Mass Transfer and Process Cont

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 74

MASS TRANSFER AND PROCESS CONTROL IN RANDOMLY PACKED COLUMN

A CASE STUDY ON CO2 ABSORPTION WTH NaOH

A THESIS

Presented to the Department of Chemical Engineering

California State University, Long Beach

In Partial Fulfillment

of the Requirements for the Degree

Master of Science in Engineering

Committee Members

Larry Jang, Ph.D. (Chair)


Sergio Mendez, Ph.D.
Gregory Smith, Ph.D.

College Designee:

Antonella Sciortino, Ph.D.

By Srinivas Gavini

B.Tech., Jawaharlal Nehru Technological University, Hyderabad, India

January 2017




ProQuest Number: 10251697




All rights reserved

INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.






ProQuest 10251697

Published by ProQuest LLC (2017 ). Copyright of the Dissertation is held by the Author.


All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.


ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
ABSTRACT

MASS TRANSFER AND PROCESS CONTROL IN RANDOMLY PACKED COLUMN

A CASE STUDY ON CO2 ABSORPTION WTH NaOH

By

Srinivas Gavini

January 2017

This thesis studies the mass transfer performance and presents a unified approach in

process modeling of a randomly packed column useful for controlling industrial scale CO2

emissions. A case study was conducted on the carbon dioxide absorption with aqueous sodium

hydroxide in the packed column. The case study conducted helped in evaluating the mass

transfer by using the method of transfer units, which involves the estimation of the height of

transfer unit, number of transfer units, and overall gas-phase mass transfer coefficients. The

process model developed was an automatic online control and uses a feedback Proportional-

Integral-Derivative (PID) Controller. Internal Model Control (IMC) and Ziegler-Nichols (Z-N)

tuning methods were used to tune the PID controller based on the process dynamic model and

model parameters obtained from the step tests. The overall gas-phase mass transfer coefficient

(K a) values obtained from the experimental data are comparable to the published data in the

literature.

ii
ACKNOWLEDGEMENTS

First, I would like to thank Department of Chemical Engineering at the California State

University, Long Beach for accepting me into the graduate program. I would like to thank my

advisor, and department chair Dr. Larry Jang, who diligently advised me in research with

countless conversations and set up a great model as an advisor for me to follow. I particularly

appreciate his encouragement, patience, support, and understanding when I have financial

troubles. I am grateful to have Dr. Roger C. Lo, Dr. Sergio Mendez and Dr. Bikau Shukla in my

Department whose advices helped me to succeed in completion of thesis. The questions and

suggestions from them enriched the thesis in a great extent.

I am thankful to the lab technician Mr. Minh Tran and facility services Mario who helped

me to install the computers and plumbing work in the chemical engineering laboratory. My class

mate Raja Sekhar Kalavacherla, I would like thank him for his value addition in the means of

clarifying chemical engineering concepts. I would like to mention my friend Sai Pradeep. Potluri

who was with me since the day one and encouraged me towards the success and completion of

thesis.

Finally, I would like to thank my family. My parents, DurgaPrasad Gavini and

Annapurna Gavini who were always supportive to me and provided me an environment allowing

me to pursue graduate study. My sister, Anusha Gavini, has a major place in my success, who

always guided and mentored me towards my goals. I appreciate what she has done for the me.

iii
TABLE OF CONTENTS

ABSTRACT .................................................................................................................................... ii

ACKNOWLEDGEMENTS ........................................................................................................... iii

LIST OF TABLES .......................................................................................................................... v

LIST OF FIGURES ....................................................................................................................... vi

LIST OF SYMBOLS ................................................................................................................... viii

LIST OF ABBREVIATIONS ........................................................................................................ ix

1. INTRODUCTION ................................................................................................................ 1

2. LITERATURE REVIEW ..................................................................................................... 6

3. METHODOLOGY ............................................................................................................... 8

4. AUTOMATION AND INSTRUMENTATION................................................................. 19

5. EXPERIMENTAL WORK................................................................................................. 37

6. RESULTS AND DISCUSSIONS ....................................................................................... 41

7. CONCLUSIONS AND FUTURE WORK ......................................................................... 51

APPENDICES .............................................................................................................................. 53

A. EXPERIMENTAL DATA OF AIR AND CO2 SUPPLIED TO THE PROCESS .............. 54

B. EXPERIMENTAL DATA OF ZIEGLER-NICHOLS TUNING METHOD....................... 57

REFERENCES ............................................................................................................................. 60

iv
LIST OF TABLES

1. PID Controller Settings Using IMC Tuning ............................................................................. 15

2. Ziegler–Nichols PID Controller Settings .................................................................................. 18

3. Dakota Instruments Mass Flow Controller specifications ........................................................ 24

4. Design data on the Randomly Packed Column......................................................................... 38

5. Hatta number in the Literature and the Experimental data ....................................................... 42

6. Results for the Experimental runs of both the 0.003 M and 0.01 M NaOH solutions.............. 44

7. Typical KGa values from Literature data and Experiment conducted ...................................... 45

8. Process Model Data for the Open-Loop System obtained from LOOP-PRO .......................... 48

9. IMC tuning for PID Controller Settings ................................................................................... 48

10 . Ziegler–Nichols tuning for PID Controller Settings .............................................................. 49

v
LIST OF FIGURES

1. Packed column gas phase and liquid phase distribution. .................................................... 10

2. General example of a feedback control loop. ..................................................................... 13

3. A PID controller in a feedback loop. .................................................................................. 14

4. Ziegler–Nichols closed-loop tuning. ................................................................................... 16

5. A data acquisition unit. ....................................................................................................... 19

6. The FieldPoint unit used in this work. ................................................................................ 20

7. FieldPoint port divisions. .................................................................................................... 21

8. FieldPoint terminal numbers. .............................................................................................. 21

9. Schematic of the MAX work flow. ..................................................................................... 22

10. MAX control panel snapshot. ........................................................................................... 23

11. Air and CO2 mass flow controllers. .................................................................................. 24

12. DB 15 D-SUB VGA 15-pin female terminal. ................................................................... 25

13. MFC pin functions. ........................................................................................................... 25

14. MFC jumper positions. ..................................................................................................... 26

15. LabVIEW setting for an air MFC. .................................................................................... 27

16. LabVIEW front panel display of the air flow. .................................................................. 27

17. LabVIEW display for the CO2 MFC. ............................................................................... 28

18. LabVIEW front panel display of the air flow. .................................................................. 28

19. An MFC’s zero potentiometer ports. ................................................................................ 29

20. Arrangement of probes at both ends of the randomly packed column. ............................ 29

21. GMP 220, CO2 probe/sensor............................................................................................. 30

22. Chamber for the Vaisala probe. ........................................................................................ 31

vi
23. Vaisala CARBOCAP® receiver and transmitting device.................................................. 31

24. Flow sheet of probe and sensor positions. ........................................................................ 32

25. LabVIEW setting for the top probe and sensor. ............................................................... 32

26. LabVIEW setting for the bottom probe and sensor. ......................................................... 33

27. Hydraulic actuator-based control valve. ........................................................................... 34

28. Flow rate vs. valve opening. ............................................................................................. 34

29. Flow sheet of the control valve and its circuit. ................................................................. 35

30. LabVIEW program to execute controller actions. ............................................................ 35

31. LabVIEW program for the pressure sensor. ..................................................................... 36

32. Process flow diagram. ....................................................................................................... 36

33. The packed column used in this work............................................................................... 37

34. Hacketten tri-pack, image courtesy of Hebei Sinta FRP Products. .................................. 38

35. Gas-phase and liquid-phase concentration profiles for rapid reactions. ........................... 42

36. Gas phase and liquid phase distribution in the packed column. ....................................... 43

37. LOOP-PRO results for the 0.003 M NaOH concentration data........................................ 47

38. LOOP-PRO results for the 0.01 M NaOH concentration data.......................................... 47

39. Bode diagrams for the 0.003 M NaOH concentration. ..................................................... 48

40. Bode diagrams for the 0.01 M NaOH concentration. ....................................................... 49

41. PID-controlled set point tracking of the 0.003 M concentrate NaOH solution. ............... 50

42. PID-controlled set point tracking of the 0.01 M concentrate NaOH solution. ................. 50

vii
LIST OF SYMBOLS

Uppercase Latin Letters

Symbol Definition Units


′ Overall volumetric mass transfer coefficient mol.m-3. s-1. atm-1
(based on partial pressure driving force)
Overall volumetric mass transfer coefficient mol.m-3. s-1
(based on mole fraction driving force)
D Diffusivity m2. s-1
H Henry’s law constant liter.atm. mole-1
KC Proportional gain of a PID control (% controller output) *(mole fraction)-1
KCU Ultimate proportional gain of a PID (% controller output) *(mole fraction)-1
controller
KP Process gain of process model (mole fraction) *(% controller output)-1
NA Number of moles CO2 transferred from gas mole.min-1
phase to liquid phase
PT Total system pressure atm
S Tower cross section area m2
TU Ultimate time period sec
V Volume of NaOH solution liter
Z Height of packing inside the column m
ΔH° Change in the enthalpy kJ.mol-1
Lowercase Latin Letters

Symbol Definition Units


a Interfacial mass transfer area per volume of the column m2.m-3
d Diameter of the packed column m
l Length of the packed column m
n Moles of CO2 mole. liter-1
u(t) Controller output -
wt. Weight of solute (NaOH) gram
x Liquid phase mole fraction (NaOH solution) -
y CO2 mole fraction -
Greek Letters

Symbol Definition Units


δ Film thickness m
θP Process dead time sec
Process time constant sec
Integral time of a PID controller sec
ω Angular frequency rad.sec-1
Critical angular frequency rad.sec-1
Φ Phase angle Rad
ε Error liter.min-1

viii
LIST OF ABBREVIATIONS

AI Analog Input

AMFC Air Mass flow controller

AMP 2-Amino-2-Methyl-1-Propanol

AO Analog Output

CMFC Carbon dioxide Mass flow Controller

CO2 Carbon dioxide

COX Carbon oxide derivatives

DAU Data Acquisition Unit

DC Direct Current

EPA Environmental Protection Agency

EIA Energy Information Administration

FOPDT First order plus dead time model

FP FieldPoint

HCO3- Bicarbonate

I/O Input/output

IP Internet Protocol

LAN Local Area Network

MAX Measurement and Automation Explorer

MDEA Methyl diethanolamine

MEA Monoethanolamine

MFC Mass Flow Controller

COX Carbon oxide derivatives

ix
MMT Million Metric Tons

Na2CO3 Sodium carbonate

NaOH Sodium hydroxide

NI National Instruments

NOX Nitrous oxide derivatives

OH- Hydroxide ion

OS Operating systems

PID Proportional Integral Derivative

ppm parts per million

SOX Sulphur oxide derivatives

STP Standard Temperature Pressure

VDC Volts of Direct Current

Z-N Ziegler-Nichols

x
CHAPTER 1

INTRODUCTION

Human activities have changed and continue to change the earth’s atmospheric

composition and surface. Air pollutants adversely affect the environment by interfering with

animal species, the climate, plant physiology, and entire ecosystems, as well as with human

property in the form of man-made structures and agricultural crops [1]. The exhaust gases from

vehicles and industry include COX, SOX, NOX, hydrogen sulfide (H2S), and other volatile organic

compounds (VOCs) [2, 3]. Carbon dioxide is a greenhouse gas that becomes undesirable when

its concentration exceeds permissible levels in the earth’s atmosphere [4], and carbon dioxide

emissions are a major reason for global warming, which causes melting of the polar ice caps,

desertification, and increased earth temperatures [5, 6]. Hence, there is a need to control CO2

emissions into the atmosphere. The U.S. Energy Information Administration (EIA) reported in

2016 that CO2 emissions from fossil fuels amounted to 2,306 million metric tons (MMTs) from

petroleum products, 1,490 MMTs from natural gas, and 1,375 MMTs from coal [7].

Scientists to date have developed a number of techniques to control exhaust gases, which

is important in modern industry from an environmental perspective, and the research is still

ongoing. Among the available techniques, carbon dioxide scrubbers or packed columns are

widely used in processing, power plants, cement manufacturing, and other industries to reduce

CO2 emissions [8]. Packed columns are mass-transfer equipment that absorb the carbon dioxide

found in the exhaust gases from industrial plants or the exhaled air from life support systems,

such as rebreathers in spacecraft, submersible craft, and airtight chambers. Some of the

technologies that can apply packed columns are amine scrubbing, chemicals and zeolites,

regenerative carbon dioxide removal, activated carbon, and metal-organic frameworks [9–14].

The chemicals and zeolites are traditional and widely used in laboratory operations.

1
An array of different types of industrial equipment and chemical absorbents have been

used to clean emitted exhaust gases before they are discharged into the atmosphere. One

commonly used CO2 absorbent is sodium hydroxide (NaOH) solution because of its low cost and

availability in the market [9]. Because CO2 absorption is governed by chemical reaction rate and

mass transfer rate, industry and academia have conducted a great deal of research to quantify

reaction and mass transfer rates in processing equipment, including spray, bubble, tray, and

packed columns [15]; all of this equipment provides larger contact areas for the mass transfer

from the gas phase to the liquid absorbent [15, 16].

The California State University–Long Beach (CSULB) Chemical Engineering

Department has a packed column with a randomly packed plastic Hacketten tri-pack that is

suitable for studying CO2 absorption with NaOH, and the absorption performance of this packed

column has not been studied before now. For this thesis work, I collected experimental results

for absorbing CO2 from the air stream. My data analysis focused on calculating the Hatta number

to identify the governing mechanism, and after I calculated the Hatta number, I calculated the

mass transfer coefficient based on the method of transfer units. Additionally, I made a concerted

effort to develop feedback control using LabVIEW software technology to ensure that the

absorption goals were reached. Building on this research, I highlighted a real-world procedure

for tuning the control loop using established theories.

Packed Column Mass Transfer

Packed columns can be studied using mass transfer theories, which help in calculating the

rate of mass transfer between the gas and liquid phases. The driving force for mass transfer is

typically a difference in the chemical potential coupled with other thermodynamic gradients.

Chemical species move from areas of higher to lower chemical potential, and thus, the maximum

theoretical extent of mass transfer is typically based on the equilibrium that can be attained as a

2
result of the mass transfer from one phase to another. To study the behavior of mass transfer in

packed columns, I identified four theories in the literature [15]:

• The surface renewal theory (Danckwerts, 1951)

• The penetration theory (Higbie, 1935)

• The two-film theory (Whitman, 1923)

• The film-penetration theory (Toor and Marchello, 1958)

Using the above-mentioned theories under different operating conditions and

assumptions, researchers have developed many mechanisms to describe absorption in packed

columns. However, the dominant mechanism (diffusion of CO2 to a gas/liquid interface,

dissolution of CO2 on the gas/liquid interface, diffusion of CO2 from the gas/liquid interface

toward a bulk liquid, and reaction of CO2 with an absorbent) may vary with different

concentrations of absorbent (NaOH), distribution patterns, contact surface geometries, and gas

and liquid phase flow rates.

Industry uses a dimensionless number called the Hatta number (Ha), which I will define

later, to identify the dominant mechanism in packed columns; once that mechanism is identified,

one can then clearly compute the apparent or observable mass transfer coefficient that

characterizes the absorption speed in a given piece of equipment. By calculating the Ha, which

could suggest whether the mass transfer (diffusion) or the chemical reaction controls the whole

absorption process [16, 17], one can then calculate the overall number of transfer units (NOG), the

height of the transfer units (HOG), and the overall gas-phase mass transfer coefficient (K a) [15,

18, 19]. These theories help to identify the amount of CO2 transferred from the gas phase to the

liquid phase, which could also be called CO2 scrubbing. I discuss all of these theories in detail in

the section on Ha.

3
Control Theory in Packed Columns

Because of increasing energy costs and more stringent government safety and

environmental regulations, many industries are making significant efforts to modernize and

automate production facilities. Additionally, with the widespread use of computers for automated

online control, industries are attempting to apply robust control theories [20]. One conclusion to

the problem of the process control gap between theory and real-world practice is that there exists

an acute need for systematic studies of the design and implementation of process control models

for pilot-scale industrial processes, and I discuss in detail the process control theories required

for this work in the methodology chapter in the section on process control.

Objectives of the Thesis

I selected a randomly packed column as the chemical process to study because of its

complexity, inherent control difficulties, and extensive industrial importance. I sought in my

thesis to study the randomly packed column installed in chemical engineering laboratory to

minimize the difficulties associated with process modeling by providing an accurate unified

approach to calculating the mass transfer and develop a random pack columns process control

model that can be used to control industrial CO2 exhaust emissions. The objectives of this work

were as follows:

1. To conduct a case study on CO2 absorption with a NaOH solution in a counter-current

mass transfer operation

2. To develop an advanced program in LabVIEW for the process model and data analysis

and to reduce the full model to one that is suitable for control structure design and online

automatic control; to set up automatic online control using software and hardware that I

discuss later in “Automation and Instrumentation”

4
3. To calculate the Hatta number, which suggests that the controlling mechanism is either

the mass transfer (diffusion) or the chemical reaction between the gas and liquid phases

4. To calculate the mass transfer coefficients using transfer units, including calculating the

number and height of transfer units and the KG a using the experimental data, and to

compare the experimental mass transfer coefficients with the literature data to compare

the performance of the packed column

5. To develop a process control model that uses automatic online control for the packed

column involving mass transfer using a feedback PID controller and applying methods

such as internal model controller (IMC) and Ziegler–Nichols (Z–N) tuning based on the

process dynamic model and the model parameters obtained from the step tests

6. To obtain a unified process modeling approach for randomly packed columns that can be

used in industrial-scale CO2 emission control

5
CHAPTER 2

LITERATURE REVIEW

The study of carbon dioxide absorption into sodium hydroxide solutions requires a wide

evaluation of mass transfer and process control theories. This literature review provides a brief

overview of the transfer unit method entailed in finding the number of gas phase transfer units

(NOG), the height of the transfer units (HOG), and the Hatta number, which are used to calculate

the K ); in it, I also apply process control theory to randomly packed columns.

Currently, there are many technologies available for carbon dioxide scrubbing, as

discussed in the opening chapter. Of these, amine scrubbing is the most widely used in many

industries to remove CO2 from exhaust gases [14]. Another technology mentioned earlier was the

use of zeolites and chemicals to absorb carbon dioxide. Using NaOH with zeolites and other

minerals has similar potential to that of the current industrially practiced amine scrubbing [9, 10].

The negative properties of amine scrubbing, such as its continuous topping and toxic and highly

volatile nature, make it uneconomical for removing carbon dioxide from exhaust [14]. Hence, in

this work, I used the previous standard technique of absorbing CO2 with a NaOH solution.

Zeman and Lackner outlined a specific technique for using this CO2 method [10], and I applied

their technique in this thesis to frame the chemical reaction-initiated mass transfer. Specifically,

my method absorbed the carbon dioxide from the air and CO2 mixture (gas phase) using a

sodium hydroxide solution (liquid phase), and I used Ha to determine the speed of the reaction

between the NaOH and CO2. Ha, the Hatta number, is a dimensionless parameter that assists in

understanding the reaction speeds and the concentration profiles of the gas and liquid phases as

well as aiding in calculating mass transfer coefficients. To understand the theory behind the

Hatta number, I referred to Bird and Hartland [16, 17].

6
The process model I designed in this work aimed at identifying a unified approach to

process control in using randomly packed columns to remove carbon dioxide from exhaust gases.

I developed the model to be controlled online automatically, that is, applying control theory to

regulate the process without direct human intervention. I used a feedback PID controller for the

online control and conducted a complete background study of how to apply these controllers to

packed columns [20–22]. The optimum process parameters are obtained by tuning the feedback

PID controller, and I used IMC and Ziegler–Nichols tuning following a complete literature

review of automatic control, PID controllers, and IMC and Z–N tuning [20–27].

7
CHAPTER 3

METHODOLOGY

Mass Transfer in Randomly Packed Columns

Absorption can be either physical or chemical, and many industries use gas absorption

accompanied by chemical reactions in the bulk liquid phase. These chemical reactions enhance

the absorption and increase the capacity of the liquid solution to dissolve the solute compared

with the regular physical absorption systems.

Applying Zeman and Lackner to Randomly Packed Columns

Using Zeman and Lackner’s method [10], CO2 is absorbed by an aqueous NaOH solution

to produce dissolved sodium carbonate:

2NaOH (aq) + CO2 (g) → Na2CO3 (aq) + H2O (l), ΔH° = −5.3 kJ.mol−1 (1)

This NaOH and CO2 reaction is a strongly exothermic gas–liquid reaction, and the absorption in

this method takes place in randomly packed columns.

Absorption is a mass transfer operation, and in this experiment, the gas–liquid absorption

was conducted in a counter-current direction. Identifying the mass transfer from the gas to the

liquid phases in gas–liquid reactions requires the Ha, HOG, NOG, and height of packing (Z), while

identifying HA and the transfer unit variables in randomly packed columns requires studying

theories of mass transfer. As I identified in “Packed Column Mass Transfer” earlier, the literature

discusses four theories of mass transfer from gas to liquid: the two-film theory, in which mass

transfer is postulated to proceed via stationary molecular diffusion in a stagnant film of thickness

(Whitman, 1923); the penetration theory, in which the residence time (θ) of a fluid element at the

interface is the characteristic parameter (Higbie, 1935); the film penetration theory, in which the

film and penetration theories are applied together (Toor and Marchello, 1958); and the surface

renewal theory, which introduces a probability of replacement (Danckwerts, 1951) [15].

8
In this work, I used a diffusion-enhanced chemical reaction to transfer the CO2 molecules

from the bulk gas stream to the NaOH solution; this reaction applies the film-penetration theory.

As the name suggests, this theory combines film theory and penetration theory. Mass transfer

from a gas to a liquid proceeds via the interface between the gas and liquid phases, where the

penetration theory applies, and the liquid forms a film on the surface of the packing material; the

gas molecules that penetrate the gas–liquid interface then react with the film following the film-

penetration theory. To calculate the rate of mass transfer using the film-penetration theory, I

worked in the following sequence: Hatta number > transfer units > mass transfer coefficients (Ha

> NOG > HOG > KG a).

Hatta Number

The Hatta number is a dimensionless number for identifying the speed of gas–liquid

reactions. Its significance can be expressed as a rate of two terms constants [15–17]:

√ ∗
Ha = = = ! "#
(2)

where k = reaction velocity constant (s−1), D = diffusivity of CO2 through aqueous solution of

NaOH (m2.s−1), and k % = physical mass transfer coefficient (m.s−1).

A significant Ha (>2) [17–19] implies that the diffusion time is much greater than the

reaction time and hence the considerable reaction can be expected to take place in the liquid film

itself. The concentration will drop significantly in the liquid film for large Ha values.

Transfer Units

A newer concept in the analysis of packed columns centers on transfer units. This method

is more appropriate because the changes in the compositions of the liquid and gas phases occur

differently in packed columns than in the stepwise fashion of trayed columns.

Height of transfer unit: The HOG is a measure of the separation effectiveness of the

packings. The more efficient the mass transfer, the smaller the HOG.
9
Number of transfer units: The NOG required is a measure of the difficulty of the

separation. A single transfer unit gives the composition change of one of the phases equal to the

average driving force that produces the change.

Packing height: Using transfer units, the required Z can be determined based on either

the gas or the liquid phase. For this work, the flows of the gas and liquid phases, mole fractions

of the input gas and liquid phases, and the Z inside the column could be visualized using Figure

1.

FIGURE 1. Packed column gas phase and liquid phase distribution.


Note: &' , &) : the mole fractions of the gas phase entering and leaving; +' , +) : the mole
fractions of the liquid phase; ,' , ,) : the mass flow rates of gas entering and leaving the
column; -) , -' : the mass flow rates of the liquid phases entering and leaving the column; Z:
height of the packing material.

Formula (4) gives Z with respect to HOG and NOG:

53 2345678 5
Z = H0 159 234562545∗ 6
(3)

Z = HOG*NOG (4)

where, N0 = number of overall gas phase transfer units, H0 = height of overall gas phase

transfer unit, and Z = height of tower packing. To calculate the NOG, formula (5) best fits

randomly packed columns [17–21]:


5 45
N0 = 2545
;
∗6
<
(5)
=

10
The log mean mole fraction difference of the denominator in formula (5) is expanded into

the following equation (6):

25; 45∗; 6425< 45∗ < 6


2y − y ∗ 6%@ = ∗
CD ED F
(6)
A [ ; ;∗ ]
CD< ED< F

In equation (6), y3∗ and y ∗ 9 , the gas phase solute mole fractions, are in equilibrium with

the bulk phase liquid concentrations x1 and x2 because randomly packed columns are

predesigned with known Z, and NOG is found using equation (5). The unknown quantity is HOG,

which is calculated using equation (4).

Mass Transfer Coefficients

The mass transfer coefficient is a rate constant of diffusion that relates to mass transfer

rate, mass transfer area, and concentration change or pressure differences as driving forces [19–

21]; mass transfer coefficients can be identified in either the gas or the liquid phases.

Experimentally, the individual mass transfer film coefficients k % and k are difficult to measure

because the concentration differences across the phases are significant [18]. Under these

circumstances, the overall coefficients K a and K % a can be calculated based on the overall gas or

liquid phase driving force. In this thesis, instead of carrying out regular liquid-phase calculations

or conducting titrations in the liquid phase to determine the amount of mass transfer, I calculated

the overall mass transfer coefficients based on the gas phase. To determine the overall K ′ a, I

used the following equations taken from Perry et al. [19]:

K
K J a = M∗N∗O L∗5 (7)
P =

K a = PR ∗ K J a (8)

To calculate NA in equation (7), I found the number of CO2 moles transferred using the

following equation (9):

NS = 22nCO9 6 − 2nCO9 66 (9)


11
The number of CO2 moles that enter and leave the packed column (9) are calculated

using the mass flow controllers and CO2 analyzer data at standard temperature and pressure

(294.25 K,1 atm, according to the manufacturers). I used equations (10) and (12) to calculate NA.

Z0< [ \ A ! # A ] !" " " "! R & O


nVW9 XY = (10)
A"! \ A _" " " "! R & O

Because air is an inert component whose flow rate does not change throughout the

column, the molar flow rate of CO2 leaving the system (nZ0< ) can be calculated from the mole

fraction of CO2 leaving the column (`9 ), according to the mass balance equations below:

`9 =
ab< cd
(11)
ab< cd e fg

5< fg
nZ0< cd
= 2345< 6
(12)

Substituting equations (10) and (12) in equation (9) helps in calculating the NA. I obtained the

required K a by multiplying total system pressure (PT) by K J a; see equation (8). Note that K J a −

mole.min−1.m−3.atm−1, K a − mole.min−1.m−3, NA = number of moles of CO2 transferred (gas

phase to liquid phase) or absorbed (mole.min−1), Z = height of packing in the tower (m), S = total

cross-sectional area (m2), PT = total system pressure (atm), and y%@ = log mean mole fraction

difference from the equation.

Process Control

Process control is the branch of engineering that concerns the mechanisms and algorithms

necessary to maintain process outputs under required conditions. Process control models

empower mass production of consistent products from continuously operated processes, such as

oil refining, paper manufacturing, chemicals, power plants, and many others. Process control

enables automation, which allows a small staff of operating personnel to operate a complex

process from a central control room.

12
FIGURE 2. General example of a feedback control loop.

Process control uses feedback or open loops based on the program or algorithm; in this

experiment, I used a PID control loop feedback. Generally, feedback control systems compare

their measured values with the desired values and then adjust the manipulated variables

accordingly and control the process, as shown in the figure.

PID Controller

A proportional–integral–derivative (PID) controller is a feedback control loop

mechanism commonly used in industrial process control systems. It calculates error values in

real time as the differences between desired set points and measured process variables. The

controller attempts to decrease the errors over time by adjusting a controller output (t) such as the

position of a control valve to a newly determined weighted sum [20]:

3 st
h2i6 = jk lm + o 1 m qi + rs su v (13)
p

where u (t) = controller output, KC = controller gain, ε = error (YNO − Y), τy = integral time

constant (s), and τs = derivative time constant (s). In the experiment for this thesis, I designed

the PID controller feedback loop using LabVIEW (see Figure 2); in the loop, the manipulated

variable was the control valve output, and the process variable was the top sensor response (the

CO2 probe at the top of the column). A step change was given by the manipulated variable,

which caused a change in the flow rate of the NaOH solution.


13
FIGURE 3. A PID controller in a feedback loop.

|2}6 3
zk 2{6 = t 2}6 = jk l1 + o } + rs {v (14)
p

In the experiment for this thesis, I designed the PID controller feedback loop using

LabVIEW (see Figure 2); in the loop, the manipulated variable was the control valve output, and

the process variable was the top sensor response (the CO2 probe at the top of the column). A step

change was given by the manipulated variable, which caused a change in the flow rate of the

NaOH solution. I assigned a desired set point value to the CO2 top probe and compared the set

point with the process variable and the error generated. The PID controller minimized the error

using the proper control parameters (K Z ,τy ,τ ) obtained from IMC and Ziegler–Nichols tuning.

Tuning the PID. Tuning the controller entails adjusting the control parameters (K Z , τy ,

τ ) to the optimum values to obtain the desired response, and there are many proven methods for

tuning these controllers. For this work, I used IMC and Z–N tuning to identify the optimum

parameters.

In the open-loop step test, the process is first allowed to reach a steady state in the manual

mode; a step change is made in the controller output for a certain percentage, and the response of

the process variable is recorded until it reaches a new steady state. To determine the parameters,

the input and the output data obtained from the open-loop step test can be processed by

14
software such as LOOP-PRO (Control Station, Inc.). I chose a first-order plus dead-time

(FOPDT) model to fit the manipulated variable (controller output, or input variable) and the

process variable (CO2 analyzer value at the top of the tower, `9 , or output variable) and time (s)

to obtain the model parameters: process gain (K [ ), time constant (τ[ ), and dead time (Ө[ ).

Obtaining these process dynamic model parameters makes it possible to obtain appropriate

tuning parameters for the feedback controller using well-established methods such as IMC,

Ziegler–Nichols, Tyreus–Luyben, and many other tuning methods [23].

IMC tuning. The internal model control tuning method was developed for self-

regulating processes. A self-regulating process always attains a new steady state after a step

change is made in the controller output in the manual mode [24]. To control the process using a

closed-loop PID controller, it is necessary to select the desired closed-loop time constant (τZ );

using a larger τZ will result in a slower control loop, whereas using a smaller value results in a

faster loop [24].

I calculated the final required PID control settings using IMC tuning and the equations

shown in Table 1.

TABLE 1. PID Controller Settings Using IMC Tuning

ۥ
Controller Gain, K # = ‚
• ∗2€ƒ eӨ[6

Integral Time, τy = τ[

Derivative Time, τ = 0

The obtained PID control settings from the IMC tuning were then commanded on the PID

controller, and the controller was switched to the automatic mode; the control loop response can

15
be sped up by increasing K Z or decreasing τy . Normally, the derivative mode is turned off with

2τ = 06 for processes in which measured process variables are inherently “noisy” [24].

Ziegler–Nichols tuning. Ziegler and Nichols used the following definition of acceptable

stability as a basis for their controller tuning rules. The ratio of the amplitudes of the subsequent

peaks in the same direction (owing to step changes in the controller) is approximately 0.25 [13].

Ziegler and Nichols operated their process model as a closed loop (see Figure 4).

FIGURE 4. Ziegler–Nichols closed-loop tuning.

Z–N tuning entails analyzing frequency response to generate Bode diagrams for

identifying the sustained oscillation conditions followed by determining control parameters by

introducing appropriate gain and phase margins [25–27]. There are multiple approaches for

identifying these sustained conditions; for my thesis, I selected theoretical frequency response

analysis on an open loop.

Frequency response analysis. When a linear system is subjected to a sinusoidal input

perturbation, its ultimate response after a long time also becomes a sinusoidal wave [26–27]. Because this

process can be described with a FOPDT model, the response to the sinusoidal input can be explained

through equations.

For FOPDT models, the transfer function in the Laplace domain is given by equation

(15):

16
52 6 ‚• EӨ•‡
G[ 2s6 = = (15)
2 6 ۥ e3

If a sinusoidal input is made to the process model (G[ 6, the input signal in the time domain is

given by the following equation (16):

u(t) = A sin ωt (16)

The response of the process variable becomes another sinusoidal function:

‚S
y 2t6 = √‰< sin{ωτp+tan-1(-ωτp)} – (ωӨp) (17)
€< e3

Bode diagrams. Bode diagrams are used to conveniently represent the frequency

response characteristics of the process dynamic model. The diagram for first-order plus dead-

time processes has the following relationships [28]:

‚[
AR = (18)
•‰< €• < e3

φ = tan−1(−ωτ‘ ) − (ωӨp) (19)

where AR = amplitude ratio, φ = phase difference, A = amplitude of oscillation, ω = angular

frequency, τO = time constant of open loop model, and Өp = dead time.

All of the above equations (15) to (19) are required to calculate the parameters in Z–N

tuning. Specifically, the following are the Z–N steps for tuning a PID controller:

1. The key procedure to obtain sustained oscillation is calculating the critical angular

velocity 2ωk 6, which results in the phase angle of −π. Then, we can determine the

ultimate proportional gain for the controller K Z’ and the ultimate oscillation period T’ .

2. Using the LOOP-PRO process model parameters (K [ , τ[ , Өp6 and frequency analysis,

when ωk gives a phase angle equal to −π, the PID controller output is in phase with the

process input.

17
3. If we set the proportional gain on the PID controller, then KC = KCU = 2AR Z 6−1 as shown

in equation (21), the amplitude of the controller output is the same as the process input.

We then achieve sustained oscillation because both the controller output and the process

input are in phase. The period of sustained oscillation is then given by equation (20).
9”
T’ = ‰ (20)
ƒ

3
K Z’ = (21)
S• –

To obtain appropriate PID controller command settings, it is necessary to introduce gain

and phase margins so that the closed loop will be operating on the stable side of the sustained

oscillation. To do this, I multiplied a factor smaller than 1 (e.g., 0.45, 0.30, 0.16) by KCU to

obtain KC and divided the ultimate oscillation period by 1.2 to obtain TU.

Finally, I calculated the required PID control settings using the equations in Table 2.

TABLE 2. Ziegler–Nichols PID Controller Settings


Controller Type KC τy

PID 0.3*KCU 0.83*TU

The mass transfer methodology and process modeling I have presented so far helped to

minimize the difficulties associated with mass transfer and process modeling in randomly packed

columns by providing an accurate unified approach to process control and performance.

18
CHAPTER 4

AUTOMATION AND INSTRUMENTATION

Automation, or Automatic Online Control

Automatic control in the fields of engineering and technology is a broad and generic term

that covers applying mechanisms to process operation and regulation without continuous direct

human intervention. Automation uses a number of different control systems for operating

equipment, such as process instruments, electric motors, and many others. The opposite of

automation is manual control, which requires human effort and thus has high potential for errors

in accuracy and precision.

FIGURE 5. A data acquisition unit.

Automation increases production quality and quantity with the least energy consumption.

For this thesis, I used different blocks of hardware combined with software technology for the

automation; specifically, I used Laboratory Virtual Instrument Engineering Workbench, Version

8.1 (LabVIEW) by National Instruments.

19
LabVIEW

LabVIEW is a developmental system design environment for visual programming. It is

used on a laboratory scale for data acquisition (see Figure 5), and I used it for data acquisition,

modeling, and simulation following [22].

FieldPoint

National Instruments also supplies FieldPoint, a modular distributed input/output system

with industrial-grade performance and unmatched software integration. FieldPoint can be

compared with the programmable logic circuits that are used to integrate measuring or

controlling instruments and transmit data to the computer for additional processing; the data

collected from FieldPoint are programmed in LabVIEW.

Figure 6 shows the FieldPoint unit I utilized for this thesis. It is divided into a power port,

analog input, and analog output, as shown in Figure 7. All instruments that deliver output in

voltage or current are connected to FP-AI-110 (the analog input port), and all instruments that

require voltage or current to perform their commands are connected to FP-AO-220.

FIGURE 6. The FieldPoint unit used in this work.

20
FIGURE 7. FieldPoint port divisions.

The model I used in this work has both the FP-AI and FP-AO ports and has seven

channel divisions to connect instruments based on whether they are voltage or current based. To

aid in understanding the channel divisions for assigning the instruments to the respective

terminals, the manufacturer provides the channel and terminal information shown in Figure 8.

FIGURE 8. FieldPoint terminal numbers.

The data received by FieldPoint from the instruments must be transmitted to analyze the

experimental data, and the available LAN ports allow for working online. The IP address through

which FieldPoint is linked is the key source of recognition for instruments in LabVIEW.

21
Measurement & Automation Explorer

Measurement & Automation Explorer (MAX) is software that provides access to

FieldPoint. It configures National Instruments hardware and software and creates and edits the

channels, tasks, interfaces, scales, and virtual instruments connected to the system, and it allows

for viewing all connected devices.

Figure 9 describes how MAX aids in understanding the working of instruments

FIGURE 9. Schematic of the MAX work flow.

Figure 10 shows a snapshot of the MAX operating panel. First, FieldPoint is connected to

the LAN and the IP address is noted. To identify and successfully instruments to FieldPoint and

to the computer that has LabVIEW installed, MAX must go through the following steps: FP@IP

> Select the Instrument in (AI or AO) > Select the channel > Go to > Start > check Value,

Range, Status.

22
FIGURE 10. MAX control panel snapshot.

Mass Flow Controllers

A mass flow controller (MFC) is an instrument that measures and controls the flow rates

of liquids and gases; they are designed and calibrated for specific gases or liquids at predesigned

flow rate ranges, and they control the flow rates based on the set points given. MFCs have a set

point range from 0 to 100% of their full-scale range, but in practice, they are most accurate when

they are operated within 10% to 90% of the total scale. They are either analog or digital, and

some companies offer both in a single instrument. For this work, I used one of each type for ease

of operation.

MFCs have an inlet, outlet, mass flow sensor, and proportional control valve. Most are

fitted with a closed-loop control system in which the operator gives an input signal that the

controller compares with the mass flow sensor, adjusting the proportional control valve to

23
achieve the required flow, and the flow rate is specified as a percentage of the calibrated full-

scale flow and supplied to the MFC as a voltage signal. For this work, I required two MFCs to

control the mass flow rates of the compressed air and the CO2, and Figure 11 shows the two

used. They operated at STP conditions of temperature and pressure at 294.25 K and 1atm

according to the Dakota Instruments operating manual. Table 3 shows the complete data sheets

for the units used.

TABLE 3. Dakota Instruments Mass Flow Controller specifications

Instrument Serial No Flow range Max Input Voltage Output

(liter.min−1) pressure (volts) Signal

(psi) (mA)

Air Mass Flow Controller 406787-1 0-200 500/34 24 4-20

CO2 Flow Controller 406789-1 0-50 1000/68 24 4-20

FIGURE 11. Air and CO2 mass flow controllers.

24
Installation and Operation

To supply the power and to collect the analog output signals, MFCs contain 15-pin D-

connector male pins on their bodies. The male pins then connect to external DB 15 D-SUB VGA

15-pin female adapter jack terminal breakout PCB boards; see Figure 12.

FIGURE 12. DB 15 D-SUB VGA 15-pin female terminal.

The male pin connects to the female pin, and the female pin must be connected to

FieldPoint to measure the MFC readings. To connect the MFC with FieldPoint, the pin function

data should be used as shown in Figure 13.

FIGURE 13. MFC pin functions.

For my work on this thesis, I mounted the jumpers locally rather than remotely to

increase the ease of controlling the mass flow rates of compressed air and CO2; see Figure 14. To

25
adjust jumper positions to be either local or remote, MFCs have accessible primary windows

behind them. If the jumpers are set to “Local,” the instrument can be controlled using the local

set point potentiometer [NR7] located behind the MFC at the primary window.

FIGURE 14. MFC jumper positions.

Air Mass Flow Controllers

For my experiment, I used an air mass flow meter to control the flow rate of compressed

air from the external source to the required flow rate between 0 and 200 liter.min−1 with

maximum operating pressure of 34 atm. The voltage required for the instrument was supplied by

the 24 VDC adapters provided by Dakota Instruments, and the output signal delivered by the

device was 4-20 mA. The instrument was analog and digital; the digital display helps the user

view the air flow rate through the device instantly if the process is under closer examination. The

AFC connected to FieldPoint through the analog input signal port (FP-AI-110) at channel 4. To

read the real-time data, AFC must be programmed in LabVIEW. The LabVIEW software

receives the flow rate only in amperes; Figure 15 displays how LabVIEW must be set in order to

convert flow rate to liter.min−1, which helps the user understand the exact flow of the air through

the instrument, unlike ampere flow rate data.

26
FIGURE 15. LabVIEW setting for an air MFC.

Figure 16 shows the LabVIEW front panel display following my programming and after I

gave the program start command.

FIGURE 16. LabVIEW front panel display of the air flow.

Carbon Dioxide Mass Flow Controllers

I used a carbon dioxide mass flow controller (CMFC) to control the flow rate of

compressed CO2 from the external cylinder source to the required flow rate of between 0 and 50

liter.min−1 with maximum operating pressure of 68 atm. The voltage required for the instrument

was supplied by the Dakota Instruments 24 VDC adapters, and the device delivered an output

signal of 4-20 mA. The instrument was both analog and digital, and the digital display helped in

instantly viewing the rate of CO2 flowing through the device. The CMFC connected to

FieldPoint via FP-AI-110 (the analog input port) at channel 1. To read the real-time data, it had

27
to be programmed in LabVIEW, as shown in Figure 17, and Figure 18 shows the LabVIEW, the

front panel display after the programming.

FIGURE 17. LabVIEW display for the CO2 MFC.

FIGURE 18. LabVIEW front panel display of the air flow.

Instrument Safety

MFCs are highly sensitive and inappropriate for humid or moist conditions; both the

compressed air and the CO2 should be moisture free or the device might be subject to internal

circuit damage. Each flow meter is designed only for specific gases, and using any other gases

can damage the calibration. Without the flow of gas, the power supply heats up the controller and

damages it.

Zero Adjustment of Potentiometers

MFCs are pre-calibrated in the industry using R52, R33, R38, R39, and R40, which are

located at the secondary window fixed with a seal, but the set point (NR7) and zero (R34)

potentiometers are reserved for user control and control valve adjustments. To ensure that no gas

leaks into the meter, the gas flow into the MFC should be temporarily disconnected. Using a

28
multimeter and an insulated screwdriver, the zero potentiometer (R34) can be adjusted through

the access window behind the MFC for 0 VDC or 4 mA at zero flow; see Figure 19.

FIGURE 19. An MFC’s zero potentiometer ports.

Carbon Dioxide Analyzers

To identify the CO2 concentrations in process models, many instruments are available,

but for operation lower ranges, 0–30% on the laboratory scale, infrared gas analyzers are useful.

This was the type of analyzer I used in this work to identify the amount of CO2 absorbed by the

sodium hydroxide solution. To determine these amounts, I needed two analyzers placed at both

ends of the randomly packed column; see Figure 20. The two analyzers I used were GMT220

series carbon dioxide transmitters from Vaisala Instruments.

FIGURE 20. Arrangement of probes at both ends of the randomly packed column.

29
The Vaisala CARBOCAP® CO2 analyzer is designed to measure carbon dioxide in harsh

and humid environments. It has two main divisions, the probe and the analyzer. The probe

detects the CO2 concentration at the measurement point and transfers the data through the cable

to the analyzer, which then transmits the data to FieldPoint in the form of a current signal.

CO2 Probes and Installation in Chambers

The two CO2 measuring probes are connected to their respective analyzers, and the

measuring ends of the probes must be placed in the gas (air and CO2) streams that pass into the

packed column and the gas that leaves the column. The probes are exchangeable based on the

required CO2 ranges, and I used GMP 220 probes with the CO2 range of 0–20%; I ensured that

the probes made no contact with any moisture per the industry rules.

The probes are long and cylindrical, 9.5 cm in length and 1.9 cm in diameter 1.9 cm (see

Figure 21), and they were installed in a closed space in which their tips could make contact with

the incoming gas stream. As shown in Figure 22, for this probe, I designed a chamber and

ensured that it did not leak.

FIGURE 21. GMP 220, CO2 probe/sensor.

30
FIGURE 22. Chamber for the Vaisala probe.

The probe is placed perpendicular to the controlled gas flow, which enables the probe to

measure the CO2 in the gas stream flowing through the chamber.

CO2 Analyzer Installation

The process model requires two analyzers to identify the amount of CO2 absorbed in the

process; they are mounted at the top and bottom of the column. The CO2 analyzer receives the

analog signal from the probes and transmits the measured carbon dioxide concentration signal to

FieldPoint; it operates with 24VDC, and FieldPoint supplies the required voltage. This analyzer

is analog and has no digital display (see Figure 23). The analog output is in the range of 4–20

mA, which is programmed into LabVIEW for conversion to ppm and then to percentage of CO2.

FIGURE 23. Vaisala CARBOCAP® receiver and transmitting device.

31
Configuration to FieldPoint and LabVIEW

The CO2 analyzer, which receives the signal in the form of a 4–20 mA electric current

from the probe, transmits the data to FieldPoint. Those data are then transferred through LAN to

the computer with LabVIEW. Figure 24 displays a schematic flow sheet of connections of the

probes, analyzer, FieldPoint, and LabVIEW. LabVIEW is designed to convert the incoming (4–

20) mA to ppm then to percentage of CO2 measured where the gas enters and leaves the packed

column. For ease in understanding, LabVIEW also refers to the probe as a sensor. Because my

process model used two analyzers, I had to program both analyzers in LabVIEW to view the

amounts of CO2 that my process measured; see Figures 25 and 26.

FIGURE 24. Flow sheet of probe and sensor positions.

FIGURE 25. LabVIEW setting for the top probe and sensor.

32
FIGURE 26. LabVIEW setting for the bottom probe and sensor.

Measurement in LabVIEW

LabVIEW enables converting ppm to percentage of CO2. The two probes I used in this

experiment identified the amounts of CO2 absorbed in the process by the NaOH solution. I

obtained the real-time values of the absorbed CO2 by deducing the bottom and top probe

percentages with respect to the flow of the NaOH solution.

Instrument Safety

CO2 analyzers run only in zero moisture with dry operating chambers, and they should be

operated at room temperature or the infrared sensor inside the probe is subject to damage.

Control Valves

The control valve I used in this work was a flow control valve with an actuator on it that

controlled the rate of the NaOH solution that flowed through the valve. The valve was

automatically controlled with a hydraulic actuator with input power of 110 volts AC, and the

FieldPoint signal to control the valve was 4–20 mA; see Figure 27. I manually tested the control

valve to understand its response to the LabVIEW commands for manual mode operation, which

Figure 27 shows. Figure 28 shows that the valve was proportionally controlled and responded

exponentially; changes in controller output linearly changed the NaOH solution flow rate.

33
FIGURE 27. Hydraulic actuator-based control valve.

Flow rate vs Valve opening


20
18
16
14
Flow rate

12
10
8 Flow rate vs Valve
6 opening
4
2
0
0 50 100 150
Valve opening in %

FIGURE 28. Flow rate vs. valve opening.

FieldPoint and LabVIEW Configuration

The control valve was connected to a 110V AC power supply to control the valve

connected to the FieldPoint analog output port. The entire arrangement of the control valve is

described in the flow sheet shown in Figure 29.

34
FIGURE 29. Flow sheet of the control valve and its circuit.

Automatically controlling the valve requires either a P, PI, or PID model with applicable

tuning rules. I modeled a PID controller using LabVIEW (Figure 30) to automate the control

valve function.

FIGURE 30. LabVIEW program to execute controller actions.

Pressure Sensor and LabVIEW Configuration

A pressure sensor is installed at the center of the packed column to identify the pressure

inside the column. The device has a range of 0–10 atm, and its output current ranges from 4 to 20

35
mA. This device does not require any input power; it converts the internal column pressure to an

analog signal and transfers that to FieldPoint. The FieldPoint data are then programmed into

LabVIEW for conversion to the required pressure units, pounds per square inch; see Figure 31.

Figure 31. LabVIEW program for the pressure sensor.

Process Flow Diagrams

Process flow diagrams (PFDs) are used in chemical and process engineering to indicate

plant process and equipment flows. They display the relationships between the major equipment

at plant facilities but do not include minor details, such as those regarding piping or designations.

Figure 32 shows a PFD that illustrates the circuit of the process instruments and media through

which my data were transferred for this work.

FIGURE 32. Process flow diagram.

36
CHAPTER 5

EXPERIMENTAL WORK

The Randomly Packed Column Design

The randomly packed column I used in this work was predesigned and installed 15 years

ago in the CSULB chemical engineering laboratory. My challenge was that the design details

were not available (perhaps lost over time), and thus I had to identify the column design details

such as length, packing height, diameter, packing material and properties, the electric motor

required to pump the NaOH solution into the column, a rotameter to identify the NaOH solution

flow rate, and capacity of the NaOH solution storage tank. I manually gathered all of these

details by measuring, evaluating, and calculating during my early work on this thesis. Table 4

shows the design details for the randomly packed column I used, and Figure 33 shows the

packed column itself.

FIGURE 33. The packed column used in this work.

37
TABLE 4. Design data on the Randomly Packed Column
S.no Design Data

1. Length of packed column (l) 4m

2. Diameter of packed column (d) 0.16 m

3. Height of packing inside column (Z) 3m

4. Electric motor for the pump 0.5 HP

5. Rotameter (0-10) gallons per minute (gpm)

6. Storage tank of NaOH solution 250 liters (maximum capacity)

Packing Material

The packing material is one of the important factors in absorption; the structure and size

of the materials contribute to absorption rates, and the material provides a large area of contact in

the column between the liquid falling from the top and the gas entering at the bottom. Packings

are classified into two types, random (material is dumped into the column) and structured

(packing is organized inside the column). I randomly packed the column for this experiment with

Hacketten tri-pack, a plastic packing material from Hebei Sinta FRP Products Trading Co., Ltd.

The materials were 25 mm in diameter and had individual surface areas of 460 m2.m−3, and the

pieces are shown in Figure 34.

FIGURE 34. Hacketten tri-pack, image courtesy of Hebei Sinta FRP Products.
38
Materials and Solution Preparation

I ensured maximum purity in the materials that were required for the NaOH solution to

absorb the CO2. I prepared solutions of 0.003 M and 0.01 M concentration by mixing,

respectively, 14.9988 g and 49.99625 g of NaOH with 125 liters of water at room temperature

using the formula of molarity (M):


]
M= (22)
@˜ ∗ ™2A !6

where M = molarity, GMW = gram molecular weight of NaOH, V = volume in liters, and wt =

weight of NaOH in grams. I conducted the NaOH solution CO2 absorption experiments at room

temperature on a molar basis by dissolving known weights of chemicals in water. I also noted the

total volume of water required to make each solution.

Experimental Procedure

I conducted two experimental runs for this work, the first with the 0.003 M NaOH

solution and the second with the 0.01 M concentration. Once I had prepared each aqueous

solution, I counter-currently treated each concentration with the air + CO2 gas mixture using a

0.5 HP electric motor pump at a rate of 14.39 liter.min−1 for each run. I controlled the solution

flow rate into the packed column using a control valve placed in the flow line. Under the gas

stream, compressed air from an external source and CO2 from a pressurized cylinder passed

through air and the CO2 MFCs, respectively; the air blended with the CO2 to provide the capacity

for it to run until the end of the packed column, and I set the required gas flow rate using the

MFCs. Once I completed the counter-current operation of the NaOH solution and the air + CO2

mixture, the CO2 probes placed at the entering and leaving gas streams (see Figure 20)

transmitted the real-time data to the CO2 analyzer. The analyzer then transmitted the data to the

FieldPoint and further to LabVIEW through the LAN. I obtained the amounts of CO2 absorbed

39
through LabVIEW (see Figures 24 and 25) by subtract the percentage of CO2 entering the gas

stream with that of the exiting CO2 percentage.

40
CHAPTER 6

RESULTS AND DISCUSSIONS

This thesis provides the basis for concerted theoretical and experimental calculations in

mass transfer and process control structure design for a randomly packed column by presenting

in-depth mass transfer analysis of a specific randomly packed column and process control

analysis of a single-variable first-order plus dead-time model. The theories of mass transfer

referred to have been defined in the mass transfer and Hatta number sections earlier, and I

compare my results here. The process control model applied on the randomly packed column in

this work are defined precisely in the mass transfer coefficient section, and I discuss those results

here as well. In fact, the basic results for my process model can be broken down into mass

transfer and process control, which I discuss in detail.

Mass Transfer

As noted in the section on the methodology of mass transfer, all calculations refer to the

gas phase. Thus, it was convenient to calculate the overall gas phase mass transfer coefficients,

HOG and NOG, on the gas side. I calculated the overall K a (mol.m−3. s−1) for the packed column

based on the Hatta number, packed column height, NOG and HOG using equations (1) to (7). I

present the results for both NaOH solution concentrations in Table 5, and detailed calculations

are available in the Appendix (Tables 11, 12, 13, 14, and 15). Table 5 shows that the

experimental data are in the range of the literature findings and the experimental Ha exceeded 2

at both the 0.003 M and 0.01 M concentrations of the NaOH solution. Following [17–19], Ha

greater than 2 indicates a rapid reaction between the CO2 and the NaOH solution.

41
TABLE 5. Hatta number in the Literature and the Experimental data

Data kind Concentration k k% D Hatta

of NaOH (M) (rate physical mass Diffusivity number

constant) transfer m2. s−1 (Ha)

(s−1) coefficient at 25oC √ ∗


=
(m.s−1)

(randomly

packed

columns)

Literature > 0.001 ≥10 1 * 10−4 (1.6 * 10-9 to 1.8 * ≥2

10−9)

Experimental 0.003 and 15 1 * 10−4 (1.75 * 109, 1.79 * >2

0.01 10−9)

As shown in Figure 35, the concentration profiles of the rapid reactions plotted against

the mole fractions of the gas and liquid phases for the two runs have similar plots.

FIGURE 35. Gas-phase and liquid-phase concentration profiles for rapid reactions.

42
Figure 36 shows `3 , `9 : the mole fractions of the gas phase entering and leaving; š3 , š9 :

the mole fractions of the liquid phase entering and leaving; G1,G2: the mass flow rates of the gas

entering and leaving; and L2, L1: the liquid phases entering and leaving the column.

FIGURE 36. Gas phase and liquid phase distribution in the packed column.

Based on the Ha shown in Table 5, the reaction in the packed column was rapid, and the

available NaOH solution was consumed by the CO2. Thus, the mole fraction of the bulk liquid

phase was zero (x = 0); the Henry’s law relationship was y* = H x [6].

Both Figure 35 and Ha again, the bulk gas-phase mole fraction of CO2 in the NaOH

solution was also zero (x = 0), and thus, the corresponding equilibrium gas-phase mole fraction

(y*) was also zero according to Henry’s law.

Because the x above was zero, y* tended to zero, and then y3∗ = 0 and y ∗ 9 = 0. The gas

phase solute mole fractions in equilibrium with bulk phase liquid concentration x3 and x9 are

zero. Hence, equation (6) becomes equation (23) below:

25; 45< 6
y%@ = D , (23)
A œ ;•
D<

and substituting equation (23) in the denominator of equation (5) gives equation (24):
5; 45<
N0 = žŸ
. (24)

43
Equation (24) calculates the number of transfer units. The packed column was

predesigned, and packed column height Z was measured manually. Using the NOG from equation

(24) and the available Z, I calculated the height of the transfer units from equation (4). Table 6

summarizes the results of the total work.

TABLE 6. Results for the Experimental runs of both the 0.003 M and 0.01 M NaOH
solutions
Materials and terms 0.003 M 0.01 M Units

NaOH NaOH

Air 73.26 50 liter.min−1

CO2 6 5 liter.min−1

Volumetric flow rate of NaOH solution 14.38 14.38 liter.min−1

Molar volume of gas 24.31 24.31 liter. mole−1

Number of moles of air at STP 3.036 2.072 mole.min−1

Number of moles of CO2 at STP 0.2487 0.2072 mole.min−1

Fraction of CO2 entering the packed column 0.0834 0.10622 -

Fraction of CO2 leaving the packed column 0.03928 0.0281 -

Materials and terms 0.003 M 0.01 M Units

NaOH NaOH

Mole fraction of CO2 entering (`3 ) 0.0757 0.0909 -

Mole fraction of CO2 leaving (`9 ) 0.03928 0.0281 -

Number of moles of CO2 transferred (NA) 0.1245 0.1473 mole.min−1

Log mean mole fraction difference (y%@ ) 0.0555 0.0534 -

Total system pressure, (PT) 1 1 Atm

Height of packing, Z 3 3 m

44
TABLE 6. Continued

Materials and terms 0.003 M 0.01 M Units

NaOH NaOH

Cross section area, S 0.020096 0.020096 m2

Overall gas phase mass transfer coefficient, 37.20578 45.67184 mol.m−3. min−1

K a

HOG 4.572 2.554 m

NOG 0.656 1.174 -

Tower diameter 0.16 0.16 m

Height of packed column 4 4 m

Gas mass flow rate 0.2062 0.1718 mole.m−2. s−1

Liquid mass flow rate 0.0357 0.0357 mole.m−2. s−1

As noted earlier, detailed calculations appear in the Appendix (Tables 11–15). The

overall gas-phase K a data in my experiments are comparable with the industry standard data, as

shown in Table 7. The table supports that the process model I designed in this work is acceptable

for industrial-grade absorption systems.

TABLE 7. Typical KGa values from Literature data and Experiment conducted

Gas-phase reactant Liquid-phase reactant Industrial KGa Experimental KGa

(mol.min-1.m-3) (mol.min-1.m-3)

CO2 NaOH 25 to 75 37.27 to 45.67

The industrial operating conditions data are courtesy of the Norton Company. All the

time-based data are in seconds, but for comparison with the experimental data, I present K a in
45
minutes in Table 7. The industrial operating conditions were as follows: 25 mm plastic Hackett

tri-packs; solute gases, 0.5–1.0 percent mole; pressure, 1 atm; temperature, 16–25 °C; gas rate =

1.3 kg (m2. s)−1 = 1.1 m. s−1; liquid rates = 3.4 to 6.8 kg (m2. s)−1; packed height, 3.00 m; and

tower diameter, 0.76 m. These are the specific industrial-grade operating conditions I scaled

down to the operating conditions for my experiments for this research.

Process Control

For this experiment, I designed a process control model to control the flow rate of the

NaOH solution from reacting away with the continuously supplied CO2, using a control valve

with an actuator to control the NaOH solution flow rate. I also used a PID controller feedback

loop to automatically control the flow rate, and I discuss the IMC and Ziegler–Nichols tuning

method results below:

• Figures 37 and 38 show the step change and open-loop controller plots of the two

different NaOH solution concentrations (0.003 M and 0.01 M), which subsequently

assisted me in determining the process model parameters using the LOOP-PRO Control

Station.

• I present the LOOP-PRO open-loop process model data for the two NaOH concentrations

in Table 8. I further processed the data using IMC tuning.

IMC Tuning Results

The open-loop process model parameters process gain (K O ), time constant (τO ), and dead

time (Өp6 shown in Table 8 assisted me in establishing the closed-loop PID controller settings

using the relationships in Table 1 and in choosing values for τ# and τy = τ[ . Table 9 shows the

PID settings for the two concentrations after the IMC tuning.

46
FIGURE 37. LOOP-PRO results for the 0.003 M NaOH concentration data.

FIGURE 38. LOOP-PRO results for the 0.01 M NaOH concentration data.

47
TABLE 8. Process Model Data for the Open-Loop System obtained from LOOP-PRO

Parameters 0.003 M NaOH run value 0.01 M NaOH run value

KO 0.9644 0.7911

τO 86.41 s 36.58 s

Өp 45.49 s 10 s

TABLE 9. IMC tuning for PID Controller Settings


IMC tuning parameters 0.003 M NaOH run value 0.01 M NaOH run value

KC 1.368 2.313

τ¡ 20 s 10 s

τy = τ[ 86.41 s 36.6 s

Ziegler–Nichols Tuning Results

After I obtained the data shown in Table 8 from LOOP-PRO Control Station, I processed

them further using frequency response analysis (see Appendix Tables 16 and 17) and Bode

diagrams. The Bode diagrams of the frequency response analysis results for both the 0.003 M

and 0.01 M NaOH concentrations are presented in Figures 39 and 40.

FIGURE 39. Bode diagrams for the 0.003 M NaOH concentration.

48
FIGURE 40. Bode diagrams for the 0.01 M NaOH concentration.

Table 10 presents the Z–N tuning rules I arrived at after my frequency response analysis

and using the Bode diagrams and the PID control parameters from Table 2, and again, the

complete data on determining the critical frequency response values are available in Tables 16

and 17 of the Appendix.

TABLE 10 . Ziegler–Nichols tuning for PID Controller Settings

Z-N Parameters 0.003 M NaOH run value 0.01 M NaOH run value

KCU 3.7853 8.09

TU 154.63 s 36.38 s

KC 1.14 2.42

τy 128.86 s 30.31 s

The PID controller settings from both the IMC and Ziegler–Nichols tuning were

sufficiently close for me to set the controller to optimum operating conditions. I achieved PID

control in the experiment, that is, automatic control of the controller using the parameters I

obtained from the two tuning methods; see Figures 41 and 42.

49
FIGURE 41. PID-controlled set point tracking of the 0.003 M concentrate NaOH solution.
Note: The Ziegler–Nichols tuning parameters were moderate, −(KC = 1.14,τI = 128.86 s).

FIGURE 42. PID-controlled set point tracking of the 0.01 M concentrate NaOH solution.
Note: The Ziegler–Nichols tuning parameters were moderate, −(KC = 0.37, = 12.21 s)

50
CHAPTER 7

CONCLUSIONS AND FUTURE WORK

Conclusions

In this thesis, I present experimental data on the liquid–gas reactions in a randomly

packed column, specifically regarding the absorption of carbon dioxide into aqueous sodium

hydroxide at concentrations of 0.003 M and 0.01 M.

1. I used the data obtained from the mass flow controllers and CO2 analyzers to

calculate the overall gas phase mass transfer coefficients (K a).

2. Based on my Hatta number calculations, the reactions in this work were rapid.

3. Mass transfer coefficients increased with the increase in the liquid phase

concentration even with the decreased gas flow rates, and this was possibly

because of the rapid reactions in the column.

4. The mass transfer in the packed column was characterized by calculating the

overall gas phase mass transfer coefficients and the number and height of transfer

units.

5. Table 6 shows that I used the overall K a data I obtained to quantify the amount

of mass transfer from the gas to the liquid phase. Quantifying the mass transfer

helps in designing and manufacturing separate equipment that can meet specified

requirements, estimate potential real-world situations (e.g., chemical spill) and

potential outcomes.

6. I applied my experimental data on open-loop control to obtain the process model

parameters using LOOP-PRO Control Station and used the open-loop process

model parameters in Ziegler–Nichols and IMC tuning to obtain optimum PID

controller settings.

51
7. The work I present in this thesis was intended to identify the mass transfer in a

packed column using a case study of CO2 absorption with a sodium hydroxide

solution and to minimize the difficulties associated with process modeling, and I

was successful.

Future Work

Research could continue to identify design equations for randomly packed columns and

the limitations of the mass transfer and time constant (τ) relationships. My research focused on

characterizing a randomly packed column by revealing the design details, performing the

necessary mass transfer calculations, and developing a unified approach to process control, but I

leave the chemical kinetics calculations for future work. In addition, researchers could study the

stripping column adjacent to the packed column in the chemical engineering laboratory. It is well

known that the time constants for process models in dynamic response are inversely proportional

to the mass transfer coefficients, and I also leave the detailed development of the relationships

between dynamic response and mass transfer coefficients for future work.

52
APPENDICES

53
APPENDIX A

EXPERIMENTAL DATA OF AIR AND CO2 SUPPLIED TO THE PROCESS

54
1. Evaluation of Air and CO2 supplied in the experiment

Run Air CO2 TA molar volume moles of air at STP

number flow rate flow rate Absolute at STP, n air,

measured at measured at temperature VSTP = air flow rate*nstp-1

STP STP =

using MFC using MFC (22.4*T)*Ta-1

T, P (294.25 T, P (294.25

K ,1 atm) K ,1 atm)

(liter.min-1) (liter.min-1) o
C liter. mole-1 mole.min-1

1 (0.003M 73.26 6 273.15 24.1303 3.036

-NaOH)

2 (0.01M- 50 5 273.15 24.1303 2.072

NaOH)

2.Mole fraction evaluation of the experimental data

Run no number of moles of CO2 in number of moles of CO2 out

!"# Z0< ∗ fg
nCO2 in= CO2 flow rate*(VSTP )-1 nCO2 out= 234 !"# Z0< 6

mole.min-1 mole.min-1

0.003M NaOH 0.2486 0.1241

0.01 M-NaOH 0.2072 0.0599

55
3. Mole fraction and related data of the experiment

Run NaOH y1 y2 NA y%@

nCO9 nCO9 = nCO9 − 5; 45<


= =
no solution = D
A œ ;•
nCO9 + n" ! nCO9 + n" ! D<

nCO9

(mole.min-1)

1 0.003 M 0.0757 0.0393 0.1245 0.0555

2 0.01 M 0.0909 0.0281 0.1473 0.0535

4. System pressure, packing height, and cross section area data of packed column

PT Z S

Total system pressure Packing height Cross section area of the column

(atm) (m) (m2)

1 3 0.0201

1 3 0.0201

5.Mass transfer coefficients, number of transfer units, height of transfer unit

K 5; 45< M
K′Ga = M∗N∗O L∗5 KG a = K′Ga *PT NOG = HOG = K
P = 5 = b¢
-1 -3
(mol.min .m )
( mol.min-1.m-3.atm-1) (-) (m)

37.206 37.206 0.656 4.573

45.671 45.671 1.174 2.555

56
APPENDIX B

EXPERIMENTAL DATA OF ZIEGLER-NICHOLS TUNING METHOD

57
1. Ziegler-Nichols frequency response analysis for 0.003M NaOH solution

Ziegler-Nichols Tuning for PID controller

ω (radians. s-1) AR Φ (radians) Φ+π (radians)

0.01 0.729713 -1.16752 1.97407

0.02 0.483005 -1.95603 1.185559

0.03 0.347095 -2.56733 0.574265

0.04 0.268026 -3.10877 0.032825

0.040631 (critical value) 0.264177 -3.14163 -4.1E-05

0.06 0.182646 -4.10966 -0.96806

0.07 0.157304 -4.59125 -1.44966

0.08 0.138072 -5.06633 -1.92474

0.09 0.122996 -5.53701 -2.39542

0.1 0.110868 -6.00458 -2.86299

58
2. Ziegler-Nichols frequency response analysis data for 0.01M NaOH solution

Ziegler-Nichols Tuning for PID controller

ω (radians. s-1) AR Φ (radians) Φ+π (radians)

0.07 0.287611 -1.89866 1.24293

0.08 0.255651 -2.04169 1.099906

0.09 0.229778 -2.17606 0.965536

0.1 0.208479 -2.30408 0.837511

0.172712 (critical value) 0.123596 -3.14102 0.000574

0.3 0.071743 -4.47997 -1.33838

0.4 0.053904 -5.5026 -2.361

59
REFERENCES

60
REFERENCES

[1] E. Birch, “A review of ‘climate change 2014: Impacts, adaptation, and vulnerability’ and
‘climate change 2014: mitigation of climate change,’” Journal of the American Planning
Association, vol. 80, no. 2, pp. 184–185, 2014.

[2] A. Kean, R.Harley, D. Litttlejohn, and G.Kendall, “On-road measurement of ammonia and
other motor vehicle exhaust emissions,” Environmental Science & Technology, vol. 34,
no. 17, pp. 3535–3539, 2000.

[3] F. Popescu, “A performant state-of-art tool to assess cross-border impact of industrial


activities, a transboundary air pollution case study,” Procedia Technology, vol. 22, pp.
440–444, 2016.

[4] K. Darjee, “Climate change: miserable greenhouse gases,” The Initiation, vol. 2, no. 1,pp 30-
100, 2010.

[5] T. Suzuki. “The greenhouse warming and the impact of CO2 emission limitation on energy
economics,” Journal of the Fuel Society of Japan, vol. 68, no. 4, pp. 268–280, 1989.

[6] J. Johnson, “EPA sets schedule for CO2 regulation,” Chemical & Engineering News, p.
112410122417, 2011.

[7] U.S. Energy Information Administration (EIA). (2016). Environment.

[8] A. Jha. (2008, June). The cost of cleaning up fossil fuels - and the price of doing
nothing. The Guardian. [Online]. Available: https://www.theguardian.com/environment/
2008/jun/13/carboncapturestorage.fossilfuels.

[9] S. Choi, J. Drese and C. Jones, “Adsorbent materials for carbon dioxide capture from large
anthropogenic point sources,” ChemSusChem, vol. 2, no. 9, pp. 796–854, Sep. 2009.
Digital object finder: 10.1002/cssc.200900036

[10] F.S. Zeman and K.S. Lackner, “Capturing carbon dioxide directly from the
Atmosphere,” World Resources, vol. 16, no. 2, pp. 157–172, 2004.

[11] J.R. Jaunsen, “The behavior and capabilities of lithium hydroxide carbon dioxide scrubbers
in a deep-sea environment.” U.S. Naval Academy Technical Report, vol. 157, no. 2008-
06-17, 1989.

[12] Y. Ohashi, T. Ogawa and N. Egami, “Development of carbon dioxide removal system from
the flue gas of coal fired power plant.” Energy Procedia, vol. 4, pp. 29–34, 2011. Digital
object identifier: 10.1016/j.egypro.2011.01.019

61
[13] Li, Jian-Rong, “Carbon dioxide capture-related gas adsorption and separation in metal-
organic frameworks.” Coordination Chemistry Reviews, vol. 255, no. 15-16, pp. 1791–
1823, 2011.

[14] G. Rochelle, “Amine scrubbing for CO2 capture.” Science, vol. 325, no. 5948, pp. 1652–
1654, 2009.

[15] R. Treybal, Mass-Transfer Operations, 1st ed. New York: McGraw-Hill, 1980.

[16] R. Bird, Transport Phenomena, 2nd ed. New York: Wiley, 1960, pp. 520–570.

[17] S. Hartland, Counter-current extraction, 1st ed. Oxford: Pergamon Press, 1970, pp. 5–30.

[18] J. D. Seader and J. Ernest. Henley, Separation process principles. New York: Wiley, 2005.

[19] R. Perry, D. Green and J. Maloney, Perry’s Chemical Engineers’ Handbook. New York:
McGraw-Hill, 1984.

[20] D. Seborg, T. Edgar, D. Mellichamp and F. Doyle, Process Dynamics and Control, 3rd ed.
New York, NY: Wiley, 1989, pp. 1–2.

[21] D. Atherton. “Almost six decades in control engineering [Historical Perspectives].” IEEE
Control Systems Magazine, vol. 34, no. 6, pp. 103–110, 2014.

[22] P. Srinivas. “Modelling and simulation of complex control systems using


LabVIEW.” International Journal of Control Theory and Computer Modeling, vol. 2, no.
4, pp. 1–19, 2012.

[23] M. Foley et al. “A comparison of PID controller tuning methods.” The Canadian Journal of
Chemical Engineering, vol. 83, no. 4, pp. 712–722, 2008.

[24] D. Rivera, M. Morari and S. Skogestad, “Internal Model Control: PID controller
design.” Industrial & Engineering Chemistry Process Design and Development, vol. 25,
no. 1, pp. 252–265, 1986.

[25] T. Yucelen et al. “Self-tuning PID controller using Ziegler–Nichols method for
programmable logic controllers, in Proc. 1st Intl. Fed. Automatic Control Workshop on
Applications of Large Scale Industrial Systems, 2006, pp. 11-16.

[26] K. Åström and T. Hägglund. “Revisiting the Ziegler–Nichols step response method for PID
control.” Journal of Process Control, vol. 14, no. 6, pp. 635–650, 2004.

[27] T. Šekara and M. Mataušek. “Revisiting the Ziegler–Nichols process dynamics


characterization.” Journal of Process Control, vol. 20, no. 3, pp. 360–363, 2010.

62
[28] T. Cavicchi. “Phase margin revisited: phase-root locus, bode plots, and phase
shifters.” IEEE Transactions on Education, vol. 46, no. 1, pp. 168–176, 2003.

63

You might also like