Citation
Chemical engineering education

Material Information

Title:
Chemical engineering education
Alternate Title:
CEE
Abbreviated Title:
Chem. eng. educ.
Creator:
American Society for Engineering Education -- Chemical Engineering Division
Publisher:
Chemical Engineering Division, American Society for Engineering Education
Creation Date:
2003
Frequency:
Quarterly[1962-]
Annual[ FORMER 1960-1961]
Language:
English
Physical Description:
v. : ill. ; 22-28 cm.

Subjects

Subjects / Keywords:
Chemical engineering -- Study and teaching -- Periodicals ( lcsh )

Notes

Citation/Reference:
Chemical abstracts
Additional Physical Form:
Also issued online.
Dates or Sequential Designation:
1960-June 1964 ; v. 1, no. 1 (Oct. 1965)-
Numbering Peculiarities:
Publication suspended briefly: issue designated v. 1, no. 4 (June 1966) published Nov. 1967.
General Note:
Title from cover.
General Note:
Place of publication varies: Rochester, N.Y., 1965-1967; Gainesville, Fla., 1968-

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
01151209 ( OCLC )
70013732 ( LCCN )
0009-2479 ( ISSN )
Classification:
TP165 .C18 ( lcc )
660/.2/071 ( ddc )

UFDC Membership

Aggregations:
Chemical Engineering Documents

Downloads

This item has the following downloads:


Full Text









EDITORIAL AND BUSINESS ADDRESS:
Chemical Engineering Education
Department of Chemical Engineering
University of Florida Gainesville, FL 32611
PHONE and FAX: 352-392-0861
e-mail: cee@che.ufl.edu


EDITOR
Tim Anderson

ASSOCIATE EDITOR
Phillip C. Wankat

MANAGING EDITOR
Carole Yocum

PROBLEM EDITOR
James 0. Wilkes, U. 1.fi. 1/., i,

LEARNING IN INDUSTRY EDITOR
William J. Koros, GC .. -i.;, Institute of Technology


PUBLICATIONS BOARD --

CHAIRMAN *
E. Dendy Sloan, Jr.
Colorado School of Mines

MEMBERS *
Pablo Debenedetti
Princeton University
Dianne Dorland
Rowan University
Thomas F. Edgar
University of Texas at Austin
Richard M. Felder
North Carolina State University
Bruce A. Finlayson
University of Washington
H. Scott Fogler
University of Michigan
Carol K. Hall
North Carolina State University
William J. Koros
Georgia Institute of Technology
John P. O'Connell
University of Virginia
David F. Ollis
North Carolina State University
Ronald W. Rousseau
Georgia Institute of Technology
Stanley L Sandier
University of Delaware
Richard C. Seagrave
Iowa State University
C. Stewart Slater
Rowan University
Donald R. Woods
McMaster University


Spring 2003


Chemical Engineering Education

Volume 37 Number 2 Spring 2003


> DEPARTMENT
82 University of Maryland Baltimore County, Taryn Bayles, Douglas Frey,
Theresa Good, Mark Marten, Antonio Moreira, Gregory Payne, Govind Rao,
Julia Ross

> EDUCATOR
88 Robert H. (Rob) Davis of the University of Colorado, Christopher Bowman

> RANKINGS
94 Productivity and Quality Indicators for Highly Ranked ChE Graduate
Programs, Phillip E. Savage

> LABORATORY
100 Building Multivariable Process Control Intuition Using Control Station,
Douglas J. Cooper Danielle Dougherty, Robert Rice
142 Optimum Cooking of French Fry-Shaped Potatoes: A Classroom Study of Heat
and Mass Transfer, Jimmy L. Smart
154 Using a Commercial Movie for an Educational Experience: An Alternative
Laboratory Exercise, Martin J. Pitt, Janet E. Robinson

> RANDOM THOUGHTS
106 FAQS. VI: Evaluating Teaching and Converting the Masses,
Richard M. Felder, Rebecca Brent

> CLASSROOM
108 A Solids Product Engineering Design Project, Dhermesh V Patel, Agba D.
Salman, Martin J. Pitt, M.J. Hounslow, I. Hayati
126 Mathematical Modeling and Process Control of Distributed Parameter Systems:
Case Study. The One-Dimensional Heated Rod,
Laurent Simon, Norman W Loney
136 Personalized, Interactive, Take-Home Examinations for Students Studying
Experimental Design, William A. Jacoby
156 Using Molecular-Level Simulations to Determine Diffusivities in the Class-
room, D.J. Keffer, Austin Newman, Parag .

> CURRICULUM
114 Collaborative Learning and Cyber-Cooperation in Multidisciplinary Projects,
Jetse C. i:.. .. Hendry Siepe, Liya E. Yu, Chi-Hwa Wang
132 Process Simulation and McCabe-Thiele Modeling: Specific Roles in the
Learning Process, Kevin D. Dahm

> OUTREACH
122 The Value of Good Recommendation Letters, Gary L. Foutch

> CLASS AND HOME PROBLEMS
148 An Exercise for Practicing Programming in the ChE Curriculum: Calculation of
Thermodynamic Properties Using the Redlich-Kwong Equation of State,
Mordechai Shacham, Neima Brauner Michael B. Cutlip

120 ChE Division of ASEE Program for Annual Conference
124 Letter to the Editor
125 Call for Papers

CHEMICAL ENGINEERING EDUCATION (ISSN 0009-2479) is published quarterly by the Chemical Engineering
Division, American Society for Engineering Education, and is edited at the University of Florida. Correspondence
regarding editorial matter, circulation, and changes of address should be sent to CEE, Chemical Engineering Department,
University of Florida, Gainesville, FL 32611-6005. Copyright 2003 by the Chemical Engineering Division, American
Society for Engineering Education. The statements and opinions expressed in this periodical are those of the writers and not
necessarily those of the ChE Division, ASEE, which body assumes no responsibility for them. Defective copies replaced if
notified within 120 days of publication. Write for information on subscription costs and for back copy costs and availability.
POSTMASTER: Send address changes to Chemical Engineering Education, ChemicalEngineering Department., University
of Florida, Gainesville, FL 32611-6005. Periodicals Postage Paid at Gainesville, Florida and additional post offices.










[ef department


ChE at


University of Maryland


Baltimore County


TARYN BAYLES, DOUGLAS FREY, THERESA GOOD, MARK MARTEN, ANTONIO MOREIRA, GREGORY
PAYNE, GOVIND RAO, AND JULIA ROSS
University of Maryland Baltimore County Baltimore, MD 21250


It all began twenty years ago. An MOU (Memorandum of
[tludl ,ii.idin-.ii, was signed in 1983 that created a satel-
lite program in engineering at the University of Mary-
land Baltimore County (UMBC) campus. There was only one
state-supported College of Engineering in Maryland at that time,
at the University of Maryland College Park (UMCP), but in the
late seventies and early eighties, sufficient economic develop-
ment had taken place in the Baltimore region to draw legisla-
tive attention to the educational needs of the Baltimore region.
The original program created in 1983 envisaged the UMBC
operation as a satellite campus, with an Associate Dean re-
porting to the Dean of Engineering at UMCP. Programs were
set up in mechanical, chemical, and electrical engineering,
with program directors in charge who would report to the
respective department chairs at UMCP. The BS degree was
approved in 1985 and the MS/PhD degree in 1986.
The founding fathers in chemical engineering wisely de-
cided to call the UMBC program "Chemical and Biochemi-
cal Engineering" and made a strategic early decision to focus


the graduate program exclusively on biochemical engineer-
ing, while offering the undergraduate degree in traditional
chemical engineering. In 1986, Greg Payne joined the fac-
ulty as the first "bio" hire, followed in 1987 by Govind Rao.
The program subsequently grew rapidly, with several addi-
tional hires joining the faculty (due to space limitations, only
current faculty are mentioned). By 1991, engineering at
UMBC had grown sufficiently to necessitate the creation of
a freestanding college with its own dean, and the programs
were renamed as "Departments" with corresponding "Chairs."
The bio focus has turned out to be a great boon for the
department. UMBC was the first chemical engineering de-
partment in the country to have such a focus, and it continues
to this day to be the country's only chemical engineering de-
partment to focus its graduate program exclusively on the
bio area. From the beginning, this specialization attracted a
great deal of attention, particularly from prominent biochemi-
cal engineering faculty at other institutions. One of the most
exciting moments in our young history was when Professor


Copyright ChE Division ofASEE 2003


Chemical Engineering Education











Daniel Wang from MIT spent half of his first (and
only!) sabbatical at UMBC (with the other half spent
at CalTech). We learned a great deal from him and
through similar interaction with Professors Arthur
Humphrey and Michael Shuler. Interestingly, a com-
mon thread of advice from all of these distinguished
visitors during our formative years was to stay the
course and keep building the program, and to resist
the temptation to move into non-bio areas. Everyone
felt that the concentration of faculty in the bio area
and the unique location of UMBC in a bio-dense re-
gion of the country would eventually result in a strong
and vibrant department.

THE PRESENT
The department's more recent history has proven
that the strategy of focusing its graduate program
exclusively on the bio area was a sound decision.
Although the department went through its share of
growing pains and tough times in the beginning, the
end result is a strong and stable department with ex-
ceptional facilities and equipment and outstanding
faculty, staff, and students. For example, all faculty
members in the department have active research pro-


Peter Harms (N5ar (Gracauate iellowJ aadusts a nign
throughput microbioreactor.

grams with substantial external funding, and every eligible junior fac-
ulty member has received an NSF CAREER award. Table 1 lists the
current faculty and staff in the department, along with their interests
and responsibilities.
A great asset of being a high-profile department at a relatively small
institution (see UMBC profile in Table 2) is an unusually close con-


TABLE 1
Current Personnel at UMBC


* Dr. Taryn Bayles Lecturer
BS, New Mexico State University
MS (Petroleum), MS and PhD, University of Pittsburgh
Undergraduate education and outreach; transport phenomena

* Dr. Douglas Frey Professor
BS, .. -. 'University
MS and PhD, University of ..' .... Berkeley
Chromatography of biopolymers

* Dr. Theresa Good Associate Professor
BS, Bucknell University
MS, Cornell University
PhD, University of Wisconsin-Madison
Cellular engineering; optimization of chemotherapy and other
problems in biocomplexity

* Dr. Mark Marten Assistant Professor
BS, State University of New York, Buffalo
MS and PhD, Purdue University
Bioprocessing, proteomics, and genomics; microbial responses to
real-life environments


Support Staff
Mary Anderson IT Support Associate
Laurie Botto Office Assistant
Mike Frizzell Technician
Victor Fulda Technician
Denise Kedzierski Administrative Assistant


* Dr. Antonio Moreira Professor and Vice Provost
BS, University of Porto, Portugal
MS and PhD, University of Pennsylvania
Post Doc, University of Waterloo, Canada
Regulatory/GMP issues, scale-up; downstream processing

* Dr. Gregory Payne Professor
BS and MS, Cornell University
PhD, University of Michigan
Biomolecular engineering; renewable resources

* Dr. Govind Rao Professor and Chair
BS, IIT (Madras)
PhD, Drexel University
Fluorescence-based sensors and instrumentation; fermentation and
cell culture

* Dr. Julia Ross Associate Professor
BS, Purdue University
PhD, Rice University
Cell and tissue engineering; cell adhesion in microbial infection and
thrombosis


Research Faculty
Dr. Yordan Kostov Research Assistant Professor
Dr. Nandakumar Madayiputhiya Research Associate
Dr. Leah Tolosa Research Assistant Professor
Dr. Pyon Kyun Shin Research Associate
Dr. Haley Kermis Research Associate


Spring 2003











nection with administration. Everyone from the uni-
versity President on down is literally at arms reach
and is tremendously responsive and supportive of de-
partmental needs.
Another unusual aspect is the close ties our depart-
ment has with the Biology and Chemistry Depart-
ments as a result of many common faculty research
interests. At its inception, our department occupied
research space and facilities generously loaned to it
by the Chemistry Department, and it also received
strong support from the Biology Department. All of
our faculty members also participate in the Molecu-
lar and Cell Biology and in the Chemistry-Biology
Interface Programs at UMBC. These two programs
have resulted in biology graduate students working
in chemical/biochemical engineering laboratories and
vice versa, leading to a creative interdisciplinary mix
in our laboratories.


HIGHLIGHTS

We are fortunate to be at the leading edge of a revo-
lution. Biotechnology has become a dominant aspect
of the US economy. Indeed, just as the previous cen-
tury witnessed enormous strides in chemistry- and
physics-based technologies, this century is poised to
herald advances based on biology. The human ge-
nome has been sequenced, and unprecedented oppor-
tunities are opening up in the biotech/pharma world.
We plan to exploit these opportunities with a vigor-
ous research and education program that targets its
bioprocess aspects, and through bioengineering ap-
plications that focus on cellular interactions in dis-
ease-causing states.
Our current undergraduate curriculum (see Table
3, Column 1) has little to differentiate it from other
departments across the country that offer the chemi-
cal engineering major. This is changing, however. Our
bio-focused graduate research program, coupled with
enormous growth in the pharma/biotech industry, pro-
vided the inspiration for a new biotechnology/bioengi-
neering track at the undergraduate level that we be-
gan in 2001 (Table 3, Column 2). While we plan to
offer both the traditional track and the new track
within the chemical engineering major for the next
few years, we anticipate that the new track will ulti-
mately emerge as a new major, depending on enroll-
ment and acceptance of its graduates by employers
and graduate/medical schools.
An unusual aspect of UMBC's graduate offerings,
developed by Tony Moreira, is the four-course se-
quence in Biochemical Regulatory Engineering.
Regulatory Issues in Biotechnology


TABLE 2
UMBC Facts, 2002-2003

[1 President Freeman A. Hrabowski, III

[1 Faculty 680 full time and 350 part-time

a[ Students, Fall 2002
11,711 enrolled
Undergraduate, 9,549
Graduate, 2,162
Full-time, 8,779
Part-time, 2,932
Freshman Class 2002
First-time freshmen, 1,370
Living on campus (74%), 1,007
SAT percentiles
25th- 1120
75th 1290
Average SAT
Top Quartile 1374

a[ Chemical and Biochemical Engineering Statistics
Undergraduate, 100
Graduate, 34
Faculty, 10 FTE

E Academic Programs
UMBC offers 37 majors and 32 minors or certificate programs in the physical
and biological sciences, social and behavioral sciences, engineering, mathemat-
ics, information technology, humanities, and visual and performing arts. New
degree programs include environmental science, financial economics, and a
B.F.A. in acting.

UMBC's Graduate School offers 27 master's degree programs, 21 doctoral
degree programs, and seven graduate certificate programs. Programs are offered
in education, engineering, imaging and digital arts, information technology, life
sciences, psychology, public policy, and a host of other areas of interest. A new
gerontology PhD program is one of only six in the United States.

E Achievements
Ranked in top tier of nation's research universities-Doctoral/Research
Universities-Extensive-by the Carnegie Foundation
Six-time Pan-American Intercollegiate Team Chess champions
National Science Foundation ranking for federally funded research in science
and engineering jumped by nearly 50 places (from 200 to 153) in less than five
years
Named a "Hot School" by the 2003 Kaplan/Newsweek College Guide
Only Maryland university rated a "Best Value" by the 2001 Kaplan/Newsweek
College Guide
Ranked 16th nationwide in NASA funding
Named "Chess College of the Year" by Chess Life magazine in 2000
Won the NCAA Northeast Conference Commissioner's Cup in 1999, 2000,
2001, and 2002
Recognized as a college that builds character by The Templeton Guide
Awarded Phi Beta Kappa chapter in 1997
Only Howard Hughes Medical Institute Investigator at a Maryland public
university
Two-time recipient of U.S. Presidential Award for Excellence in Science,
Mathematics, and Engineering Mentoring
Consistently ranked among the top five research universities nationally in
production of bachelor's degrees in Information Technology
Designated a Center of Academic Excellence in Information Assurance by the
National Security Agency


Chemical Engineering Education











Good M,I ni,.. ni in.. Processes for Bioprocess
Quality Control and Quality Assurance for Biotechnol-
ogy Products
Biotechnology GMP Facility D. D.-..,. Construction, and
Validation
This course sequence is also available as a stand-alone cer-
tificate program that is highly sought after by biotechnology
industry professionals. Graduate students who complete this
certificate program are highly attractive to industry-these
issues are of critical importance to industry and programs of


this type are not generally available at most institu-
tions.
While the primary focus of our graduate program
is on PhD students, we are also mindful of industry's
need for trained Master's students. This, coupled with
an attractive integrated BS/MS option available to un-
dergraduates, will result in significantly more MS de-
grees being granted over the next few years. Ulti-
mately, this is primarily a resource issue, as the ma-
jority of the faculty is involved in long-term research


TABLE 3
BS Degree in Chemical Engineering: Traditional (left) and Bio (right) Tracks


Freshman Year
CHEM 101 Principles of Chemistry I (4) CHEM 101 Principles of Chemistry I (4)
MATH 151 Calculus and Analytic Geometry I (4) MATH 151 Calculus and Analytic Geometry I (4)
ENES 101 Introductory Engineering Science (3) ENES 101 Introductory Engineering Science (3)
GFR electives (6) GFR electives (6)

CHEM 102 Principles of Chemistry II (3) CHEM 102 Principles of Chemistry II (3)
CHEM 102L Introductory Chemistry Lab (2) CHEM 102L Introductory Chemistry Lab (2)
PHYS 121 Introductory Physics I (4) PHYS 121 Introductory Physics I (4)
MATH 152 Calculus and Analytic Geometry II (4) MATH 152 Calculus and Analytic Geometry II (4)
ENES 110 Statics (3) BIOL 100 Concepts of Biology (4)
GFR electives (3) GFR electives (3)

Sophomore Year
CHEM 351 Organic Chemistry I (3) CHEM 351 Organic Chemistry I (3)
ENCH 215 Chemical Engineering Analysis (3) ENCH 215 Chemical Engineering Analysis (3)
MATH 251 Multivariable Calculus (4) MATH 251 Multivariable Calculus (4)
PHYS 122 Introductory Physics II (4) BIOL 302 Molecular and General Genetics (4)

CHEM 351L Organic Chemistry Lab I (2) BIOL 303 Cell Biology (3)
MATH 225 Introduction to Differential Equations (3) BIOL 303L Cell Biology Laboratory (2)
Advanced Science elective (3) CHEM 352 Organic Chemistry II (3)
ENES 230 Introduction to Materials (3) MATH 225 Introduction to Differential Equations (3)
GFR electives (6) GFR electives (6)

Junior Year
CHEM 301 Physical Chemistry I (4) CHEM 301 Physical Chemistry I (4)
CHEM 311 Advanced Laboratory I (3) CHEM 437 Comprehensive Biochemistry I (4)
ENCH 300 Chemical Process Thermodynamics (3) ENCH 300 Chemical Process Thermodynamics (3)
ENCH 425 Transport Processes I (3) ENCH 425 Transport Processes I (3)
GFR electives (3) GFR elective (3)

CHEM 302 Physical Chemistry II (3) CHEM 438 Comprehensive Biochemistry II (4)
ENCH 427 Transport Processes II (3) ENCH 427 Transport Processes II (3)
ENCH 440 Chemical Engineering Kinetics (3) ENCH 440 Chemical Engineering Kinetics (3)
ENCH 442 Chemical Engineering Systems Analysis (3) ENCH 442 Chemical Engineering Systems Analysis (3)
ENGL 393 Technical Writing (3) ENGL 393 Technical Writing (3)

Senior Year


ENCH 437 Chemical Engineering Laboratory (3)
ENCH 444 Process Engineering Economics and Design I (3)
ENCH 445 Equilibrium Stage Computations (3)
ENCH XXX Chemical Engineering elective (3)
GFR electives (3)

ENCH 446 Process Engineering Economics and Design II (3)
ENCH XXX Chemical Engineering elective (3)
ENCH XXX Chemical Engineering elective (3)
GFR electives (6)


ENCH 444 Process Engineering Economics and Design I (3)
ENCH 445 Equilibrium Stage Computations (3)
ENCH XXX Bioengineering elective (3)
ENCH XXX Bioengineering elective (3)
GFR elective (3)

ENCH 446 Process Engineering Economics and Design II (3)
ECH 485L Bioengineering Laboratory (3)
ENCH XX Bioengineering elective (3)
GFR electives (6)


Spring 2003













UMBC was the first chemical engineering department in the country to have such a
focus, and it continues to this day to be this country's only chemical engineering
department to focus its graduate program exclusively in the bio area.


projects that require the continuity
and time investment of longer-term TA
PhD students. At the present time, Former Chemical/B
financial assistance is primarily di- Meyerh
rected at incoming PhD students "*" .. .
(with some exceptions). "()" indicates cur
How does a small department degree
handle so much? Part of the answer Stephanie Bates Cle
Stephanie Bates Clem
is Taryn Bayles, a full-time faculty Christy Butler Case
member devoted to education and
Adetokunbo Eniola Pi
outreach. Her infectious enthusi-
asm and energy are largely respon- Andre Johnson Emplc
sible for the high profile enjoyed Ray Onley Georgia Te
by the department. An example of Bradley Peterson MIT
her creative talents is demonstrated Lee Pitts Johns Hopki
by teaching innovations incorpo- Simone Stalling Penn
rated into her courses, such as a Kendra Sarratt Penn *
design project where freshman en- Jeremiah Tabb Georgi
gineering students had to build and Felicia Boone Employ
operate a water-balloon-launching Kafui Dzirasa Duke *
trebuchet that featured her as the Alexis Hillock Georgi
target! In addition, Taryn's out- Michael Johnson -UM]
reach efforts extend to several lo- Camelia Owens Dela\
cal schools and have served to in- Jason Pinnix Penn (I
crease both UMBC's visibility Natasha Powell -Unkn
and the community's awareness Marc Price Employed
of engineering. Frederick Scott UMB
In addition, several faculty mem- Jason Thorpe Georgia
bers are involved in electronic in-
structional media development. For
example, Doug Frey has developed a highly useful separa-
tions course web page that is available to anyone (found at
), and Julie
Ross, in collaboration with faculty in the medical school, is
developing innovative XML-based teaching modules.
We have close ties to industry-several faculty members
have research interactions with a number of pharmaceutical/
biotechnology companies. In addition, UMBC's location puts
us within an hour's drive of top-notch Federal facilities in-
cluding NIH, ONR, NIST, USDA, FDA, and DOD. Several
of our faculty members and students have benefitted by us-
ing these unique research facilities.

MEYERHOFF PROGRAM
UMBC is home to the nationally recognized Meyerhoff
program, which has a strong track record for graduating mi-
86


BLE
iochi
off S

*ntly w
require,

son U
Western
enn (
yed
ch (
* (Ph
ns (P
* (MD
(PhD
a Tech
'ed
(MD/]
a Tech
BC (
vare *
PhD)
own

C* (M
Tech


nority students and sending them on
4 to top-ranked PhD programs. The
emical Engineering program was started in 1994 by
scholars President Freeman Hrabowski with
... ... student, a grant from the Meyerhoff Foun-
'orking on graduate dation and has since attracted na-
ements tional recognition.

university MS To date, the Meyerhoff Scholarship
Reserve (MD/PhD) Program has produced 296 graduates
(the first degrees were awarded in
1993). One-hundred and forty-eight
students (148) are currently enrolled
MS) in PhD, MD/PhD, or other graduate
hD)* or professional degree programs at in-
PhD) stitutions ranging from Yale, Harvard,
)/PhD) and Stanford to MIT, Johns Hopkins,
) Carnegie Mellon, and Berkeley. An
S(PhD) additional 107 students have already
completed graduate-degree require-
PhD) ments and are working as research-
(PhD) ers and teachers at some of the finest
PhD) institutions and companies in the
(PhD) world. Research studies have demon-
strated that when compared to a
(MD/PhD) sample of high-achieving non-
Meyerhoff African-American stu-
[D) dents, Meyerhoff scholars have a sig-
* (PhD) nificantly higher incidence of attend-
ing medical school or graduate school
in the sciences, engineering, or math.
These findings have been substanti-
ated by the fact that the National Science Foundation and the
National Institutes of Health have identified UMBC, a pre-
dominantly white institution, as having one of the most ef-
fective programs contributing to minority-student success in
science in the nation. Table 4 lists the Meyerhoff students
from Chemical & Biochemical Engineering.


LUMPKIN MEMORIAL LECTURE
Janice Antoine Lumpkin was one of the first African-Ameri-
can female faculty members in the chemical engineering field
in this country. She graduated from MIT and Penn with BS
and PhD degrees, respectively, and joined UMBC in 1989,
initially as a part-time faculty member. She later converted
to a full-time position and brought her catalysis skills to bear
on understanding the mechanisms and kinetics of protein
Chemical Engineering Education











4 Sungmun Lee (left), Theresa Good (center), and Wanida
Wattanakaroon (right) purifying and testing
photoimmuno conjugates for
T-cell cancer treatment.


Graduate student Swapnil Bhargava (right) instructs
undergraduate Seth Miller (left) on the operation
of a 20-liter fermentor in Mark Marten's lab. V


lI1 /ir


oxidation. Tragically, she passed away in 1997 after the birth
of her fourth child. The department has honored her memory
in the form of a high-profile memorial lecture that is part of
UMBC's annual Life Sciences Day celebration. An eminent
person is invited to deliver the Lumpkin Memorial Lecture for
this celebration. Past Lumpkin lecturers include Arthur
Humphrey, Daniel Wang, Douglas Lauffenberger, Sangtae Kim,
and Barry Buckland. AIChE has also instituted a travel award
in her name for attendance at its annual national meeting.

THE FUTURE
We share a sense of excitement and anticipation about the
future. Biotechnology is transforming life as its early prom-
ise is maturing. There is an unusual atmosphere shared by all
members of this department-indeed, the feeling one gets is
more like being in a small biotech company than in a tradi-
tional university setting. Our strengths and the challenges we
face as we look into the future are
Strengths
Focus on Biotechnology and Bioengineering: this is a
major factor in our ability to achieve excellence. A
traditional chemical engineering department faces
competition for resources from other subspecialties
such as catalysis, polymers, etc. This is never an
issue for us.
Outstanding Faculty: Our faculty members are as
productive as those at higher-ranked peer institutions.
We are a young group and are aggressive and
passionate about both research and teaching. Further-
more, the environment is extremely collegial and
friendly.
Well-Equipped Laboratories: Our research areas are
well supported with state-of-the-art equipment, and
we truly have unmatched equipment resources
Spring 2003


compared to our much higher-ranked peers. Again,
this is partly due to our focus on one area.
Outstanding Geographical Location: We are located
in an area where biotech-driven growth is inevitable,
given our proximity to leading biomedical and
biotechnology companies. Maryland ranks third in
the nation for the number of biotech companies
located in a state.
Outstanding Foreign Graduate Students: UMBC is
just about the only chemical engineering department
that can guarantee an incoming graduate student that
he or she will work on a bio-related project. This
gives us a significant competitive edge in attracting
students.
Challenges
Obtaining greater resources for building on our base
in a tough budget environment.
Few domestic graduate students-a situation that is
not unique to us and that is slowly changing
Growth in the number of faculty members. We would
like to do more!

ACKNOWLEDMENTS
We thank Tim Ford for the photographs and Greg Simmons
for the Meyerhoff Program statistics. J


ff- f









educator


CHRISTOPHER BOWMAN
University of Colorado Boulder, CO 80309-0424
As engineering faculty, each one of us is asked to per-
form at an exceptional level in research, education,
and service to our universities and to our profession.
These tasks often seem to be in conflict, and time pressures
often force each of us to focus on one aspect at the expense
of the others. For the eleven years that I have been at the
University of Colorado, however, I have witnessed and
worked with one faculty member who personifies those ide-
als-one who is committed to research at the highest level,
to educating undergraduate and graduate students in the class-
room and through the discovery process, and to serving his
colleagues, his university, and his profession.
That person is Professor Robert H. Davis, Dean of the Col-
lege of Engineering and Applied Science and Patten Profes-
sor of Chemical Engineering at the University of Colorado.
He has been a prototype for what a faculty member should


be during his twenty years on the faculty. In fact, he is the
only faculty member in the 110-year history of the College
of Engineering and Applied Science at the University of Colo-
rado who has received all three College awards for Outstand-
ing Research, Teaching, and Service. He has not only dem-
onstrated exceptional performance in each of those individual
areas, but he has also focused on the synergistic interaction
that exists between them.
As a hallmark of his career, Rob has worked tirelessly to
develop programs that use research to assist educational ef-
forts and to develop educational programs that impact research
efforts. In addition to numerous research, teaching, and ser-
vice awards within the University of Colorado, he has also
been recognized with several national awards, including (most
recently) the American Society for Engineering Education's
Dow Lectureship Award.


Copyright ChE Division ofASEE 2003


Chemical Engineering Education


Robert



H.



(Rob)



Davis


of the
University of Colorado













Two of his favorite faculty
from U.C. Davis, Ruben
Carbonell (left) and
Steve Whitaker (right)
relaxing on a 1978
road trip with Rob and his
Rob. Rob and his
Rob. PhD
advisor,
Andy Acrivos,
in Cesaria, Israel,
in 1984.
V


HISTORY
Rob was born on March 26, 1957, in Paris, France, where his
dad was stationed as a military advisor at the U.S. Embassy.
Within three months of his birth, his family moved back to the
United States, first to Garden City, New York, and then further
west to Walnut Creek, California, when he was three years old.
Fortunately, Rob was exposed to great educators throughout
his life; his mother taught college mathematics and his father
taught elementary school and piano after retiring from the Navy.
Rob attended Ygnacio Valley High School in Concord, Califor-
nia, where he was named the outstanding senior in both math-
ematics and science. When he entered the University of Califor-
nia at Davis, intending to major in either math or chemistry, the
teaching assistant for his freshman chemistry class si i.- icd il.ii
he could combine those subjects and major in chemical engi-
neering instead. Like many entering freshmen in our field, prior
to that time Rob 'had not heard the words chemical and engi-
,. in:. used together in the same sentence!'
Rob displayed an early knack for leadership at Davis. During
all four years he volunteered 15-20 hours a week to work with
junior-high and high-school students through Young Life. In his
senior year, he was President of the AIChE Student Chapter, which
hosted the regional AIChE Student Chapter Conference. He also
organized the First-Annual Kronecker Delta golf tournament,
named in honor of a "favorite" tensor used by Professor Steve
Whitaker in transport courses. Somehow, Rob also found time to
study, and he received the University Medal in 1978 as the out-
standing graduate from U.C. Davis in all disciplines.
For graduate school, Rob moved across the San Francisco Bay
to Stanford, where he had the good fortune of working with Pro-
fessor Andreas Acrivos. "I was the second in a line of several
PhD students who studied the Boycott Effict with Andy," Rob
notes, "which refers to the phenomenon of an enhanced clari-
fication rate in sedimentation vessels with inclined walls."
Rob's dissertation work involved a combination of theory and
experiment, a hallmark of his own research program ever
since that time.
Spring 2003


Before leaving Stanford for his postdoctoral position, Rob
interviewed for a number of faculty positions and ultimately
accepted an offer to come to the University of Colorado.
Interestingly, this interview and selection process became
the subject of an article written by Rich Felder regarding
his observations while he was spending his sabbatical at
Colorado."1 At the time, it was clear that Rob would be an
exceptional teacher, although his future research career and
success was not as obvious. Rob notes, "I have always loved
to teach, but I was less certain about research when I was
interviewing for a faculty position. Fortunately, I quickly
learned how much fun research can be, especially when
working with students."
More than twenty PhD students of Andy Acrivos have
gone on to successful academic careers, including several
(John Brady, Dave Leighton, Ashok Sangani, and Eric
Shaqfeh) who overlapped with Rob. Many of these students
did postdoctoral research in the Department of Applied
Mathematics and Theoretical Physics (DAMTP) at the
University of Cambridge, and Rob dutifully took up the
call after completing his PhD in 1982. He was a NATO
Postdoctoral Fellow at DAMTP for a year, working with
89










Rob's first responsibility after becoming Dean in July
2002 was to buy a tuxedo for the black-tie functions
that he and his wife, Shirley, would attend.


-1', r*
Rob enjoys teaching students of all ages, even if only
half of the class pays attention! With daughters
Grace (right) and Allie (left) in 1993.

Professor George Batchelor on particle aggregation and with Dr. John
Hinch on elastohydrodynamic collisions and rebound.
Rob has always enjoyed working with young people, both inside
and outside of the university setting. While in graduate school, he
continued to spend 15-20 hours a week (and often more) leading a
Young Life club. Young Life is a nondenominational Christian out-
reach to primarily non-church kids, and Rob led weekly club meetings, Bible
studies, camping trips, and social events, in addition to co-leading and train-
ing a team of other volunteers.
Near the end of his time in graduate school, Rob became a student leader of
the Menlo University Fellowship and met Shirley Giles, a member of the group.
They married in December 1982, a few months after Rob finished his PhD
and then part of his Postdoctoral year, while Shirley completed a BA in
Communications from Stanford and then a mission experience in Banga-
lore, India. Rob and Shirley returned to the United States in late summer
1983 and moved to Colorado for Rob to begin the faculty position he had
lined up the year before.
Shortly after moving from England to Colorado, Rob and Shirley began
doing volunteer work with the high school program of the First Presbyterian
Church in Boulder. After a year, they began working with the University Chris-
tian Fellowship, a program for CU-Boulder students sponsored by the same
church. Rob was the volunteer director of this program for several years, and
he and Shirley continue to be associates in the program. Their activities over
the years have included teaching a Sunday class, leading Bible studies, hous-
ing interns, organizing retreats, and chairing the Messenger Committee to send
teams of university students on summer projects in foreign countries.
Rob was promoted from Assistant to Associate Professor after only five
years on the faculty and was promoted to full professor in 1992. In 1990-91,
he received a Guggenheim Fellowship for his first sabbatical, which he took
at the Massachusetts Institute of Tccin 'l, :,. At MIT, he enjoyed interactions
with Professors Bob Armstrong, Howard Brenner, Bob Brown, Clark Colton,
and Greg Stephanopoulos, among others, as well as with Howard Stone at
Harvard University. "I also enjoyed getting to know several bright PhD stu-
dents and postdocs," Rob recalls, "including Nick Abbott, Stephanie
Dungan, Gareth McKinley, and Ron Phillips, who have all gone on to
90


Daughters Grace and Allie today, well on
their way to being teenagers, on a
trip to Santa Barbara.


Chemical Engineering Education











"Punting" on the
river Cam, a welcome
break from
postdoctoral
studies
at the
University of
Cambridge
in 1982-83. V


Rob (on the left) leading songs for a Young
Life retreat in 1980, with Robert Aguirre
(now a Professor of English).


Rob in his Stanford office in
1982, explaining the concept
of inclined settling. The T-
shirt depicts his love of
bicycling-he still rides
a bike to work
every day!


successful academic careers."
During this year at MIT, Rob and Shirley lived in
the Back Bay area of Boston. While Rob walked across
the Massachusetts Avenue bridge over the Charles
River to MIT, Shirley walked upriver to Boston Uni-
versity, where she completed an MA degree in broad-
cast journalism.
After they returned to Colorado, their first daugh-
ter, Grace, was born in December of 1991, followed
by their second daughter, Allison, born in June of 1993.
"I never thought that I would enjoy young children as
much as I enjoyed high-school and college students,"
Rob says, "but I've changed my mind, now that I have
my own children." In the year between his daughters'
births, Rob became Department Chair (1992). Al-
though his teaching load was slightly reduced to ac-
commodate his new activities, throughout his ten years
as department chair, Rob maintained his research pro-
gram at its usual high level.
Rob took his second sabbatical in 1997-98, this time
at the University of California at Santa Barbara, hosted
Spring 2003


by Professor Gary Leal. Besides providing time for uninterrupted re-
search, it was also a great opportunity for Rob to spend more time
with Shirley and their young daughters. He notes that they had a pic-
nic in their backyard or at the Goleta beach several evenings every
week. The close-knit family now often travels with Rob for confer-
ence/vacation trips, especially to foreign countries. Closer to home,
they love to camp, hike, bike, and ski, and Rob often brings the girls
with him when he can't stay away from the office on Saturdays!
More recently, Rob was appointed Dean of Engineering and Ap-
plied Science at the University of Colorado (July, 2002). While he
took this position out of a sense of duty to the institution that has served
him well for the past twenty years, he has found his new responsibili-
ties "surprisingly fun." In the current economic climate of limited re-
sources for the traditional "dean-type" activities of adding new build-
ings, supporting new initiatives, and increasing the faculty, he remains
excited about the challenges of nurturing faculty for excellence in both
teaching and research, educating students in both traditional and ac-
tive-learning environments, and allocating resources wisely to invest
in excellence for the long term.
"I expect to be Dean for ten, plus or minus eight, years," Rob jokes,
"so making personal plans for the future is difficult." He anticipates
continuing a vibrant research program, although perhaps more modest
in size. His current research group consists of nine PhD students and
two research associates. Rob hopes to return to classroom teaching
someday and plans to remain active in serving the profession. Most
importantly, we expect Rob to continue to balance his priorities of
family and faith along with his service to students, faculty, and the
profession.













The research that he has performed and the


EDUCATION
Rob is an outstanding classroom teacher and has
won several departmental and college-wide teaching
awards. He is respected by students for his high stan-
dards, superb organization, compelling lectures and
demonstrations, as well as his compassion and fair-
ness. In fact, Professor Bill Bentley (University of
Maryland), a former PhD student who also had Rob
as a professor, indicates that "Rob was singularly the
best educator I've ever encountered, anywhere."
The lasting influence of Rob's educational work in-
cludes a half-dozen publications on teaching methods
in peer-reviewed journals, the development of six new
courses (five that are now taught by other faculty),
organization of a special issue of Chemical Fi .ii. I-
ing Education on teaching fluid-particle tclkmiii ,h -.,,
and development of the Interdisciplinary Biotechnol-
ogy Program at the University of Colorado. Addition-
ally, he directs or co-directs three Graduate Assistant-
ships in Areas of National Need (GAANN) programs
funded by the U.S. Department of Education, which
support graduate-student training throughout the De-
partment of Chemical Engineering. As part of these
programs, Rob thoroughly enjoys taking the students
on retreats and road trips. Despite his recent ascen-
sion descensionn?) to the Deanship, Rob has contin-
ued to be active in these programs, including attend-
ing the retreats and other student interactions.
Rob is also an outstanding mentor and spends count-
less hours helping students and young faculty to think
critically, to learn through discovery, and to commu-
nicate effectively. For the past three years, he has
served as a faculty mentor to graduate students par-
ticipating in an NSF-funded outreach program to lo-
cal high schools and middle schools. He has also been
research mentor to over 120 undergraduates, 50 gradu-
ate students, and 10 postdocs. As one significant mea-
sure of his success and lasting impact, ten of his former
graduate students and postdocs are now full-time fac-
ulty members. As has been noted by several of these
former students, the framework that Rob estab-
lished, his mentoring style, and his concern for his
students are all aspects that these former students
hope to emulate.

RESEARCH
Rob's research philosophy is to perform fundamen-
tal research on problems selected from or motivated
by practical engineering applications. He is a world
92


Rob with some members of his research group on a hike in
the Colorado Rocky Mountains in 2000.


leader in the hydrodynamics of complex fluids, and his group has
applied fundamental theory and principles in this area to an astonish-
ing variety of problems.
In his twenty-plus-year academic career, Rob has published more
than 160 papers and has received over $18 million in grants to sup-
port his research program. Worth noting is the fact that, as evidenced
by his references, publications, and funding, he has had a significant
impact on three distinct research areas: fluid mechanics, biotechnol-
ogy, and membrane separations. As one example of his creativity, Rob
and a PhD student, Kim Ogden (now at the University of Arizona),
showed that productive cells could be separated from unproductive
cells and recycled in a continuous-flow bioreactor by coupling ge-
netic markers for flocculation with the gene for the product of inter-
est, so that the productive cells settled rapidly as flocs with fractal
structures. Rob and his group later became the first to apply funda-
mental engineering principles to pioneer new bioreactor strategies for
enzymatic production of ribonucleic acids, by immobilizing DNA tem-
plates on small beads and then recovering both DNA and enzyme
(due to binding) along with the beads to achieve substantially im-
proved yields of RNA product.
As another example, Rob applied fundamental transport principles,
including the newly recognized phenomenon of shear-induced hydro-
Chemical Engineering Education













impact he has had on other lives will last for many lifetimes.


dynamic diffusion, to establish widely used models for
crossflow membrane filtration. More recently, his group has
developed and analyzed several novel strategies for mem-
brane-fouling control: rapid backpulsing, dynamic second-
ary membranes, and surface modification by photografting.
In more basic research on multiphase flow, Rob developed
the first elastohydrodynamic theory (with coupled solid and
fluid mechanics) for particle collisions with other particles
or surfaces in liquids or gases, to predict whether particle
rebound or adhesion occurs, and then later elucidated the fric-
tion/lubrication nature of particle contacts in liquids. This
pioneering work is now used in diverse fields such as granu-
lation, wet granular flow, suspension flow, and air filtration.
Moreover, his group has analyzed the related problems of
drop and bubble interactions in near contact, showing how
small deformations due to lubrication forces retard coales-
cence and how large deformations may promote alignment,
breakup, and/or coalescence.

SERVICE AND LEADERSHIP

When Rob became the Department Chair, it was one of the
best possible things that could happen to our department. As
Chair, Rob undertook a major program to improve the De-
partment in all areas and at all levels, including undergradu-
ate students and programs, graduate students and programs,
and faculty. Since Rob took over, the number and quality of
the undergraduate and graduate student populations have im-
proved, funding and publications per faculty member have
more than doubled, and the faculty has grown in size-half
of the current faculty were hired while Rob was Chair. Fac-
ulty have also received numerous national and international
awards from professional societies (Materials Research So-
ciety, AIChE, ACS, and ASEE) and foundations (Dreyfus,
Packard, Sloan, Howard Hughes Medical Institute) that rec-
ognize its progress, with most of these awards based on nomi-
nations that Rob carefully prepared for his colleagues. In fact,
in just the last three years, three different faculty have won
singular national awards from ASEE (two Curtis W. McGraw
Awards and Rob's selection as the 2002 Dow Lectureship
winner). The State of Colorado has also twice designated the
Department as a Program of Excellence.
Rob is a tireless advocate for chemical engineering educa-
tion and research, as well as for the people involved in those
activities. In addition to numerous responsibilities at the Uni-
versity of Colorado (including his service as Chair (1992-
2002), with only one sabbatical break, and now as Dean), his
professional activities have included organizing the IUTAM
Symposium on Hydrodynamic Diffusion of Suspended Par-
Spring 2003


tiles in 1995, the technical program of the AIChE Annual
Meeting in 1999, and the technical program of the North
American Membrane Society Annual Meeting in 2000. He
co-organized a series of workshops on "Teaching Fluid-Par-
ticle Processes" for the 1997 ASEE Summer School for
Chemical Engineering Faculty, and he served as Guest Edi-
tor of a special-feature section of Chemical Ei.-.i.,. I, .* Edu-
cation in 1998, which contained seven articles related to the
recommendations of this workshop. He also served as the
Director of the Colorado RNA Center (1992-2001) and co-
Director of the Colorado Institute for Research in Biotech-
nology (1987-2001), in statewide efforts to promote research,
student training, and industry/university cooperation, includ-
ing management of an annual symposium, seed grants pro-
gram, graduate fellowships, and student internships. Rob was
the co-Chair (along with Scott Fogler and Mike Cutlip) of
the 2002 ASEE Summer School for Chemical Engineering
Faculty, held last July at the University of Colorado.
In 1995, Rob was invited to make a presentation at the
AIChE Young Faculty Forum, and he chose the subject "Get-
ting Along With (and the most out of) Your Department Chair."
Based on session evaluations, his presentation received the
Outstanding Paper Award for the 1995 AIChE Annual Meet-
ing. As the co-Chair for that session, it was readily apparent
to me that Rob's advice to the younger faculty, as well as to
those aspiring to be young faculty, was extremely well re-
ceived. He was also not afraid to challenge the common as-
sumptions about what young faculty should do-he chal-
lenged them to participate in service activities that had a high
outcome-to-input ratio and not to simply neglect service un-
til after being tenured. Excerpts of his advice to young fac-
ulty are soon to be submitted as an article in Chemical Engi-
;. ., i ;,,. Education.

SUMMARY

If your vision is for one year, plant wheat.
If your vision is for ten years, plant trees.
If your vision is for a lifetime, plant people.
Old Chinese Proverb
In fact, that is exactly what Robert Davis has spent the last
twenty years doing! As a researcher, he has trained PhD and
undergraduate research students who will lead the next gen-
eration; as a teacher, he makes sure that his students know
the basic principles and fundamentals; and as a Department
Chair and Dean, he has mentored faculty and provided a
framework in which all are encouraged and enabled to be
successful. The research that he has performed and the im-
pact he has had on other lives will last for many lifetimes. 1
93











MRg^ rankings


PRODUCTIVITY AND QUALITY

INDICATORS

For Highly Ranked ChE Graduate Programs



PHILLIP E. SAVAGE
University of Michigan Ann Arbor, MI 48109-2136


Comparative assessments of graduate programs have
been made for at least eighty years. Such assessments
are useful to prospective students and to those seek-
ing an academic position. They are also used in the political
arena to make or justify policy and appropriations decisions.
Within engineering, the most visible rankings are those from
U.S. News,[1] the NRC Report,[2] and the Gourman report.[3]
The U.S. News ranking is arguably the best publicized and
most widely used ranking today.
U.S. News ranks the graduate programs for individual en-
gineering disciplines. These discipline-specific rankings are
based exclusively on a department's reputation as determined
from a peer-assessment survey. Engineering deans (or their
designees) nominate up to ten departments in a particular dis-
cipline (e.g., chemical -i-.iicc iii.;-. and the total number of
respondents who nominate a department determines its rank.
The most recent rankings11 of graduate programs was com-
piled in January 2002, based on data from a survey distrib-
uted in the fall of 2001.
This article expands the reputation-based U. S. News
rankings of chemical engineering departments by providing
and comparing quantitative quality and productivity indica-
tors for the top twenty chemical engineering departments in
its 2002 ranking. One objective of this study was to deter-
mine how well the rankings, which are based exclusively on
reputation, correlate with different publicly available produc-
tivity and quality indicators. A second objective was simply
to assemble the database of quantitative indicators, an exer-
cise that has not been completed for at least ten years.
The productivity indicators examined here are the number
of published articles and reviews and the number of bach-
elor, master, and doctoral degrees granted annually. The qual-
ity indicators are the number of NAE members, the number
of AIChE Institute awards received, the number of highly
cited papers, the number of citations per paper, and the total


number of citations to the department's published articles and
reviews. This last quantity is an indicator of both quality (ci-
tations) and productivity (number of publications).
The study also included data on the research expenditures
for each department. Some would contend that total research
expenditure is not an indicator of productivity or quality, but
research funding is a necessary input for a high-quality gradu-
ate program. Moreover, one could argue that the ability to
compete successfully for peer-reviewed federal funds is an
indicator of quality. Therefore, the study included data for
federally funded research expenditures for each department.
None of the indicators used in this study are perfect or com-
plete measures of quality or productivity. They are simply
quantities that most chemical engineering educators would
likely agree are among the most relevant indicators. Simi-
larly, the indicators used in this study do not constitute an
exhaustive set of all relevant indicators. Other relevant indi-
cators (e.g., non-AIChE awards, patents, faculty appointments
for PhD recipients, etc.) exist but were excluded here to make
the demands of data gathering consistent with the resources
available for the task.
Many of the indicators considered here have been used pre-
viously to rank graduate programs. Diamond and Graham[4]
argued that per capital citation density (citations per faculty
member) is perhaps the best single indicator of a program's

Phillip Savage is Professor of Chemical Engi-
neering at the University of Michigan. He re-
ceived his BS from Penn State in 1982 and his
MChE (1983) and PhD (1986) from the Univer-
sity of Delaware, all in Chemical Engineering.
His research and teaching interests focus on the
rates, mechanisms, and engineering of organic
chemical reactions. Current research deals with
reactions that can be used for hydrogen produc-
tion from liquid fuels, for environmentally benign
chemical synthesis and for waste treatment.


@ Copyright ChE Division ofASEE 2003


Chemical Engineering Education










excellence. Their article also provides an interesting discus-
sion of the history and limitations of subjective peer assess-
ments reputationall rankings). Angus, et al.,E1 proposed a rank-
ing system that uses data for publications, citations, research


support, and awards. Their system included a
greater variety of awards than NAE membership
and AIChE Institute awards, which are the only
awards considered here.
Both articles provided rankings of chemical
engineering programs. These rankings were based
on the publication and citation data that appeared
in the 1995 NRC report. The data were gathered
in 1993, so the rankings in these articles as well
as in the NRC report itself reflect the landscape
as it existed ten or more years ago. Additionally,
there were inaccuracies in some of the citation
data in the 1995 NRC report.E41
It is worthy of note that the NRC is currently
evaluating various methodologies for its next
comparative study of graduate programs at U.S.
universities, release of which is anticipated to be
in 2005. Given that at least a decade has passed
since a comprehensive set of indicators has been
assembled for chemical engineering graduate pro-
grams, we set out to develop such a database for
the top twenty programs in the U.S. News 2002


One obje
of this s
was t
determ
how wel
rankin
which
base
exclusive
reputat
correlate
differ
public
availa
product
and qu
indicate


rankings. One purpose in doing so is to assess the degree of
correspondence between the subjective rankings and the vari-
ous publicly available quantitative indicators.

METHODOLOGY
This study examines data for both productivity indicators
and quality indicators for the twenty chemical engineering
departments ranked by U.S. News. One of the quality indica-
tors is the number of faculty members in a chemical engi-
neering department who are also members of the National
Academy of Engineering (NAE). This information was com-
piled by comparing the list of NAE membersE61 at each insti-
tution with the list of faculty in each department.E17 A sec-
ond quality indicator is the number of AIChE Institute
awards that faculty members in a given chemical engi-
neering department in 2002 had received between the
years 1992 and 2001.81]
Additional productivity and quality indicators involve pub-
lications and citations. The average annual number of publi-
cations from each chemical engineering department was de-
termined for 2000 and 2001. These data were obtained from
a search of ISI's Web of Science.E91 The search was conducted
by department (or school) and not by each individual faculty
member. It provided all publications in which at least one
author self-identified with the specific department.
The search included only "articles" and "reviews." It pro-
Spring 2003


vided no data specifically for the chemical engineering pro-
grams at Caltech and at Minnesota. Caltech authors in both
chemistry and chemical engineering identified themselves
with the Division of Chemistry and Chemical Engineering.
Thus, the search returned publications for both
departments and no attempt was made to deter-
ctive mine the subset that could be attributed to chemi-
tudy cal engineering. Minnesota's chemical engineer-
0o ing program is part of the Department of Chemi-
ine cal Engineering and Materials Science, and the
search returned papers published by the entire
I the department. These departmental totals were in-
'gs, cluded in this study because the chemical engi-
are neering portion of that department is easily the
d larger of the two.
ly on The ISI database was also used to discover the
ion, total number of citations made to each "article"
with and "review" published by a given chemical en-
ent gineering department in 1998 and 1999. This
ely search also provided the total number of articles
ble and reviews published by that department dur-
ivity ing that two-year span. The number of citations
ality reported is the number as of the dates of the
ors. searches (May 23-24, 2002). Thus, the citation
S. statistics reported herein are for papers that had
been in print for two-and-one-half to four-and-
one-half years. There may be a benefit to using a
longer time in print for the citation analysis (capture more
completely the total impact of the articles), but there also
exists a disadvantage (using older papers makes the cita-
tion data less reflective of the impact of a department's
recent work).
The number of papers published by most departments in
1998/99 was within 10% of the number published in 2000/
01. Since the departmental publication rates are similar for
these four years, and since the citation statistics are for only a
two-year sample, the citation statistics are not likely to suffer
from a publication-rate-profile bias.i101 Moreover, the cita-
tion statistics presented herein are free of many of the "pit-
falls" enumerated by Grossmann.111 Other authorst4 51 have
also discussed the strengths and weaknesses of using ci-
tations as an indicator of quality so these issues will not
be rehashed here.
Note that ISI computes statistics for the citation impact in
chemical engineering for different institutions. These statis-
tics are determined from citations to all publications from a
given university in a set of journals ISI classifies as "chemi-
cal engineering" journals. Thus, a portion of the data will be
from articles contributed by other departments, and more
importantly, work published by chemical engineering faculty
will be excluded if it is published outside the traditional chemi-
cal engineering journals. It is for these reasons that this sta-
tistic was not used in the present study. Finally, note that one
95











could devise a scheme to calibrate the citation statistics (per-
haps using impact factors for journals or fields) to account
for field-to-field differences in citation frequency. This cali-
brated citation frequency could be a useful complement to
the total citation frequency reported here.
Another indicator of productivity is producing engineer-
ing graduates. The ASEE websiteE121 provided the number of
bachelor, master, and doctorate degrees, respectively, granted
in chemical engineering in 2000 and 2001. The data avail-
able for the University of Minnesota includes chemical engi-
neering and materials science together. This site also pro-
vided the number of full-time, tenured, or tenure-track fac-
ulty in each department for these two years. Note that these
data do not account for fractional academic appointments nor
do they include non-tenure-track faculty. Accurate data for
the number of faculty full-time equivalents in each depart-
ment would have been useful, but such data do not appear to
reside in a publicly available database.
Finally, the study included information regarding research
expenditures made by each department. The ASEE website
provided the total annual expenditure for 2000 and 2001.
These total research expenditure figures include both spon-
sored and internally funded research. No research expendi-
ture data were available on the ASEE website for Caltech,


Georgia Tech, or Northwestern. The National Science Foun-
dationE131 also compiled and reported research expenditure
data. The most current data are for fiscal year 2000, and both
the total and the federally sponsored research expenditures
are available for all of the departments of interest.

RESULTS AND DISCUSSION
Table 1 provides the data for each department. The first
column, "Rank," provides the U.S. News ranking. "NAE" is
the number of faculty members in a chemical engineering
department who are also members of the National Academy
of Engineering. The next column shows the number of AIChE
Institute awards that faculty members in a given chemical
engineering department in 2002 received between 1992 and
2001. The column "Pubs" shows the average annual number
of publications from each department. "B," "M,", and "D"
are the mean number of bachelor, master, and doctorate de-
grees, respectively, granted annually in chemical engineer-
ing for 2000 and 2001. "FTF" is the mean number of full-
time (tenured or tenure-track) faculty.
The first "Total Research Expenditure" column is an an-
nual average for 2000 and 2001, as compiled by ASEE, and
the other two Research Expenditure columns contain data
from NSFE131 for fiscal year 2000. The next column lists the


TABLE 1
Extensive Indicators for Chemical Engineering Departments


AIChE


NAE Awd Pubs B M D FTF


1. MIT
2. Minnesota3
3. UC Berkeley
4. Caltech
5. Wisconsin
6. Stanford
7. Texas
8. Delaware
9. Illinois
10. Princeton
11. Michigan
12. UC Santa Barbara
13. Georgia Tech
13. Purdue
15. Carnegie Mellon
16. Cornell
16. Pennsylvania
18. Northwestern
19. Penn State
20. Texas A&M


Research Expend. ($K)

Total Total2 Federal2


17,958
7,551
13,205


16,106
9,057
4,842


10,131
5,682
1,880


4 4 n.d. 11 6 6 10 n.d. 5,105 2,772
2 1 90 93 6 16 17 8,862 7,317 4,295
1 1 58 14 31 6 11 6,019 6,424 5,378
3 1 91 126 20 18 20 5,405 7,469 3,823
2 2 86 45 10 19 21 3,380 5,890 2,940
1 0 54 79 18 9 13 2,825 5,160 3,001
3 7 70 27 2 11 17 3,644 3,130 1,564
0 3 79 130 22 10 17 4,143 3,623 2,315
6 2 73 17 3 10 19 4,610 4,995 3,907
3 2 62 135 10 15 34 n.d. 5,938 2,460
0 1 63 112 8 12 21 6,699 6,624 2,403
3 2 78 41 9 13 19 3,603 3,379 2,223
1 0 36 58 12 8 13 3,397 3,020 1,647
2 2 38 30 14 7 9 1,738 1,777 1,300
1 2 45 46 6 10 15 n.d. 4,086 2,084
1 1 50 141 8 6 20 3,172 14,257 8,491
0 1 49 116 14 15 18 11,826 9,364 1,963


Cit. >50
Cit. per Pub Cites

3751 12.0 8
2283 6.7 2
1577 8.6 1
n.d. n.d. n.d.
1210 7.0 1
1068 10.8 2
1412 7.2 0
1168 6.8 2
675 5.8 0
1412 9.8 1
1267 8.6 5
2648 15.9 11
793 6.0 1
655 5.0 0
1029 7.3 0


718 6.1 1
381 4.9 0


Chemical Engineering Education


From ASEE data
From NSF data
For chemical engineering and materials science











total number of citations to all articles and reviews published
by a given department in 1998 and 1999. The mean number
of citations per research publication appears in the next col-
umn. This quantity was calculated as the total number of ci-
tations divided by the total number of articles published dur-
ing those two calendar years. The final column lists the total
number of articles in the sample that had been cited more


TABLE 2
Top Ten' Departments in Different Productivity or Quality Indic

Citations/Pub' Citations' Publications2 NAE Members Doctorate
1 UCSB MIT Minnesota MIT(9) Minn
2 MIT UCSB MIT Minnesota (7) M
3 Stanford Minnesota Berkeley UCSB (6) Dela
4 Princeton Berkeley Texas Caltech (4) Te
5 Pennsylvania Princeton Wisconsin Berkeley (3) Wise
6 Michigan Texas Delaware Texas (3) Georg
7 Berkeley Michigan Michigan Princeton (3) Texas
8 Cornell Wisconsin CMU CMU (3) Ber
9 CMU Delaware UCSB Georgia Tech (3) CT
10 Texas Stanford Princeton 3 depts w/2 Pur

1 Of the 20 ranked by U.S. News
2 Excluding Caltech because of lack of data


TABLE 3
Intensive Indicators for Chemical Engineering I

AIChE Reset
Rank NAE Awd Pubs B M D TotaP
1 MIT 0.28 0.31 4.14 2.42 1.18 1.03 553
2 Minnesota3 0.22 0.06 4.47 4.86 0.30 1.33 236
3 Berkeley 0.17 0.11 5.22 4.42 0.36 0.69 734
4 Caltech 0.40 0.40 n.d. 1.10 0.55 0.55 n.d.
5 Wisconsin 0.12 0.06 5.42 5.64 0.36 0.94 537
6 Stanford 0.09 0.09 5.27 1.23 2.82 0.55 547
7 Texas 0.15 0.05 4.55 6.28 1.00 0.88 270
8 Delaware 0.10 0.10 4.17 2.17 0.46 0.93 165
9 Illinois 0.08 0.00 4.28 6.28 1.44 0.68 226
10 Princeton 0.18 0.41 4.12 1.59 0.09 0.65 214
11 Michigan 0.00 0.18 4.62 7.62 1.29 0.59 244
12 UC Santa Barbara 0.32 0.11 3.84 0.89 0.16 0.50 243
13 Georgia Tech 0.09 0.06 1.85 4.01 0.30 0.43 n.d.
13 Purdue 0.00 0.05 2.98 5.31 0.38 0.55 319
15 Carnegie Mellon 0.16 0.11 4.22 2.22 0.49 0.68 195
16 Cornell 0.08 0.00 2.88 4.64 0.96 0.64 272
16 Pennsylvania 0.22 0.22 4.17 3.33 1.50 0.78 193
18 Northwestern 0.07 0.14 3.07 3.14 0.41 0.66 n.d.
19 Penn State 0.05 0.05 2.56 7.23 0.41 0.31 163
20 Texas A&M 0.00 0.06 2.69 6.44 0.78 0.81 657

From ASEE data
From NSF data
3For chemical engineering and materials science


Spring 2003


than fifty times as of the date of the citation search.
Different sources sometimes report different values for the
same statistic. A manifestation of this discrepancy is appar-
ent in the "Total Research Expenditure" data in Table 1. Sub-
stantial differences between the NSF and ASEE databases
appear for four departments (Berkeley, Delaware, Illinois, and
Penn State). The NSF data are for fiscal year 2000
and the ASEE data are for the academic year,
but it is difficult to envision such large differ-
ators ences being attributable to different ending dates
for a fiscal and an academic year. The chemical
Degrees engineering programs at Berkeley and at Illinois
nesota do not reside within the College of Engineering,
IT so this administrative structure might play a role
in the discrepancies. Data reported by different
aware
sources for degrees granted by a given depart-
xas
ment also exhibited variability (but not as much
as the research expenditure data).
ia Tech
The data in Table 1 afford an opportunity to
determine which departments had the highest
eley values for the different extensive quality and pro-
MU ductivity indicators. Table 2 lists the top ten de-
due apartments (of the twenty considered) in several
of the categories. For each of the five indicators
in Table 2, at least half of the departments listed
are also among the top ten
in the U.S. News ranking.
In fact, the only top-ten
Departments schools absent in more
than two of the columns
arch Expend. ($K)
in Table 2 are Stanford
Total FederaP Cit. >50 Cites and Illinois.
and Illinois. Carnegie
496 312 115 0.25 Mellon (CMU), Michi-
283 178 71 0.06 gan, and UC Santa Bar-
bara (UCSB) are the only
511 277 n.d. n.d.
443 260 73 0.06 schools ranked in the sec-
584 489 97 0.18 ond ten by U.S. News to
373 191 71 0.00 appear in at least two of
287 143 57 0.10 the columns. Each of
413 240 54 0.00 these schools appears on
184 92 84 0.06 three or four of the lists.
213 136 75 0.29
263 206 139 0.58 All of the data in Table
177 73 24 0.03 1 except for citations per
315 114 31 0.00 paper are extensive indi-
183 120 56 0.00 cators of the productivity
242 132 62 0.00 or quality of each depart-
197 144 71 0.22 ment. That is, they are
282 144 44 0.00 total quantities and their
731 435 37 0.05 values can depend on the
520 109 21 0.00
size of the department. To
analyze the data more
thoroughly, intensive in-
dicators were obtained by
97










dividing all of the statistics in Table 1 by the number of full-
time, tenured/tenure-track faculty (FTF) listed for each de-
partment. Table 3 lists these intensive indicators for each de-
partment. The data in Table 3 afford an opportunity to deter-
mine which departments had the highest values for the dif-
ferent intensive quality and productivity indicators. Table 4
lists the top ten departments (of the twenty considered) in
several of the categories.
For each of the five indicators above, at least seven of the
ten schools listed are also among the top ten in the U.S. News
ranking. Illinois is the only top-ten school absent in more
than two of the lists above for productivity or quality indica-
tors on a per-FTF basis. CMU, Pennsylvania, Michigan, and
UCSB are the only schools ranked in the second ten by U.S.
News to appear on at least two of the lists above. CMU, Michi-
gan, and UCSB also surfaced as the second-ten departments
that most frequently appeared on the top-ten lists in Table 2
for the different extensive productivity or quality indicators.
It appears that the chemical engineering graduate programs
at CMU, Michigan, Pennsylvania, and UCSB have higher
values for their productivity and quality indicators than one
might expect based on their U.S. News rankings.
The data in Tables 1 and 3 allow identification of the indi-
cators that correlate best with the U. S. News ranking. Table
5 presents the results of the correlation analysis in terms of
the correlation coefficient (R) for each indicator. This coeffi-
cient was calculated as the covariance of the two data sets
(the indicator and the ranking) divided by the product of their
standard deviations. A negative correlation in Table 5 simply
indicates that an increase in that particular quantity was ac-
companied by an improvement in the ranking.
The quantities in Table 5 with the largest correlations (ab-
solute value) are the annual number of publications, publica-
tions per FTF, the number of times cited, the num-
ber of NAE members, the number of citations per
FTF, and the number of doctorate degrees. This
strong correlation between the ranking of a chemi-
cal engineering program and its publication out-
put and citation rate was also evident in the re- Pu
sults of the 1995 NRC report on graduate pro-
gram quality. Note that three of the four most
strongly correlated quantities are extensive (sys- 2 s
tem-size dependent) variables; that is, they are the 3 B
absolute numbers of publications, citations, and 4 N
NAE members. Note too that each of the top four 5
indicators (number of publications, citations, NAE 6 M
members, and doctorate degrees) correlates bet- 7
ter with ranking when considered on an absolute 8
rather than a relative (per FTF) basis.
Table 5 shows that the U.S. News rankings do 10 Pen
not correlate as strongly with research expendi- Of the
tures as with the other indicators itemized above. Exc udi
hExcludi
The data from the ASEE show the strongest cor-
98


relation, but keep in mind that this data set is missing entries
for three departments. The NSF database included expendi-
tures for all twenty schools, and these data show a poorer
correlation with ranking. That there is a modest correlation
between total research expenditures and ranking is evident,
however, in that eight of the departments in the top ten in
expenditures in 2000 (NSF) were in the U.S. News top twenty.
That the correlation is not strong is evident in that only two
of the next ten in total research expenditures were in the U.S.
News top twenty. The schools with large research expendi-
tures (according to the NSF survey) that were not among the
top 20 in the U.S. News ranking were NC State (2nd in total
expenditures), Case Western (10th), Auburn (11 th), Oklahoma
(12th), Utah (13th), Johns Hopkins (14th), South Carolina (15th),
Florida (18th), New Mexico Institute of Mining & Technol-
ogy (19th), and New Mexico State 2 i"',i
One must keep in mind that the correlation analysis simply
shows where correlations exist. It provides no direct infor-
mation about causative effects. One might be tempted to con-
clude, for example, that a department that wants to improve
its ranking should work hard at getting more NAE members
on its faculty. Such an action might succeed, but the logic
leading to that conclusion is faulty in that it is not supported
by the mere existence of a correlation. It is possible, for ex-
ample, that the correlation between a department's ranking
and the number of NAE members on its faculty exists be-
cause it is easier for faculty at a top-ranked department to be-
come NAE members. That is, the high-ranking of the depart-
ment (or variables causing 11 i.i Ii.i i i.iiikiii. ) could be a partial
cause of the high number of NAE members, not the result of it.
Finally, it is worth noting that the correlations found herein
to exist between ranking and some indicators of productivity
and quality for the twenty departments ranked by U.S. News


TABLE 4
Top Ten' Departments in Different
Intensive Productivity or Quality Indicators

blications' Citations' Pubs w/>50 cites2 NAE Members Doctorate Degrees
isconsin UCSB UCSB Caltech Minnesota
tanford MIT Michigan UCSB MIT
lerkeley Stanford MIT MIT Wisconsin
lichigan Berkeley Pennsylvania Pennsylvania Delaware
Texas Princeton Stanford Minnesota Texas
[innesota Michigan Delaware Princeton Texas A&M
Illinois Wisconsin Minnesota Berkeley Pennsylvania
CMU Minnesota Wisconsin CMU Berkeley
elaware Pennsylvania Princeton Texas Illinois
insylvania Texas Berkeley Wisconsin CMU
20 ranked by U.S. News
ng Caltech because of lack of data


Chemical Engineering Education











likely become weaker as one includes more departments
in the analysis. Previous analysisE15 showed that the cor-
relation between reputational rankings and objective
indicators is much weaker for departments that are not
highly ranked.

CONCLUDING REMARKS
This article provides objective indicators of the pro-
ductivity and quality for the twenty chemical engineer-
ing departments ranked most highly by U.S. News. The
indicators that correlated most strongly with the rankings
were the number of publications, citations, NAE mem-
bers, and doctorate degrees. For each of these four indi-
cators, the extensive quantity was more strongly corre-
lated with the ranking than was the intensive quantity.
This result suggests that departments with more faculty
members tend to be more highly ranked than departments
with fewer, but equally excellent, faculty members.


TABLE 5
Correlation of U.S. News Ranking
with Different Indicators


Indicator
Number of Publications'
Publications per FTPF
Number of Times Cited2
NAE Members
Citations per FTPF
Doctorate Degrees
Doctorate Degrees per FTF
NAE Members per FTF
Total Research Expenditures'
AIChE Institute Awards
Federal Research Expenditures3
Total Research Expenditures per FTF1
Master Degrees
AIChE Institute Awards per FTF
Papers with >50 citations2
Citations per paper2
Full-Time Tenured/Tenure Track Faculty
Federal Research Expenditures per FTF3
Total Research Expenditures3
Master Degrees per FTF
Total Research Expenditures per FTF3
Bachelor Degrees
BS Degrees per FTF


Correlation Coefficient
-0.819
-0.718
-0.675
-0.602
-0.572
-0.525
-0.518
-0.511
-0.489
-0.402
-0.354
-0.333
-0.302
-0.285
-0.285
-0.278
(FTF) -0.273
-0.242
-0.207
-0.109
-0.077
0.034
0.244


StI.' .., schools for which no expenditure data were
reported, ASEE
2I .'.. .. Caltech
3NSF report


There have been calls4 5]1 for departmental rankings to use objec-
tive criteria that indicate excellence rather than relying solely on repu-
tation. Rankings based solely on peer assessment surveys are akin to
preseason college football polls that are good at identifying teams
that have a history of sustained excellence, but which typically un-
dervalue teams that are on the rise and overvalue teams that are de-
clining. At the end of the season, though, those polled can use statis-
tical data and won/loss records to assess the excellence of the teams.
These year-end rankings, whether exclusively from a poll or from a
combination of poll results and objective data (e.g., the Bowl-Cham-
pionship Series, or BCS, formula) provide a reasonable sorting of the
different teams by their likely ability to win football games. Like-
wise, rankings of engineering programs could be improved by in-
cluding some quantitative measures of objective indicators of pro-
ductivity or quality. Survey respondents could use these indicators,
along with their subjective judgment, to assess different programs
(as in a coaches' or writers' poll in college football). Alternatively,
these indicators could be used in some formula, along with survey re-
sults, to determine rankings (as in the BCS formula). That the ranking
systems in college football make better use of objective indicators of
excellence than the ranking systems used for chemical engineering gradu-
ate programs is revealing.

ACKNOWLEDGMENTS
This article had its genesis in an internal review of the chemical
engineering department at the University of Michigan. Mark Burns,
Sharon Glotzer, Mike Solomon, and Steve Yalisove served on the re-
view committee with me. I greatly appreciate their participation in this
review and their insights regarding the data presented in this article.

REFERENCES
1. "Best Graduate Schools, 2003 Edition, U.S. News & World Report; portions
available on-line at engindex.htm>
2. Goldberger, M.L., B.A. Mahler, and P.E. Flattau, eds, Research-Doctorate Pro-
grams in the United States: Continuity and ....... National Academy Press,
Washington, DC (1995)
3. Gourman, J., The Gourman Report: A Rating of Graduate Professional Pro-
grams in American and International Universities, 8th ed., Princeton Review
(1997)
4. Diamond, N., and H.G. Davis, "How Should We Rate Research Universities?"
(. ... .. 2-14 (July/August 2000)
5. Angus, J.C., R.V. Edwards, and B.D. Schultz, "Ranking Graduate Programs:
Alternative Measures of Quality," Chem. Eng. Ed., 33(1), 72 (1999)
6. Membership directory available on-line at . Emeritus fac-
ulty were not included in this count.
7. Chemical Engineering Faculty Directory 2001-02, AIChE, New York, NY
8. From Institute Awards lists posted at
9. Available at
10. Shacham, M., and N. Brauner, "The Effect of Publication Rate Profile on Cita-
tion Statistics," Chem. Eng. Ed., 35(1), 32 (2001)
11 Grossmann, I.E., "Some Pitfalls with Citation Statistics,." i i 34(1),
62 (2000)
12. ASEE Directory of Engineering Colleges-I .'.. available online at www.asee.org/publications/colleges/>
13. National Science Foundation, Division of Science Resources Statistics, Aca-
demic Research and Development Expenditures: Fiscal Years 2000, NSF 02-
308, Project Officer, M. Marge Machen, Arlington, VA (2002) Tables B-48 and
B-49, available at J


Spring 2003











EI n= laboratory


BUILDING MULTIVARIABLE


PROCESS CONTROL INTUITION


USING CONTROL STATION



DOUGLAS J. COOPER, DANIELLE DOUGHERTY, ROBERT RICE
University of Connecticut Storrs, CT 06269-3222


Mutivariable loop interaction is a well-known con-
trol problem that is discussed in a host of popular
texts.E1-4] Computer tools such as Matlab/Simulink
enable instructors and students alike to explore the phenom-
ena by providing a high-level programming environment use-
ful for simulating process control systems. The topics to be
covered in a process control course, however, are numerous
relative to the time allotted to them in the typical curriculum.
Instructors must decide for themselves whether or not time
spent with programming issues is time well spent in a pro-
cess dynamics and control class. Many feel it is an appro-
priate use of time, and valid arguments can be made to
support that viewpoint.
An alternative chosen by more than 150 college and uni-
versity instructors around the world is the Control Station
training simulator. Control Station lets students design, imple-
ment, and test control solutions using a computer interface
much like one they will find in industrial practice. It pro-
vides hands-on and real-world experience that the students
will be able to use on the job. One of the primary benefits
according to instructors who use the program is that the soft-
ware is easy to use, permitting them to focus on teaching
process dynamics and control issues rather than on program
usage. Many students have related that because Control Sta-
tion is so visual in its presentation, they believe it enhances
their learning and knowledge retention.
Control Station provides a platform where broad and rapid
experimentation can help students build fundamental intu-
ition about a broad spectrum of process dynamic and control
phenomena. Some of the topics that can be explored using
the software include
Dynamic modeling of plant data
Using process models parameters for controller tuning
Tuning P-Only, PI, PID, and PID with Filter controllers


Cascade controller design and implementation
Feed forward control with feedback trim
Smith predictor design for dead time compensation
Parameter scheduling and adaptive control
Dynamics and control of integrating processes
Single and multiloop dynamic matrix control (DMC)
This paper will show how students can use Control Station
to investigate the nature of multivariable loop interaction and
how decouplers can minimize this undesirable behavior. The
examples will demonstrate how students can use the soft-
ware to quickly develop a host of multivariable process be-

Doug Cooper is Professorof Chemical Engineer-
ing at the University of Connecticut. His long-term
research focus is on developing control meth-
ods that are both reliable and easy for practitio-
ners to use. He is currently studying whole-plant
control, multivariable adaptive control, and the
control of the direct methanol fuel cell process.



Danielle Dougherty received a BS from Wid-


engineeering. Her thesis was on multivariable
adaptive model predictive control. Her current
post-doc research focuses on modeling and
controlling direct methanol fuel cell processes.




Robert Rice received his BS from Virginia Poly-
technic Institute in 2000 and is currently work-
-* ing toward his PhD in chemical engineering at
the University of Connecticut under the direc-
tion of Doug Cooper. His research involves
multivariable model predictive control of un-
stable processes.


Copyright ChE Division ofASEE 2003


Chemical Engineering Education










haviors for exploration and study, and how they can then test
the performance of control strategies using methods found in
their text. Students performing this or similar study will cer-
tainly strengthen their understanding and intuition about this
challenging subject.

MULTIVARIABLE CASE STUDIES
Multivariable process control is increasingly important for
students to understand at an intuitive level because in many
industrial applications, when one controller output signal is
changed, more than one measured process variable will be
affected. Control loops sometimes interact and even fight each
other, causing significant multivariable challenges for pro-
cess control. Control Station provides a means for students
to gain a hands-on understanding of multivariable process
behavior and to practice how to design and tune controllers
that address these behaviors.
One multivariable case study available to students is the
multitank process. As shown in Figure 1, the process com-
prises two sets of freely draining tanks positioned side by
side. The two measured process variables are the liquid lev-
els in the lower tanks. To maintain liquid level, two level
controllers manipulate the flow rate of liquid entering their
respective top tanks. In this process, each of the upper tanks
drain into both lower tanks. This creates a multivariable in-
teraction because manipulations by one controller affect both
measured process variables.
The distillation column case study is shown in Figure 2.
This is a binary distillation column that separates benzene
and toluene. The objective is to send a high percentage of the
benzene out the top distillate stream and a high percentage of


We do not believe that the
training simulator should replace real
lab experiences since hands-on studies are
fundamental to the learning process, but a
training simulator can provide a broad
range of meaningful experiences
in a safe and efficient fashion.

the toluene out the bottom stream. To separate benzene from
toluene, the top controller manipulates the reflux rate to con-
trol the distillate composition. The bottom controller adjusts
the rate of steam to the reboiler to control the bottoms com-
position. Any change in feed rate to the column acts as a dis-
turbance to the process.
Multivariable loop interaction occurs in this process be-
cause when the benzene composition in the top distillate
stream is below the set point, the top controller responds by
increasing the cold reflux into the column. This cold liquid
eventually spills to the bottom, cooling it and causing the
bottom composition to move off the set point. The bottom
controller "fights back" by increasing the flow of steam into
the reboiler. The result is an increase of hot vapors traveling
up the column that counteract the increased reflux by heating
the top of the column.

MULTIVARIABLE CUSTOM PROCESSES
Control Station's multiloop Custom Process graphic, used
to simulate general multivariable systems created from dy-
namic models, is shown in Figure 3. Following the nomen-
clature established in popular texts,E1-41 G represents the dy-


F1 ([m3.min) F2(m3?m
| 11 : J 133

C01 ( .) _t C02


--- T 1





Set Point 1 I 3.25
W Lel 1 mW




D1(m3 min)
[ 1.0 F (m3mrin)
(Disturbance) |
I ---- >


in)



.5




2






D2 (m3lmin)
I 1.0u
(Disturbance)


F (m3lmin)
j 10.3
---,-


Figure 1. Control Station's multitank case study.
Spring 2003


Figure 2. Control Station's distillation column
case study.











namic behavior of the ith measured process variable
response to the jth controller output signal. Hence,
as can be seen in Figure 3, process G,, describes the
direct dynamic response of measured process vari-
able PV1 to changes in controller output CO,, and
interaction G21 describes the cross-loop dynamic
response of PV2 to changes in CO,.

RELATIVE GAIN AS A
MEASURE OF LOOP INTERACTION

Before exploring different multivariable process
behaviors, we introduce the concept of relative
gain. [1 Relative gain, k, is popular because it

Provides a convenient measure of loop interaction
Is easy to compute
Is dimensionless, so it is not T..: i...I by the units of
the process data

Relative gain is computed from the steady-state pro-
cess gains of the process models (K11 and K22) and
the cross-loop interaction models (K12 and K21) that
best describe observed process behavior (that results
from model fits of process data). Following the no-
menclature above, relative gain is computed as

K11K22 12 K(1)

In the remainder of this paper, we will show how
Control Station helps students explore what the size
and sign of k implies for multivariable loop inter-
action and the ease with which a process can be con-
trolled. Before starting that study, consider that our
process has two controllers (CO1 and CO2) that regu-
late two process variables (PV1 and PV2). The con-
trollers are connected to the process variables by
wires and the connections can be wired one of two
ways:
1) CO, controls PV1 and CO2 controls PV2
2) CO1 controls PV2 and CO2 controls PV1
Each combination yields a different value of k.
An important lesson students learn is that control
loops should always be paired (wired) so the rela-
tive .... ,in is positive and as close as possible to one.

EFFECT OF Kp ON CONTROL LOOP
INTERACTION
The students are taught the usefulness of relative
gain as a measure of multivariable loop interaction
by considering a variety of cases such as those listed
in Table 1. These particular cases are simulated and
studied here using Control Stations's Custom Pro-
cess module, as shown in Figure 3.
102


All of the direct process and interaction models used in the simulation
studies are first order plus dead time (FOPDT). For each simulation case
study, the direct process and cross-loop gains are listed in the table. All of
the time constant and dead time parameters for the simulation case stud-
ies given in Table 1 are
Process time constant: Tp = 10
Dead time: Op = 1
Also, all of the investigations use two PI (pr pi, ii ii.il-ilic-1.il) control-
lers with no decoupling and with
Controller gain: K = 5
Reset time: Tz = 10
For all examples, when one PI controller is put in automatic while the
other is in manual mode, that controller tracks set point changes with an
appropriately small rise time and rapid damping. The issue the students
study is process behavior when both PI controllers are put in automatic at


TABLE 1
Exploring Relative Gain, k, as a Measure of Loop Interaction

direct cross-loop cross-loop direct
CO1 -PV1 CO1 -> PV2 CO2 -> PV1 CO2 -> PV2
Case K11 K21 K12 K 22
1 1 1.1 1.1 1 -4.8
2a 0 1.1 0.5 1 0.0
2b -1 3.0 1.1 1 0.2
2c 1 -3.0 0.5 1 0.4
3 1 -1.1 0.5 1 0.6
4 1 0 0.5 1 1.0
5 1 1.1 0.5 1 2.2
6 1 1.1 .85 1 15.4


Figure 3. Control Station's multiloop custom process.
Chemical Engineering Education











the same time.

Case 1: X < 0 When the cross-loop interaction gains
are larger than the direct process gain, as is true for Case 1 in
Table 1, then each controller has more influence on its cross-
loop measured process variable than it does on its own direct
measured process variable. As listed in the table, the relative
gain, X, computed by Eq. (1) for this case is negative.
Figure 4 shows the set point tracking performance of the
Case 1 process when both loops are under PI control with no
decoupling (remember that for all simulations, Tp = 10 and
Op = 1; also, K = 5 and Tc = 10). As each controller works to
keep its direct measured process variable on its set point, every
control action causes an even larger disruption in the cross-
loop process variable-and the harder each controller works,
the worse the situation becomes. As can be seen in Figure 4,
the result is an unstable, diverging system.
A negative relative gain implies that the loop pairing is in-
correct. That is, each controller is wired to the wrong mea-
sured process variable. The best course of action is to switch
the controller wiring. This switches the cross-loop gains in
Table 1 to the direct process gains and vice versa.
Switching the loop pairing recasts Case 1 into a process

The Process is Unstable When Relative Gain, k < 0

50 --
48
S100
o75
o 50
54
52
>52 ---------- ; :------
a. 50 ------------ I -------
IL> 50
50
O 25
C-)-
25 30 35 40 45 50 55 60 65
Time (time units)
Tuning: Gain = 5.00, Reset Time = 10.0, Derv Time = 0.0, Sample Time = 0.10
Tuning: Gain 5.00, Reset Time = 10.0, Deriv Time = 0 0, Sample Time = 0.10

Figure 4. Incorrect loop pairing and an unstable process
under PI control indicated by k = 0.

Relative Gain and Multi-Loop Interaction
51 -
g 50 -. '.
60 (case 3) (case 4) (case 5)
50
O 40
51


60
o 50
40
50 100 150 200 250 300 350
Time (time units)
Tuning: Gain= 5.00, Reset Time = 10.0, Deriv Time 0.0, Sample Time= 0.10
Tuning: Gain = 5.00, Reset Time = 10.0, Deriv Time = 0.0, Sample Time = 0.10

Figure 5. Impact of X on PI control loop interaction with
no decoupling.
Spring 2003


with a relative gain of X = 5.8, which is a loop interaction
behavior between Case 5 and Case 6. As we will learn, a
process with this relative gain is challenging to control, but it
is closed-loop stable and the loops can be decoupled using
standard methods.
Case 2: 0 < X 0.5 For the relative gain to be exactly
zero ( = 0), one of the direct process gains must be zero. A
direct process gain of zero means that a controller has no
impact on the measured process variable it is wired to. Clearly,
there can be no regulation if a controller has no influence.
Case 2a in Table 1 has Ki = 0, implying that CO1 has no
influence on PV1. Yet because the cross-loop gain K12 is not
zero, changes in CO2 will disrupt PV1. If a measured process
variable can be disrupted but there is no means to control it,
the result is an unstable process under PI control (no figure
shown). Because both cross-loop gains are not zero in Case
2a, the loop pairing should be switched in this case to give
each controller direct influence over a measured process vari-
able. This would recast Case 2a into a process with a k = 1.0,
which is the interaction measure most desired. We study such
a process in Case 4 below.
When the relative gain is near zero (0 < k< 0.5), then at
least one of the cross-loop gains is large on an absolute basis
(e.g., Case 2b and 2c). Under PI control with no decoupling
and using the base tuning values of K = 5 and Ti = 10, both
of these processes are unstable and show considerable loop
interaction (no figure shown). Detuning both controllers to
Kc = 2 and Tc = 10 restores stability, but control-loop inter-
action is still significant.
Again, the best course of action is to switch the loop pair-
ing. With the wiring switched, Case 2b yields k = 0.8 and
Case 2c yields k = 0.6, putting both relative gains closer to
the desired value of one. While both processes still display
loop interaction, the processes become stable under PI con-
trol with no decoupling, even with the base case PI controller
tuning values.
Case 3: 0.5 < Xk < 1 When the relative gain is between
0.5 and one, the cross-loop interactions cause each control
action to be reflected and amplified in both process variables.
As shown in the left-most set point steps in Figure 5 for a
case where k = 0.6, this interaction leads to a measured pro-
cess variable response that includes significant overshoot and
slowly damping oscillations.
This amplifying interaction exists when stepping the set
point of either loop. It grows more extreme and ultimately
leads to an unstable process as k approaches zero (see Case
2). Moreover, the interaction becomes less pronounced as k
approaches one (see Case 4).
Case 4: X = 1 A relative gain of one occurs when
either or both of the cross-loop gains are zero. In Case 4, K21
is zero, so controller output CO has no impact on the cross-
loop measured process variable PV2. Since K12 is not zero as
103











listed in Table 1, however, changes in CO2 will impact PV1.
The second set point steps in Figure 5 show the control
performance of the Case 4 process when the set point of PV1
is changed. As expected, the set point tracking actions of CO1
have no impact on PV2. While not shown, a set point step in
PV2 would cause some cross-loop disruption in PV1 because
of loop interaction.
When both cross-loop gains are zero, the loops do not in-
teract. Such a system is naturally and completely decoupled
and the controllers should be designed and tuned as single-
loop processes.
D- Case 5: k > 1 Opposite to the observations of Case 3,
when the relative gain is greater than one, the control loops
fight each other. Specifically, the cross-loop interactions act
to restrain movement in the measured process variables, pro-
longing the set-point response. The third set point steps in
Figure 5 illustrate this behavior for a case where 2 = 2.2.
As stated earlier, a process with a relative gain that is posi-
tive and close to one displays the smallest loop interactions
(is better behaved). For Case 5, switching the loop pairing
would yield a very undesirable negative 2. This means that
the loops are correctly paired and the significant loop inter-
action is unavoidable.
Case 6: >> 1 As the cross-loop gain product, Kl2K 21
approaches the direct process gain product, K11K22, the rela-
tive gain grows and the restraining effect on movement in
the measured process variables discussed in Case 5 become
greater. This is illustrated in the right-most set point steps in
Figure 5 for a case where 2 = 15.4. Again, switching the
loop pairing would yield a negative 2, so the loops are cor-
rectly paired and the significant loop interaction is unavoid-
able. Interestingly, as the cross-loop gains grow to the point
that their product is larger than the direct process gain prod-
uct (when K12K21>K11K22), then k becomes negative and we
circle back to Case 1.

DECOUPLING CROSS-LOOP Kp INTERACTION
After gaining an appreciation for the range of open-loop
dynamic behaviors, students then explore decoupling con-
trol strategies. A decoupler is a feed-forward element where
the measured disturbance is the action of a cross-loop con-
troller. Analogous to a feed-forward controller, a decoupler
is comprised of a process model and a cross-loop disturbance
model. The cross-loop disturbance model receives the cross-
loop controller signal and predicts an "impact profile," or
when and by how much the process variable will be impacted.
Given this predicted sequence of disruption, the process model
then back calculates a series of control actions that will coun-
teract the cross-loop disturbance as it arrives so the measured
process variable, in theory, remains constant at set point.
Here we explore how perfect decouplers can reduce cross-
loop interaction. A perfect decoupler employs the identical
104


models in the decoupler as is used for the process simulation.
Using the terminology from Figure 3, these decouplers are
defined in the Laplace domain as


Gll(s)
D12(s) G11(s)


G21(s)
and D21(s) G22(s)


Students are reminded to be aware that in real-world applica-
tions, no decoupler model exactly represents the true process
behavior. Hence, the decoupling capabilities shown here must
be considered as the best possible performance.
Case 1: k < 0 A negative relative gain implies that the
loop pairing is incorrect. Decoupling is not explored because
the best course of action is to switch the controller wiring to
produce a process with a relative gain of k = 5.8. This loop
interaction behavior is somewhere between Case 5 and Case
6 discussed below.
Case 2: 0 < 0.5 A relative gain of exactly zero (2
= 0) implies that at least one controller has no impact on the
measured process variable that it is wired to. There can be no
regulation if a controller has no influence. Hence, decoupling
becomes meaningless for this case and is not explored here.
When the relative gain is near zero (0 < k < 0.5), PI con-
trollers with no decoupling must be detuned to stabilize the
multivariable system. When the PI controllers are detuned
and perfect decouplers (the identical models are used in the
decouplers as are used for the process simulation) are in-
cluded, the result is an unstable system (no figure shown).
Detuning the decouplers (lowering the disturbance model
gain) will restore stability, but interaction remains signifi-
cant and general performance is poor. Again, the best course
of action is to switch loop pairing.
Case 3: 0.5 k < 1 When the relative gain is between

Perfect Decouplers Minimize Interaction for Moderate k
Process: Multi-Loop Custom Process Cont. 1: PID w/ Decoupler
Cont. 2: PID w/ Decoupler
51.0 : i I i
50.5
0. 500
60 ------- (case 3) ---! (case 4) -i-- (case) -----

o 40
50.5
IL 50.0 ------,- - ------ ------------------------
6 0- --49 .- - - - - - - - ---- -------- -- -- --- ---+--- A--- -- -- -- -- -
49. 50


0 40
20 40 60 80 100 120 140 160 180 200
Time (time units)
Tuning: Gain 5.00, Reset Time 10.0, Sample Time= 0.10
Process Model: Gain(Kp) 1.00, T1l 10.0, T2 0.0, TI 0.0, TD 1.00
Disturbance Model: Gain(Kd) 0.50, TI 10.0, T2 0.0, TL = 0.0, TD 1.00
Tuning: Gain 5.00, Reset Time 10.0, Sample Time 0.10
Process Model: Gain(Kp) = 1.00, T1 = 10.0, T2 0.0, TL = 0.0, TD 1.00
Disturbance Model: Gain(Kd) ..... T 10.0, T2 0.0, TL 0.0, TD 1.00


Figure 6. Decouplers work well when k is near 1.
Chemical Engineering Education











0.5 and one, the cross-loop interactions cause each control
action to be reflected and amplified in both process variables.
As shown in the left-most set-point steps in Figure 6 for the
case of k = 0.6, PI controllers with perfect decouplers virtu-
ally eliminate cross-loop interactions. This is not surprising
since the relative gain is positive and close to one.
Case 4: X = 1 A relative gain of one occurs when either
or both of the cross-loop gains are zero. In Case 4 of Table 1,
K21 is zero, so controller output CO1 has no impact on the
cross-loop measured process variable PV2. Consequently, a
perfect decoupler will provide no benefit for this loop, and as
shown in Figure 6 for the middle set-point steps, while a per-
fect decoupler causes no harm, a decoupler implemented on
a real process will likely have imperfect models and would
then create loop interaction.
Table 1 shows that K12 is not zero, so changes in CO2 will
impact PV1. A perfect decoupler will virtually eliminate cross-
loop interaction for information flow in this direction (no fig-
ure shown). Thus, the Case 4 system can address the multi-
variable loop interaction with a single decoupler on the CO2
to PV1 loop.
Case 5: X > 1 When the relative gain is greater than
one, the cross-loop interactions act to restrain movement in
the measured process variables. The third set point steps in
Figure 6 for the case where k = 2.2 illustrate that perfect
decouplers substantially eliminate both this restraining ef-
fect and the level of loop interaction, Again, this is not sur-
prising since the relative gain is positive and reasonably close
to one.
Case 6: X >> 1 As the relative gain grows larger, the
restraining effect on movement in the measured process vari-

Detuning the Decouplers for Case 6 ( >> 1)
Process: Multi-Loop Custom Process Cont. 1: PID w/ Decoupler
Cont. 2: PID w/ Decoupler
51

5- detuned cross-loop decoupler perfect cross-loop decoupler
100 (K, = 1 0) 44(K4 = 1.1) 4
50
0 V
51
N 50


0o 0
0
40 60 80 100 120 140 160 180 200 220
Time (time units)
Tuning: Gain = 500, Reset Time = 10.0, Sample Time 0.10
Process Model: Gain(Kp) = 1.00, TI 10.0, T2 = 0.0, TL = 0.0, TD = 1.00
Disturbance Model: Gain(Kd) = 0.85, Tl = 10.0, T2 = 0.0, TL = 0.0, TD = 1.00
Tuning: Gain = 5.00, Reset Time = 10.0, Sample Time = 0.10
Process Model: Gain(Kp) 1.00, Tl 10.0, T2 0.0, TL 0.0, TD = 1.00
Disturbance Model: Gain(Kd) = 1.10, TI = 10.0, T2 = 0.0, TL = 0.0, TD = 1.00
lower cross -loop gain >

Figure 7. Decouplers can cause stability
problems for large k.
Spring 2003


ables due to loop interaction becomes greater. Case 6 in Table
1 is interesting because K21 is greater than K22. This means
that PV2 is influenced more by a change in controller output
CO1 (its cross-loop disturbance) than it is by an equal change
in its own controller output CO2. Switching loop pairing of-
fers no benefit as this makes the relative gain negative.
With perfect decouplers as shown in the, ..lit set-point steps
in Figure 7 (the decoupler employs the identical models as
are used for the process simulation), the system is unstable.
This cannot be addressed by detuning the PI controller be-
cause even with lower values for controller gain, Kc, the sys-
tem is unstable.
For a decoupler to be stable, the gain of the cross-loop dis-
turbance model must be less than or equal to the gain of the
process model, or in this case, K21 < K22. That is, a decoupler
must pass through at least as much influence of a controller
output to its direct process variable as it does for any distur-
bance variable.
To address this, we detune the decoupler by lowering the
cross-loop disturbance gain of the bottom loop so that in ab-
solute value, K21 < K22 and K21 < K Repeating the test in the
left set-point steps of Figure 7 reveals a stable and reason-
ably decoupled system.


CONCLUSION

We have presented examples of the lessons and challenges
associated with multivariable process control and shown how
Control Station can provide a better understanding of these
complicated systems. Space prohibits the presentation of other
multivariable studies available in Control Station, including
the use of dynamic matrix control for multivariable model
predictive control.
We do not believe that the training simulator should re-
place real lab experiences since hands-on studies are funda-
mental to the learning process, but a training simulator can
provide a broad range of meaningful experiences in a safe
and efficient fashion. The training simulator can be used to
bridge the gap between process control theory and practice.
If readers would like to learn more, they are encouraged to
contact Doug Cooper at cooper@engr.uconn.edu, or visit
.


REFERENCES
1. Luyben, M.L., and W.L. Luyben, Essentials of Process Control,
McGraw-Hill, New York, NY (1997)
2. Ogunnaike, B.A., and W.H. Ray, Process Dynamics, Modeling, and
Control, Oxford, New York, NY (1994)
3. Seborg, D.E., T.E Edgar, and D.A. Mellichamp, Process Dynamics
and Control, Wiley, New York, NY (1989)
4. Smith, C.A., and A.B. Corripio, Principles and Practice ofAutomated
Process Control, Wiley, New York, NY (1997)
5. Bristol, E.H., "On a New Measure of Interaction for Multivariable
Process Control," IEEE Trans. on Automated Control, AC-11, p. 133
(1966) 1











Random Thoughts ...






FAQS. VI


Evaluating Teaching

and Converting the Masses



RICHARD M. FIELDER, REBECCA BRENT
North Carolina State University Raleigh NC 27695


our years ago we raised ten questions that frequently
come up in our teaching workshops,E11 and since then
we have devoted five columns to answering eight of
them.* In this column we take up the last two:

1. My department head says that we can't count teach-
ing in promotion and tenure decisions because there
is no good way to evaluate it. Is there a meaning-
ful way to evaluate teaching?

2. Most people who go to teaching workshops are al-
ready good teachers-the ones who most need them
wouldn't go to one under any circumstances. How
can staunchly traditional professors be persuaded
to use proven but nontraditional teaching meth-
ods?


Evaluating Teaching
We have written several columns about evaluating teach-
ing and so will simply provide a synopsis with references
here.
The key to meaningful evaluation is t, ,,, .il.i;.i. --et-
ting data from several different sources. Student ratings ob-
viously should be included: students are the best judges of
(among other things) whether instructors are effective lectur-
ers, encourage active participation, are available and support-
ive outside class, and treat all of their students with respect.
Extensive research attests to the validity of student ratingsE2]
and several things can be done to maximize their effective-
ness at both evaluating and improving teaching.[31

* All of the FAQ columns can be viewed on-line at
< ...-. I .... .. ... .. ,..'_teaching/Columns.html>.


While necessary, however, student ratings are not sufficient.
Most students are not equipped to judge certain aspects of
teaching, such as the depth of an instructor's knowledge of
the subject, the appropriateness of the course content and its
compatibility with the department's curricular objectives, and
the fairness of assignments and tests. Only other faculty mem-
bers are in a position to make those judgments. Peer review
is therefore another important component of teaching evalu-
ation.
A proven approach to peer review (as opposed to the tradi-
tional unreliable one-shot classroom observation) calls for
two raters to observe at least two class sessions, complete
rating checklists for both sessions and other checklists for
evaluating course materials, assignments, and tests, and rec-
oncile their ratings.[J] Research-supported checklist items can
be selected from lists provided by Weimer, et al.E1
Additional evidence of teaching effectiveness can be ob-
tained from retrospective senior evaluations and alumni evalu-
ations, student performance on common examinations, and
instructor self-evaluations. Student ratings taken over sev-
eral quarters or semesters may be combined with peer rat-
ings and outcomes of some of these other assessments into a


Richard M. Felder is Hoechst Celanese Professor Emeritus of Chemi-
cal Engineering at North Carolina State University He received his BChE
from City College of CUNY and his PhD from Princeton. He is coauthor
of the text Elementary Principles of Chemical Processes (Wiley 2000)
and codirector of the ASEE National Effective Teaching Institute
Rebecca Brent is an education consultant specializing in faculty devel-
opment for effective university teaching, classroom and computer-based
simulations in teacher education, and K-12 staff development in lan-
guage arts and classroom management. She co-directs the SUCCEED
Coalition faculty development program and has published articles on a
variety of topics including writing in undergraduate courses, cooperative
learning, public school reform, and effective university teaching.


Copyright ChE Division of ASEE 2003
Chemical Engineering Education










;I.,.. bitI,.. portfolio,E61 which provides the basis for an excep-
tionally meaningful evaluation of teaching.


Converting the Masses
At almost every workshop we give, we are informed that


we are preaching to the choir, and the faculty
who most need to change wouldn't go to a teach-
ing workshop at gunpoint. Some of our infor-
mants then ask how such individuals can ever
be persuaded to change to more effective teach-
ing methods.
We offer several notes of encouragement in
response.


I
simp
bes
teaci
know
_- -j -


In part due to ',. .i,-., such as the UnU
National F0t ...i i. ,.h1iin;.. Institutey71 and yo
local campus faculty development efforts, wi
the number of faculty members ,,. i; *.. coll
proven but (in, ,*.. in .,) nontraditional incline
;1., o..li:- methods has risen dramatically in it,
the past decade, and the number is almost rel
certain to keep ,.i ii studez
In a 1999 survey of engineering faculty mem- jus
bers in the eight institutions that comprised the
SUCCEED Coalition, 65% of the 511 respon-
dents reported writing instructional objectives
for their classes, 60% assigned small-group exercises, and
54% gave team assignments. Demographic data established
that the respondents were truly representative of the entire
1621-person faculty and not disproportionately "true believ-
ers."E81 The survey results support our own observations. In
the workshops we have given for over a decade, when we
describe active learning (getting students to do things in class
other than watch and listen to the instructor) we usually ask
for a show of hands of the participants who regularly use this
approach in their classes. Ten years ago, two or three hands
would typically go up; now, one-third to one-half of them do.
ABET and the new accreditation criteria have been and
will continue to be a driving force for the continuation of this
trend. If we are to produce engineering graduates with mas-
tery of such skills as communication and multidisciplinary
teamwork, we must clearly do something in the preceding
four years to equip them with those skills. Equally clearly,
lecturing alone won't do it, but instructional methods such as
active, cooperative, and problem-based learning when done
correctly can promote development of all of the skills in ABET
Outcomes 3a-3k.E91 Engineering instructors who are currently


the only ones in their departments using those methods are
unlikely to be alone much longer.

It is not necessary to convert the masses.

It's certainly true that some instructors will never attend
teaching workshops or use any of the methods promoted in


fyou
ly do the
t job of
thing you
how to do
hare what
i know
th any
eagues
ed to hear
rou can
ix-the
its will be
t fine.


them, but it's also not worth losing sleep over.
Students can still learn in classes taught by
skilled lecturers who do nothing else, and even
if an instructor does not use cooperative learn-
ing, many or most students figure out the ben-
efits of group work for themselves and form
study groups on their own. As long as some in-
structors provide an optimal classroom environ-
ment-one that weans the students away from
their dependence on professors and teaches them
to rely on themselves and their peers as the pri-
mary sources of learning-the skills they acquire
will carry over to their less expertly taught
courses and later to their careers.E10111]
In short, there is no need for all of your col-
leagues to see the light. If you simply do the
best job of teaching you know how to do and
share what you know with any colleagues in-
clined to hear it, you can relax-the students
will be just fine.


References
1. Felder, R.M., and R. Brent, "FAQs," Chem. Engr. Ed., 33(1), 32 (1999)
2. Felder, R.M., "What DoTheyKnowAnyway?" (. .... i. i 26, ),
134 (1992)
3. Felder, R.M., "What Do They Know Anyway? 2. Making Evalua-
tions Effective," Chem. Engr. Ed., 27(1), 28 (1993)
4. Brent, R., and R.M. Felder, "It Takes One to Know One," Chem. Engr.
Ed., 31(1), 32 (1997)
5. Weimer, W., J. L. Garrett, and M. Kerns, How am I Teaching? Forms
and Activities for Acquiring Instructional Input, Magna Publications,
Madison, WI, (1988)
6. Felder, R.M., and Rebecca Brent, "If You've Got It, Flaunt It: Uses
and Abuses of Teaching Portfolios," ( .. i .-- i 30(3), 188 (1996)
7. National Effective Teaching Institute Web Site, felder-public/NETI.html>, accessed 3/5/03
8. Brawner, C.E., R.M. Felder, R.H. Allen, and R. Brent, "A Survey of
Faculty Teaching Practices and Involvement in Faculty Development
Activities," J. Engr. Ed., 91(4), 393 (2002)
9. Felder, R.M., and R. Brent, "Designing and Teaching Courses to Sat-
isfy the ABET Engineering Criteria," J. Engr. Ed., 92(1), 7 (2003)
10. Felder, R.M., "A Longitudinal Study Of Engineering Student Perfor-
mance And Retention: IV. Instructional Methods And Student Re-
sponses To Them," J. Engr. Ed., 84(4), 361 (1995)
11. Felder, R.M., "The Alumni Speak," (.... i. i 34(3), 238 (2000)
a


Spring 2003


All of the Random Thoughts columns are now available on the World Wide Web at
http://www.ncsu.edu/effective_teaching and at http://che.ufl.edu/~cee/











classroom


A SOLIDS PRODUCT


ENGINEERING DESIGN PROJECT




DHERMESH V. PATEL, AGBA D. SALMAN, MARTIN J. PITT, M.J. HOUNSLOW, I HAYATI*
The University of Sheffield Sheffield S1 3JD, United Kingdom


he design project forms an integral part of an under-
graduate degree in chemical engineering accredited
by the Institution of Chemical EngineersE1 and the
Institute of Energy. It is a four-year program in which a ma-
jor design project contributes one-quarter of the credits in
the third year. Usually, different projects (supervised by an
academic staff member) are assigned to groups of 3 to 6 stu-
dents. The groups have a period of one year to work through
a typical process industry problem. For assessment purposes,
the students make a verbal and visual presentation to staff
and peers in the first semester and prepare a joint poster and
a detailed individual design dissertation in the second.
The project provides a necessary understanding of process
design of unit operations such as separators, distillation col-
umns, heat exchangers, and other process components as well
as bringing together other elements of the degree course, such
as thermodynamics, transport phenomena, and process safety.
In addition to supplementing the technical skills learned in
lectures, the project develops transferable skills such as com-
munication, organization, and team-working.
Traditionally, the design project has been geared toward
designing a theoretical process for manufacture of a com-
modity chemical that dominated the chemical industry dur-
ing the twentieth century, such as cumene or ethanol. It is
relatively unusual for projects to involve much solid process-
ing, despite its importance in industry.E2-41 This is in part be-
cause of the intrinsic difficulty and in part because data and
design procedures are less readily available. In addition, stu-
dent projects normally use purity as the main or sole measure
of the product's quality. For many solid products, however,
the particle size distribution, flowability, and functionality in
use may be equally or more important.
In the past, chemical engineers have designed processes

* Borax Europe Ltd, (.. .' GU2 8XG, United Kingdom


but have largely left product specification to others. In recent
years, however, it has been suggested that they should be
actively involved in product design, particularly for solids.15,61
Courses and theoretical projects on product design now exist
in some European and North American universities,17 and
there is now an undergraduate textbook.E8'
This project was restricted to MEng students. They are first-
degree students who have achieved a higher minimum stan-
dard (55% instead of 40%) in earlier courses and who com-
plete an additional year of study compared with BEng stu-
dents. The project was offered to allow such capable and well-
motivated students to actively engage in the design process
for a real industrial product. Before starting the project, they
had completed two years of laboratories and a mini-project.
In the current scheme, students study particle science in the
second year and particle processes in the third year.
A number of universities include experimental work as part
of the MEng degree scheme, but they are generally research


Dhermesh V. Patel graduated with a first class degree in Chemical Pro-
cess Engineering and Fuel Technology. While at Sheffield University he
received the British Coke Research Association Prize and the Vacation
Work Prize, and was awarded the Mappin Medal for his outstanding final
year performance.
Agba Salman is a Chartered Physicist, formerly with the Open University
and now a Lecturer in Chemical and Process Engineering at the Univer-
sity of Sheffield. Both his teaching and research are concerned with par-
ticle technology. He was the principal supervisor for the project.
Martin J. Pitt is a Chartered Chemical Engineer, with industrial and aca-
demic experience, currently the Co-ordinator of Design Teaching in Chemi-
cal and Process Engineering at the University of Sheffield.
Michael Hounslow has a doctorate in chemical engineering from the
University of Adelaide. He is currently Head of Department and leader of
the Particle Products research group in Chemical and Process Engineer-
ing at the University of Sheffield.
Igan Hayati studied chemical engineering at Leeds University. He then
joined the Interface Science group at Imperial College and obtained his
Ph.D. in 1985. He worked for ICI, Paints Division for three years and then
joined the research Department at Borax in 1990. He is now responsible
for new product development and new applications at Borax Europe.


SCopyright ChE Division of ASEE 2003
Chemical Engineering Education










projects (albeit of an applied and often interesting nature).
The project described here is unusual in that it uses labora-
tory measurements as part of a design exercise and uses real
industrial materials.

BACKGROUND
Boron is one of seven essential micronutrients required for
normal growth and fruiting of most agricultural crops. Soil
testing and plant tissue analyses have detected that of the es-
sential micronutrients, boron is usually the most deficient in
crops. Therefore, annual applications of boron are required for
high yields and improved quality and to offset losses from crop
removal and leaching. Plants, depending on soil type, manage-
ment level, and method of application, require only small quan-
tities of boron-around 0.2 to 4 kg per hectare per year.E91 Also,
since borates are toxic to wood-boring insects but beneficial for
plant growth, they are used as a wood preservative.
Boron is conventionally supplied in the form of granular
borate compositions. In applications that require spraying,
such as in agriculture, aqueous borate suspensions are pre-
pared by heating, dissolving, and rapid cooling of the gran-
ules, incurring practical difficulties and additional costs.
These suspensions are not self-structured and thus require
the addition of a thickening agent to maintain the stability
of the suspension.
Borax, a global supplier of borates, now has a patented self-
structured aqueous borate suspensiont101 that does not require
a thickening agent to suspend the particles. Sodium
pentaborate pentahydrate (NaB O.5H20) is formed by re-
acting boric acid (H3BO3) with borax pentahydrate
(Na2B407.5H20) in the presence of water.

6H3BO3 +Na2B407.5H20- 2NaB5O8.5H20+4H20

The product conveniently referred to as "borate cream" is
sodium pentaborate pentahydrate and is, as yet, the only stable
aqueous borate suspension found. As a consequence there is
limited knowledge of the intermediate steps involved in its
production. The suspension has a high solid content of 46
wt% and a 10% boron content. The reason the suspension is
self-structured is believed to be caused by the weak attrac-
tion of the particles by van der Waals forces.
The self-structured suspension has many advantages over
the previously prepared borate suspensions. Not only is the
suspension physically stable, but it is also pourable (unlike
the stiff compositions previously produced). The suspension
can then be readily diluted in water for application, provid-
ing greater convenience for the consumers.
The demand for the "borate cream" in plant nutrition and
wood preservation is expected to increase in the future. Other
applications for the cream could also be discovered due to
the diverse properties it exhibits.
To date, the cream has been entirely produced in a batch
Spring 2003


process. Thus, Borax is collaborating with the University of
Sheffield to establish the feasibility of increasing production
by implementing a continuous process.

PROJECT WORK
A group of four third-year chemical engineering students
were given the task of meeting the project objective. The group
first arranged a meeting with a Borax representative to dis-
cuss the requirements of the plant design and to obtain a more
complete description of the current process. The company
recommended features of the plant layout, sizing of equip-
ment, and a production of 50,000 tonnes per year of 10%
boron content cream. Product stability was critical, so it was
necessary to investigate additives to prevent syneresis (sepa-
ration) of the cream.
From this data, a preliminary overview of the process was
devised. The process could be divided into four sections: si-
los, premixing and storage, solid conveying, and reactors.
In order to design and select the appropriate equipment for
the handling and storage of these components, essential data
on the nature of the bulk solids and water would have to be
determined. The physical and chemical properties of water
could be readily obtained from published data, but due to the
originality of the process, relevant design data for the solid
additives (such as specific densities and particle sizes) were
less readily available. Therefore, these parameters were mea-
sured experimentally by the students to give the properties
directly for the conditions that would be encountered in the
design. In addition, the cream was produced by using the
current batch procedure to give the students further insight
into the process. This data could then be used to design the
individual components of the continuous production plant.
Finally, a collective consideration of the design, econom-
ics, and safety and environmental aspects of the final pro-
posed design was performed. The following paragraphs de-
scribe the work of the students in terms of both experimental
and design work.

EXPERIMENTAL WORK
The experimental work consisted of three major constitu-
ents: measuring the physical properties of the reactants and
products, determining the formation characteristics, and mea-
suring the viscosity of the final cream.
Properties of the Reactants The physical properties of
the components that are vital in the design of the mixers,
storage silos, and transportation system are the angle of
repose, the bulk density, and the particle size, shape, den-
sity, and porosity.
The repose angle is required in the design of the storage
silos and belt conveyors. It is the angle the particles make
with a flat surface when a quantity of solid is allowed to form
a heap. A standardized test procedure was used to give the
109











"poured" angle of repose for a given bulk solid.
In powder-handling systems, particle size is a key
parameter in design calculations. There are a variety of
particle-size-measurement techniques available in the
department, but due to the smaller particles present in
the powder distributions, the preferred measurement
techniques were the laser-diffraction technique (LDT)
and sieving. The general operating principle of LDT is
that the angle of diffraction of a beam of light passing
through the particles depends on the wavelength of the
light and the size of the particles. We used the appara-
tus located in the department to give the particle size
for powders within the measurement range of 4.5 to
875 pm, with sieving used for all other circumstances.
The frequency distribution obtained for additive "A" is
shown in Figure 1.
The shape of particles can have a significant bearing
on the packing and flow behavior of the bulk solid.
Samples of boric acid and borax pentahydrate were
analyzed underneath a low-powered microscope with
suitable photographic attachments to determine the
particle shapes. Micrographs of boric acid and borax
pentahydrate are given in Figure 2.
It was evident from the micrographs that the particles
were non-spherical and that some were even agglom-
erates. Hence, since nonspherical particles can af-
fect the flow behavior and equipment wear, the
particle's shape was taken into account in the silo
and conveying design.
Also, the particle density, bulk density, and porosity
of the powders were measured using standard tests. All
the measurements performed on the four bulk solids
are summarized in Table 1.
Formation of the 'cream' The quality of the cream
generated is highly dependent on three conditions of
formation: agitation rate, concentration of solid com-


ponents, and temperature of operation.
The cream was produced under batch conditions in the laboratory to
determine the optimum conditions to produce it in a continuous pro-
cess, taking into account production, economic, and safety factors.
Initially, the cream was produced without any anti-settling agents,




12 -









Figure 1. Frequency particle size distribution for additive 'A'.


a) Borax
Pentahydrate









b) Boric acid


Figure 2. Micrographs of reagents:
a) borax pentahydrate b) boric acid.


Property Boric acid Borax pentahydrate Additive 'B' Additive 'A'

Angle of Repose 34.50 36.50 32.00 51.00

Particle x16 142.80 190 2.70 77.2
Particle -- ----- ------ ----------
Size X50 255.75 450 8.65 182.5
Parameters Xs4 444.70 860 32.50 316.8
(plm)
w 1.76 2.13 3.49 2.03

Elongation 1.25 1.20
Bulk Density (kg/m3) 950 1060
Particle Density (kg/m3) 1520 2420 720 1760
Porosity 0.375 0.562

Table 1. Summary of physical measurements of the bulk solids.
Chemical Engineering Education











but syneresis occurred when the suspension was allowed to stand for long
periods, resulting in two distinct layers: an aqueous and a solid phase. Al-
though mixing could readily restore homogeneity, this would greatly affect
large-scale production. We found that adding two anti-settling agents to the
water prior to adding the reagents minimized separation. Hence, later cream
productions involved an additional premixing stage to hydrate the anti-set-
tling agents prior to adding the reagents.
Understanding the mechanism of the cream formation is essential since it
can allow possible improvements to the manufacturing process and could
lead to the production of other borate suspensions. Therefore, samples taken
at various intervals during the production of the cream were analyzed (in
relation to temperature and pH readings) to determine the stages of produc-
tion. It is known that mechanism of the formation of crystals involves two
major phases: dissolution and nucleation.
A general temperature trend during the reaction phase was a U-shaped curve,
as shown in Figure 3. Upon addition of the reactants, the temperature dropped
sharply, resulting in a minimum temperature after approximately 20 minutes.
We found that this period corresponded to the dissolution of the solid re-
agents in water via an endothermic process.
The temperature increased steadily between 20 and 60 minutes of reacting,
but remained below the temperature before the addition of the reactants. This

Variation of Temperature With Time


Figure 3. Temperature profile during cream production at
varying initial temperatures.

Variation of Viscosity With Shear Stress of All Samples
450 --Wiou
Addltves
400
350 30degC, 45 mm
300 Additives
2 50 A Additives
2 -\ 3MegC, 45 mn
S" "Cream" with additives Add...es
35degC, 60 mi!

---Preixing
S-- Stage

Anti-settling agents S age 2
0 'I' SIIg 2 i^^---- '---------------------, -----
0 10000 20000 30000 40000 50000 60000 70000 8000
Shear Stress (mPa)
Figure 4. Variation of viscosity with shear stress for the cream produced
with and without additives and the anti-settling agents.
Spring 2003


temperature rise was found to be due to the
commencement of nucleation, which is an exo-
thermic process.
The experimental data allowed us to deter-
mine the effect of conditions on dissolution and
nucleation (and hence the product quality)
and were used to design the reactor stage
under the optimum conditions for the desired
product quality.
Viscosity Measurements The viscosity is a
fundamental fluid property that is necessary to
predict the manner in which a fluid will react
to applied forces such as pumping. Since the
cream is non-Newtonian and exhibits complex
flow behavior such as separation, the viscosity
had to be determined experimentally for the
conditions that would be encountered.
We measured the viscosities of the samples
in a Rheomat 115 rotational viscometer located
in the department. Its coaxial measuring sys-
tem operates according to the Searle principle.
The control instrument enables the rotational
speed to be varied and the torque readout to be
recorded. The shear rate and shear stress are
determined from the rotational speed of the bob
and the braking torque indication, respectively,
allowing the rheogram to be plotted.
The viscosity of the cream should be as low
as possible to allow ease of handling and to al-
low the cream to be dispersed in water during
application. The variation of viscosity with
shear stress for the cream produced with and
without the additives and the anti-settling agents
is shown in Figure 4. The experimental data
suggest that for the cream produced with the
anti-settling agents, high mixing time during
the reaction stage and high temperature tend to
give lower viscosity. The conclusions from this
investigation were again incorporated into the
process design.


DESIGN WORK

Design of the proposed process was divided
into two main sections: design of the individual
components of the process and overall process
design.
Component Design The proposed process
was divided into four distinct phases: contain-
ment of solids, solid conveying, premixing, and
reaction. A flow diagram of the proposed con-
tinuous plant is given in Figure 5 (next page).
The process features a storage silo for each of
111


0 20 40 60 80 100 120 140
Time (min)


- -Run 1 -- Run 2 -o- Run 3


160 180










the solid components. The two solid reagents have storage
and feed silos to contain the large quantities of materials re-
quired. The silo design included material properties deter-
mined from shear testing using a Jenike shear cell.J111
Pneumatic conveyors transport the material from storage
to the feed silos, with cyclones positioned adjacent to the
feed silos to separate the particles from the air stream. The
solid conveying design included evaluation of the required
air flow rate to give steady operation, design of a cyclone
separator, and specification of a suitable air mover and ro-
tary valve to discharge the cyclone and act as an air lock. As
there are many system specifications that could be used, evalu-
ation of the best design was performed. All the solid compo-
nents are gravity fed from the feed silos to the first CSTR.
Since the two anti-settling agents comprise only a small
part of the final cream, they are only stored in feed silos that
are refilled manually. The anti-settling agents are hydrated in
two batch premixers working alternately. The components
need hydrating for a specific time of 1 hour, so the process has
to be performed in batches. A continuous feed to the reactors is
obtained by allowing the premixers to work alternately.
The reaction occurs in four reactors in a series arrange-
ment. By increasing the number of reactors in operation, the
process shifts from a continuous to a batch process, thus in-
creasing the likelihood of complete reaction. Due to economic
factors, however, four CSTRs in series were chosen. The re-
actors were designed by scaling up the laboratory data so
that dissolution of
the solid reagents in PnfidW-,ar
water occurs in the
first CSTR and r2
nucleation in the cF Fc
subsequent reactors. It] 1W


The finished cream
product is then stored
in two large tanks
prior to transporta-
tion to consumers.

Overall Process *
A collective consid-
eration of the design,
economics and safety
and environmental
aspects of the final
proposed design was
performed. A de-
tailed cost analysis
was carried out on
the final plant design.
The figures derived
by the students were
given and the capital
112


cost of the plant was estimated by using the factorial method.
The purchase cost of equipment was obtained from quota-
tions, when possible, to increase accuracy of the analysis.
The plant should be inherently safe since the process is
enclosed and safe operation is inherent in the nature of the
process. The solid reagents, the additives, and the 'borate
cream' are not flammable, combustible, or explosive. Addi-
tive 'A' is combustible, but since it comprises only 0.1% of
the cream, the danger is likely to be minor. Dust exposure
can be controlled by a combining engineering and process
control to prevent airborne dust concentrations. Basic safety
and fire preventative measures were included in the design.
Overall, the process should cause negligible damage under
foreseeable circumstances.
A hazard and operability study (HAZOP)E121-a systematic,
critical examination of the operability of a process-was per-
formed to indicate potential hazards that could arise from
deviations from the intended design. A partial HAZOP Re-
sult Sheet is shown in Table 2. Any additional safety fea-
tures from the analysis were included in the final P&I dia-
gram. The unit P&I diagram for the premixing stage is
given in Figure 6.

CONCLUSIONS
Borax expressed its delight on a vital piece of work that
would otherwise have been performed by the company. The
exercise showed that students are capable of taking an active


Figure 5. Flow diagram for continuous production of the cream.
Chemical Engineering Education











role in an industrial design, not as part of an industrial year
but as a major assessed project carried out at the university.
The use of experimental measurements to define product per-
formance rather than simply to collect property data was a
valuable experience.
Student feedback indicates design projects tend to develop
teamwork, presentation, and technical skills. In this particu-
lar case, the students also felt that obtaining experimental
data on the product properties and producing the cream in
the laboratory gave greater insight into the fundamental as-
pects of the process and provided a better means to meet the
process objectives. Another beneficial aspect was dealing with
company representatives and actual components rather than
just theoretical data.


TABLE 2
Partial HAZOP Result Sheet


No flow Blockage in line 9/10
Valve V3/V4 failure
PLC faulty
More flow Valve V3/V4 failure


Less flow Valve V3/V4 failure


Early flow Timer faulty on PLC for V3/V4
Late flow Timer faulty on PLC for V3/V4


Hydration does not occur



Insufficient hydration
PLC faulty
Insufficient hydration
PLC faulty


Put



Put


For traditional design projects, the problem with under-
graduates getting experimental data is that materials in many
traditional processes are toxic and the conditions involve high
temperatures and pressures. In comparison, the materials in-
volved in this process were relatively benign and the condi-
tions were moderate. Many other industrial processes in-
volving solids would also fall into this category. In the
future, we hope to have more design projects that com-
bine laboratory work with engineering design, preferably
based on actual problems-collaboration that benefits stu-
dents and industry alike.

ACKNOWLEDGMENT

We would like to acknowledge the work of group mem-
bers Jenny Richardson, Richard
Heath, and Andrew Brown. Fi-
nally, we are grateful to Borax for
allocating the project to the depart-
ment and providing essential in-
flow indicator in line 9/10 formation and assistance.


REFERENCES
weight control in premiers 1. Institution of Chemical Engineers,


tO detect quantity o01 material
Put weight control in premixers
to detect quantity of material


Feed may enter in previous batch Regular checks on timer on PLC
Feed may enter in next batch Regular checks on timer on PLC


Figure 6. Unit piping and instrumentation diagram.


"Accreditation of University Chemi-
cal Engineering Courses," IChemE,
Rugby, UK (2001)
2. Bridgwater, J., "Particle Technology,"
Chem. I. 5 11 4, 4081 (1995)
3. Nelson, R.D., and R. Davies, "Indus-
trial Perspectives on Teaching Par-
ticle Technology," Chem. Eng. Ed.,
32(2), 98 (1998)
4. Chase, G.G., and K. Jacob, "Under-
graduate Teaching in Solids Process-
ing and Particle Technology," Chem.
Eng. Ed., 32(2), 118 (1998)
5. Villadsen, J., "Putting Structure into
Chemical Engineering," Chem. Eng.
Sci., 52(17), 2857 (1997)
6. Cussler, E.L., "Do Changes in the
Chemical Industry Imply Changes in
Curriculum?" Chem. Eng. Ed., 33(1),
12 (1999)
7. Shaeiwitz, JA., and R. Turton,
"Chemical Product Design," Chem.
Eng. Ed., 35(4), 280 (2001)
8. Cussler, E.L., and G.D. Moggridge,
Chemical Product Design, Cam-
bridge University Press (2001)
9. Borax Company website (2001):

10. Hayati, I., Aqueous Borate-Contain-
ing Compositions and Their Prepara-
tion, Patent # Wq 99/20565
11. ASTM: D6128-00 Standard Test
Method for Shear Testing of Bulk
Solids Using the Jenike Shear Cell
12. Kletz, T., Hazop and Hazan, 4th ed.,
IChemE, Rugby, U.K. (also AIChE)
(1999) 1


PuCdWWAtr


TO RC2


Spring 2003











curriculum


COLLABORATIVE LEARNING


AND CYBER-COOPERATION

In Multidisciplinary Projects


JETSE C. REIJENGA, HENDRY SIEPE, LIYA E. Yu,* CHI-HWA WANG*
Eindhoven University of Technology The Netherlands


he National University of Singapore (NUS) and the
Eindhoven University of TCd iii .1-:, (TU/e) recently
formed a strategic alliance with the aim of offering
joint PhD programs. Existing scientific contacts between both
universities and the preparation of this strategic alliance ini-
tiated the additional concept of joint collaborative learning
among several interested departments at both universities.
The Department of Chemical and Environmental Engineer-
ing (ChEE) at NUS consists of more than forty faculty mem-
bers and a thousand-plus student body. The undergraduate
programs train over six hundred students who go on to foster
the growth of chemical and environmental engineering in
Southeast Asia. The quality of teaching in the ChEE depart-
ment has been greatly enhanced by its in-depth and integrated
research, which requires multidisciplinary expertise and can
be generally categorized into the areas of chemical engineer-
ing fundamentals, environmental science and tchliii .b .,,
materials and devices, and process and systems engineering.
TU/e is one of fourteen Dutch universities dedicated to edu-
cating over five thousand students in technical scientific edu-
cation and research. It comprises eight faculties offering
twelve full engineering degree programs (for the Dutch
"ir" title). The five-year degree programs lead to an aca-
demic title equivalent to a Master of Science degree in
engineering. In addition, TU/e offers a 3-year BSc and a
4-year PhD program.
Research teams at both TU/e and NUS carried out certain
tasks to meet the objectives given by a company, Global Cool-
ing, under comparable, yet different, settings. The TU/e team
designed a photovoltaic refrigerator with a Stirling cooler,
while the NUS team incorporated a direct-current compres-
sor with an identical refrigerator. The project was partially
sponsored by Global Cooling, with additional support sup-
plied by the multidisciplinary project (MDP) program at TU/e
and the Undergraduate Research Opportunity Program

* Chemical/Environmental Engineering, National University oJ *.
114


(UROP) at NUS. The company participated by supplying the
Stirling cooler and feedback on the design. Various overseas
communication methods were established to facilitate com-
munication and to ensure that the parameters and experiments
were conducted under comparable conditions.

UROP PROGRAM AT NUS
The Undergraduate Research Opportunities Program
(UROP) initiated by the faculty at NUS is a special program
that helps undergraduate students strengthen their research
experience and their life-long learning ability. The program
encourages research that involves cross-departmental partici-
pation, allowing undergraduate students to enhance and ap-
ply their knowledge of the latest tch'lciii .1, .:. Due to the sig-
nificance of the program, the National Science and Technol-
ogy Board in Singapore elevated it to the national level by
holding an annual UROP congress where the participating
students could present their research findings and receive com-
mendable recognition.

Jetse C. Reijenga is Associate Professor of Chemical Engineering and
Chemistry. He received his PhD (1984) and MSc (1978) in chemical engi-
neering from the Eindhoven University of Technology His research inter-
ests include fundamentals and mathematical modeling of electro separa-
tion techniques and the application of information and communication tech-
nologies to education in chemical engineering and chemistry.
Hendry Siepe is an Academic Staff Member at the Center of Technology
for Sustainable Development. He received his BSc degree in mechanical
engineering from the HTS in Groningen (1987), his Master Degree in Psy-
chology from the University of Groningen (1994), and his degree of Master
of Technological Design from Eindhoven University of Technology (1997).
Liya Yu isAssistant Professor of Environmental Engineering. She received
her PhD (1997) and MSc (1990) in civil engineering from Stanford Univer-
sity and her BSc in environmental engineering from Natn'l Cheng-Kung in
1988. Her research interests include size distributions in soot during com-
bustion and investigation of ambient NPAC concentrations.
Chi-Hwa Wang is Assistant Professor of Chemical Engineering. He re-
ceived his PhD (1995) and MA (1993) in chemical engineering from
Princeton, his MSc in biomedical engineering from Johns Hopkins (1991),
and his BSc in chemical engineering from Natn'l Taiwan (1987). His re-
search interests include solid/liquid separation, drug delivery systems, and
flow and dynamics of granular materials.


Copyright ChE Division ofASEE 2003
Chemical Engineering Education










The students participating in the UROP projects were re-
quired to start their research during their second- or third-
year of study to ensure its completion. A minimum of 65 hours
over two consecutive semesters was scheduled to complete a
satisfactory project. Each student had to submit a 4-page pa-
per for final assessment, and a pass-or-fail grade was awarded.
It should be noted that additional requirements for the project
were given due to the special nature of its international con-
nection with the MDP program at TU/e. Specification of the
requirements and assessment are discussed in detail below.
The working team at NUS comprised eight undergraduate
students who were in their second year of study. Supervision
was provided mainly by two full-time academic staff mem-
bers in the ChEE department, while other engineering de-
partments (such as the mechanical engineering and electri-
cal/computer engineering departments) were occasionally
consulted for relevant technical questions.

MDP PROGRAM AT TU/E
The inter-departmental Centre for Sustainable Tc'hiiim, ,h .1,**
at TU/e played a key role during the 1990s in initiating
multidisciplinary project work as an optional activity for stu-
dents of different departments to work together on a sub-
ject related to sustainability. Participating departments
include chemical engineering and chemistry (400 MSc
students), mechanical engineering (700 MSc students),
and applied physics (100 MSc students). Multidisciplinary
projects are now a compulsory part of the curriculum for
most TU/e departments.
In the departments of chemical engineering/chemistry and
applied physics, the MDP program is placed in the fourth
year of study, at the beginning of Master-degree work, so the
students will have sufficient background to apply their knowl-
edge and integrate different expertise from other students.
On the other hand, other departments at TU/e, such as me-
chanical engineering, place MDP projects during the third
year of the curriculum in order to conclude the phase of ful-
filling the Bachelor degree. As a result, the various research
teams of MDP programs often consist of students with dif-
ferent backgrounds in educational experience (different years)
and scientific/engineering training (different departments).
An MDP group at TU/e usually consists of 5 to 7 students,
preferably with different backgrounds. A 6-credit unit is
awarded, requiring approximately 240 working hours to com-
plete the project within a single trimester (10-12 weeks). The
students usually work on the design of a prototype based on
literature study. For the current project, the team at TU/e con-
sisted of six undergraduate students from three different de-
partments (chemical engineering/chemistry, mechanical en-
gineering, and applied physics), some of whom had previous
experience in collaborative project work. In addition to the
supervision facilitated by two full-time faculty members, the
students were encouraged to search for additional expertise,
Spring 2003


both inside and outside the university.

EDUCATIONAL GOALS
The proposed international Multidisciplinary Project (MDP)
was a design-oriented collaboration with a specific economic
and societal context. The operating procedures in the project
were conducted in parallel by two research teams at NUS
and TU/e. The educational goals to be achieved included
Working on projects
Dealing with practical problems
*Applying already-acquired integrated (technical) knowledge
Localizing and acquiring new knowledge and information
Working on a team with students from iliffe'rnt backgrounds
Developing and applying communicative skills, presentation
skills, and discussion techniques
The purpose of an MDP is to involve undergraduate stu-
dents in ongoing collaborative design work. MDP should ben-
efit students by
Enhancing their knowledge of the newest technology
Providing an opportunity to acquire skills for the intellectual
process of inquiry
Encouraging students, faculty members, and client companies
to interact and form closer ties
Rewarding students with .. i.. ,.. ofparticipationfor
successful completion of an MDP project
Exchanging information and ideas with a parallel group
abroad
In addition, to focus on the goal of group dynamics, a num-
ber of team-building sessions were held to address some of
the aspects that play an important role within a group, such
as decision making, leadership, communication, conflict han-
dling, group-style inventory, and pilot peer-review.
The NUS group found that the project involved acquisi-
tion of new knowledge because the group members were only
equipped with two years of undergraduate education and were
still under basic training in chemical engineering. Hence, the
group spent a substantial amount of time on self-study to fa-
miliarize themselves with the project-related subjects.

THE INTERDISCIPLINARY STRUCTURE
The students operated as two teams of engineers from the
virtual company MDP International (the virtual contractor)
within a (virtual) budget agreed to by Global Cooling. Esti-
mation of various costs was included as part of the project.
Students participating in the program were from the Depart-
ment of Chemical and Environmental Engineering at NUS,
and the Departments of Chemical Engineering and Chemis-
try, Applied Physics, and Mechanical Engineering at TU/e.
Global Cooling and MDP International agreed on a con-
tract and the groups were responsible for documenting and
periodically reporting on the virtual cost. Global Cooling
supplied the Stirling cooler and knowledge, while the team
at TU/e purchased the refrigerators and (initially) the solar
115










panels for both parties, to ensure that the parameters and ex-
periments were conducted under comparable conditions. On
the other hand, the National Undergraduate Research Oppor-
tunity Program and the Centre for Advanced Chemical Engi-
neering at NUS jointly supported the NUS group by offering
the necessary facilities and funding for the purchase of a DC
compressor, along with the construction materials and re-
quired accessories.

OBJECTIVE OF THE JOINT PROJECT
The objective of the project was to design a photovoltaic
refrigerator. The World Bank estimates that in today's world,
about two billion people have no access to modem energy
services. They live, for the most part, in developing coun-
tries in parts of Africa, Asia, and Latin America. For their
energy supply, they are dependent on often-scarce biomass
sources such as wood and dried dung. Photovoltaic (PV)
energy technologies now make it possible to offer sustain-
able modem energy services to those who live relatively far
from a central electric grid.1,21 In most countries, there are
three major areas in which PV will be preferably applied:
lighting, communication, and cooking and cooling. This
project focused on building a solar-powered cooling system.
The objective of the project was to design and manufac-
ture two PV refrigerator prototypes to function as efficiently
as possible, using either the Stirling cooler or the DC com-
pressor. A test protocol had to be created that would enable
comparison of the results for the two systems (PV-refrigera-
tor connected to PV-panels). Finally, a testing report compar-
ing both systems had to be presented. Efficiency was consid-
ered in terms of the conversion of sunlight energy to maintain
the cooling chamber at desired temperatures. The teams used
identical refrigerators and solar panels as their base material.
The requirements regarding the functioning of the refrig-
erator were
At environmental temperatures between 32 C and 43 C, the
inner temperature of the cabinet should remain between 00C
and 8C
With respect to cooling rate, a minimum of 2 liters of water
should be cooled down to 50C within 24 hours
The system was limited to using a thermal storage ..iiT. ,
(such as water), while the use of a chemical battery was not
allowed
Without sunlight, the thermal storage should be able to
maintain the refrigerator at temperatures between 00C and
8 C for at least 24 hours
The refrigerator using the Stirling cooler was required to
meet two additional conditions of
It should have a thermal siphon at the cold and the hot end of
the system
It should preferably have a maximum temperature diffi'len over the heat exchanger of 50C per side
The variable factors in this project were the selection of
116


the cooling system and the interaction between the cooling
system and the solar panel. The NUS team used a DC com-
pressor as a cooling engine, while the TU/e team used a
Stirling cooler. Initially, both groups focused on the theoreti-
cal research of the subject matter and individual components.
Next, some experiments were conducted to assess individual
components regarding the working properties, which included
Heat leakage in the Samsung refrigerator
Variation of the output voltage of the solar panels with the in-
tensity of light
COP and capacities of the DC compressor at various condi-
tions
Apart from the actual design, attention was also given to
areas such as safety, environmental concerns, and market-
ability. One of the major problems in producing equipment
for markets in developing countries is the initially limited
volumes to be marketed. The chances of a PV refrigerator
being produced in substantial numbers would significantly
depend on the richer parts of the world also presenting a
market for such a device. One of the niches for this device
could be the outdoor (sporting and c.Inpini.i market.
An appreciable amount of attention was directed toward
the question of sustainability. The subject of the MDP shows
close relevance with the use of sustainable tc:'liii .1 -.,, and
therefore sustainable technology had to be a key feature of
the research question. That is, in addition to the technical
aspects of the subject, students had to research environmen-
tal and social aspects of the subject and had to consider
sustainability aspects. In this way, students were required to
integrate their specific technical skill with knowledge of
sustainability in their design and final report.
The students on both teams had assistance from techni-
cians in building the prototype, to ensure sufficient progress.
The main areas in which assistance was required were the
construction of the buffer container and the disassembly of
the original refrigerator.
A market analysis was conducted simultaneously with the
construction of the photovoltaic refrigerators. Factors that
were considered included pricing the photovoltaic refrigera-
tors so that it would be attractive to targeted customers, namely
the medical sectors in developing countries or sport and camp-
ing companies in developed countries. Other aspects included
in this economic analysis were production volume, shipping,
and assembly.

TIME TABLE
Schedules of the academic year at TU/e and NUS vary
greatly (trimester vs. semester), a severe drawback when
scheduling such inter-university projects. The initial sched-
ule was planned through a consensus between the staff mem-
bers from both universities, with preliminary input from stu-
dents being solicited. During the first videoconference, the
schedule was modified subsequent to a discussion between
Chemical Engineering Education










two student teams. To achieve comparable progress for evalu-
ation, a 17-week timetable was eventually compiled (from
September 2000 to June 2001) that accomodated the holi-
days and examination periods at the respective universities.
Based on the expected 240 hours per student at TU/e, this
corresponded to roughly 14 hours per week.

PHASING OF THE PROJECT
The various phases of the project spanned 17 weeks and
included the components of research, coupling, testing, mar-
keting, and ending the project. It should be noted that some
of the phases had to be done simultaneously to achieve proper
progress. The following paragraphs contain more details about
the activities planned for the various phases of the project.
1st Phase (week 1 through week 4) This phase, which
took about one-quarter of the total project time, was divided
into two parts: orientation and purchase. Orientation was fo-
cused on gathering and processing information on the vari-
ous elements of the photovoltaic refrigerator. The aim was to
gain as much insight as possible regarding its operation and
the efficiencies of the individual elements, which were an
important consideration in the calculation of the required
power of the solar panels.
A lot of self-reading and sales research was carried out in
parallel to find a suitable DC compressor (the Stirling cooler
was provided by Global C, .% lii i. During this phase, a project
plan was devised that required deliverable goals and realistic
planning in detail. A financial budget that met the target range
of the project served to conclude the first phase. The budget
proposed by both teams actually showed virtual expenses.
The "virtual" budget consisted of four primary costs: wages,
equipment and material, working facilities, and stationery
costs. The total virtual budget was around US $14,000. In
contrast, the real project budget, excluding the cost for
wages and working facilities, came to about US $2,500.
The overall expenditures were about 92% of the proposed
project budget (NUS team), which is a valuable outcome
for executing the project.
2nd Phase (week 5 through week
8_) During the second phase, which
spanned the same length of time as TA
the first phase, students started their Gradin
research relevant to the project. At-
tention was paid primarily to the de-
sign of couplings between the vari- Assessment
ous elements. Couplings between Effectiveness of Team Wo
the refrigerator and the DC compres- Project Plan
sor or Stirling cooler, between the Interim Report
solar panels and the refrigerator, and Interim Presentation
between the buffer and the refrig- Final Report
erator were investigated. The theo- Final Presentation
retical design was accomplished in Total
the last two weeks of this phase,
Spring 2003


while the prototype design was consolidated in the 6th week.
At the end of the second phase, students had to produce an
interim report with details about the relevant choices and as-
sumptions that they had made, along with a report of their
progress and possible adjustments for the remaining project.
In addition, students had to present their up-to-date results.
3rd Phase (week 9 through week 17) The third phase
comprised the major milestones of the project over 9 weeks
(half of the project time). During this phase, development of
a test protocol was initiated. Both student teams used the ini-
tial period of this phase to clarify and streamline the mea-
surement standards and criteria for reasonable comparisons
between the prototypes. The first round of testing was car-
ried out during weeks 13 and 14, and both teams conducted a
second round of testing as well as some extra tests (which
differed for each team) during the 15th week. In preparing
the final report, each of the team members worked on a dif-
ferent chapter, with the results being compiled by a team edi-
tor. At the end of the third phase, students were expected to
finalize their project and submit the final report. A final presen-
tation during a videoconference concluded this MDP project.

GRADING
Table 1 shows the assessment scale of the various grading
criteria of the project. The grading criteria included (with
corresponding weighing factor in parentheses) the final re-
port (3), the final presentation (2), the project plan (2), in-
between oral and written presentations (2), and group par-
ticipation (1). The (sub) grades are on a 1-to-10 scale, rounded
off to multiples of 0.5. Evaluation of teamwork effectiveness
assessed delegation among group members and organization
of the research work. The MDP students also had to give a
formal interim presentation on their preliminary results to
their respective project tutors at TU/e and NUS. They were
asked to focus on the project progress as compared to the
original project plan. In addition to the interim and final re-
port, feedback from the client, Global Cooling, also played
an important role in evaluating the final deliverables of the
individual groups.
A pilot peer review that included
E 1 individual and mutual assessment
proach was part of the MDP educational
goal at TU/e. It was first exercised
Supervisor of Global on a trial basis halfway through the
ale EUTORNUS Cooling project. Students were asked to
1 evaluate each other on two aspects:
2 1) specific (1p, ,ii i\ c') ways a mem-
S ber contributed to teamwork and 2)
1 additional improvement the student
3 should strive for. In addition to dis-
2 cussion, the students compiled a
<0 brief confidential report for the su-
pervising staff. This peer review
117


BLI
g Ap]


Sca
)rk






I










was also exercised at the end of the project to evaluate the
progress of individual students in light of the previous sug-
gestions from team members.
A final report (in electronic format) to the client and the
tutors at both TU/e and NUS had to be submitted for grading
a week before the final videoconference. A final evaluation
was then completed by the client and staff members toward
the end of the final videoconference. Within one week after
grading, students were expected to submit a corrected report
that addressed the remarks provided by the staff members at
the respective universities and Global Cooling, so that the
printed edition could be processed in time.

ROLES OF CLIENT/COACH/ORGANIZER
There are a number of roles played by different people dur-
ing the project. One role is the "contractor"-the person who
has a research question and who is highly interested in the
project's outcome. This person is often an expert on the sub-
ject. The contractor can co-decide on the quality of the project
plan and on the quality of the interim and final reports and
presentations.
Another role is the "coach," who follows the progress and
process of the project and is the person to whom students can
turn with daily questions. He/she can also act (if necessary)
as an intermediary between the group and the people from
the "outside world." As a coach, this person can stimulate
and motivate the group and guide and promote their progress.
A third role is the "organizer." This person works mainly
in the background, making sure that facilities such as special
training, overall finance, and a place to work, are available.
Interaction between the three participants above and the
students was made possible via regular e-mails and ICQ ses-
sions. In addition, there were four videoconferences held
during the program that facilitated idea exchange via direct
"face-to-face" discussion. The MDP students were also re-
quired to give a formal interim presentation on their prelimi-
nary results to their respective project tutors at both universi-
ties. They were asked to focus on the project progress rather
than the original project plan.
The feedback and comments from the client (Global Cool-
ing) were considerations in grading the interim report, the
presentation, and the final report. The MDP students used
multimedia facilities to record the relevant project materials
in electronic form (e.g., CD-ROMS). These materials were
mailed or e-mailed to the respective client, coaches, organiz-
ers, and partner-group members for their comments. The feed-
back was subsequently incorporated into the latter part of the
MDP project work and report.

COMMUNICATION FORMATS
Since the groups came from different cultures, mutual un-
derstanding between them was very important for stimulat-
118


ing constructive working dynamics and for enhancing com-
parable interpretation of the project. The leaders of both
projects communicated at least once a week to monitor the
groups' progress and to ensure achievement of the short-term
goals. In addition, frequent communication between the sev-
eral subgroups at both universities took place via e-mail and
ICQ sessions (real-time "chat" communication over the
internet). Four videoconferences were scheduled to obtain
mutual understanding and to enhance cohesive execution of
the research project. Furthermore, there was communication
between the academic staff members at both universities to
resolve questions that arose and on administrative matters
such as scheduling and the agenda of the videoconference.
Meeting minutes included actions taken, results obtained,
and decisions made and were mailed to the other teams and
coaches in order to achieve the desired synchronization. Each
TU/e student had a notebook computer, and the group as a
whole had its own MDP room with network connections. In
addition, they had a group e-mail account and a separate
website for communication purposes. Additionally, the stu-
dents frequently used ICQ accounts for exchanging ideas and
making decisions with the counter group abroad. The MDP
groups at TU/e had a weekly meeting in which the academic
staff members were present.
The NUS group members were given laboratory space in
the engineering workshop that was equipped with networked
personal computers and the necessary facilities for regular
meetings. Individual group members took turns organizing
the meetings to discuss the project's progress.

EXPERIMENTAL RESULTS
Individual prototypes built with a Stirling cooler and a DC
compressor were accomplished at the end of the project. Both
teams performed comprehensive and identical tests, compar-
ing the efficiency of the systems. Daylight cycles were char-
acterized and calibrated in both countries to ensure that the
testing environments were comparable. Due to different volt-
age requirements by the Stirling cooler and the DC compres-
sor, the exact daylight cycle and various parameters of the
DC compressor, such as suction pressure, input voltage, and
current, were investigated before the final tests. Initial ex-
periments were conducted with varying buffer amounts and
container types to obtain an estimate of the heat leakage rate
from the refrigerator.
Water was chosen as the buffer material, due mainly to its
availability and well-known properties. Using energy con-
servation laws, an estimate of the buffer amount was obtained
after considering the heat transfer (enhanced by fins) between
the buffer surroundings inside the refrigerator and the buffer
itself. Due to the different power-supply levels, designing the
buffer container and the fins was different for both groups.
While certain additional adjustments were made by both teams
before the final performance tests of the refrigerators coupled
Chemical Engineering Education











with solar energy, a mutually agreed test
protocol was established to assess the ef-
ficiency of the individual designs.
To examine the efficiency of the refrig-
erators, three major stages were evaluated
as a function of time needed: the start-up
stage, the temperature-maintenance stage,
and the cool-down stage. The test results
successfully met the requirements posed Sta
by Global Cooling. Table 2 shows one of Co
the tests for both refrigerators. In general, Ke
the Stirling-cooler system showed a higher Sta
efficiency and demonstrated more steady Co
temperature profiles, shorter start-up time, He
and longer "keep-cool" time. In contrast,
the DC-compressor system gave faster
cool-down, with favorable temperature profiles.


EVALUATION
Part of the project evaluation was devoted to illustrating
how the project objectives were achieved.
Team Work Overall, the multidisciplinary project exposed
students to a research project in a practical way. Although the
initial period of team formation was fraught with difficulties in
work allocation and coordination, the members learned to work
with one another and coordinate advanced planning, establishing
infrastructure, decision making, critical thinking, self-evaluation,
corresponding improvement, dealing with conflicts, and overcom-
ing differences.
Applying Technical Knowledge The various tasks enabled
students to apply learned knowledge and to acquire new knowl-
edge. For example, foundation training may suffice to test the
solar panels, but in-depth studies were required to resolve more
complex problems such as the proposed power conditioning unit.
Students also found that theories given in class don't always agree
with real life, so they developed creative approaches and
independent thinking to properly interpret data for situations
beyond their academic expertise.
Resolving Practical Problems Students experienced several
practical problems, such as how to best design the buffer
container for the refrigerator powered by a compressor. Such first-
hand experience in problem solving is not offered by current
academic courses.
Developing and Applying Communication Skills The
students learned to refine their communication skills to efficiently
pin-point useful resources, clearly convey problems, and
effectively communicate with others. The videoconferencing
presentations reinforced students' technical communication skills,
and they found it a challenging way to interact with overseas
counterparts.

CONCLUSIONS
This project contained the uniqueness of multidisciplinary,
international, and industrial collaboration. Students were par-
ticularly challenged to apply fundamental knowledge, use
their creativity, and interpret results. Furthermore, they ex-
Spring 2003


TABLE 2
Test Results of
Refrigerator Performance
(Courtesy .. i.i, ii iot,., .,,
by both MDP ... -"'". -

Stirling DC
Cooler Compressor
.rt-up Time (hrs) 152 177
ol-down Time (hrs) 12 5.8
ep-cool Time (hrs) 47 31.6
.rt-up COP (-) 1.32 0.97
ol-down COP (-) 0.88 0.56
at Leak (W) 16.6 11.8


proved cooperation.
The importance of video conferencing for decision mak-
ing was overestimated, whereas the usefulness of chat ses-
sions was underestimated. Chatting was preferred by the stu-
dents in spite of local time differences. Direct communica-
tion proved essential for mutual understanding and agree-
ment on important points.
Different academic calendars at the two universities made
it difficult to plan the project, but spreading it over the entire
academic year proved essential because of its practical and
experimental aspects (i.e., material delivery times, construc-
tion of and debugging the prototype, testing experiments).
The students were enthusiastic about the multicultural com-
munication aspect and the opportunity for experimental de-
sign and consequently spent 70% more time on the project
than originally intended.

ACKNOWLEDGMENTS
The authors thank TU/e and NUS for the support of MDP
(at TU/e) and CAChE and UROP (at NUS). Contributions
from the MDP students are also appreciated: Arjan Buijsse,
Paul Scholtes, Bastiaan Bergman, Ronny de Ridder, Maarten
Blox, and Thijs Adriaans from TU/e, and Josephine Yeo Siew
Khim, Ng Chwee Lin, Wuang Shy Chyi, Ashwin
Balasubramanian, Kwong Bing Fai, Jason Chew Sin Yong,
George Ng Ming Horng, and Ong Guan Tien from NUS.
We also thank Dr. Suryadevara Madhusudana Rao for his
technical support.

REFERENCES
1. Fahrenbruch, A.F., and R.H. Bube, Fundamentals of Solar Cells: Pho-
tovoltaic Energy Conversion, Academic, New York, NY (1983)
2. Zweibel, K., Harnessing Solar Power: The Photovoltaic ( .'.'..
Plenum, New York, NY (1990)
3. Reijenga, J.C., H. Siepe, L.E.Yu, and C.H. Wang, "Collaborative Learn-
ing and Cyber-Cooperation in Multidisciplinary Projects," BITE Con-
ference, Eindhoven, The Netherlands (2001) bite> I


perienced the importance of communi-
cation skills and learned the importance
of a constructive attitude.
Coordination of such a project is com-
plicated and requires a lot of effort. It
provides, however, a unique learning op-
portunity in working with peers, with dif-
ferent knowledge backgrounds and dif-
ferent cultural backgrounds.
The impact of different backgrounds
was underestimated. It was late in the
project that these differences were iden-
tified, because they resulted in misunder-
standings. Solving these misunderstand-
ings by intensive communication brought
both groups much closer and greatly im-























June 22-25, 2003 Nashville, Tennessee


Technical Sessions



Sunday, June 22,2003
Session 0413: Teaching Teaming, Writing, and Speaking
Moderators: Steven W Peretti, Lisa G. Bullard, Chris Anson, Deanna P. Dannels
ABET requirements, as well as communication-across-the-curriculum initiatives, have focused faculty attention on how to effectively integrate teaming, writing,
and speaking instruction within the engineering curriculum. Sponsored by an NSF-Action Agenda grant, a multidisciplinary faculty team at North Carolina State
University has developed a set of teaming, writing, and speaking instructional materials for an engineering design course and an engineering laboratory course.
Workshop participants will receive a CD and hard copy of instructional materials for both courses, evaluation rubrics for written and oral reports, and recommenda-
tions on effective implementation models based on the size of the department, the expertise of the instructor, and available campus resources.


SMonday, June 23,2003
Session 1313: Novel Courses for ChEs
Moderators: Jason Keith and Veronica Burrows
1. "A New Chemical Engineering Senior Elective Course: Principles of Food Engineering," Mariano Savelski
2. "Integration of Microelectronics-Based Unit Operations into the ChE Curriculum" Milo Koretsky, ., :,. Shoichi Kimura, Skip Rochefort
3. "Pediment Graduate Course in Transport Phenomena," William Krantz
4. "Sparking Student Interest in El I- 11 .- .. I..i- .. ...... .. Robert Hesketh, Stephanie Farrell, C. Stewart Slater
5. "Teaching Packaging Engineering at Christian Brothers University," Asit Ray
6. "Fundamentals, Design, and Applictions of Drug Delivery Systems," Stephanie Farrell

Session 1413: Design in the ChE Curriculum
Moderators: David Silverstein and Priscilla Hill
1. "Life-Long Learning Experiences and Simulating Multi-Disciplinary Teamwork Experiences Through Unusual Capstone Design Projects," Joseph Shaeiwitz,
Richard Turton
2. "New Topics in Chemical Engineering Design," Paul Blowers
3. "An Economic Model for Capstone Design," Rudy Rogers
4. "A Web-Based Case Study for the Chemical Engineering Capstone Course," Lisa Bullard
5. "Challenging the Freshman: Freshman Design in ChE at Rose-Hulman Institute of Technology," Atanas Serbezov, Carl Abegg, Jerry Caskey, Sharon Sauer


Tuesday, June 24,2003
Session 2213: Recruitment and Outreach in ChE
Moderators: Anne Marie Flynn and Mariano J. Savelski
1. "Cookies and Diapers and Chemical Engineering," Lisa Bullard
2. "Integrating Chemical Engineering into High School Sciences Classrooms," Deran Hanesian
3. "OSU GK-12 Program for the Enhancement of Science Education in Oregon Schools," Willie Rochefort, Dan Arp, Edith Gummer, Tricia Lytton, Haack Margie
4. "Science, Technology, Engineering, and Mathematics Talent Expansion Program," Taryn Bayles
5. "Using an Enrichment Program to Introduce High School Students to ChE," Paul Dunbar, Rhonda Lee, David Silverstein, Jim Smart
6. "Chemically Powered Toy Cars: A Way to Interest High School Students in a Chemical Engineering Career," Christi Luks, Laura Ford

Session 2313: Innovations in the ChE Laboratory
Moderators: S. Scott Moor and Jim Henry
1. "The Fuel Cell-An Ideal Chemical Engineering Undergraduate Experiment," Suzanne Fenton, James Fenton, H. Russel Kunz
2. "A Novel Unit Operations Project to Reinforce the Concepts of Reactor Design and Transport Phenomena," Sundarc 1! i' Benjamin Lawrence, R.


Chemical Engineering Education












Russel Rhinehart
3. "Simple, Low-Cost Demonstrations for UO II (Mass Transfer Operations)," I Piergiovanni
4. "Use of Lab Experiments to Build Transport Concepts," Nam Kim
5. "A Novel Fluid Flow Demonstration/Unit Operations Experiment," Ronald Wille. l Bina, Ralph Buonopane, Guido Lopez, Deniz Turan
6. "Institutionalizing the Multidisciplinary Lab Experience," R. Worden

Session 2613: The Biology Interface
Moderators: I Piergiovanni and Stephanie Farrell
1. "Teaching of Engineering Biotechnology," Raj Mutharasan
2. "Seamless Integration of Chemical and Biological Engineering in the Undergraduate Curriculum," Howard Saltsburg, Gregory Botsaris, Maria Flytzani-
Stephanopoulos, David Kaplin, Kyongbum Lee
3. "Integrating Biology and Chemical Engineering at the Freshman and Sophomore Levels," Kathryn Hollar, Stephanie Farrell, Gregory Hecht, Patricia Mosto
4. \i. ...i....... i. ll. Speaking: Preparing Chemical Engineers for Careers in Life Sciences," William French
5. "ChE Power! A Hands-On Introduction to Energy Balances on the Human Body," Stephanie Farrell, Robert Hesketh, Mariano Savelski


Wednesday, June 25, 2003
Session 3213: Learning Enhancements for ChE Courses
Moderators: Jim Smart and John Gossage
1. "Improving Critical Thinking and Creative Problem Solving Skills by Interactive Troubleshooting," Nihat Gurmen, H. Scott Fogler, John J. Lucas
2. "Incorporating Computational Fluid Dynamics in the Chemical Reactor Design Course," Randy Lewis, i I! ,',
3. "Web-Based Instructional Tools for Heat and Mass Transfer," Jason Keith, Haishan Zheng
4. "Experiments in the Classroom: Examples of Inductive Learning with Classroom-Friendly Laboratory Kits," S. Scott Moor, I Piergiovanni
5. "Thermo-CD: An Electronic Text for the Introduction to Thermodynamics Course," William Baratuci, Angela Linse
6. "Development of an Intelligent Tutor for Teaching Material Balances to First-Year Students," John Harb, Paul Miller, Kenneth Solen, Richard Swan

Session 3413: Advisory Boards and Program Assessment
Moderators: James Newell and Randy Lewis
1. "Using Standardized Examinations to Assess Chemical Engineering Programs," Keith Schimmel, Shamsuddin Ilias, I
2. "Structuring Program Assessment to Yield Useful Information for ChE Faculty," Helen Qammar, Teresa Cutright
3. I..-. .... I v....wements Resulting from Completion of One ABET 2000 Assessment Cycle," Sindee Simon, Lloyd Heinze, Theodore Wiesner
4. "Departmental Advisory Boards-Their Creation, Operation, and Optimization," Michael Cutlip
5. "Involvement of the Departmental Advisory Board with Curriculum and Student Recruitment Issues," Dana Knox, Basil Baltzis
6. "Effective Use of External Advisory Boards," Kirk Schulz

Session 3513: Statistics in the ChE Curriculum
Moderators: Valerie Young and Donald Visco
1. "Variation, variation, very-action, everywhere, but... ," Milo Koretsky
2. "The Use of Active Learning in the Design of Engineering Experiments," Gerardine Botte
3. "Use of an Applied Statistical Method to Optimize Efficiency of an Air Pollution Scrubber Within an Undergraduate Laboratory," Jim Smart
4. "Designing a Statistics Course for Chemical Engineers," Valerie Young
5. "Teaching Statistical Experimental Design Using a Gas Chromatography Experiment," Douglas Ludlow, Robert Mollenkamp
6. "Integration of Statistics Throughout the Undergraduate Curriculum: Use of the Senior Chemical Engineering Unit Operations Laboratory as an End-of-Program
Statistics Assessment Course," Michael Prudich, Darin Ridgway, Valerie Young

Session 3613: Teamwork and Assessment in the Classroom
Moderators: Joseph Shaeiwitz and Andrew Kline
1. "Imbedding Assessment and Achievement in Course Los with Periodic Reflection" Franklin I Shamsuddin Ilias
2. "Assessment in High Performance Learning Environments," Pedro Arce, Sharon Sauer
3. "Developing Metacognitive Engineering Teams," James Newell
4. "Observations on Forming Teams and Assessing Teamwork," Joseph Shaeiwitz
5. "Rubric Development for Assessment of Multi-Disciplinary Team Projects," Kevin Dahm, James Newell


ChE Executive Committee Meeting
(Breakfast, Ticketed)
Monday, June 23, 2003
7:00 AM 8:15 AM


ChE Lectureship Award
Presentation
Monday, June 23, 2003
4:30 PM 6:00 PM


ChE Division Banquet
(Off-site; Ticketed)
Monday, June 23, 2003
6:30 PM 9:00 PM


ChE Chairpersons Breakfast
(Ticketed)
Wednesday, June 25, 2003
7:00 AM 8:15 AM


Spring 2003


Division Business Meeting/Luncheon
(Ticketed)
Tuesday, June 24, 2003
12:30 PM 2:00 PM











outreach


The Value of Good

RECOMMENDATION LETTERS




GARY L. FOUTCH
Oklahoma State University Stillwater, OK 74078


Whether you currently have a job, are looking for
one, are up for promotion or tenure, or are pursu-
ing some other opportunity, sooner or later you
will most likely need a supporting letter. Let's say that you've
just decided to apply for a position, or perhaps a fellowship
or an award. You've spent hours conscientiously filling out
the paperwork and you've asked the best people you can think
of to write letters on your behalf. It seems like you've done
everything right so far, doesn't it?
Well, maybe not.
What did your references say when they agreed to write a
letter for you? Did the conversation go something like, "Pro-
fessor X, I'm applying for the xyz fellowship. Would you be
willing to write a letter of recommendation for me?" with the
Professor replying, "Sure, I'd be happy to"? If that was the
limit of your communication, you may have made a big mis-
take! You've just put your hopes into the hands of someone
1) who may be too busy to write a letter that truly reflects
your talents, 2) who knows very little about you, even if you
think otherwise, 3) who is unfamiliar with the criteria that
will be used to evaluate your application, or 4) who may not
think as positively about you as you think.
Do you think that someone's willingness to write a letter
about you implies that the person supports you? If so, I sug-
gest you rethink your strategy for getting appropriate letters
of support.
I recently heard someone say, "I hear you write a good let-
ter." It was clear this person wasn't looking for a letter that
necessarily said something good about him personally, but
carried the sense that "I hear that you can write letters that
have a high probability of getting me what I want." Perhaps
this doesn't sound like much of a difference, but I can assure
you, it is quite different.
Let me begin by giving the reviewer's perspective of your
application, based on my own experience. I have served four


years as a panelist for the NSF graduate fellowship program
and four years for the Fulbright Foundation. The NSF fel-
lowship program application pool consists primarily of col-
lege seniors, while the Fulbright program that I served on
was for faculty sabbaticals in England, Ireland, and Canada.
All applicants in these national and international competi-
tions are bright, have strong backgrounds, and present good
supporting documentation. Frequently, the deciding factor will
come down to the quality of the reference letters supporting
the application. Quality in this context not only means that
the letter says good things about you, but also that it is be-
lievable and that it addresses the criteria for the award or
position. As a reviewer, I have to believe the supporting
letters-and in a tie-breaker, the most believable letter
can make the difference.
The following examples paraphrase letters I've read. How
would you feel if one of your references said something like
I can't believe Joe Bob asked me to ....i him a
recommendation. He was a horrible student in my
class-when he bothered to show up. There must be
someone more Jc'.\t'inf,, of this award.
What do you think of Joe Bob's chances for a highly com-
petitive award if his application contained such a recommen-
dation? Or, how would you like to be mentioned in a letter
that said


0 Copyright ChE Division ofASEE 2003


Chemical Engineering Education


Gary L. Foutch is Kerr-McGee Chair and Re-
gents Professor at Oklahoma State University,
having joined the School of Chemical Engineer-
ing in 1980. He received all his degrees in chemi-
cal engineering from the University of Missouri-
Rolla, with part of his PhD work at the Techical
University of Munich-Weihenstephan. His re-
search is in the area of transport-limited kinetics
and separations, with current projects on
ultrapure water processing and high-temperature
reactor design.










I am Diiii,..,,;-.,,. 3 Professor X. I have a Nobel Prize
in Chemistry. I know Joe Bob. Award him a fellowship.
What has the committee learned about Joe Bob from this
masterful piece of writing? All I learned was that he knows
an egotistical chemistry professor. I learned nothing about
Joe Bob himself.
Perhaps you think I'm making these letters up, but I assure
you that within a word or two, I have seen
them-the excerpts are as factual as my
memory allows (we can't keep copies of ap- All a
plications). The good news for most of you in these
(but not, unfortunately, for Joe Bob) is that of and ei
the approximately 1200 letters I've read, I es-
timate that only about 10 were that bad. com
are
An example of a reference writer not under- av
standing the criteria for an award is demon- ha
strated by an excerpt from a supporting letter andpr
for a Fulbright that stated and pr
I can think of no better reward for sup
Professor X's accomplishments at
Di.ii .'1,,-.1. U than col. -n ir.'- him and Frequ
his lovely wife to enjoy a relaxing year at decide
C.,,,,1., ,.... will co
At the time, the criteria for the award for which the qu
Professor X was being considered focused on referee
research and/or teaching collaboration between supp
U.S. and foreign scientists and long-term ben- app
efits to both the visitor's and the host's insti-
tutions were important. A reward for past ac-
complishments, or a vacation in the English countryside, was
most certainly not a goal of the program!
There is another type of letter that hurts an application.
Some letter writers make up things, or cut and paste from
other letters, or simply have no idea what to say about the
applicant. These letters quite often contain errors in fact
or actually contradict the body of the application. An ex-
ample follows.
The NSF panels have twenty to thirty reviewers sitting in
the same room who are, for the most part, reading. Occasion-
ally, however, a comment will be made about a statement in
an application. During one of these panels, a colleague noted
that according to the department chairman's supporting let-
ter, two students from the same class of about twenty had
ranked in the "top 5% of the class." (Engineers appreciate
these little mathematical oddities-it's just part of our na-
ture!) This doesn't sound like a big deal so far, but then some-
one else remembered they had also seen that statement. Within
a matter of minutes, seven applications that were submitted
from this same department were checked, and each contained
a letter from the department chairman indicating that each
applicant had been in the top 5% of the class.


ppli
;e n
tern
peti
bri
e sti
groin
'esei
por
neni
lent
ing
me
ality
nce
ortii
lica


Spring 2003


Those letters no longer contained any credibility.
Another possibility is that your references simply do not
remember that much about you, or that they don't remember
what you remember. A few years ago I had a wonderful stu-
dent who I enjoyed teaching and who has kept me updated
once or twice a year through e-mails. Several months ago he
relocated and sent me a note with his new address, adding a
personal note of a memory from his school
days. He related that one day when he was
giants walking down the hall after class, he met me
and two visiting chemical engineers, and that
itional I had invited him to go to lunch with us. He
national said that at the time he had been considering
tions leaving chemical engineering, but that listen-
ght, ing to the industry guys talk about their jobs
wrong and other general topics had revitalized him,
unds, and he ended up staying in the program and
nt good getting his degree. He wanted me to know and
ting to thank me for that lunch invitation. I'm
ati on. afraid that I have no recollection about that
lunch whatsoever! I'm glad I did something
fya the to help him stay committed to engineering,
factor but if he hadn't mentioned it I would never
down to have known. While this is exactly the type
of the of personal story that could be used in a
letters letter of recommendation to show commit-
ig the ment and dedication, it can't be related if
tion. it isn't remembered.
How can you help yourself? There are sev-
eral things I recommend in order to get sup-
porting letters worthy of the time and effort you devote to
your application:
Determine if the letter-writers actually support your ap-
plication. This is easily determined-just ask! Don't start
with, "Will you write a letter of recommendation for me?"
Instead, tell them that you are interested in applying for
a particular program or award and ask them what they
think your chances are. Do they feel you would be com-
petitive? Ask if they have any advice on how to compete
for the job or award. What do they know about your
strengths and weaknesses that would allow you to be
successful if you applied? Ask if they would be support-
ive of your application. DO NOT ask them to write a
letter of support until you have heard their responses to
the above and are convinced that they have your best
interests in mind. If you're not sure, say thanks and walk
away. After some thought, you may conclude that they
should be one of your references after all, and in that
case approach them again with "...remember the con-
versation we had the other day...."
Educate your reviewers. Most potential reviewers will
not know the criteria of the specific award or program.










Even if your letter-writers were familiar with the pro-
gram several years ago, don't assume the criteria are the
same today and that your references are up to date on
them. You need to be sure they understand the criteria
upon which you will be evaluated. Feel free to commu-
nicate which criteria you believe best match your skills
and which you think need the most support.
D Tell your reviewers something about yourself. Tell them
why this award or position is the perfect match for you.
Allow them to make the letter as personal as possible.
They won't have the perspective you have; you have more
knowledge about yourself and why you should be the
recipient than they do. If you can sell them on your
dreams, they will be able to focus that energy into a let-
ter that can truly support you.
- Meet their timetable! Don't ask for a letter that's due
tomorrow. To ensure all deadlines can be met, I suggest
planning ahead by at least two weeks. A rushed letter
will most likely have omissions that could hurt your ap-
plication.
D Consider having an extra letter sent. One too many is
better than one too few. Read the application details or
call the program administrator. Usually, an extra letter
just goes into the file, but the bottom line is not to be a
letter short of the required number. Feel free to get con-
firmation that letters were sent. Some application pro-


cesses have a return postcard so you can be sure.
Try to guide the letter so it matches the narrative appli-
cation and forms you have written. Don't write the letter
for your reference, and if they suggest that you do so, I
recommend you find someone else to do it. You want a
sincere and honest opinion from a conscientious sup-
porter. I suggest that you prepare a letter to your refer-
ence that contains the criteria and a bullet list of items
you feel the letter should consider. A bullet list allows
them to add their own prose as they address key points
so that all letters won't sound alike. Also, just in case, if
you have similar bulleted lists for different references,
mix the order so they don't go down the line and hit the
same points in the same sequence.
Let me add a note specifically to those of you applying for
a Fulbright or other international award. For the high-demand
locations such as England and Germany, you can assume that
all applicants have invitation letters offering a desk and com-
puter access. Look for real ties to your host institution. In
today's world where it's easy to have collaborators from
around the globe, you need to give the judges a reason for
physically being there. Help your references explain why you
have to be overseas. If possible, in addition to the host letter,
have another colleagues) within the same or a nearby coun-
try describe what your presence will mean to them.
Good luck! 7


BjRWletter to the editor


To the Editor;
Regarding the article NJl.ikim Phase Equilibrium More
User-Friendly" by Michael J. Misovich,E11 we endorse some
of the points made, but are also concerned by some gen-
eral attitudes expressed about teaching this subject (and
by extension, chemical engineering thermodynamics in
general, since he makes passing reference to chemical
reaction equilibrium).
On the positive side, we commend the considerable em-
phasis on the calculation of properties and presentation of
the data graphically. We also agree with the importance of
developing an intuitive understanding related to such things
as order-of-magnitude values of thermodynamic quantities,
and the likelihood of the occurrence of azeotropes.
On the other hnd, some statements are made that seem to
place the subject matter in a very limited position relative to
other courses that he mentions. For example
"Phase equilibrium in which abstract concepts are


presented to the near exclusion of practical examples."
"... most phase equilibrium courses (sic) do not connect
these (calculations) to real processes or equipment."
"... this class deals with techniques for generating data...
to the total exclusion of applications."
It seems no wonder then that "students who perform calcula-
tions satisfactorily seem confused over the meaning of what
they have learned." These statements also tend to run counter
to Felder's TIP 1,12] notwithstanding the subsequent empha-
sis on graphical presentation.
To the contrary, we believe that teaching this subject with-
out overtly involving applications (processes and equipment)
amounts to emasculation of it. One thing that should be em-
phasized is that thermodynamics (as the umbrella subject)
provides limiting or boundary solutions to problems, but is
silent on cllicic'i," in various guises, that translates the
limiting-case results into actual results. It is inevitable that


Chemical Engineering Education













CALL FOR


* PAPERS


for the Fall 2003 Graduate Education Issue of


Chemical Engineering Education

We invite articloes on graduate education and research for our fall 2003 issue. If you are interested in contributing,
please send us your name, the subject of the contribution, and the tentative date of submission.

Deadline for Manuscript Submission is June 1. 2003


Respond to: cee@che.ufl.edu


this requires, however, the introduction of actual processes
(and equipment), and in this way bridges can be built to these
other courses.
The author is undoubtedly aware of many such applica-
tions, as he indicates, and we mention only a few (not neces-
sarily directly related to phase equilibrium, but to equilib-
rium in general):
Separation of a condensable from a noncondensable
species (cooler-condenser); also related to
himidification and dehumidification
Eutectic behavior related to the use of ethylene glycol
antifreeze coolant (automobile engine) and its vapor-
liquid counterpart in "steam distillation"
Vapor-compression refrigeration (compressor)
Energy conversion (fuel cell or electrochemical cell in
general
Equilibrium reaction yields, equilibrium species dis-
tribution in general (equilibrium-limited reactor,
whether batch or flow system)
The author also expresses a strong preference for the use
of computer spreadsheets, although he acknowledges the
possible alternative use of metacomputing software (such
as Maple1m3), which, in our opinion, is more efficient. In
addition, this software does not require the trial-and-error
or iteration approaches mentioned by the author for some
of his assignments.
If the goal is to produce graphical visualization of behav-
ior, then spreadsheets have the inherent limitation that the
explicit generation of data must precede the generation of
graphs. Spreadsheets can only easily generate such data if
the equations are available in analytical form; otherwise, trial-
and-error or iterative procedures must be used, as he notes.
In contrast, metacomputing software provides graphing com-
mands that do not require such explicit prior data generation.


Furthermore, any required data can be obtained separately,
without trial-and-error or iterative procedures.
As an example, if plotting the graphs P(x1) and P(yl) for
the ideal system in his Figure 1 is the objective of a student
assignment, Maple requires only the following statements
(only the first two lines are required for the phii i the other
lines relate to cosmetic aspects of the display):


fsazc:= va lue);sazc:=[vaue) ;
plot(Psat2+(Psat2-Psatl)*x,Psa


'sat2/(Psatl+x*(Psat2


?sazi) ) ,x=u..i,
axes=BOXED,xtickmarks=10,labels=["xl,yl","P/mm Hg"],
labeldirections= [HORIZONTAL, VERTICAL],title= [P-x-y diagram"] )
Using the implicitplot command, we can readily construct
Txy diagrams with Maple, for both ideal and nonideal sys-
tems, without trial-and-error or iterative procedures.
As a further example of the use of metacomputing soft-
ware in phase equilibria, we note that Dickson, et al.,1] have
demonstrated the use of Mathcad1E5 to obtain 3-dimensional
vapor-liquid equilibrium envelopes.
In conclusion, although we agree with much of what the
author says, we believe that there is more than he allows in
iii.ik m phase equilibrium more user-friendly."
R.W. Missen
University of Toronto
W.R. Smith
University of Ontario Institute of Technology
References
1. Misovich, M.J., "Making Phase Equilibrium More User-Friendly,"
Chem. Eng. Ed., 36(4), 284 (2002)
2. Felder, R.M., "How to Survive Engineering School," Chem. Eng. Ed.,
37(1), 30 (2003)
3. MAPLE is a registered trademark of Waterloo Maple, Inc.
4. Dickson, J.., J.A. Hart, IV, and Wei-Yin Chen, "Construction and Vi-
sualization of VLE Envelopes in Mathcad," Chem. Eng. Ed., 37(1), 20
(2003)
5. MathCAD is a registered trademark of MathSoft, Inc. 1


Spring 2003











classroom


MATHEMATICAL MODELING AND

PROCESS CONTROL OF

DISTRIBUTED PARAMETER SYSTEMS

Case Study: The One-Dimensional Heated Rod



LAURENT SIMON, NORMAN W. LONEY
New Jersey Institute of Technology Newark, NJ 07102


Distributed parameter systems (DPS) such as chemi-
cal vapor deposition (CVD), nanostructured coat-
ings processing, population balance, transdermal
drug delivery, or film growth are normally represented by
partial differential equations (PDEs). They have important
industrial applications, but controlling them presents theo-
retical and practical challenges.El One of the methods em-
ployed to control first- and second-order systems uses an exact
reduction of a distributed parameter system to a lumped one.E21
The theory of lumped parameter systems can then be used to
design a controller that meets user specifications and desired
quality objectives. Laplace transform is a common technique
used to derive the lumped parameter system. Although the
conversion is straightforward, the inversion of the resulting
Laplace transform equation is usually not trivial.
This paper shows that certain materials covered in math-
ematical modeling and process control courses are good start-
ing points for designing controllers for these systems. The
work is divided into
Section 1, .1. ,/*1,.- with the solution of a one-dimen-
sional rod in the Laplace domain
Section 2, ,,.i ,.* the residue theorem to invert the
Laplace transform
Section 3, .. ,0,, .- with the .. i.. of a PI controller for
set-point ti, .. ,in..
Section 4, includes experiences in .,,.. I,, .. courses in
mathematical methods and chemical process control

SOLUTION OF THE
ONE-DIMENSIONAL ROD PROBLEM
Consider a one-dimensional rod (see Figure 1). The bound-
ary conditions are such that heat from a steam chest is added


to the system at z = 0, while the other end, z = 1, is perfectly
insulated.E21 The variables are
x(z, t) T -Td (1)
u(t) = T Twd (2)
where T and Tw are the temperature of the rod and steam
chest, respectively. Variables x and u represent deviations from
the set-point values Td and Twd. The model equation is
ax(z, t) a2x(z, t)
at az2
The boundary conditions are


ax
a = P(x u)
az


z=0


ax
S= 0 z = 1 (5)
az
The initial condition is
x(z, 0) = 0 (6)
To solve Eqs. (3) to (6), we first take Laplace transforms with
respect to time:

sX(z,s)-x(z,0)= d (7)
dz2


Laurent Simon is Assistant Professor of Chemical Engineering at New
Jersey Institute of Technology He graduated from NJIT with a bachelor's
degree and obtained his Master and Doctorate degrees from Colorado
State University, all in chemical engineering. His current interests are in
bioseparations, process modeling, and control.
Norman W Loney is Associate Professor of Chemical Engineering at New
Jersey Institute of Technology. He has studied chemical engineering at
NJIT and applied mathematics at Courant Institute of Mathematical Sci-
ence. In addition, Dr. Loney has practical experience in process develop-
ment, process design, and in-plant engineering.


Copyright ChE Division ofASEE 2003
Chemical Engineering Education










dX
= (X- U) z=0
dz
dX
-=0 z=l
dz
Using the initial condition, Eq. (7) becomes

d2X
sX(z, s) = 2
Sdz2
In operator form
(s -D2)X= 0

The characteristic equation is
s-r2 =0
with roots
r = _+-s
The general solution is

X= ale-()z +a2e+(T)z
in terms of exponential function, or
X = c1 sinh(zv) + c2 cosh(z-s)
in terms of hyperbolic function.
Using the boundary condition, Eq. (8), one obtains


dXz
dz zo


(8) -XU(s)tanh(4) s (U4+ (s) c4 )
P + .Ftanh(; v P+VstanhV v)


(10) X(s,z) tanh() sinh(zVT) + cosh(z;S)]
U(s) 3+ -s tanh(-)

(22)
Since thermal energy is continually transferred from the up-
per end of the metal rod to the lower end, it is of particular
interest to study the temperature profile at the lower end of
(12) the rod and the time it takes this temperature to settle down
to equilibrium.
At z = 1, Eq. (22) becomes
(13)(s,1) _
p [- tanh()sinh(.)+cosh(.)] (23)
U(s) P + 4s tanh(;4S) tn(
(14) Recall that
cosh2 (z)- sinh2 (z) 1 (24)
(15) so Eq. (23) can also be written as
X(s,1)_ Psech(;S) (25)
U(s) p + tanh(; )


c1 S cosh(0) + c2 s sinh(0)


P[ci sinh(0) + c2 cosh(0) U(s)] (16)
or
C1IS = P[c2 U(s)] (17)
Furthermore, the boundary condition given by Eq. (9) yields
dX= = c= cosh( ) + c2 sinh(V) = 0 (18)
dz zi VS/ v


cl + c2 tanh(4) = 0


(19)


Solving Eqs. (17) and (19) results in
-13U(s) tanh(A4 ) 2 = pu(s) (20
c1 = sta and c2 = US) (20)
P + -4 tanh(4s) P + VF tanh(-)
Therefore, Eq. (15) becomes



Steam chest I-
Tw


Figure 1. A one-dimensional rod heated by a steam chest
of temperature T,. The temperature of the rod at
position z and time t is denoted by T(z,t).
Spring 2003


INVERSION OF THE LAPLACE TRANSFORM
In principle, control design for lumped parameter linear
systems can be used to analyze Eq. (25), but the analysis is
not trivial since the zeros and poles are not easily obtainable.
We seek an expression of the form

G(s) = (s,1) P(s) (26)
U(s) Q(s)
where P and Q are polynomials in s and Q(s) is of higher
degree than P(s).[3]
The inverse transform of G(s) is given by

L-IG(s)} = Res[F(s)est, k] (27)
k=1
where the sum is taken over all the residues of the complex
function F(s)e"t. The function

pk(t)= Res[F(s)est, Sk] (28)
is the residue ofF(s) at the singularities (poles) sk. Its value is
given by

p(t) P(Sk) eSkt (29)
Q'(sk)
where Q'(sk) is the value of dQ/ds evaluated at the singular
points of interest.[3,4]










The quantity P(sk)/Q'(Sk) can be written as

P(sk) P(s) P(s)

SS-Skk
When sk is a multiple pole of order m of F(s), then
W t2 t m-1
Pk(t)=esk(t) A +tA2+-A3+...+ Am (31)
2! (m 1)!
or
m i_1
pk(t)= ek( A (32)
i=1 1)!
where

Ai = lim i) -- i [(s- sk)m F(s)] (33)
s ->sk(m -i)! ds-
Recall that
L a tn -bt = (34)
L{n! +b(s+b)n+1 (34)

G(s) can now be written as a ratio of polynomials. In the
discussion that follows, we will first show that for a step
change in the amount of heat added to the steam chest, the
temperature at z = 1 follows a time trajectory before settling
to a steady-state value. The amount of heat is usually deter-
mined from steady-state analysis, which is very common in
chemical engineering. This is the case of most controlled
membrane devices in which a specified drug concentration
in the donor cell is used in order to reach a required steady-
state concentration in the receiver cell.
This work shows that it is possible to change the heat from
the steam chest in order for the temperature at z = 1 to reach
the desired value in a predetermined manner. In other words,
both the system performance and the final value can be set a
priori. A standard PI controller can be used for this purpose.
With p =1. Eq. (25) becomes

X(s, 1) sech (;)
U(s) 1+ ;s tanh(-4)
The identification of P(s) and Q(s) is not difficult in the case
of polynomials. For expressions involving transcendental
functions, one has to make certain that the numerator does
not involve a singularity. Since the hyperbolic secant func-
tion does not have a singularity, the denominator is repre-
sented by

Q(s) = 1+ 4 tanh(,4s) (36)

The first four poles are s1 = -0.7402, s2 = -11.7349, s3 =
-41.4388, and s4 = -90.80821. The function "FindRoot" in
Mathematical was used to compute these roots. Figure 2
shows a plot of Q as a function of s. Although an infinite
number of poles are obtained, it is customary to use the first
128


two poles since they dominate the system response. Four poles
are taken in this work for increased accuracy.
By taking the derivative of the denominator, Q'(s), one
obtains

1 2 tanh( s)
Q'(s)- sech2 (;)+ (37)

From Eqs. (27) and (29), the inverse Laplace transform is

L tG(s)= P(sl) slt P(s2) S2t P(S3) s3t ,P(s4) eS4t
L- G(s)}= )e + ( + Q(e +-0 e + QTS4)e
Q(Sl) Q s2) Q s3) Q S4

(38)

L- G(s)}= 0.8284e-0.7402t -1.7801e-117349t +
1.9308e-41.4388t -1.9676e-90.8082t (39)
G(s) is then
G(s) 0.8284 1.7801 1.9308 1.9676 (40)
s+0.7402 s+11.7349 s+41.4388 s+90.8082
or
X(s,1)_ -0.988583s3-24.1254s2-1300.31s+32435.1
U(s) s4 +144.722s3 +5421.46s2 +48092.0s+32684.7

CONTROLLER DESIGN
In practice, the size A of the step change in U(s) (the steam
chest temperature) necessary to get a desired value for X(s, 1)
(the temperature at the end of the rod) is usually known from
steady-state analysis or experiments. Therefore, X(s,1) then
becomes

X(s1)= -0.988583s3-24.1254s2-1300.31s+32435.1 A
s4 +144.722s3 +5421.46s2 +48092.0s+32684.7 s
(42)

An important concept in process analysis and control is the
steady-state gain defined as the ratio of steady-state changes

Q(s)








-100 50 100


Figure 2. Characteristic equation Q as a function of
poles s. The poles are in the range (-100, 100).
Chemical Engineering Education










in the process output to sustained changes in process input.J"5
Using the property
lim x(t, 1)= lim [sX(s, 1)] (43)
t-4> s->0
Eq. (42) becomes
lim[sX(s, 1)] = 32435.1 A = 0.9924 A (44)
s-liO 32684.7
The process steady-state gain is 0.9924, which means that
each degree increase or decrease in u(t) will correspond to a
change in x(t) of size 0.9924. To graph the response, one can
invert Eq. (42) to get
x(t,1)=
A[0.9924+0.02167 exp(-90.8082t)-0.04659 exp(-41.4388t)+
0.1517exp(- 1.7349t)-1.119lexp(-0.7402t)] (45)


Figure 3. Response x(t,1) as a result of an input step in-
crease of size 2. This plot is generated using the "step "func-
tion in Matlab. The manipulated input variable is T .


Figure 4. Block diagram of a general feedback control loop.
The transfer functions Go(s), Go(s), G (s), Gd(s), and G(s)
represent the dynamics of the controller, actuator, process,
disturbance, and sensor, respectively. The inputs D(s), E(s),
and Y (s) (in the Laplace domain) are the disturbance, er-
ror, and setpoint, respectively. The output of the system is
denoted by Y(s).
Spring 2003


For example, if we use an A-value of 2 (step change size),
the response x(t,1) is as shown in Figure 3. This plot is ob-
tained by using the "step" function in Matlab. It is widely
accepted that the response reaches its final value when it is
within +5% of its final value and remains constant.J61 The final
value is 1.9847. By setting x(t,1) to 95% of the final value
(1.8855), Eq. (45) is solved to give a time t = 4.2097 seconds,
which is the time it takes the system to reach steady state.
The performance of the system can be improved by pole
placement (also called direct synthesis). The main idea of
pole placement is to design a controller such that the system
has closed-loop poles at desired locations. In this work, only
the methodology and the final results are outlined. Further
details and derivations can be found in the literature.5-71
Consider the block diagram of a general feedback control
loop (seen in Figure 4).E71 The transfer functions Gc(s), G,(s),
G (s), Gd(s), and G (s) represent the dynamics of the control-
ler, actuator, process, disturbance, and sensor, respectively.
G represents how the sensor responds to a change in the tem-
perature. In our example, Y(s) stands for the temperature at z
= 1, which is measured by a thermocouple (Gs). The mea-
sured variable is then compared with the desired value Y.,
yielding an error Y-Y~et (see Figure 4). This deviation is sent
to a controller Gc(s). The output of the controller (which is
the temperature of the steam chest) goes to an actuator or
final control element (i.e., a steam valve) that regulates the
temperature of the chest. Assuming no disturbance to the pro-
cess (D(s) = 0), it can be shown that the closed-loop transfer
function for set-point tracking is given byE71

Y(s) Gp (s)Ga (s)Gc (s)
Gcl(s)= Y(s) Gc(s)Gp(s)Ga(s)Gs(s)+1 (46)

This equation relates the process output to the set point. The
equation
Gc(s)Gp(s)Ga(s)Gs(s)+1= 0 (47)
is called the characteristic equation of the feedback loop. the
roots of this equation are the poles of the feedback process.
Consequently, they determine the response of the process.
For our example, assuming Ga(s) = G (s) = 1, we obtain

Gp (s)Gc (s)
Gcl(s)= G(s)Gs) (48)
Gc(s)GP(s)+1
Solving for Gc(s),

Gc (s= Gel(S) (49)
Gp[1-Gcl(s)]

The pole-placement problem consists of placing the closed-
loop poles at desired locations to meet performance specifi-
cations. A general controller can then be derived using this
procedure. Based on issues related to pole-zero cancellations,
however, and the fact that PID controllers are more avail-
able, we will derive a PI controller. The first step in the pro-
129










cedure is to approximate the plant as a first-order system with
dead time

Gp,(s)= KPe (50)
ps+1
where G p(s) is the transfer function of the first-order system.
The gain Kp = (Ay/Au) is the steady-state process gain. The
quantity Op is the time delay and Tp is the time constant. A
reasonable choice for G (s) isE51

Gce(s)= (51)
Tc s+1
such that the closed-loop transfer function also contains a
time delay. The time constant determines the dynamic path
of the process as it approaches the new steady state. The pa-
rameters 0c and Tc are pre-specified design parameters. The
condition O, -Oc >0 must hold since the controller cannot
respond to a set-point change in less than Op time units.E15
From Eq. (49)
e-Ocs

Gc(s)= KpePS1 (52)
K"ePS e-es


or

Gc(s)= ps+l (53)
Kp Ts+1 e-ecs)

with 0c =Op. Using a first-order Taylor series expansion,
e-es=1-Os (54)


Equation (53) becomes

TS+1 sps+l Tp
Kc Kp (, +0s) Kp (Th +0p )s Kp (T+ p )


which is the form of the PI controller with


K (Cc + 0


1 = "p


where Kc is the controller gain and Ti is the reset
time.
The original plant can now be approximated by
0.9924e-0.1672s
GpI (s) 1.379s+1 (57)
Figure 5 shows that Eq. (56) is a very good model
of the plant dynamics.
Assuming that one wants to reduce the time
constant by a half and one third (in this case
T,=1.379/2=0.6895sec and Tc/3=0.45970sec,
respectively), let us study how the system re-
sponds to a unit step change in the temperature
130


of the steam chest with these design parameters using a PI
control. By using Eq. (56), one obtains cT = 1.3790 sec in
both cases. The controller gains (Kc) are 1.6220 and 2.2166
for time constants of 0.6895 and 0.4597sec respectively. Fig-
ure 6 shows the implementation of the controller using
Simulink. Two loops are shown with the same plant transfer
functions. The first loop is the closed-loop response with the
PI controller, the second one is an open-loop. Both responses
are recorded in block "scope." Figure 7 compares the open-
and closed-loop responses. From the figure, the performance
of the system is greatly improved. Figure 7 shows the system
responses to an input step change of size two.
Using a discrete form for the transfer functions, one can
easily implement the controller at desired sampling intervals.
The Matlab function "c2d" converts the continuous system
to a discrete-time system with specified sample time. The
velocity form of the PI controller can then be used.161 The end


0 1 2 3 4 5 6 7 8

Figure 5. Comparison of the true (solid line) and approxi-
P n mated (dashed line) plant dynamics. The approximated
(55) plant is represented by a first-order plus delay (FOTD)
model.


Figure 6. Diagram of the PI controller using Simulink. The step change is
implemented by the block "Setpoint." The loop G is the closed-loop re-
sponse with the PI controller, G is part of the open-loop configuration.
Both responses are monitored in block "scope." The block "simout" al-
lows the closed and open-loop responses to be saved in the workspace.
Chemical Engineering Education











result is that at sampling time, a process operator can manu-
ally calculate the controller output using a hand calculator.
Since a PI controller is relatively inexpensive, however, and
in view of the increased performance of the system, it is ad-
vantageous to use one in line with the systems and in com-
puting the best adjustable parameters computed off-line. It
should be noted that a PI controller could be tuned with much
less effort using classical tuning approaches such as field tun-
ing, Cohen and Coon, and Ziegler-Nichols tuning methods.E-5
7] Pole placement, however, can be used to develop a general
controller (which may not have a PID or PI structure) de-
signed to meet preset performance criteria.

TEACHING MATHEMATICAL METHODS,
DYNAMICS, AND CONTROL
First-year chemical engineering graduate students at NJIT


with PI control, rc = T/2
--- with P1 control, T = T /3
without PI control


08


S0.6


0.4


02


0 1 2 3 4 5 6 7 8 9 10
I _t[s]
Figure 7. Closed-loop and open-loop response (-) for a
simulation time of 10 seconds. The time constants were
reduced by a half(---) and one third (-*-). A unit step change
was used.


Figure 8. Closed-loop and open-loop response (-) for a
simulation time of 10 seconds. The time constants were
reduced by a half(---) and one third (-*-). A step change of
size 2 was used.
Spring 2003


take a 3-credit class in "Applied Mathematical Methods" in
chemical engineering practice (see textbook[3]). They are also
exposed to an undergraduate 4-credit course that deals with
process dynamics and control. A course in the control of dis-
tributed parameter systems has not yet been offered in the
department, but the potential is being explored through col-
laborative efforts among faculty members, mini-projects with
industrial applications, and extensive research. Such prob-
lems are also ideal for independent studies.
Undergraduate chemical engineering students at NJIT re-
act positively to the process dynamics and control of lumped
parameter systems. With time, they understand Laplace trans-
forms and have no difficulty analyzing dynamic behavior of
feedback-controlled processes. The inversion of Laplace
transforms is, usually, the most challenging part. Solving prob-
lems in class and completing homework assignments help
considerably. These students are also encouraged to use math-
ematical software to plot, find roots, and take derivatives of
special functions (e.g., Mathematica, Matlab).
The students are given examples of industrial chemical
processes in which they use their fundamental knowledge in
mathematics to analyze the system open-loop and closed-loop
dynamics. While using the techniques learned in class (transfer
functions, closed-pole analysis, controller tuiiii.ii) to solve
practical chemical engineering problems, two things become
apparent to them: first, the skills that they are learning are rel-
evant and in demand; second, that they are imbued with knowl-
edge and insight to solve these problems. The students respond
very well to this approach, and some have even become inter-
ested in doing research in control of drug delivery systems.

CONCLUSION
A one-dimensional perfectly insulated rod was solved in
the Laplace domain with given boundary conditions. The
solution in Laplace domain was inverted to the time domain
using the residue theorem. The temperature profile (at the
right end z = 1) was approximated as a first-order system
with a time delay of 0.167 sec and a time constant of 1.379
sec. A proportional-integral (PI) controller was then used to
decrease the time constant of the process by 50 and 33%.

REFERENCES
1. Christophides, P.D., "Control of Nonlinear Distributed Process Sys-
tems: Recent Developments and Challenges,"AIChE J., 47, 514 (2001)
2. Ray, W.H., Advanced Process Control, McGraw-Hill Book Company,
New York, NY (1981)
3. Loney, N.W., Applied Mathematical Methods for ( .... .' i ....
CRC Press LLC, New York, NY (2001)
4. Loney, N.W., "Use of the Residue Theorem to Invert Laplace Trans-
forms," Chem. Eng. Ed., 35, 22 (2001)
5. Seborg, D.E., T.F. Edgar, and D.A. Mellichamp, Process Dynamics
and Control, John Wiley & Sons, Inc., New York, NY (1989)
6. Stephanopoulos, G., Chemical Process Control: An Introduction to
Theory and Practice, PTR Prentice Hall, Englewood Cliffs, NJ (1984)
7. Riggs, J.B., Chemical Process Control, 2nd ed., Ferret Publishing,
Lubbock TX (2001) 1


2 ,
18



1.2
11
08



0
0 1 2 3 4


,,, f i ,o-N .ol, =
S with PI control, = P /3
- without PI control


6 7 8 9 10










curriculum


PROCESS SIMULATION

AND McCABE-THIELE MODELING


Specific Roles in the Learning Process


KEVIN D. DAHM
Rowan University Glassboro, NJ 08028-1701


S standard texts on equilibrium staged separations1,2]
present the McCabe-Thiele, graphical approach as a
primary tool for modeling and designing staged sepa-
ration processes such as distillation, absorption, extraction,
and stripping. The development of process simulation soft-
ware, however, has impacted the way this material is taught.
In a recent surveyE3l of U.S. chemical engineering departments,
57% of the respondents indicated that they now use process
simulators in teaching equilibrium-staged operations, and this
number is presumably still growing. Recently, authors have
discussed methods of integrating process simulators into
lecture coursesE41 and of using simulators to facilitate ma-
jor project work.E51
Simulators certainly have not, and should not, entirely re-
place "hand" solution techniques. The primary pedagogical
concern regarding process simulators is that they function as
black boxes. In many cases students can use them to solve
specific problems without necessarily understanding the
physical process they are modeling.E31 They are likely to ac-
cept the results of the simulation blindly, with no thought of
the potential limitations of the modeling approach used.
One merit of traditional graphical approaches is that they
provide some insight into what the simulator is actually do-
ing. A further consideration is that graphical approaches pro-
vide a convenient framework for visualizing the process.
WankatE6' points out that even experienced engineers "com-
monly use McCabe-Thiele diagrams to understand or help
debug simulation results." But the merit of extending the hand
calculations significantly beyond simple graphical models,
such as using the Ponchon-Savarit method to include the en-
ergy balance, is less clear in the era of process simulation.E1
It is such considerations that led Wankat to recommend
"an eclectic approach that includes classical graphical and


The course organization is consistent with
what is known about cognition and the
progression of student understanding,
and it appeals to students with
varied learning styles.

analytical methods, computer simulations, and laboratory ex-
perience."E6l This paper examines how an effective balance
between these various components can be attained, using re-
search into cognition and the learning process as a guide.
Over the past three years, the author has taught a 2-credit-
hour, 14-week course (two 75-minute periods per week) on
equilibrium staged separations (see Table 1 for a summary
of its content). Enrollment varied between 14 and 22 first-
semester juniors. In the fall of 1999, the course was taught
using a lecture format almost exclusively. Material was
presented in a purely deductive manner, closely follow-
ing Wankat's textbookE~' and making little use of process
simulation.
In the fall 2000 and 2001 semesters, the course was orga-
nized as described in this paper (still using the Wankat text-

Kevin Dahm is Assistant Professor of Chemical
Engineering at Rowan University He received
his PhD in 1998 from Massachusetts Institute
of Technology Prior to joining the faculty at
Rowan University, he served as Adjunct Pro-
fessor of Chemical Engineering at North Caro-
lina A& T State University. His primary technical
expertise is in chemical kinetics and mecha-
nisms, and his recent educational scholarship
focuses on incorporating computing and simu-
lation into the curriculum.


Copyright ChE Division of ASEE 2003


Chemical Engineering Education











book). Active learning exercises were employed throughout,
with lab demonstrations, McCabe-Thiele modeling, and pro-
cess simulation playing specific, complementary roles that
are discussed in detail in this paper. Significantly, restructur-
ing the course did not affect the class time requirements sum-
marized in Table 1 and required no increase in preparation
time on the part of the instructor aside from the one-time
investment of learning to use HYSYS.

COURSE ORGANIZATION
In a series of articles in Chemical F E,,n., u i n.. Education,
HaileE8-121 discussed the operation of the human brain and the
learning process. This paper discusses how these insights on
cognition were used to guide the course's organization and


TABLE 1
Topics in Equilibrium Staged Operations and
Approximate Number of Class Periods Spent on
Each

Topic (Number of 75-minute periods devoted to it)
Introduction to Separations (1)
Vapor-Liquid Equilibrium, Bubble/Dew Points (3)
Flash Distillation, VLE Models (3)
Binary Column Distillation (6)
Multi-Component Distillation, Shortcut Methods (4)
Absorption and Stripping (3)
Liquid-Liquid Extraction (4)



TABLE 2
Levels of Understanding in the Special Hierarchy as Described by
They Might Manifest in Students Learning about Disti

Level of Understanding Examples of Student Capability
1. Making Conversation Describe in general how distillation works
Recognize a distillation column when seen
2. Identifying Elements Compare/contrast column distillation to flash distilla
Identify individual components of a column and expl
3. Recognizing Patterns Correctly predict relationships between column parai
happens to the heat duty in the reboiler when you rai
4. Solving Problems Use McCabe-Thiele model to determine the number
required, given reflux ratio, top and bottom product
feed rate and composition
5. Posing Problems Use McCabe-Thiele model to solve a variety of distill
which different sets of variables are used as "givens"
6. Making Connections Apply the McCabe-Thiele model to a column configi
heating, multiple feed, side stream product) that the s
seen before
7. Creating Extensions Recognize that the McCabe-Thiele model is not vali
application and articulate how to modify the modeling
the problem at hand


Spring 2003


the specific role McCabe-Thiele modeling and process simu-
lation should play. This paper uses column distillation as an
example, but the approach is readily applied to other physi-
cal processes and was integrated throughout the course.
Haile describedE91 a "special hierarchy"-a progression of
seven levels at which a student can understand concepts. These
levels are summarized in Table 2 along with examples of ca-
pabilities of students who understand distillation at a particu-
lar level. The table assumes McCabe-Thiele is the primary
modeling tool used.
Haile11] also described a general hierarchy of modes of
understanding that includes

Somatic 1n.,, i, *..1,.. Tactile learning. Observing
and handling something lays the groundwork for
understanding it at higher, more abstract levels.J131
Mythic U[. ,. i, li .- Oral traditions. Levels 1 and
2 of the special hierarchy fall within this realm.
Romantic [ .. .,1. 1lii, Characterized by abstrac-
tions such as writing and graphs. Level 3 of the
special hierarchy is an example.
IP il, *., /.. n.j l1. I.I./.0-1. Logical reasoning.
Levels 4 through 7 of the special hierarchy require
a philosophic understanding.

The progression from Somatic to Philosophic understand-
ing, in this case, suggests a course structure in which stu-
dents are first exposed to a real distillation column, then they
are exposed to an abstract model of a
column (such as a HYSYS model) that
is already complete, and finally they
learn to derive their own abstract model,
Haile1 and How namely the McCabe-Thiele model. The
llation special hierarchy is also a useful guide.
In Chapter 5 of Wankat's book, for ex-
ample, the McCabe-Thiele model is de-
rived and then used as a framework for
illustrating such patterns as the trade-off
tion between reflux ratio and the number of
a their function stages. The special hierarchy, however,
meters, e.g., what suggests an alternative organization in
se the reflux ratio? which students are exposed to such con-
of equilibrium stages cepts and patterns first (levels 1 through
compositions, and 3). This was accomplished by using
HYSYS to generate simulated experi-
llation problems in mental data supporting an inductive pre-
sentation of the patterns. Derivation of
duration (open steam a model came later in the context of solv-
tudent has never ing problems (levels 4 and 5).

The following sections give a step-by-
d for a given step discussion of strategy for advanc-
g technique to solve
ing the students through the levels of un-
derstanding and the tools used to facili-
133










tate each transition.


* Introduction to Column
Distillation

HaileEl8 stated that because lc.imni creates new structures
in the brain by modifying existing structures, learning can
only begin from things the student already knows." Flash, or
single-stage, distillation is the logical lead-in for column dis-
tillation. The limitations of flash distillation were demon-
strated by an example problem in which it took five flash
stages to produce a desired product of >98% pure A from a
feed of 50% A and 50% B. (This is similar to the presentation
in Chapter 4 of Wankat's text.) Students began to calculate
flow rates and compositions for all streams, given equilib-
rium data, but they quickly recognized that, practically speak-
ing, the process makes no sense. The "saleable" product
stream had a tiny flow rate and there was a clear need to
somehow recycle the intermediate fractions.
The class then moved to the Unit Operations Laboratory,
where the ten-stage distillation column had been prepared
and was operating at steady state. The instructor explained
the counter-current functioning of the column and discussed
the purposes of the various components of the column (con-
denser, reboiler, etc.). Next, the instructor posed the ques-
tion, "How is this like flash distillation and how is it differ-
ent?" This exercise followed the active learning strategy ad-
vocated by Felder, et al. [14] The class broke into groups of two
to three students each, where they brainstormed lists of simi-
larities and differences, and then the instructor led the full
class in a discussion.
These activities were viewed as a vehicle to bring the
students to Level 2 of the special hierarchy (Table 2). The
next step, as outlined above, was to expose the students
to an abstract model of the process and to help them rec-
ognize patterns.


* Use of HYSYSfor
Inductive Presentation of Concepts

Induction consists of starting with observation and infer-
ring the governing physical principles, as opposed to deduc-
tion, which consists of deriving the specifics of the case at
hand from the general principles. Educators have begun to
recognize that induction is a more natural learning mode,E15,161
but most traditional textbooks are written deductively. The
chemical engineering department at Rowan University has
previously implemented experiments to promote inductive
learning of heat and mass transfer.E171 Here, the students gained
a qualitative understanding of the physical process of distil-
lation inductively, using the simulator as a rapid way to gen-
erate simulated "experimental data."


After seeing the real column, students moved to the com-
puter lab and loaded a HYSYS model of a distillation col-
umn, which had been prepared and converged ahead of time
by the instructor. Students then went through a short (about
five minutes) tutorial on the software, learning how to access
significant column parameters (Qc, Qr, reflux ratio, product
compositions, temperature profile, internal liquid and vapor
flow rates) and how to specify them. The class discussed why
each of these parameters is of interest to the engineer-for
example, the reboiler heat duty is significant because energy
is expensive.
Next, the students were asked to collect simulated data in
order to quantify certain patterns, such as
The 0 .. t of reflux ratio on product purity
The. i.. i offered ,.i... location on product purity
The i. .. t of reflux ratio on condenser and ,. l...,/.
heat duty
The. ,.:. t of number of t* ,.. on product purity
In response, the students took the column through a series of
configurations and plotted graphs of the relevant data. After
collecting the information, students broke into small groups
to brainstorm physical explanations for the trends in prepara-
tion for full-class discussion.
During this stage of the process, students also observed
that liquid and vapor flow rates throughout the column were
nearly uniform. The physical reason for this, involving the
energy balance on each individual stage, was another topic
for discussion. Students were thus exposed to the physical
justification for the constant molal overflow approxima-
tion before they knew of its significance in simplifying
by-hand calculations.
HYSYS was specifically chosen for this process as part of
a department-wide effort to introduce students to process
simulation before the senior design sequence. Bums and
Sung,E181 however, have created McCabe-Thiele models on
spreadsheets and used them for comparable classroom dem-
onstrations. The McCabe software packageE19,201 developed at
the University of Michigan is also ideally suited for induc-
tive exploration of cause/effect relationships within a column.
The activities described in this section are viewed as a ve-
hicle to instill a romantic understanding (Level 3 in the
special hierarchy) of distillation in the students. The tran-
sition to a philosophic understanding (Level 4) was
achieved by challenging students to devise their own
model of the process.


* Hand Calculations

After receiving this thorough introduction to the physical
process, students were able to derive the model equations
with relatively little guidance from the instructor beyond the
Chemical Engineering Education










simple posing of questions. The sequence of questions is given
here; for each, the students spent time working in teams be-
fore the full class discussed the results.
1. The instructor drew a control volume around the
entire column and asked the students to list the
process variables and brainstorm which of them
would ,'. \i be ...i. and which would likely be
unknown.
2. The instructor then asked the students to write balance


equations l,i;,.. these variables to
each other The. ,, ,,, in,. discussion led to
a determination of the number of./. .,. ,
of freedom in a column and the most
likely ways rffC,, (1ll,1 them.
3. Next, the class wrote lists of variables
and constraints (mass balance, energy
balance, and equilibrium) for an
individual ,i,. .. and determined that no
"new" .J.. .* of freedom are introduced
when one ,i,..*. is added to the column.
At this point, the instructor pointed out that
HYSYS models a column by solving these equa-
tions simultaneously with the constraint that all
stages are at equilibrium. Thus, the function of
the "black box" is elucidated.
Next, students were given an example prob-
lem involving a ten-stage distillation column and
were able to demonstrate that the number of
variables and constraints were equal-thus it
was possible to attain a complete solution of all
column parameters of interest. They also rec-
ognized the complexity of solving this many
simultaneous equations "by hand." The strategy
of solving a system of equations that includes


to appreciate the capabilities and limitations of the McCabe-
Thiele model (in whatever form) and less likely to regard it
as an arbitrary ritual.

HIGHER LEVELS OF UNDERSTANDING
The activities outlined in the previous sections required, in
total, approximately two weeks of class time. Progression
through the higher levels (Levels 5 through 7) of the special
hierarchy requires practice in problem solving through rep-


... because
"learning
creates new
structures in
the brain by
modifying
existing
structures,
[and] learning
can only begin
from things
the student
already knows,"
[flash], or
single-stage,
distillation is
the logical
lead-in for
column
distillation.


mass balances and equilibrium constraints by plotting both
on the same y-x diagram was familiar to the students from
the module on flash distillation. The instructor reminded the
class of their observation that liquid and vapor flow rates
throughout the column were essentially uniform and pointed
out how the assumption of constant molal overflow led to
mass balances in the form of straight operating lines. Stu-
dents then learned the graphical technique of stepping off
stages. This completed a deductive derivation of the McCabe-
Thiele method, which was primarily carried out actively rather
than in a lecture format.
While the McCabe-Thiele method was presented as a "pen-
cil and paper" technique, the spreadsheet modelsE181 or
McCabe19 20] software package mentioned above could also
be introduced at this stage. The crucial point is that the stu-
dents have received a thorough exposure to the physical pro-
cess, intended to provide the philosophic understanding re-
quired for true model building. They are therefore more likely

Spring 2003


petition and examination of variations.[101 In the
fall of 2000 this was done exclusively using
the McCabe-Thiele model for both in-class
examples and homework problems, but in
2001 some homework problems were also
completed on HYSYS so that students would
have the experience of constructing models
from scratch on the simulator. The final as-
signment in the 2001 module on distillation
was one in which students designed the same
two-column system both by hand and with
HYSYS, comparing the results. This was in-
tended to reinforce the students' understand-
ing of the assumptions and methodology be-
hind both modeling approaches and the limi-
tations of each, consistent with the highest
levels of Haile's special hierarchy of student
understanding.

LEARNING STYLES
The course structure presented here used
both process simulation and McCabe-Thiele
modeling in a sequence that is logical accord-
ing to the learning progression described by
Haile. It was also consistent with the variety
of learning stylesE211 represented in any class


Visual vs. Verbal L. in,. The students spent most of
their class time discussing the system, either in small groups
or with the full class. Throughout the process, however, vi-
sual learners were also stimulated. Introduction to distilla-
tion was carried out in the lab with a real, working col-
umn. Students transcribed the simulated data from HYSYS
into graphical form and used the graphs as the basis for
the discussion.
Active vs. Reflective L. ,111 .* Small-group, active learn-
ing exercises were a feature of the entire course. The full-
class discussions allowed the instructor to insure that the work
from these activities was accurate and that no salient points
were missed. But they were also intended to benefit the re-
flective learners in the class.
Sensory vs. Intuitive L ,,, .: Students were quickly
immersed in studying and explaining physical phenomena, a
Continued on page 141.
135












classroom


PERSONALIZED, INTERACTIVE,


TAKE-HOME EXAMINATIONS

For Students Studying Experimental Design



WILLIAM A. JACOBY
University of Missouri Columbia, MO 65211


In this day and age, many chemical engineers seek jobs
traditionally filled by engineers from other disciplines,
and the chemical engineering curriculum, particularly
electives, can help enhance their prospects in that respect.J1
One crosscutting skill set that facilitates this trend is ex-
pertise in statistical methods.E2J Employers particularly
value knowledge of the techniques of experimental de-
sign and quality control.E341
The University of Missouri-Columbia's Department of
Chemical Engineering offers a three-semester-hour course
called "Experimental Design and Statistical Quality Control
for Chemical Engineers." It is the most popular undergradu-
ate elective, perhaps because it can be taken in lieu of a re-
quired course in probability and statistics offered in the Col-
lege of Arts and Sciences. Graduate students, who must com-
plete an additional semester project, also take the course.
The examinations described in this article are personalized
and interactive in the sense that the students are allotted a
prescribed number of experiments. Using a sequential ap-
proach in which some fraction of the experimental budget is
expended in the first submission, each student submits a care-
fully formatted table of experimental conditions (factor-
levels for each of the variables under consideration). The in-
structor uses a computer model that includes a random error
term as a virtual laboratory to efficiently generate a unique
data set for each submission. After interpreting the data from

William A. (Bill) Jacoby received his PhD
from the University of Colorado in 1993. He
workedas a research engineeratthe National
Renewable Energy Laboratory until 1997
when he joined the faculty at the University of
Missouri-Columbia. His research interests in-
clude photocatalysis, thermal catalysis, and
biotechnology.


the first set of experiments, the student submits additional
experiments and receives additional sets of unique data until
his or her experimental budget is expended. The appropriate
set of experimental designs must be combined with accurate
calculations and insightful analysis to arrive at "the truth,"

TABLE 1
List of Topics in "Experimental Design and Statistical
Quality Control for Chemical Engineers"

1. Normal distribution and the central limit theorem
2. Statistical quality control: creating, maintaining, and interpreting
SQC charts
3. Statistical quality control: rational subgroups and interpretation
4. Significance testing
5. Z distribution
6. t distribution
7. Statistical dependence
8. Random sampling
9. Randomization
10. Blocking
11. Confidence intervals
12. Inferences about variances
13. Error propagation
14. Comparing more than two treatments
15. Empirical and theoretical models
16. Analysis of variance
17. Multiple comparisons
18. Randomized blocks with replication
19. Designs with more than one blocking variable
20. Balanced incomplete blocked designs
21. Full factorial designs
22. Interpreting the results of full factorial experiments
23. Determining significance of effects in factorial experiments
24. Applications of statistical quality control
25. Partial factorial designs
26. Design resolution
27. Confounding patterns
28. Sequential design of experiments; additional runs
29. Analysis of Residuals
30. Parsimony in empirical models
31. Linear regression
32. Nonlinear regression


Copyright ChE Division ofASEE 2003


Chemical Engineering Education












TABLE 2
Problem Statement
You have accepted a job at Cavitron, a small start-up company. Cavitron is attempting to
commercialize a turn-key, skid-mounted "pump-and-treat" system for use in oxidizing the
organic and chlorinated organic compounds in aqueous mixtures.
Hydrodynamically induced cavitation is the operating principle for the treatment device,
which is referred to as a "jet reactor." When polluted water is pumped at high pressure and
high velocity through an appropriately designed nozzle and around an appropriately de-
signed obstruction, microscopic bubbles form and implode in the fluid. Local temperatures
reaching 800C and local pressures in excess of 5,000 psi accompany the formation and
implosion of the bubbles. Organic vapors predominate (relative to water vapor) in the bubbles.
In the presence of dissolved oxygen and other oxidative species, as well as a miscible fluid
catalyst (with appropriate vapor pressure), each bubble is a microreactor in which some
fraction of the organic vapor is oxidized.
Your first project for Cavitron is to set up and operate a skid-mounted system for treating
the leachate from a hazardous waste landfill. You will draw polluted water from the con-
tainment pond, treat it, and pump it back into the pond. Since each waste stream is different,
the operating conditions for this application must be optimized. The response to be opti-
mized is single-pass conversion (treatment efficiency). Table 3 lists seven standard process
variables routinely evaluated at each installation. Factor level settings that experience has
shown are in the proper experimental spaces are also provided. Your first task involves
determining the effect of these seven "standard" process variables on treatment efficiency.
The Research and Development Department would also like you to evaluate four experi-
mental modifications to the jet reactor. Field data is essential to verify laboratory results. At
some point during your experimental campaign, you are to install the experimental modifi-
cations and proceed with testing. Table 3 also lists these experimental modifications (vari-
ables) and their factor level values. Your second task involves evaluating the effect of these
experimental variables of treatment efficiency.
Your tasks are tabulated more specifically below.
Task la: Determine the sign and magnitude of the significant main effects and interactions
of the standard process variables on the treatment efficiency of the unit.
Task ib: Formulate an empirical model and evaluate its validity.
Task lc: Recommend operating settings for these seven variables.
Task 2a: Determine the sign and magnitude of the significant main effects and interactions
of the experimental modifications on the treatment efficiency of the unit.
Task 2b: Appropriately modify your empirical model from Task lb and evaluate its valid-
ity.
Task 2c: Make recommendations about whether these four modifications should be adopted
in future production units.
Time and budget constraints will allow you to perform 24 experiments. These may be
submitted in whatever increments you choose over the next five days. Submit your sets of
experimental conditions electronically and you will receive your data via return e-mail.


TABLE 3
Standard Variables and Experimental
Variables/Modifications and Factor Levels
Standard Variables and Factor Levels
Symbol Description Level + Level
P Pressure in the nozzle 2000 psi 3000 psi
L Length of the pretreatment capillary 10 m 20 m
T Temperature of the pretreatment capillary 25 C 70C
C Concentration of the catalyst 0.05 M 0.10 M
A Angle of the obstruction 0 5
D Diameter of the obstruction 5 cm 8 cm
X Distance between nozzle and obstruction 0.5 mm 0.75 mm
Experimental Variables and Factor Levels
Symbol Description Level + Level
S Supersaturated oxygen Off On
K Catalyst type Standard Experimental
0 Ozone generator Off On
N Nozzle design Standard Experimental

Spring 2003


an accurate estimate of the parameters of the model used
to generate the data.


COURSE STRUCTURE

The latest rendition of the course (spring semester,
2002) met for 50-minute sessions on Mondays, Wednes-
days, and Fridays for fifteen weeks. Table 1 lists the top-
ics discussed. They were selected to provide a practical
statistical toolbox to chemical engineers in research, pro-
cess engineering, and manufacturing.
The availability of computational tools, principally a
personal computer and associated software, has allowed
an increase in the complexity of calculations presented
in chemical engineering classes, as well as in the home-
work assignments. In this class, most lectures (as well as
all examples and homework solutions) were performed
using the Excel" spreadsheet program. These spreadsheets
were made available, at the appropriate time, to the stu-
dents via e-mail. This allowed the use of a relatively old
but well-written and classic text that does not explicitly
employ computer techniques or software.E[5 Fortunately,
on Monday and Wednesdays the course met in a com-
puter lab where each student had access to a computer. The
use of a computer lab during class, however, is not required
in the administration of this type of examination.


DESCRIPTION OF THE EXAMINATION
It is difficult to give a comprehensive examination in a
computationally intensive course when there are constric-
tions of class duration and/or access to computers in the
classroom. Most chemical engineering examinations are
completed during a single class period without the aid of
computers. The availability of a computer lab does not
circumvent the time constraint. The challenge for the in-
structor under these circumstances is to write an exam
that promotes learning, discriminates among the students,
and is consistent with the course content and homework.
Take-home examinations are an attractive option, but
raise another problem: academic dishonesty. Although the
percentage of students who collaborate improperly on
take-home examinations is small, there is an opportunity
for a minority to gain an unfair advantage. A take-home
exam in which each student has a unique data set gener-
ated from a model including a random-error term elimi-
nates the opportunity for one student to copy another's
work. The use of several different models to generate the
students' data sets provides a further obstacle to dishon-
est collaboration, but must be accounted for during record-
keeping and grading.
Table 2 is the problem statement from a personalized,
interactive, take-home examination based on this concept.
Prior to the class in which it was presented, an electronic
137














version was e-mailed to each of the students as a worksheet
in an Excel spreadsheet. This spreadsheet also included a
worksheet containing Table 3, which includes the standard
variables and the experimental variables/modifications as well
as their factor-level settings. Also included was an abbrevi-
ated version of Table 4 (no factor levels, data, etc.), which
was formatted for submission of experiments.

An individual student has a budget of 24 experiments. For
a particular experiment, the model shown as Eq. (1) gener-
ates a data point:

P T D S 0O SxO
y = I+P-X +-XT +XD +Xs +-Xo +-2Xsxo +E (1)


where y is the response, the single-pass conversion (%), and
I is the overall average response (I = 15%). The X-variables
(X XT, XD, Xs, Xo) have a value of -1 for the experiments in
which the indexed variable is set at the minus level and +1
for the experiments in which the indexed variable is set at the
plus level. Xsxo is the factor level of the interaction between
the S variable and the 0 variable, and its value is the sign of
their product. The magnitudes of the main effects and inter-
actions used in the model to generate the data are shown in
Table 5. The student chooses the values of all 11 variables for
each experiment. The variables that are not included in the model
used to generate the data set (L, C, A, X, K, N) are inert.

Equation 1 is an empirical model used to interpret data from
factorial experiments. Theoretical models can also be ap-
pended with error terms to generate unique data sets for take-
home examinations in core subjects such as thermodynamics
and transport phenomena. More empirical curricula (e.g., ki-
netics) are even more amenable to the technique.


The student submits a total of 24 experiments via e-mail
over a period of five days. Most students submitted three sets
of eight experiments each. It took about two minutes to open
an e-mail, open the experimental design, insert the student's
input into the model to generate a data set, save the data set,
attach it to a return e-mail, and send. For example, if a class
had 20 students, they would request 60 data sets, requiring
the instructor to spend two hours generating data. The data
generation process could be easily automated. The time
required to write and grade this exam is similar to a con-
ventional exam.

Based on the individualized data sets, the student must de-
termine which of these variables has a significant effect on


TABLE 5
Summary of Results

Main Effect/ Model Estimate Recommended
Interaction Parameter from Data Error Setting

P 2.0 2.1 5% +




T -1.5 -1.8 -23%

A -
D -1.5 -1.7 -16%


S 3.0 2.9 4% +

K -
0 2.0 2.5 27% +

SxO 3.0 3.0 0% NA


En. Variables


T AI T D X= I
iL I C i PL Pir Lx I Pirr IP S K IO N


Empirical Model
Prediction (%)
111I
15.0
14,6
149
12.8
167
129
132
Enppirical Model
Prediction (%)


TABLE 4
Summary of Experimental Campaign


Standard Prcess Variables


Single Pass
Conversion (%)
118
148
13.9
15.5
13.2
16.4
12.6
13.4
Single Pass
Conversion (%)
13.4
12.6
129
16.2
164
18.7
12.2


Ep. # P


Exn. ISxK L C SU O A KxO


-1 -1 1 1 -1


T A


S K 0 N


Conversion () IPreiction (mu
Conversion (%) Prediction (%)


Residual (%)
07
-0.2
-07
0,6
0.4
-04
-04
02

Residual(%)
00
-0.4
-0,1
-06
0.0
0.3
-0,4
-93

Residual (%)


Main Eect Abbreviated
Variable (%) Confounding Pattern
P 2.2 P+ LxT + CxA + DxX
L -0.2 L +PxT +CxD +AxX
C -0.1 C +PxA + LxD+TX
T -1.0 T +PxL + AxD + LxX
A -0 1 A + PxC + TxA + PX
D -1.6 D +LxC+TlxA+PxX
X -U.2 X + CxT + LxA + PD
Average 13.9
Main Effect Abbreiated
Variable (%) Confounding Pattern
P 2.0 P + SxK + TxD + OxN
T 1.5 T + PD + SxO + KxN
D -19 D+PNT+SxN+KxO
S 28 S+PK+TsO+D DxN
K -0 3 K + PxS + TxN + DxO
0 27 0 +PxN + TiS + DxK
N 1-0 N + PxO + TxK + DxS
Average = 15.1
Variable or Main Effect
Interaction (%)
T -1.8
S 3.0
O 2.4
TxS 0 1
TxO 0 2
SxO 3.0
TxSxcO 0.0
Average = 15.5


Chemical Engineering Education


I -.. M


I


I


I


r










the single-pass conversion and which are inert. The sign and
magnitude of the significant main effects must also be deter-
mined. Further, any significant interactions among the stan-
dard variables must be identified and their signs and magni-
tudes estimated. The student must also formulate an empiri-
cal model and evaluate its validity and recommend operating
settings for these seven variables. Finally, each student must
similarly assess the effects and interactions of four additional
experimental variables of interest to the Research and Devel-
opment Department.
Performing a full factorial experiment with eleven vari-
ables would require 2,048 experiments. As the experimental
budget is about 1% of this amount, the use of highly fraction-
ated partial factorial designs is required.

SOLUTION TO THE EXAMINATION
The first step in one of many effective solution strategies is
to design and perform a 27-4 partial factorial experiment fo-
cusing on the standard process variables. This is a resolution
III "main effects" design because it estimates the main ef-
fects subject to a confounding pattern including two-way in-
teractions. Aspects and advantages of this type of design are
discussed in the course textbook.J5
The first eight experiments shown in Table 4 prescribe this
design. Pressure in the nozzle (P), length of the pretreatment
capillary (L), and concentration of catalyst (C) are taken as the
"live" variables. Their factor levels are assigned in standard
order, as they would be for a 23 full factorial experiment.
The four remaining variables in the standard process vari-
able set are temperature in the pretreatment capillary (T), angle
of obstruction (A), diameter of obstruction (D), and distance
between the nozzle and the obstruction (X). The levels of
these variables are set according to the four combinations of
interactions possible among the three live variables (i.e.,
T=PxL, A=PxC, D=LxC, and X=PxLxT). Since all of the
possible interactions among the three live variables were used
as aliases for the additional variables, the design is referred
to as fully saturated. The experimental variables/modifica-
tions are held at the minus (standard or unmodified) level for
the first set of experiments.
Eight experiments were performed-therefore, eight pa-
rameters (the average and the seven main effects) can be es-
timated from the data. Each main effect is subject to con-
founding by fifteen other interactions. An abbreviated con-
founding pattern, including only the confounding two-way
interactions, is also shown in Table 4. The data in the column
headed "Single-Pass Conversion (%)" were generated using
the model shown in Eq. (1).
Quantitative methods of determining significant effects are
discussed in the course textE41 and will not be covered here.
Examination of Table 4 reveals that the first eight experi-
ments correctly indicate that P, T, and D may be important
Spring 2003


Personalized, interactive, take-home
examinations are not subject to the
constraints of class duration and
availability of computers they
can be more complex and thorough.


variables, while the remaining standard process variables (L,
C, A, and X) may be relatively inert.
After evaluating the first set of experiments, the principle
of sequential design of experiments must be practiced in the
second design. This solution strategy involves another set of
eight experiments, shown as experiments 9 through 16 in
Table 4. In this design, the intent is to begin investigation of
the experimental variables/modifications, while confirming
and improving the estimates of the three standard variables
judged to be significant. The experimental modifications/vari-
ables supersaturated oxygen (S), catalyst type (K), and ozone
generator (0) are the live variables in a second 2 7-4 partial
factorial experimental design. The alias for the final experi-
mental variable/modification, nozzle design (N), is the three-
way interaction among the live variables (N=SxKxO). This
design is also fully saturated in that the remaining three pos-
sible interactions among the live variables are used as aliases
for the three variables judged to be significant during the first
set of experiments (P=SxK, T=SxO, D=KxO). Table 4 also
includes the data for these experiments, the abbreviated con-
founding pattern, and the parameter estimates based on the data.
The parameter estimates show that the experimental vari-
ables/modifications S and 0 may be significant, while K and
N may be inert. Further, the estimates of the standard vari-
able parameters P and D are confirmed. These estimates are
subject to entirely different confounding patterns, lending cre-
dence to the assumption that it is these main effects and not
their confounding two-way interactions that are significant.
The temperature variable, T, however, is a different matter.
While both the first and second sets of experiments resulted
in estimates of similar magnitude, the sign changed. This
suggests the presence of a significant interaction. Careful
examination of the abbreviated confounding patterns for both
the first and second sets of experiments reveals that an inter-
action between S and 0 is the most likely candidate, as both
are significant variables whose interaction has not been pre-
viously aliased to an inert variable. Therefore, the final eight
experiments in the experimental budget are expended per-
forming a 23 full factorial experiment using variable T, S,
and 0. This design has the advantage that all interactions are
explicitly estimated. As shown in Table 4, this design pro-
vides an unambiguous estimate of the T effect, confirms and
refines the estimate of the S and 0 effects, and reveals an
important two-way interaction oxygen supersaturation and
139











ozone generation (SxO).
Table 5 summarizes the information gleaned from the ex-
perimental campaign and compares it to the actual param-
eters of the model used to generate the data. Three of the
standard process variables were found to be significant, while
the other four were determined to be inert. Two of the four
experimental variables/modifications were determined to be
significant, while the other two were inert. For all eleven vari-
ables, these determinations were correct (in agreement with
the model used to generate the data). Further, the signs of all
the effect and the interaction were also correct and the mag-
nitudes were accurate between +/-30%. A column of recom-
mended settings is also included in Table 5. For the inert vari-
ables, decisions about the settings are based on what might
be expected to be easiest and cheapest.
Two empirical models can be developed from the data. The
first
2.1 -1.8 -1.7
y = 13.9 + --Xp + -- XT + --1XD (2)

predicts the single-pass efficiency of the jet reactor in its stan-
dard configuration (unmodified, all experimental variables/
modifications at the minus level) as a function of the three
significant standard variables. This model was used to gener-
ate the predicted values of the single-pass efficiency for the
first eight experiments in Table 4.
The second model

2.1 -1.8T -1.7 2.9 2.5 3
y=15.3+-Xp+-X --XT +Z XD+--Xs +-Xo +-Xso
2 2 2 2 2 2

(3)

predicts the performance of the jet reactor in its experimental
configuration. It has a higher average and includes the S and
O effects as well as their interaction. This model was used to
generate the predicted values of the single-pass efficiency
for the final sixteen experiments in Table 4.
Variables for both models are defined as in Eq. (1). Equa-
tion 3 is the experimental estimation of "the truth," as de-
scribed by Eq. (1). Analysis of the residuals, tabulated in Table
4, was undertaken according to standard procedures and con-
firms the validity of the models. 51


STUDENT FEEDBACK
An interactive learning environment was established and
persisted throughout the week of the exam. This excitement
was felt by both the students and the instructor. Table 6 shows
the results of a feedback survey administered to the class.
There were 19 respondents. The results document that a
personalized, interactive, take-home examination is not
only a good learning tool, but is also popular with the
students. Three estimates of the central tendency are in-
cluded to aid in interpretation.
140


TABLE 6
Summary of Anonymous Feedback Survey
(1- Agree; 2 -Agree; 3- Neutral; 4- Disagree; 5- Disagree)
Avg. Median Mode
I understand the partial factorial experimental designs
better as a result of this exam. 1.7 2 2
I understand the sequential nature of experimental
design better as a result of this exam. 1.6 2 2
I like the individualized data concept. 2.0 2 2
I liked this exam. 2.2 2 2
I like take-home exams. 1.6 2 1
I lemed a lot while working on this exam. 2.1 2 2
This exam was a superior learning experience relative
to the other exams for this class. 2.4 2 2
This exam was a superior learning experience relative
to exams in other engineering courses. 2.6 3 3
I spent more time on this exam relative to the other
exams for this class. 2.1 2 2
I spent more time on this exam relative to exams in
other engineering courses. 2.6 3 2
This exam really sucked. 4.0 4 4


CONCLUSIONS
Personalized, interactive, take-home examinations are not
subject to the constraints of class duration and availability of
computers. Therefore, they can be more complex and thor-
ough. Because a unique data set is generated for each stu-
dent, the opportunities for dishonest collaborations are re-
duced. The use of several models to generate the students'
data sets is a further barrier to cheating. Taking advantage of
ubiquitous e-mail connectivity and the speed and storage ca-
pacity of modem personal computers, data generation and
dispersal is expeditious.
The interactive aspects of the examination and the pre-
scribed experimental budget allow a hands-on exploration of
the concept of sequential design of experiments. Student feed-
back regarding the exam was favorable.
This type of examination can be adapted for use in other
chemical engineering courses. In the future, elimination of
the instructor interface during data generation will stream-
line the process.

REFERENCES
1. Cussler, E.L., "Do Changes in the Chemical Industry Imply Changes
in Curriculum?" Chem. Eng. Ed., 33(1) (1999)
2. Fahidy, T.Z., "An Undergraduate Course in Applied Probability and
Statistics," Chem. Eng. Ed., 36(2) (2002)
3. Moen, R.D., T.W. Nolan, and L.P. Provost, Quality Improvement
Through Planned Experimentation, 2nd ed.,McGraw Hill, New York,
NY (1999)
4. Montgomery, D.C., Introduction to Statistical Quality Control, 2nd
ed., John Wiley & Sons, New York, NY (1991)
5. Box, G.E.P, W.G. Hunter, and J.S. Hunter, \: ... Experiment-
ers, John Wiley & Sons, New York NY (1973) 1
Chemical Engineering Education












Process Simulation
Continued from page 135.
process that should appeal to an intuitive learner. They did
this, however, in a practical context that would also appeal to
a sensory learner; they first saw a real column and did an ex-
ample validating its importance, and then they used HYSYS,
which is recognizable as a tool
used by "real engineers."


Sequential vs. Global Learn-
ni,.. The structure was
methodical and well-suited for
sequential learners, but was
also interspersed with "big pic-
ture" insights that were meant
to benefit all students, particu-
larly global learners. The first
thing the class learned about
column distillation was why it
was useful. The class discussed
the significance of each pro-
cess parameter before attempt-


ing to calculate it or to even relate it to anything else.

STUDENT RESPONSE
The course structure described in this paper was used in
the fall 2000 and fall 2001 semesters at Rowan University.
Table 3 summarizes the results of the course and teacher evalu-
ations of it. Feedback was very positive, both toward the use
of HYSYS for inductive teaching on concepts and toward
the overall course. Specific student comments included,
"Learning HYSYS and seeing what actually happens in a
distillation column, etc., was very helpful," and "The in-
class HYSYS days were helpful for seeing how the whole
process works."

SUMMARY
In assessing how modern process simulators should affect
teaching of separations, chemical engineering educators have
suggested a blend of simulation with traditional graphical
modeling approaches. This paper describes an effective strat-
egy for using these two modeling approaches that was suc-
cessfully implemented in the fall 2000 and fall 2001 semes-
ters at Rowan University. Students' first introduction to dis-
tillation was exposure to a real column and discussion of the
practical significance of distillation. Process simulation was
used as a tool for inductive presentation of concepts to pro-
mote a thorough understanding of the physical process. This
was followed by a deductive derivation of the McCabe-Thiele
model. The course organization is consistent with what is
known about cognition and the progression of student under-
standing, and it appeals to students with varied learning styles.
It was an effective presentation, as evidenced by student feed-
back. This paper focused on column distillation as an example,
Spring 2003


but the approach is readily extended to other physical processes.

REFERENCES
1. Wankat, P.D., Equilibrium Staged Separations, Prentice Hall,
Englewood Cliffs, NJ (1988)
2. Seader, J.D., and E.J. Henley, Separation Process Principles, John
Wiley & Sons, New York, NY (1998)


TABLE 3
Summary of Course and Teac
Responses were on a scale from 1-5

Question

Were the additional activities (HYSYS) hel
for understanding the subject matter?
Considering everything, how would you rat
Considering everything, how would you rat


3. Dahm, K.D., R.P. Hesketh, and
M.S. Savelski, "Is Process Simu-
lation Used Effectively in
Chemical Engineering
Courses?" Chem. Eng. Ed.,
36(3), 192 (2002)
her Evaluations 4. Wankat, PC., "Integrating the
, with 5 being best. Use of Commercial Simulators
into Lecture Courses," J. Eng.
Fall Fall Ed., 91(1) (2002)
2000 2001 5. Mackenzie, G.H., W.B. Earl,
pful 4.63 4.88 R.M. Allen, and I.A. Gilmour,
"Amoco Computer Simulation
in Chemical Engineering Educa-
e this teacher 5.00 4.71 tion," J. Eng. Ed., 90(3) (2001)
e this course 5.00 4.65 6. Wankat, PC., "Teaching Sepa-
rations: Why, What, When, and
How?" Chem. Eng. Ed., 35(3)
(2001)
Wankat, PC., R.P Hesketh, K.H. Schulz, and C.S. Slater, "Separa-
tions: What to Teach Undergraduates," Chem. Eng. Ed., 28(1) (1994)
Haile, J.M., "Toward Technical Understanding: Part 1. Brain Struc-
ture and Function," Chem. Eng. Ed., 31(3) (1997)
Haile, J.M., "Toward Technical Understanding: Part 2. Elementary
Levels," Chem. Eng. Ed., 31(4) (1997)
Haile, J.M., "Toward Technical Understanding: Part 3. Advanced Lev-
els," Chem. Eng. Ed., 32(1) (1998)
Haile, J.M., "Toward Technical Understanding: Part 4. General Hier-
archy Based on the Evolution of Cognition," Chem. Eng. Ed., 34(1)
(2000)
Haile, J.M., "Toward Technical Understanding: Part 5. General Hier-
archy Applied to Engineering Education," ( ...*-- i .- 34(2) (2000)
Godiwalla, S., "What is Inside that Black Box and How Does It Work?"
Chem. Eng. Ed., 32 (1998)
Felder, R.M., D.R. Woods, J.E. Stice, and A. Rugarcia, "The Future of
Engineering Education: Part 2. Teaching Methods that Work," Chem.
Eng. Ed., 34(1) (2000)
Bransford, J.D., A.L. Brown, and R.R. Cocking, eds., How People
Learn, National Academy Press, Washington DC (2000)
Felder, R.M., and L.K. Silverman, "Learning and Teaching Styles in
Engineering Education," Eng. Ed., 78(7) (1988)
Farrell, S., and R.P., Hesketh, "An Inductive Approach to Teaching
Heat and Mass Transfer," Proc. ASEE Ann. Conf. and Exposition, St.
Louis, MO, June (2000)
Burns, M.A., and J.C. Sung, "Design of Separation Units Using Spread-
sheets," Chem. Eng. Ed., 30(1) (1998)
Fogler, H.S., S.M. Montgomery, and R.P. Zipp, "Interactive Computer
Modules for Chemical Engineering Instruction," Comp. Appl. Eng.
Ed., 1(1) (1992)
Montgomery, S., and H.S. Fogler, "Selecting Computer-Aided Instruc-
tional Software," J. Eng. Ed., 85(1) (1996)
Felder, R.M., "Reaching the Second Tier: Learning and Teaching Styles
in College Science Education," J. College Sci. Teach., 23(5) (1993)
Wankat, PC., and F.S. Oreovicz, Teaching Engineering, McGraw Hill,
New York, NY (1993)
Felder, R.M., "Meet Your Students: 1. Stan and Nathan," Chem. Eng.
Ed., 23(2) (1989)
Felder, R.M., "Meet Your Students: 2. Susan and Glenda," ( .... i .
Ed., 24(1) (1990) J











EI n= laboratory


OPTIMUM COOKING OF

FRENCH FRY-SHAPED POTATOES

A Classroom Study of Heat and Mass Transfer



JIMMY L. SMART
University of Kentucky Paducah, KY 42002


affles. Ridges. Piil Iul Tater Skins. What do
these trade names share? They are offered to the
consumer as the perfect potato chip. And how
might this so-called perfect potato chip be defined? Probably
in terms of quality of taste and texture...balanced against a
reasonable cost.
Along with pizza, students are seriously interested in po-
tato chips-for the obvious reasons. At the University of
Kentucky, we are always looking for new ways to stimulate
learning in the classroom. Although chemical engineers do
not traditionally study food engineering, we believe the ex-
ploration of various methods to cook the common potato helps
motivate students to learn and apply the engineering prin-
ciples of heat and mass transfer.
The preparation and manufacture of potato chips is a com-
plex subject, spawning complete industries and intense re-
search. Even doctoral dissertations have been devoted to the
preparation of potato chips. Much of the recent research ef-
fort has been directed toward evaluation of cooking oils and
seasonings, nutritional content, and product preservation.
Other work has been done to optimize storage life with
various protective barriers/packing materials and appli-
cation of preservatives.
The following laboratory exercise deals with the optimiza-
tion of french fry-shaped potatoes (rather than chip geom-
etry) and is offered as an initial exploratory exercise for stu-
dents. The complete exercise may be too lengthy for some
laboratory allotments and portions may be modified or elimi-
nated where appropriate. Faculty and students are invited to
consult other excellent resources for further discussion of the
technical aspects of food engineering.[1-41 Two other related
articles recently featured in Chemical F,,. *.,,. i, ii Educa-
tion include a study of heat and mass transfer with micro-
wave dryingE51 and the use of a mathematical model for


cooking potatoes.[6] Finally, a recent popular article in The
New Yorker[71 traced the origins of the development and
optimization of the french fry in the U.S. by Ray Kroc of
McDonald's fame.

MOTIVATION
Students receive and learn information in accordance with
three modalities: visual, auditory, and kinesthetic. Generally,
academic environments appeal to these modalities by com-
bining classroom theory and lab experimentation. In Kolb's
four-stage learning model,E81 he calls this process reflective
observation, abstract conceptualization, active experimenta-
tion, and finally, concrete experience fcclii-:. We believe
most students (reported to be as high as 60%[9]) learn better
when "hands-on" applications (active experimentation) are
presented concurrently with classroom theory. Traditionally,
students often wait between one to two years to apply a pre-
viously learned theory to an actual application in an experi-
mental laboratory setting.
At the University of Kentucky, we offer an undergraduate
course in the chemical/materials engineering curriculum
called "Heat and Mass Transfer." Recently, our department
has made concerted efforts to bring more experimental ap-
plications back into the classroom. One such experiment in-
corporated into the classroom environment is the study of

Jimmy Smart is Assistant Professor of Chemi-
cal and Materials Engineering at the University
of Kentucky He received his BS from TexasA&M
University and his MS and PhD from The Uni-
versityof Texas, Austin, all in chemical engineer-
ing. He has over twenty years of industrial ex-
perience with companies such as IBM and
Ashland Chemical. His research areas include
applications of membranes to purify water sup-
l plies and treatment of hazardous waste.


@ Copyright ChE Division of ASEE 2003
Chemical Engineering Education










heat and mass transfer and how it applies to a simple thing
such as cooking a potato. Please note: these types of com-
bined classroom/short experimental components are not in-
tended to replace an existing separate laboratory experimen-
tal course. Instead, they are designed to complement and en-
hance traditional classroom theory.

SCOPE AND OBJECTIVES
The purpose of this exercise is not to conduct an in-depth
investigation into the best methods of producing potato chips,
but rather to use fundamental principles of heat and mass
transfer to demonstrate what effects these principles have upon
possible food quality. Traditionally, the food industry has
taken a "cook-and-look" approach to development of new
foods. There is some evidence, however, that it is starting to
take a more scientific approach because such an approach
can reproduce successes and lead to more interesting dif-
ferences in food textures.E10] The students in this exercise
take advantage of the opportunity to explore some of the
cooking variables involved in the preparation of products
in the food industry.
Since the science and art associated with preparing the "per-
fect potato chip" is so complex, conditions in this exercise
have been simplified to examine only fundamental compo-
nents of the food preparation process. Potato chips are usu-
ally fried or prepared with various cooking oils, although there
has been some interest lately in baking chips to reduce the fat
levels. Using cooking oils, antioxidants, or seasonings (in-
cluding salt) will not be considered in this exercise. Instead,
various heat transfer equipment will be used to judge their
effect on the drying (mass transfer) and cooking (heat trans-
fer) of potato slices. Cooking equipment will include the
conventional oven, a convection oven, a microwave oven,
and a pressure cooker.
One might wonder-what is cooking and what is happen-
ing during the actual cooking process? The general cooking
process is largely a matter of how heat is applied to a food
product. In terms of unit operations, cooking is a combina-
tion of heat transfer and drying operations coupled with
chemical reaction. Actually, cooking involves modifications
of molecular structures and formation of new compounds,
the killing of dangerous organisms, modification of textures,
and the drying/browning of food materials. A typical potato
is made up of water, starch, reducing sugars, pectin, and com-
plex organic molecules.E11 During the cooking process, mois-
ture levels and flavor components change. Also, bond
strengths within the vegetable pectin are altered, which af-
fects the mechanical properties of the potato.J121
A word about the potato chip geometry: In our initial cook-
ing experiments, the edges of the potato chips curled, which
interfered with mechanical testing. Teflon holders were con-
structed to hold the chips in an upright position to promote
heat transfer and to reduce edge curling. In the end, this chip
Spring 2003


geometry was not the most desirable shape for heat-transfer
modeling. Finally, a rectilinear geometry (french fry shape)
was selected for ease of mechanical testing and approxima-
tion to cylindrical geometry for heat-transfer calculations.
Using a conventional oven to cook a potato stick, the stu-
dent is prompted to define an "optimum potato" in terms of
quantitative factors of mechanical hardness/deflection and
qualitative factors of color, taste, feel, and smell. During the
cooking process, there are two simultaneous phenomena oc-
curring in the small potato stick. The inside of the potato is
"cooked" during the process of unsteady-state heat transfer
as heat progressively moves from the outside surface to the
center of the potato. In a reverse gradient, mass is transferred
as volatiles (water and organic molecules) move from the
center of the potato to the outside surface during the drying
process. Once the potato optimum is defined with a conven-
tional oven, the student is challenged to reproduce the potato
quality in other cooking equipment (convection oven, mi-
crowave, and pressure cooker).

EQUIPMENT AND MATERIALS
Heat transfer (c'k, kiin.- equipment includes a conventional
oven, a convection oven, a microwave oven, and a pressure
cooker. A gravimetric scale, capable of 0.01 g, is used to
monitor loss of volatile materials during the cooking pro-
cess. Surface firmess of cooked potatoes is monitored with a
durometer.* A compression force gage** is used to test potato
material strength by monitoring deflection. Dimensions of
each potato test specimen are measured with a micrometer,
and a thermocouple is used to monitor oven temperature. A
french fry potato extruder*** is used to provide consistent-
size test specimens.

PREPARATORY STEPS
Before the actual cooking procedure is started, the avail-
able temperature ranges of the four ovens should be verified.
To execute the heat transfer models, it is desirable to have
the same temperature setting in each of the ovens. The con-
ventional oven poses no problem because it can be varied
from 380C to 2600C (1000F to 5000F), but the temperature
settings for the convection and pressure cookers will usually
be pre-set by the equipment manufacturer. The temperature
of the pressure cooker will be fixed by the pressure rating of
the vessel. For example, our 6-quart pressure cooker is de-
signed for 10 psig, or about 1160C (2400F).
All experimental equipment and plans should be carefully
assembled before the potatoes are sliced. Raw potatoes readily
turn brown upon exposure to air and this will affect the as-
sessment of product color during the cooking test.

McMaster-Carr Supply, Cleveland, OH; Shore 00 range, model
1388T232m /#450)
** McMaster-Carr Supply, Cleveland, OH; model 2115T11, $65
*** HALCO french fry cutter model K375, $120











GENERAL PROCEDURE

O Select large, white baking potatoes (Russett variety)
from one bag (same lot). Peel the potatoes and use a
french-fry cutter to prepare consistently sized test
specimens. Cut potato strips into 10.2 cm lengths (4.0
in) and pierce with short lengths of bamboo skewers
so that the samples resemble a "carpenter's saw-
horse" (see Figure 1). Record the samples' weight,
including skewers, and place them in a conventional
oven set at a moderately high temperature (2040C) to
drive-off moisture and other volatile materials.
Prepare a drying curve by plotting free moisture loss
versus time.[131 This will entail removing the potato
samples from the oven approximately every five
minutes and recording weight changes. Weigh the
samples and promptly replace them in the oven, as
they will begin to cool and absorb humidity from the
ambient air. See Figure 2 and 3 for typical examples
of drying curves by students. Note the insertion of
solid lines in Figure 3 to approximate the heat-up,
constant-rate, and falling-rate regimes of drying.
Much data scatter was the result of the potatoes
removal from and reinsertion into the oven. If it is
available, a laboratory drying oven with integral
scale would allow more precise construction of
classical drying curves.
Divide the drying curve into six segments: three
points in the constant-drying-rate period and three
points in the falling-rate period. Prepare seven new
potato samples with skewers and place them in the
conventional oven. Remove individual samples from
the oven at those times corresponding to the points
previously selected on the drying curve. Let the
samples come to equilibrium in ambient air, and then
conduct deflection tests, hardness tests, and panel
evaluations tests on the samples as described below.
0 Repeat steps 1 and 2 for the conventional oven at a
lower oven temperature setting (1210C).


Figure 1. French-fry geometry on bamboo skewers.


2.50
0 o 204'C (400'F)
2.25 1210C (2500F)

2.00

S1.75

0 0 A
3 o A
S1.25
a o
1.00











time, mi
S0.75 o













conditions in a conventional oven.
o 0
0.50 o

0.25 -

S0.002
0 20 40 60 80 100 120 140 160 180 200
time, min
Figure 2. Free moisture versus time for constant drying
conditions in a conventional oven.

0.004





0.003




S0.002






0.001 ,


0.0 0.5 1.0 1.5 2.0 2.5
free moisture, (g moisture/g dry solid)
Figure 3. Drying curve for conventional oven.
Chemical Engineering Education


2040C (4000F)
1210C (2500F)










G Follow the same general procedure for sample testing
in the convection oven, the microwave oven, and the
pressure cooker.

OPERATION OF HEATING EQUIPMENT
Conventional Oven Locate a thermocouple near the po-
tato samples to accurately measure the temperature, as
deadbands on oven thermostats are known to vary widely.
Convection Oven Forced circulation is used to improve
heat transfer and reduce cooking time. In order to make heat-
transfer calculations, the specific fan rating (standard cubic
feet per minute, or scfm) for the oven must be determined.
Depending on the oven design, the air flow can be measured
in one of two ways: 1) if the air is recirculated within the
oven, a sheet metal shroud/duct apparatus can be constructed
and pop-riveted to the air suction or discharge. A pitot tube
and micromanometer can then be used to measure air veloc-
ity through the known diameter duct (see Figure 4). 2) If the
oven design uses once-through air, this flow can be measured
by a technique similar to one used by environmental engi-
neers to measure breathing losses from atmospheric storage-


Figure 4. Student measurement of air velocity in a
convection oven.

80-




E

6501 2, 1 I 2 4 ,i 5
6 5 4 3 2 1 0 1 2 3 4 5 6

Radial Distance at Mid Height (cm)
Figure 5. Experimental radial temperature profile in a
cylindrical geometry of roast beef heated with micro-
waves.
Spring 2003


tank discharge vents. With the oven at a very low heat set-
ting, tape a plastic bag over the discharge vent of the oven to
capture all air flow. Cut one hole along the outside edge of
the plastic bag and insert a tube into it to measure static pres-
sure with a micromanometer (resolution of +0.001 inches
water). Cut another hole, with precisely measured diameter,
approximately in the middle of one face of the bag. This hole
will act as an orifice through which the air in the inflated bag
will escape at a controlled rate. Use the following relation-
ship to determine the cfm of the oven fan:

q= CoAFgAp (1)

where
q gas flow rate (=) ft3/sec
Co correction coefficient for orifice -0.61
A orifice area (=) ft2
g, gravitational conversion factor
Ap pressure drop across orifice (=) lb/ft2
p gas density (=) lbm/ft2
As was done with a conventional oven, prepare a drying curve
and conduct the testing protocol (deflection, hardness, panel-
evaluation test) on the cooked potato sticks.
Microwave Oven Using a microwave oven in cooking
potatoes is advantageous because it results in faster and more
uniform heating. Microwaves penetrate through various foods
and their added energy causes dipoles of the water molecules
to rotate in an alternating field. This alternating-rotation ef-
fect causes friction and provides a source of heat, which ei-
ther thaws or cooks food. The governing energy equation for
microwave heating is[141
6T= oUV2T+ Q (2)
6t pCp

where T is temperature, t is time, ac is thermal diffusivity, p
is density, and C is the specific heat of the material. Note
that the equation contains a heat-generation term, Q, that rep-
resents the conversion of electromagnetic energy to heat. For
small-size food samples where spatial variations in tempera-
ture are negligible, such as our potato sticks, Eq. (2) can be
simplified to

Q=pC 6 (3)

For larger size food materials, the temperature distribution
may vary significantly. Figure 5 shows the experimental ra-
dial temperature profile in a cylindrical geometry of roast
beef heated with microwaves. 151 Note the higher tempera-
tures just inside the edge of the cylindrical wall of the roast
beef due to surface evaporation of moisture.
For our small geometries, thermal gradients within our
potato samples are not expected to be significant. The gener-
alized boundary condition for microwave heating is











-k = h(T- T) +(T4 T+mw (4)

where k is the thermal conductivity, n represents the normal
direction to the boundary, h is the convective heat transfer
coefficient, and T is the convective air temperature. The sec-
ond term is for radiant heat transfer (to be ignored in our
experiment), where E is the surface emissivity and o is the
Stefan-Boltzmann constant. The third term describes evapo-
ration at the surface, where mw is the mass of water and k is
the latent heat of evaporation. This evaporation term is more
important in the microwave cooking versus cooking in a con-
ventional oven because moisture moves rapidly from the in-
terior to the outside (due to uniform Ihc.iiii.-.
Although microwave heating provides a constant heat
source, the highest temperature initially within foods that have
large quantities of water (such as our potatoes) is the boiling
point of water. After most of the moisture had been evapo-
rated from the food, the temperature will rise to higher val-
ues and eventual surface charring will occur.
When cooking at different settings of a microwave oven,
the power is not attenuated. Instead, different power set-
tings cause the oven to cycle off and on. For example a
50% power setting means the oven is on at full power
only 50% of the time.
One other unusual phenomenon that occurs with micro-
wave heating of food that is not observed with conventional
heating methods concerns the movement of internal mois-
ture. A potato can be modeled as a capillary, porous body.
With microwaves, thermal gradients within the potato can
usually be ignored since essentially all parts of the potato are
heated simultaneously. In conventional heating methods,
moisture usually diffuses from inside the potato to the out-
side as a result of thermal and concentration gradients. With
microwave heating, an additional driving force for moisture
migration is due to generation of substantial pressure gradi-
ents within the potato. Positive pressures can build up within
the potato that cause moisture to rapidly move to the surface,
where it evaporates.
Prepare drying curves for potato sticks at maximum mi-
crowave setting.

Pressure Cooker An added dimension of cooking is of-
fered by using a pressure cooker. In addition to temperature
and heat transfer effects, students can assess how elevated
pressure affects cooking times and final product quality. With
standard home-cooking pressure cookers designed for pub-
lic consumers, low operating pressures are used for obvious
safety reasons. By measuring the diameter of the opening in
the top of a cooker and weighing the top floating element,
students can determine the pressure rating (psi) of the cooker.
Boiling water within the cooker is used to generate a fixed
pressure, and therefore only one temperature is available to
146


cook potatoes with this device. There are expensive pressure
cookers that allow some control over the cooking pressure,
but the pressure setting of the inexpensive models are pre-set
by virtue of the weight of the top floating element. The pres-
sure setting for our cooker was 10 psig, and our potatoes
cooked at a temperature of 1160C (2400F). With the water
boiling, place seven potato sticks with skewers in the bottom
of the cooker (but out of the water), and tighten the lid. With
a small-volume cooker, the pressure should build rapidly.
Once operating pressure is attained, by evidence of escaping
pressure, begin timing the cooking process. Every three min-
utes, quickly release pressure from the cooker and remove a
potato stick. Retighten the cooker lid and resume pressure
levels to cook the remaining potato sticks.
With a standard pressure cooker, there is no quick way to
release pressure from the vessel. Pressure-cooker procedures
instruct the operator to place the pan in cool water or wait
until it cools to room temperature before removing the lid.
This is for obvious safety reasons. For purposes of this exer-
cise, our pressure cooker was modified by welding a half-
inch ball valve (with Teflon seats) to the pan top. This pro-
vided a quick-relief method to depressurize the pan so that
potato sticks could be removed and the pan expeditiously
returned to steady-state operation. Note: in constructing and
welding the ball valve to the lid, be careful to install the valve
so that the integrity of the pan and the secondary relief de-
vice is not compromised. Once the valve is attached, test the
final apparatus behind a safety hood to ensure a safe vessel
prior to having students work with the unit.

TESTING PROTOCOL
Initially, a "potato optimum" base case is established in a
conventional oven. This optimum is defined by the student
in terms of surface hardness (measured with a durometer),
mechanical strength (determined with a compressive force
gage), and qualitative factors (assessed by a product panel
test). Once the optimum is defined, the student is chal-
lenged to predict this same optimum in other heat trans-
fer equipment (convection and microwave ovens and a
pressure cooker).
Hardness Material hardness is a common material test-
ing characteristic used to gauge surface hardness of rubbers,
polymers, metals, textiles, printing, and forestry products. A
raw, uncooked potato has a firm surface. As it is cooked, the
surface will become softer as pectin bonds begin to loosen.
As the potato is progressively heated, its surface become drier
until finally it becomes quite firm if overcooked. Using the
durometer hardness tester, stages of potato-surface hardness
can be tracked over time during the cooking process.
Deflection There are many ASTM (American Society
for Testing and Materials) testing methods available
(www.astm.org) to measure compression, torsion, and ten-
sion of solid materials. Zhao[161 found that potatoes lose me-
Chemical Engineering Education











chanical strength during the cooking process and determined that
compressive losses were due to the release of pectic substances
within the potato. In our experiment, a potato stick of length 10.2
cm (4.0 in) is progressively tested for deflection during the cook-
ing process. A raw potato stick is very firm and has good me-
chanical strength. As it is cooked, chemical bonds within the veg-
etable pectin are broken and the potato loses mechanical strength.
To perform the test and track this loss of strength during the cook-
ing process, support the length of the potato stick with fulcrums
at each end (about 1.25 cm from each end). Using a compressive
force gage fitted with a large bearing surface, apply the in-


PRODUCT EVALUATION SHEET
PANEL TESP*
Run Coonkmg Setting Time Deflect Hardness Color Texture Feel Odor last
device


Figure 6. Product evaluation sheet.


Spring 2003


strument probe at the mid-top surface of the potato stick.
Apply downward pressure to deflect the stick a vertical
distance of 6.35 mm (0.25 in). Record the force neces-
sary to deflect the potato stick.
Panel Evaluation The Product Evaluation Sheet is seen
in Figure 6. Criteria of color, texture, feel, odor, and taste
are to be evaluated for potatoes during progressive stages of
cooking. Use these criteria, coupled with hardness and de-
flection, to define a "potato optimum." Taste and odor of
beverages and foods is a complex, subjective process. In
many cases, organic molecules responsible for taste and
odors in various foods have been identified, but the defini-
tion of ideal taste will always remain a subjective experi-
ence. In the case of potatoes, potato aroma is attributed to
the pyrazin family of organic molecules, namely 2,5-dim-
ethyl pyrazin and 2-ethyl pyrazin.[17] The specific fresh po-
tato aroma is attributed to 3-methylmercaptopropanal.E18'

HEAT-TRANSFER CALCULATIONS
Once a "potato optimum" is established in a conventional
oven (natural convection), heat-transfer calculations are used
to predict the same optimum in a forced-air convection oven.
Trying to predict or reproduce the identified potato opti-
mum in conventional and convective ovens is a study in
unsteady-state heat transfer. The student charts the tempera-
ture history within a long cylinder as hot air is passed trans-
versely across the outside surface of the french-fry geom-
etry. This is a case of heating a conducting body having an
initial uniform temperature, under conditions of negligible
surface resistance. Heisler charts,[191 Gurney charts,E201 or
Carslaw/Jaeger correlationsE21' are useful resources for nu-
merical solutions to the classical Fourier series of heat con-
duction. Graphical correlations for Nusselt number versus
Reynolds number for flow normal to single cylindersE22] are
used for approximate modeling of natural and forced con-
vective heat transfer to the rectilinear french-fry geometry.
These correlations allow determination of heat transfer co-
efficients for the unsteady-state heating process. See Table
1 for physical property data for potatoes.
Heat-transfer calculations in the microwave oven are com-
plex and the students are instructed to prepare drying curves
only from the microwave. A priori predictions of potato
optimums based on heat-transfer data collected from con-
ventional and convection ovens were not assigned. Also,
heat-transfer calculations were not performed with the pres-
sure cooker apparatus because use of elevated pressure con-
ditions and the "non-browning" option made it difficult to
perform a direct comparison to potato cooking in conven-
tional and convective ovens. Students determined potato op-
timums in the microwave oven and pressure cooker and
qualitatively compared cooking times and final overall po-
tato characteristics among the various cooking appliances.
Continued on page 153.
147


* Color: Owhite, @off white, *yellow, Obrown, brownsh-black, Oburnt
Texture: Omoit, @smooth, slghtly-rough, Ownkled uneven, rough
Feel: Oraw, soggy, *rubbery, Oabout right, Odry, @overdone
Odor: Ono aroma, @shght potato, e potato, Oslightly bunt, Oburnt, @unidentified
Taste: Oflavorless, @slght potato, Opotato. Ouidentified, 0after-taste, bunt


TABLE 1
Potato Properties

Thermal properties of potatoes depend on porosity, structure,
moisture, and chemical constituents.
Estimates are provided from the following sources:

1. Specific heat:[231
CP = 0.216 + 0.780W (W = % moisture > 0.50) kcal/kg K
and
Cp = 0.393 + 0.437W (W = 0.20 0.50) kcal.kg.K
2. Thermal conductivity:1131
k = 0.554 W/m K
3. Equilibrium moisture content'l2
7 to 10% at relative humidity of 30 to 50%, respectively
4. Heat transfer coefficient of fried potatoes in oil[251
330 335 W/m2 C for top, and 450 480 for bottom
After crust formation, coefficient dropped to 70-150 and 150-190










[W =class and home problems


AN EXERCISE FOR

PRACTICING PROGRAMMING

IN THE ChE CURRICULUM


Calculation of Thermodynamic Properties

Using the Redlich-Kwong Equation of State



MORDECHAI SHACHAM, NEIMA BRAUNER,1 MICHAEL B. CUTLIP2
Ben-Gurion University of the Negev Beer-Sheva 84105, Israel


any students find it difficult to learn programming.
One source of difficulty has to do with the com-
plexity and relevance of the examples and exer-
cises being used. Exercises that are simple enough for a stu-
dent to write a working program in a reasonable length of
time, without too much frustration, are often irrelevant to their
chemical engineering studies. Consequently, they often do
not see the benefit in learning programming and lose inter-
est. More complex and realistic exercises, however, may re-
quire a long and frustrating debugging period, causing them
to lose faith in their ability to make the program run and dis-
couraging them from further programming attempts.
A good exercise to help students learn programming would
be one of practical importance that can be constructed gradu-
ally in several steps. At each step, new types and more com-
plex commands would be added to the program, but only
after debugging of the previous step had been completed.


: Tel-Aviv University, Tel-Aviv 69978, Israel
2 University of Connecticut, Storrs, CT 06269


This paper presents such an exercise-one that involves
analytical solution of the Redlich-Kwong equation for the
compressibility factor and consequent calculation of molar
volume, fugacity coefficient, isothermal enthalpy, and entropy
departures. The solution is demonstrated using MATLAB,E1l
but other programming languages (such as C or C++) can
also be used.

Mordechai Shacham received his BSc (1969) and his DSc (1973) from
Technion, Israel Institute of Technology He is a professor of chemical en-
gineering at the Ben-Gurion University of the Negev His research inter-
ests include analysis, modeling, regression of data, applied numerical
methods, computer-aided instruction, and process simulation, design, and
optimization.
Neima Brauner is professor and head of mechanical engineering under-
graduate studies at Tel-Aviv University. She received her BSc and MSc in
chemical engineering from the Technion Israel Institute of Technology, and
her PhD in mechanical engineering from Tel-Aviv University. Her research
has focused on the field of hydrodynamics and transport phenomena in
two-phase flow systems.
Michael B. Cutlip is a BS and MS graduate of The Ohio State University
(1964) and a PhD graduate of the University of Colorado (1968), all in
chemical engineering. He is coauthor with Mordechai Shacham of the
POLYMATH software package and a recent textbook on numerical prob-
lem solving.


Copyright ChE Division ofASEE 2003


Chemical Engineering Education


The object of this column is to enhance our readers' collections of interesting and novel prob-
lems in chemical engineering. Problems of the type that can be used to motivate the student by
presenting a particular principle in class, or in a new light, or that can be assigned as a novel home
problem, are requested, as well as those that are more traditional in nature and that elucidate
difficult concepts. Manuscripts should not exceed ten double-spaced pages if possible and should
be accompanied by the originals of any figures or photographs. Please submit them to Professor
James 0. Wilkes (e-mail: wilkes @umich.edu), Chemical Engineering Department, University of
Michigan, Ann Arbor, MI 48109-2136.










Calculation of the Compressibility Factor and
Derived Thermodynamic Properties
Using the Redlich-Kwong Equation of State

The two-parameter Redlich-Kwong (R-K) equation of state
has an accuracy that compares well with more complicated
equations that incorporate many more constants (when ap-
plied to non-polar compounds[2]). The R-K equation is a cu-
bic equation in the volume (or in the compressibility factor)
for which analytical solutions can be found.[3] After solving
for the molar volume (or compressibility factor), several
important thermodynamic functions (such as fugacity co-
efficient, isothermal enthalpy, and entropy departures) can
be calculated.
In this exercise, the molar volume, the compressibility fac-
tor, the isothermal enthalpy departure, the isothermal entropy
departure, and the fugacity coefficients are calculated and
plotted for water vapor in the supercritical region. The val-
ues of reduced pressure and reduced temperature used are
shown in Table 1.
Equations and Numerical Data
The R-K equation is usually written 41
RTa
P = T-a (1)
V -b V(V+b)T (1)

where


a = 0.42747R2T5/2
P O c 6

b = 0.08664 -- L
I PC


P pressure (atm)


The exercise presented here
enables students to start a programming
assignment at a fairly simple level and to
build it up gradually to a more complex
assignment of practical importance...

V molar volume (liters/g-mol)
T temperature (K)
R gas constant [R=0.08206 (atmliter/g-mol K)]
T critical temperature (K)
P critical pressure (atm)
Eliminating V from Eq. (1) and writing it as a cubic equa-
tion of the compressibility factor, z, yields

f(z) = z3 -z2 qz- r = 0 (4)
where


r = A2B

q =B2 +B-A2

A2 = 0.427474 R
TR )

B = 0.08664 --
TR


in which PR is the reduced pressure (P/Pc) and TR is the re-
(2) duced temperature (T/Tc).
Equation (4) can be solved analytically for three roots, some
of which may be complex. Considering only the real roots,
(3) the sequence of calculations involves the steps

C = (f)3 2 (9)

where
f 3q-1 (10)
3
S-27r 9q 2 (11)
S(127
If C > 0, there is one real solution for z:
z=D+E+1/3 (12)
where


D, 1/3


/ -_, 1/3


If C < 0, there are three real solutions for z:

Sf( 2r(k- 1)+ 1
zk=2 -cos[ + -1) k=1,2,3
S3 3 3 3


Spring 2003


TABLE 1
Reduced Pressure and Reduced Temperature
Values for Example 1

Pr Pr Pr Pr Pr Tr
0.1 2 4 6 8 1
0.2 2.2 4.2 6.2 8.2 1.05
0.4 2.4 4.4 6.4 8.4 1.1
0.6 2.6 4.6 6.6 8.6 1.15
0.8 2.8 4.8 6.8 8.8 1.2
1 3 5 7 9 1.3
1.2 3.2 5.2 7.2 9.2 1.5
1.4 3.4 5.4 7.4 9.4 1.7
1.6 3.6 5.6 7.6 9.6 2
1.8 3.8 5.8 7.8 9.8 3
10










where

g2 /4
= a cos (16)
f/ 27
In the supercritical region, two of these solutions are nega-
tive, so the maximal zk is selected as the true compress-
ibility factor.
After calculating the compressibility factor, the molar
volume (V), the isothermal enthalpy departure (AH*), the
isothermal entropy departure (AS*), and the fugacity co-
efficient (f) are calculated from[41


V = zRT (17)
P

AH*RT 2bRT3 n l+ b-_(z_l) (18)
RT 2bRT ( V

AS* a ( b_(z Pb (
R 2bRT3/2 ( V1 RT (

A = exp{z -1 nLz( b)] bRT 3/2 n + b) (20)
V bRT ( V

The numerical data needed for solving this problem in-
clude R = 0.08206 literatm/g-mol-K, critical temperature
for water T. = 647.4 K, and critical pressure of water P. =
218.3 atm.
Recommended Steps for Solution

1. Prepare a MATLAB m-file for solving the set of
equations for Tr = 1.2 and Pr = 5 (C, in Eq. 9, is
positive) and Tr = 10 and Pr = 5 (C, in Eq. 9, is
negative). Compare the results obtained with
values from generalized charts of thermodynamic
properties.
2. Convert the program developed in part 1 to a
function and write a main program to carry out the
calculations for Pr = 5 and the set of Tr values
shown in Table 1.
3. Extend the main program to carry out the calcula-
tions for all Pr and Tr values shown in Table 1.
Store all the results of z, V, enthalphy and entropy
departures, and fugacity coefficients in column
vectors. Display the various variables versus Pr
and Tr in tabular and graphic forms.

Solution
The MATLAB program (m-file) for solving the set of
equations for one value of Tr and Pr and displaying the
values of selected variables is shown in Figure 1. Prepa-
ration of the program requires that students rewrite the
equations using the MATLAB syntax. This stage includes
changing some variable names to valid MATLAB names,
changing some algebraic operators, and changing some
150


intrinsic function names (such as converting In to log). The use of
the "max" intrinsic function to select the maximal compressibil-
ity factor from the values obtained in Eq. (15) requires storing
these values in a vector.
The equations must also be reordered according to a proper com-
putational order (thus a variable is not used before a value is as-
signed to it). This can be most easily achieved by first entering

R = 0.08206; % Gas constant (L-atm/g-mol-K)
Tc = 647.4; % Critical temperature (K)
Pc = 218.3; % Critical pressure (atm)
a = 0.42747*R^2*Tc^(5/2)/Pc; % Eq. (2), RK equation constant
b = 0.08664*R*Tc/Pc; % Eq. (3),RK equation constant
Pr = 6; % Reduced pressure dimensionlesss)
Tr = 1.2; % Reduced temperature dimensionlesss)
Asqr = 0.42747*Pr/(TrA2.5); % Eq. (7)
B = 0.08664*Pr/Tr; % Eq. (8)
r=Asqr*B; % Eq. (5)
q=BA2+B-Asqr; % Eq. (6)
f = (-3*q-)/3; % Eq. (10)
g=(-27*r-9*q-2)/27; % Eq. (11)
C= 3 1 3+ig 2 2;% Eq.(9)
if (C>0)
D=((-g/2+sqrt(C))^(1/3)); % Eq. (13)
El=(-g/2-sqrt(C)); % Eq. (14)
E = ((sign(E1)*(abs(El))^(1/3))); % Eq. (14)
z = (D+E+1/3) % Compressibility factor (dimensionless)Eq. (12)
else
psii = (acos(sqrt((g^2/4)/(-f^3/27)))); % Eq. (16)
zv(1) = (2*sqrt(-f/3)*cos((psii/3))+1/3); % Eq. (15)
zv(2) = (2*sqrt(-f/3)*cos((psii/3)+2*3.1416*1/3)+1/3); % Eq. (15)
zv(3) = (2*sqrt(-f/3)*cos((psii/3)+2*3.1416*2/3)+1/3); % Eq. (15)
z = max(zv) % Compressibility factor dimensionlesss)
end
P = Pr*Pc % Pressure (atm)
T = Tr*Tc % Temperature (K)
V = z*R*T/P % Eq. (17), Molar volume (L/g-mol)
Hdep = (3*a/(2*b*R*TA1.5))*log(l+b/V)-(z-1)
% Eq. (18), Enthalpy departure dimensionlesss)
Sdep = (a/(2*b*R*T^.5))*log((l+b/V))-log(z-P*b/(R*T))
%Eq. (19), Entropy departure dimensionlesss)
f _coeff = exp(z-1-log(z*(1-b/V))-a/(b*R*TA1.5)*log(l+b/V))
% Eq. (20), Fugacity cefficient dimensionlesss)

Figure 1. MATLAB program for calculating compressibility
factor and thermodyamic properties for one value of Re and Pr.


TABLE 2
Comparison of Calculated and Generalized
ChartE5' Values for P = 5

Tr=1.2 Tr=10
Calc. Chart Calc. Chart
Compressibility factor 0.7326 0.7 1.0373 1.0
Enthalpy departure AH*/Tf(cal/g.mol K) 6.0167 6.5 -0.5515
Entropy departure AS* (cal/g.mol K) 3.4616 4 0.0183
Fugacity coefficient V 0.4579 0.47 1.0376 1.05

Chemical Engineering Education











% A script file for calculating compressibility factor and derived
% thermodynamic properties using the Redlich Kwong equation of state.
clear, clc, format compact, format short g
Tc = 647.4 ; % Critical temperature (K)
Pc = 218.3; % Critical pressure (atm)
Pr = 5; % Reduced pressure dimensionlesss)
Tr list=[l 1.05 1.1 1.15 1.2 1.3 1.5 1.72 3];
forj=l:10
Tr = Trlist(j); % Reduced temperature dimensionlesss)
[P,T,V,z,Hdep,Sdep,f_coeff]=RKfunc(Tc,Pc,Tr,Pr)
end

Figure 2. Main program for carrying out the calculations for one
Pr and ten Tr values.

Tr list=[1 1.05 1.1 1.15 1.2 1.3 1.5 1.7 2 3];
Pr list=[0.1 0.2];
i=2;
while (Pr list(i)<9.9)
i=i+l;
Pr_list(i)=Pr list(i-1)+0.2;
end
n_Tr=10;
n Pr-size(Pr list,2);
for i=1:n Tr
forj=l:n Pr
[P(j,i),T(j,i),V(j,i),z(j,i),Hdep(j,i),Sdep(j,i),f coeff(j,i)]=RKfunc(Tc,Pc,Tr lis


% Print tabular results

for i= 1:n Tr
disp( ');
disp(['Tr =' num2str(Tr_list(i))' T(K)=' num2str(Tr_list(i)*Tc)]);
disp(' ');
disp(' Pr P(atm) V(L/g-mol) z Hdep Sdep
Res=[Pr_list(:) P(:,i) V(:,i) z(:,i) Hdep(:,i) Sdep(:,i) f_coeff(:,i)];
disp(Res);
pause


the equations to a program that automatically reorders
them (POLYMATH, for example). The ordered set of
equations can then be pasted into the MATLAB editor.
In addition to the .,IIIicil" statements, this simple
program requires only the "if" statement. No commands
for printing the results are used, but selected variables
are shown during the program execution by selective ad-
dition or removal of the semicolon from the ends of the
commands. Good programming practice requires clear
descriptions of the variables and the equations by add-
ing comments.
The results obtained for compressibility factor, en-
thalpy and entropy departures, and fugacity coefficient
by the MATLAB program are compared
to values of generalized charts (KyleE11)
in Table 2. The differences between the
calculated values (presumed to be more
accurate) and the generalized chart val-
ues are small enough to validate the cor-
rectness of the MATLAB program. For
Tr = 10, no generalized chart values are
available for enthalpy and entropy depar-
ture, but the calculated values match the
t(i),Pr list(j)); trend observed in the generalized chart.


f coeff);


% Plot results

plot(Pr list,z(:,l),'.',Pr list,z(:,2),'-',Pr list,z(:,3),':',Pr list,z(:,4),'-.',Prlist,z(:,5),'--'...
Pr list,z(:,6),'*',Pr list,z(:,7),'o',Pr list,z(:,8),'+',Pr list,z(:,9),'v',Pr list,z(:,10),'^');
legend('Tr-l','Tr=1.05','Tr-1.l','Tr-1.15','Tr=1.2','Tr=1.3','Tr=1.5','Tr-1.7','Tr=2','Tr-3');
title(' Compressibility Factor Versus Tr and Pr')
xlabel('Reduced Pressure Pr');
ylabel('Compressibility Factor(z)');
pause
plot(Pr list,f coeff(:,1),'.',Pr_list,f_coeff(:,2),'-',Pr list,f coeff(:,3),':',Prlist,f coeff(:,4),...
'-.',Prlist,f_coeff(:,5),'--',Prlist,f_coeff(:,6),'*',Pr_list,f_coeff(:,7),'o',Pr_list,fcoeff(:,8),....
'+',Pr _list,f coeff(:,9),'v',Pr list,f coeff(:,10),'^');
legend('Tr= l','Tr=1.05','Tr-=1.1','Tr- 1.15','Tr-1.2','Tr-1.3','Tr=1.5','Tr-1.7','Tr=2','Tr-3');
title(' Fugacity Coefficient Versus Tr and Pr')
xlabel('Reduced Pressure Pr');
ylabel('Fugacity Coefficient(f/P)');
pause

Figure 3. Part of the main program in its final form.
Spring 2003


The principal change that has to be in-
troduced in the program, when proceed-
ing to the second step of the develop-
ment, includes the addition of the func-
tion definition statement

ftmction[P,T,Vz,Hdep,Sdepqf_coet]=RKfunc(Tc,Pc,Tr,Pr)

and removal of the definition of the vari-
ables Tc, Pc, Tr, and Pr. The Tr and Pr
are the parameters that are changed in
the main program. Putting the definition
of Tc and Pc in the main program en-
ables easy modification of the program
for different substances. All the variables
that should be displayed in tabular or
graphic form are included in the list of
returned variables. The main program
that calls this function in order to per-
form the calculations for Pr = 5 and the
ten Tr values (shown in Table 1) is dis-
played in Figure 2.
The program starts with commands
that are not specific to the problem at
hand and fall into the category of "good
programming practice." The workspace
and the command window are cleared
and the preferred format for printing is
defined. The ten specified Tr values are
stored in a row vector Tr list and a "for"










statement is used to call the function while changing
the parameter values. The results are displayed in a
very rudimentary form, just by omitting the semico-
lon after the call to the function.
After verifying that this function works properly,
the assignment can be finished by adding to the main
program the set of Pr values shown in Table 1, stor-
ing the results, and displaying them in tabular and
graphic forms. Part of the main program in its final
form is shown in Figure 3.
In this program, a "while" statement is used to in-
put the required Pr values into the row vector Pr_list.
The intrinsic function "size" is used to determine the
number of elements in Pr list. The values returned
from the function are stored in two-dimensional ma-
trices, one column for each Tr and one row for each
Pr value. Tables of results are printed for a constant
Tr value, where the respective columns of the results
matrices are united into a single matrix, "Res" which
is displayed.
Only the code for plotting the compressibility fac-
tor and the fugacity coefficient is shown, and the ad-
ditional variables can be plotted similarly. The plots
of the compressibility factor versus Tr and Pr and the
fugacity coefficient versus Tr and Pr are shown in
Figures 4 and 5, respectively. These plots are almost
identical to the generalized charts that can be found
in the thermodynamics textbooks.

CONCLUSION
The exercise presented here enables students to start
a programming assignment at a fairly simple level
and to build it up gradually to a more complex as-
signment of practical importance in chemical engi-
neering. It demonstrates several aspects of good pro-
gramming practice:
The use of comments to clearly .. ... t.
equations and variables
Clearing the workspace and command window
before .iti,:.., execution
Proper -... iin,.. of the equations
Modular construction of the .. *,.*,,,. where
each module is tested separately before its
imn. *.#.'ii. -ii with the other components
A variety of the variable types (i.e., scalar and ma-
trix), intrinsic functions, and simple and complex
commands are used. Thus, the exercise can cover a
considerable portion of a programming course.
Because of the gradual increase of difficulty in
building this program, most students can successfully
complete it and thus gain confidence in their ability
152


Figure 4. Plot of the compressibility factor versus reduced
temperature and pressure.


Figure 5. Plot of the fugacity coefficient versus reduced
temperature and pressure.


to write a "real" program. The outcome of the exercise, the set of dia-
grams that for many decades has been a very important component in all
thermodynamic textbooks, provides an excellent demonstration of the
importance of programming in chemical engineering.

REFERENCES
1. MATLAB is a trademark of The Math Works, Inc.
2. Seader, J.D., and E.J. Henley, Separation Process Principles, John Wiley & Sons,
New York, NY, page 55 (1999)
3. Perry, R.H., C.H. Chilton, and S.D. Kirkpatrick, eds, Perry's Chemical Engineers
Handbook, 4th ed., McGraw-Hill, New York, NY, pages 2-10 (1963)
4. Cutlip, M.B., and M. Shacham, Problem Solving in Chemical Engineering with Nu-
merical Methods, Prentice-Hall, Upper Saddle River, NJ (1999)
5. Kyle, B.G., Chemical and Process Thermodynamics, 3rd ed., Prentice-Hall, Upper
Saddle River, NJ (1999) 1
Chemical Engineering Education












Optimum Cooking of Potatoes
Continued from page 147.

STUDENT DELIVERABLES
1. Prepare single drying curves for potato samples cooked in a
conventional oven, a convection oven, and a microwave
oven. Construct two drying curves (low and high tempera-
ture settings) in a conventional oven. Compare and contrast
all drying curves.
2. Determine the "potato optimum" cooking time (based on
results from hardness, deflection, and panel tests) at a low
temperature setting in a conventional oven. Using heat-
transfer calculations, predict this optimum at a high
temperature setting in the conventional oven and at low and
high temperature settings in a convection oven.
3. Using a microwave oven, determine the potato optimum.
Discuss how this optimum compares to other optimums
obtained in other heat-transfer equipment. Discuss the
advantages and disadvantages of potato cooking with a
microwave oven. Place a damp paper towel over the potato
stick and cook under previous "optimum" conditions. What
happens to the potato quality and why?
4. Using a pressure cooker, determine the potato optimum.
Discuss the nature of this optimum and how it compares to
other optimums obtained in other heat-transfer equipment.
Show calculations to determine the pressure and tempera-
ture conditions within the cooker.

STUDENT FEEDBACK AND OUTCOMES
Students found this exercise to be both energizing and
meaningful in engineering education. Applying principles of
heat and mass transfer to foods they commonly consume gen-
erated considerable interest. Student feedback on the exer-
cise during class evaluations was extremely positive. As an
instructor, I like this exercise because students appear moti-
vated, the experimental setup is relatively inexpensive, and
the activity integrates multiple concepts of drying operations,
conduction, and convective heat transfer.
The outcomes achieved from this classroom experience
were:
Enhanced total learning experience from combining
classroom theory with an experimental component
Reinforcement of ABET outcomes criteria, including (b)
an ability to conduct experiments and to analyze/interpret
data, and (d) an ability to function in multidisciplinary
teams
Letting students address the open-ended question of what
the "optimum potato" is and how it might be produced
Examination and appreciation of temperature and pressure
effects on heat and mass transfer in a food-engineering
application.

CONCLUSIONS
Students found this simple exercise to be a welcome addi-
tion to traditional classroom theory of heat and mass transfer.

Spring 2003


This experimental application seemed to be both motivational
and an excellent learning vehicle. It provided application of
fundamental engineering principles learned in the classroom
to an everyday kitchen environment. Based on calculated
rates of heat transfer, students could evaluate the effects
of cooking and drying operations on something they fre-
quently eat-the common potato.


REFERENCES
1. Barham, P., The Science of Cooking, Springer Verlag, Berlin (2001)
2. Fellows, P.J., Food Processing Technology: Principles and Practice,
CRC Press, Boca Raton, FL (2000)
3. Singh, R.P, and D.P. Heldman, Introduction to Food Engineering,
Academic Press, New York, NY (2001)
4. Grosch, W., and M.M. Burghagen, Food ( .... .- Springer Verlag,
Berlin (1999)
5. Steidle, C.C., and K.J. Myers, "Demonstrating Simultaneous Heat and
Mass Transfer with Microwave Drying," Chem. Eng. Ed., 33(1), 46
(1999)
6. Chen. X.D., "Cooking Potatoes: Experimentation and Mathematical
Modeling," Chem. Eng. Ed., 36(1), 26 (2002)
7. Gladwell, M., "The Trouble with Fries: Fast Food is Killing Us. Can It
be Fixed?" The New Yorker, March 5, 52 (2001)
8. Kolb, D.A., Experiential Learning: Experience as the Source ofLearn-
ing and Development, Prentice-Hall, Englewood Cliffs, NJ (1984)
9. Solen, K.A., and J.N. Harb, "An Introductory ChE Course for First
Year Students," Chem. Eng. Ed., 32(1) (1998)
10. "Why is a Soggy Potato Chip Unappetizing?" Science 293,1753 (2001)
11. Schuette, H.A., and G. Raymond, "What is a Potato Chip?" Food
Indus., 9(11), 54 (1937)
12. Rogers, M.C., CF. Rogers, and A.M. Child, "The Making of Potato
Chips in Relation to Some Chemical Properties of Potatoes," Am. Po-
tato J., 14, 269 (1937)
13. Geankoplis, C.J., Transport Processes and Unit Operations, 3rd ed.,
Prentice Hall, New Jersey, 537 (1993)
14. Datta, A.D., "Heat and Mass Transfer in the Microwave Processing of
Food," Chem. Eng. Prog., 47, 47 (1990)
15. Nykist, W.E., and R.V. Decareau, J. Micro. Power 11, 3 (1976)
16. Zhao, Y., and Y. Wang, "Relationship Between Compressive Strength
of Cooked Potato Slice and Release of Pectic Substances," Shipin
Kexue, 22(5), 16 (2001)
17. Deck, R.E., J. Pokorny, and S.S. Chang, "Isolation and Identification
of Volatile Compounds from Potato Chips," J. Food Sci., 38(2), 345
(1973)
18. Guadagni, D.G., R.G. Buttery, and J.G. Turnbaugh, "Odor Thresholds
and Similarity Ratings of Some Potato Chip Components," J. Food
Sci., 23(12), 1435 (1972)
19. Heisler, M.PR, "Temperature Charts for Induction and Constant-Tem-
perature Heating," ASME Trans., p. 227, April (1947)
20. Gurney, H.RP, and J. Lurie, "Charts for Estimating Temperature Distri-
butions in Heating and Cooling Solid Shapes," I. & E. Chem., 15(11),
1170 (1923)
21. Carslaw, H.S., and J.C. Jaeger, Conduction of Heat in Solids, 2nd ed.,
Oxford University Press (1959)
22. Welty, J.R., C.E. Wicks, R.E. Wilson, and G. Rorrer, Fundamentals of
Momentum, Heat, and Mass Transfer, 4th ed., John Wiley & Sons,
New York, NY (2001)
23. Yamada, T., "Thermal Properties of Potato," Nippon Nogei Kagaku
Kaishi, 44(12), 587 (1970)
24. Tomkins, R.G., L.W. Mapson, and R.J.L. Allen, "Drying of Vegetables:
III. Storage of Dried Vegetables," J. Soc., Chem. Ind., 63, 225 (1944)
25. Sahin. S., and S.K. Sastry, "Heat Transfer During Frying of Potato
Slices," Food Sci. Tech., 32(1), 19 (1970) 1











[I n= laboratory


USING A COMMERCIAL MOVIE

FOR AN EDUCATIONAL EXPERIENCE

An Alternative Laboratory Exercise


MARTIN J. PITT, JANET E. ROBINSON
University of Sheffield Sheffield S1 3JD, United Kingdon


I have used a commercial film, Acceptable Risks, m edu-
cationally for ten years. I give it to small groups of stu-
dents in the timetable slot for a laboratory exercise and
then have them write a report on it. It is not an educational
film; it is a commercial cinema thriller-a "disaster" movie
centered around a chemical plant. It is a drama involving
human beings and is actually surprisingly sympathetic to those
who work in the chemical industry. It is available on video
for a modest price (vastly less than what is charged for some
educational films). Although it did not get the media atten-
tion of The China Syndrome,E21 (which was about a nuclear
power plant, released at about the same time as the Three-
Mile Island incident), it is equally dramatic and watchable.
In some respects, it resembles the Bhopal disaster, but it takes
place on American soil and has characters that we get to know.
Brian Dennehy plays the manager of a Citychem chemical
plant in Oakbridge, under pressure from his bosses to main-
tain production and keep costs down. Eventually there is a
toxic chemical release.
For chemical engineering students, however, there are many
lessons to be learned. More than any other film I have seen
(including specifically educational ones), it shows the tech-
nology and working practices of a plant, from the label-
ing of tanks to operating procedures; it shows what people
actually do in a plant...management, operators, and tech-
nicians in particular.
There are technical issues. Understanding what goes wrong
in this film and witnessing the consequences can give stu-
dents insight into safety tc'lmiii' l ,-. and techniques. More-
over, there is the human side. Perhaps one day some of these
students will find themselves, like characters in the film, un-
der pressure to speed up production and/or to save money.
They see how there are conflicts and interactions between
various groups, or how the company may go under if they
cannot meet the price or order date, resulting in major job


losses and devastating effects on the local economy, or they
see the conflict between politicians and environmentalists who
fight for their own agendas.
As the students themselves recognize, this exercise demands
some intellectual effort and provides a different learning ex-
perience from a traditional experiment and report. Analyzing
what went wrong is more complex than just interpreting ex-
perimental data.

USING THE VIDEO AS
AN ASSESSED PRACTICAL EXERCISE
Typically, I give the film to second-year students in the
time period allotted to a laboratory exercise. Three to six stu-
dents in a room with a video player are told to watch the
movie through to the end. The film takes an hour and a half,
and the students have three hours for the practical. They then
have to write a three-part report:
1) Write a news item for The Chemical Engineer (the main UK
subject journal) reporting on the events as if they had just

Martin Pitt has a Master's and a PhD degree
in chemical engineering from the Universities
of Aston in Birmingham and Loughborough, re-
spectively. He worked in industry as a project
chemical engineer and a chemical plant man-
ager before becoming an academic in 1985. He
looks after the second-year pilot plant laborato-
ries and third-year design projects.





Janet Robinson is a third-year student of
chemical and process engineering at the Uni-
versity of Sheffield. When she wrote the report
contained in this paper she was a second-year
student.


SCopyright ChE Division of ASEE 2003
Chemical Engineering Education











happened, remembering that the details will not yet be known
and that the publication is subject to the libel laws. Their
reading audience will expect to be told the company's name and
the chemicals involved (so far as they are known) as accurately
as possible.
2) Make a personal assessment of what went wrong and who was to
blame.
3) Report on how valuable the experience of watching the film was.
Did it give any insight into industrial practice in chemical
plants? Did it affect your ideas about industrial safety? Was it a
worthwhile alternative to a laboratory exercise?

STUDENT RESPONSES
The student response has been overwhelmingly favorable.
The small number of negative comments acknowledge that
the student would have preferred to do an actual hands-on
practical. Some of the responses to part three of the report
were:
The film allowed me to picture the kind of work I might be
involved with in the future and the quick thinking that is
necessary in a chemical plant in an emergency.
Although the film is about things going wrong, it would
have been pretty dull had it not. It did not put me off
working in the chemical industry. Indeed, it may have
confirmed that this is what I want to do.
In particular, it reminds us that monetary gains should not
be played off against human safety. In addition, the issue of
plant location is raised, something that is currently very
topical because of the recent disaster in Toulouse.
I did consider the film worth watching. I think it was an
insight into the chemical industry from a perspective that I
might not otherwise have had. It highlighted many impor-
tant safety, economic, and social issues.
It was a challenging exercise, and I had to redevelop writing
skills, very different from those I would use in writing
laboratory reports, that I have not really used since I was
studying GCSE English.
Having watched this film, my awareness for the importance
of safety in industry has definitely been increased.
In the course of watching the film, I have learned how a
chemical plant operates, about industrial practice, and about
the safety procedures inside a plant.

A Student's Appraisal (Janet Robinson)

Personally, I think I gained quite a lot from watching the video
and writing this report. Not just about the chemical plant and
industrial practice, but also about writing in a new style com-
pared to my normal work. I actually found the task a lot harder
than writing a traditional lab report. I had to think in more depth
about what I was going to write and make sure that, in the first
place, I did not blame anyone, and in the second place, that I
contributed my own opinions and not just what I had been told.
That is considerably harder than it seems because there are quite
a few people who could be blamed and it was hard to sort out the
correct procedures from the incorrect ones since I have never
been in a situation such as that.


The film showed me just how important safety issues within a
chemical plant are-even simple but very serious i,, i. such as
;ildcir 'tiffin;' and an out-of-date evacuation plan. That sort of
Ih;, should be high on the agenda and should be sorted out
before .... i 1, i. is produced. It has also shown me that you should
not skimp on safety procedures just because a certain amount of
chemical has to be produced. Safety should always come first, no
matter how much pressure you are under I think this is a very
valuable 4i,, i. to know when I go into industry.
I feel the film was worth watching and it taught me a lot. I think
it is an acceptable alternative to the laboratory experiment and
should be made compulsory for a number of reasons. It breaks up
the traditional lab report. You gain valuable new skills such as
writing in a dliffi'ernt manner I also think it teaches a lot about the
day-to-day running of an industrial plant and shows that slight
errors in procedures can have disastrous effects.

CONCLUSION
Watching a commercial film can be a valid educational
experience if students are required to analyze and comment
on it. Chemical engineering is not just about technical pro-
cesses-it is also about people. It is clear that students have
gained insights from watching this film that they did not get
from visiting a plant. I also find this film a useful preparation
for my course in Process Safety and Loss Prevention (where
I show films about Bhopal and Feyzin).
A video can be a useful back-up if some laboratory experi-
ments are temporarily unavailable. It can also be used as a
timetabled class or borrowed for a project. Other films of
relevance to chemical engineering are The China Syndrome[2]
(about problems in the nuclear industry), Erin ... I .. i.. I
(about the effects of chemical pollution), and Ti,, %.il (about
purifying water, with a real chemical engineering finale). The
film Silkwood is briefly concerned with the 1970s nuclear
industry, but has, I think, little value in this context.
Since many chemical engineering departments now have
teachers with degrees in other subjects and no industrial
experience, Acceptable Risks might be a useful primer for
them also.

REFERENCES
1. Acceptable Risks, (film 1986, video 1992) distributed by Prism Home
Entertainment (USA, NTSC, ASIN 6302447569) and Odyssey Video
(UK, PAL, ODY775)
2. The (.... ....... (1979) Columbia/Tristar; NTSC, PAL, DVD (A
particular point that is worth discussing is the human side of safety.
For example, the control room staff take action believing a faulty level
indicator and do not think to look at its duplicate.)
3. Erin Brockovich (2000) Universal Studios, NTSC, PAL, DVD (Sup-
posedly based on a true story about people being poisoned by con-
tamination of water supplies by hexavalent chromium. No real pro-
cess information, but you could ask students to research Cr(VI) and
water supplies; possibly also useful for discussion of ethical issues.)
4. Thirst (1997) New Line Studios, NTSC (TV movie. The hero is prob-
ably a civil engineer, but the story is about bugs in the water supply
getting through filters. There are technical and environmental issues.
The resolution is definitely chemical engineering.) 1


Spring 2003











classroom


USING

MOLECULAR-LEVEL SIMULATIONS

TO DETERMINE DIFFUSIVITIES

In the Classroom



D.J. KEFFER, AUSTIN NEWMAN, PARAG ADHANGALE
University of Tennessee Knoxville, TN 37996-2200


When engineers require a diffusivity for a chemical
species in a fluid mixture for which experimental
data is not available, there are several methods of
obtaining a value. The most obvious method is to experimen-
tally determine the value of the diffusivity, but frequently
time and money constraints rule out this method. In that case,
a theoretical approach to obtain the diffusivity can be used.
There are a variety of established methods for theoretical
determination of diffusivities. For self-diffusivities and trans-
port diffusivities of binary systems in gases, we can obtain
values from kinetic theory and corresponding states argu-
ments, a corresponding state chart, and the Chapman-Enskog
theory,E1-31 and for self-diffusivities and transport diffusivities
of binary systems in liquids, we can use the Wilke-Chang
equation.11 There are also a variety of other empiricisms sum-
marized in the literature.[4E Needless to say, these empiricisms,
while valuable, are limited in terms of the types of systems
that they describe.
An alternative to obtaining the self-diffusivities for fluid
mixtures, including those with an arbitrary number of com-
ponents, is to conduct equilibrium molecular dynamics simu-
lations of the system.[5-71 Engineers have been calculating self-
diffusivities with this method for a number of years, but us-
ing molecular dynamics to obtain self-diffusivities has not
yet become a common alternative in chemical engineering
transport classes because of the historically extensive com-
putational resources required to conduct the simulations.
In this paper we describe our efforts and our results in in-
corporating molecular-level simulations into a graduate trans-
port phenomena course. Above all, our philosophy was to
provide a utilitarian tool that could be used in a manner analo-


gous to existing techniques, such as the Wilke-Chang equa-
tion, to obtain transport diffusivities. Our target audience is
the general graduate students in chemical engineering who
will not necessarily perform molecular-level simulations as
part of their thesis project. In the implementation of this work,
we remain keenly aware of constraints due to time, computa-
tional resources, money, and target-audience qualifications.


@ Copyright ChE Division ofASEE 2003


Chemical Engineering Education


David Keffer has been an assistant profes-
sor in the Department of Chemical Engineer-
ing at the Universityof Tennessee since Janu-
ary, 1998. His research involves, among other
things, computational description of the be-
havior of nanoscopically confined fluids, us-
ing molecular-level simulation techniques.


Austin Newman is in the process of com-
pleting his degree requirements for a Mas-
ter of Science in Chemical Engineering at
the University of Tennessee. He is working
with statistical mechanical models that de-
scribe the transport of fluids in nanoporous
materials.


Parag A dhangale received his BS from the
University of Bombay and his MS from North
Carolina Agricultural and Technical State
University both in chemical engineering. He
is currently pursuing a PhD at the University
of Tennessee. His research involves molecu-
lar and process simulations of adsorption of
multicomponent systems in nanoporous
materials.










BACKGROUND
Academic Preparation
This transport course is taken in the second semester of the
first year of graduate school. The students have already
had graduate courses in thermodynamics, advanced math-
ematics, and fluid mechanics. The advanced mathematics
course includes numerical solution of systems of ordinary
differential equations (ODEs) and partial differential equa-
tions (PDEs).
The course is roughly divided into two components. The
first is the generation of transport properties, such as
diffusivities. The second component is solution of transport
equations, which are most generally systems of coupled para-
bolic PDEs representing transient material, energy, and mo-
mentum balances. Since the students are already equipped to
tackle the equations numerically, the course, while demand-
ing practical solutions with numerical values, focuses on con-
ceptual understanding of transport phenomena.

Molecular-Level Simulation
In an equilibrium molecular dynamics simulation, we se-
lect an appropriate potential that describes intramolecular and
intermolecular interactions.[5-71 A typical potential for the in-
termolecular interaction of spherical molecules is the Lennard-
Jones potential, for which parameters are widely available.[1-21
With a potential, we can generate the classical equations
of motion, which for N spherical molecules result in a
system of 3N coupled second-order nonlinear ODEs. For
the calculation of diffusivities in a bulk fluid, N is gener-
ally in the range from 200 to 1000 molecules. We solve
the ODEs numerically, obtaining positions and velocities
as a function of time.
By collecting, analyzing (using the Einstein relation for
diffusivity), and regressing the trajectories as a function of
time, we can obtain mutual self-diffusion coefficients.515 There
is one mutual self-diffusion coefficient for each species in
the mixture. These coefficients are a function of the thermo-
dynamic state (temperature, density, and composition). They
are self-diffusion coefficients because they were calculated
from an equilibrium simulation in the absence of macroscopic
concentration gradients.
The mutual self-diffusion coefficients provide a quantita-
tive description of each component's mobility in the system,
but they are not transport diffusivities (also called Fickian
diffusivities). We require transport diffusivities if we in-
tend to use them in Fick's Law in a transport equation
(material balance) in order to obtain the solution to an
applied engineering problem.

Irreversible Thermodynamics
The connection between mutual self-diffusivities and trans-


port diffusivities is provided in the framework of irreversible
or nonequilibrium thermodynamics. One commonly used
equation relates the transport diffusivity to the self-diffusivity
via the thermodynamic partial derivativeE1l

Dij = Dselfi c tn(p) (1)
cd an(cj)

Equation (1) contains numerous, potentially serious, assump-
tions. (Critical discussions of the applicability of the equa-
tion are available elsewhere.[9-171) Regardless, Eq. (1) is widely
used for lack of an alternative. (One alternative is to perform
full-blown nonequilibrium molecular dynamics simulations,
which has also been done.181) For a binary mixture, Eq. (1)
yields four diffusivities, which are intended to be used in
Fick's lawE131


NA = CAA = -DAAVCA DABVCB

NB = CBIB = -DBAVCA DBBVCB


Traditional ChE Description of Diffusion
Chemical engineers know that the diffusive behavior of a
binary system can be completely described by a single
diffusivity. Traditionally, we write Fick's law relative to a
molar average velocity, *, and Fick's law is written (for a
binary mixture)


S- CA(VA *)

J -B CB(vB )


-cDBSLVxA

-cDBSLVXB


where DBSL is the only independent diffusivity.11
In the course we begin by calculating diffusivities for bi-
nary mixtures using the traditional correlations and theories,
following the formalism and notation used in Reference 1. In
order to compare the diffusivities of molecular dynamics
simulations to traditional methods, we must present the
diffusivities in Eq. (2) as a single number that can be directly
compared to DBSL in Eq. (3).
We have derived this relationship for a binary mixture and
it is given as

DBSL =XB(DAA -DAB)+XA(DBB -DBA)+


(XAXBDAA XADBA + XDAB XAXBDBB) \ (4)

If the fluid is an ideal gas or we make some other assumption
in which the density is not a function of composition, then
(dc/dxA) is zero. We can use Eq. (4) to obtain a single trans-
port diffusivity for the binary mixture. Since this diffusivity
is the same property with respect to the same frame of refer-
ence that is generated by traditional methods of estimating


Spring 2003











In this paper we describe our efforts and our results in incorporating
molecular-level simulations into a graduate transport phenomena course. Above all,
our philosophy was to provide a utilitarian tool that could be used in a manner analogous to
existing techniques, such as the Wilke-Chang equation, to obtain transport diffusivities....
In the implementation of this work, we remain keenly aware of constraints due to
time, computational resources, money, and target-audience qualifications.


the diffusivity of a binary system (such as the Wilke-Chang
equation), we can make a direct comparison.

A FEASIBILITY STUDY
Time, Money, and Computational Constraints
Part of the reason that using molecular-level simulations to
determine diffusivities isn't as prevalent in chemical engi-
neering classrooms as it could be lies with the perception
that the simulations simply require too much computer power.
While this was true as recently as the 1990s, it is no longer
true. Rigorous molecular-level simulations generating
diffusivities (with error bars small enough to permit pub-
lication) now take only a few minutes on a several-year-
old processor (for example, an AMD Athlon 850 MHz
processor). In the example below, we provide specific
program clocking. Certainly the impediment is no longer
computational resources.
Efficient molecular-level simulations do require a FOR-
TRAN or C compiler. Using a software platform that inter-
prets code rather than compiling it is not an alternative due to
the computational efficiency. In our example, we ran the simu-
lations on
Compaq FORTRAN in the Microsoft Windows environment
Intel FORTRAN in the Linux environment
Matlab in the Microsoft Windows environment
We shall show that a software platform that interprets code,
rather than compiling it, is about four orders of magnitude
slower than a structured code, and is thus not an option. Of
the first two choices above, both have advantages. The ad-
vantage of the Windows environment is its ubiquity-the dis-
advantage is that the FORTRAN compilers for the Windows
environment are relatively expensive. The advantage of the
Linux environment is that both it and the FORTRAN com-
piler software for it are free.


Constraints Based on Target-Audience Qualifications
Our target audience are first-year chemical engineering
graduate students, including those who do not intend to per-
form simulations as part of their thesis work. With this in
mind, we structured the course to address that audience. Each
part of the process that generates the diffusivity (including
the molecular-level simulation, the irreversible thermodynam-
ics, and the traditional description of diffusion) is presented


with a pragmatic attitude: we are engineers who need a trans-
port diffusivity; we first want to understand the techniques
used to obtain the diffusivity; after we understand it, we
want a simple, methodical, (preferably foolproof) algo-
rithm to follow that generates a reliable transport
diffusivity that we can use in material balances describ-
ing applied engineering systems.
The course is in no way intended to be an exhaustive sur-
vey of molecular-level simulation techniques, or of irrevers-
ible thermodynamics, or of the numerical solutions of ODEs
and PDEs. On the contrary, the course describes a procedure
that incorporates each of these elements. As we stated be-
fore, the students have obtained enough background dur-
ing their first semester as graduate students to make this
course content feasible.
When discussing molecular dynamics, we present a com-
plete, self-enclosed description of the procedure.E191 We dis-
cuss only equilibrium molecular dynamics in the
microcanonical ensemble, since that is the simplest sys-
tem to simulate. A "base-case" FORTRAN code for this
system is provided and discussed subroutine-by-subrou-
tine in a lecture.[19]

AN EXAMPLE
As a practical example, we work problem 17.A.5 of Bird,
Stewart, and Lightfoot.El' The problem asks the students to
calculate the quantity, cD BL, of an equimolar mixture of ni-

TABLE 1
Simulation Parameters

Number of N, molecules 108
Number of CH, molecules 108
Volume (A ) 1.88153 x 105
Intermolecular potential Lennard-Jones
ON2 (A) 3.667
GC2H6 (A) 4.388
FN2 (K) 99.8
CH6 (K) 232
Integration algorithm Gear fifth-order predictor corrector22
Time step (fs) 4
Long-range cut-off distance (A) 12
Number of equilibration steps 50000
Number of data production steps 500000


Chemical Engineering Education











trogen and ethane at 288.2 K and 40 atm. In the problem, the
student is instructed to solve the problem using (a) an experi-
mental data point and kinetic theory and (b) correlations and
kinetic theory. If we obtain the concentration, c, of the mix-
ture via the Lennard-Jones equation of state'201 with standard
mixing rules,E211 the values of DBSL are (a) 3.04 x 10-3 cm2/s
and (b) 2.78 x 10-3 cm2/s.
The students then perform a molecular-level simulation
using the parameters given in Table 1. From the simulations,
they obtain self-diffusivities. They use Eq. (1) to generate
transport diffusivities from the self-diffusivities, and they use
the Lennard-Jones equation of state to provide the thermody-
namic derivatives in Eq. (1). They use Eq. (4) to obtain a
single transport diffusivity for the binary system.


Following this procedure, the students obtain a value of
DBSL of 2.98 x 10-3 cm2/s, which is nicely bracketed by the
two estimates obtained via traditional means. A summary of
the results of the molecular-level simulation that generated
the self-diffusion coefficients, as well as the thermodynamic
partial derivatives obtained from the Lennard-Jones equation
of state and used in Eq. (1), are provided in Table 2.
Two notes of explanation are in order for Table 2. The tem-
perature and pressure in the molecular dynamics simulation
do not exactly match those stipulated in the problem. Be-
cause this is not a course in molecular simulation, we limit
ourselves to simulating in the microcanonical ensemble,
which is the simplest ensemble. In using the microcanonical
ensemble, we fix the number of molecules of each species,


Equation-of-State Results


T(K) 293 T(K) 288.2
p(atm) 39.7 p(atm) 40
XN2 0.5 XN2 0.5
c(molecules/A3) 1.148 x 10 c(molecules/A3) 1.148 x 10

DselfN (cm2 / s) 3.445 x 103 pN2- N2 1.1025
seCN2 dPN2

D .(cm2 /s) 2.333 x 10 pN2 CC2H6 -0.2150

CCzH6 PCzH6
DN2,N (cm2 /s) 3.798 x 10 PC2H6 CN,2 2.3916

CC2H16 PC2H6
DN2 ,CH6 (cm / s) -7.405 x 10- PC2H6 DCC2H6 0.7850
DN21C2H tern PC216 aC2H


DC2H6N (cm2 /s)


2.392 x 104


ac
aXN, (molecules/A3)


-4.1060 x 10"


DC 2H,C2H (cm2 /s) 1.832 x 103


2.98 x 103


TABLE 3
CPU Usage
(The Matlab time is projected for a simulation of 550,000 steps, using the fact that a
simulation of 20,000 steps used 179,360 seconds of CPU time.
All codes were run on an AMD Athlon 850 MHz processor.)


Operating System


COMPAQ Visual FORTRAN 6.5
Intel FORTRAN Compiler 5.0
Matlab 5.1


Windows XP Professional
Red Hat Linux 7.1 with Kernel 2.4.2-2
Windows XP Professional


CPU Usage

443
324
4.932 x 106


the total system volume, and the total energy.
Since the problem asks for the diffusivity at a
given temperature and pressure, we estimate the
density that corresponds to the requested T and
p, using the Lennard-Jones equation of state.
We then equilibrate at that density, maintain-
ing a constant temperature with velocity scal-
ing. For data production, we run in the
microcanonical ensemble, which fluctuates
about the set temperature, because there is
no driving force pushing the temperature to
another value.
Second, we see that DNz,CzH6 is negative. It
is acceptable to have a negative diffusivity in a
Fick's law of the form of Eq. (2). This simply
indicates that, all other things being equal, ni-
trogen would diffuse up the ethane gradients.
This negative term, however, is roughly five
times smaller in magnitude than the positive
DN2,N2, which yields a net positive trans-
port diffusivity.
If we were to assume that the molar volume
was not a function of composition, an assump-
tion which is true for, among other systems,
ideal gases, then the latter term from Eq. (4)
would drop out and we would have a numeri-
cal value of DBSL equal to 3.07 x 10-3 cm2/s, as
compared to the value from the complete ver-
sion of Eq. (4), which was 2.98 x 10-3 cm2/s.
The effect of that term is to lower the diffusivity
from a more ideal case.
In Table 3 we provide the CPU usage for our
three cases on an AMD Athlon 850 MHz pro-
cessor. Clearly, either of the FORTRAN cases
makes this calculation a very reasonable home-
work problem, requiring less than 8 minutes of
CPU time. We have solved a system of 648 (3
dimensions x 216 molecules) second-order


TABLE 2
Simulation and Equation-of-State Results


Molecular Dynamics Results


DB (cm2s)


Spring 2003











ODEs over 550,000 time increments (2 nanoseconds of data
production-more than enough time to establish a self-
diffusivity for this system) in 8 minutes.
In this demonstration, we computed the transport diffusivity
of a high-density gas that could be adequately described with
traditional methods, but there is nothing in the simulation
code that limits it to a binary mixture, which therefore greatly
expands the capabilities of molecular-level simulation.

CONCLUSION
We have presented work describing the practical use of
molecular-level simulations to determine diffusivities in a
course targeted at the general audience of first-year chemical
engineering graduate students. We have shown how the simu-
lation techniques can be used to directly complement tradi-
tional methods for obtaining diffusivities. We have provided
an algorithm by which students can generate transport
diffusivities that can be used in material balances that de-
scribe practical engineering applications. In the implementa-
tion of this work, we have shown that it is computationally
feasible to include numerical simulations in the classroom.
We have also shown that it is a financially modest approach
for chemical engineering departments.

ACKNOWLEDGMENTS
DJK would like to thank the Departmental Chair, Dr. John
Collier, for encouraging him to incorporate molecular-level
simulations into the required graduate student curriculum.
He would also like to thank Dr. Hank Cochran for his helpful
discussions and encouragement. Finally, he acknowledges the
students of ChE 548 who conducted these simulations and
demonstrated that this was a worthwhile task: Keith
Bailey, Yang Gao, Bangwu Jiang, Tudor lonescu, Prajakta
Kamerkar, Vishal Koparde, Austin Newan, Yizhong Wang,
and Jiandong Zhou.

NOMENCLATURE
c total molar concentration
c1 molar concentration of component i
D sf, self-diffusivity of component i
D Darken transport diffusivity
D BSL single independent diffusivity for a binary system

J-A flux of component i, relative to molar average velocity

N flux of component i, relative to laboratory frame of
reference
p total pressure
p1 partial pressure of component i
T temperature
v velocity of component i
v* molar average velocity
x mole fraction of component i


ji intermolecular potential well-depth of component i

0i collision diameter of component i

REFERENCES
1. Bird, R.B., W.E. Stewart, and E.N. Lightfoot, Transport Phenomena,
2nd ed., John Wiley & Sons, New York, NY (2002)
2. Hirschfelder, J.O., C.F. Curtiss, and R.B. Bird, Molecular Theory of
Gases and Liquids, John Wiley & Sons, New York, NY (1954)
3. Chapman, S., and T.G. Cowling, The Mathematical Theory of Non-
uniform Gases, 2nd ed., Cambridge University Press, Cambridge
(1952)
4. Reid, R.C., and T.K. Sherwood, The Properties of Gases and Liquids:
Their Estimation and Correlation, 2nd ed., McGraw-Hill, New York,
NY (1966)
5. Haile, J.M., Molecular Dynamics Simulation, John Wiley & Sons, New
York, NY (1992)
6. Allen, M.P, and D.J. I l.1 I Computer .....' .. f Liquids, Ox-
ford Science Publications, Oxford, England (1987)
7. Frenkel, D., andB. Smit, .. ... -- Molecular Simulation, Aca-
demic Press, San Diego, CA (1996)
8. Darken, L.S., "Diffusion, Mobility, and Their Interrelation through
Free Energy in Binary Metallic Systems," Trans. Am. Inst. Mining
andMetall. Engrs., 175, 184 (1948)
9. Carman, PC., "Self-Diffusion and Interdiffusion in Complex-Form-
ing Binary Systems," U. Phys. Chem., 71(8), 2565 (1967)
10. Carman, PC., "Intrinsic Mobilities and Independent Fluxes in Multi-
component Isothermal Diffusion. I. Simple Darken Systems," J. Phys.
Chem., 72(5), 1707 (1968)
11. Carman, PC., "Intrinsic Mobilities and Independent Fluxes in Multi-
component Isothermal Diffusion. II. Complex Darken Systems," J.
Phys. Chem., 72(5), 1713 (1968)
12. McCall, D.W., and D.C. Douglass, "Diffusion in Binary Systems," J.
Phys. Chem., 71(4), 987 (1967)
13. Ghai, R.K., H. Ertl, and F.A.L. Dullien, "Liquid Diffusion of Non-
electrolytes, Part I,"AIChE J., 19(5), 881 (1973)
14. Ghai, R.K., H. Ertl, and F.A.L. Dullien, "Liquid Diffusion of
Nonelectroclytes, Part II," AIChE J., 20(1), 1, (1974)
15. Jolly, D.L., and R.J. Bearman, "Molecular Dynamics Simulation of
the Mutual and Self-Diffusion Coefficients in Lennard-Jones Liquid
Mixtures,"Mol. Phys., 41 (1), 137 (1980)
16. Schoen, M., and C. Hoheisel, "The Mutual Diffusion Coefficient D12
in Binary Liquid Model Mixtures. Molecular Dynamics Calculations
Based on Lennard-Jones (12-6) Potentials. I. The Method of Determi-
nation," Mol. Phys., 52(1), 33 (1984)
17. Karger, J., and D.M. Ruthven, Diffusion in Zeolites and Other
Microporous Solids, John Wiley & Sons, Inc., New York, NY (1992)
18. Heffelfinger, G.S., and F. van Swol, "Diffusion in Lennard-Jones Flu-
ids Using Dual Control-Volume Grand-Canonical Molecular Dynam-
ics Simulation (DCV-GCMD)," J. Chem. Phys., 100(10), 7548 (1994)
19. Keffer, D., "A Second-Semester Course in Advanced Transport Phe-
nomena for Chemical Engineers," course website at clausius.engr.utk.edu/che548/index.html>, Department of Chemical
Engineering, University of Tennessee (2002)
20. Nicolas, J.J., K.E. Gubbins, W.B. Street, and D.J. I .. I "Equa-
tion of State for the Lennard-Jones Fluid," Mol. Phys., 37(5), 1429
(1979)
21. Sandler, S.I., (..... ,.. i .... ..-.... ..Thermodynamics, John Wiley
& Sons, New York, NY p. 318 (1989)
22. Gear, C.W., "The Numerical Integration of Ordinary Differential Equa-
tions of Various Orders," Argonne National Laboratory, ANL-7126
(1966)
23. Gear, C.W., Numerical Initial Value Problems in ( 0' ... Differen-
tial Equations, Prentice Hall, Inc., Englewood Cliffs, NJ (1971) 1
Chemical Engineering Education




Full Text


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E5LVIPR1S_864GAQ INGEST_TIME 2011-05-04T19:22:14Z PACKAGE AA00000383_00002
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ELFE3Q8KI_MUFHT4 INGEST_TIME 2011-09-15T19:11:06Z PACKAGE AA00000383_00002
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES