for this type of
determination, was to perform a Rayleigh distillation of the standard batch. Triplicate
simple distillations were performed in an attempt to eliminate errors in terms of the
congener concentrations present in the mash. Each simple distillation utilized 250 mL of
standard mash and the distillation was allowed to continue until the temperature on the
thermometer read 98° C for five minutes. By distilling to this point, the total recovery of
the congener compounds is assumed. The distillate was analyzed by gas chromatography

as outlined in chapter 4.

73

Table 5.4 The composition of the standard batches for these studies. The

same concentrations were used for the large batches (750 L) and

 

 

small batches (30 L).
Concentration (Volume
Fraction)
Compound Batch 1 Batch 2
Ethanol 8.000 8.000
Methanol 0.133 0.080
Acetaldehyde 0.040 0.050
Benzaldehyde 0.020 0.01 0
T-Butanol 0.067 0.010
1-Pr0panol 0.120 0.080
Isoamyl Alcohol 0.100 0.080
Acetone 0.060 0.050
Ethyl Acetate 0.027 0.020
Ethyl Formate 0.020 0.020
Water 91.413 91.600

 

 

 

Table 5.5 shows the composition of the congeners present in the distillate of the
simple Rayleigh distillations of the first standard batch, after normalization to a
concentration of 8% ethanol. The compositions in most cases were reasonably close to
the actual composition present in the distilled spirits. In the third set of simple
distillations larger errors could be accounted for due to the time that the sample sat before
being distilled in the simple distillation apparatus. In all of the other cases, the simple
distillation was carried out on the same day as the large scale distillation, and the day
following the solution preparation.

The second standard batch was not completely distilled the day after the solution
was made. Two large scale distillations were carried out the day after the solution was
made, and the first two sets of simple distillation data were taken and run on those days.

The rest of the batch was distilled two days later, accounting for evaporative changes to

74

the solution composition before distillation, and altering the composition present after the
simple distillation.

Table 5.5 The composition of the first standard batch and the average
composition normalized from the simple distillation procedure. The
concentration of the components after the simple distillation were

normalized to a concentration of an 8% ethanol solution.

 

 

Actual Simple Simple Simple

Composition Distillation Distillation Distillation
Compound % WV (1) % WV (2) % WV (3) % v/v
Ethanol 8.000 8.000 8.000 8.000
Methanol 0.133 0.154 0.122 0.272
Acetaldehyde 0.040 0.038 0.066 0.1 35
Benzaldehyde 0.020 0.031 0.045 0.054
2-Methyl-2-Propanol 0.067 0.009 0.031 0.014
1-Propanol 0.120 0.1 13 0.1 18 0.209
Isoamyl Alcohol 0.100 0.120 0.083 0.132
Acetone 0.060 0.052 0.045 0.079
Ethyl Acetate 0.020 0.01 1 0.001 0.013
Ethyl Formate 0.027 0.015 0.001 0.019
Water 91 .413 91.455 91.491 91 .073

 

 

 

The simple distillation has been able to generate concentration ranges similar to
those of the actual standard batch. Error can be expected from evaporative losses, as well
as incomplete distillation during the simple distillation procedure.

For the small scale distillations (10 L) of standard mashes, the same procedure
was followed as above. The concentration of the standard batches and resulting
concentrations from the Rayleigh distillate analysis can be seen in table 5.6. The
composition of Rayleigh distillate is shown as a mean and standard deviation of the

triplicate runs of the simple distillate on the standard batch.

75

Table 5.6 The average composition of the congener components of the second

standard batch used for the small (10 L) scale distillations. Three

Rayleigh distillations were averaged for these results.

 

 

 

 

Standard Rayleigh Distillate
Mixture Concentration

Standard

Compound Composition Average Deviation
Ethanol 8.000 8.000 0.0920
Methanol 0.080 0.076 0.0017
Acetaldehyde 0.050 0.052 0.0023
Benzaldehyde 0.01 0 0.009 0.0065
2-Methyl-2-Propanol 0.010 0.001 0.0007
1-Propanol 0.080 0.079 0.0017
Isoamyl Alcohol 0.080 0.086 0.0040
Acetone 0.050 0.038 0.0008
Ethyl Acetate 0.020 0.008 0.0000
Ethyl Formate 0.020 0.009 0.0000

 

 

The concentrations from the simple Rayleigh distillation and the actual
composition of the mixture were entered into the CHEMCADTM computer program. The
CHEMCADTM data was then normalized to concentrations of % v/v in 40% ethanol. This
concentration range will be used for comparison of the distillate from the large
distillation, with that of the CHEMCADTM predictive values. The 40% ethanol solution

is the common drinking/bottling strength of brandies, vodkas, whiskeys, gins and most

other distilled beverages].

76

5.2.2 Fruit Mash Batches

A mixture of apple varieties was used for the fermentation mash for both scales.
This general apple mash was prepared in the standard manner, of crushing the fruit
through a rolling mill, and fermented inside a temperature controlled ferrnenter with
stirring. An approximate volume of 600 L of mash was fermented using saccharomyces
cerevisiae as the yeast. The fermentation lasted two weeks and three distillations were
carried out on both the large (150 L) and small (10 L) scale. Before each distillation two
250 ml samples were distilled using the Rayleigh distillation apparatus discussed above.
Analysis of the distillate from the Rayleigh distillation conveyed congener
concentrations, which were then normalized to a concentration of 8% ethanol. The

results of these Rayleigh distillations of the fruit mash are presented in table 5.7.

77

Table 5.7 Mean congener concentration and standard deviation from analysis of

the Rayleigh distillate from apple fermentation. The concentrations

have been normalized to an ethanol concentration of 8% v/v.

 

 

 

Mean Standard

Compound Concentration Deviation
Ethanol 8.0000 0.0980
Methanol 0.0981 0.0059
Acetaldehyde 0.01 1 1 0.0004
Benzaldehyde 0.001 9 0.0026
T-Butanol 0.0036 0.0004
1-Propanol 0.1401 0.0183
Isoamyl Alcohol 0.0525 0.0155
Acetone 0.0036 0.0008
Ethyl Acetate 0.001 3 0.0018
Ethyl Formate 0.0000 0.0000

 

 

5.3 Batch Distillation with Reflux

Batch distillation of the standard and fruit mashes were carried out in the same
manner. The procedure for using the multistage batch still with reflux was outlined in
chapter 5. The sample volume for the larger Christian-Carl still is 150 L and the volume
for the smaller Holstein still is 10 L. The antifoam agent was only added to the fruit
distillations as the standard distillations did not exhibit excessive foaming. Also, the lack
of a stirrer in the smaller still caused some of the apple mash to bake onto the copper pot

of the still, making cleaning more difficult.

5.3.1 Standard Solution Multistage Batch Distillation with Reflux

The batch composition of the standard "mash" solution can be seen in the tables

above. The 10 L still distillations each produced nine fractions of 75 mL each for a total

78

collected volume of 675 mL. Sixteen fractions of 750 mL were collected from each of
the 150 L still distillations for a total distillate volume of 12000 mL. The concentration
of the congener compounds in these distillate fractions were then determined by gas
chromatography. The results of these distillations were then compared to the predicted
values produced by the CHEMCADTM program. The comparison of CHEMCADTM
simulations with the experimental distillations for each compound can be seen below for

each compound.

5.3.2 Fruit Mash Multistage Batch Distillation with Reflux

Apples used in these batch distillations were allowed to ferment for two weeks,
before distillation. The fermentation mash was then pumped/poured into the pot of the
still and the sample for the Rayleigh distillation was taken from the pot of the still before
the batch distillation commenced. The volume fractions of the distillations were taken at
the same intervals above, and the same procedure was used for analyzing the

concentrations of the compounds present in each fraction.

5.4 CHEMCADTM Predictive Modeling

The CHEMCADTM program has many different thermodynamic models that can
be used for the analysis of chemical processes. Many of these thermodynamic models
are not applicable to alcohol distillations. The Thermodynamic Wizard function within
CHEMCADTM identified the NTRL model as the best thermodynamic model for the
compounds used in these experiments. Also used in these experiments were the ESD,

UNIFAC, Modified UNIFAC, UNIFAC LLE, and UNIQUAC thermodynamic models.

79

The predictive models of CHEMCADTM generate the volume fraction of each
compound at given time intervals. The concentration of these compounds was then
normalized to their concentration in a 40% v/v ethanol solution and the time function was
converted to cumulative distillate volume using the flow rate information entered into the
CHEMCADTM program. These data are then compared against the concentration of each

compound present in the distillate of the fruit and standard batch distillations.

5.5 Compound Comparison

Comparing the CHEMCADTM predicted values with the actual values from the
batch distillations are done as concentration of congener in 40% ethanol against
cumulative distillate volume. The three thermodynamic models within CHEMCADTM
that fit the results of the actual distillation are shown to improve clarity. Each compound

has sets of thermodynamic models that best fit the profile expressed during the batch

distillation process.

5.5.] Ethanol

The ethanol data shown use values of concentration (% v/v) against cumulative
distillate volume. Figures 5.5, 5.6, 5.7 and 5.8 show the composition predicted by the
CHEMCAD program with the concentrations achieved with experimental batch
distillations. The thermodynamic models that best matched the concentration of the
distillate were the Modified UNIFAC, NTRL and ESD models. The larger scale batch

distillations show better agreement between the predicted values and the experimental

80

concentrations than is present in the smaller scale distillations. This is due to the
increased holdup present on the trays of the 10 L still.

Because the concentration of the fruit mash is approximated to 8% v/v which is
comparable with that which is found in a fruit mash, the average ethanol concentration of
the fruit mash is complimentary between the thermodynamic models and the
experimental distillation. For example the average concentration of the 10 L fruit
distillation is 71.11 i 0.47 % v/v. The average concentration for the NTRL and ESD
models is 68.63% v/v and the average concentration for the Modified UNIFAC model is
70.60 % v/v. These represent the concentration of a blend of the nine fractions collected.
Table 5.8 illustrates the average ethanol concentration of distillate collected against that

predicted via CHEMCAD“.

81

Table 5.8 Average ethanol concentration of a blend of all cuts taken during
distillation. The total yield of ethanol produced is similar to those

values predicted by the CHEMCAD“ thermodynamic models shown.

 

 

 

Batch Modified Total Volume

fi Distillation Error NTRL ESD UNIFAC (mL)
10 L Fruit
Distillation 71 .1 1 0.47 68.63 68.63 70.60 675
10 L
Standard
Batch 69.1 1 1 .50 68.10 66.80 69.70 675
150 L Fruit
Distillation 75.94 1 .36 74.71 73.56 76.02 12000
150 L
Standard
Batch 74.13 1 .44 74.01 72.86 75.1 1 12000

 

 

82

85

 

l—o— 10 1. Fruit Distillation?
l—x— NTRL I

l
1 “EB- ESD
Ff 1-35.9"1919111569

l
l

 

75'

65 4

Ethanol Concentration (% vlv)

 

 

55

 

0 100 200 300 400 500 600 700 800
Cumulative Distlllate Volume (mL)

Figure 5.5 The ethanol concentration of a 10 L fruit distillation compared to the
values predicted by the CHEMCADTM computer model using three

thermodynamic models.

83

85

 

75

65

Ethanol Concentration (% v/v)

 

l—o— 10 L Standard BatchI
I Distillation

—x— NTRL

 
    

I
-i*""'<‘\ ...... I ESD
I

1+ Modified UNIFAC

 

 

\ at
55 l . .
0 1 00 200 300 400 500 600 700 800
Cumulative Distillate Volume (mL)
Figure 5.6 The ethanol concentration of a 10 L standard batch distillation

compared to the values predicted by the CHEMCADTM computer

model using three thermodynamic models.

84

85

 

+150 L—Fruit Distillation '
jar—NRTL
. ESD '

fir-Modified UNIFAC L ’

   
    

751

65 ~-

 

Ethanol Concentration (% vlv)

 

 

0 2000 4000 6000 8000 10000 12000
Cumulatlve Distillate Volume (mL)

Figure 5.7 The ethanol concentration of a 150 L fruit distillation compared to the
values predicted by the CHEMCADTM computer model using three

thermodynamic models.

85

85

 

 

 

 

 

 

/ A.
'51
‘2 g.
o ’7.
c 75 ~ g g
2 .
‘5
E
3 .- - m .. L.
3 +150 L Standard Batch l
o l Distillation '
g 65 - l-)K—NRTL }
2 I l
111 ' ESD l
I l
:-+—-Modified UNIFAC ;
I __,_,__,r-___ _-_ .
55 . . . .
0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)
Figure 5.8 The ethanol concentration of a 150 L standard batch distillation

compared to the values predicted by the CHEMCADTM computer

model using three thermodynamic models.

86

5.5.2 Methanol

The health hazards associated with methanol require regulation of the
concentration of methanol in these type of spirits. The concentration of methanol is
higher in the heads portion of the distillate, decreases in the hearts, and then increases
again in the tails requiring the distiller to make cuts of the distillate that will be low
enough in methanol concentration so that it will be able to be consumed.

Table 5.9 illustrates the average concentration of methanol in the distilled spirits
when the cuts are blended together. This type of analysis will give the distiller an idea of
where to make the cuts so that their distillate is consumable. The NTRL and Modified
UNIFAC models do the best at predicting the concentration of the methanol that is
present in the actual distillate.

Figures 5.9 through 5.12 show the experimental concentration of methanol
present in the distillate compared with the concentrations predicted by the CHEMCADTM
thermodynamic models NTRL, BSD, and Modified UNIFAC. The Modified UNIFAC
thermodynamic model, similar to the case with ethanol, is the best at modeling the

concentration found in the distillate.

87

Table 5.9

Methanol concentration of blended cuts for each distillation type and

the predicted values from CHEMCAD”.

 

Methanol Concentration (% v/v in 40% Ethanol)

 

 

Modified
10 L Fruit Batch Distillation Error NTRL ESD UNIFAC
All Cuts 0.416 0.010 0.350 0.209 0.354
Cuts 2-9 0.386 0.005 0.355 0.214 0.358
Cuts 3-9 0.371 0.004 0.360 0.222 0.361
10 L Standard
Batch
All Cuts 0.426 0.011 0.355 0.379 0.362
Cuts 2-9 0.402 0.005 0.361 0.384 0.366
Cuts 39 0.397 0.006 0.367 0.388 0.368
150 L Fruit Batch
All Cuts 0.377 0.024 0.372 0.392 0.385
Cuts 216 0.370 0.023 0.375 0.395 0.387
Cuts 3-16 0.364 0.022 0.375 0.393 0.381
Cuts 3-15 0.360 0.024 0.369 0.388 0.377
150 L Standard
Batch
All Cuts 0.372 0.014 0.314 0.334 0.328
Cuts 2-16 0.365 0.015 0.317 0.336 0.329
Cuts 3-16 0.363 0.014 0.317 0.334 0.324
Cuts 3-15 0.360 0.015 0.311 0.330 0.321

 

88

 

 

(170

(150:

(130*

Methanol Concentration (% vlv In 40% Ethanol)

 

OJCI'

:Ifiufoétiiatiin"
I —)l(-— NRTL

,' ESD I

1...... Modified UNIFAC l
' 20:95‘1/1'339‘Etiflfll

l
l

 

 

 

 

Figure 5.9

r T l m

75 150 225 300 375 450 525 600 675
Cumulative Distillate Volume (mL)

Methanol concentration of a 10 L fruit distillation normalized to a
40% ethanol solution. The three thermodynamic models shown are

the closest in approximating the actual data.

89

0.70

 

l +Standard Batch Distillation .
~ +NTRL '
'. ESD '

——t.—- Modified UNIFAC I
l __ : 2-2-_3§_‘%.ngyleti9.n_. r 1 -11

0.50 1

 

0.30

Methanol Concentratlon (% vlv In 40% Ethanol)

 

 

 

0.10 -- . . . . . L
0 75 1 50 225 300 375 450 525 600 675

Cumuatlve Distillate Volume (mL)

Figure 5.10 Methanol concentration of a 10 L standard batch distillation

normalized to a 40% ethanol solution.

90

  
    
        

 

(15
I—e— Fruit Distillation l
l—x—NRTL l
l ESD j _
I Modified UNIFACl /'
l~~_~--__0.§5:% Regulatigrj " I

(145 .

 

(14-

(135

 

(13

Methanol Concentratlon (% vlv In 40% Ethanol)

 

 

 

0.25 - l I I I
0 2000 4000 6000 8000 10000 12000

Cumulative Distillate Volume (mL)

Figure 5.11 Methanol concentration of a 150 L fruit distillation normalized to a

40% ethanol solution.

91

F’
or

 

 

 

 

 

 

E l l3—61516echl‘EiatErlpb—i61—1151iota7i
g I +NRTL '
g 0.45. l " “ESD
3 l -+-Modified UNIFAC I
5 l1,'_”0°35% Regulation—“n I
5 0.4
E
C
.2
,3; 0.35
C
O
0
5
3 0.3
C
U
'5
O
a

0.25 . . . . .

0 2000 4000 6000 8000 10000 12000

Cumulative Distillate Volume (mL)

Figure 5.12 Methanol concentration of a 150 L standard batch distillation

normalized to a 40% ethanol solution.

92

5.5.3 F use] Alcohols

Three fusel alcohols were studied in this experiment; t-butanol, l-propanol, and
isoamyl alcohol (3-methyl-l-butanol). 1-propanol and isoamyl alcohol represent the bulk

2’ 3' 4. T-butanol is

of the total fusel alcohol concentration present in these distilled spirits
higher in concentration in pomace fruit than stone fruits. In all cases, the three
thermodynamic models in CHEMCADTM that best model the concentration of the fusel
alcohols were the UNIFAC, Modified UNIFAC, and UNIFAC LLE models.

Figure 5.13 shows the profile of t—butanol present in these distilled spirits in small
scale control distillations. The profile of t-butanol is consistent with that of the fruit
distillation at the same scale. The concentration of t-butanol is higher than predicted in
the early cuts of the distillate, and lower than predicted in later cuts. The same trend is
seen in larger scale distillations. Figure 5.14 shows the distillate concentration of t-
butanol in terms of % v/v in 40% ethanol. The larger scale distillation does a better job
of approximating the concentration of the t—butanol than the smaller scale. The UNIFAC
LLE thermodynamic model is the best for mimicking the concentration profile of :-
butanol.

The next fusel alcohol studied is l-propanol. Figures 5.15 and 5.16 show the
concentration of l-propanol compared with the predicted values. The larger scale
distillation is better described although the concentration range is not correct when
compared to the actual values. A similar response can be seen for isoamyl alcohol in
figures 5.17 and 5.18.

The fusel alcohol concentration in distilled spirits is important because the quality

of the distilled spirits decreases after the maximum of the fuse] alcohol concentration is

93

reached. The undesired aroma associated with fusel alcohols increases in concentration
in the later portion of the distillation after the maximum concentration of l-propanol and
isoamyl alcohol is reached. Figure 5.19 shows the predicted CHEMCADTM values of the

additive concentration of l-propanol and isoamyl alcohol, as well as the actual distillate

concentration.

94

0.45

 

 

 

 

E , 4Standard—BatciiBistilla00d I

a ,

5 0'4 I +NTRL I

m o o

2.; 0.35 I +Modlfled UNIFAC I

v 1 o UNIFAC LLE I

c .77- __________‘__._._A_v*

E 0.3

>

it; 0.25

5 6\V'.“\

g 02 \N)

c

8 0.154

C

O

‘3 0.1 -

g -

«3 0.05 ‘Q

l- - ° .
o . e . . . . .

o 75 150 225 300 375 450 525 600 675

Cumulative Dlstlllate Volume (mL)

Figure 5.13 Standard 10 L batch distillation compared with predicted values from

CHEMCADTM for t-butanol.

95

T-Butanol Concentratlon (% vlv in 40% Ethanol)

(112

(108 -

(104~

 

 

 

 

F—o—T *F‘Fuit [Sisrtifiéiioh‘ 4
I+NRTL I
-~+— Modified UNIFAC I
I .
(W. t _ .. UNIFASLLLE __ I
l
l“
\l
T wk. -“‘:-0 I o _‘, . . 9 "'9" -A t
2000 4000 6000 8000 10000 12000

Cumulative Distillate Volume (mL)

Figure 5.14 Comparison of predicted values from CHEMCADTM compared with

the t-butanol concentration in fruit spirit distillate in 150 L

distillation.

96

'1" -_ .Q.

1.2

 

I -;F;uit Distiiation
i -0—UNIFAC I
, -—--r--Moditied UNIFAC I
l o- UNIFAC LLE I

1 -Propanol Concentration
(% vlv in 40% Ethanol)

 

    
 

 

 

.f-a'" «'0' ’- “*‘*.g-VMN"N9
’9’ "' _ _G:_— “has.“ ‘
z ’ Mgr: -9
0.8 ’0’ “*1:
/
/
’0’
0.6 . .
0 75 150 225 300 375 450 525 600 675

Cumulative Distillate Volume (mL)

Figure 5.15 The concentration of l-propanol present in 10 L fruit distillation

compared to predicted values.

97

 

 

“I; Erfiit'oiéifilatién ”I
_°_UNlFAC \
,--*- Modified UNIFACI

I} UNIFACLLE j

 
  

”'6-

.0
c»

 

0.6 4

1-Propanol Concentratlon
(% vlv in 40% Ethanol)

.0
is
a
\

 

 

0.2

 

0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)

Figure 5.16 The predicted values of l-propanol concentration and the
experimental distillate concentration of l-propanol in 150 L batch

distillations.

98

(1400

(1300

Isoamyl Alcohol Concentration
(% vlv in 40% Ethanol)
g:
N
8

(1100

Figure 5.17

 

 

q IIErMEtTIIaTibfi— I

l

.I I—o—UNIFAC 1
. her-Modified UNIFACI
\ .«rkw‘rg‘ , L: UNIFAC LLE I

 

 

 

 

T

75 150 225 300 375 450 525 600 675
Cumulative Distillate Volume (mL)

The 10 L distillate concentration of isoamyl alcohol and the values
predicted by CHEMCAD” using the UNIF AC thermodynamic

models.

99

 

L2

 

 

14.3.0: Disfinétion ,
I—o—UNIFAC i

1 A Fir—Modified UNIFAC.
. ,,°_, ENIIFACILLE_

F)
a:

Isoamyl Alcohol Concentration
(% vlv in 40% Ethanol)
;> c:
b O)

 

0 . I \‘o
J?"J’
02 h I. . ’
\ 2"”

 

 

0 2000 4000

6000 8000 10000 12000

Cumulative Distillate Volume (mL)

Figure 5.18 The 150 L batch concentration of isoamyl alcohol present in distilled

spirits compared with the predicted CHEMCAD“ values.

100

 

 

-0— Fruit Distillation

 

 

 

I i

J I
§§ I -o—-UNIFAC ,
' u 1.5 . 1 .
5 5 1 «iv—Modified UNIFAC I xe“~e
El.“ I i >8- ‘fl‘\
3 32 . <> UNIFAC LLE 3 ~ “x
.. 3 ._ -.- __.. m- _ a ___..l ,- “x. 0
5 .s 1’ » \-'\
'3 i 1 g‘ ’ i ,0” \
8 * \ I x
3.6 w“
e g a
'; g \ 2”“
g c 0.5 \ 9,2
58 ..,
<

o s . . .
0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)
Figure 5.19 The additive values of 1-propanol and isoamyl alcohol which

represent the bulk of the fusel alcohol concentration present in

distilled spirits.

101

'w v. (I 1*T‘J

5. 5.4 Aldehydes

The aldehydes studied in this experiment are acetaldehyde and benzaldehyde.
Acetaldehyde, with a low boiling point, is the first compound eluted during the
distillation. Acetaldehyde does not have positive aromatic characteristics associated with
it requiring the culling of the heads from the rest of the distillate. At the tails end of the
distillate, benzaldehyde has a positive aroma characteristic associated with it.
Benzaldehyde is the last of the compounds studied in this experiment to elute from the
distillation column. A high concentration of benzaldehyde can overpower the distillate,
often requiring increasing the volume of the tails fraction (i.e. taking the cut earlier) to
reduce influence of benzaldehyde on the hearts fraction of the distillate.

Acetaldehyde has a higher concentration in the earlier cuts than the hearts cuts.
This can be seen in figures 5.20, 5.21, and 5.22. The concentration predicted by the
CHEMCADTM thermodynamic models UNIFAC, Modified UNIFAC, and UNIFAC LLE,
are not as accurate in the first few cuts of the distillation than in the hearts and tails cuts.
The predictive power of these thermodynamic models is to determine where the rate of
acetaldehyde concentration change begins to level out. A smaller distillation does not
show the dramatic inflection point predicted by CHEMCADTM in the larger scale
distillation.

Benzaldehyde is not modeled well with the CHEMCADTM thermodynamic
models. The small scale distillations, an example of which can be seen in figure 5.23, do
not have similar inflection points in the concentration of the predicted values and the
actual values, although the concentration is approximately the same. An example of the
150 L scale distillation can be seen in figure 5 .24. The UNIFAC LLE thermodynamic

102

model predicts the volume of the concentration maximum for benzaldehyde, however,
the concentration of benzaldehyde is not accurately predicted.

Overall the benzaldehyde and acetaldehyde concentrations assist in determining
where the heads/hearts cut and hearts/tails cut should be made. The prediction models
within CHEMCADTM are more accurate in the larger scale distillation than the smaller

one for identifying the volumes where these cuts should be made.

103

0.400

0.300

(% vlv In 40% Ethanol)

0.100 ,

Acetaldehyde Concentratlon

0.000

Figure 5.20

 

0.200 .

 

I-O-Fibit Dist—ill—aiiSnF I

I-O-UNIFAC

I—+— Modified UNIFAC I

I o UNIFAC LLE

I

I

 

 

 

 

75 1 50 225 300

375

450 525 600
Cumulative Distillation Volume (mL)

Acetaldehyde concentration of 10L fruit batch distillation.

104

-.I

:.*“..f"'..

 

 

 

   
  
    

 

 

 

1s~ ,LL1L____W,-
I—O—Standard Batch I
I Distillation
=-O—UNIFAC

: I

3 A rte-Modified UNIFAC

“ _

g 3 1.2« I

3 g I .. UNIFAC LLE

1: ii I w A___.,______ ___ _g

8 a;

0 v

E. .5

33 E 0 6

8 32 '

0 V

0

<

0 . . . . -
0 100 200 300 400 500 600 700 800

Cumulative Distillate Volume (mL)

Figure 5.21 Acetaldehyde concentration of 10L standard

105

batch distillation.

rumu~4z "-: a

 

 

 

 

 

0.6 -
I+Fruit Disitllation I
I-O—UNIFAC .
”var-Modified UNIFAC

I: .:-<z-;UM:A9 9;: r '

.2

g g 0.4

I: a

8 5

: Ill

8 .\°

0

o v

E. .5

.C

8 S

3 °\o 0.2 .

0 V r

0

< \

\m.
o . I “T” ‘ m o . e . __.
0 2000 4000 6000 3000 10000 12000

Cumulative Distillate Volume (mL)

Figure 5.22 Acetaldehyde concentration of 150 L fruit batch distillation.

106

r-fim (*5 o -

.o .o
—L cut
[0 A

(% v/v in 40% Ethanol)
g o

Benzaldehyde Concentratlon

.0 .o
O O
N A

O

 

.0
O
on

 

-0—UNIFAC

. I ~+~Modified UNIFAC
o - UNIFAC LLE

-O-Standard BatchDistillation If

   
 

 

 

75

150

225

Cumulative Distillate Volume (mL)

300

107

375

450

Figure 5.23 Benzaldehyde in 10 L standard batch distillation.

525

600

675

FTT—flmr—t‘l

Wm. . M. as
I

03

 

   

I—O— Fruit Distillation
I-O— UNIFAC

I-+- Modified UNI PAC
.1: ,UNIEAQELE _ - I

' I
I
I
I
I

F3
A)

Benzaldehyde Concntration
(% vlv in 40% Ethanol)
.0

 

 

 

0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)

Figure 5.24 Benzaldehyde in 150 L fruit batch distillation. The benzaldehyde
concentration maxima is predicted by the CHEMCADTM
thermodynamic models at earlier distillate volumes than the
experimental distillation data shows. The concentration range of the
predicted model is lower than the data from the experimental

distillation.

108

5.5.5 Other Carbonyl Compounds

The three other carbonyl compounds studied include a ketone, acetone, and two
esters, ethyl formate, and ethyl acetate. The esters presented a problem with this study as
they were consistently at or below the limit of detection of the gas chromatograph. The
ester concentration was, as expected, greater in the first three cuts of the distillate (heads);
however, the concentration of these esters varied greatly within the multiple runs of the
same sample, due to the limit of detection for the chromatography equipment. The data
for the esters therefore is not reliable and is not included.

Acetone is the final compound included in this study. The concentration of
acetone decreases through the distillation. This trend is predicted by the CHEMCADTM
program, although the concentration predicted by the CHEMCADTM program will need to
be improved. A calibration curve may need to be developed to accurately predict the
concentration of acetone in the distilled spirits. Figures 5.25 and 5.26 illustrate the
concentration of acetone predicted by CHEMCADTM and the concentration of the actual

distillation.

109

 

 

 

 

I 9:01-15 f§t§nfié§d eater? I

II Distillation I

I+NRTL I

I I

1.5 _ I~+~Modified UNIFAC I
C c | I
:9.- 8 I o UNIFAC LLE I
2 O I
25 0 I._.____L__ _______.__._
m r
O V o
o .5
.65
3
.1 ii
0.5
. o T. \
9“ \‘
. o - . --o ____
0 75 1 50 225 300 375 450 525 600 675

Cumulative Distillate Volume (mL)

Figure 5.25 Predicted and experimental concentration of acetone in a 10 L

standard batch distillation.

llO

215

 

ILL 750? "566836?
I Batch DistillationI
.9 I +NR I
I-—ir-hflodfikfli
I UNIFAC
I 6 UNIFAC LLE

15-

Acetone Concentration
(% vlv in 40% Ethanol)

(15*

 

 

 

 

0 2000 4000 6000 8000 10000 12000
Cumulative Distillate Volume (mL)

Figure 5.26 Predicted and experimental concentration of acetone in a 150 L

standard batch distillation.

111

fit-fl .E-

 

 

5.6 Summary

The thermodynamic models within CHEMCADTM have varying ability to predict
the concentration of the compounds studied over time within the distillate. Table 5.10
shows the best fit model for each compound and the two thermodynamic models next
closest to fitting the experimental data as well. Overall the three UNIFAC distillation
models were best for identifying the distillation profile of the compounds studied by
inspection of the data. The BSD and NRTL models were accurate as predictive models
for the simple alcohols ethanol and methanol, but were poor comparison models for the
fusel alcohols. The Modified UNIFAC model was the best overall model for identifying

the concentration profile for the variety of compounds within this study.

 

The predicted concentration of the distillate does not always agree with the
experimental concentrations. This can clearly be seen in the case of benzaldehyde. This
is due to inaccurate feed values obtained from the Rayleigh distillations. The mass
balance for benzaldehyde holds for the predicted values when carried out to longer
distillation time, and it can be concluded that the concentration of the benzaldehyde in the
experimental mash was greater than the concentration entered as the feed concentration

of benzaldehyde in the CHEMCADTM model. Therefore the Rayleigh distillation method

does not work well for the prediction of the benzaldehyde concentration.

112

I J-l I

 

 

 

Table 5.10 The best fit thermodynamic models within CHEMCADTM for each of
the compounds studied in these experiments.
Thermodynamic Models
Compound Best Fit Next Best Fit
Ethanol Modified UNIFAC ESD NRTL
Methaol Modified UNIFAC ESD NRTL
T-Butanol UNIFAC LLE Modified UNIFAC NTRL
1-Propanol UNIFAC UNIFAC LLE Modified UNIFAC
Isoamyl Alcohol UNIFAC Modified UNIFAC UNIFAC LLE
Acetaldehye UNIFAC LLE Modified UNIFAC UNIFAC
Benzaldehyde UNIFAC LLE Modified UNIFAC UNIFAC
Acetone UNIFAC LLE NRTL Modified UNIFAC

 

 

 

113

 

5.7

Literature Cited

United States Code of Federal Regulations, Title 27—Alcohol, Tobacco Products
and Firearms, Chapter I, Part 5, Subpart C—Standards of Identity for Distilled
Spirits. <www.access.gpo.gov/nara.cfr/cfr-table-search.html>.

Tanner, H., and Brunner, H. R., Fruit Distillation Today 3’d ed., (English
translation), Heller Chemical and Administration Society, Germany, 1982.

Piggott, J. R., and Patterson, A. ed., Distilled Beverage Flavour: Recent
Developments. Ellis Horwood Ltd., 1989.

Guymon, J.F., Higher Alcohols in Beverage Brandy: Feasibility of Control of
Levels. Wines and Vines, 53, (1) 1972.

114

F.—_r¢.=-p= .1.“er

 

6. SUMMARY AND CONCLUSIONS

6.1 Process

The process of performing a Rayleigh distillation on a fermentation mash, and
then analyzing the distillate for congener compound concentration to determine the
approximate concentration of the con gener compounds present in the mash worked well
for both the standard solution and fermentation mash. Because this method was not
accurate at determining the exact concentration of the mash, the concentrations predicted
with the CHEMCADTM thermodynamic models were not exact. While the concentration
of the compounds were not always precise, the important factor in this process is the
ability of CHEMCADTM to mimic the trends of the compounds. By imitating the
concentration profile over the distillate volume, a distiller can determine where they need
to make fractions, in order to maximize the yield of potable spirits.

The low benzaldehyde concentration from the Rayleigh distillation is likely
responsible for the poor prediction of the concentration of the benzaldehyde in the
distillate when modeled in the CHEMCADTM program. A modification of the procedure

for determining the benzaldehyde content in the mash is warranted.

6.2 Fruit Distillation Modeling

Determining the flow rate of the distillate of an ethanol/water mixture is necessary

to determine the reflux ratio of the still so that the distillation modeling of the

CHEMCADTM program can be more accurate. When the initial modeling of the

115

'nu‘.

 

 

‘Fr_

distillation was done, the concentration of the compounds present was not complimentary
to the actual concentration or pattern of concentration over time. By adding the reflux
ratio data, and modeling the distillation as a multiple step process, the distillation
modeling by CHEMCADTM was more accurate.

The goal of this project was to determine if the CHEMCADTM program could be
used to improve the quality of distilled fruit spirits. With the process outlined in this
experiment, a distiller can analyze their fermentation mash using a Rayleigh distillation
and gas chromatograph. The resulting data, when input into the CHEMCADTM program
can predict the concentration of ethanol, methanol, and acetaldehyde in the distillate. The
rest of the compounds studied did not yield accurate prediction of concentration by
CHEMCADTM. The general profiles of these compounds, i.e. concentration maxima,
were determined using this method. This process allows the distiller to modify the
amount of distillate collected in the heads, hearts, and tails to improve the quality of the
distilled fruit spirits. Inaccurate predictions of compound concentrations is due to
incomplete recovery of these compounds from the Rayleigh distillation. The feed

concentrations for the CHEMCADTM models are incorrect resulting in the discrepancy.

6.3 Improving Cut Determination for Methanol

As shown in table 5.9, the predicted concentration of methanol in the spirits can
be used to determine which cuts should be blended together to meet the regulatory limit
of 0.35 % v/v methanol in fruit spirits. This will allow the distiller to make three cuts of

heads, hearts and tails, based on the volume of the hearts cut that would produce a

116

methanol concentration that would meet the federal guidelines. By doing this the distiller

would not be required to make many small fractional cuts.

6.4 Still Size Comparison

The component concentrations of the 10 L Holstein still were not as accurately
predicted as the concentrations of the 150 L industrial scale Christian Carl still
distillations. The 10 L still is not necessarily a good indicator of the efficiency of a
process and the results from the 10 L still to a 150 L industrial scale distillation do not

scale well.

6.5 Still Characteristics

Each still has unique characteristics regarding distillate flow rate and reflux ratio.
A procedure was developed to determine these characteristics for each still, to improve
the modeling of the distillations. The nuances each still possess that alter internal flow
parameters, and other parameters, cause differences in the distillation of each still. These
characteristics can be considered within this procedure, by determining the reflux ratio
and flow rate of a binary ethanol/water mixture of concentration equivalent to that of the
feed mash. The variation in distillation from one still to the next can be accounted for in

the thermodynamic models by using this method.

117

 

 

6.6 Conclusions "

1. Measuring the flow rate of the distillate from an 8% ethanol/water mixture can be
used to determine the reflux ratio for the operating steps involved in the modeling of
distilled spirits. This in turn can be used to account for subtle manufacturing differences
between one still and another within the thermodynamic models.

2. Rayleigh distillation can be used to determine the concentrations of the
compounds present in fruit mash. These values can then be used as the feed for the
distillation simulation software. However, the benzaldehyde concentration was not
accurately predicted using this method.

3. The modeling of concentration of the compounds present in distilled spirits is
possible using thermodynamic models available on the CHEMCADTM computer program.
The method for determining the feed values from a Rayleigh distillation of the feed mash
works for ethanol, methanol, and acetaldehyde. Improvements need to be made to more
accurately predict concentration values for the other compounds.

4. The process does not scale-up well from the 10 L Holstein still to the 150L
Christian Carl still. The 150 L scale is more accuratly modeled by the computer
simulations than the 10 L scale.

5. By analyzing the concentration of methanol in each distillation cut, the
concentration of methanol in the distilled spirits can be approximated using the above
method and "blending" the above cuts. This will allow the distiller to use fewer fractions
based on the results of the CHEMCADTM predictive models.

6. The ethanol concentration predicted by this procedure and the thermodynamic

models is a very good approximation of the experimental results. It is important to

118

remember that the concentration of ethanol in the distilled spirits is the most important
factor to the distiller. This process of extracting ethanol, in the form of fruit brandy, from
the fermentation mash is the goal of these distillations. This method approximates the
concentration of the ethanol better than any of the other compounds. In the end, the
distiller is more concerned with the concentration of ethanol than any of the other

compounds present.

119

 

 

8. FUTURE WORK

The relationship between the predicted values and the concentration of the fruit
brandies for most of the compounds can be studied further and calibration data can be
generated to improve the concentration data prediction for all the components present.
Improvement in the compound recovery from the Rayleigh distillation should improve
the concentration prediction of the CHEMCADTM models.

A more in depth examination of the flow rate and reflux ratio of the distillation
may also improve this process. In these experiments data was collected in increments of
750 mL on the 150 L still. Smaller volume increments will lead to more information
about the behavior of the flow rate of the distillate, and the reflux ratio in the still.
Measurement of the reflux ratio and flow rate for industrial stills should improve the
ability to model these distillations.

Experimental VLE measurements would increase the ability to model these
distillations. Collecting VLE data for multicomponent mixtures using these batch stills
will improve the understanding of the process involved in the distillation of fruit
brandies.

Finally, as other thennodynamic models become available, they should be
studied, in an attempt to better mimic the concentration of the components present in

these type of spirits.

120

APPENDICIES

Appendix A. Distillation Models
Appendix B. Thermodynamic Models
Appendix C. Models Used

Appendix D. Data Tables

Literature Cited

121

122

125

129

138

151

 

 

 

A. DISTILLATION MODELS
A.1 Batch Distillation Mathematical Model and Describing Equations

Batch distillation is inherently an unsteady state process in which the transient
behavior is caused by an imbalance in the inlet and outlet streams over the entire period
of operationl. In most applications, the column is brought to steady state at total reflux
before product withdrawal begins. It is usually not of interest to follow the behavior
while total reflux steady state is being established, since no product is being withdrawn.
Therefore, the conditions representing the solution of the total reflux equations based
upon a specified initial charge and column parameters are taken as the initial conditions
for the first operating step.

The basic unsteady state describing equations employed in the CHEMCAD
program are essentially the same as those given by Distefano (1968). The main
differences are in terms introduced to handle feeds, side products, heat losses and side
heaters, and additional equations for the side product accumulators. According to the
CHEMCAD literature, these equations consist of the rigorous unsteady state mass and
enthalpy balances and the phase equilibrium relationships. Equilibrium stages with
constant volume, mass or molar liquid holdup were assumed. Vapor holdup was
neglected and the column hydraulics were not included as part of the simulation.
Thermodynamics are completely rigorous, employing the entire selection of K-value,
enthalpy, and property methods in CHEMCAD.

The describing equations presented here are based on the constant volume holdup

assumption rather than the less realistic constant mole or mass.

122

 

‘u
I

b.
I.
’4
i
1

Component Mass Balance

d H X.
iii)— : fin + Ln-an—l —(n +WlJr)Xin _(Vn +an )Yin +Vn+iYn+l

Total Mass Balance

d(H..)
dt

 

: Fri +Ln-1_Ln —Wbi —Vn -WW1 +V"+I
Enthalpy Balance

d H H
ij—t—Jiz : HFnFn + HLn-an—l _ HI): (Ln +Wbr)—an(vn +an)+I-1Vn+l‘/n+l +Qn

Phase Equilibrium

Constitutive

20¢. —1)X,, = 0

Holdup Equations

_ ann
" M1,.(Xn)

A.2 Physical Property Models

Specification of the flow rates, compositions, enthalpies and locations of the feed
streams, the locations and molar flow rates of the side product streams, the internal stage
heat duties, the column pressure profile and the volume holdups are required for solving
the rigorous equations. With this information, at each time interval the composition,

temperature, flow rate, and holdup profiles are solved for by these equations.

123

A series of operations steps define the operation of a batch distillation, with
specified topping criteria. For this work we change the reflux ratio at each operation step
for example. The CHEMCAD software utilizes these operation steps to accommodate
these changing parameters for the model equations].

The describing equations that must be solved over each operation step consist of
the unsteady state mass and enthalpy balances and the phase equilibrium relationships. In
this work, equilibrium stages with constant volume, mass or molar liquid holdup were
assumed. Vapor holdup was neglected and the column hydraulics were not included'.

The unsteady state mass and enthalpy balances form a large set of coupled,

nonlinear ordinary differential equations'.

124

 

 

 

B. THERMODYNAMIC MODELS

B.l Guidelines to Selecting Thermodynamic Methods

These guidelines are outlined in the help function of the CHEMCAD program.
CHEMCAD provides an extensive array of the most up-to-date methods for performing
heat and material balances. These techniques cover applications ranging from straight
hydrocarbon applications, to chemical models, to a wide variety of special applications
involving electrolytes, salt effects, amines, sour, water, to other specialty chemical
applications. These methods have been field tested over a period of years and have been
demonstrated to give highly accurate results.

To achieve accurate results; however, it is necessary to select the proper method
for a given application. This section of the manual provides guidelines for making these
selections. It should be noted, however, that the guidelines we will be discussing are for

the selection of K-value (phase equilibrium) and enthalpy methods only.

B.2 The K-Value Wizard

The K—value wizard function is designed to recommend a thermodynamic model
for the components and application used in the CHEMCAD program. To determine the
best fit model, the following criteria are used:

1. First, what general type of model is required (i.e. equation-of—state, activity
model, etc) is determined from the component list.
2. Next, the temperature and pressure ranges input by the user is used to determine

which equation within a given category is best at the limits of those ranges.

125

 

3. If the model is an activity model, the program then looks at the BIP database to I»
see which model has the most data sets for the current problem. It then calculates the
fractional completeness of the BIP matrix. If that fraction is greater than the BIP

threshold parameter, it uses the chosen activity method; if not it uses UNIFAC.

8.3 Conclusions and Recommendations for Thermo Models

CHEMCAD classifies liquids into five categories: ideal, regular, polar (highly
non-ideal), electrolytes, and special. For these classifications the following distinctions
and qualifications are made:

1. Ideal Solutions are systems where the vapor phase behaves essentially as an ideal
gas (low pressure) and all the molecules in the liquid phase are virtually the same size. In
addition, no intermolecular forces of attraction are assumed to exist. Vapor—liquid
equilibria are determined using Raoult's law:

P.
K.- =—V'
P

where VP,- is calculated using the selected vapor pressure equation.
2. Regular Solutions are systems where the non-idealities stem from moderate
physical interactions, i.e., from differences in the size and shape of the molecules.
Intermolecular associations are assumed to be minimal. These systems are best modeled
using equations-of-state such as PR, SRK, APIS, BWRS, CS/GS and MSRK.

In all of these cases, both the vapor and the liquid phases are assumed for form
regular (i.e., mildly non-ideal) solutions, and K-values are calculated like so:

_q)li
" <I>

vi

126

 

3. Polar (Highly Non-Ideal Solutions) are systems where the liquid phase non- [I
idealities arise predominantly from molecular associations. These systems must be
modeled using activity coefficient methods which generally require BIPs for accuracy.

The vapor phase is taken to be a regular solution, therefore,

K‘ : Yifl:
PCD vi
where
fuo = VP.- and <1)...- = l;0r

 

fr

The equations which fall into this category are: NRTL, UNIFAC, UNIQUAC, Wilson,
T.K.Wilson, I-Iiranuma, Van Laar, Margules, and GMAC. Wilson, NRTL, and
UNIQUAC are the recommended methods. UNIFAC may be used where data is absent.
This is the type of solution used in this work.

4. Electrolyte Solutions are not utilized for these experiments.

 

5. Special Systems are provided for the simulation of common applications which
do not lend themselves to the above approaches. Therefore no information is included.

A summary of the recommendations of the K-values can be seen below:

Hydrocarbons

0 Soave-Redlich-Kwong High - moderate P & T’s.

- API Soave General HC. High - moderate P & T's.

- Peng-Robinson High - moderate P & T's.

- Benedict-Webb-Ruben-Starling High - moderate P & T's.
- Grayson-Streed Moderate pressures and temperatures.

- Maxwell-Bonnell K-charts Low pressure, heavy material.

- ESD Hydrocarbon -water; hydrocarbon-gases

- SAFI‘ Hydrocarbon -water; hydrocarbon-gases

Chemicals

- UNIFAC T 2 275K - 475K; P = 0-4 atm.; two liquid phases. Non-ideal; group contribution;
predictive.

- Wilson Highly non-ideal.

127

Vapor Pressure Ideal solutions.

NRTL Highly non-ideal and 2 liquid phases.

UNIQUAC Highly non-ideal and 2 liquid phases.

Margules Highly non-ideal and 2 liquid phases. (4 suffix)

T. K. Wilson Highly non-ideal and 2 liquid phases.

Hiranuma (HRNM) Highly non-ideal and 2 liquid phases.

Regular Solution Moderately non-ideal (Predictive).

Van Laar Moderately non-ideal.

Modified SRK (4 parameter) Polar compounds in regular solutions.

Predictive SRK Polar compounds in non-ideal solutions. Better than UNIFAC at high pressures.
0 Wilson Salt Non-ideal solutions with salts dissolved in them.

Special Techniques

0 Henry's Gas Law Gases dissolved in water

0 Amine (MEA DEA) Gas sweetening

0 Sour Water Acid gases and NH3 dissolved in H2O

- K Tables User K's

0 Polynomial User K's

- User-Added Subroutine User K's

° TSRK Methanol system; particularly with light gases.

- PPAQ General, but electrolyte systems is most common application
- TEG Dehydration of hydrocarbon streams using tri-ethylene-glycol
- FLOR Flory-Huggins method for polymers

UPLM UNIFAC for polymers

ACTXUser specified activity coefficients

ESD Hydrogen bonding; hydrogen bonding at high pressure
SAFI‘ Hydrogen bonding; hydrogen bonding at high pressure

128

 

C. MODELS USED

C.1 Renon NRTL Model

The NTRL model is recommended for highly ideal solutions and 2 liquid phases.
The NT RL method was the model recommended by the K-value wizard by CHEMCAD.
The NTRL method can use many different parameters to determine the separation of the
components in distillates. The UNIQUAC and NTRL models are widely used for both
vapor-liquid and liquid-liquid equilibria.
Because of the non-ideal nature of the solutions used in this work (i.e. polar alcohols,
variety of size, hydrogen bonding of alcohols and water, etc.) the NRTL method was
chosen as the principle model for performing the calculationsl’ 2.
Liquid phase activity coefficients are calculated by NRTL equation.

The NRTL equations has the following form':

 

 

Where

7;.- =AII.+BII/T+CII*ln(T)+DII*T
GII. =exp(—ajI *TII)
aij :aji

T =Temperature in Kelvin

The most common usage of the NRTL equation is the three parameter equation.

The NRTL equation may be used either as a three-parameter (Bji, Bi}, and (113' only), as a

129

five-parameter (Aji, AU, Bji, Bib and aij), seven parameter (Aji, AU, Bjj, By, C ji, C,-,-, and 00,-), k-
or a nine parameter equation (A,-,-, Aij, Bji, BU, ij, Cg, Dji, Di}, and org).

Converting Binary Parameters (Bips) from Literature
The CHEMCAD help manual states that many data sources use the DECHEMA equation
whereas CHEMCAD divides the actual Bips by RT. To use Bips from a source using the
alternate format, divide by R (1.97842) for the proper value and divide by T to identify
the proper Bip. For example:

Using data from DECHEMA, if A12 = A +B *T, then CHEMCAD bips are Bij=
A/RandAij =B/R'.

Regressing Bips
According to the CHEMCAD program it is possible to use the Tools menu command Bip
regression to regress Bips from data. This model is dependant upon the binary

interaction parameters of each of the components'.

C.2 The ESD Equation

The ESD equation (Elliott, Suresh, and Donohue, 1990) is similar to conventional
cubic equations of state like the Soave (1972) equation or Peng-Robinson (1976)
equation for hydrocarbons and gases, but provides accuracy competitive with
UNIQUAC, NRTL, or Wilson’s equation for hydrogen bonding mixtures. It also
provides high accuracy for hydrogen bonding mixtures at high pressure like the
interpolation methods of Dahl et al. (1991), Schwartzentruber and Renon (1989), and
Wong et al. (1992). It also provides accuracy for hydrocarbon 4» water mixtures that is

comparable with that of the adaptation of the Soave equation by Kabadi and Danner

130

(1986). The ESD equation is based on treating the hydrogen bonding interactions as I“
chemical reactions in accordance with the formalism developed by Wertheim (1986)
unlike the interpolation methods. The ESD equation uses the critical properties and
acentric factor to estimate the pure-component parameters when no parameters are listed
in the databank. Parameters estimated in this way are based on neglecting self-
association. For associating compounds, the ESD equation still matches the experimental
critical temperature, but adjusts the size and shape parameters to obtain an optimal
representation of vapor pressure data. If the accuracy is insufficient for any particular
application, it should be possible by limiting the range of conditions to develop specific
pure component parameters that provide sufficient accuracy. The accuracy of VLE
correlations by the ESD equation has been studied for a wide range of mixtures by Puhala
and Elliott (1993). The treatment of the association thermodynamics applied here is
restricted to linear Acceptor-donor association like alcohols and water and binary
associations like carboxylic acids. This restriction permits an increase in computational
efficiency of about 100-fold relative to the completely general formulation (Elliott,
1996). Evaluations to date for mixtures containing ethers, esters, and ketones, as well as
alcohols, aldehydes, amines, and glycols have shown little or no reduction in accuracy as
a result of this restrictionl.

The use of the ESD methods for this work is due to the polar mixture containing
alcohol, water, aldehydes, ketones and esters. The hydrocarbon + water base model for
the ESD equations is different from that of the polar solution models of UNIFAC,
UNIQUAC and NRTL.

Summary of Equations1

131

RT 1-1-977 1+ 1.7745(77Y)

 

P_V =1+ 4<C77) 9‘5<an> + 20.5506

where,

0.5506
Z _

8.405506 /RT
p __

8p
A306 = ZxINdI [21n(x;‘ )+ 1 - XIAI+ ZxINdIIln(XIC)+1_2XICI

77 E prJ).

(C77) 5 pZZ(Cb)ij xix]
(WW) 5 pzz(qb)ij Yijxlxi
q, a 1+1.90476*(cI -1)

YII. a exp(£I.I lkT) - 1.0617
(Cb)11 E (chl + chI)/2
(6119).. E (qibi +qlb1)/2

en. 5(8 e )“2(l—kII)

ii ii

(WY) 5 pzblYixl

 

 

XiA = the fraction of linear proton acceptor sites not bonded
XIC = the fraction of binary (carboxylic) bonding sites not bonded
The procedure for determining the fractions of bonding is given by Elliot (1996).

b, ci, and tau/k are adjustable pure component parameters.

C.3 UNIFAC

In the UNIFAC K model, the liquid phase activity coefficients for each species
are calculated from the UNIFAC group contribution method. The limitations of

UNIFAC are temperature range from 275K to 425 K and pressure up to a few

132

 

atmospheres. Also this UNIFAC reads the group contribution parameter stored in the

VLE databasel.

Because UNIFAC is a Gibbs excess model it is theoretically capable of predicting
a second liquid phase. In practice liquid-liquid behavior cannot be adequately predicted

from VLE data. The user is recommended to use UNIFAC LLE to model a two phase

system with the UNIFAC modelsl.

UNIFAC Equationl’2

in y, = in yIC + in n”

(combinatorial) (residual )

where
6.

(I). z (I).
In .C=ln——‘-+— .ln-4+I.——'— xi
7. 261. ch . X.- Z}: , ,

x.

l i

and

my,” =qI1—ln BTI
[Z J j)_ 222:1}!
2

 

where

133

 

 

Ii=—:'(ri"q.-)-r,--I) 2:10

6 _ qixi

.--——
2'ij
,-
fix,-

¢.- =—
erxj
,-

T uji _un'
~ =€X ‘-
J‘ p RT

’3' = 2‘0:an
k

 

qi = Evian
k
R; : Vaw’
15.17
Aavi
Q‘ _ 2.5X109

ln yf = Zvl‘)(1n 11 — In 175”)
k

 

ln 1“,. = Q1. 1—1n(20mwm)—Z E3“

6 = Qme
m ZQan

Umn —Unn amn
Wm" =CXp —T ZCXP --T—

T =temperature in Kelvin

C.4 UNIFAC LLE

As the UNIFAC model is a Gibbs excess energy model, it can theoretically
predict Liquid-Liquid Equilibria (LLE). In practice, binary interaction parameters

regressed from VLE data are insufficient to predict LLE behavior].

134

1" ”"““'_:l

CHEMCAD provides an alternate UNIFAC model which uses binary interaction
parameters regressed from LLE data. The UNIFAC VLE model equations are used
without change. Group interaction parameter values differ from the UNIFAC VLE
model].

It was necessary to develop additional subgroups for l-propanol (P1), 2-propanol
(P2), diethylene glycol (DEOH), trichloroethylene (TCE), methylformamide (MFA), and
tetramethylenesulfone (T MS). These additional subgroups are only for the UNIFAC
LLE model. The LLE subgroups use regressed values of Q and R which differ from

UNIFAC VLE subgroups‘.

C.5 Modified UNIFAC (Dortmund)

The Modified UNIFAC model (Dortmund) introduces temperature dependant
interaction parameters, allowing for more reliable description of phase behavior as a
function of temperature. This method also uses Van der Waals (Q and R) properties

which are slightly different than those used in the original UNIFAC method.

 

Original UNIFAC: ‘11,,” = CXP[-;ml

 

— +b T+ T2
Modified UNIFAC (Do.) \an zexp[ am m; Cm ]

The modified UNIFAC (Do.) model uses different component group and
subgroup matrices than the original UNIFAC. Published interaction parameters have
been included in CHEMCAD. Unpublished parameters can be added into CHEMCAD if

your company has access to theml.

135

 

As with the original, the Modified UNIFAC model has the activity coefficient as

the sum of a combinatorial and residual part].
In y, = ln yf + In y,”

The combinatorial term in Modified UNIFAC has changed, to allow it to deal

with compounds of very different sizes'.

. . V. V.
lnyic =1—V,. +anl. -5-q,. l-—'-+ln —‘
F.- F.-

Where V'; is calculated from relative Van der Waals volumes Rk of different groups.

C6 UNIQUAC Model

In the UNIQUAC K model, the liquid phase activity coefficients for each species
are calculated by the UNIQUAC equation. CHEMCAD supports up to 8 parameters (Aij,
Aji, Uij—Uji, Uji-Uij, Cij, Cji, Dij, Dji) for the UNIQUAC model].

UNIQUAC Equation1

¢i Z 6i q)'. N N N 6-‘T-l-
lny‘. =ln—x—+§-q, -ln-¢—+l‘. ——-ij -lj —q,. ln[26i ~Tflj+ql —q,. 2 J J
. . 1

X j 2::(6k 'Tki)

i l t I

Where

(bi = xi*r.-/(Zx,-*rj)

6i = x;*qz/(ij*q,-)

1;,- = exp [AU- - (Uij - Ujj) / RT + CU*ln(T) + D,~j-*T]

136

. '

 

‘1’

Temperature in degrees Kelvin
(z/2)*(rr-qr~)-rr+1

57‘“!

N
II II II II

10 (coordination number)

q,- van der Waals area parameter (A,,.,- / (2.5 * 10") where A,,.,- is the van der
Waals area)

r,- = van der Waals volume parameter (V,,.,~/ 15.17 where Vw; is the van der

Waals volume)

For comparison to DECHEMA format values ail- and by:

at] + bij/T = 140' + (Uij - Ufl) / RT
bij = ' (Uij ' Uii)/R
aij = Aij

The UNIQUAC binary interaction parameters Aij, (Ui j - Uj j) and (Uji - Uii) are in
cal/gmol.

The binary interaction parameters (BiPs) Cij and Dij are optional.

Several binary interaction parameters (BiPs) are in the UNIQUAC program. The
Therrnophysical menu command Edit Bips can be used to view existing Bips for the
system, if the K Value model is UNIQUAC.

The UNIQUAC equation uses qi, the van der Waals area parameter, and n, the van der
Waals volume parameter. Values for q,- and r,- are required for each component in the
mixture.

Values of q,- and r.- stored in the databank were computed for use with the UNIQUAC

model.

137

rug-mus (1:43“ I
.

 

D. DATA TABLES

Table D.1 Data for Figure 6.1. The CHEMCAD program predicted ethanol

concentrations for constant reflux ratio distillations.

Reflux Ratio 2

Cumulative Ethanol/
Distillate Modified UNIFAC Water
Volume (mL) NTRL BW RS UNIFAC UNIFAC LLE UNIQUAC ESD Distillation

750 78.22 89.74 79.61 79.61 76.31 78.16 78.16 78.39
1500 77.32 90.84 78.87 78.87 75.60 77.28 77.28 82.24
2250 76.31 90.79 78.03 78.03 74.81 76.29 76.29 82.47
3000 75.18 90.71 77.06 77.06 73.91 75.18 75.18 82.49
3750 73.94 90.55 75.96 75.96 72.90 73.97 73.97 80.98
4500 72.57 90.28 74.72 74.72 71 .77 72.63 72.63 80.03
5250 71.09 89.85 73.33 73.33 70.51 71.17 71.17 80.43
6000 69.48 89.19 71 .80 71 .80 69.12 69.59 69.59 78.27
6750 67.75 88.28 70.12 70.12 67.59 67.89 67.89 77.50
7500 65.90 86.66 68.28 68.28 65.92 66.05 66.05 77.00
8250 63.92 85.06 66.29 66.29 64.12 64.09 64.09 74.74
9000 61.81 83.18 64.13 64.13 62.16 62.00 62.00 73.67
9750 59.57 80.54 61.81 61.81 60.07 59.78 59.78 71.13

10500 57.21 76.06 59.33 59.33 57.84 57.43 57.43 66.18
1 1250 54.74 70.78 56.68 56.68 55.47 54.96 54.96 63.64
12000 52.15 64.65 53.89 53.89 52.97 52.37 52.37 60.45
Reflux Ratio 5
Cumulative Ethanol/
Distillate Modified UNIFAC Water
Volume (mL) NTRL BW RS UNIFAC UNIFAC LLE UNIQUAC ESD Distillation

750 84.52 91 .30 84.99 85.25 82.38 84.62 82.18 78.39
1500 84.25 91 .29 84.76 85.07 82.1 1 84.35 81 .89 82.24
2250 83.96 91 .28 84.51 84.87 81.80 84.05 81 .57 82.47
3000 83.61 91.27 84.21 84.63 81.45 83.70 81 .20 82.49
3750 83.21 91 .27 83.87 84.36 81.05 83.30 80.77 80.98
4500 82.74 91 .26 83.47 84.04 80.59 82.83 80.28 80.03
5250 82.19 91 .25 83.00 83.65 80.05 82.26 79.71 80.43
6000 81.51 91 .22 82.42 83.19 79.42 81 .57 79.02 78.27
6750 80.65 91.16 81.71 82.61 78.66 80.70 78.18 77.50
7500 79.52 91 .02 80.80 81 .88 77.74 79.57 77.13 77.00
8250 78.00 90.75 79.58 80.95 76.59 78.05 75.80 74.74
9000 75.93 90.21 77.91 79.70 75.12 76.02 74.05 73.67
9750 73.23 89.22 75.58 77.93 73.20 73.38 71 .79 71 .13

10500 69.87 87.72 72.43 75.27 70.65 70.07 68.93 66.18
1 1250 65.88 83.35 68.44 71.38 67.37 66.11 65.48 63.64
12000 61.30 76.45 63.62 66.26 63.33 61 .55 61 .45 60.45

138

f'fi“ it K 's ..‘I’

 

Table D.2

Experimental and predicted results for ethanol in fruit spirits and

standard batch distillations.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 L Fruit DistiTlation
Mental Distillation CHEMCAD Model Distillation
Cumulative
Distillate Volume Standard Modified
4mm % Ethanol (vlv) Deviation NRT L ESD UNIFAC UNIFAC UNIFAC LL§ UNIQUAC
75 72.96 0.33 75.55 75.55 78.75 77.18 73.92 75.55
150 75.33 0.40 78.00 78.00 $.51 79.27 76.29 78.00
225 75.76 0.30 76.60 78.60 79.41 78.05 74.98 76.64
300 74.91 0.88 74.35 74.35 77.81 76.17 73.08 74.44
375 74.25 023 70.30 70.30 74.97 72.56 69.73 70.43
450 72.69 0.51 88.41 66.41 71.65 68.82 88.29 66.58
525 69.83 0.20 62.09 62.09 87.38 64.49 62.27 6228
800 85.12 0.41 58.67 58.67 63.71 60.97 59.03 58.87
675 59.16 0.97 55.74 55.74 60.39 57.89 56.26 55.96
150 L Fruit Distillation
Exarimental Distillation CHEMCAD Model Distillation
Cumulative
Distillate Volume Standard Modified
(mL) % EM (vlv) Devi_a_tion NRT L E§D UNIFAC UNIFAC UleFAC LLE UNQUAC
75 71 .94 2.14 73.99 72.10 75.82 74.22 70.44 73.88
150 74.08 1.63 76.76 75.21 78.1 1 76.83 73.63 76.65
225 74.09 1.19 75.82 74.54 77.84 76.47 73.34 75.7
300 73.63 1.34 73.83 72.61 76.79 75.23 72.14 73.85
375 71 .95 2.27 69.98 68.75 74.43 72.12 69.35 70.08
450 69.00 1 .56 66.22 65.00 71 .45 68.66 66.21 66.3
525 86.16 0.97 61.98 60.80 67.39 84.52 62.37 62.1
600 62.98 1.73 58.60 57.50 63.87 81.13 59.25 58.81
675 58.15 0.681 55.71 J5_4.72 60.65 58.15 56.56 55.
1o L PM!”
EMmental Distillation CHEMCAD Model Distillation
Cumulative
Distillate Volume Standard Modified
(mL) % Ethanol (vlv) Deviation NRTL ESD UNIFAC UN_I_FAC UNIFAC LLE UNIQUAC
750 77.15 1 .25 79.66 78.40 81.33 , 80.06 77.26 79.54
1500 82.52 1.28 82.38 80.84 83.45 62.49 80.56 82.10
2250 82.22 0.32 82.55 81.34 84.00 83.04 81.15 82.39
3$O 82.35 1.26 81.49 80.44 83.17 82.12 79.83 81.46
3750 80.58 0.53 81.34 80.22 82.98 81.90 79.56 81.34
4500 81.29 2.68 $.44 7927 82.06 80.94 78.39 60.48
5250 $.16 1.20 $.11 78.86 81.68 80.55 77.93 80.16
6000 79.14 1.12 78.35 77.09 80.14 78.96 76.08 78.42
6750 78.04 1 .27 77.62 76.32 79.49 78.40 75.43 77.72
7500 77.97 1.1 1 75.09 73.85 77.85 76.56 73.48 7528
8250 75.47 2.22 73.64 72.40 76.65 75.50 72.47 73.89
9000 73.70 2.28 89.85 88.88 74.19 72.38 69.72 70.18
9750 71 .67 1 .87 87.43 68.33 72.32 70.12 67.85 67.76
10500 67.48 1.59 64.01 63.03 69.20 66.77 64.97 64.36
11250 84.21 0.67 60.87 60.03 65.89 63.53 62.19 8121
12000 61.07 1.28 60.49 59.78 65.01 62.94 61.96 60.62
150 L Standard Batch Distillation
Exaimental Distillation CHEMCAD Model Distillation
Cumulative
Distillate Volume Standard Modified
(mL) 96 final (vlv) W NRTL E_SD UNIFAC UNIF UNIFAC LL§ UNIQUAC ‘
750 70.25 4.38 75.97 73.77 75.77 74.35 70.61 75.84
1500 77.69 0.77 77.16 74.42 76.03 75.63 72.92 76.63
2250 78.55 1.41 79.68 78.18 80.37 79.87 78.00 79.35
3000 78.59 1.17 80.51 79.57 81.98 81.04 79.01 80.47
3750 78.41 0.90 80.86 79.88 82.28 81.26 79.21 80.83
4500 79.27 We $.40 79.31 81.81 $.61 78.32 $.37
5250 78.79 0.49 80.25 79.07 81.61 80.38 77.97 80.20
6000 78.96 0.68 78.60 77.46 $.24 78.93 76.16 78.55
6750 77.01 0.13 77.89 76.76 79.67 78.47 75.56 77.86
7500 76.88 1 .16 75.29 74.33 77.93 76.77 73.67 75.41
8250 75.20 1.99 73.83 72.89 77.00 75.82 72.77 74.03
9$0 72.72 1.96 70.04 69.17 74.65 72.82 70.15 70.34
9750 70.12 2.48 67.63 66.82 72.88 70.66 68.37 67.97
10500 67.50 2.22 64.23 63.50 69.85 67.37 65.56 64.60
11250 64.96 1.02 61.09 60.48 66.60 64.17 62.82 61.48
12000 61.26 2.24 60.73 60.23 65.79 63.62 62.63 61.11

 

 

 

 

Table D.3 Congener concentration experimental and predicted values for 10 L

fruit distillations.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 L Fruit Distillation
Congener Concentration 1% v/v in 40% Ethanol)
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.656 0.383 1.26E-01 0.054 0.701 0.075 0.313
150 0.491 0.167 1.41 E-01 0.032 0.81 1 0.084 0.324
225 0.362 0.1 19 1.42E-01 0.022 0.938 0.077 0.284
300 0.353 0.069 1 .48E-01 0.016 0.951 0.075 0.267
375 0.356 0.048 1.49E-01 0.013 1.005 0.072 0.278
450 0.369 0.028 1 .47E-01 0.010 1 .035 0.052 0.287
525 0.384 0.008 1 .39E-01 0.007 1 .057 0.045 0.299
600 0.388 0.003 1 .31 E-01 0.005 1 .002 0.000 0.278
675 0.389 0.007 1 .29E-01 0.003 0.920 0.000 0.188
Standard Deviation of Congener Concentration
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.048 0.014 7.24E-04 0.056 0.007 0.000 0.010
150 0.012 0.005 1.97E-04 0.016 0.005 0.001 0.006
225 0.003 0.003 4.47E-04 0.002 0.005 0.001 0.010
300 0.006 0.005 1 .64E-03 0.006 0.018 0.001 0.023
375 0.002 0.001 1 .825-04 0.018 0.006 0.000 0.005
450 0.004 0.001 6.64E-04 0.005 0.009 0.000 0.008
525 0.003 0.000 5.70E-04 0.001 0.004 0.000 0.003
600 0.003 0.001 4.28E-04 0.001 0.005 0.000 0.004
675 0.009 0.001 6.97E-04 0.003 0.013 0.000 0.004
NRTL
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.310 0.213 4.86E-17 0.049 1.444 0.039 0.154
150 0.318 0.176 1 .35E-18 0.043 1.331 0.037 0.088
225 0.322 0.126 2.14E-20 0.032 1 .265 0.031 0.148
300 0.331 0.090 2.13E-19 0.024 1.170 0.026 0.220
375 0.344 0.065 1 .75E-20 0.018 1 .062 0.022 0.266
450 0.358 0.048 1 .8415-18 0.014 0.962 0.019 0.277
525 0.373 0.035 7.25E-19 0.01 1 0.868 0.017 0.280
600 0.387 0.026 2.40E-18 0.009 0.780 0.014 0.278
675 0.403 0.01 8 1 .95E-1 9 0.007 0.697 0.012 0.273
ESD
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.164 0.1 13 2.57E-17 0.026 0.765 0.020 0.082
150 0.163 0.090 6.90E-19 0.022 0.683 0.019 0.045
225 0.168 0.066 1 .12E-20 0.017 0.661 0.016 0.077
300 0.178 0.048 1 .15E-19 0.013 0.629 0.014 0.119
375 0.196 0.037 9.98E-21 0.010 0.604 0.013 0.152
450 0.216 0.029 1 .1 1 E-18 0.009 0.579 0.012 0.167
525 0.240 0.023 4.67E-1 9 0.007 0.559 0.01 1 0.180
600 0.264 0.017 1 .64E-18 0.006 0.532 0.010 0.189
675 0.289 0.013 1 .40E-19 0.005 0.500 0.009 0.196

 

 

140

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 L Fruit Distillation {continued}
UNIFAC

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol

75 0.230 0.238 5.58E-1 7 0.048 0.765 0.063 0.245

150 0.221 0.189 2.53E-17 0.043 0.700 0.052 0.179

225 0.240 0.120 4.77E-19 0.031 0.757 0.036 0.230

300 0.263 0.077 1 .45E-18 0.023 0.807 0.025 0.279

375 0.293 0.051 4.88E-19 0.017 0.844 0.017 0.309

450 0.318 0.035 5.10E-25 0.013 0.849 0.012 0.300

525 0.342 0.024 9.545-18 0.010 0.844 0.009 0.281

600 0.364 0.017 1 .08E-17 0.008 0.835 0.006 0.261

675 0.388 0.012 1 .49E-19 0.006 0.823 0.004 0.239

Modified UNIFAC

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol

75 0.323 0.222 3.155-17 0.040 0.884 0.074 0.386

150 0.339 0.183 2.55E-18 0.037 0.827 0.057 0.313

225 0.337 0.122 1 .14E-19 0.029 0.861 0.036 0.337

300 0.338 0.083 2.335-18 0.023 0.878 0.023 0.334

375 0.345 0.058 3.37E-26 0.019 0.876 0.015 0.308

450 0.356 0.042 3.61 E18 0.016 0.861 0.010 0.275

525 0.369 0.030 1 .17E-22 0.013 0.843 0.007 0.245

600 0.383 0.022 6.41 E-20 0.01 1 0.823 0.004 0.216

675 0.398 0.015 6.57E-18 0.009 0.803 0.003 0.188

UNIFAC LLE

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol

75 0.320 0.288 2.985-17 0.068 0.880 0.073 0.385

150 0.324 0.210 3.05E-18 0.053 0.827 0.057 0.316

225 0.331 0.125 3.92E-18 0.034 0.859 0.038 0.339

300 0.341 0.076 6.51 E18 0.022 0.877 0.025 0.338

375 0.354 0.047 5.851520 0.014 0.880 0.017 0.317

450 0.367 0.030 4.68E-18 0.010 0.868 0.012 0.286

525 0.381 0.019 1 .36E-18 0.007 0.852 0.008 0.255

600 0.395 0.012 1 .435-24 0.005 0.834 0.005 0.226

675 0.409 0.008 7.75E-18 0.003 0.815 0.004 0.199

UNIQUAC

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol

75 0.31 1 0.225 1 .18E-16 0.045 1 .433 0.037 0.209

150 0.320 0.183 1 .30E-19 0.041 1.346 0.035 0.139

225 0.324 0.1 26 3.7651 9 0.031 1 .252 0.030 0.200

300 0.333 0.088 7.53E-19 0.024 1 .143 0.028 0.253

375 0.346 0.062 7.88E-19 0.019 1 .034 0.022 0.275

450 0.359 0.044 1 .30E-17 0.015 0.939 0.019 0.276

525 0.373 0.032 9.70E-21 0.012 0.851 0.017 0.270

600 0.388 0.023 1 .36E-18 0.010 0.770 0.015 0.263

675 0.403 0.016 1 .39E-18 0.008 0.692 0.013 0.254

 

 

141

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table D.4 Congener concentration experimental and predicted values for 10 L
standard batch distillations.
10 L §tandard Batgh Distillation
Congener Concentration (% v/v in 40% Ethanol)
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Promol T-Butanol Alcohol
75 0.616 1.710 0.140 2.296 0.905 0.405 1.102
150 0.438 1 .002 0.159 1 .341 1 .053 0.326 1 .466
225 0.319 0.668 0.173 0.966 1 .094 0.181 1 .550
300 0.331 0.398 0.182 0.777 1 .072 0.086 1 .389
375 0.354 0.257 0.182 0.637 1 .050 0.066 1 .150
450 0.383 0.131 0.174 0.494 1 .002 0.048 0.844
525 0.417 0.051 0.162 0.373 0.951 0.000 0.601
600 0.452 0.009 0.156 0.281 0.910 0.000 0.445
675 0.520 0.000 0.153 0.187 0.865 0.000 0.306
Standard Deviation of Congener Concentration
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.052 0.140 0.004 0.041 0.022 0.013 0.022
150 0.002 0.050 0.003 0.024 0.027 0.006 0.047
225 0.002 0.018 0.007 0.019 0.037 0.005 0.108
300 0.002 0.025 0.002 0.028 0.01 1 0.015 0.030
375 0.007 0.007 0.001 0.015 0.006 0.016 0.029
450 0.017 0.001 0.002 0.007 0.007 0.001 0.031
525 0.010 0.006 0.001 0.013 0.01 1 0.000 0.004
600 0.003 0.008 0.002 0.016 0.016 0.000 0.029
675 0.001 0.000 0.001 0.006 0.01 1 0.000 0.012
NTRL
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.313 0.954 3.33E-07 0.827 0.747 0.1 18 0.310
150 0.321 0.790 2.19E-07 0.732 0.689 0.1 12 0.179
225 0.326 0.564 2.90E-07 0.538 0.655 0.094 0.301
300 0.336 0.404 4.23E-07 0.403 0.606 0.079 0.451
375 0.351 0.292 7.15E-07 0.308 0.550 0.067 0.552
450 0.365 0.215 1 .06E-06 0.242 0.499 0.058 0.576
525 0.380 0.1 58 1 .55E-06 0.191 0.450 0.051 0.581
600 0.396 0.1 15 1.99E-06 0.150 0.405 0.044 0.577
675 0.412 0.083 2.41 E-06 0.1 17 0.362 0.038 0.568
ESD
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.345 1 .522 3.50E-05 0.693 0.427 0.012 0.001
150 0.354 0.990 2.12E-05 0.647 0.415 0.010 0.001
225 0.357 0.530 3.06E-05 0.509 0.422 0.012 0.001
300 0.364 0.289 4.83E-05 0.408 0.424 0.016 0.002
375 0.375 0.160 8.67E-05 0.332 0.423 0.019 0.004
450 0.386 0.091 1 .32E-04 0.277 0.419 0.022 0.006
525 0.399 0.051 1 .95E-04 0.232 0.416 0.025 0.008
800 0.41 1 0.028 2.52E-04 0.194 0.412 0.027 0.01 1
675 0.424 0.016 3.06E-04 0.162 0.407 0.030 0.013

 

 

142

 

 

10 L Standard Batch Distillation (continued)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

UNIFAC
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.233 1 .083 0.1 37 0.820 0.390 0.195 0.485
150 0.224 0.866 0.109 0.735 0.356 0.160 0.354
225 0.244 0.546 0.103 0.531 0.388 0.112 0.463
300 0.267 0.351 0.091 0.389 0.415 0.078 0.573
375 0.298 0.229 0.074 0.287 0.436 0.054 0.647
450 0.324 0.1 58 0.058 0.221 0.440 0.038 0.632
525 0.348 0.1 10 0.045 0.173 0.438 0.027 0.594
600 0.370 0.077 0.035 0.135 0.433 0.01 9 0.551
675 0.394 0.053 0.027 0.105 0.427 0.01 3 0.506
Modified UNIFAC
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.332 1.012 0.141 0.673 0.454 0.230 0.788
150 0.349 0.838 0.1 1 1 0.634 0.423 0.178 0.630
225 0.345 0.556 0.106 0.497 0.443 0.114 0.695
300 0.345 0.377 0.092 0.395 0.454 0.073 0.700
375 0.352 0.262 0.074 0.31 9 0.454 0.047 0.650
450 0.363 0.1 89 0.059 0.266 0.446 0.031 0.583
525 0.376 0.137 0.047 0.223 0.437 0.021 0.518
600 0.390 0.099 0.037 0.186 0.427 0.014 0.457
675 0.405 0.070 0.028 0.155 0.416 0.009 0.399
UNIFAC LLE
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.327 1 .322 0.094 1 .173 0.450 0.226 0.780
150 0.332 0.969 0.060 0.921 0.422 0.179 0.635
225 0.338 0.575 0.083 0.586 0.441 0.1 19 0.697
300 0.348 0.347 0.096 0.378 0.453 0.079 0.710
375 0.361 0.214 0.093 0.249 0.456 0.053 0.672
450 0.374 0.136 0.081 0.169 0.451 0.036 0.608
525 0.388 0.087 0.069 0.1 16 0.442 0.025 0.543
600 0.402 0.056 0.058 0.079 0.433 0.01 7 0.482
675 0.416 0.035 0.048 0.053 0.423 0.01 1 0.424
UNIQUAC
Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
75 0.319 1.007 1.52E-04 0.760 0.741 0.1 14 0.422
150 0.327 0.822 6.85E-05 0.691 0.696 0.108 0.281
225 0.331 0.566 1 .28E-04 0.524 0.648 0.092 0.407
300 0.340 0.394 2.77E-04 0.404 0.593 0.079 0.521
375 0.353 0.278 6.68E-04 0.31 9 0.536 0.068 0.572
450 0.366 0.200 1 .15E-03 0.257 0.487 0.059 0.574
525 0.381 0.144 1 .77E-03 0.208 0.442 0.052 0.564
600 0.395 0.1 03 2.34E-03 0.1 68 0.400 0.045 0.548
675 0.41 1 0.073 2.90E-03 0.135 0.360 0.039 0.529

 

143

 

Table D.5 Congener concentration experimental and predicted values for 150 L

 

 

 

 

 

 

 

 

fruit distillations.
150 L Fruit Distillation
Congener Concentration (% vlv in 40% Ethanol)

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.493 5.263 0.064 1 .983 0.446 0.704 0.159
1500 0.452 0.567 0.062 0.861 0.477 0.197 0.172
2250 0.443 0.261 0.069 0.464 0.532 0.062 0.215
3000 0.430 0.152 0.080 0.289 0.599 0.012 0.281
3750 0.386 0.092 0.097 0.209 0.677 0.009 0.370
4500 0.365 0.060 0.1 14 0.165 0.737 0.005 0.459
5250 0.356 0.037 0.138 0.137 0.803 0.004 0.573
6000 0.352 0.020 0.164 0.1 15 0.849 0.003 0.885
6750 0.336 0.009 0.198 0.095 0.894 0.000 0.821
7500 0.324 0.002 0.236 0.081 0.918 0.000 0.948
8250 0.335 0.003 0.275 0.072 0.912 0.000 1.023
9000 0.332 0.000 0.296 0.062 0.875 0.000 0.995
9750 0.339 0.000 0.289 0.055 0.793 0.000 0.754
10500 0.333 0.000 0.232 0.049 0.715 0.000 0.443
11250 0.349 0.000 0.183 0.042 0.658 0.000 0.275
12000 0.413 0.000 0.161 0.035 0.625 0.000 0.203

Standard Deviation of Congener Concentration

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.031 0.872 0.002 0.080 0.009 0.056 0.01 1
1500 0.037 0.102 0.001 0.049 0.012 0.024 0.007
2250 0.010 0.072 0.001 0.056 0.001 0.012 0.005
3000 0.049 0.036 0.003 0.041 0.004 0.01 1 0.014
3750 0.036 0.016 0.001 0.015 0.01 1 0.002 0.004
4500 0.054 0.013 0.000 0.014 0.017 0.002 0.010
5250 0.033 0.006 0.001 0.008 0.007 0.001 0.004
8000 0.026 0.006 0.001 0.008 0.003 0.000 0.002
6750 0.022 0.004 0.002 0.005 0.015 0.000 0.015
7500 0.019 0.003 0.003 0.005 0.015 0.000 0.024
8250 0.023 0.000 0.002 0.006 0.019 0.000 0.019
9000 0.016 0.000 0.003 0.004 0.019 0.000 0.020
9750 0.01 1 0.000 0.004 0.005 0.018 0.000 0.030
10500 0.007 0.000 0.010 0.004 0.019 0.000 0.040
11250 0.003 0.000 0.010 0.004 0.016 0.000 0.030
12000 0.006 0.000 0.006 0.003 0.017 0.000 0.018

 

 

 

144

 

 

1§0 L Fruit Distillgtim lfifinufl)

 

 

 

 

 

 

 

NRTL

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.322 0.262 5.72E-08 0.089 1 .1 14 0.046 0.059
1500 0.378 0.249 3.19E-08 0.134 0.767 0.054 0.015
2250 0.375 0.1 86 3.31 E-08 0.078 0.780 0.046 0.018
3000 0.346 0.126 4.37E-08 0.033 0.941 0.034 0.037
3750 0.348 0.093 4.61 E-08 0.023 0.937 0.029 0.045
4500 0.339 0.064 5.60E-08 0.014 0.981 0.023 0.075
5250 0.341 0.046 6.04E-08 0.010 0.953 0.020 0.096
6000 0.339 0.031 8265-08 0.007 0.950 0.016 0.199
6750 0.345 0.022 9.33E-08 0.005 0.887 0.013 0.281
7500 0.354 0.015 1 .35E-07 0.004 0.817 0.01 1 0.449
8250 0.366 0.01 1 1 .645-07 0.003 0.738 0.009 0.533
9000 0.383 0.007 2.51 E-07 0.002 0.656 0.008 0.633
9750 0.400 0.005 3.125-07 0.002 0.583 0.007 0.647
10500 0.419 0.004 4.09E-07 0.001 0.513 0.006 0.648
1 1250 0.439 0.003 5.08E-07 0.001 0.449 0.005 0.632
12000 0.459 0.002 5.06E-07 0.001 0.389 0.004 0.606

ESD

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.348 0.453 5.025-06 0.072 0.639 0.002 0.000
1500 0.415 0.426 2.45E-06 0.115 0.521 0.001 0.000
2250 0.410 0.161 2.57E-06 0.077 0.532 0.002 0.000
3000 0.373 0.049 3.67E-06 0.037 0.604 0.002 0.000
3750 0.373 0.024 3.96E-06 0.029 0.615 0.002 0.000
4500 0.364 0.010 5.09E-06 0.020 0.650 0.003 0.000
5250 0.366 0.005 5655-08 0.015 0.660 0.003 0.000
6000 0.363 0.002 8.43E-06 0.01 1 0.693 0.004 0.001
6750 0.368 0.001 9.87E-06 0.009 0.698 0.004 0.001
7500 0.374 0.001 1 .56E-05 0.007 0.709 0.006 0.001
8250 0.384 0.000 1 .96E-05 0.005 0.706 0.007 0.002
9000 0.397 0.000 3.16E-05 0.004 0.703 0.009 0.003
9750 0.410 0.000 4.02E-05 0.004 0.697 0.010 0.004
10500 0.425 0.000 5.41 E-05 0.003 0.689 0.01 1 0.005
11250 0.441 0.000 6.82E-05 0.002 0.680 0.012 0.006
12000 0.456 0.000 6.82E-05 0.002 0.671 0.013 0.008

 

14S

 

 

150 L Fruit Distillation (continued)

 

 

 

 

 

 

 

 

UNIFAC

Cumulative Distillate Isoamyl

Volume (ml) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.198 0.332 0.040 0.089 0.502 0.078 0.173
1500 0.169 0.437 0.019 0.132 0.342 0.077 0.068
2250 0.175 0.185 0.020 0.077 0.360 0.055 0.076
3000 0.197 0.057 0.029 0.032 0.456 0.035 0.131
3750 0.205 0.033 0.030 0.022 0.480 0.025 0.149
4500 0.223 0.017 0.034 0.014 0.546 0.017 0.209
5250 0.234 0.010 0.033 0.010 0.575 0.012 0.241
6000 0.260 0.006 0.035 0.006 0.662 0.007 0.361
6750 0.276 0.004 0.031 0.005 0.692 0.005 0.412
7500 0.308 0.002 0.027 0.003 0.759 0.003 0.541
8250 0.332 0.001 0.022 0.002 0.783 0.002 0.582
9000 0.374 0.001 0.017 0.002 0.818 0.002 0.633
9750 0.407 0.001 0.013 0.001 0.821 0.001 0.606
10500 0.445 0.000 0.009 0.001 0.816 0.001 0.560
1 1250 0.482 0.000 0.007 0.001 0.803 0.000 0.505
12000 0.518 0.000 0.005 0.001 0.786 0.000 0.448

Modified UNIFAC

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.353 0.304 0.038 0.070 0.620 0.096 0.365
1500 0.464 0.427 0.015 0.107 0.432 0.097 0.142
2250 0.455 0.200 0.016 0.074 0.451 0.058 0.157
3000 0.390 0.067 0.027 0.038 0.561 0.028 0.268
3750 0.385 0.041 0.028 0.029 0.582 0.018 0.296
4500 0.365 0.023 0.035 0.020 0.646 0.010 0.391
5250 0.364 0.015 0.035 0.016 0.668 0.006 0.424
6000 0.348 0.009 0.039 0.01 1 0.740 0.004 0.552
6750 0.350 0.006 0.035 0.009 0.755 0.002 0.563
7500 0.347 0.004 0.030 0.007 0.789 0.001 0.594
8250 0.354 0.003 0.024 0.006 0.788 0.001 0.551
9000 0.364 0.002 0.019 0.004 0.783 0.001 0.495
9750 0.377 0.001 0.014 0.004 0.765 0.000 0.432
10500 0.394 0.001 0.01 1 0.003 0.745 0.000 0.372
11250 0.413 0.001 0.008 0.002 0.723 0.000 0.316
12000 0.433 0.000 0.006 0.002 0.698 0.000 0.265

 

146

 

Table D.6

Congener concentration experimental and predicted values for 150 L

standard batch distillations.

 

 

 

 

 

 

 

 

 

150 L Standard Batch Distillation
Congener Concentration (% vlv in 40% Ethanol)

Cumulative Distillate isoamyl

Volume (mg Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.473 2.291 0.126 2.475 0.799 2.373 0.635
1500 0.396 0.768 0.146 1 .473 1 .038 1 .230 1.144
2250 0.377 0.610 0.142 1 .230 1 .014 0.792 1 .077
3000 0.356 0.381 0.145 0.91 1 1 .007 0.399 1 .042
3750 0.345 0.257 0.149 0.750 1 .003 0.245 1.035
4500 0.348 0.190 0.153 0.643 1.014 0.158 1.042
5250 0.336 0.159 0.155 0.517 0.986 0.020 0.969
6000 0.346 0.133 0.152 0.482 0.958 0.006 0.862
6750 0.339 0.077 0.141 0.335 0.922 0.000 0.780
7500 0.355 0.057 0.148 0.343 0.883 0.000 0.651
8250 0.360 0.025 0.142 0.262 0.828 0.000 0.509
9000 0.368 0.003 0.135 0.205 0.758 0.000 0.368
9750 0.376 0.000 0.129 0.159 0.701 0.000 0.275
10500 0.380 0.000 0.126 0.124 0.650 0.000 0.210
11250 0.388 0.000 0.123 0.088 0.598 0.000 0.155
12000 0.401 0.000 0.122 0.064 0.535 0.000 0.110

Error

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.013 0.394 0.094 0.040 0.198 0.309 0.168
1500 0.020 0.069 0.077 0.126 0.053 0.088 0.001
2250 0.018 0.008 0.077 0.067 0.065 0.039 0.006
3000 0.021 0.029 0.077 0.022 0.081 0.034 0.038
3750 0.021 0.015 0.077 0.000 0.071 0.005 0.026

4500 n/a n/a n/a n/a n/a n/a n/a

5250 0.017 0.005 0.079 0.087 0.047 0.001 0.042
6000 0.024 0.024 0.077 0.010 0.056 0.001 0.010
6750 0.015 0.027 0.094 0.186 0.055 0.001 0.029
7500 0.024 0.015 0.081 0.006 0.064 0.000 0.001
8250 0.010 0.000 0.083 0.022 0.083 0.000 0.01 1
9000 0.012 0.000 0.084 0.001 0.078 0.000 0.006
9750 0.010 0.000 0.088 0.000 0.091 0.000 0.009
10500 0.01 1 0.000 0.091 0.001 0.081 0.000 0.004
11250 0.014 0.000 0.094 0.004 0.058 0.000 0.017
12000 0.001 0.000 0.098 0.006 0.075 0.000 0.008

 

147

 

 

150 L Standard Batch Distillation (continued)

 

 

 

 

 

 

 

 

NRTL

Cumulative Distillate Isoamyl
Volume (mg Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.270 1.185 7.07E-07 1.103 0.697 0.430 0.071
1500 0.320 1 .189 4.09E-07 1 .718 0.488 0.522 0.019
2250 0.313 0.856 4.17E-07 0.992 0.493 0.438 0.022
3000 0.286 0.560 5.37E-07 0.408 0.591 0.317 0.046
3750 0.286 0.409 5.63E-07 0.284 0.589 0.271 0.056
4500 0.281 0.281 6.75E-07 0.178 0.617 0.217 0.094
5250 0.284 0.201 7.25E-07 0.1 28 0.600 0.1 85 0.120
6000 0.285 0.134 9.79E-07 0.084 0.599 0.146 0.248
6750 0.291 0.096 1 .1 0E-06 0.063 0.561 0.124 0.324
7500 0.301 0.065 1.57E-06 0.045 0.518 0.102 0.556
8250 0.312 0.047 1 .88E-06 0.034 0.469 0.086 0.661
9000 0.328 0.033 2.86E-06 0.026 0.417 0.072 0.790
9750 0.342 0.023 3.56E-06 0.020 0.371 0.062 0.809
10500 0.359 0.016 4.69506 0.015 0.327 0.052 0.810

1 1250 0.378 0.01 1 5.83E-06 0.012 0.287 0.044 0.792
12000 0.393 0.008 5.79E-06 0.009 0.249 0.037 0.761

ESD

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.300 2.077 6.04E-05 0.905 0.401 0.022 0.000
1500 0.364 2.041 3.08E-05 1.482 0.328 0.014 0.000
2250 0.353 0.762 3.15E-05 0.980 0.334 0.014 0.000
3000 0.318 0.228 4.42E-05 0.468 0.380 0.019 0.000
3750 0.317 0.110 4.77E-05 0.357 0.387 0.020 0.000
4500 0.309 0.049 6.1 1 E-05 0.245 0.409 0.025 0.000
5250 0.311 0.025 6.77E-05 0.191 0.415 0.027 0.001
6000 0.307 0.01 1 1 .00E-04 0.135 0.436 0.036 0.001
6750 0.312 0.006 1.17E-04 0.108 0.439 0.041 0.001
7500 0.317 0.003 1 .84E-04 0.083 0.445 0.054 0.002
8250 0.325 0.002 2.30E-04 0.068 0.444 0.062 0.002
9000 0.337 0.001 3.70E-04 0.054 0.442 0.077 0.004
9750 0.348 0.000 4.71 E-04 0.045 0.438 0.088 0.005
10500 0.360 0.000 6.33E-04 0.037 0.433 0.100 0.006

1 1250 0.374 0.000 7.98E-04 0.030 0.428 0.1 12 0.008
12000 0.386 0.000 7.97E-04 0.024 0.422 0.119 0.008

 

148

 

 

 

150 L Standard Batch Distillation (Qntinued)

 

 

 

 

 

UNIFAC

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.167 1 .538 0.443 1 .120 0.308 0.739 0.203
1500 0.142 2.108 0.207 1.715 0.21 1 0.737 0.081
2250 0.147 0.880 0.227 0.989 0.223 0.538 0.092
3000 0.168 0.270 0.340 0.414 0.284 0.345 0.160
3750 0.1 72 0.1 55 0.352 0.286 0.299 0.250 0.183
4500 0.187 0.081 0.409 0.178 0.341 0.169 0.257
5250 0.196 0.049 0.405 0.127 0.359 0.120 0.296
6000 0.219 0.027 0.442 0.081 0.413 0.078 0.445
8750 0.232 0.017 0.402 0.060 0.432 0.055 0.508
7500 0.258 0.01 1 0.366 0.042 0.475 0.036 0.671
8250 0.278 0.007 0.302 0.031 0.490 0.025 0.728
9000 0.313 0.004 0.238 0.022 0.514 0.016 0.798
9750 0.340 0.003 0.181 0.017 0.516 0.01 1 0.767
10500 0.372 0.002 0.134 0.013 0.513 0.007 0.710

1 1250 0.404 0.001 0.098 0.010 0.505 0.005 0.842
12000 0.433 0.001 0.070 0.007 0.495 0.003 0.569

Modified UNIFAC

Cumulative Distillate Isoamyl
Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.306 1.411 0.418 0.883 0.382 0.919 0.433
1500 0.402 2.047 0.162 1.366 0.267 0.947 0.171
2250 0.391 0.941 0.181 0.941 0.281 0.574 0.193
3000 0.334 0.314 0.305 0.480 0.350 0.289 0.332
3750 0.330 0.193 0.328 0.370 0.364 0.184 0.367
4500 0.312 0.106 0.407 0.258 0.404 0.108 0.486
5250 0.310 0.069 0.416 0.203 0.418 0.068 0.529
6000 0.297 0.040 0.484 0.144 0.464 0.040 0.691
6750 0.298 0.027 0.447 0.1 1 5 0.474 0.025 0.71 0
7500 0.295 0.017 0.410 0.087 0.496 0.015 0.756
8250 0.300 0.012 0.335 0.071 0.496 0.010 0.704
9000 0.307 0.008 0.261 0.056 0.494 0.006 0.635
9750 0.31 8 0.006 0.200 0.046 0.483 0.004 0.556
10500 0.332 0.004 0.150 0.038 0.471 0.002 0.479

1 1250 0.348 0.003 0.11 1 0.031 0.458 0.001 0.408
12000 0.384 0.002 0.080 0.025 0.441 0.001 0.342

 

149

 

 

 

 

 

 

 

 

 

 

 

150 L §tandard Batch Distillation (cgntinued)
UNIFAC LLE

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.273 1 .831 0.129 1 .617 0.383 0.886 0.442
1500 0.294 2.143 0.029 2.1 16 0.271 0.932 0.182
2250 _ 0.293 0.869 0.035 0.936 0.286 0.594 0.207
3000 0.285 0.273 0.083 0.31 1 0.354 0.316 0.351
3750 0.287 0.147 0.100 0.1 82 0.367 0.210 0.385
4500 0.289 0.073 0.165 0.097 0.406 0.129 0.499
5250 0.293 0.042 0.199 0.059 0.41 8 0.085 0.537
6000 0.297 0.022 0.350 0.033 0.460 0.052 0.683
6750 0.303 0.013 0.406 0.021 0.469 0.034 0.698
7500 0.312 0.007 0.562 0.013 0.491 0.022 0.748
8250 0.321 0.005 0.577 0.008 0.493 0.014 0.707
9000 0.335 0.003 0.573 0.005 0.495 0.009 0.654
9750 0.348 0.002 0.491 0.004 0.486 0.008 0.578
10500 0.363 0.001 0.406 0.002 0.476 0.004 0.502
1 1250 0.379 0.001 0.328 0.002 0.463 0.002 0.431
12000 0.394 0.000 0.259 0.001 0.449 0.002 0.364

UNIQUAC

Cumulative Distillate Isoamyl

Volume (mL) Methanol Acetaldehyde Benzaldehyde Acetone 1-Propanol T-Butanol Alcohol
750 0.276 1 .299 1 .37E-04 1 .004 0.741 0.402 0.130
1500 0.337 1 .480 4.67E-05 1 .809 0.569 0.455 0.037
2250 0.329 0.929 5.09E-05 0.992 0.568 0.397 0.043
3000 0.296 0.486 8.94E-05 0.435 0.635 0.312 0.087
3750 0.295 0.330 1 .02E-04 0.316 0.619 0.274 0.103
4500 0.288 0.206 1 .55E-04 0.206 0.618 0.228 0.185
5250 0.290 0.140 1 .86E-04 0.154 0.589 0.1 98 0.202
6000 0.289 0.088 3.55E-04 0.104 0.582 0.162 0.367
6750 0.294 0.061 4.66E-04 0.080 0.520 0.140 0.447
7500 0.302 0.040 9.88E-04 0.059 0.472 0.1 16 0.650
8250 0.31 1 0.028 1 .46E-03 0.047 0.426 0.100 0.713
9000 0.325 0.01 9 3.07E-03 0.036 0.380 0.084 0.773
9750 0.339 0.013 4.39E-03 0.029 0.340 0.072 0.762
10500 0.354 0.009 6.51 E-03 0.023 0.302 0.062 0.740

1 1250 0.371 0.006 8.76E-03 0.01 8 0.268 0.052 0.708
12000 0.387 0.004 9.12E-03 0.014 0.235 0.044 0.669

 

 

150

 

 

LITERATURE CITED

Chemstations CHEMCADTM Process Simulation Software. Internal Help
Function and Tutorial. 2002.

Reid, R.C., Prausnitz, J .M., and Sherwood, T.K., The properties of Gases and
Liquids, Third Edition. McGraw-Hill, New York, 1977.

Wankat, P.C., Equilibrium Staged Separations. Prentice Hall, New Jersey, 1998.

151

 

lliliiiiiiiiiiiiiiil