Chemical engineering education ( Journal Site )

Material Information

Chemical engineering education
Alternate Title:
Abbreviated Title:
Chem. eng. educ.
Physical Description:
v. : ill. ; 22-28 cm.
American Society for Engineering Education -- Chemical Engineering Division
Chemical Engineering Division, American Society for Engineering Education
Place of Publication:
Storrs, Conn
Publication Date:
annual[ former 1960-1961]


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


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:
oclc - 01151209
lccn - 70013732
issn - 0009-2479
lcc - TP165 .C18
ddc - 660/.2/071
System ID:

Full Text




Chemical Engineering Education
Department of Chemical Engineering
University of Florida Gainesville, FL 32611
PHONE and FAX: 352-392-0861

Tim Anderson

Phillip C. Wankat

Carole Yocum

James 0. Wilkes, U. Michigan

William J. Koros, Georgia Institute of Technology


E. Dendy Sloan, Jr.
Colorado School of Mines

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 I. Sandler
University of Delaware
Richard C. Seagrave
Iowa State University
C. Stewart Slater
Rowan University
Donald R. Woods
McMaster University

Chemical Engineering Education

Volume 38

Number 3

Summer 2004

162 University of Oklahoma,
Robert L. Shambaugh, Edgar A. O'Rear Lance L. Lobban

168 Mike Doherty of the University of California, Santa Barbara,
Marv E. Howe-Grant

174 A Quadruple-Tank Process Control Experiment,
Effendi Rusli, Siong Ang, Richard D. Braatz
188 A Dust Explosion Apparatus Suitable for Use in Lecture Demonstrations,
Ronald J. Willey, Edward Shanley
190 Compressible Flow Analysis: Discharging Vessels,
S.E. Forrester, A.V. Nguyen, G.M. Evans, PM. Machniewski
196 Investigation into the Propagation of Baker's Yeast: A Laboratory
Experiment in Biochemical Engineering,
Xuemei Li, Xiao Dong Chen, Matthew T Hardin
204 Using the Evolutionary Operation Method to Optimize Gas Absorber
Operation (A Statistical Method for Process Improvement),
Jimmy L. Smart
212 Laboratory Experiment on Gene Subcloning for Chemical Engineering
Students, Claire Komives, Sabine Rech, Melanie McNeil
228 An Integrated Chemical Reaction Engineering Lab Experiment,
Addlio M. Mendes, Luis M. Madeira, Ferndo D. Magalhaes,
Jose M. Sousa
236 PEM Fuel-Cell Test Station and Laboratory Experiment,
Michael W Fowler, Alfred Lam

182 A Respiration Experiment to Introduce ChE Principles,
Stephanie Farrell, Robert P Hesketh, Mariano J. Savelski

200 How to Evaluate Teaching, Richard M. Felder, Rebecca Brent

216 Solvent Recovery by Condensation: An Application of Phase Equilib-
rium and Sensitivity Analysis, Ronald W Missen

222 Freshman Design in Chemical Engineering at Rose-Hulman Institute of
Technology, Sharon G. Sauer

203 Book Review

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 2004 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 andfor back copy costs and availability.
POSTMASTER: Send address changes to Chemical Engineering Education, Chemical Engineering Department., University
of Florida, Gainesville. FL 32611-6005. Periodicals Postage Paid at Gainesville, Florida and additional post offices.

Summer 2004





The Sarkeys Energy Center, home of Chemical Engineering.

Oklahoma defies the stereotypes made famous in the
Grapes of Wrath. It has the second most var-
ied terrain of any state in the country and contains a
diverse range of fauna and flora. There are pine trees, moun-
tains, alligators, and Spanish moss in the southeast; a Nature
Conservancy tall grass prairie preserve with stalks as high as
an elephant's eye in the northeast; cacti and canyons at the
Wichita Wildlife Preserve in the southwest; and sand dunes
and salt flats in the northwest. These sights and Native Ameri-
can culture captivate visitors today just as they did Washing-
ton Irving, George Catlin, and Theodore Roosevelt in the past.
Located in the center of the state, Norman and the Univer-
sity of Oklahoma (OU) lie 17 miles to the south of Okla-
homa City. It has a lovely, 2000-acre campus covered with
trees, flowers, fountains, and sculpture. There are 28,000 stu-
dents from all 50 states and 108 foreign countries. Notable
features of the University include the outstanding De Gollyer
History of Science book collection, a leading meteorology
program, the Neustadt Prize ("American Nobel" in interna-
tional literature), the Weitzenhoffer Collection of Impression-
ist and Post-Impressionist Art, the Sam Noble Natural His-
tory Museum, and one of the top sites for sequencing of the
human genome. On an absolute and per capital basis, the Uni-
versity regularly ranks among the top comprehensive, publicly
supported institutions in number of National Merit Scholars.
Unprecedented growth has taken place under the dynamic
leadership of President David L. Boren, a Rhodes Scholar
and former U.S. Senator. The growth is evident in ongoing
construction projects on the campus totaling hundreds of

millions of dollars. Recent speakers on campus have included
such dignitaries as Mikhail Gorbachev, Margaret Thatcher,
and Desmond Tutu.
Located in a temperature and humidity-controlled floor of
the main library, the De Gollyer History of Science collec-
tion (an 85,000-volume collection that includes books by
Agricola, Galileo, Newton, and Darwin) is unique. Of the
many volumes by Galileo in the collection, four are first edi-
tions containing Galileo's handwriting, including his own per-
sonal copy of Dialogue Concerning Two Chief World Sys-
tems (1632); this copy contains his margin notes for a never-
published second edition.
The University of Oklahoma was founded in 1890-one
year after the famous land run opened up Oklahoma for settle-
ment. Its educational programs developed at the same time
that chemical engineering was becoming a discipline, and
with Oklahoma situated in the oil- and gas-producing region
of the country, the chemical engineering discipline found fer-
tile ground at OU.
Chemical engineering at OU (now the School of Chemical
Engineering and Materials Science, or CEMS), enjoys a long
and rich tradition. What chemical engineer hasn't used
McGraw-Hill's Perr 's Chemical Engineers'Handbook, prob-
ably the best-known title in chemical engineering? Robert
H. Perry, editor of the fourth through sixth editions (and son
of the original editor, John H. Perry), taught chemical engi-
neering at OU from 1958-1964 and served as the department
director from 1961-1963. Don Green, the current editor of the
Handbook, received his PhD at OU under Perry's supervision.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

University of Oklahoma
Norman, Oklahoma 73019

Another chemical engineering stan-
dard, Transport Phenomena, was re-
viewed prior to publication by Jack
Powers and undergraduates at OU in the
late 1950s. Most chemical engineers
know the BWRS equation, Ken
Starling's modification of the Benedict-
Webb-Rubin equation of state, for light
hydrocarbon calculations, or the
Carnahan-Starling hard sphere modifi-
cation of the van der Waals equation of
state. Those in the natural gas process-
ing industry are familiar with the ac-
complishments of Laurance "Bud" Reid
(the Laurance Reid Gas Conditioning
Conference, held every year at OU
since 1950, is the premier conference
for the gas conditioning industry).
Cheddy Sliepcevich has been dubbed
"the Father of LNG" for his work on
the first LNG carrier, the S.S. Methane
Pioneer. Starling, Reid, and Sliepcevich
were all long-time CEMS faculty.
The self-contained college town of
Norman (population 100,000) is close
enough to Oklahoma City to be consid-
ered a suburb of that city. The advan-
tages of this city of 1 million are easily
accessible. For example, Oklahoma
City recently completed a $200 million
dollar renovation of its historic
Bricktown area, which now includes a
navigable canal, restaurants, hotels, an
art museum, a minor-league ballpark, a
performance center, and a convention
center. There is even scull racing on the
nearby river.


The ChE department has space in the
Sarkeys Energy Center, a 400,000
square-foot academic building that has
a 2.5 acre footprint and a 15-story tower.
The building was completed in 1989 and has exci
ratory, office, and teaching facilities. Presently
full-time faculty, 6 postdocs, 242 undergradu
graduate students. Our diverse funding comes fro
DOE, DOD, American Heart Association, othe
State agencies, and industry. In the last sever
undergraduate students have consistently won
gional AIChE conferences. Our classes are a
male, 7% Hispanic, 10% Native American, an

... our department has shared in the excitement
and turmoil of new areas such as bioengineering and
nanotechnology. We have wholeheartedly embraced these
new areas as natural extensions of chemical
engineering, a discipline that is probably
the broadest and strongest of all
the engineering disciplines.

SDEPARTMENT Undergraduates are very active in
the student AIChE chapter, the re-
10 Chemical engineering program was gional AIChE competitions, and in
established as part of Chemistry the campus Engineers Club, and the
Department graduate students have an active or-
13 First BS in ChE awarded to Albert ganization, the OU CHEGS. As far
Edward Gartide as student-faculty interactions, it is
18 First masters degree in ChE awarded rare not to see a row of open doors in
to Roy Clyde Mitchell the faculty hallways-no appoint-
ment is necessary! A year-end high-
37 Chemical engineering becomes part t is necessary A yearned
of College of Engineering with R.L. light is the AIChE faculty/student
Huntington as head of Chemical roast, during which the faculty and
Engineering students make good fun of each other.
43 First BS awarded to a woman (Freda
55 Cheddy Sliepcevich leaves Michigan CONNECTION
to become chairman of chemical The department values and nur-
engineering; renewed emphasis on tures its relationships with industry.
graduate program as well as Past industrial experience is consid-
undergraduate program ered to be a valuable (although not
-63 Robert Perry (of Perry's Handbook) required) asset for a prospective fac-
is chairman of department ulty candidate. Presently, 50% of the
-86 Carl Locke is department head. In a faculty have worked in industry. Also,
two-year period he hires Harwell, all of the faculty have, or have had,
Scamehorn, O'Rear, Mallinson, and industrial support for their research.
Shambaugh The departmental faculty have 28 ac-
86 Department moves to newly- tive patents, some of which are li-
constructed Sarkeys Energy Center censed to industry.
04 Stephenson Research Center nears In 1987, the department negoti-
completion; the building will provide ated with the University for the es-
additional space for bioengineering establishment of two industrial consor-
tia. At the time, these consortia were
unique to the University and, for that

optional labo-
,there are 15
plates, and 54
:r federal and
al years, our
prizes at re-
,out 38% fe-
d 9% African

matter, to universities in general. For an established annual
fee, the consortia gave the sponsoring companies access to
research, access to prospective employees, and royalty-free
rights to any university patents developed by the consortia.
Both of these consortia are still in existence. The larger, the
IASR (Institute for Applied Surfactant Research), does
ground-breaking, applied research for the detergent, petro-
leum, health-care, and many other industries. The smaller,
the CPFR (Center for Polymer and Fiber Research), concen-

Summer 2004












trates on high-value fiber technology
for use in nonwovens and compos-
ites. Over 35 Fortune 500 companies
have been members of these consor-
tia. More consortia have recently de-
veloped in the department, and the
present dean in the College of Engi-
neering (Skip Porter) has greatly fur-
thered the cause of the industry-aca-
demic connection.
The department's industrial advi-
sory board (OKChE), active since
1969, has served as a model for the
development of similar boards in
other departments of the College of
Engineering. Program review, fund-
raising, student mentoring, and senior
exit interviews are among the many
valuable tasks they perform. Pres-
ently, Robert Purgason (Vice Presi-
dent, Williams Petroleum) is the
board's president, and Larry Evans
(CEO and founder, Aspen Technol-
ogy) is the board's vice president.

Presently, the department has re- A unit operations lab
search strengths in bioengineering, This is a page in Agr
nanotechnology, catalysis, turbulent from the University
Science" Collection.
transport, process optimization, fu-
els, surfactants, and polymers. In the
bioengineering area, the department's history spans at least
four decades. There was early work on an artificial liver, on
artificial blood, and on confirmation of mass transfer corre-
lations for large biomolecular species. Current research ranges
from tissue engineering to approaches for treating cancer,
heart attack, and stroke. At the turn of the millennia, faculty
members in CEMS helped spearhead an initiative to coalesce
diffuse activity in bioengineering across the College. Engi-
neers teamed with biomedical scientists from the Oklahoma
Medical Research Foundation (OMRF) and the Norman and
Health Sciences Center campuses of the University of Okla-
homa to write a successful application for a Special Oppor-
tunity Award from the Whitaker Foundation. As a result of
growth sparked by this grant, 40% of the CEMS faculty now
has research in bioengineering.
With highly regarded researchers in adhesion molecules
and glycobiology and as one of the top five sites for the se-
quencing of the human genome, the University of Oklahoma
has much to offer in tissue engineering (two of our new hires
work in tissue engineering). In the fall of 2004, the bioengi-
neering faculty will expand into the new Stephenson Research
and Technology Building on the south campus. This building


brings together researchers on
the Norman campus in micro-
biology, zoology, biochemistry,
and bioengineering.
In nanotechnology, one of our
faculty (Resasco) has started a
company for the manufacture of
single-wall carbon nanotubes.
This company, SWeNT (South-
west Nanotechnology) uses
Resasco's patented Co-Mo cata-
lyst system for producing
nanotubes of unusual purity and
selective chirality. Resasco's
company presently provides
nanotubes to more than 15 in-
dustrial giants, including three
Fortune 50 companies.
Miguel Bagajewicz is direc-
tor of the Center for Engineer-
ing Optimization. The main fo-
cus of this consortium is the de-
velopment of new design meth-
Al ods to minimize overall cost.
S Other goals include the minimi-
zation of the energy consump-
toryin the really old days. tion, the prevention of pollution,
I's De re Metallica (1556) and the minimization of waste
Oklahoma's "History of generated. Since doing research
in this field requires strong re-
lationships with industry, the
students work on problems of practical interest.
In fuels, the department has a long history of research in
natural gas, petroleum, and coal. Recent work involves natu-
ral gas storage, hydrogen production and storage, fuel cells,
and other alternative energy sources.
As discussed above, research in the department's other two
areas-surfactants and polymers-is heavily involved with
two industrial consortia.

Recent faculty hires have greatly strengthened the
department's bioengineering effort. Traditional strengths are
not being short-changed, however. In fact, we contemplate
growth in non-bio areas.
Vassilios Sikavitsas joined the department as an assistant
professor in 2002. A native of Thessaloniki, Greece, he re-
ceived his PhD in 2000 from SUNY Buffalo. Vassilios then
did a postdoc at Rice University where his work with Tony
Mikos on bone led to the discovery that shear stress promotes
cell differentiation (marrow stromal cells to osteoblasts). Such
results are important in preparing tissues and organs in the

Chemical Engineering Education

A more modern
laboratory scene
than the one
shown on the
facing page.
Graduate stu-
dents Olga Rueda
and Jose Herrera
are shown
operating the
XPS in Resasco's
catalysis labora-
Sliepcevich with
a model of the
S.S. Methane
Pioneer, the first LNG carrier, which was developed at OU.

laboratory on a time scale feasible for clinical use. He has
become somewhat of a star in the newspapers and on local
TV, where his research on bone constructs using synthetic
scaffolds provides visibility to the growing activity in bioengi-
neering in the department.
Peter McFetridge also joined the department as an assis-
tant professor in 2002. He was born in Rotorua, North Is-
land, New Zealand (he is a "Kiwi"), and he captivates every-
one with stories of his home country. He, like several others
in our department, conducts research in tissue engineering,
but with an emphasis on the production of vascular grafts
that he began studying while receiving his PhD under the
direction of Julian Chaudhuri and Mike Horrocks at the Uni-
versity of Bath. Pete's approach uses natural scaffolds pre-
pared by decellularization of blood vessels isolated from
umbilical cords. Recent results in his lab show promise for
fully automating the dissection protocols, which has vastly
improved the mechanical uniformity of these natural vascu-
lar constructs. Pete's background reads like a storybook. In
one past job, he worked in the New Zealand hills laying high-
pressure gas lines (in the beautiful locale where Lord of the
Rings was filmed). After a list of other activities, including
motorcycle racing and resultant broken legs, he left for the
UK and took a job driving a dust cart, a 32-ton lorry (truck)
used to transport industrial waste (and to pay for his Euro-
pean excursions).
David Schmidtke, who was born in Sheboygan, Wiconsin,
became an assistant professor in the department in 2000. Af-
ter receiving his PhD from the University of Texas at Austin

(1997), he did postdoctoral studies at Penn with Scott Dia-
mond. Although we did not know it at the time of his hiring,
Dave already had an Oklahoma connection-his grandfather
was the pastor of a Lutheran church in Oklahoma City fifty
years ago. For his research, David focuses on the phenom-
enon of tether formation during leukocyte (white blood cell)
adhesion, biosensors for diabetes, and microfluidics. His ad-
dition to the faculty builds on existing expertise in cell adhe-
sion with Ulli Nollert and Rodger McEver, the Lilly Chair at
the OMRF (Oklahoma Medical Research Foundation).
Dimitrios V. Papavassiliou joined the department as an
assistant professor in 1999. A native of Karditsa, Greece, he
received his PhD in 1996 from the University of Illinois at
Urbana-Champaign. After receiving his degree, he gained
valuable real-world experience as a Senior Research Engi-
neer at the Mobil Technology Company Upstream Strategic
Research Center in Dallas, TX. His research focus is on the
fundamental understanding and modeling of transport pro-
cesses with industrial and environmental interest. His group
develops novel computational methods that are applied to
explore turbulent transport of mass and heat, reactive flows,
turbulent jet flows, turbulent drag reduction, and flow and
transport through porous media. High Performance Com-
puters are used to conduct the numerical experiments and to
interpret the data.
Miguel J. Bagajewicz joined the department in 1995. A
native of Buenos Aires, Argentina, he received his PhD from
Cal Tech in 1987. Prior to coming to OU, he was a Staff
Associate Member, Argentine National Research Council
(Conicet) (1980-91), an Associate Professor at the Universi-
dad Nacional del Litoral, Argentina (1987-91), a Senior En-
gineer with Simulation Sciences (SimSci) (1992-95), and a
Visiting Professor at UCLA (1995). His experience at SimSci
has been particularly valuable for integration into his teach-

Summer 2004

ing of the capstone design course. As director of the Center
for Engineering Optimization at OU, his research has focused
on the design, operation, simulation, and optimization of pro-
cess plants. In the area of process operation, his group fo-
cuses on data reconciliation technology. Since plant data are
corrupted by noise and instrument malfunction, data recon-
ciliation is used to filter this information.
Brian Grady, a Chicago native, joined the department in
1994 after receiving his PhD from the University of Wiscon-
sin-Madison. He gained real-world experience as a Project
Engineer with Procter and Gamble (1987-89) prior to attend-
ing graduate school. His awards include an NSF CAREER
Award (1998) and an Alexander von Humboldt Research
Fellowship (2000). The latter was awarded for a year's stay
at the Max Planck Institute for Colloid and Interface Science
near Berlin. Brian's research is focused on polymer systems
with two different emphases: polymer systems with two dif-
ferent components such as polymer-matrix composites or
phase separated copolymers, and polymer or surfactant
nanostructures at solid-liquid interfaces. He is a Councilor
for the Society of Plastics Engineers and serves on the Board
of Directors in the Engineering Properties and Structure Di-
vision of that same organization.
Daniel Resasco, a native of Bahia Blanca, Argentina, joined
the department in 1993. He received his PhD from Yale (1984)
and has previous academic experience as a Professor (and
Chairman for part of his stay) in the Chemical Engineering
Department at Universidad Nacional de Mar del Plata (1983-
90). He also served as a Visiting Professor at Yale University
(1986-87, 1991). He received industrial experience as a Se-
nior Scientist, Sun Company, Inc., Pennsylvania (1991-93).
Recently (2001), he became an Associate Editor of the Jour-
nal of Catalysis. In his research in the area of heterogeneous
catalysis he seeks to understand the relationship between the
catalytic performance and the microscopic structure and com-
position of the material, in addition to the links between the
synthesis process and the final catalyst. His work is appli-
cable to industrial processes such as isomerization and dehy-
drogenation of lower alkanes, aromatization of paraffins, and
nitration of aromatics. Another important application of his
studies is in the area of environmental catalysis for the abate-
ment of NOx in the presence of 02, HO2, and SO2. In the last
few years, his work has progressed to an area of high impact
and visibility-the controlled catalytic synthesis of carbon
nanotubes in processes that can be scaled-up.
Matthias (Ulli) Nollert received his PhD from Cornell
in1987. Although he was born in Luray, Virginia, his parents
are from Germany. He joined the department in 1991 after a
postdoc at Rice University. Ulli's work emphasizes the adhe-
sion molecules in platelets and white blood cells and the
measurement of adhesion forces. Leukocyte adhesion to the
endothelial cells lining the blood vessels of the body is a key
step to white blood cell function and the process of inflam-

mation. There is good evidence that changes in blood flow
characteristics may lead to the development of vascular dis-
ease. Only by studying vascular cells in a flowing system
that closely mimics the environment found in the blood ves-
sels can one truly understand how these cells behave in the
body and why vascular disease occurs. His group is currently
examining alterations in protein production in blood vessel
wall cells that are exposed to fluid flow.
Roger Harrison received his PhD from the University of
Wisconsin-Madison (1975) and came to OU in 1988. He is
our "local" professor-he was raised in Altus, in the south-
western section of the State. Roger has a vast amount of in-
dustrial experience as a Research Engineer with Chevron
Research (1968-70), a Research Scientist with Upjohn (1975-
81), and as a Senior Research Engineer with Phillips Petro-
leum (1981-88). He designs hybrid proteins that are produced
in E. coli using the techniques of gene insertion and expres-
sion. Around these methods, he has developed strategies for
targeted cancer agents and for improved solubility and puri-
fication of expressed protein. For cancer, the hybrid or "fu-
sion" protein contains a component selective for a tumor cell
and a portion that will cause cell death. Fusion proteins for
improved separations use a highly soluble species linked to
the target molecule. Roger has recently published a textbook
titled Bioseparations Science and Engineering (with P. Todd,
S.R. Rudge, and D. P. Petrides). This well-written book fills
an important need for a textbook on bioseparations. The text
was first made available for adoption in courses for the spring
2003 semester. The text has already been adopted for use in
courses at thirty universities, including Carnegie-Mellon,
Princeton, Cornell, Imperial College (London), and Ohio State.
Lance Lobban, a native of McPherson, Kansas, received
his PhD from the University of Houston in 1987 and came to
OU the same year. As Director of our department, he has
been instrumental in guiding us into the new millennium. His
research focuses on catalysis and reaction engineering. His
group studies gas phase reactions at temperatures up to 8000C,
liquid phase reactions at room temperature and below, and
reactions in presence of a strong electric field. One very ac-
tive project is an investigation of methane oxidative coupling
under a variety of conditions including cold plasma condi-
tions and on different catalysts. A second project involves
the synthesis and use of novel TiO2 aerogels and binary SiO2-
TiO2 aerogels as photocatalysts. The unique properties of these
aerogels are hypothesized to allow more efficient use of UV
light to activate the photocatalysts for the complete oxida-
tion of air and water contaminants. Not only is Lance an out-
standing director and researcher, but he also continues to
maintain a teaching load. And he teaches extremely well-
the students just gave him an outstanding professor award-
for the fourth time!
Robert Shambaugh, a native of Youngstown, Ohio, came
to OU in 1983. He received his PhD from Case Western Re-

Chemical Engineering Education

serve (1976). Prior to coming to OU he gained nine years of
industrial experience at Du Pont. Besides tours at two plant
sites, he spent most of this time at the Experimental Station
in Wilmington, Delaware. At OU, he has been heavily in-
volved in the aforementioned CPFR, an industrial consor-
tium that works in the fibers area. His research group is par-
ticularly interested in melt blowing, a process wherein a high-
velocity gas stream meets a stream of molten polymer as the
polymer exits a fine capillary. The result of this impact is that
the polymer rapidly (in about 50 microseconds) attenuates
into fiber strands as fine as 0.1 micron in diameter. Extremely
interesting and potentially very strong crystal structures are
formed under these high strain rate conditions.
Richard G. Mallinson, a native of Indianapolis, Indiana,
came to OU in 1983. He received his PhD from Purdue (1983).
He was Director of the Institute for Gas Utilization Tech-
nologies of the Sarkeys Energy Center from the mid-90s un-
til 2003. He was integrally involved in the development of a
curriculum for a new interdisciplinary Masters degree pro-
gram in Natural Gas Engineering and Management that in-
cludes internet-based courses via streaming video and
netmeeting. Rick's research involves the many aspects of
energy and fuels. His energy-related research, which began
with studies of coal and shale oil liquefaction, has more re-
cently focused on natural gas conversion and gas vehicle fuel
storage and gas transportation. These studies emphasize the
understanding of the chemical and physical processes under-
lying observable thermodynamic and rate behavior. The ef-
forts are directed toward applications of reaction engineer-
ing and other chemical process operations, including process
development. Rick has also developed projects related to
polymerization and catalytic reaction engineering.
Edgar O'Rear came to OU in 1981 after receiving his PhD
from Rice. A native of the Appalachian hill country of Jasper,
Alabama, Ed explains that in his hometown the term "run-
ner" does not necessarily refer to someone who likes to jog.
Ed has gained valuable experience in Japan as a Visiting Se-
nior Researcher with Hitachi Central Research Laboratory
(1988) and as a Visiting Scientist at RIKEN, the Institute for
Physical and Chemical Research (1992). He was also Pro-
gram Director, Interfacial, Transport and Separations Pro-
gram, National Science Foundation (1993-94). Ed is currently
the President of the International Society of Biorheology. As
the Director of the University of Oklahoma Bioengineering
Center, he has been a tireless driving force behind the bioengi-
neering program at OU. Work in Ed's laboratory involves
polymeric encapsulation of clot-busting drugs or plasmino-
gen activators for heart attack and stroke. Recent results have
elucidated the mechanism of accelerated thrombolysis,
yielding as much as an order of magnitude reduction in
the time for reperfusion with these agents and demonstrat-
ing a unique mode of action for a polymeric drug deliv-
ery system. Ed's other research interests include

admicellar polymerization and novel surfactants.
Jeffrey H. Harwell, a native of Texas, arrived at OU in
1982. He received his PhD at the University of Texas at Aus-
tin (1983). Jeff's research concerns the use surfactants in con-
trolling interfacial properties in engineering systems. His re-
search ranges from environmental remediation (in-situ ground
water remediation and ex-situ soil washing), to polymer com-
posites (modification of silica fillers and reinforcers by
admicellar polymerization), to the use of surfactants in creat-
ing novel adsorbents and catalysts. Applications of this re-
search range from field tests of remediation technologies at
hazardous waste sites, to new tire treads, disk drive lubri-
cants, natural-gas-fueled vehicles, and indoor air decontami-
nation. Jeff presently serves as Executive Associate Dean in
the College of Engineering, but he still finds time to main-
tain an active research program and teach some classes. Jeff
has an interesting, multi-talented background-prior to re-
ceiving his PhD, he received a divinity degree.
John Scamehorn, a native of Nebraska, arrived at OU in
1981. He received his PhD at the University of Texas at Aus-
tin (1980). His industrial experience includes time as a Re-
search Engineer with Conoco (1974-77) and a Research En-
gineer with Shell Development (1980-81). Here at OU, he is
the cofounder and Director of IASR (Institute for Applied
Surfactant Research). Among his many other activities, he is
Chair of the Surfactants and Detergents Division of the Ameri-
can Oil Chemists' Society. A primary area of his surfactant
research is the use of surfactants in novel separation tech-
niques for the cleanup of polluted wastewater, groundwater,
and air streams. One approach he has taken is to dissolve
surfactants in water under conditions in which contaminants
associate with surfactant aggregates, which are easy to ultra-
filter from solution in a subsequent step. Another technology
he has developed is the use of surfactants to regenerate spent
activated carbon beds. In addition, he also investigates the
solution properties of surfactants relevant to improving de-
tergent formulations.

We are now in the 21"t century, and the department is only
a few short years from celebrating its centennial in 2010. Like
many other chemical engineering departments, our depart-
ment has shared in the excitement and turmoil of new areas
such as bioengineering and nanotechnology. We have whole-
heartedly embraced these new areas as natural extensions of
chemical engineering, a discipline that is probably the broad-
est and strongest of all the engineering disciplines. At the
same time, we have not neglected our traditional departmen-
tal-and chemical engineering-strengths in such areas as
transport phenomena, thermodynamics, and reaction engi-
neering. Only by maintaining these strengths can we survive
as a viable discipline whose graduates are sought by a wide
diversity of employers. [

Summer 2004

" educator



of the
University of California, Santa Barbara

Chemical Engineering Professor Michael Doherty says, "The
great thing about being a scholar is that by teaching stu-
dents you really get to learn the material yourself! Research
is incredibly important, but a scholar is different from a researcher-
scholars distill a coherent story about a subject out of their own
research, the literature, and their personal experience. It's in teach-
ing and writing textbooks that the material is really mastered-and
passed along."
Mike Doherty still loves teaching and the opportunities for inter-
action with students that it brings, although teaching at the univer-
sity level gains people's respect but rarely serves as a means to
professional advancement. Mike likens teaching to being an actor
in a play, where an enormous amount of time goes into the prepara-
tion. He's taught dozens of design courses to seniors and industrial
practitioners, yet it's new and challenging every time! And when
he's teaching, it consumes him.
Although Mike can teach many courses in the chemical engi-
neering curriculum, there's only one he feels he can teach better
than most anyone else-the senior design course. Why? Because
while not many people know the coherent body of knowledge that
goes into creating a process flow sheet from scratch, Mike learned
"the touch" at U Mass from Jim Douglas, one of the greatest in the
field. Douglas, during much of the latter half of his career, struggled
successfully to develop a cohesive framework for teaching design.
Many people in chemical engineering believed that design can
only be learned by experience, and that is certainly how the course
has traditionally been presented. As a consequence, design projects
usually left the students feeling deflated right at the time they were
graduating and going out into the world. Douglas, convinced there
was a better way, developed a methodology, and the publication of
his book, Conceptual Design of Chemical Processes, in 1988, revo-
lutionized the approach to this subject. Design can now be taught

On a bet that he wouldn't dare do it, Mike taught
the first class in his junior fluid mechanics
course in Fall 1979 dressed as a Redcoat.

in a systematic way.
Rather than thinking serially about design, i.e. devel-
oping each section of a plant in excruciating detail and
then throwing the developed sections together at the end,
Douglas showed that it's possible to take a hierarchical
approach to the design process. Moreover, the result of
using this methodology is a much more satisfactory
outcome. The procedure a chemical engineer goes
through can be compared to that of an artist painting a
picture, where one starts with the big concept, produc-
ing a series of sketches later to develop the details. For
a chemical process that means starting only with what
comes in and what goes out. Moving downward into
the complexities, one layer at a time, decisions are made
that influence what comes next. Separations systems,
Mike's forte, is, of course, one of the big blocks.
Mike has been teaching the senior design course since
the mid-1980s. The current version, a two-quarter se-
Copyright ChE Division of ASEE 2004

Chemical Engineering Education

quence that has an annual average enrollment
of around 30 students, is nontraditional and, says 1
Mike, "... fun to teach. It doesn't focus on syn- wi
thesis or analysis of processes alone-it also choosE
focuses on decision-making. Plus, undergradu-
ates are the last group of people who will ever
believe you. Seniors are the last chance to make
an impression."
The course shows students that they can ac-
tually invent something from scratch. For the
students it is truly a capstone experience as they
get to decide almost everything about the pro-
cess. Using the hierarchical methodology, they
work in teams of two (any more and some team
members become just passengers), making the
simple decisions first. They can't really fail in
this initial step in the hierarchy of decisions.
Rather, they explore different sets of alterna-
tives. Then, by disentangling the information
between the layers, they learn first-hand about
the trade-offs required in process design, e.g.,
that the benefits to production in using extra ma-
terials are offset by the increased demand placed
upon the separations system. What students get
out of this course they carry with them for life,
whether they remain engineers or choose an-
other career path. They learn how to go about Reseac
making decisions in a logical way, how to look
at a large problem and go about breaking it down to make the
big decisions first. Then, once a set of decisions is made,
they are taught how to evaluate those decisions.
In the business world, chemical engineers must incorpo-
rate more than just good engineering in their designs. Monty
Alger, a chemical engineer at GE Plastics in Pittsfield, MA,
says, "The role of the engineer in industry is to make, evalu-
ate, and justify technical decisions in support of business."
So, in the first two-and-one-half weeks of the senior design
course's second term, the focus turns to a business challenge
problem, originally developed by Alger. To be successful,
the students have to understand the chemistry of the process-
plus, they need to be able to cut manufacturing costs through
good process engineering and to expand the business through
strategic investments in product development and market-
ing. Mike tells his students, "You are not a chemist and you
are not a business major, but you need to be able to take what
a chemist produces and turn it into a viable business. You
cannot do this with engineering alone; you have to be able to
make things happen that are a surprise to both sides. If not,
then you won't be needed as engineers because the company
can do it without you."
The business focus arose after many graduates from good
programs, confident they could design anything and think-
ing they had all the answers, were confronted with business

What students get out of [Mike's] course they carry
th them for life, whether they remain engineers or
e another career path. They learn how to go about
making decisions in a logical way, how to look
at a large problem and go about breaking
it down to make the big decisions first.

:hers like to have fun. Mike (left) and Mike Malone (right).

decisions in the real world and would, according to Alger,
seem to rely on magic, not logic, to solve the problem. Mike
wasn't so sure that his students would be so clueless. So Alger
approached Mike with a proposal in the Fall of 1997: "I'm
going to form a realistic chemical process business problem
and give it to your students at the end of your design course
and see how they do." Alger worked nights and weekends
and finally rolled out a business problem game complete with
an extensive database for providing end-of-year income state-
ments for the many technical-business decisions made by the
players. The students were to prepare and submit a budget
statement (with appropriate technical justification for pro-
cess modifications) that would salvage a loss-making busi-
ness and turn it around to profitability within three budget
cycles. The database ensured that the paths and results de-
pended on the decisions that were made. Sure enough-in-
stead of approaching the problem with the logical methods
they had learned, the students started relying on magic.
Mike was initially devastated. He'd gone out of his way to
teach students to be systematic. Then, when placed in a real-
world situation, they had indeed failed to attack the problem
logically. There was only one solution: incorporate more real-
life business situations into the course.
Mike absolutely believes in the design process he teaches.
He's involved in an off-campus company in which the con-

Summer 2004

ceptual engineering team is able to make technical/economic
decisions so rapidly that they influence the research direc-
tions and goals of the discovery chemists in real time. "We
work faster and better than anybody else in the business,"
he says, because our chemists can focus their research
on solving the problems with the greatest economic and
engineering impact."
A chemical engineer can design a chemical system to work
in a way chemists might not
think about. Using the hierar-
chical methodology developed
by Douglas allows the engineer
to work quickly. Generally the
decisions made at the very be-
ginning, designated "Level I,"
are the most important. And, in
order to make those decisions
logically, it is absolutely essen-
tial that the engineer under-
stand the chemistry of the pro-
cess. Otherwise, the process it-
self will never be understood.
To be successful, it is also nec-
essary to understand the busi- chr of
As co-chair ofFO(
ness side of the equation. Colorado, Mike
Real-life challenges tend to mechc
be bigger than academic ones. Usually, a discovery chemist
making one gram (or smaller) samples is going about the pro-
cess all wrong in terms of commercial production. The chemist
focuses on making high yield of material, not on the more
than 95% selectively necessary to the commercial process.
But the company can't afford tons of waste, so the engineer
needs to be able to get back to the discovery chemist quickly
to carry out experiments under other sets of conditions, as
well as to link directly to the business community.
Just as in the design course, Mike has come to realize that
engineering is more than science. "I'm beginning to preach-
just like Jim!" Mike exclaims wryly.
Ask Mike Doherty about his friends and he'll tell you
they're a bit unconventional. It all started when he went to
Imperial College, University of London, in the Fall of 1970.
When Mike entered Imperial, he stepped into a whole new
world. Excited, yet strangely at home, he never looked back.
The early '70s was an amazing time, especially in London-
the place to be! There were people who had green or brilliant
red or bright yellow hair right next to short-haired, well-coifed
men in pinstripes. And at Imperial, located in the center of
South Kensington (between Knightsbridge and Chelsea), the
dorms circle the campus and the Royal College of Art is just
across the street. Mike loved every bit of it.
Born in 1951 and educated through high school in Manches-
ter (England), Mike had studied hard, done well, and knew


he was interested in studying science. All the better students
at his high school who continued their studies in science went
on to the University of Manchester, but Mike was determined
to escape his provincial confines. London was a giant step
away from life as he knew it, and he embraced it.
Mike's first year at Imperial was, understandably, spent
taking in both the social and academic scene and developing
a sense of himself. He was immediately exposed to other
cultures and socio-economic
classes-his first-year room-
mate was the son of the Brit-
ish High Commissioner to Fiji.
Mike often ate at the Royal
College of Art and made
friends there. His third-year
roommate was an artist.
Mike also explored aca-
demically. He'd entered Impe-
rial as an engineering student,
but in his second year he de-
cided he wanted to try medi-
cine. His Department Chair
talked him out of it, however.
4 in Snowmss, Then in his third year, he con-
D-94 in Snowmass,
first crack at the sidered switching again-to
aI bull. physics-but again the De-
partment Chair dissuaded him.
These yearnings vanished when he started work on his re-
quired independent research project and fell in love with
thermodynamics. Mike, who had never doubted his abil-
ity to do the work, was now excited about it. He was more
than happy to work long and arduous hours and to devote
himself to his project.
Mike and his partner were under the supervision of two
advisors: John Rowlinson, the great molecular thermodynami-
cist, served as the strategist, and Graham Saville was the tac-
tician. The project consisted of computing phase equilibria
from the newly formulated Bender Equation of State, using
twenty constants to represent PVT data. They derived ex-
pressions for the chemical potential and then solved for equal
potential across phases. The brand new mainframe computer
at the students' disposal (a CDC 6600) required that they pro-
gram it in machine language-with a three-feet-long deck of
computer cards that required overnight runs.
Mike was certain he had found his calling and Rowlinson
did his part to ensure Mike's success. In an unusual step at
that time, Rowlinson took Mike, an undergraduate, to a ther-
modynamics research conference, paid his way, and intro-
duced him to many of the participants. As a result, arrange-
ments were made for Mike to pursue his doctoral studies in
chemical engineering at Rice University under the supervi-
sion of Tom Leland, a distinguished statistical mechanician.
In mid-1973 Mike Doherty, Imperial College degree in hand

Chemical Engineering Education

Family Album...

4 Mike (left) and his
long-time friend,
LU AU L Dave Pinardi,
jazz musician
extraordinaire, drove
the entire length of
Route 66 on their way
from Massachusetts to
Santa Barbara (2000)

Mike and Peggy prepare to take a
sunset cruise in December, 2002. V

and his eye on his academic future, set out for the U.S. His
plan was to stop off in New York to see a woman, a Barnard
College student whom he'd met the previous summer, then
to continue on for a vacation in California and Mexico be-
fore reporting in at Rice in August. But, in the words of Rob-
ert Burs, "The best laid schemes o' mice an' men, gang aft
a-gley." Mike, whom no one ever accused of being satis-
fied with the status quo, never made it beyond New York
city on that trip.
By mid-August, he was still in this woman's apartment in
Greenwich Village and was working as a cocktail waiter in a
night club. Although a bit uncertain about his future, he'd
decided not to go to Rice. Then, as fate would have it, Mike's
father called from Manchester to report that Trinity College,
Cambridge, had been pestering the family, asking whether
Mike would be coming in the fall to pursue graduate stud-
ies-and Cambridge didn't start up until early October.
Mike sent a telegraph, "Hold place, am coming," left NYC
on October 1st, and went straight to Cambridge. The first
person Mike happened to see when he arrived at Cambridge
was a fellow who had been a third-year undergraduate stu-
dent at Imperial College when Mike was in his first year there.
John Perkins was incredibly smart, a whiz kid, who had stayed
on at Imperial for his PhD in the field of control, completing
it in two-years. The encounter was a surprised to each of them.
Cambridge University was definitely where Mike wanted
to be, but he was disappointed that no one in the chemical
engineering department was doing statistical mechanics. So
he wandered around wondering what he might do until
Christmas, thinking there would be nothing for him at
Cambridge unless he was willing to work on fluidized

Ar .3urua, -t;gy, uUiu lvilu A 1 Auy
breakfast on Cape Ann, MA.

beds. Then it dawned on him! To stay he needed to de-
velop a project of his own!
Mike had just read a very descriptive book on extractive
distillation that contained enough thermodynamics, he
thought, to use as a starting point for developing a mathemati-
cal model. Moreover, he felt the model should be generic and
global enough to apply to other types of azeotropic distilla-
tions. So Mike went to John Perkins, whose work was highly
mathematical, and said, "You need students and I need an
advisor. Let's develop a theory of azeotropic distillation that
can be applied globally for all mixtures." And John came to
Mike's rescue, saying, "I will be happy to be your advisor
but I'm not going to be your foreman."
So Mike Doherty became John Perkins' first student and
the only one working on separations system design. It took
longer than Mike expected to formulate a useful theory of
extractive and azeotropic distillation, but eventually he found
the right mathematical framework, using geometric methods
from nonlinear differential equations. Their first paper on the
mathematical theory of residue curve maps for azeotropic
distillation has been cited hundreds of times. Once Mike un-
raveled the thread of extractive distillation, he discovered it
was only the tip of the iceberg. Indeed, the main topics of his
highly mathematical dissertation subsequently led to topics
for about a dozen PhD students of his own. Years later this
led to "the best collaboration I could imagine" with his col-
league Mike Malone. During a twenty-year period, theirjoint
group produced about twenty graduates, dozens of papers (two
of which won Best Paper of the Year awards in the journal
Computers and Chemical Engineering, and one of which was
published in the journal Nature), and a recent textbook,

Summer 2004

Conceptual Design of Distillation Systems, published by
McGraw-Hill in 2001.
Perkins proved to be a great advisor. Thinking back, Mike
says, "John was young, very smart, full of confidence, will-
ing to explore unconventional ideas, and then finish the day
over a pint. John can be pretty intimidating and many people
are unable to get close to him, but I was lucky to get to
know him while we were both young. I learned a great
deal about how to advise students from him. He taught
me some great lessons."
Mike's personal life also stabilized somewhat during his
years at Trinity. In 1974, his second year, Peggy, his friend
from New York, came to England to be with him. In looking
for work, she immediately found herself in a bureaucratic
Catch 22. Because she was a foreigner, she needed a work
permit to get a job...but a job was a requirement to obtain a
work permit. After several months and down to their last few
pennies, they were getting desperate, so Mike called his fa-
ther to announce, "We're getting married tomorrow." Not only
did married students make more money and have better
housing, Peggy, as the wife of a British citizen, would be
able to get a work permit.
Mike's father responded, "I'll send you 100 pounds to tide
you over. Come home to Manchester and get married with
all the family around. Your mother and sisters will make
the wedding dress and you can borrow one of my suits."
So they did just that.
When Mike started his doctoral work in 1973, the chemi-
cal industry was in the throes of change. Post-World War II
manufacturers had concentrated on hydrocarbon production.
Processes often employed distillation of ideal mixtures. By
1960 the phase behavior and design of these systems were
very well understood. Beginning in the 1960s, however, the
emphasis within the chemical industry began to shift away
from hydrocarbons and toward chemicals, e.g., specialty
chemicals, polymer precursors, etc. These chemical processes
were more complex, most often involving nonideal mixtures,
the phase behavior of which was not well understood. Thus,
there was little or no predictability with regard to process de-
sign. Separations systems resulted mainly from trial and error.
To devise a systematic way of describing nonideal mix-
tures so they would be susceptible to mathematical analysis,
was nontrivial. A completely new way of thinking was needed
for the design of these distillation processes, especially those
containing azeotropes. Mike found a way to represent the
phase diagram as a set of paths or curves in space employing
a set of differential equations. The resulting set of curves pro-
duced an equivalent residue map for the particular system. In
his doctoral dissertation, he formalized this methodology,
known as Residue Map Analysis, whereby a set of curves
can be developed for each and any phase diagram.
Mike knew what he wanted to do upon completion of his

doctorate-he wanted to be an academic. Moreover, he re-
ally wanted to be in the United States. He had an American
wife and he'd hitchhiked around the States a bit as a college
student. So, while in his third and last year at Cambridge,
Mike started applying for academic positions in the U.S. In
England there had been no pressure to publish, so Mike had
no papers to his name. As a result, no one in the U.S. had
heard of him. Moreover, few had even heard of his young
advisor, Perkins, or of the theory Mike had developed. So,
sadly, Mike received a fistful of rejections and no invitations
to interview for a position.
At Easter time, one of the Cambridge faculty stopped by
Mike's office to tell him that, "Professor Aris is coming to
town next week and wants to take you to lunch." Rutherford
Aris, at the height of his distinguished career, was Head of
the Chemical Engineering Department at Minnesota, which
had already rejected Mike for an advertised job. Mike's
officemate, Rob was studying an Aris paper at the time and
having some difficulty with it. When Aris arrived, Rob jumped
up and exclaimed, "Prof. Aris, I have read your paper and I
was hoping to ask you a few questions about it." Aris re-
sponded, "If you have read that paper recently, you surely
know more about it than I do!"
Aris and Mike went to lunch, and Aris offered Mike "a
nonrenewable position for one year only to teach," i.e., a teach-
ing postdoctoral. Aris was going to CalTech in the fall of
1976 on a fellowship and needed someone to teach his course.
The catch was that Mike had to be in Minnesota by August
1st-with his thesis submitted.
Mike decided he could do it. In order to finish by August,
he had to work seven days a week, from 7 in the morning
until midnight each day. His friends would come around regu-
larly to take Peggy out to the local pub while Mike stayed
home and slogged away.
Mike learned one of the more valuable lessons of his life
during that period of time. He and Peggy had a beautiful sec-
ond-story apartment at Trinity, and Mike had set up a desk
for himself in a big bay window. Both the typed and the hand-
written copies of his almost-complete thesis were on the desk
when one afternoon he and Peggy decided to go downstairs
to a phone box in front of their apartment to make a call.
They opened the phone box door and saw a package with a
clock strapped to it. They immediately closed the phone box
door-as gently as possible-since the IRA bombing cam-
paign in England was at its height at that time, and ran to a
nearby hotel and called the police, who arrived on the scene
within three minutes. Mike and Peggy were instructed to hold
up traffic while the police evacuated the whole block. A ro-
bot was brought in to blow up the suspicious package, and,
fortunately, there were no explosives. But Mike had come
too close to losing all copies of his thesis, and to this day, he
never keeps just one copy of anything. He backs up his work
in multiple places.

Chemical Engineering Education

On August 8, 1976, Mike submitted his thesis ("only a few
days behind schedule"). The next day, he and Peggy leased a
large shipping container, packed up their things, and flew to
New York the next day. After spending a few days with
Peggy's parents, they bought a VW minibus and drove West,
arriving in Minnesota during a mid-August heat wave.
On Mike's first day on the job, he rolled in to the depart-
ment at a respectful hour (at any rate, for a British university
in August), around 11 a.m., and was talking to Aris and some

others when at 11:30 they said, "Let's go to lunch."
Mike, not thinking about the impression he was mak-
ing until later, replied, "I just had breakfast." He was
a fast learner, however-the next day he was at the
office before 8 am.
The working day schedule was not the only way in
which Minnesota differed from England. Mike and
Peggy had also never experienced the kind of heat
they were subjected to throughout August and into
September. What's more, they didn't have the appro-
priate clothes for it. Then, in October, the coldest
weather they had every experienced arrived, and they
had no clothes for it either. At the beginning of No-
vember, Peggy's mother called to ask, "What would
you like for Christmas?" Mike and Peggy weren't shy
about responding, "goose down jackets. .but please
don't wait until Christmas, send them today."
When Mike flew home over Christmas to defend
his thesis and Peggy went to New York to see her
family, they did what they had always done in England
to economize. They turned off the heat in their apart-

of 1984, Mike and Peggy went to Berkeley where Peggy had
a fabulous postdoc position in the Psychology Department
and Mike did a sabbatical. Although Peggy's mentor wanted
her to stay, they returned to U Mass in January 1985. Their
second child, Max, was born that same month.
Mike was certain he would never leave U Mass. He loved
his friends, his colleagues, and his house and land (all sev-
enty acres). And during the time he was Department Head
(1989-1997), he'd helped to build up the department. Most

came to
for one
year of


ment. Upon their return in January, they were greeted with
a frightful mess-the water pipes had burst. Another les-
son learned!
The University of Minnesota turned out to be a wonderful
career break for both Mike and Peggy. The Chemical Engi-
neering Department had many of the best people in the world
and they were wonderful colleagues. Mike learned a great
deal, both through his teaching and through his interactions
with other members of the department. He also wrote his thesis
papers during this period. Meanwhile, Peggy, who had her
undergraduate degree in classical Arabic, talked her way into
Minnesota's top-rated Institute for Child Development and
became a PhD student.
Mike and Peggy left Minnesota in the summer of 1977 as a
two-career family and went to the University of Massachu-
setts at Amherst where Mike was extremely happy. The stu-
dents he attracted were really first class. He had ". a great
run. I can't imagine being able to do better work or publish
better papers or work with a better colleague than Mike
Malone anywhere else in the world. Our students did not leave
much on the table." Mike and Peggy's first child, Sarah, was
born in 1981 and Peggy finished her PhD in 1983. In January

of the young faculty that were hired during this time
won NSF Career Awards; two won Packard fellow-
ships. In 1999, Mike took a six-month sabbatical.
Moreover, U Mass gave Mike one of their Conti
Fellowships-a very special, highly prized, year-
long, and full-salaried fellowships that doesn't in-
terfere with sabbatical (or other) clocks. Mike spent
most of the year working on the textbook. He also
traveled around, giving talks. One of them was given
at UC Santa Barbara where the faculty expressed
an interest in having him join them. After many
months of anguish Mike, made a return visit to
Santa Barbara in February 2000, and this time
Peggy accompanied him. Early one morning
(which happened to be wedding anniversary),
they were sitting at a beachside restaurant enjoy-
ing the view, and Peggy leaned over and said,
"You can leave me here."

S The more Mike visited, the more he appreciated
the people and the work at UCSB. Not only were
the professional aspects enticing, but the location
was physically perfect, so in September 2000, the
family moved to Santa Barbara. Mike, who had already
started to shift his research focus away from distillation, found
that UCSB was perfect for emphasizing his new interest, the
crystallization of organic materials. He believes that in order
to develop a great process, engineers really need to under-
stand the chemistry. UCSB is a world center for materials
and Mike, still very much an engineer interested in the big
picture of process design, is trying to build a bridge between
materials and engineering. At UCSB, the right mix of col-
leagues, courses, and facilities provide the perfect environ-
ment for his interests and his students. Five of the seven people
in Mike's group are currently working in solid-state to ex-
plain the evolution of crystal growth.
In 2004, Mike is happily working in that niche in the de-
sign world which is the bailiwick of chemical engineers.
Peggy is a case manager at Santa Barbara's Devereux Cen-
ter, a nationally known residential facility for the develop-
mentally disabled. Sarah is a Junior at Smith College in
Northhampton, MA, majoring in economics, and Max is a
freshman at UC Berkeley, majoring in political science. To
sum it all up, Mike says, "In 1976 I came to America for one
year. It's been the longest and best year of my life." I

Summer 2004

. M laboratory



University of Illinois at Urbana-Champaign Urbana, IL 61801

while analytical calculations and process simula-
tions1,21' should be a key component in the educa-
tion of a chemical engineer, students gain a deeper
understanding of the nonidealities of industrial processes by
carrying out experiments. Many industrial control problems
are nonlinear and have multiple manipulated and controlled
variables. It is common for models of industrial processes to
have significant uncertainties, strong interactions, and/or non-
minimum phase behavior (i.e., right-half-plane transmission
zeros). Chemical engineering students especially find the
concept of right-half-plane transmission zeros to be more
subtle than other concepts.
We designed a quadruple-tank process that was constructed
to give undergraduate chemical engineers laboratory experi-
ence with key multivariable control concepts (see Fig. 1). By
changing two flow ratios in the apparatus, a range of multi-
variable interactions can be investigated by using only the
one experimental apparatus. Since the spring of 1999, this
quadruple tank process has been used to teach students at the
University of Illinois to
> Understand control limitations due to interactions, model
uncertainties, non-minimum phase behavior, and unpredict-
able time variations
Design decentralized (often called "multiloop") controllers,
and understand their limitations
0 Implement decouplers to reduce the effect of interactions,
and understand their limitations
Implement a fully multivariable control system
Select the best control structure, based on the characteris-
tics of the multivariable process
The quadruple-tank apparatus is a variation on an appara-
tus described in the literature 31 where we introduced a time-
varying interaction between the tanks. This time-varying char-
acteristic is caused by an irregularity in the fluid mechanics
of splitting the stream into the upper and lower tanks, which
results from the capillary effect of the tubing and dynamics

of the multiphase flow of liquid and air in the tubing. The
consequence of combining these factors is an enhanced sen-
sitivity and stochasticity of the flow ratio to manipulated vari-
able movements. The apparatus can exhibit a time-varying
qualitative change in its dynamics, between conditions that
are controllable to those that are uncontrollable. Although
this uncontrollability issue has been reported as a major is-
sue in large-scale industrial processes,14' this appears to be
the first educational laboratory experiment designed to clearly
illustrate it and its effects on the control system.
The apparatus is small (1 ft x 1 ft x 6 in, not counting com-
puter equipment) and is designed so the students, teaching
assistants, and instructor can determine at a glance if the stu-
dents are controlling the apparatus successfully. The small
size enables experimental data to be collected rapidly and
keeps the cost low. The apparatus is designed to be self-con-
tained (that is, there are no requirements for continual access
to water, steam, vacuum, or gas) and is environmentally
friendly-the only chemical used is ordinary tap water, which
is recycled during the experiments.
Past studies with 4-tank apparatuses implemented decen-
tralized PI control,'31 multivariable H control,[3' multivari-
able internal model control,5' and dynamic matrix control.'51

Siong Ang received his BS in chemical engineering from the University of
Illinois under a Singapore Armed Forces Overseas Merit Scholarship and
his MS from Stanford University. As part of his undergraduate thesis, he
constructed quadruple-tank-process experiments and wrote the visual-
programming control interface currently used in the undergraduate pro-
cess control laboratory.
Effendi Rusli received his BS from the University of Wisconsin and his
MS from the University of Illinois at Urbana-Champaign, both in chemical
engineering. He is currently completing his PhD thesis on the analysis
and control of systems described by multiscale simulation codes, with the
main application being the electrodeposition of copper films and wires in
microelectronic devices.
Richard Braatz received his BS from Oregon State University and his MS
and PhD from the California Institute of Technology. After a postdoctoral
year at DuPont, he joined the faculty at the University of Illinois, where he
is a Professor and University Scholar. His main research interest is in
multiscale systems theory and its application. He has published three
books, one of which is a textbook on fault detection and diagnosis.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

The main educational focus of Ref. 3 was providing an appa-
ratus with highly idealized and reproducible dynamics for
use in illustrating multivariable interactions and multivari-
able transmission zeros. The main educational focus of Ref.
5 was to provide hands-on experience for students in imple-
menting advanced control algorithms. In contrast, our main
educationalfocus is to aid students in understanding the ad-
vantages and disadvantages of the different control struc-
tures (e.g., decentralized, decoupling, multivariable) when
applied to a multivariable process with interactions and dy-
namics ranging from highly ideal to highly nonideal.
First, the construction of the apparatus will be described
here in enough detail for duplication. Enough information
will be provided for a technician or student to construct the
control apparatus and for an instructor (who may not be an
expert in control) to see how to use the experiment in the
laboratory. This will be followed by motivation and back-
ground on the modeling and control for the apparatus.
Some experimental results obtained by two students will
also be presented to show how the apparatus illustrates
some key control principles that are not addressed by past
control experiments.

A table of all equipment needed to construct the apparatus,
including costs, can be obtained from the website at>. Four cylindrical tanks are mounted ver-
tically on an acrylic board and are arranged in a symmetric 2
x 2 fashion, as shown in Figure 1. A small hole is drilled at
the bottom of each tank to channel the water from each to a

Pump 1 Pump 2
P Tank 1 Tank 1[ 2

Safety Overflow (only used if leaks or splashing occurs)

Figure 1. Schematic of the quadruple-tank process appa-
ratus. To simplify the figure, not shown are a tube between
Tank 3 and Tank 1, a tube between Tank 4 and Tank 2, a
tube from Tank 1 to the water reservoir, a tube between
Tank 2 and the water reservoir, and a lid on the water res-
ervoir. These tubes and lid are used to reduce evaporation.

differential pressure sensor via a 3/16-in tubing.
MASTERFLEX tubings transport water between the tanks.
Taking into account the maximum capacity of all four tanks
(750 ml) and the dead volume inside the entire length of the
tubings, a 1000-ml cylindrical beaker is enough to store and
recycle water for the experiments. Two MASTERFLEX volu-
metric pumps are used. A 5-gallon tank immediately below
the apparatus contains any spillage or splashing from over-
flow in any of the four tanks.
A Y-junction is used to divide the flow such that water is
channeled to a bottom-level tank and the upper-level tank
diagonal to it. This arrangement makes both levels in the
bottom two tanks a function of both pump-flow rates. By
adjusting the valve knob, the process can be operated so that
one of the multivariable transition zeros is in the right-half
plane, the left-half plane, or switches between the two planes
in a stochastic time-varying manner.
The low cross-sectional area of the tanks makes level varia-
tions easy to see with the naked eye. Hence, students, teach-
ing assistants, and instructors can assess the performance of
the closed-loop system with a glance. The tank heights are
small, so the closed-loop controllers that perform poorly lead
to overflow in the tanks, which is an indication that the con-
trol system needs either better tuning or an alternative con-
trol structure, or the interactions need to be changed to make
the process more controllable.
The visual programming control interface used in the labo-
ratory161 was modified for use with this apparatus. It enables
students with a minimum background in computer program-
ming to make changes in the control structures and is avail-
able for download at a web site.17 Readers can find more
details in the references.

There are several advantages to including a quadruple-tank
process experiment in an undergraduate chemical engineer-
ing laboratory. One is that the experiment can demonstrate a
range of interactions from slight to very strong. The appara-
tus allows students to investigate the extent to which a de-
centralized controller is capable of controlling the process as
the interactions increase. They can also implement partial or
full decoupling as a first step to reduce process interactions.
This enables students to obtain hands-on experience in how
decoupling can improve the closed-loop performance in some
situations (when there are some interactions, but not too
strong), while having significant limitations when the inter-
actions become sufficiently strong.[8'
The quadruple-tank dynamics have an adjustable multivari-
able transition zero, whose position can be in the left- or the
right-half plane, depending on the ratio of flow rates between
the tanks. This enables students to investigate performance
limitations due to right-half-plane transmission zeros. For the
particular quadruple-tank apparatus at the University of Illi-

Summer 2004

nois, under certain conditions the transmission zero can move
between the left- and right-half planes, with varying levels of
stochasticity depending on the operating condition. This leads
to some interesting time-varying dynamics.


The experimental apparatus is used to teach important prin-
ciples of process control while familiarizing chemical engi-
neers with control structures used in industry. In the labora-
tory reports, students describe each learned control principle
in words and illustrate the principle for the quadruple-tank
process by first-principles modeling, applying control theory
learned in lecture, and experimental verification. This draws
a close connection between what the students learn in the
lecture and what they practice in the laboratory.
The material balance equations using common assumptions
and the transfer function matrix obtained by linearizing and
taking deviation variables are given in Figure 2. The second-
order transfer functions correspond to the contributions to
the bottom two tanks by the upper two tanks. The linearized
system G(s) in Figure 2 has two multivariable transmission
zeros, which are determined by the zeros of its determinant

detG(s)= clc212 +s ( T3)(l+ST4)1- 1-72)
u H 4_(1+sTi)1 712

It is important to determine the location of these zeros in the
complex plane since right-half-plane zeros limit the closed-
loop performance achievable by any control system.'891 For
the sake of convenience, the parameter is introduced as

(1- )(1- 72) (2)

where 9 e (0, -). Since the numerator of Eq. (1) is a quadratic,
the zeros can be computed analytically

-(T3+T4) + (T3- T4)2+4T3T4 (3)
zl,2(4)= (3)

Given T3 > 0 and T4 > 0, the function in Eq. (3) is differen-
tiable for le (0, O ).

Z,2()= + (4)
I(T3 -T4)2 +4T3T4T1

The derivatives exist for all conditions where T3, T4. When
r = 0, the zeros are z, = -1/T3 and z2 = 1T4. As q approaches
0, it is straightforward to deduce from Eq. (3) that

Z -* l/T3T4 and z2->- TI/T3T

Because the derivative functions in Eq. (4) are monotonic, z,
is strictly increasing and z2 is strictly decreasing. This im-

plies that the transmission zero z will cross from the left-
half plane to the right-half-plane with increasing "q. The
crossing occurs at 1q = 1. With a little algebra, these re-
sults can be written in terms of the flow ratios y, and y,,
as shown in Table 1.
The process is minimum phase when the total flow to the
lower tanks is greater than the total flow to the upper tanks (1
< "Y + Y2 < 2). The process is non-minimum phase (that is,
has a multivariable transmission zero in the right-half plane)
when the total flow to the lower tanks is smaller than the
total flow to the upper tanks. For operating conditions where
the total flow to the upper tanks is nearly the same as the total
flow to the lower tanks, small variations in the flows due to
irregular behavior in the tubing can cause the transmission
zero to move between the two half planes in an irregular
manner, in which case the process becomes uncontrollable.[9'

a, -2,+, a, 2_ j + y,,
2gh, + V2gh,
A, A, A,
a 2gh, +'- 2gh(+Yk
A, A, A,

a,4 2 A+ ( I- )k, v ,
A4 A4
A, A,

cross-section of Tank i
cross-section of the outlet hole
water level in Tank i
steady-state value for the water level in Tank I
the fraction of water flowing to Tank i from Pump i
voltage applied to Pump i
flow from Pump i
fraction of flow going to Tank i from Pump i
acceleration due to gravity

( iC, (-, r )c,
G(s) ( ,)c, y,)c,
(I+sT4)(l+sT,) l+sT,


A, 2h,
a, g

c, T,k, /A,
c = T,k, / A
Figure 2. Physical models for quadruple-tank process.13/

Location of Zeros on the Linearized System as a Function
of the Flow Ratios y, and y,

z, Z2
1 <.y + y 2 negative negative minimum phase
y,+ y2 = 1 zero negative boundary
0 < ', + 7' < I positive negative nonminimum phase

Chemical Engineering Education

i =1, .*,4.

More precisely, the steady-state determinant of the transfer
function [G(0) in Figure 2] switches sign when y, + y, crosses
1, indicating that it is impossible to control the process with a
linear time invariant feedback controller with integral action.191
This is a generalization of the single-loop notion that the sign
of the steady-state gain must be either consistently positive
or consistently negative for the process to be controllable with
a linear feedback controller with integral action.

Students have applied decentralized control, decoupling
control, and fully multivariable control on the same appara-
tus. They compare different multivariable control structures
and judge for themselves the most effective method to con-
trol the apparatus. It is important for the students to realize
that the same structure can perform very differently and they
will face a new set of limitations when conditions change.
This is especially relevant to this particular apparatus in which
under some conditions the process becomes uncontrollable
during movement of the transmission zero across the imagi-
nary axis within a single setpoint or disturbance response.

After investigating decentralized control, decoupling is
implemented as the first step to deal with loop interactions.
Students verify the improvement/deterioration in closed-loop

E (1,1)
0 500 1000
time (s)
r (2,1)
0 10


0 500 1000 1500
time (s)


0 500 1000
time (s)

E (2,2)


0 500
time (s)

E (1,1)


0 500 1000 1500
time (s)
S (2,1)


0 500 1000
time (s)
S (1,2)


-10 -
a s00 1000
time (s)
6 (2,2)

1000 0 500 1000 1500
time (s)

Figure 3. Model predictions plotted with experimental data
for the four elements of the transfer function matrix: data
used to fit model (left), data used to verify model (right).
The row and column numbers are reported in the upper-
left-hand corner.

performance caused by the implementation of decoupling.
The details of the implementation vary depending on the type
of decoupling (steady-state, dynamic, partial, full), but all of
these are easy to implement using the control interface.161 Stu-
dents also investigate the effects of model uncertainties, which
are especially important for this apparatus.


The identification experiment is an ascending series of steps
followed by subsequent descending steps for identification,
which is a better input for characterizing process
nonlinearities. The use of ascending and descending step in-
puts has the educational benefit of permitting visual moni-
toring of the change in the dynamics with different level of
operating regime and checking of the reproducibility of the
process response.

To ease them into the process, students are prompted to
first operate the apparatus so that most of the flow goes to the
bottom two tanks. The first step is to determine the transfer
function matrix for the experimental apparatus for compari-
son to the theoretical model. Various student teams have fit
first-order-plus-time-delay, state space, and ARMAX mod-
els to experimental data. For brevity, only transfer functions
determined using the program ms2th,[l01 which is a MATLAB
built-in identification subroutine, to compute the least-squares
estimate of both discrete and continuous model parameters,
are reported here for one operating condition:

y(s) = G(s)u(s) + H(s)e(s) (5)

121.4s +1
G(s) =
(84.73s + 1)(3.109s +


2.074s +1

8.75s2 + 5.75s +
(3.109s + 1)(84.73s + 1)


84.73s + 1

1 (1.410s + 1)(7.970s +1)
(121.4s + 1)(3.967s +1)

1 2.725s+1
84.73s +1


where u(s) is the vector of voltage signals from the two pumps,
y(s) is the vector of heights of Tanks 1 and 2, and H(s) has
been normalized so that the noise signal e(s) is uncorrelated
with unit variance. Comparing the theoretical transfer func-
tion model in Figure 2 with Eq. 7 gives the nominal esti-
mates of the physical parameters

Yi = 0.63 72 = 0.64 cl = 18.94 c2 =18.10

which would indicate that both transmission zeros are in the
left-half plane (see Table 1). Figure 3 compares the predic-

Summer 2004

tions of the model (6) and experimental data. There is some
variation in the gains, which agrees with an experimental
observation that the flow ratios vary depending on the oper-
ating conditions and that the gains are a function of the flow
ratios (see Figure 2). Using basic statistics,11t the 95% confi-
dence intervals for the flow ratios are 0.48 < y, < 0.79 and
0.49 < y2 <0.80, which suggest that the transmission zero
may move into the right-half plane under some operating
conditions. This has serious implications on feedback con-
troller design, as seen below.

Full multivariable control such as model predictive control
can be implemented that manipulates the signals to the pumps
to control levels in the bottom two tanks.5'1 While such mul-
tivariable controllers are being increasingly implemented in
industry, other types of controllers have been applied in the
chemical industries, such as
Decentralized control: a noninteracting controller with
single-loop controllers designedfor each tank. Control loop
1 manipulates theflow through Pump 1 (via a voltage signal)
to control the height of Tank 1, while Control loop 2
manipulates the flow through Pump 2 to control the height of
Tank 2.
Partial decouplingfollowed by decentralized control.
Full decouplingfollowed by decentralized control.
Various students have implemented these control strategies
on the quadruple-tank process apparatus during the past five
years. Students implement up to three control strategies in a 7-
week period, where the scheduled lab time is 3 hours per week
and the lab report requirements include first-principles model-
ing, analysis, and comparison between theory and experiment.
The relative gain for the nominal plant (6) is 1.5, which
indicates that Pump i should be paired to the level in Tank I
(8). Decentralized Internal Model Control Proportional-Inte-
gral (ICM-PI) controllers are tuned to trade off robustness
with performance'8,9,12 (see Table 2 and Figures 4 and 5 for
two levels of tuning). Due to model uncertainties, the differ-
ences between model predictions and experiments are large
when the IMC-PI controllers are tuning too aggressively.
Implementing a partial dynamic decoupler and multiply-
ing by the transfer function matrix in Figure 2 gives


Y7c, 0

(1-y )c2 Y2C20
(1 + sT4)( + sT2) ( +sT2)( + sT3)( + sT4)

S=(l+sT3)(1+ST4 )- 1 Y2
Y iY2

Proportional Gains, K, and Integral Time Constants, T1 for
Decentralized Controllers with Aggressive and
Sluggish Tuning

Aggressive C, 1 121.4
C2 1 84.73
Sluggish C, 0.378 121.4
C2 0.395 84.73



M 6

0 500 1000 1500
time (s)


T 10

4 L
0 500 1000 1500
time (s)

f J

0 500 1000 1500 2000 2500 3000 3500 4000
time (s)


4 -5- --

0 500 1000 1500 2000 2500 3000 3500 4000
time (s)
Figure 5. Decentralized controller with sluggish tuning:
setpoint (dashed line), experiment (dots), model prediction
(solid line).

Chemical Engineering Education

0 500 1000
time (s)
Figure 6. Partial decoupling with aggressive tuning: set-
point (dashed line), experiment (dots), model prediction
(solid line).

10 10

0 500 1000 1500 2000 2500 3000 3500 4000
time (s)

0 500 1000 1500 2000 2500 3000 3500 4000
time (s)
Figure 7. Partial decoupling with sluggish tuning: setpoint
(dashed line), experiment (dots), model prediction (solid

Proportional Gains, K, and Integral Time Constants, T,
for Partially Decoupled Controllers with Aggressive and
Sluggish Tuning

K 1
Aggressive C, 1 121.4
C, 1 84.73
Sluggish C, 0.3 121.4
C, 0.3 84.73

(1+sT)(l+sT3 )( +sT4


(1+sT2)(1+sT3)(1 +sT4)




I 8



E 10


4 -

Summer 2004

The transmission zeros are values of s in which 0 = 0 (see
Eq. 5). Both transmission zeros appear in the second control
loop. This results in a degradation of the closed-loop perfor-
mance for the second control loop when one of these trans-
mission zeros is in the right-half plane. Using model (6), the
results of tuning IMC-PI controllers to trade off robustness
with performance are shown in Table 3 and Figures 6 and 7.
Both aggressive and sluggish tuning shows some interactions
between the control loops, due to plant/model mismatch. The
differences between the model predictions and experimental
data are larger for the aggressive tuning.

While full dynamic decoupling is not common industrial
practice, for educational purposes it is useful to compare full
dynamic decoupling with partial dynamic decoupling to il-
lustrate how full decoupling can lead to worse closed-loop
performance than partial decoupling. Implementing a full
dynamic decoupler and multiplying by the transfer function
matrix in Figure 2 gives

The transmission zeros appear in both control loops. When
one transmission zero is in the right-half plane, its effect on
both loops implies that the closed-loop performance can be
worse for full decoupling than for partial decoupling, since
the right-half-plane transmission zero will affect both con-
trol loops. Using model (6), the model predictions and ex-
perimental data using the IMC-PI controllers in Table 3 are
shown in Figures 8 and 9 (next page). The closed-loop re-
sponses with full decoupling are much worse than for the
decentralized or partial decoupling controllers. In most cases
when stepping up the setpoint, there appears to be inverse
response exhibited in both control loops, suggesting that the
closed-loop system is stable but a transmission zero has moved
into the right-half plane. That interpretation would be incor-
rect. While it is correct that a transmission zero moves into
the right-half plane when the setpoint is stepped up, the closed-
loop system becomes locally unstable when this occurs. This
is because the steady-state gains in (9) change sign, switch-
ing the controllers from negative to positive feedback. This
is a common issue in large-scale industrial systems, which
can be masked when physical constraints are present.[413-14]

To see the change in sign, consider the entry (2,2) in the
transfer function matrices (8) and (9)

2C2(S) (10)
g (s)= (10)
(1+ sT2 )(1+ sT3 )(1 + ST4

Since y c, > 0, the sign of the steady-state gain of g,(s) is

500 1000
time (s)

equal to the sign of 0(0) = l-(l-Y1)(l-y2)/y 1y2, which changes
sign when the process operating condition switches from
minimum phase to non-minimum phase and vice versa (that
is, when y, + y2 crosses 1).
This local instability causes the initial decrease in tank lev-
els. Decreasing the tank levels changes the relative magni-
tudes of the total flow rates between the top two tanks and
the bottom two tanks, causing the right-half-plane transmis-
sion zero to move back into the left-half plane, the steady-
state gain to change back to its original sign, and the closed-
loop poles to move back into the left-half plane. The closed-
loop system stabilizes, causing the tank levels to increase back
towards the desired setpoints. This switch from closed-loop
stability to instability and back to stability is why the initial
decreases followed by increases in the tank levels are sharper
than expected for a smooth system consisting of only low-
order processes. There is no apparent "inverse response" in
either control loops when stepping down the setpoint. While
hysteresis is common in industrial process units such as
valves, the case here is more interesting because it involves
the movement of a transmission zero between the left- and
right-half planes and a change in sign of the steady-state gains,
resulting in very poor closed-loop performance obtained for
a linear controller. (Although the essence of the argument is
valid, for the student's sake this interpretation involves some
simplification, since the real system is nonlinear.)
For this particular valve knob setting, the full decoupling
controller induces this behavior more readily than the decen-
tralized or partial decoupling controller. This illustrates the
important point that when interactions are large enough,
decoupling control can do more harm than good.['91 Full
decoupling control has increased sensitivity to uncertainties
in the transfer function model, which causes the ratios of the
total flow rates in the bottom tanks and top tanks to vary
more than for the other controllers. If the valve knob is shifted
so that the transmission zero easily moves between the right-
and left-half planes for the whole operating range (instead of
only for some conditions, as in Figures 8 and 9), then good
setpoint tracking is unobtainable by a linear controller, no
matter how sophisticated.191
The second important point is that hysteresis effects are
common in industrial control loops and should be considered
when troubleshooting. The third point is that the cause of
unexpected dynamic behavior in control loops is often more
subtle than what is often first assumed. But such phenomena
can be understood with some thinking and judicious applica-
tion of undergraduate-level process control analysis tools. This
understanding is needed to determine whether a particular
control problem can be resolved by better controller tuning,
a different control structure, by changing the process design,
or by changing the operating conditions.
For the next experiment, the quadruple-tank process was
made more interactive by using the Y-junctions to increase

the proportion of flow to the top tanks. Closed-loop responses
with decentralized control are shown in Figure 10. Due to
the higher interactions, as well as some nonlinear effects, the
closed-loop responses were highly oscillatory around the
setpoints. The student was unable to obtain controller tuning
parameters that would stabilize the closed-loop system when
either steady-state or dynamic decoupling was used. The best
closed-loop response obtained by dynamic decoupling is
shown in Figure 11. The initial closed-loop performance was
acceptable up to 200 s, but the level in Tank 2 deviated from

0 500 1000 1500 2000 2500
time (s)
Figure 8. Full decoupling with aggressive tuning: setpoint
(dashed line), experiment (dots), model prediction (solid


10 I-- It:



0 500 1000 1500 2000 2500 3000
time (s)




0 500 1000 1500 2000 2500 3000
time (s)
Figure 9. Full decoupling with sluggish tuning: setpoint
(dashed line), experiment (dots), model prediction (solid

Chemical Engineering Education





0 500 1000 1500
time (s)

N /

6 /

2000 2500

the setpoint for t > 400 s, indicating that the closed-loop system was not
locally asymptotically stable. In addition, there was a consistent steady-state
offset exhibited by the level in Tank 1. Again, this illustrates to students that
a process that is designed poorly can be difficult or impossible to control.
Different student teams are given different valve settings in the Y-junc-
tions, and students are encouraged to share their results with other teams.
Students who consistently have >80% of the flow going to the bottom tanks
observe that decoupling control can provide better closed-loop performance
than multiloop control. Decoupling control performs worse than decentral-
ized control when the interactions are increased. When the total flows to the
top and bottom tanks are equal or nearly equal, no linear controller can pro-
vide acceptable closed-loop performance.


A 4-tank apparatus was introduced in which a multivariate transmission
zero can cross the imaginary axis during a single closed-loop response, which
is used to illustrate the effects of time-varying dynamics, changes in the sign
of the steady-state gain, and hysteresis. Example student results illustrated
how the apparatus is used to teach many important points that are ignored in
most process control lectures and laboratories: 1) the effect of time-varying
dynamics should be considered when designing control systems; 2) the sign
of the steady-state gain should always be considered when designing control
systems for multivariable processes; 3) the cause of unexpected dynamic
behavior in control loops is often more subtle than what is first assumed; 4)
under some conditions, full decoupling can lead to significantly worse per-
formance than partial decoupling; 5) decoupling control can do more harm
than good; 6) hysteresis effects should be considered when troubleshooting
control problems. This level of understanding is needed for students to select

3.5 101
Tank2 9
2.5 > 7 controller 1
V 2 Tank 1. 6I
S1.5 i $

0 0 100 200 300 400 500 600
0 100 200 300 400 500 600
time (s) time (s)

Figure 10. Responses to decentralized control with setpoint heights of 3
inches in a strongly interacting system.

3 10 -
Se ,or.I 9 -
S 8.,, controller 1
E. 2 7 1
Q, 6

: ; "Tank 2 5 41

0.51 0 2
0 0 20 0 60 80 00 1 controller
0 200 400 600 800 1000 0 200 400 600 800 100(
time (s) time (s)

Figure 11. Responses to dynamic decoupling control with setpoint
heights of 3 inches in a strongly interacting system.

Summer 2004

the proper multivariable control structure and
to determine whether a particular control prob-
lem can be addressed by better controller tun-
ing, by a different control structure, by chang-
ing the process design, or by changing the op-
erating conditions.
Although not reported here, the apparatus has
been used to implement partial and full steady-
state decoupling, to compare with dynamic
decoupling. Also, it would be educationally
valuable to investigate the development of feed-
back linearizing controllers to enable a single
controller to provide good performance for a
wider range of operating conditions.1I5


The University of Illinois IBHE program is
acknowledged for support of this project.

1. Cooper, D.J., "Picles: A Simulator for 'Virtual World'
Education and Training in Process Dynamics and
Control," Comp. Appl. Eng. Ed., 4, 207 (1996)
2. Doyle III, F.J., E.P. Gatzke, and R.S. Parker, Process
Control Modules: A Software Laboratory for Con-
trol Design, Prentice Hall, Upper Saddle River, NJ
3. Johansson, K.H., "The Quadruple-Tank Process: A
Multivariable Laboratory Process with an Adjustable
Zero," IEEE Trans. Cont. Sys. Tech., 8, 456 (2000)
4. Featherstone, A.P., and R.D. Braatz, "Integrated Ro-
bust Identification and Control of Large Scale Pro-
cesses," Ind. Eng. Chem. Res., 37, 97 (1998)
5. Gatzke, E.P., E.S. Meadows, C. Wang, and F.J. Doyle
III, "Model Based Control of a Four-Tank System,"
Comp. Chem. Eng., 24, 1503 (2000)
6. Braatz, R.D., and M.R. Johnson, "Process Control
Laboratory Education Using a Graphical Operator
Interface," Comp. Appl. Eng. Ed., 6, 151 (1998)
8. Ogunnaike, B.A., and W.H. Ray, Process Dynamics,
Modeling, and Control, Oxford University Press,
New York, NY (1994)
9. Morari, M., and E. Zafiriou, Robust Process Con-
trol, Prentice Hall, Englewood Cliffs, NJ (1989)
10. Ljung, L., System Identification Toolbox: User's
Guide, The Mathworks Inc., Natick, MA (1995)
11. Ljung, L., System Identification: Theory for the User
Prentice Hall, Englewood Cliffs, NJ (1987)
12. Braatz, R.D., "Internal Model Control," in Control
Systems Fundamentals, W.S. Levine, ed., CRC Press,
Boca Raton, FL, pp. 215-224 (2000)
13. Russell, E.L., and R.D. Braatz, "The Average-Case
Identifiability and Controlability of Large Scale Sys-
tems," J. Proc. Cont., 12, 823 (2002)
14. Nunes, G., S. Kincal, and O.D. Crisalle, "Stability
Analysis of Multivariable Predictive Control: A Poly-
nomial Approach," Proc. Am. Cont. Conf., 3, 2424
15. Ogunnaike, B.A., "Controller Design for Nonlinear
Process Systems via Variable Transformation," Ind.
Eng. Chem. Proc. Des. Dev., 25, 241 (1986) C

r, curriculum


To Introduce ChE Principles

Rowan University Glassboro, NJ 08028

Rowan's two-semester freshman clinic sequence in-
troduces all freshman engineering students to engi-
neering in a hands-on, active learning environment.
Engineering measurements and reverse engineering methods
are common threads that tie together the different engineer-
ing disciplines. Previous reverse engineering projects have
involved common household products such as automatic cof-
fee makers,31' hair dryers, and electric toothbrushes.[41
Recently, the human body was added to the repertoire of
familiar machines to be reverse engineered. In a semester-
long project, freshman engineering students explore the in-
teracting systems of the human body. They discover the func-
tion, interaction, and response to changing demands of vari-
ous systems in the human body-the respiratory, metabolic,
cardiovascular, electrical, and musculoskeletal systems.
This paper describes a laboratory experiment in which stu-
dents are introduced to engineering measurements and cal-
culations, mass balances, and process simulation through their
application to the respiratory system. The experiment and
module are appropriate for a freshman engineering course or
a sophomore material balances course. A subsequent, related
experiment introduces students to the chemical reactions in-
volved in the oxidation of foods and concepts associated with
energy balances, but these concepts are not addressed here.
According to Webster,51 inspiration is the "action orpower
of moving the intellect or emotions"-something all profes-
sors strive to do in the classroom. A second definition is also
given: "the act of drawing air into the lungs"-something
we all do and a physiologic process with which we are all
familiar. It is the familiarity of the physiologic process of
breathing that represents its primary appeal as a framework
for teaching engineering principles.
In a hands-on experiment, students measure physiologic
variables such as breathing flow rate and respiratory gas com-
positions both at rest and during moderate exercise on an ex-
ercise bicycle ergometer. Using their data, students perform

mass balances to determine the rates of oxygen consump-
tion, carbon dioxide production, and water loss. They use a
psychrometric chart to obtain water content for inhaled and
exhaled air and compare those results to calculated values.
Finally, the students create a process flow diagram using a
HYSYS161 process simulator and perform mass balance cal-
culations on the lungs.
This experiment and the associated course content are used
in the Freshman Engineering Clinic. Our goal in this course
is to give students a first exposure to real engineering mea-
surements, principles, and calculations, and to provide moti-
vation for future in-depth study of mass balances during the
sophomore year. The experiment and module could also be
used very effectively in a sophomore-level material balances
course where mastery of the same engineering principles is

Stephanie Farrell is Associate Professor of
Chemical Engineering at Rowan University. She
received her BS in 1986 from the University of
Pennsylvania, her MS in 1992 from Stevens
Institute of Technology, and her PhD in 1996
from New Jersey Institute of Technology. Her
teaching and research interests are in controlled
drug delivery and biomedical engineering.

Robert Hesketh is Professor of Chemical Engi-
neering at Rowan University. He received his BS
in 1982 from the University of Illinois and his PhD
from the University of Delaware in 1987. His re-
search is in the areas of reaction engineering,
novel separations, and green engineering.

Mariano Savelski is Assistant Professor of
Chemical Engineering at Rowan University. He
received his BS in 1991 from the University of
Buenos Aires, his ME in 1994 from the Univer-
sity of Tulsa, and his PhD in 1999 from the Uni-
versity of Oklahoma. His technical research is in
the area of process design and optimization.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

required. The objectives of the module are
To perform measurements of gas concentra-
tion and flow rate during breathing to
To perform mass balances on the lungs rever
To represent the process in terms of the
relevant unit operations, and to prepare a fi
simple process flow diagram using a process
To use a process simulator to perform mass
balances on the lungs
To use a psychrometric chart to estimate the rate of
water loss during respiration.
The module comprises two 3-hour laboratory periods and
three 50-minute classes of combined lecture and cooperative
learning exercises. One of the laboratory periods is used to
perform the respiration experiment, and the second labora-
tory period is used for the HYSYS simulation exercise. Each
3-hour laboratory period is sufficient to afford a brief (45-
minute) introduction or wrap-up class. The three 50-minute
classes are used to introduce the relevant engineering con-
cepts and to perform example calculations using student data
in a cooperative learning environment. It should be empha-
sized that the module is taught inductively-students begin
by making experimental observations and afterwards (with
structured guidance) "discover" the underlying engineering
principles that explain their observations.

The air we inspire (inhale) is approximately 21% 02 and
79% N2 on a dry basis, while the expired (exhaled) gas from
the lungs contains approximately 75% N2, 16% 02, 4% CO2,
and 5% H20.17'8] The inspired air is at ambient pressure, tem-
perature, and humidity, while the expired air is saturated at
body temperature and ambient pressure. The lungs serve as a
mass transfer device that allows rapid and efficient exchange
of 0O, CO2, and H20.
A flow diagram for the breathing process is shown in Fig-
ure 1. Streams information shows the measured variables,

T = 20C T= 37C
P-759 mm Hg P=759 mm Hg
47% R.H. Saturated with water
Inhaled Lungs Exhaled
air Lungs air
V" = ? V"' =13.08 L/min
y O'= 0.21 Yo"' =0.185
y =0.79 d =?
Y2 0.0 Exchange with body Yco2 = 0.023
Po =?
Vco2 = ?
SH20 =?
Figure 1. Flow chart showing the measured variables in
the respiration experiment. The values given represent the
resting data for a 19-year-old female subject (125 lb, 66 in).

S. the human body was added
the repertoire of familiar machines to be
se engineered. In a semester-long project,
freshman engineering students explore the
interacting systems of the human body.

with the values given being those for a 19-year-old, 125-
pound, 66-inches-tall female student. The ambient tempera-
ture, pressure, and relative humidity are recorded; inlet gas
compositions are assumed to be 21% oxygen and 79% nitro-
gen on a dry basis. The experimentally obtained data for flow-
rate of exhaled breath ( out,B'PS) is reported at BTPS (body
temperature and pressure, saturated) conditions, while the gas
compositions ( y" and yo2 ) are reported on a dry basis. The
ideal gas law and concepts of relative humidity, Raoult's law,
and dry/wet bases are therefore employed in the mass bal-
ance calculations on this multicomponent system.
To illustrate the concepts that are applied through the cal-
culations associated with this experiment, the basic solu-
tion procedure for the calculation of oxygen consump-
tion and carbon dioxide production rates, and the rate of
water loss is outlined below.
For ease of calculation, the volumetric flow rate of exhaled
air obtained experimentally is first converted to a molar flow
rate (nout) using the ideal gas law,
pVoutBTPS = noutRT (1)

The body temperature, Tb, is 37C (310K).
The total flow rate of saturated air at body temperature and
barometric pressure are next converted to the flow rate of dry
air (at the same temperature and pressure). The partial pres-
sure of water in saturated air at 370C is 47 mm Hg. The mole
fraction of water in the exhaled air is therefore 47 mm Hg/P,
and the molar flowrate of exhaled air on a dry basis is

nout,dry ftout(l 47 mn Hg (2)
P ()

Next, the fraction of nitrogen in the exhaled air is determined
on a dry basis
out,dry l out,dry +youtdr (3)
YN2 0, 1 "CO2
The flow rate of each species in the exhaled breath is next

out out,dry out,dry
"02 YO,2

not out,dry out,dry (4)
out out,dry *lout,dry
NJ2 =N2

Summer 2004

Since nitrogen is inert, the calculations for the inlet air are
begun by equating the inlet and outlet molar flowrates of ni-
*ifn = out O
nN, =N, (5)
The total flowrate of inhaled air can now be calculated on a
dry basis, assuming a composition of 79% nitrogen and 21%
Sin in,dry fin,dry (6)
nN, YN (6)
The molar flowrates of oxygen and carbon dioxide in the
inhaled air can be calculated next

hin yindry *in,dry (7)
no, = yo' (7)
o in in,dry lin.dry
"co, = Yco,

Finally, the rates of oxygen and carbon dioxide transfer to
the body can be calculated using a component balance

nin out +* body0 = 0 (8)
o in out body =
co, -nco, +nco,

The rate of water loss involves calculation of the rates of
water vapor inhaled and exhaled, using relative humidity
measurement and known flow rates. The molar rates of wa-
ter vapor in the exhaled or inhaled breath are (respectively)

out out out
nH,O yHO (9)
flin Yin *iin
H20 = YH n (10)

When yYou must be determined from the relative humidity
and the vapor pressure at the appropriate temperature (body
temperature for outlet conditions, ambient temperature for
inlet conditions)
20o= RH p (11)

nin can be determined from the molar flow rate of dry air
(Eq. 5) and the mole fraction of water in the inhaled air
S dry,in pvap
i"n ---RH- (12)
in" P
1-yH0 Po
Although most of the calculations are done using molar
flow rates, the conversion to volumetric flow rates are more
meaningful to students and are well worth one extra calcula-
tion step.


The equipment used for all cardiorespiratory measurements
was a gas-exchange system coupled with a cycle ergometer.
The MedGraphics CPX/D cardiorespiratory gas-exchange
system includes capability for direct oxygen and carbon di-

oxide measurement and ventilation (flow) determination. The
system interfaces with a cycle ergometer (Lode Corvial) for
exercise testing. To prevent cross contamination between pa-
tients (students), disposable PreVent'" pneumotachs were used
once and then discarded. The system was purchased from
MedGraphics (St. Paul, MN) for approximately $35,000.
While this may be prohibitively expensive for an engineer-
ing program if it is not used for research purposes, many uni-
versities have such equipment available in a physiology or
exercise science laboratory. In addition, several companies
offer human physiology teaching kits in the $3,000 range that
allow respiratory flow and volume measurements (e.g.,
Biopac Systems, Santa Barbara, CA; ADInstruments, Colo-
rado Springs, CO; Iworx, Dover, NH).

Prior to commencing the experiment, the MedGraphics
CPX/D system pneumotach is calibrated for air flow rate us-
ing a calibration syringe. Gas calibrations for oxygen and
carbon dioxide are performed using a reference gas (21%
oxygen, balance ni-
trogen) and a calibra-
tion gas (12% oxy-
gen, 5% carbon di-
oxide). In addition,
the barometric pres-
sure and ambient
relative humidity are
entered manually to
the MedGraphics
Breeze Suite soft-
One student per
team of four students
is selected as the test
subject for the experi-
ment. Using the
MedGraphics CPX/D
cardiorespiratory test
system coupled with
the Corvial Cycle er-
gometer, measure-
ments are taken at
rest (for four min-
utes) and during ex-
ercise (for four min- Figure 2. A student performing
utes, pedaling at 70- the respiration experiment.
80 rpm at 30 W brak-
ing power). A student
is shown performing
the experiment in Figure 2.
The following quantities are measured directly and dis-
played using MedGraphics Breeze Suite software:

Chemical Engineering Education

/out out,dry out,dry indry in,dry
SYo ,Yco, 'Yo, 'Yco,
and braking power. As mentioned previously, the experimen-
tally obtained data for flowrate of exhaled breath ( ,out,BTPS)
is reported at BTPS (body temperature and pressure, satu-
rated) conditions, while the gas compositions are reported on
a dry basis. The software offers many options for the conve-
nient display of automatically calculated values for quanti-
ties such as oxygen consumption rate, carbon dioxide pro-
duction rate, and energy expenditure, but for educational pur-
poses it is preferable to perform calculations by hand. If
desired, the calculation/display options can be exercised
in order to provide numbers against which students can
check their calculations.
For their laboratory report, students perform all calcula-
tions by hand. In a subsequent laboratory period, they are
introduced to the process simulator, HYSYS, and in an in-
class activity, they use HYSYS to draw a simple process flow
diagram of the respiration cycle. They provide their data and
allow HYSYS to perform material and energy balances on
the respiration process, and then they compare the results of
the simulation to their hand calculations.
Finally, the psychrometric chart is introduced and students

use it to determine the
water content of in-
haled and exhaled air.
The point on the chart
for inhaled air is iden-
tified using the ambient
temperature and rela-
tive humidity, and the
exhaled air point is at
body temperature (37
C) and is saturated
with water vapor. The
water content of each
stream is read off the
chart, and the water
loss during respiration
is calculated and com-
pared with previously
calculated values. In
this exercise, the rela-

and heart rate. By comparing the resting and exercise gas
exchange measurements, students quantify this physiologic
response. Table 1 shows gas exchange measurements and cal-
culated values for the respiration experiment for a 19-year-
old female student (125 Ib, 66 in). The calculations reveal
that oxygen is consumed and carbon dioxide is produced
during breathing, and that both of these rates increase during
the very mild (30 W) exercise performed in this experiment.
In addition, it is interesting to note that the total volume of air
inhaled is smaller than the volume exhaled; while there is a
slight change in the molar flow rate, the difference is prima-
rily due to the temperature change. The volumetric liquid
equivalent of the rate of water loss is 0.48 mL/min at rest-
this increases to 0.75 mL/min during the mild exercise.
There are three sources of variance in the measurements
that are examined by students in the experiment:
1. Reproducibility of the equipment used for experi-
mental measurement
2. Breath-by-breath variation on a single subject
3. Person-to-person physiologic variations
The first is illustrated by taking five consecutive measure-
ments of the ambient air composition and determining the
average and standard deviation of the oxygen concentration.
The second is explored
by observing ten con-

Gas Exchange Measurements and Calculations
at Rest and During Cycling Exercise.

* vout is reported at BTPS conditions.
* All molefractions ar reported on a dry basis.
SVin, Vo0 and Vtco are calculated at ambient conditions.
* Ambient conditions: T = 20 C, P = 759 mm Hg, RH = 47%.

Measured Variables

Power Vout.BTPS yOut
(W) (L/min)

Calculated Values


13.08 0.185 0.023
20.50 0.171 0.031

tively small changes in oxygen and carbon dioxide com-
positions are ignored, as is any slight deviation from at-
mospheric pressure.

Gas-exchange measurements were taken at rest and during
exercise as described above. Nearly everyone is aware of the
body's physiologic responses to exercise: the body's increased
demand for energy is met with an increased breathing rate

Vo, Vco, Vin
(L/min) (L/min) (L/min)


secutive breath-by-
breath analyses of flow
rate and gas composi-
tions for a single sub-
ject. The third is ex-
plored by examining
software-predicted re-
sults and experimental
results between dif-
ferent students. Fac-
tors such as gender,
height, and weight
are considered.

The oxygen measure-
-0.303 0.267 11.81 0.476 en e eyre
ments are very repro-
-0.665 0.563 18.51 0.767 ducible (0.03%), as is
required of equipment
used for medical test-
ing. The breath-by-breath analysis demonstrates a much
higher level of variation, and students observe that the varia-
tion decreases a few minutes into the test. This is a common
phenomenon witnessed with respiratory testing-the subject
has difficulty breathing naturally and regularly at the begin-
ning of the protocol when he or she is thinking about the test,
but after a few minutes when the subject's thoughts are not
as focused on breathing, it exhibits a much more normal and
regular pattern. Still, the breath-to-breath variation is about
5%. The person-to-person variations are by far the greatest

Summer 2004

source of variance, and it is not uncommon to observe
+30% variations in oxygen-consumption rates between
male and female students.

Process simulators have become an essential tool in mod-
ern chemical engineering education. In the past, most chemi-
cal engineering programs viewed process simulation as a task
inherent to the capstone plant design course, but chemical
engineering programs have recently been integrating process
simulators throughout the entire curriculum.191 At Rowan
we have vertically integrated the use of process simula-
tors in most of the chemical engineering courses, starting
with the freshman clinic.
Using a HYSYS process simulator, the experimental gas-
exchange resting-measurement data are used to simulate the
process of respiration. The process is represented by two unit
operations: a heater that heats the inhaled air to body tem-
perature and a humidifier that saturates the inhaled air
with water. The HYSYS respiration process flow diagram
is shown in Figure 3.
The HYSYS flow sheet has been set up to simulate the
respiration process by providing the experimentally measured
values of flow rates, composition, temperature, pressure, and
relative humidity. Students enter the ambient conditions of
temperature, pressure, and relative humidity into a spread-
sheet operator called the "weather station." A hidden spread-
sheet takes these data and calculates the mole fraction of water
in the inhaled air, using the Antoine equation to determine





Inhaled ;W Warm
Humid Heater Humid
Air Air

the vapor pressure of water at the ambient temperature. These
steps were necessary because HYSYS requires a water vapor
mole fraction rather than relative humidity to calculate water
content of a given stream.
The "inhaled humid air" stream represents inspired air at
ambient temperature, pressure, and relative humidity. The
stream called "exhaled warm saturated air" represents the
exhaled air at body temperature and pressure, saturated with
water vapor. Students supply temperature, pressure, flow rate,
and composition of this stream using their experimental data.
They also supply temperature and pressure values for the in-
termediate streams called "warm humid air" and "moisture
from lung tissue." HYSYS completes the material balances
and students compare their process simulation results with
their hand-calculated results.

An assessment plan based on the rubrics developed by
Newell, et al.,11l was developed to map student work directly
to the individual learning outcomes of these freshmen. The
learning outcomes specifically address ABET criteria and
AIChE- and program-specific goals. This assessment was
based on reasonable expectations for freshman students
who have had their first introductory exposure to engi-
neering principles.
Four instruments were chosen for the evaluation: a team
laboratory report, an individual in-class quiz, a formal oral
presentation, and an interactive poster presentation. The labo-
ratory reports and quizzes were evaluated by the course in-




Figure 3. The HYSYS respiration process flow diagram.

Chemical Engineering Education

structor only, and the formal and poster presentations were
graded by two engineering professors. These were evaluated
for two consecutive years. The first column in Table 2 shows
the stated objectives/outcomes that were evaluated on a four-
point ordinal scale to describe student performance, as dis-
cussed previously in the paper by Newell.lo01 The second col-
umn provides numerical results that indicate the average score
for the four instruments.
We believe that the student and faculty scores indicate that
we were successful in achieving our stated learning objec-
tives. In using traditional classroom surveys, the students re-
sponded that the module contributed to their enthusiasm for
engineering, as evidenced with a score of 3.6 out of 4.

This paper describes a simple and exciting laboratory ex-
periment in which a wide range of chemical engineering prin-
ciples are introduced through application to the process of
respiration. Students take measurements of physiologic vari-
ables both at rest and during exercise, and then perform cal-
culations involving mass balances. Through these calcula-
tions, students apply the ideal gas law and concepts of partial
pressure and relative humidity. Students are also introduced
to chemical process simulation software when they simulate
the process of respiration using HYSYS.
Basic physiologic responses are already familiar to students
through "common knowledge" and sensory experiences, and
most of them have a natural curiosity to learn how their own
bodies work. In a series of hands-on experiments that use
engineering measurements and reverse engineering methods,

Assessment Results
(1 = low to 4 = high)

Objective/Outcome (to demonstrate...) and Mapped Av. Fac. Av. Stud.
to Goal Score Score
A working knowledge of ChE principles (mass balances,
psychrometric chart, unit operations): AIChE Pro-
fessional Component 3.4 3.2
A working knowledge of chemistry (ideal gas law, vapor
pressure, partial pressure): AIChE Professional Component 3.3 3.0
An ability to function on multidisciplinary and/or
diverse teams: ABET-D 3.4 3.5
An ability to approach tasks involving experimental results
in a logical and systematic fashion (measurements, recording,
analysis, and interpretation : Program 3.1 N/A
An understanding of contemporary issues relevant to the
field (current technical material, finding relevant current
information and use in curricular assignments: ABET-J 2.9 3.7
An ability to use techniques, skills, and modern engineering
tools necessary for engineering practice (spreadsheets,
word processors, and process simulators) to assist in
problem solving: ABET-K 3.6 3.8
Effective oral and written communication skills: ABET-G 3.4 3.5

these physiologic responses are quantified. This establishes
a framework within which new engineering concepts are
introduced through analysis of the data. Using a familiar
system, sensory experiences, and hands-on active learn-
ing are thought to increase understanding and retention
of the new concepts.

Funding for this project was obtained from the National
Science Foundation Course, Curriculum, and Laboratory
Improvement Program (NSF DUE #0088437).

n molar flow rate (mol/min)
P pressure (barometric) (mm Hg)
p'p vapor pressure of water (mm Hg)
R universal gas constant
RH relative humidity (fraction)
T temperature (K)
V volumetric flow rate (L/min)
y mole fraction
a ambient
b body
BTPS body temperature and pressure, saturated
dry on a dry basis
in inlet stream or inhaled air
out outlet stream or exhaled air

1. Hesketh, R., and C. Stewart Slater, "Demonstration of Chemical En-
gineering Principles to a Multidisciplinary Engineering Audience,"
Proc. Ann. Conf. ASEE, June (1997)
2. Marchese, A.J., R.P. Hesketh, K. Jahan, T.R. Chandrupatla, R.A.
Dusseau, C.S. Slater, and J.L. Schmalzel, "Design in the Rowan Uni-
versity Freshman Engineering Clinic," Proc. Ann. Conf. ASEE, June
3. Hesketh, R.P., K. Jahan, A.J. Marchese, C.S. Slater, J.L. Schmalzel,
T.R. Chandrupatla, and R.A. Dusseau, "Multidisciplinary Experimen-
tal Experiences in the Freshman Engineering Clinic at Rowan Univer-
sity," Proc. Ann. Conf ASEE, June (1997)
4. Ramachandran. R., J. Schmalzel, and S. Mandayam, "Engineering Prin-
ciples of an Electric Toothbrush," Proc. Ann. Conf ASEE, June (1999)
5. Merriam Webster's Collegiate Dictionary, 1 Oth ed., Merriam Webster,
Inc., Springfield, MA (1998)
6. HYSYS, version 2.4.1, Hyprotech Ltd (2001)
7. McArdle, W.D., F.I. Katch, and V.L. Katch, Exercise Physiology: En-
ergy, Nutrition, and Human Performance, 4th ed., Lea and Febiger,
Philadelphia, PA (1996)
8. Adams, Gene, Exercise Physiology Laboratory Manual, W.C.B.
McGraw Hill, NY (1998)
9. Dahm, K.D., R.P. Hesketh, and M.J. Savelski, "Is Process Simulation
Used Effectively in ChE Courses?" Chem. Eng. Ed., 36(3) (2002)
10. Newell, J.A., K.D. Dahm, and H.L. Newell, "Rubric Development
and Inter-Rater Reliability Issues in Assessing Learning Outcomes,"
Chem. Eng. Ed., 36(3), 212 (2002) 0

Summer 2004

,W laboratory




Northeastern University Boston, MA 02115

Classroom demonstrations enhance the sensing learn-
ing style that many engineering students use. This
simple apparatus, which can be constructed with com-
ponents found at a local discount store for under $20, effec-
tively demonstrates the nature of dust explosions. The dem-
onstration can be used to complement a lecture in thermody-
namics (combustion, heating, and rapid expansion of gases),
process design (hazards involved with solids handling), or
process safety (dust explosions and vent sizing).
Any combustible solid that can be reduced to a fine pow-
der has the potential for involvement in a dust explosion. The
Oxford Dictionary[" defines dust as
Earth or other solid matter in a minute or fine state of
subdivision so that the particles are small and light enough
to be easily raised and carried in the wind; any substance
comminuted or pulverized; powder
Typically, these solids are 1 to 50 microns in particle size.
For an excellent overview about dusts and their hazards, the
World Health Organization has an html document available
on the web.'21
Many accidental dust explosions occur during manufac-
turing operations associated with the preparation or use of
such materials as pharmaceutical powders, wheat flour, wood
processing, metallic powders, powdered coal, powdered sugar,
powdered confectionery ingredients, etc. Eckhoff[3] reports
that during the past twenty years, dust explosions have ac-
counted for several hundred deaths and hundreds of millions
of dollars in property damage. A recent dust explosion in
Kinston, North Carolina, demonstrates how damaging these
explosions can be (six people were killed and dozens were
injured).'4' It is critical that controls be in place to prevent
these events from happening.

* Address: 31 Granite Ridge Road, Cumberland, ME 04110

Some additional teaching aids on dust explosions are avail-
able through SACHE products (CCPS-AIChE). Two prod-
ucts particularly suited to this experiment are the "Dust Ex-
plosion Control"f5' video/slide/lecture and the "Explosion"E61
video. The "power" of a dust explosion is quantized by a
value called the deflagration index, Kst[71

Kst (dP V3 (1)
K dt )ax
This value can be used to predict the over-pressure rate at a
boundary using a scaled distance (volume to the 1/3rd power
in the above equation). Aluminum powder, for example, has
one of the highest Kst at 415 bar m/s. For comparison, the
material used in the experiment described in this paper has a
Ks, of 151 bar m/s8. The textbook by Crowl and Louvar'91
provides more details related to mechanism and predictions
involved with dust explosions. A source for minimum flam-
mability concentrations for many types of dusts (including
agricultural products, carbonaceous dusts, chemical dusts,
metal dusts, and plastics) can be found in Appendix D of the

Ronald Willey is a Professor of Chemical
Engineering at Northeastern University. He is
a member of SACHE (Safety and Chemical
Engineering Education) and a director in the
AIChE Safety and Health Division.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

SEd S. Shanley is a retired Vice President of
Arthur D. Little, Inc., in Cambridge, Massachu-
setts, and specializes in chemical process

NFPA 68 Guide for Venting ofDeflagrations.["'1

A transparent plastic food storage box with a snap-on lid, about
15 centimeters square and 18 centimeters in height, available at
department and general merchandise stores (hereinafter called "the
box"), provides a satisfactory container for the demonstration. The
exact size of the box is not important, but its side walls should be
transparent. An aluminum foil heat shield should be attached to the
inside surface of the snap-on lid-sticky tape is sufficient to do the
job. A quarter-inch hole should be drilled in the side wall at one of
the bottom corners of the box.
A small tray to hold the flammable dust can be constructed of
heavy kitchen-type aluminum foil. A piece of foil is cut into the
form of a regular trapezoid with edge dimensions of about six-by-
five-by-six-by-three centimeters. A pencil stub or similar-sized
object about four centimeters long is attached (with tacks or adhe-
sive) to what will become the bottom of the five-inch side. Over-
hanging foil is bent upward to form the outboard end of a shallow
tray. The opposite, three-centimeter, side is bent upward to form an
open trough about 1 centimeter in width. The finished object is a
small tray about six centimeters long, almost flat at the wide, out-
board end and formed into an open channel about one centimeter
deep at the inboard end. Placed on a flat surface, the tray will have
a gentle upward slope toward the outboard end. Figure 1 provides a

Figure 1. Materials used for dust explosion

photographic illustration of the individual parts of the
apparatus described above.
The dimensions noted above are not critical. The ob-
jective is to contain the flammable dust in such a way
that a puff of air at the inboard end will lead to disper-
sion of the sample into the air space of the box.
Flammable dusts of many different sorts have been in-
volved in destructive explosions. Nevertheless, dust
samples that have been held in storage for any length of
time become more difficult to ignite, possibly because
of agglomeration during storage or adsorption of mois-
ture from the surroundings. Lycopodium powder, readily
available from reagent suppliers, is exceptional in retain-
ing its easy ignitibility, even after prolonged storage. It
is the most satisfactory fuel for dust explosion demon-
strations. Part of the reason is that its equilibrium mois-
ture content is low as compared to cornstarch and other
combustible powders that could be used.
The lycopodium dust sample, about 0.5 cm3 in vol-
ume, is placed in the narrow end of the tray after that end
has been positioned directly in front of the 1/4-inch hole
drilled near a bottom corner of the food storage box
(which is intended for accepting the discharge end of the
turkey-baster tube). The tray should be oriented diago-
nally toward the opposite corner of the box.
A short candle of the type used in candle lamps, about
1/2-inch high and 1 1/2-inch diameter, serves as a con-
venient ignition source. It is placed on the inside bottom
of the box, diagonally across from the dust sample tray.
If the candle is provided with a handle, say of coat-hanger
wire, it can be ignited before positioning in the box. Safety
glasses should be worn, and without undue delay, the
snap-on top of the container should be put in place, mak-
ing sure that it is firmly seated. Inserting the discharge
end of the turkey baster into the quarter-inch hole in the
container wall and gently squeezing the baster's bulb can
lead to immediate dispersion and ignition of the dust
cloud, filling the container with a burst of flame and blow-
ing off the snap-on top with a satisfying "pop." The candle
can be repositioned a bit to one side if the air jet extin-
guishes it before the dust cloud is ignited.
Figure 2 shows the flame propagation rate in the appa-
Continued on page 195.

Propagation (70 ms) 140 ms 200 ms 270 ms
Figure 2. Sequential photographs of explosion recorded during the demonstration.

Summer 2004

B 1 laboratory


Discharging Vessels

University of Newcastle Callaghan, New South Wales 2308, Australia

ompressible flows are usually observed for gases and
characterized by a significant change in the fluid den-
sity with a change in either the pressure or the tem-
perature. They represent an important topic within the under-
graduate curriculum due to their common occurrence through-
out the field of chemical engineering, e.g., high-pressure gas
jets used for mixing and chemical reaction. This paper re-
ports on the second stage of an undergraduate laboratory ex-
periment that was developed to illustrate some of the impor-
tant concepts of compressible flow. The first stage of the
experiment dealt with filling the vessels and was published
previously in this journal.'"
A schematic representation of the experimental rig is shown
in Figure 1. It consists of an insulated pressure vessel, the
inlet to which is connected to the air mains while the outlet is
connected to a converging nozzle. The first stage of the ex-
periment involves pressurizing the vessel from initial atmo-
spheric conditions up to a predetermined elevated pressure
and measuring the corresponding temperature change within
the vessel.' ] The second stage of the experiment involves dis-
charging the vessel to atmosphere via a converging nozzle
and recording the pressure-versus-time relationship for this
process. This paper considers only the second (or discharg-
ing) stage, and the principal objectives are
To measure the pressure-versus-time relationship as the
vessel discharges to the atmosphere
To develop a theory for this pressure variation for compari-
son with the experimental data
The significance of this stage of the experiment concerns
developing the theory for the time variations in vessel pres-
sure. An important part of any theoretical analysis is specifi-
cation of the simplifying assumptions on which the model is
based, to enable the equations to be solved and ideally to
produce an analytical solution. This study demonstrates that
it is possible for an assumption to be justified by greatly re-

* Warsaw University of Technology, PL-00-645 Warsaw, Poland

during the complexity of a problem while generating a solu-
tion that adequately describes the experimental process,
even though the assumption may appear to be invalid. This
is the opposite situation to modeling the filling process,
where the importance of correctly choosing the system
boundary is highlighted.

The experiment described in this paper involves discharg-
ing a pressure vessel to the atmosphere through a convergent
nozzle (as shown in Figure 1). In this section the theory for
the discharging process is considered with the objective of
determining an expression for the variation in vessel pres-
sure with time.
The first step is to define the "system," or control volume,
on which to perform the thermodynamic analysis; this is
shown in Figure 1 and was described previously by Forrester
and Evans.[" For the discharge stage of the experiment, the
system is initially isolated at some elevated pressure, P(0),
while the experiment ends when the vessel pressure has
dropped to that of the surroundings, i.e., P(t) = PA. The prin-

Stephanie E. Forrester is Research Associate in Chemical Engineering
at The University of Newcastle, Australia. She received her degrees from
the Universities of Edinburgh (BE Hons) and Cambridge (PhD). Her research
and teaching interests include fluid mechanics and computational modeling.
Anh V. Nguyen is Associate Professor in Chemical Engineering at The
University of Newcastle, Australia. He received his degrees from the Tech-
nical University of Kosice, Czechoslovakia (BE Hons and PhD). His teach-
ing and research interests include multiphase processes, colloidal hydro-
dynamics and chemistry, and computational modeling.
Geoffrey M. Evans is Professor in Chemical Engineering at The Univer-
sity of Newcastle, Australia. He received his degrees from the University of
Newcastle (BE Hons and PhD). His teaching and research interests are in
the areas of fluid mechanics, thermodynamics, and multiphase systems.
He is also interested in developing sustainability courses as generally ap-
plied to the chemical engineering discipline.
Piotr M. Machniewski is currently a lecturer in the Department of Chemi-
cal and Process Engineering at the Warsaw University of Chemical Tech-
nology, from which he received both his undergraduate and PhD degrees.
His research interests are in mass transfer, fluid mechanics, and chemical
engineering education.
Copyright ChE Division of ASEE 2004

Chemical Engineering Education

This paper reports on the second stage of an undergraduate laboratory experiment that was
developed to illustrate some of the important concepts of compressible flow.
The first stage of the experiment dealt with filling the vessels
and was published previously in this journal.

ciple assumptions required in the analysis presented below
The temperature andpressure within the vessel remain at
stagnation conditions throughout the entire discharge

converging nozzle

ambient pressure, PA

nozzle exit:
pressure, Pe
temperature, Te
area, A
velocity, U

pressure, P
temperature, T
area -> o
velocity 0

Figure 2. Flow through the converging nozzle.

Isentropicflow through the nozzle
The pressure vessel can be regarded as adiabatic
The air behaves as a perfect gas
The vessel temperature is constant throughout the experi-
The nozzle is choked at all times
The nozzle discharge coefficient is equal to one

Applying continuity to the system outlined in Figure 1 gives
-=-q(t) (1)
where m is the mass of gas in the pressure vessel and q is the
mass flow rate through the nozzle. Applying the perfect gas
law, the mass of gas in the vessel is given by

m= (2)
where P and T are the (stagnation) pressure and temperature
inside the vessel, V is the external volume of the vessel, and
R is the specific gas constant. The mass flow rate of gas exit-
ing the vessel through the discharge nozzle, q, is given by

q(t)=p(t)AU (3)
where p is the gas density, A is the cross-sectional area of the
nozzle, and U is the mean gas velocity, all measured at the
nozzle exit plane as shown in Figure 2. Substituting Eqs. (2)
and (3) into Eq. (1) gives

V dP
=-p(t)AU (4)
RT dt
Applying the perfect gas law gives

p(t)= (t) (5)
where Pe and Te are the pressure and temperature at the nozzle
exit plane, respectively. For isentropic flow through the nozzle
one has[21

U=Ma yRTe (6)

where Ma is the Mach number and y is the ratio of specific
heats. Substituting Eqs. (5) and (6) into Eq. (4) gives

V dP Pe(t) AMa (7)
RT dt Te
The next stage is to relate the pressure and temperature at the
nozzle exit plane, P. and Te, to the stagnation conditions in-
side the vessel, P and T, respectively. This can be achieved
using the following identities for isentropic flow:[2]

Summer 2004

=P 72
Pe = P + 2Ma2 Y-1

Te = (9)
fl + 'Y 1 M a2
2 J
Substituting Eqs. (8) and (9) into Eq. (7) and rearranging gives

dP_p AMa Ma- 27+2(10)
dt V 2

Under conditions where the nozzle is choked, i.e., Ma = 1,
Eq. (10) can be simply integrated between the starting time, t
= 0, up to some later time, t, giving
n P(t) = -kt (11)
where k is a rate constant (with the dimension of reciprocal
of time) described by
kA + _-Ma2) 27+2 (12)
yRT 1+ Ma2
Eq. (11) suggests that a plot of fn[P(t) / P(0)] against time, t,
for the vessel discharge process is a straight line with the
gradient equal to k.
The range of vessel pressures for which the nozzle is choked
and Eq. (11) is valid can be determined by substituting Ma =
1 and Pe = PA into the isentropic flow relationship given by
Eq. (8), leading to

P(t) A (13)

where PA is the pressure at the nozzle exit as shown in Figure
2. For a convergent nozzle discharging air to the atmosphere,
PA = 101 kPa and "y = 1.4, thus Eq. (13) predicts that the
nozzle will be choked for all P(t) 2 192 kPa.

A schematic representation of the experimental rig is shown
in Figure 1. Three different discharge nozzles, of diameters
2.5 mm, 3.0 mm, and 4.0 mm, can be attached to the pressure
vessel outlet. The first stage of the experiment involves pres-
surizing the vessel from the air mains, the procedure for which
is described elsewhere.' tAt the end of the pressurizing stage,
both the inlet and exit globe valves are closed and the vessel
is allowed to reach steady state (or equilibrium) conditions
of temperature and pressure.
The second, or discharge, stage of the experiment is con-
ducted by first setting the speed on the chart recorder accord-
ing to which discharge nozzle is being used: 2 cm/min for the
2.5 mm nozzle, 6 cm/min for the 3.0 mm nozzle, and 20 cm/
min for the 4.0 mm nozzle. The outlet valve is then opened

and the vessel is allowed to discharge (with a muffler placed
over the exit nozzle to reduce noise pollution). As the vessel
pressure drops, the chart recorder switch is flicked every 35
kPa (5 psi) change in pressure, and this continues until the
vessel pressure has dropped to the ambient conditions (which
typically takes around 2 to 3 minutes). The vessel is then left
for approximately 5 minutes to reach steady state before the
next experimental run is commenced. The discharge stage is
conducted for the three discharge nozzles defined above and
a single value for the initial vessel pressure of approximately
730 kPa.
Important Note: There are a number of inherent safety
implications associated with this experiment that should
be noted. First, although air is a benign material under
atmospheric conditions, it can become hazardous at el-
evated pressures (4-10 atmospheres in this case). Sec-
ond, the experiment should only be attempted using a
properly certified vesselfitted with the appropriate pres-
sure-relief system; this is absolutely essential when un-
dertaking any experiment under pressurized conditions,
but especially important when gases are involved since
rupture can result in catastrophic explosion of the vessel.

Variation in the Vessel during Discharge
Typical experimental and theoretical results for the vessel
pressure, plotted in the form en[P(t) / P(0)] as suggested by

Input Parameters for the Theoretical Calculations

Parameter Value Notes
V 0.102 m
R 287 J/kg/K For air
y 1.4 For air
P(0) 726 kPa Nozzle diameter = 2.5 mm
T 295.9 K
P(0) 733 kPa Nozzle diameter = 3.0 mm
T 295.6 K
P(0) 733 kPa Nozzle diameter = 4.0 mm
T 289.0 K

Experimental and Theoretical Results for the Rate of
Vessel Discharge

diameter kth
(mm) (I/s)


k,p,, Difference
(1/s) (%)


k .hk, Difference
(i/s) (%)


Chemical Engineering Education

Time (s)
50 100 126

186 20

S2.5 mm diameter nozzle


- Theon. Eq (11)

Time (s)

Eq. (11), as a function of time are shown in Figure 3. The theoretical
results are based on the input parameters listed in Table 1. These plots
clearly demonstrate good absolute agreement in the variation of ves-
sel pressure with time throughout the discharge process, confirming
the applicability of the theoretical analysis presented above. In par-
ticular, the experimental data show a close-to-linear variation in
(n[P(t) / P(0)] with time as predicted by the theory.
It is possible to calculate the gradient of the experimental data points
for comparison with the theoretical values of k determined from Eq.
(12). In this study, two experimental gradients have been found, in-
Using all the experimental data points, kexp all
Using only the experimental data points in the region over which
the nozzle is choked, k
exp, choke
The results are given in Table 2 and confirm the very good agree-
ment between the theoretical and experimental characteristic vessel
discharge rates, the maximum deviation being 6 percent. The data in
Table 2 also indicates that the absolute value of k increases as the
nozzle diameter increases, corresponding to an increasing rate of ves-
sel discharge. This is confirmed in Figure 3, which illustrates that for
the 2.5-mm nozzle it takes 186 s for the vessel to discharge to ambi-
ent pressure conditions, for the 3.0-mm nozzle it takes 130 s and for
the 4.0-mm nozzle it takes only 74 s. In addition, as the nozzle diam-
eter increases, the slight discrepancy between the experimental and
theoretical values for the discharge gradient also increases. As the
nozzle diameter increases, the rate at which the vessel discharges also
increases, leading to larger experimental error in the manual opera-
tion of recording the time for each 35 kPA (5 psi) drop in vessel pres-
sure. The final point to note from Table 2 is that the experimental
discharge gradients based on all data points, kxp, all, give slightly bet-
ter agreement with the theoretical values derived assuming choked
flow throughout, keo", compared to those based only on the data points
in the choked flow region, kxp. choke. This is unlikely to be of physical
significance, but is simply a result of experimental error and the as-
sumptions used in deriving theoretical expression.

The results shown in Figure 3 clearly illustrate very good agree-
ment between experiment and theory, indicating the applicability of
the model described above to predict the pressure-versus-time rela-
tionship for a pressure vessel discharging to atmosphere. It is also of
interest, however, to use the experimental data to examine the applica-
bility of the principal assumptions underlying the theoretical analysis.
In particular, two of the assumptions used in the model for the dis-
charging vessel can be investigated based on the experimental results.
First it is assumed that the nozzle is choked throughout the entire
discharge process, which (as discussed earlier) is only true under con-
ditions where P(t) 2 192 kPa. The actual region over which the nozzle
is choked is illustrated in each of the plots in Figure 3; all the nozzles
are choked for the first two-thirds (68 percent) of the discharge pro-
cess, but not for the final third (32 percent). Clearly, the experimental

Figure 3. Typical experimental and theoretical re-
sults for the variation in pressure inside the vessel
with time during the discharge process at different
nozzle diameters.

Summer 2004

data points give good agreement with the theory of Eq. (12)
throughout the entire discharge process, under both choked
and unchoked conditions. There may be a very slight drift in
the experimental data away from the theory as the nozzle
enters the unchoked region-however, the difference remains
insignificant even when the discharge process reaches comple-
tion. Therefore, even though the nozzles are choked for only
the first two-thirds of the discharge process, it can reason-
ably be assumed they remain close to being choked through-
out the remaining period of the discharge process in deter-
mining the pressure-versus-time relationship for the vessel.
Second, the assumption that the vessel (stagnation) tem-
perature, T, remains constant throughout the discharge pro-
cess can be assessed. The chart-recorder output, used to mea-
sure the pressure-versus-time discharge relationship, indicates
a significant drop in the internal vessel temperature, typically
by about 10K over the duration of the experiment. To deter-
mine the impact of this temperature change on the theory for
the rate of vessel discharge, the following relationship be-
tween the discharge gradient and the stagnation temperature
can be obtained from Eq. (11)

kaT1/2 (14)
Hence a AT change in the value of T over the duration of an
experiment can be related to a Ak change in the value of k by

Ak ( AT 0.5
=1+T) -1 (15)

where T' and k* represent the conditions at the start of the
discharge process. Eq. (15) predicts that for a temperature
drop, AT, of 10 K based on an initial stagnation temperature,
T = 300 K, the corresponding value of k decreases by only
1.7%, which is clearly insignificant. Hence, even though the
stagnation temperature may change significantly during the
vessel discharge, this will have a negligible impact on the
rate of discharge, and it is therefore reasonable to assume
that T is constant in determining the pressure-versus-time re-
lationship for the discharge process.
If, however, one really wants to solve the pressure equa-
tion for the variation in the stagnation temperature with time,
the expression is described by

---We (16)
dt L dt J
where the discharge rate constant, k, is determined by Eq.
(12) with T being replaced by the stagnation temperature at
time, t = 0, and the scaled variable, 0, is a function of time,
which is defined by 0 = T(t) / T(0). For small variations in the
stagnation temperature with time, 0 is of the order of unity,
and Eq. (16) reduces to Eq. (10), as expected. In the general
case, Eq. (16) is a nonlinear differential equation of the first
order for the stagnation pressure, P, with respect to t, which
has to be numerically integrated using, for example, the four-

step Runge-Kutta scheme. The initial condition includes P =
P(0), 0 = 1, and time t = 0. Therefore, this topic can be suit-
ably extended to advanced undergraduate students.


In this paper, we have described an undergraduate experi-
ment on compressible flow, based on the discharge of an adia-
batic pressure vessel through a converging nozzle. In par-
ticular, the vessel is emptied from an initial pressure of ap-
proximately 730 kPa to ambient conditions, and the varia-
tion in vessel pressure with time is recorded; the process is
repeated for three different nozzles of diameter 2.5 mm, 3.0
mm, and 4.0 mm. Discussion of the experimental results has
involved a qualitative description of the variation of internal
vessel pressure as a function of time, the development of a
theoretical model for the process, and a comparison of the
resulting model predictions with the experimental data.
Both the experimental and theoretical results show a linear
relationship between fn[P(t) / P(0)] and time, with the abso-
lute value of the gradient increasing with increasing nozzle
diameter, corresponding to an increasing rate of vessel dis-
charge. Furthermore, there is good absolute agreement be-
tween the experimental and theoretical gradients for all three
nozzles used, the maximum discrepancy being 6%, confirm-
ing the applicability of the theoretical analysis. The theory
predicts a discharge gradient of -0.00960/s for the 2.5-mm
nozzle, -0.0138/s for the 3.0-mm nozzle, and -0.0243/s for
the 4.0-mm nozzle.
The experimental results also allow the applicability of
some of the principal assumptions used in the theoretical de-
velopment to be assessed. First, it is assumed that the nozzle
is choked throughout the discharge while experimentally the
nozzle is observed to be choked for only the first two-thirds
of the experiment. Second, it is assumed that the vessel stag-
nation temperature remains constant throughout the discharge,
while experimentally it is observed to drop by around 10K
during the experiment. Although these assumptions have been
shown to not apply strictly, they are justified in allowing the
development of a simple analytical model describing the pres-
sure variation during discharge, which provides an excellent fit
to the experimental data. This is the opposite situation to the
case of filling the vessel where it was very important to cor-
rectly define the system boundary and its initial conditions.

A cross-sectional area of nozzle at the exit plane [m2]
q mass flowrate of gas through the nozzle [kg/s]
m mass of gas in the pressure vessel [kg]
Ma Mach number [-]
P stagnation pressure [N/m2]
PA ambient pressure [N/m2]
Pe pressure at the nozzle exit plane [N/m2]
R specific gas constant [J/kg/K]

Chemical Engineering Education

t time [s]
T stagnation temperature [K]
T temperature at the nozzle exit plane [K]
U gas velocity at the nozzle exit plane [m/s]
V internal volume of pressure vessel [m3]
k discharge rate constant defined [1/s]
y ratio of specific heats [-]
p gas density at the nozzle exit plane [kg/m3]
0 dimensionless temperature, defined as 0 = T(t) / T(0)

1. Forrester, S.E. and G.M. Evans, "The Importance of System Selection
on Compressible Flow Analysis: Filling Vessels," Chem. Eng. Ed.,
2. Kay, J.M. and R.M. Nedderman, Fluid Mechanics and Transfer Pro-
cesses, Cambridge University Press, Cambridge, United Kingdom
(1985) 0

Continued from page 189.
ratus. In this experiment, the time from ignition to full in-
volvement was on the order of (200 70) ms, or 130 ms for
propagation through about 20 cm. This corresponds to a propa-
gation rate of roughly 1.5 meters per second, extremely slow
by explosive standards. For example, black gunpowder
propagates at a rate of about 400 meters per second, while
typical high explosives such as TNT propagate at about
4000 meters per second.["1
Flammable dusts rarely, if ever, constitute a hazard in the
open air. Operations capable of creating dust explosion haz-
ards are usually conducted inside buildings such as flour mills
and grain elevators, as well as in facilities associated with
the manufacture and/or use of such products as edible flours,
powdered sugar, metallic pigments, etc. Dust concentrations
capable of ignition are reported to contain on the order of at
least 30 g/m3.'01 This is much higher in solids content than
could be tolerated by human operators. For example, it has
been noted that minimum flammable concentrations of most
dusts would limit visibility to a meter or so. Accordingly,
flammable dust-air compositions are usually found in closed
processing containers or in isolated areas within a manufac-
turing facility. An ignition source is also required-perhaps
a pilot flame, a welding spark, an electrical fault, or the like.
The original explosion may be too small to cause appre-
ciable damage. The resulting shock wave may, however, dis-
lodge additional dust from horizontal surfaces, cracks and
crevices, storage areas, and the like. A new and perhaps larger
dust cloud is formed and may be ignited by the original source
or by hot embers. This cycle, typical of dust explosions, may
repeat itself four or five times or more and culminate in com-
plete destruction of the facility. Cleanliness counts in keep-
ing control of dust explosions.
Dust explosions in closed containers are reported to gener-

ate pressure on the order of 3 to 7 atmospheres.J21 Buildings
housing ordinary manufacturing facilities will not support
such internal overpressures. Quite modest excess pressure,
on the order of a fraction of an atmosphere, may cause roofs
to rise and walls to bulge, leading to a complete collapse of
the structure. 12' This collapse represents most of the energy
released during the incident. Keep in mind that the initial
dust explosion had only a small fraction of that energy. The
dust explosion energy probably served only to move or dis-
tort structural elements upon which the building was sup-
ported. A little can do a lot.

We have provided a simple system to demonstrate the ex-
plosiveness of dusts. Students witnessing these experiments
are always impressed and tend to remember this demonstra-
tion for many years thereafter. The experience creates an
awareness of the explosiveness of dust and of the necessity
to prevent such experiences from happening inadvertently.

As we were preparing this paper, a high school teacher,
Mr. David Barr, Cranston High School West, pointed out to
us a similar experiment used during Halloween that is de-
scribed on the internet.113,141

1. Simpson, J., and E. Weiner, The Oxford English Dictionary, 2nd ed.,
Oxford University Press, Corby, UK (1989)
2. Hazard Prevention and Control in the Work Environment: Airborne
Dust, World Health Organization, WHO_SDE_OEH_99.14.pdf> (1999)
3. Eckhoff, R.K., Dust Explosions in the Process Industries, Butterworth-
Heinemann, Boston, MA (1997)
4. West Pharmaceutical Services Plant Explosion and Fire, Kinston, NC,
Chemical Safety Board, completed_investigations/info.cfm?ID=34> (2003)
5. Louvar, J.E, and R.W. Schoeff, Dust Explosion Control, Center for
Chemical Process Safety, AIChE, New York, NY (1995)
6. Explosions, Center for Chemical Process Safety, AIChE, New York,
NY (1999)
7. NFPA 68-Guide for Venting ofDeflagrations, NFPA, 1 Batterymarch
Park, Quincy, MA (2002)
8. ASTM Standard, E1226,ANSI, Washington, DC (from Dr. Joe Sencal
Kidde-Fenwal, Inc.)
9. Crowl, D.A., and J.F. Louvar, Chemical Process Safety: Fundamen-
tals with Applications, 2nd ed., PR Prentice Hall, Engelwood Cliffs,
NJ, p. 259 (2002)
10. NFPA 68, "Venting of Deflagrations," National Fire Protection Asso-
ciation, Batherymarch Park, Quincy, MA, App. D (1988)
11. Cook, M.A., ACS Monograph No. 139, The Science of High Explo-
sives, Reinhold Pub. Corp., New York, NY, p. 45 (1985)
12. Crowl, D.A., and J.E Louvar, Chemical Process Safety: Fundamen-
tals with Applications, 2nd ed., PTR Prentice Hall, Engelwood Cliffs,
NJ, Table 6.9, p. 267) (2002)
13. Geyer, M., Subject: A Halloween Story, found at> (2003)
14. Flinn Scientific, Inc., Lycopodium Powder The Mini Grain Elevator
Explosion, found at lycopodiuml.pdf> (2003) 0

Summer 2004

e, laboratory



A Laboratory Experiment in Biochemical Engineering

The University ofAuckland Auckland City, New Zealand

Nowadays, biocatalysts (including microorganisms
and enzymes) and bioreactors are applied not only
to the bioprocess industries but also to problems in
waste treatment and remediation of contaminated sites. Use
of concepts related to them is important to the new genera-
tion of chemical engineers, but many students are not ex-
posed to biology through the standard chemical engineering
curriculum. In order to introduce these concepts and extend
traditional chemical engineering principles, we have designed
a laboratory experiment in biochemical engineering that dem-
onstrates a typical growth pattern of microorganisms, as well
as the fermentation process involving multistage scale-up.
Saccharomyces cerevisiae, mainly in the form of baker's
yeast, represents the largest bulk production of any single-
cell microorganism in the world. Several million tons of fresh
S. cerevisiae cells are produced yearly for use in human food.t
In several areas of fundamental biological science, S.
cerevisiae has been extensively studied and serves as a valu-
able model eukaryotic cell in such studies."l The large-scale
manufacture of baker's yeast involves a multistage propaga-
tion of a specially selected S. cerevisiae strain on molasses.[21
The large-scale production of many other microorganisms
also follows the same principles that are employed here. The
propagation of baker's yeast is applied in this demonstration
because it is a widely used and well-studied system using
easily obtainable raw material.
In this experiment students become acquainted with prac-
tical microbiology techniques such as preparing and steriliz-
ing media and equipment by autoclaving, inoculating yeast
into shaker flasks, and inoculation of the bioreactor. The ef-
fect of temperature and pH on the propagation of baker's yeast
is also studied. Fermentation processes are operated in both
batch and fed-batch modes.
Because of the relatively slow response time of biological
systems, experiments span one-and-a-half days. Students usu-
ally came to the laboratory to prepare the media and auto-

clave the bioreactor and other stuffs in the afternoon. They
began cell cultures in the shake flasks and then transferred to
the bioreactor in the following day. We did not expect one
student to be in the laboratory for such a long period, so four
or five students would work together as a group in shifts.
Each group repeated the experiment three times, varying op-
erating conditions (temperature or pH). Students should
change their roles in the group to make sure that each of them
can go through the whole process.

Yeast Propagation Mathematics
The propagation of baker's yeast follows a typical micro-
bial growth pattern: a lag phase where no growth takes place,
an exponential phase where the growth follows a first-order
reaction scheme, a stationary phase, and then a death phase.031
Growth during the exponential phase can be given by
S= tX (1)
where X is biomass concentration and 9i is the specific growth
As long as gt is constant, the differential equation above
can be integrated with the initial condition of t = 0, X = X0

Xiao Dong Chen is a professor and fellow of the Royal Society of New
Zealand, Fellow of the Institution of Chemical Engineers, and a member
of AIChE. His main teaching areas are heat and mass transfer, food pro-
cess engineering, and bioprocessing. His main research areas include
freeze concentration, drying, fouling, cleaning, and separation.
Matt Hardin is now a lecturer at the Division of Chemical Engineering,
University of Queensland. He teaches undergraduate lab courses as well
as bioprocessing. His research interests include ethanol and carbohydrate
derived fuels and plastics and value adding of food processing wastes.
Grace X.M. Li is a postdoc research fellow and a part-time tutor at the
Chemical and Materials Engineering Department, the University of
Auckland. She is a biochemical engineer with expertise in enzymatic re-
actions, enzyme immobilization technology, andprobiotic bacteria deacti-
vation during drying.

Copyright ChE Division ofASEE 2004

Chemical Engineering Education

X = Xoet (2)
Taking the natural logarithms of Eq. (2) gives
fnX = nX0 (3)
This equation indicates that a plot of fn X versus time will
give a straight line with slope ut.
It has been proven121 that the constant g is substrate depen-
dent and can be expressed by the Monod equation

S= m S (4)
S concentration of the growth limiting substrate
.m maximum specific growth rate
Km is a constant and the value is very small [e.g., at tempera-
ture of 300C, pH 4.0, with glucose as inhibiting substrate,
Km = (3.6+0.5) x 10-4 mol/L][21

Aerobic Growth and Alcohol Fermentation
The presence of oxygen is necessary for yeast cell multi-
plication, which dispels the general notion that yeast can grow
truly anaerobically.111 This is because, as well as providing a
substrate for respiratory enzymes during aerobic growth, oxy-
gen is required for certain growth-maintaining hydroxylations
such as those involving the biosyntheses of sterols and un-
saturated fatty acids. Oxygen should therefore be regarded
as an important yeast growth factor.
Propagation of yeast cells and production of alcohol by
yeast are two quite different industrial processes. In the first
case, where optimization of respiratory growth is important,
sufficient oxygen must be maintained in bioreactors to sup-
port rapid yeast growth (e.g., in commercial fermentation, 1
volume of air per fermentor volume is passed through the
medium per minute'41). In contrast, production of alcohol by
yeast should be carried out without aeration. In this case, yeast
reproduces while producing ethanol and carbon dioxide. The
yield of yeast, based on the amount of fermentable sugar, is
low-often not more than 10%. But in aerobic systems, the
colonies grow up to 20 times faster than those without aera-
tion. A yield of up to 50% of the weight of fermentable sugar
can be obtained under some special conditions.
Actual levels of dissolved oxygen (D.O.) in a bioreactor
can be determined by oxygen electrodes. They are frequently
calibrated by saturating the fermentation medium (without
yeast) with air and by equating the instrument response with
"air saturation." Instrument readings during the actual fer-
mentation can then be expressed as "% of air saturation." A
useful approximation of the actual amount of oxygen present
in an air-saturated bioreactor is 7 ppm.[41

Microorganism and Medium
Saccharomyces cerevisiae (dried baker's yeast packed for
Goodman Fielder Milling & Baking N.Z. Ltd.) was grown in

YEPD medium51 with the following composition: glucose,
20 g/L; yeast extract, 10 g/L; peptone, 20 g/L; and commer-
cial antifoam, 10 drops/L.
Equipment and Experimental Procedure
Shaker Culture 100 mL of medium was added to a 250-
mL flask. After being sterilized at 1100C for 30 minutes, 1 g
of dried baker's yeast was added, then cultured in the shaker
at 300C and 200 rpm for 60 to 90 minutes. The broth was
sampled at the beginning and at the end of the culture.
Small-Scale Bioreactor Culture 2 L of medium was added
to the 3 L bioreactor (New Brunswick Scientific Co., Inc.)
with a working volume of 2.5 L. The complete assembly was
sterilized by autoclaving at 110C for 30 minutes. The pH
probe (Ingold Electrodes, Inc.) should be calibrated before
sterilization, while the D.O. probe (Mettler-Toledo Process
Analytical, Inc.) should be done after sterilization. 400 mL
of fermentation broth was then inoculated from flasks into
the bioreactor. The agitation rate of the bioreactor was set at
400 rpm. The fermentation process was operated for 4 or 5
hours and sampled every half hour. At the end of the fermen-
tation the air supply was shut off, and the change in D.O. was
recorded every 15 seconds until it reached a constant value.
One of the student groups repeated the experiment three times,
setting the temperature at 28C, 300C, and 350C. The other
group ran it without pH control the first time and then added
sodium carbonate to keep the pH at 6.0 and 5.5, respectively,
for the second and third times. Some of the students tried to
operate in a fed-batch mode.
Analytical Methods
Cell Concentration was measured by three methods: 1) by
counting numbers of the cells using a hemocytometer; 2) by
measuring the wet weight of yeast after centrifuging 10 mL
broth samples for 10 minutes at 4500 rpm and decanting the
supernatant liquid; and 3) by measuring the dry weight of
yeast after dehydrating the remaining pellets in the centri-
fuge tubes at 65 C for 48 hours.
Glucose Concentration was determined by reacting the glu-
cose with Glucose Trinder reagent (Sigma Diagnostics, Inc.)
to yield a red color solution, and the change in color was then
measured by spectrophotometer.


Microorganism Growth Pattern
A series of typical microorganism growth curves was ob-
tained by students from the propagation of baker's yeast in
the bioreactor for 4 or 5 hours. A typical result is shown in
Figure 1. The growth pattern in which an initial lag phase
was followed by an exponential phase and a stationary phase
obviously appeared in all three curves, including cell num-
ber, wet weight, and dry weight against time.
In the wet weight curve, however, many experimental data

Summer 2004

seemed erratic, sometimes in opposition to the increasing trend. This
might be caused by variation in the water content of the remaining
pellets in the centrifuge tubes. In this instance, the average water
content accounted for about 87.5% of the wet weight, and the varia-
tion in the water content between different samples was up to 5%.
For example, if the dry weight of biomass was 3.5 g/L, the wet weight
should be 28 g/L, but the experiment value would vary from 23.33
to 35 g/L with the variation in the water content of 5%. The experi-
mental error was too large to be acceptable.
It was impossible for cells to be absolutely well distributed in the
fermentation broth, so the result of counting cells in 0.1 pL of di-
luted sample could only be statistically representative. In the 4-hour
propagation process when the biomass doubled, the cell number did
not change as much as observed by counting, using the hemocytom-
eter. This might be the reason for differences between the cell num-
ber and the cell weight curves. Another explanation might be the
differences in individual cell weight at different times. As shown in
Figure 1, the increase in cell number seemed to stop after 2 hours,
while the total weight of biomass still went up. The continuous in-
crease of individual cell weight could contribute to the difference. Of
the three methods, dry weights seem to be most accurate, although cell
counts and wet weights give more rapid results.

According to Eq. (3), a plot of In X versus time gave a straight
line with slope R at the exponential phase (X is derived from dry
weights). For this experiment, the specific growth rate was found to
be 0.3 hr' (see Figure 1). The doubling time was 2.34 hr.
Substrate Utilization and the Yield
During the 4-hour propagation of baker's yeast, the glucose con-
centration was seen to decrease corresponding to the growth patterns
(Figure 1). Initially, in the period of the lag phase, the rate of glucose
consumption was slow, and then a steady decrease in glucose concen-
tration was seen until the glucose was completely consumed.
The total yield of yeast (dry weight), based on the amount of glu-
cose consumed, was 0.18 g/g for this experiment. The result was less
than what was expected from the literature.'41 This is because the
specific growth rate in this experiment reached 0.30 hr- higher than
a given value of 0.25 hr-. Above this value, the fermentable sugars
will be fermented into alcohol and the yield of yeast will be reduced.[4'
In order to obtain the maximum yield, the supply of fermentable
sugars must be limited and their concentrations must be extremely
low, generally below 0.0004%.[4] This can be achieved by continu-
ous addition of fermentable sugars to the bioreactor without any re-
moval of fermentation broth, i.e., fed-batch operation mode. Here,
in our laboratory, the fermentation process was operated in fed-
batch mode only once due to the limiting course schedule. Since
no valid conclusions can be drawn from a single experiment,
results are not shown.
Change of D.O. and Oxygen Demand
As shown in Figure 2, over the course of the propagation, dis-
solved oxygen (D.O.) concentration initially increased in the first
half hour and then gradually fell for the remainder of the experi-
ment. D.O. concentration fluctuated at the beginning because the

contents of the bioreactor were changed when the fer-
mentation broth with yeast was inoculated from flasks.
After the lag phase, the consumption of D.O. went up
with increasing cell numbers until the amount of aeration
was less than the demand of D.O. by yeast in the bioreac-
tor, so D.O. concentration was observed to decrease.
In the fermentation process it is very important to keep
a balance between the amount of oxygen supply and de-

x Wet Weight
a Dry Weight x
*Cell Number = x x
A *

y = 1.7248eo.30ox
S R2 = 0.9954

"* { .
1 ,
0 1 2 3 4 5
Time (hr)

Figure 1. Ln (concentrations of glucose, wet weight,
dry weight, and cell number) versus time over the
course of the propagation of baker's yeast in the

140 6.5
120 1 a .
100- .
80 *
d 60
0 "" 5.5
40 D.O.
20 pH
0 ,- 5
0 1 2 3 4 5
Time (hr)

Figure 2. The change of D.O. and pH over the course
of the propagation of baker's yeast in the bioreactor.


80 y=-0.9535x+107.93



0 40 80 120 160 200
Time (second)

Figure 3. The change in D.O. after the air supply was
shut off in the bioreactor.

Chemical Engineering Education

mand. D.O. concentration is always used to judge whether
the oxygen supply is enough, as well as to indicate the state
of cell growth and abnormal changes in the process. The
amount of oxygen required by a given amount of biomass
can be estimated by recording the change of D.O. concentra-
tion soon after the air supply is shut off. A typical result of
this experiment is plotted in Figure 3. Initially, the D.O. con-
centration decreased slowly due to air bubbles remaining in
the fermentation broth continuing to transport oxygen into
the solution. Then the D.O. concentration declined rapidly,
following a straight line when the decrease of D.O. was equal
to the oxygen consumed by yeast. Here, according to Figure
3, the rate of oxygen consumption was about 0.95%/sec. With
an approximation of the oxygen in an air-saturated
bioreactor of 7 ppm, the oxygen required by a given
amount of yeast (dry weight) was 0.041 mmol oxygen/g
of yeast/sec for this experiment.
Change ofpH over the Propagation
Figure 2 also shows the pH readings over the time course
of the experiment. The pH declined from a value of 6.3 to 5.5
over the 4-hour propagation course corresponding to the
growth pattern. Normally during the propagation process,
apart from yeast cells, carbon dioxide, water, and energy in
the form of heat are also produced. Among these, the carbon
dioxide produced could form carbonic acid and result in a
decrease of pH in the fermentation broth.
In commercial fermentation, pH can be partly controlled
by the ratio of ammonia to ammonium sulfate in the feed.
When all the required nitrogen for a fermentation has been
supplied, the pH can be regulated by adding sodium carbon-
ate.141 In our experiment, peptone was applied as a nitrogen
source. The YEPD medium appeared to be a buffer solution.
The pH of the fermentation broth did not change too much
during the 4-hour propagation course without adding sodium
carbonate, ammonia, or ammonium sulfate.

From the propagation of baker's yeast in a 3L bioreactor
for four or five hours, we obtained generally good results of
typical microorganism growth curves in which an initial lag
phase was followed by an exponential phase and a stationary
phase. The glucose consumption, pH, and dissolved oxygen
curves were also observed to correspond to the growth pat-
tern. Among the results from three kinds of methods to deter-
mine the cell concentration in the fermentation broth, dry
weights gave the most accurate values, while cell numbers
and wet weights could be obtained more rapidly. In this ex-
periment, the total yield was less than the value from litera-
ture because the specific growth rate did not fall into the ex-
pected region where the best yield can be obtained. The oxy-
gen demand by a given amount of yeast was also estimated
by recording the change of D.O. concentration soon after the
air supply was shut off.

The fermentation processes were operated in both batch
and fed-batch modes. The effects of pH and temperature for
the propagation of baker's yeast were studied by different
groups of students, but the data were problematic since few
data points were available due to the laboratory's time con-
straints. Further, the requirements for careful attention to de-
tail and sample handling could not be met by every student.
According to the results of the experiment we have run this
year, we suggest that the time of each course of the propaga-
tion might be extended from four or five hours to over eight
hours if there is no strict time constraint for the laboratory.
Therefore, a lower inoculum size would be applied at the
beginning and a much greater percent increase in cell mass
over the course could be observed.
The results for the cell mass concentration might be ob-
served by measuring the optical density (a much easier
method) and using a previously prepared calibration curve.
If a larger bioreactor (for example, with a volume of 30 L) is
employed in the laboratory, the concentration of oxygen and
carbon dioxide in the exit gas could also be measured, and
further, these data could be used to find the mass transfer
coefficients, oxygen uptake, and respiratory quotient.
In summary, processes using living cells are quite different
from those with chemicals and materials. The production of
cells always involves a multistage propagation of a specially
selected strain. An aseptic environment is necessary for the
propagation of pure culture, and contamination should be
avoided as much as possible. The activities of the cells are
affected by many variables and play the most important part
in the bioprocesses. We believe that it is necessary to intro-
duce a meaningful, challenging, but practicable microorgan-
ism propagation experiment to chemical engineering students
who lack any prior exposure to biology. The student response
to the experience has been positive.

K Monod constant
S concentration of the growth-limiting substrate
t time
X biomass concentration
X0 initial biomass concentration
Greek letters
lL specific growth rate
[m maximum specific growth rate
1. Walker, G.M., Yeast Physiology andBiotechnology, John Wiley & Sons,
Chichester (1998)
2. Berry, D.R., I. Russell, and G.G. Stewart, Yeast Biotechnology, Allen
& Unwin, London, UK (1987)
3. Crueger, W., and A. Crueger, Biotechnology: A Textbook oflndustrial
Microbiology, Science Tech, Inc., Madison (1984)
4. Gerald, R., and H.J. Pepper, Yeast Biotechnology, Avi Publishing
Company, Inc., Connecticut (1973)
5. Atkinson, B., and F. Mavituna, Biochemical Engineering and Biotech-
nology Handbook, 2nd ed., Stockton Press, New York, NY (1991) O

Summer 2004

Random Thoughts...



North Carolina State University Raleigh, NC 27695

Student ratings of teaching get a bad rap in some aca-
demic circles. Faculty members are repeatedly and au-
thoritatively assured that "They're just popularity con-
tests," "High ratings go to the easy graders," and "If I get low
ratings it's only because I set high standards and students
don't like demanding teachers."
In fact, student ratings have been repeatedly shown to have
a high level of validity, and those complaints about them have
been debunked by research."t31 Students are in a better posi-
tion than anyone else to judge certain aspects of teaching,
such as how clear, interesting, respectful, and fair a course
instructor is, and they're the only ones who can say how an
instructor has influenced their attitude toward the course sub-
ject, their motivation to learn it, and their self-confidence.
For these and other reasons, student ratings should be con-
sidered an essential component of faculty teaching perfor-
mance evaluation.
But it makes little sense to use only student ratings. Few
students are equipped to judge whether a course is accurate
and up-to-date, the assignments and tests are appropriately
challenging, and the content and learning objectives are con-
sistent with the course's intended role in the department (for
example, to serve as a prerequisite to other departmental
courses or to address certain outcomes in the department's
accreditation plan). Only faculty colleagues are in a position
to make such judgments.
Moreover, classroom teaching may only be a small part of
a faculty member's educational activities. He/she may also
advise students, develop new courses and redesign old ones,
adapt and develop courseware and innovative teaching strat-
egies for use in both traditional classroom instruction and
distance education, coordinate departmental preparation for

accreditation, offer seminars, workshops, consulting, and
mentoring to help faculty colleagues and/or graduate students
improve their teaching skills, write textbooks, and conduct
educational research. All of these activities can have a dra-
matic effect on a department's teaching quality, student re-
tention, and chances of receiving full accreditation, but stu-
dent ratings don't indicate whether and how well an in-
structor is doing them.
In short, a key to effective teaching evaluation is to collect
data from multiple sources (triangulation), making sure that
all education-related activities are rated by the people best
qualified to rate them. Figure 1 presents a multiple-source
evaluation model designed to work that way. The remainder

Richard M. Felder is Hoechst Celanese Pro-
fessor Emeritus of Chemical Engineering at
North Carolina State University. He received
his BChE from City College of CUNYand his
PhD from Princeton. He is coauthor of the
text Elementary Principles of Chemical Pro-
cesses (Wiley, 2000) and codirector of the
ASEE National Effective Teaching Institute

Rebecca Brent is an education consultant
specializing in faculty development for effec-
tive university teaching, classroom and com-
puter-based simulations in teacher education,
and K- 12 staff development in language arts
and classroom management. She co-directs
the ASEE National Effective Teaching Insti-
tute 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 2004

Chemical Engineering Education

of this column briefly elaborates on the model components.

Peer Ratings
The usual form of peer evaluation, in which an observer
visits a lecture and jots down whatever happens to catch his
or her attention, has its own drawbacks. Most obviously, a
single observed class may not be representative of someone's
normal teaching. Even if it is, faculty members have widely
disparate ideas of what constitutes good teaching, so that the
same class could get an excellent rating from one observer
and a poor rating from another. More importantly, a single
class observation provides no assessment data at all on as-
pects of teaching performance other than lecturing.
A far more effective procedure is for two or more review-
ers to use standardized checklists to rate instructional materi-
als and at least two class observations independently and then

to reconcile their ratings.[4] The checklists should consist of
items taken from a list of attributes known to correlate with
effective teaching,5'61 and should be approved by the depart-
ment faculty before they are used. This procedure has a high
level of inter-rater reliability and includes measures to ad-
dress commonly expressed concerns about peer review, in-
cluding possible rater bias and excessive time demands im-
posed on reviewers.[4]

Student Ratings
Tested forms for student evaluation of teaching are given
in a recent National Research Council publication,t71 and more
information about how to make student evaluations effective
is provided in that reference and by Felder.s8' Faculty perfor-
mance evaluations should take into account student ratings
collected over a period of several years, with relatively little

a Including assignments, tests, graded products, & mechanisms for getting student
b Including availability outside class and helpfulness in office hours
c Including research supervision
d Including syllabus, learning objectives, policies and procedures, test & course grades
e Including teaching, advising, mentoring (students and colleagues), developing courses,
creating instructional materials, and educational research. Materials in the last category
should be included in the summary of the faculty member's research, and the rest of the
materials in the figure should be assembled into a teaching portfolio.
f Including letters from students, alumni, local faculty, and faculty at other institutions

Figure 1. Teaching performance evaluation model.

Summer 2004

weight being attached to ratings of someone's first semester
of teaching.

The Teaching Portfolio
Just as some performance assessment data can best be pro-
vided by students and some by peers, certain important in-
formation can only be supplied by the faculty member being
reviewed. Instructors should assemble materials summariz-
ing all of their education-related activities, including devel-
oping new courses and redesigning old ones, developing and
evaluating innovative instructional methods, advising and
mentoring students, writing new texts and courseware, pro-
viding instructional development to faculty colleagues and
graduate students, and carrying out educational research. All
of these materials except those related to educational research
(which we discuss in the next section) should be incorpo-
rated into a teaching portfolio, along with summaries of stu-
dent ratings over the past two or three years, peer ratings,
and reference letters from alumni and colleagues at other in-
stitutions who are familiar with the instructor's educational
activities. The portfolio provides a solid basis for evalu-
ating the faculty member's teaching performance and con-
tributions to education.[9-11

The Scholarship of Teaching and Learning
When done properly, educational research is every bit as
demanding, rigorous, and important to the future of an aca-
demic discipline as traditional disciplinary research.[12] There
is no legitimate reason to separate the two categories of re-
search by making educational scholarship just another com-
ponent of teaching performance, or worse, not to count it at
all in faculty performance reviews. Any material related to
educational research (including lists of grants, publications,
presentations, and awards, along with supporting letters)
should be combined with documentation of disciplinary re-
search in faculty activity reports and in tenure and promotion
dossiers, and the same high standards should be applied to
the evaluation of performance in both research categories.

Consistency of Multiple-Source Ratings
For triangulation to be most effective, data from different
sources should overlap to the greatest extent possible. For
example, items on student rating forms related to aspects of
teaching that both students and peers are equipped to evalu-
ate (e.g., the instructor's preparedness, clarity, responsive-
ness to questions, and respect for students) should parallel
items in peer review checklists. If the two sets of ratings lead
to the same conclusions, it affirms the validity of both, while
if they disagree substantially it suggests that at least one of
the sets is suspect and further investigation should be under-

taken. For example, the department head might bring in
someone from outside the department (such as a consult-
ant from the campus center for teaching and learning) to
conduct focus group interviews with students related to
the issues in question.

Summative and Formative Evaluation
Evaluation of teaching may be summative (to provide data
for use in making decisions regarding reappointment, tenure,
promotion, and merit raises, and for selection of award re-
cipients) orformative (to improve the teaching of the instructor
being evaluated). The full procedure depicted in Figure 1 and
described above should be implemented for summative evalu-
ation. Once the portfolio is assembled, only minor effort
should be required to update it in successive years. For for-
mative evaluation, a subset of the procedure should be car-
ried out (for example, only one peer rater may be used), and
the results should be shared only with the instructor rather
than being passed on to the department head or a performance
review committee. Carrying out formative reviews in the
first few years of a faculty member's career should sub-
stantially increase the chances that a subsequent
summative review will be favorable.

1. Cashin, W.E., "Student Ratings of Teaching: The Research Revisited,"
IDEA Paper No. 32, Kansas State University Center for Faculty Evalu-
ation and Development, , September (1995)
2. Felder, R.M., "What Do They Know, Anyway?" Chem. Eng. Ed., 26(3),
134 (1992),
3. McKeachie, W.J., "Student Ratings: The Validity of Use," Amer. Psy-
chologist, 52(11), 1218 (1997)
4. Brent, R., and R.M. Felder, "A Protocol for Peer Review of Teach-
ing," Proc. 2004 An. ASEE Meet., ASEE, June (2004),>
5. Chism, N. Van Note, Peer Review of Teaching, Anker Publishing,
Bolton, MA (1999)
6. Weimer, M., J.L. Parrett, and M. Kems, How am I Teaching? Magna
Publications, Madison, WI (1988)
7. National Research Council, Evaluating and Improving Undergradu-
ate Teaching in Science, Technology, Engineering, and Mathematics,
Washington, DC, National Academies Press (2003)
8. Felder, R.M., "What Do They Know, Anyway? 2. Making Evalua-
tions Effective," Chem. Eng. Ed., 27(1), 28 (1993),>
9. Seldin, P., The Teaching Portfolio: A Practical Guide to Improved Per-
formance and Promotion/Tenure Decisions, 2"d ed., Anker Publishing
Co., Bolton, MA (1997)
10. Edgerton, R., P. Hutchings, and K. Quinlan, The Teaching Portfolio:
Capturing the Scholarship in Teaching, American Association for
Higher Education, Washington, DC (1991)
11. Felder, R.M., "If You've Got It, Flaunt It: Uses and Abuses of Teach-
ing Portfolios," Chem. Eng. Ed., 30(3), 188 (1996)>
12. Huber, M.T., and S. Morreale, eds., Disciplinary Styles in the Scholar-
ship of Teaching and Learning: Exploring Common Ground. AAHE/
Carnegie Foundation for the Advancement of Teaching, Washington,
(2002) 0

Chemical Engineering Education

2 book review

DISCUSSION OF THE METHOD: Conducting the Engineer's Approach to Problem Solving
Billy Vaughn Koen, Oxford University Press, 272 pages, $60 (2003)

Reviewed by
Dendy Sloan
Colorado School of Mines

This monograph changes one's views about the engineer-
ing process. More importantly, the thesis of the monograph
is all-encompassing-the engineering process is normally ap-
plied by everyone in our everyday life.
In 1637, Rene Descartes' book with a similar title effec-
tively began modem philosophy. Discours de la Methode was
one of the first statements of the scientific method, where the
last three steps (analysis, synthesis, and evaluation) recur in
the modem higher-order thinking skills of Bloom's pedagogi-
cal taxonomy.1" Descartes suggested that his method could
be applied to an individual's life perspective. Potentially,
Koen's book could have a comparable impact.
Discussion of the Method: Conducting the Engineer's Ap-
proach to Problem Solving has six chapters:
1. Situations calling for engineering talents are described
in this chapter.
2. In this chapter the engineering method is defined as the
use of heuristics to bring about the best change in a
poorly understood situation within the available
resources. A heuristic, sometimes called a rule-of-
thumb, is defined as a guide to action that has four
a) It does not guarantee a solution
b) It may contradict other heuristics
c) It reduces the search time to solve a problem
d) Its acceptance depends on the immediate context
rather than on an absolute standard.
3. Examples of heuristics the engineer uses are given in
this chapter. While there are 59 heuristics in Koen's
book, four from engineering design are
a) Allocate resources as long as the cost of not
knowing exceeds the cost of finding out
b) Allocate resources to the weak link
c) Work at the margin of solvable problems
d) Make small changes in the state of the art
Chapter three also provides examples from general
engineering practice related to heuristics, such as the
Embarcadero Freeway, the Tacoma Narrows Bridge,
and the Golden Gate Bridge. Chemical engineering

Summer 2004

heuristic examples, mostly from Austin professors like
John McKetta or Matthew van Winkle, include
"Engineers always give an answer," or "In the absence
of other information, a distillation column should have
20 stages."
4. In chapter four, the engineering method is generalized
to the universal method for causing any change under
5. A summary of the engineering method is given in this
6. In Chapter 6 an application of the method to construct
a new society of learning or "Eutopia" is found.
The book is as interesting as it is challenging. After read-
ing a draft in 1982, and rereading it several times since, I
recommend that Koen's book be read in two parts. The first
part (Chapters 1, 2, and 3) defines the engineer and the engi-
neering method, illustrating the concepts of heuristics and
state-of-the-art, and how they distinguish engineers. This
first part can be read at a single sitting in a few hours, to
get the central concept of Koen's definition of the engi-
neering method.
This first part of Koen's book was published by ASEE in
1985 as Definition of the Engineering Method. It relies upon
and enhances our engineering education background. For this
first part alone, the book is valuable. It provides an educa-
tional guide to engineering students who want to understand
what it means to be an engineer.
The second part of the book (Chapters 4, 5, and 6) extends
the engineering method to a universal method, to be used by
everyone (including scientists) in all aspects of life. Com-
pared to a definition of engineering, this second part is a much
more daunting challenge. The author rightly feels the need to
briefly compare this method to all preceding Western (and
some Eastern) thinkers who proposed a universal method,
beginning with the pre-Socratic lonians and proceeding
through modem authors such as Goodman, Kuhn, Popper,
and Wittgenstein.
In his second part, Koen departs from Descartes, who used
reason alone, without citations or footnotes, because Descartes
Continued on page 211.



University of Kentucky Paducah, KY 42002
evolutionary Operation (EVOP) is a statistical tool de-
veloped for incrementally moving the operation of a
dynamic process in the direction of some optimum
set of conditions. The EVOP methodI" was introduced in the
late 1950s as a field application technique for improving ex-
isting industrial processes. It was to be applied to an existing
manufacturing process that was currently producing accept-
able product. By exploring small incremental changes in an
existing set of process conditions, the process could be im-
proved and moved in the direction of some process optimum.
There are other advanced statistical methods, such as strat-
egies of experimentation,[21 simplex optimization,[3] response
surface methodology,[4] and advanced factorial design,151 but
they are more complex and require a great deal of training
for reliable application and interpretation. Most of the
methods deal with an initial strategy of experimentation
when formulating a set of bench-scale experimental runs.
The goal of experimental design is to minimize the num-
ber of runs while at the same time maximizing the amount
of useful information.

EVOP is a simple technique that is relatively easy to apply
and provides intuitive, yet statistically based results. In the
chemical engineering undergraduate laboratory at the Uni-
versity of Kentucky, students operate a carbon dioxide scrub-
ber to gain training in using the EVOP method. Not only do
they acquire knowledge of how to design a scrubber, but they
also learn how the interplay of various operating parameters
affects the overall scrubbing performance of the device. De-
tails of this student experience were previously delivered at

the annual ASEE conference in 2003.[6]
Typically, a gas scrubber is a packed column that uses liq-
uid media such as water to absorb and remove contaminants
from polluted industrial gas streams. This scrubbing process
often serves as a final process step prior to release of the
"clean" air into the environment. Chemical engineering stu-
dents in the undergraduate curriculum learn to design packed
columns for use as air pollution control devices. Most cur-
ricula include some hands-on training with these devices in
the laboratory environment. The design techniques learned
by students are approximate methods based on such operat-
ing variables as column-pressure drop, packing factors, and
mass transfer coefficients. These approximations often serve
as a fundamental basis for design and construction of air pol-
lution control devices, but in reality, final optimization and
fine-tuning are often performed by engineers in the field on
already-installed operational equipment. The EVOP technique
is ideally suited for optimizing existing equipment operation.
In addition to EVOP, some investigators have successfully
applied other advanced statistical methods to the dynamic
optimization of a packed gas absorber.[7]

Jimmy Smart is Associate Professor of
Chemical and Materials Engineering at the
University of Kentucky. He received his BS
chemical engineering degree from TexasA&M
and his MS and PhD from the University of
Texas at Austin, all in chemical engineering.
He has over 20 years industrial experience
with companies such as IBM and Ashland
Chemical. His research areas include appli-
cations of membranes to purify watersupplies
and treatment of hazardous wastes.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

oratory )




To Optimize Gas Absorber Operation

(A Statistical Method for Process Improvement)

The purpose of this laboratory exercise is to introduce students to the use and application of
the EVOP method.... Not only do they acquire knowledge of how to design a scrubber,
but they also learn how the interplay of various operating parameters
affects the overall scrubbing performance of the device.

The purpose of this laboratory exercise is to introduce stu-
dents to the use and application of the EVOP method. Using
sodium hydroxide solutions to scrub CO2 emissions is not
commonly found in industry-aqueous solutions of potas-
sium hydroxide or amines, in conjunction with arsenite cata-
lysts, are usually more desirable from an economic stand-
point.J8 Sodium hydroxide solutions were selected for this
exercise, however, because of their ready availability in
educational laboratories.
In this exercise, students are presented with a packed scrub-
ber that is currently being used to remove CO2 from a simu-
lated industrial stack gas. Gas flow rate, CO2 inlet concentra-
tion, and column diameter are fixed operational parameters.
Existing conditions include use of once-through ambient water
flowing countercurrent to the gas flow in a column packed
with spherical packing material. Students use principles of
EVOP to optimize column performance by selecting appro-
priate column packing, liquid recirculation rate, caustic con-
centration, and temperature. Column performance in this ex-
ercise is defined as lowest CO2 emission, not lowest cost.

Since EVOP's introduction in the late 1950s, many books
and journal articles have been published discussing the
method. This article is not intended as a survey review of
EVOP; the reader is invited to consult the original publica-
tion or other excellent discussions9- 121 for detailed informa-
tion and its applications. The purpose of this article is to in-
troduce undergraduate students in the chemical engineering
laboratory curriculum to EVOP and to provide a procedure
and list of equipment for faculty who might wish to set up a
similar experiment. This introductory EVOP exercise pro-
vides a basis for further study leading to response surface
theory and culminating in contemporary quality strategies in
a manufacturing plant environment, such as Total Quality
Managementtl13 and Six Sigma.114]
In a research laboratory environment, strict requirements
for formulating strategies of experiments can usually be sat-
isfied. Usually, all independent and dependent variables can
either be measured or carefully controlled. Principles behind
orthogonal design, generation of response surfaces, random-
ization, experimental replication, and factorial designs can
be successfully met. In a manufacturing plant environ-
ment, however, there are forces at work not subject to

control by operating personnel, including economic fac-
tors, product demands, and other undefined influences.
Quite often, orthogonal designs are not compatible with
production requirements.
EVOP is a procedure designed to meet the needed
flexibilities inherent to the plant environment. It should be
emphasized that EVOP is a routine method for permanent
process operation, not an experimental procedure. It is to be
applied in an existing plant operation rather than used at the
pilot/laboratory scale. It was developed to avoid undesirable
characteristics of full-scale process experiments that require
specially trained personnel and the subsequent produc-
tion of off-specification product. EVOP requires no spe-
cial staff and can be used by existing plant operators after
a brief training period.
Plant operators find EVOP appealing because of its intui-
tive approach. EVOP philosophy says to explore the effects
of process variables near current operating conditions and
make adjustments that will drive the process in a direction
that offers improvement, whether it be quality, reduced cost,
greater output, or less waste. Another added bonus behind
implementation of EVOP is that it improves overall under-
standing of the process itself. Plant personnel gain a better
understanding of the effects of process variables upon prod-
uct quality. Also, subtle effects are often discovered that were
not previously known to exist.


The basic idea behind using EVOP is to improve the sig-
nal-to-noise ratio of an existing process in an effort to un-
cover relationships between operating variables. The signal
is increased by deliberately introducing carefully chosen
minor variations about an existing operational point, called
the "works process." Noise within the process arises from a
variety of sources, such as variability of raw materials, in-
ability to precisely control process inputs, and instrument and
measurement error. The final variation in the product yield is
a composite of all these sources. The magnitude of the varia-
tion is measured by the standard deviation.
The first step in implementing EVOP is to identify perti-
nent process variables associated with an existing process.
Then, a cycle of process runs is designed around the normal
or existing values of the process variables. We deliberately
introduce small changes in the process signal or process out-

Summer 2004

puts and investigate their effects. Differences between nor-
mal and proposed values are kept necessarily small to avoid
production of off-specification product. Generally, it is im-
practical to investigate the effects of more than three vari-
ables in an industrial process, so a 23 factorial design is
arranged. It investigates the effects (response) of low and
high levels of three process variables. Cycles of runs are
repeated to replicate operational conditions and to reduce
experimental error. Here, we reduce the noise level in the
process by repeatedly measuring the process output at a
fixed set of operating conditions.
Assume a 23 factorial design is set up around our first
phase (ambient temperature), where the variables to be stud-
ied are recirculation rate, caustic concentration, and pack-
ing material. A "phase" is defined as a set of variables to be
tested, and it forms the cubic geometry of the 23 factorial
design, as shown in Figure 1. One complete "cycle" is de-
fined as a complete collection of process runs from point 1
to point 8 of the cube (phase). Anywhere from three to six
cycles are run for each phase to provide a valid statistical
analysis. The output response to be optimized is the % CO2
concentration in the stack gas. From Table 1, the experi-
mental runs are set up to explore the low and high varia-
tions of these variables and the output responses for each
set of conditions are noted at each apex of the cube
shown in Figure 1.
Once a factorial design is arranged, it is normal statisti-
cal practice to randomize the order of the runs within each
cycle. Randomization ensures that if systematic trends oc-
cur from untested variables, these effects will not be mis-
taken for effects from deliberately introduced changes in
tested variables. Randomization also validates our analy-
sis that assumes that errors within cycles are independent
of each other. In an actual plant manufacturing environ-
ment, it is difficult to organize a random-run sequence, but
by following different run sequences over the course of
various cycles, randomization is assured.
After four cycles are completed for phase one, the re-
sults are averaged for each variable effect (actual results
are shown in Table 1). The main effects and interaction
between effects are determined from using the averages at
each apex of the cube shown in Figure 1.

Figure 1. 23 factorial design for phase one.

Eight Sets of Conditions of a 23 Factorial Design
(Phase I containing 4 cycles)

A B C Average CO,
Run Recirculation Caustic Packing in Stack Gas
Rate/ Concentration Type % by volume
Umin %by wt

1 2.5 0.0 #1 9.87
2 4.5 0.0 #1 8.97
3 2.5 1. #1 7.45
4 4.5 1.0 #1 7.28
5 2.5 0.0 #2 9.90
6 4.5 0.0 #2 9.93
7 2.5 1.0 #2 610
8 4.5 1.0 #2 5.18

Analysis of Main Effects and Interactions

std. recirc/ caustic packing
mean dev. temp

Phase 1 8.09 1.77 -0.49 -3.17 -0.62 -0.06 0.05 -1.11 -0.42 +1.25
Phase2 4.64 1.53 -3.31 -2.11 1.56 -0.33 -0.60 -0.26 -0.23 1.08

Chemical Engineering Education

Main effects are calculated as

A= 14(Y2 +Y4 +6 +Y8)- 14(YI +Y3 +5 + 7)
B= 4(Y3 +y4 + 7 +8)- 4(Y1 + 2 +Y5 +Y6) (1)
= /4(y5 +6 +Y7 +8)- 4(Y + Y2 +Y3 +Y4)

Two-factor interactions are calculated as

AB= 14(Yi + 4 + Y5 + Y8)- 4(Y2 + 3 + Y6 + Y7)
AC = 4(y + 3 +6 +8)-~4(2 + 4 +5 +7) (2)
BC= 4(1 +Y2 +Y7 +8)- 4(Y3 +4 +Y5 +6)

Three-factor interaction is calculated as

ABC= /4(Y2 +Y3 + 5 + Y)-14(Y1 + Y4 + 6 + 7) (3)
Results from these calculations are summarized in Table 2.
A negative sign on the main effect or interaction is desir-
able in this exercise because it indicates a reduction in %

CO2 stack gas emissions.
After main effects and interactions have been tabulated,
we must use a statistical tool to help us decide what variable
effect is significant, i.e., what response is above the "noise
level" of the process. Traditional statistical tests are not ap-
propriate in assessing the uncertainties associated with EVOP
because of the small number of observations. Instead, the
most practical way to evaluate EVOP uncertainty has been
found to be the use of two standard-error limits. A standard
error (S.E.) is the estimated standard deviation of the vari-
able of interest. If the true standard deviation, cr, of the pro-
cess variable was known, 2 S.E. limits would represent ap-
proximately a 95% confidence limit. The true standard de-
viation is not known, but an estimate of the standard devia-
tion, s, can be calculated. We use this estimate to formulate
our 2 S.E. limits to guide us in interpreting what effects
and interactions are significant. Variances and standard er-
rors can be calculated from Table 3.~[1 An estimate of the true
variance, U2, is calculated as

n(yi -y)2
s2 i= (4)

where s2 is an estimate of the true variance, y, is an individual
observation, y is the mean, and n is the total number of ob-

Figure 2.
Gas Absorption Column
1. Recirculation sump (40-L capacity)
2. Liquid recirculation pump
3. Air blower
4. Rotameter for recirculated liquid (1-10 L/min)
5. Packing material (clear plastic column)
6. Scrubber liquor inlet
7. Scrubber discharge stack (gas outlet)
8. Pressure taps for checking Ap across the packed column
9. Rotameter for CO, (0-20 L/min)
10. Rotameter for air (20-180 L/min)

Summer 2004

Variances (a') and Standard Errors for Main
Effects and Interactions Estimated from a 2P
Factorial Design after c Cycles with
Estimate of the Standard Deviation, s.

Design 2P 22 23
Variance 4c2/c2P /2/c -2/2c
Standard Error 2s/(c2P)"~ s/cl" s/(2c)"2

servations. An estimate of the true standard deviation is just
]s2. For a 23 factorial design, the standard error of effects is
calculated as s/12c. Therefore, the 2 S.E. limits for the ef-
fects in our design are provided by

effect 2i (5)

Final calculated results for standard error limits are shown in
Table 2. Data collected from phase two of this exercise where
temperature, caustic concentration, and packing material were
evaluated, are shown in Table 4.
NOTE: In the interest of reducing student laboratory time to
a reasonable period, the EVOP method in this laboratory
exercise has not been strictly followed in that
Larger ranges of temperature (+130C) were selected to
clearly demonstrate their effect on scrubbing effi-
ciency. In a plant environment, smaller temperature
ranges (5 C) would probably be selected for each
factorial design (phase). Remember that one of the
advantages of the EVOP method is the generation of
minimal quantities of off-spec product.
In a plant environment, four cycles would typically be
used to average the output response. Instead of cycling
through all eight apexes of the factorial cube before
beginning the second cycle, four distinct samples for
each given set of experimental conditions were
The run order was not strictly followed. For example,
in phase one, in order to delay use of caustic, we
followed the run order of 1-2-5-6-3-4-7-8.

An existing experimental laboratory scrubber package by
Armfield, Ltd., (see Figure 2) was modified for this exercise.
The column is made of clear plastic with dimensions of 9 cm
OD by 1.7 m tall. The overall column consists of two packed

Type #1:
Y2" diameter
polypropylene ball

sections, each having a packed bed depth of 55 cm and one
liquid redistributor. Three types of packing were offered
to students for optimization purposes: 1/2" (1.3 cm) di-
ameter polypropylene balls, 3/8" (1 cm) glass Pall rings,
and 3/16" (0.5 cm) Jaeger stainless steel slotted rings (see
Figure 3 and Table 5).
During operation of the scrubber, CO, is supplied from a
standard gas cylinder, regulated through a rotameter, and
mixed with air from an air blower to provide a 10% by vol-
ume mixture of CO2 in air (compare this to CO2 emissions
from coal-fired power plants that are typically 14%). This
gas mixture is routed to the bottom of the packed column and
allowed to flow upward through the column. Scrubber liquor
enters the top of the column and flows downward through
the packed bed where it acts to absorb (scrub) CO2 from the
gas stream. "Clean" gas exits through the top of the scrubber.
The purpose of this exercise is to find the combination of
operating parameters to minimize the percent CO2 leaving
the scrubber stack.
An inexpensive ($350) Bacharach Fyrite gas analyzer was

Type #2: Type #3:
3/8" glass pall ring 3/16" Jaeger
stainless steel
slotted ring

Figure 3. Random packing used for scrubber internals.

Chemical Engineering Education

Eight Sets of Conditions of a 23 Factorial Design
(Phase 2 containing 4 cycles)

A B C y, Avg. CO2
Run Temperature Caustic Packing in Stack Gas
TC Concentration Type % by volume
% by wt
1 12 1.0 #2 5.05
2 38 1.0 #2 4.53
3 12 2.0 #2 3.30
4 38 2.0 #2 2.58
5 12 1.0 #3 7.25
6 38 1.0 #3 5.98
7 12 2.0 #3 5.43
8 38 2.0 #3 3.05


used to measure CO, concentrations in air for both scrubber
inlet and exit. Other, more expensive, gas analyzers and ana-
lytical equipment provide greater accuracy, but 0.1 vol %
was adequate for the purposes of this exercise. Precision, or
repeatability, under experimental conditions proved to be
8%. The Fyrite provides a quick and easy method for mea-
surement of CO, in air. It employs the Orsat method of volu-
metric analysis involving chemical absorption of carbon di-
oxide into a potassium hydroxide solution. A rubber bulb is
used to draw the gas sample into the indicator solution. The
instrument is inverted and the percentage of gas absorbed by
the Fyrite fluid is immediately read from the scale (0-20%).
Temperature of the scrubber recirculation sump was controlled
by a bath circulator (NESLAB Instruments, RTE-100) fitted
with an external cooling coil.
Teams of three to five students can be formed to complete
this exercise. Safety procedures should include familiarity
with the NaOH Material Safety Data Sheet (MSDS) and Fyrite
device, location of the eyewash/shower, and wearing appro-
priate personal protective equipment (PPE) for handling caus-

Properties of Random Packing

Packing Surface Pressure drop:
Packing Type pieces/cm3 arealvol Ap/L (cm water/cm)
(cm2/cm3)@ 2.5 ml liquid/min

#1: 1/2"-diameter P/P sphere 4.2 21.3 0.16
#2: 3/8" glass Pall ring 5.5 35.2 0.06
#3: 3/16: Jaeger s.s. slotted ring 18.1 49.0 0.05

36 o 1% caustic scrubbing liquor at 120 C
34 2% caustic scrubbing liquor at 38 C
2 24
E 20
0 18
1 16
0' 14
0 12

4 -
05 10 15 20 25 30 35 40 45 50 55 60 65 70 75
time, min
Figure 4. Unsteady-state nature of scrubber performance:
recirc rate, 4.5 L/min, #2 packing (lines
added to guide the eye).

tic. The first set of data (varying concentration at ambient
temperature) can be collected in one afternoon, but the more
lengthy elevated/reduced temperature settings require comple-
tion on subsequent days.
After students become familiar with the equipment and have
characterized existing scrubber performance, they are ready
to apply the EVOP method. A 23 factorial experimental ma-
trix is set up to measure the response (level of CO2 emis-
sions) from small changes in operating variables (liquid flow
rate, caustic concentration, and temperature). Standard error
limits are tabulated to build response surfaces to judge posi-
tive and negative effects of parameter changes for a given
packing material. Results from these small changes are
used to guide the students in making judgments as to what
direction parameters should be adjusted to achieve a final
process optimum.
The final submitted report should include safety and op-
erational procedures, collected data, calculations, results,
sources of experimental error, and a discussion that includes
which direction the final process optimum will lie and how
to proceed to locate this optimum.


In actual scrubbers at industrial power plants, Ca(OH)2,
called slaked lime, is used to remove the hazardous air pol-
lutant, sulfur dioxide, from stack gases. In this experiment,
another strong base, NaOH, is used to remove CO2 from our
simulated stack gas. The primary reactions that occur during
the scrubbing process are

CO2(g)+ H20(0) = H2CO3()
NaOH(s)+H2CO3(0) t NaHCO3(s)+ H20(e)
2 NaOH(s)+ H2CO3 () Na2CO3(s)+ 2 H20(t)

In Eq. (6), a weak dibasic acid, carbonic acid, is formed when
carbon dioxide is mixed with water. As the caustic solution
(a strong base) contacts the carbonic acid, either sodium bi-
carbonate is formed from Eq. (7) or sodium carbonate from
Eq. (8). The reactions can be found as titration curves in any
standard analytical chemistry textl"51 and can be monitored in
the scrubber with a pH probe.
In the absorption process, the overall rate is governed by
diffusion and chemical reactions occurring in the liquid phase.
The reaction is a pseudo first-order reaction between dissolved
CO2 and OH in the liquid and is of the same order of magni-
tude as the rate of diffusion."61
For an industrial scrubber, a tank supplying scrubber me-
dium typically contains several thousand gallons. Unfortu-
nately, the working volume of the small recirculation sump
in our experimental apparatus is only about 38 L. Based on
an assumption of plug flow at a recirculation rate of 4.5 L/
min, where the pump suction and recirculated liquor return

Summer 2004

are located at opposite ends of the sump, a working time of
only 8.5 minutes is available. The unsteady-state nature of
the scrubbing action is charted in Figure 4. Sure enough, for
about 8 minutes, the percent CO2 by volume in air is 5.0.
After this period, the percent CO2 slowly climbs to a steady-
state value of 8.8 after 44 minutes. During the initial 8 min-
utes, the sodium carbonate salt (Eq.8) is probably being
formed. After the initial period there are probably equilib-
rium competitions between Eqs. (7) and (8), until a steady-
state condition is attained.
As previously mentioned, repeatability of experimental
measurements with the gas analyzer was 8%. This value
would be viewed as high in a laboratory setting, but is prob-
ably a realistic value in a plant environment. The lack of re-
peatability was not due to the instrument itself, but was pri-
marily due to gas-flow drifting fluctuations in the rotameter,
Joule-Thompson effects of gas expansion across the CO2 cyl-
inder regulator, and observed channeling effects of liquid flow
in the packed column.

The main effects and interactions between process vari-
ables are summarized in Table 2. In phase 1, recirculation
rate (A), % caustic (B), and packing material (C) were evalu-
ated. From consideration of each effect 2 S.E., it appears
the increase of caustic from 0% to 1% had a strong negative
effect (decreasing the percent of CO2 in the stack gas, which
is a desirable outcome). The other effects are not significant
since they are below the error limits and can be considered to
be within the noise of the process. One surprising result is
the fact that the main effect (A) of the recirculation rate was
not significant. There was no advantage to increasing the recir-
culation rate from 2.5 to 4.5 L/min. Evidently, the scrubbing
effect is not diffusion-rate limited, but reaction-rate limited.
In phase 2 of Table 2, the temperature (A), % caustic (B),
and packing material (C) were evaluated. From consideration
of each effect 2 S.E., it appears only the main effects of
temperature (A), % caustic (B), and packing material (C) were
significant. The result of increasing the temperature and %
caustic had a strong negative effect (decreasing the % CO2 in
the stack gas). On the other hand, switching the packing from
Pall rings to more open stainless steel slotted rings had a strong
positive effect (increasing the % CO2 in the stack gas). This
is a somewhat surprising result, since the slotted rings offer
more surface area/volume and less pressure drop (see Table
5). This can be explained (as was shown from results of phase
1) by the fact that the scrubbing effect is limited by reaction
rate, not the diffusion rate. The slotted rings offer more open
geometry (less Ap) and therefore less liquid holdup within
the packed beds. With glass Pall rings, there is more liquid
holdup within the column, which favors reaction between CO2
to form the carbonates. In the case of packing #1 (P/P spheres),
this geometry offers a much reduced surface area/volume,

which lowers the reaction rate between the gas and liquid
phases within the column.

From this abbreviated application of EVOP, we can draw
some conclusions as to what direction the optimum for this
scrubbing process might lie. The optimum will reside in a
direction of elevated operating temperature and higher % caus-
tic in the scrubbing liquor. Packing material #2 (Pall rings) is
more desirable in decreasing the overall % CO2 in the stack
gas, but is just slightly out of the error limits. In the final
analysis, packing #3 (slotted rings) may be a more favorable
choice because of the lower operating cost (less Ap/length of
packing height).
From the EVOP analysis, use of higher % caustic reduced
CO2 emissions from our scrubber stack. Where might the
overall process optimum reside? From the MSDS, sodium
bicarbonate is soluble in water to about 8% b.w. at 180C.
Therefore, this condition would be the limiting factor on what
maximum concentration of NaOH to use in the scrubbing
liquor. Anything above about 8% to 10% would cause salt-
ing-out of solids that would foul and possibly occlude scrub-
ber packing. It is interesting to compare the main effect of %
caustic for phase 1 and phase 2. Moving from 0% caustic to
1% had a very strong negative effect (CO2 reduced in stack
gas), whereas moving from 1% to 2% had a less strong nega-
tive effect. The response surface is not linear. This situation
calls for additional investigation, as a process optimum may
reside somewhere between 2% caustic and the recommended
maximum limit of 10% caustic. Other investigatorsE'71 have
found the optimum mass transfer coefficient to reside at a
2M NaOH solution (about 8% b.w.).
Where might the elevated process optimum temperature
lie? A cost analysis would have to be performed on a better-
defined response surface to identify optimum temperature.
In a plant environment, unless low-pressure waste steam is
available, the energy costs to heat the scrubber liquor is prob-
ably not justified. One point to consider, however, is that usu-
ally inlet stack gases fed to scrubbers (especially those from
power plants) are often at elevated temperatures. Another al-
ternative is to add 50% b.w. caustic with enough water in a
mixing tee to form a 10% b.w. caustic solution just prior to
its introduction to the scrubber. This method would allow
some elevation of temperature due to heat of solution.
This exercise has been prepared to provide an undergradu-
ate student with experience in the use and application of the
EVOP method in a laboratory environment. As demonstrated,
the student can set up additional variable ranges to be tested
for phases 3 and 4, and so forth, to identify a final optimiza-
tion of the overall scrubbing process.

1. Box, G., and N.R. Draper, Evolutionary Operation, John Wiley & Sons,

Chemical Engineering Education

New York, NY (1969)
2. Box, G.E.P., W.G. Hunter, and J.S. Hunter, Statistics for Experiment-
ers, John Wiley & Sons, New York, NY (1978)
3. Walters, F.H., S.L. Morgan, L.R. Parker, and S.N. Deming, Sequential
Simplex Optimization, CRC Press, Boca Raton, FL (1991)
4. Montgomery, D.C., Design and Analysis of Experiments, John Wiley
& Sons, New York, NY (2001)
5. Box, G., W.G. Hunter, and J.S. Hunter, Statistics for Experimenters,
John Wiley & Sons, New York, NY (1978)
6. Smart, Jimmy L., "Use of an Applied Statistical Method to Optimize
Efficiency of an Air Pollution Scrubber Within an Undergraduate Labo-
ratory," ASEE Conf. Proc., Nashville, TN, June (2003)
7. Hoerner, G.M., and W.E. Shiesser, "Simultaneous Optimization and
Transient Response Evaluation of Packed-Tower Gas Absorption,"
Chem. Eng. Prog. Symp. Ser, 61(55), p. 115 (1965)
8. Danckwerts, P.V., and M.M. Sharma, "The Absorption of Carbon Di-
oxide into Solutions of Alkalis and Amines," The Chem. Engr., p.
244, October (1966)
9. Klingel, A.R., and R.G. McIntyre, "An Experimental Strategy for In-

vestigating Commercial Processes," Appl. Statistics, 11(2), p. 79, June
10. Spendley, W., G.R. Hext, and FR. Himsworth, "Sequential Applica-
tion of Simplex Designs in Optimisation and Evolutionary Operation,"
Technometrics, 4(4), p. 441, November (1962)
11. Carpenter, B.H., and H.C. Sweeny, "Process Improvement with Sim-
plex Self-Directing Evolutionary Operation," Chem. Engr, p. 117, July
5 (1965)
12. Scarrah, W.P., "Improve Production Efficiency via Evolutionary Op-
eration, Chem. Engr., p. 131, December 7 (1987)
13. Schmidt, S.R., M.J. Kiemele, and R.J. Berdine, Knowledge-Based
Management, Air Academy Press, Colorado Springs, CO (1997)
14. Breyfogle, F.W., hnplementing Six Sigma: Smarter Solutions Using
Statistical Methods, John Wiley & Sons, New York, NY (1999)
15. Fritz, J.S., and G.H. Schenk, Quantitative Analytical Chemistry, 4th
ed., Allyn & Bacon, Boston, MA, p. 184 (1979)
16. Sherwood, T.K., and R.L. Pigford, Absorption and Extraction,
McGraw-Hill, New York, NY, p. 358 (1952)
17. Tepe, J.B., and B.F. Dodge, Trans. Am. Inst. Chem. Engrs., 39 (1943)

Book Review: Discussion of the Method
Continued from page 203.

did not wish to appeal to Greek authority. Readers must ei-
ther accept Koen's synopsis of such major thinkers as Kant,
Godel, and Wittgenstein, or they must be conversant with the
history of Western thought, principally in philosophy. This
requires either an abstraction of the thinkers cited, or many
years of reading.
William Perry121 suggested that the intellectual development
of college students consists of stages that progress from au-
thoritarian dualism (Stages 1 through 3) through the slough
of relativism (Stages 4 through 6) to committed action (Stages
7 through 9), and indicated that relativism is where many
college students get stuck. The relative equality of opinion
and the absence of authority lead to the lack of commitment
that is sometimes predominant in education today.
As applied to pedagogy, Koen's book also suggests that a
departure from authority (dualism) is good, but going beyond
relativism is better. Koen's method for doing so is through
heuristics. Heuristics, or general rules-of-thumb, are particu-
larly important guides in the absence of absolutes.
Koen's book provides some guidance in dealing with am-
bivalence of contrasting heuristics, often incorporated in
society's aphorisms. How does one balance the contrasting
heuristics of "Look before you leap," with "He who hesitates
is lost"? It is clear that the triage advice, "When you hear
hoofbeats, think horses not zebras," has a geographic limita-
tion-it applies more in the Western world than Africa. Ac-
cording to Koen, contradictions require judgment to obtain a
basis for action, to get beyond relativism.
In the score of years since its original publication, Koen's
ASEE book has been used in a freshman honors seminar
"Paradoxes of the Human Condition," with between 12 and

15 students per year, in an effort to find a way beyond Rela-
tivism in the Absolute versus Relative paradox. One week
the students discuss the Absolute through Descartes' effort to
break from Greek authority. The following week the students
discuss Koen's Definition of the Engineering Method in par-
allel with Perry's model of intellectual development.
Our students' essays indicate that a study of Koen's heuris-
tics initiates progress away from a Relativistic position. In
other words, even though an absolute is not known, heuris-
tics show the way to take appropriate action, or to choose
between two actions. As such, the students readily embrace
Koen's perspective of heuristics, and their combination into
a state-of-the-art, or paradigm.
Professor Koen's book suggests a startling, explicit state-
ment of a new way to think about engineering and life, but a
method which may already be implicit in the subconscious
of most practicing engineers. If we, as educators, wish to pre-
pare our students for engineering practice, the techniques in-
dicated in this book provide a philosophical underpinning
for dealing with risks associated with engineering actions and
designs, when there is insufficient applicable science. The
interesting extension of Koen's engineering philosophy to life
is, at a minimum, worthy of our consideration.


1. Bloom, B.S., Taxonomy of Educational Objectives.
Handbook I. Cognitive Domain, Addison-Wesley Pub-
lishing Co. (1984)
2. Perry, William O., Intellectual and Ethical Development
in the College Years: A Scheme, Holt, Rinehart & Win-
ston, New York (1968) 0

Summer 2004

MR% laboratory

Laboratory Experiment on


For Chemical Engineering Students

San Jose State University San Jose, CA 95192-0082

Biochemical engineering is the application of chemi-
cal engineering principles to biological processes.
Processes for the production of fine and commodity
chemicals that are catalyzed by whole cells or enzymes are
developed and analyzed through the principles of biochemi-
cal engineering. Surveys of the industrial employment of
chemical engineers show a steady increase in the number of
graduates entering the biotech area.'1 In addition, many
chemical companies have new developments in industrial
biocatalysis. Including the kind of course described in this
paper in the undergraduate chemical engineering curriculum
can encourage students to consider biotechnology companies
as a career by increasing their exposure to the types of projects
they may encounter in such a job.
To introduce students to the fundamentals of laboratory
practice in biochemical engineering, we have developed a
laboratory course that includes protein isolation and purifi-
cation, basic molecular biology methods, microbial kinetics
and energetic, enzyme kinetics, and operation of bioreac-
tors. For the first five weeks of the 15-week course, students
practice basic molecular biology procedures-microbial
transformation, restriction digest, DNA ligation, plasmid
preparation, DNA and SDS-PAGE electrophoresis, and asep-
tic techniques. The time block for the laboratory class each
week is five hours long.
Five hours are needed each week for experiments, with time
for 75- to 125-minute lectures while students are waiting for
experiments to be completed. For example, the thermal cy-
cler runs for close to two hours, so the lecture can be given
during that time. The subcloning experiment involves the am-
plification of the gene for green fluorescent protein from a
plasmid, followed by insertion into a separate plasmid de-
signed for high levels of protein expression. The five-week
sequence is summarized in this article, with presentation of
data from the students' results where appropriate.

The model system used in the lab course is a green fluores-
cent protein (GFPuv) that has been shuffled to yield fluores-
cence that is enhanced from that emitted by the native pro-
tein found in Aequorea victoria.[2] This system is conve-
nient for the students because by using a hand-held UV
lamp, it is easy to detect the protein by observing the fluo-
rescence produced. The use of a long-wave lamp avoids
any harmful UV exposure.
We have found that students have a much better under-
standing of the process of DNA subcloning after concluding
the experiment, which conforms to the general trend in engi-
neering education of increasing opportunities for visual and
hands-on experience. As stated in an NSF Report,131 "Shap-
ing the Future" with regard to undergraduate technical edu-
It is important to assist them to learn not only science
facts but, just as important, the methods and processes of
research, what scientists and engineers do, how to make

Claire Komives is Assistant Professor in the Chemical and Materials
Engineering Department at San Jose State University. She obtained a
BS degree from Tufts University and a PhD degree from the University
of Pittsburgh, both in Chemical Engineering. She teaches thermodynam-
ics, heat transfer in electronics, and biochemical engineering courses.
Her research is in the areas of olfactory G-protein coupled receptors
and process development studies with whole cell biocatalysts.
Sabine Rech is Assistant Professor in the Department of Biological Sci-
ences at San Jose State University. She obtained a BS in Biology from
Santa Clara University, an MA in Microbiology from San Jose State Uni-
versity and a PhD in microbiology from UC Davis. She is teaching courses
in general microbiology, microbial physiology, and microbial diversity.
Her research interests include the isolation of natural products, gene
expression in environmentally important bacteria, and the study of mi-
crobial diversity in the soils of salt marshes in the process of restoration.
Melanie A. McNeil is Professor of Chemical Engineering. She received
her PhD degree in Chemical Engineering from the University of Califor-
nia at Santa Barbara.. She teaches chemical kinetics and reactor de-
sign, biochemical engineering, heat transfer, fluids, safety and ethics,
and statistics. Her research projects include nanowire processing,
bioremediation, bioinformatics, and enzyme kinetics.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

informed judgments about technical matters, and how to
communicate and work in teams to solve complex
The laboratory course has proved to be valuable for students
entering the biotechnology workforce, even when they do
not continue to do molecular biology.
For process development projects involving recombinant
host strains, chemical engineering graduates will likely work
closely with molecular biologists for improvement of the
desired biocatalyst. The experience gained in this course is
targeted to assist them to work in teams with scientists and
other engineers.
The objective of the experiment is
to subclone the GFPuv gene into a
plasmid that has been designed for c
high expression levels and to enable the numb
purification of the expressed protein Including the
after fermentation of the host organ- undergr
isms. The fermentation and protein
purification experiments comprise

eight of the remaining weeks of the

The plasmid system chosen for this
work is the pET plasmid from Novagen.13-6] The pET plasmid
has a T7 promoter that transcribes mRNA for the target gene
in the presence of T7 RNA polymerase. The T7 transcrip-
tion/translation system originated from a bacteriophage and
works efficiently for the expression of very high levels of
protein. E. coli that have a gene for the T7 RNA polymerase
under the control of a lac promoter can express the poly-
merase in the presence of lactose or isopropyl-P-D-
thiogalactoside (IPTG).
For the subcloning, the pET29a plasmid is cut with the same
enzymes as the amplified GFPuv gene insert. The GFPuv
gene can then be ligated into the plasmid and the presence of
the insert can be identified by DNA electrophoresis. The
pET29a-GFPuv plasmid can then be used to transform com-
petent B121 E. coli containing a T7 RNA polymerase for en-
hanced expression of proteins from the pET expression sys-
tem. The students do not need to prepare competent B121
E. coli cells because they are ready for transformation in
the Novagen kit. Ampicillin is used as a selection agent
for the pGLO plasmid and kanamycin is used with the
pET29a plasmid.

(Summary of the Experiment)
Chemicals and Materials
Yeast extract

Ethanol (200 proof)
Ethidium bromide
Glacial acetic acid
Ethylene diamine tetraacetic acid (EDTA)
were all purchased from Fisher. Primers for the polymerase
chain reaction (PCR)

Surveys of the industrial employment of
chemical engineers show a steady increase in
er of graduates entering the biotech area....
kind of course described in this paper in the
aduate chemical engineering curriculum can
encourage students to consider biotechnology
companies as a career.

were purchased from Operon. Restriction enzymes
were purchased from Promega, and the Promega 10x multi-
core buffer was used for the digest.
Taq polymerase for the PCR was included with the
Promega master mix. The MgCI2 solution and nuclease-free
water were supplied with the Taq polymerase. A 1-kb DNA
ladder was purchased from GeneChoice because it can be
used to estimate the amount of DNA on the bands (Cat.
The 6x loading dye for the DNA electrophoresis was pur-
chased from Promega (G190A). Petri dishes and all
disposables were purchased from Fisher, and the five-minute
ligation kit was purchased from GeneChoice (Cat. #62-6104-
20). The pGLO plasmid was purchased from Bio-Rad
(pGLO Bacterial Transformation Kit; Cat. #166-
0003EDU, approximately $50), which also contains de-
hydrated e. coli (strain H-101).
The pET29a expression system kit was given by Novagen
free of charge to use in the class. Additional competent cells
were purchased from Novagen (Novablue singles, Cat
#70181, B121 (DE3) singles, Cat #70235). Jellyfish software
was purchased from Labvelocity on-line at>. Kits for plasmid pu-
rification and gel extraction were purchased from Qiagen
(Cat. #27104 and #28704, respectively).

Summer 2004

Kanamycin and ampicillin stocks were prepared prior to
the course with 30 mg/ml and 100 mg/ml, respectively. The
solutions were sterile filtered and stored in 1-ml aliquots
at -300C.
Special storage conditions: The enzymes were stored at
-300C; the competent cells were stored at -800C; frozen stocks
of all strains were stored with 30% glycerol at -30C.
Circulating waterbath (Brinkman LAUDA RM6)
Incubator (Fisher Isotemp)
Incubator shaker (New Brunswick G25)
Eppendorf mastercycler gradient thermal cycler
UVP PhotoDOC-IT with 4912 camera and software
Agilent bioanalyzer 2100
HP 8452A UV/visible spectrophotometer
Turner Designs picofluor fluorometer
Agarose gel electrophoresis chamber and power sup-
The methods are described in the weekly laboratory ses-
sion below.
Week 1 -
LB agar (1% Bacto-tryptone, 0.5% yeast extract, 0.5%
NaCI, 2% Difco agar in DI water, addition of 640 JL) was
prepared, followed by autoclaving at 1210C for 15 minutes.
Antibiotic was added to the agar from a 1000x stock prior to
pouring. From an overnight culture of pGLO E. coli, 3 mls
were distributed to each group and the plasmid was purified
according to the Qiagen kit instructions. The bacteria were
viewed under a microscope. In addition, the students were
able to study the primer binding sites on the pGLO plasmid
and the restriction cutting sites and insertion region on the
pET29a plasmid using Jellyfish software. This software pro-
vides a numbered image of the plasmid and identifies the
restriction sites for the user-selected enzymes. Information
on the primers is also given to facilitate the PCR design. TAE
buffer (0.04M Tris-acetate, 0.001M EDTA) was prepared for
the following week.
Week 2 -
PCR was performed to amplify the GFPuv gene. The PCR
reaction mixture consisted of 25 pl of the Promega
Mastermix, 1 Ipl of each primer (1 pLM final concentration
each), 3 l1 of purified pGLO plasmid from the mini-prep of
the previous week, additional MgCI, to make a 4 mM final
concentration, and nuclease-free water added to make 50 pl
total solution. The annealing temperature of 520C is optimal.
While the PCR reaction was running, a restriction digest on
the pGLO plasmid purified the previous week was performed
to check the length of the plasmid. The restriction digest was
incubated at 370C for one hour in a circulating water bath,
where the reaction mixture comprised 10 pil of pGLO plas-

mid from mini-prep, 1 pl of KpnI enzyme, 2 pl of 10x
Promega multi-core buffer, 0.2 p1 of concentrated Bovine
Serum Albumin (BSA, frozen stock solution comes with the
restriction enzyme), and 6.8 pl1 of nuclease-free water to make
the mixture 20 pJL in total volume. The PCR product was run
on a 1% agarose gel along with cut plasmid to check the con-
centration of the purified plasmid from the mini-prep and to
check the length of the PCR product. A 1-kb ladder is ad-
equate for both measurements. The PCR product bands were
purified from the gel with the Qiagen gel-extraction kit ac-
cording to the instructions.
Week 3 -
A restriction digest was performed to cut the GFPuv gene
from the PCR product with KpnI and XbaI. The reaction mix-
ture contained 1 pl of each enzyme, 0.2 pl BSA, 2 pl 10x
multi-core buffer, and 15.8 pLI of the purified PCR product.
The cut gene was eluted from the spin tubes with Tris buffer
supplied with the gel extraction kit. At the same time as the
cutting of the PCR product, the pET29a plasmid was cut with
KpnI and XbaI. To prepare for the ligation and transforma-
tion, 1 pLg of plasmid was used. For the plasmid cutting ex-
periment, three separate restriction digests should be per-
formed: 1) cut with KpnI only; 2) cut with Xbal only; 3) cut
with both KpnI and XbaI. The digests were run on a 1% aga-
rose gel and the bands from the double-cut PCR product and
double-cut plasmid were cut out of the gel and cleaned with
the Qiagen gel-extraction kit. The two single cuts are per-
formed to ensure the enzymes are working properly, and
should have identical bands on the gel that are the length of
the full plasmid.
The ligation of GFPuv gene into the cut pET plasmid was
performed with the five-minute ligation kit by Genechoice.
The reaction mixture for the ligation included approximately
30 ng of cut PCR product and 50 ng of plasmid. The reaction
was carried out at room temperature. Competent "Novablue"
E. coli cells were transformed with the ligation product and
with control plasmid provided in the Novagen kit according
to instructions in the pET system manual. The transformants
were plated on LB/kanamycin plates and incubated for 18
hours at 37C.

Week 4 -
Prior to the class, the TA prepared overnight cultures from
the students' plates in LB broth with kanamycin. During the
lab period, the students first performed a mini-prep of colo-
nies that were grown in liquid culture and then checked for
the insert in the plasmid by a restriction digest with KpnI and
XbaI. The students again carried out the cutting with three
separate restriction cuts: one with KpnI only, one XbaI only,
and one with both enzymes to compare the length of the plas-
mids. The double-cut plasmid should run ahead of the single-
cut plasmid due to the removal of the insert. A pET plasmid
without the insert was also cut with KpnI and XbaI. The in-

Chemical Engineering Education

sert including the GFPuv gene should be 778 bases long. The
digests should be compared on a 1% agarose gel. Given that
the pET-GFPuv plasmid was successfully prepared, as dem-
onstrated by the presence of the insert, competent B121DE3
E. coli cells were transformed with the pET-GFPuv plasmid
and with the pET plasmid without the insert for a control.
Week 5 -
The TA prepared overnight cultures of both the pGLO E.
coli and the pET-GFPuv E. coli. For these cultures, 1 mM
IPTG and 0.1% arabinose were used as inducers, respectively.
During the lab period, the students compared the fluorescence
and the optical density of both the pGLO E. coli and the pET-
GFPuv E. coli. A determination of the fluorescence normal-
ized with the optical density for both strains enables a quan-
titative comparison of the amounts of correctly folded pro-
tein per bacteria. A complete SDS-PAGE experiment may
not be feasible for a five-hour lab session. It is possible to
make this comparison if a microfluidic analysis can be per-
formed, such as with the Agilent 2100 Bioanalyzer.

The experiment begins with the bacterial culture that can
be generated from the components of the Bio-Rad pGLO
Bacterial Transformation Kit. The Biotechnology Explorer
products are designed for high school students to get some
hands-on experience with biotechnology, and the pGLO plas-
mid that comes with the kit serves as the starting point for the
subcloning experiment performed in our course. The gene
for the green fluorescent protein (GFPuv) is located in the

1 2 3 4
Plasmid template e--i

4 800bp
PCR product -- 600 bp

Figure 1. Gel from lab session 2. The expected band of
795 was obtained by groups running in lanes 1-3.

pET plasmid
5268 bases --

Insert 778 bases 0

Figure 2. pET-GFPuv plasmid cut with Kpnl and XbaI
showing the cut insert at 778 bases in length.

pGLO plasmid downstream from the pBAD promoter, which
can be induced with arabinose. The pGLO plasmid has both
the ampicillin resistance gene as a selection marker for en-
abling only the growth of bacteria successfully transformed
with the plasmid, as well as an arabinose promoter that en-
ables user-control of the induction of the gene for GFP by
addition of arabinose to the agar or broth. Transformation is
necessary if the experiment will be performed in future
courses, as the bacterial hosts provide a means to produce
additional plasmid. The transformation should be performed
ahead of the course and an overnight culture prepared prior
to the first lab session.
In the second lab session, the students should be told that
the order of the experiments is usually to check the results of
the mini-prep with the restriction digest, followed by the aga-
rose gel, prior to performing the PCR. Normally, one should
be assured that the mini-prep was successful before running
the PCR reaction, but the five-hour lab session does not per-
mit this. If the students do not put correct amounts of the
reagents into their tubes or do not follow steps accurately,
their experiments will not work. Our philosophy is to main-
tain a stock of the necessary plasmid and gene products from
the different experimental steps for use if any experiments
fail. In that way the students can continue through the five
weeks even if one of their experiments is not successful.
Figure 1 shows a section from the gel of a Week 2 lab ses-
sion. The students are given the option of modifying the choice
of annealing temperature and MgCl2 concentration to further
optimize conditions. The gel in Figure 1 shows three PCR
reaction products (lanes 1-3), with the ladder in lane 4. The
band between 600 and 800 bp is the desired PCR product
containing the GFPuv gene. Because the students purified
their own plasmids and chose different conditions for the PCR
experiment, the brightness of the bands shows variations.
Students whose bands are particularly bright are able to share
some of their PCR product with other groups that got little or
no band at the proper length.
The bacterial transformation of Week 3 is one of the most
challenging steps in the five-week experiment. It is impor-
tant to explain to the students that the times and temperatures
listed in the protocol are critical, and if they change a step it
will likely have a deleterious effect on their transformation.
It has proven helpful to go over the procedure with them just
prior to the actual experiment. The Novablue cells are used
because they have a high transformation efficiency and lack
the T7 RNA polymerase gene necessary for protein expres-
sion. After a ligation, it is favorable to use competent cells
with a high transformation efficiency. Therefore, only after
the Novablue cells are shown to contain the plasmid and the
plasmid can be purified through a mini-prep are the B121 (DE3)
cells transformed with the plasmid. While it is possible to
include arabinose in the LB/Ampicillin plates to induce the
Continued on page 221.

Summer 2004

Son class and home problems

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 eluci-
date difficult concepts. Manuscripts should not exceed fourteen double-spaced pages and should
be accompanied by the originals of any figures or photographs. Please submit them to Professor
James O. Wilkes (e-mail:, Chemical Engineering Department, University
of Michigan, Ann Arbor, MI 48109-2136.


An Application of Phase Equilibrium

and Sensitivity Analysis

University of Toronto Toronto, Ontario, Canada M5S 3E5

ondensation is one method that can be used for re-
C cover of a volatile solvent from a gas stream or for
similar industrial operations.['1 In this method, the
temperature (T) of the stream containing solvent vapor (con-
densable) and an inert gas (noncondensable) is lowered suf-
ficiently to allow (partial) condensation of the solvent vapor.
Examples of processes that may use condensation for this
purpose are the recovery of tetrachloroethylene (C2C14) in
dry-cleaning systems, and n-hexane (C6H14) in vegetable oil
extraction. Condensation may be part of a more comprehen-
sive system.[11
Apart from heat transfer considerations, the process design
conditions for operation of the involved condenser can be
guided by application of phase equilibrium and elementary
sensitivity analysis.121 Phase equilibrium provides an indica-
tion of temperature and composition behaviors during con-
densation, and the upper and lower limits of variables in-
volved, including the extent of condensation. Sensitivity
analysis shows the effects of changing the system operating
variables (which are usually within the control of the opera-
tor) and any governing constitutive parameterss; it also pro-
vides assessment of the effect of errors in the constitutive
The purpose of using solvent recovery by condensation from
a noncondensable gas as a "class and home problem," apart
from illustrating some important aspects of the process, is to
demonstrate an actual case for these applications of phase
equilibrium and sensitivity analysis. The thermal design of

the heat exchanger (condenser), itself, is outside the scope; it
is discussed, for example, by McAdams, by Rohsenow, et
al., and by Frank.p[-5] We first pose a problem statement in-
volving model assumptions and several questions, and then
provide solutions to these questions in general terms. Fol-
lowing this, a numerical example is given.

Figure 1 shows a schematic flow diagram for partial sol-
vent recovery by condensation. The condenser, C, is depicted
as though it were a single-pass, tube-side, horizontal heat
exchanger,[4] although different configurations involving a
vertical arrangement,'11 or a dephlegmater,[5] can be used. For
simplicity, we consider a binary stream consisting of a con-
densable solvent (component 1, e.g., C2Cl4) and a
noncondensable inert species (component 2, e.g., air or N2).
(A condensable species has a critical temperature, Tc, higher
than the stream temperature, and conversely for a

Ronald W. Missen is a Professor Emeritus
(Chemical Engineering) at the University of
Toronto. He received his BSc and MSc de-
grees in chemical engineering from Queen's
University and his PhD degree in physical
chemistry from the University of Cambridge.
He is coauthor of Chemical Reaction Equilib-
rium Analysis and Introduction to Chemical Re-
action Engineering and Kinetics.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

noncondensable species.)
In Figure 1, the inlet vapor stream (at T'", P'", and xj",
where P is total pressure and x, is mole fraction of compo-
nent 1) enters condenser C, in which Tn1 is lowered eventu-
ally to Tex. The resulting two-phase (liquid + vapor) stream
enters drum D for separation by gravity into an overhead va-
por stream and a bottom stream of liquid 1. Because of fric-
tional pressure drop, the exit pressure from D, Pex, is somewhat
lower than pin.
The fraction of entering solvent condensed in the exit liq-
uid stream (of C or D) is
in -ex ex
fex = n1 1 i 1- (1)

where niPn is the inlet rate of flow of component 1 (e.g., mol
s-1) and fix is the rate of flow of component 1 in the exit
vapor stream from D.

We assume the following:
(Al) The process is at steady-state.
(A2) The vapor is an ideal-gas mixture.
(A3) The effect of P on the molar volume of pure liquid 1 is
(A4) The vapor and liquid streams are at thermodynamic
equilibrium at any point considered for two-phase coexist-
(A5) The solubility of component 2 in liquid 1 is negligible.
(A6) The inlet stream to C is single-phase vapor.
(A7) The coolant in C (water or refrigerant) enters at Tnol, and
there is a nonzero-approach-temperature difference (ATappr)
at the exit of C, where
ATappr = Tex T (2)

Sex ex ex
n, X, n2

Liquid 1
6in ,ex
nI ni
Figure 1. Schematic flow diagram for solvent
recovery by condensation.

Summer 2004

(A8) Tc2 (A9) If required, the vapor pressure of component 1 ( p,) is
represented by the Antoine equation16] in the form

logo P7 = A (3)
where A, B, and C are constants, each > 0; values of the
constants for pt in mm Hg and T in C are given by Dean.171

For this problem, we posed the following questions/deter-
minations. The answers are first given in general terms in-
volving chosen system variables and constitutive
parameterss. Numerical values are then used in an example.
(Q1) For given (P'i, T"i), what is the upper limit of xin for
the entering stream to be a single-phase vapor?
(Q2) At what temperature does condensation begin, and does
the temperature change as condensation proceeds?
(Q3) Derive an expression for flex in terms of a chosen set
of system variables, and in so doing establish the num-
ber of degrees of freedom (F) among all the quantities
chosen for description of the system.
(Q4) For any condensation to take place in C, what is the
range of allowable values (lower and upper limits) of
each of the chosen system variables, taken separately?
(Q5) What are the corresponding lower and upper limits of
condensation, fl (LL) and fe (UL), for each of the
values in (Q4)?
(Q6) From the result in (Q3), obtain expressions for all rel-
evant first-order sensitivity coefficients,121 including
for any constitutive parameterss.
(Q7) From the results of (Q6), state conclusions about the
direction of change of flfe with specified direction of
change of each system variable and constitutive pa-
(Q8) Express the uncertainty in flc in terms of the uncer-
tainty of any constitutive parameter.

(Q1) The question stems from assumption (A6). Equilib-
rium considerations set an upper limit for the inlet
composition, x'n (UL), in establishing the condition for
saturation of the gas stream with respect to compo-
nent 1. If we equate the chemical potentials of com-
ponent 1 in vapor and (virtual) liquid phases accord-
ing to assumption (A4), together with assumptions
(A2) and (A3), we obtain
xln(UL)= p* (Tin)/Pin (4)

from which x l'(UL) can be calculated once Tn' and P'n
are established. Then, typically, x in tual operation.
(Q2) At the inlet to the condenser, from the discussion sur-

rounding Eq. (4), the partial pressure p, of component
1 is such that
pin(= xpin) Pi= X (5)
From the inlet to the point at which condensation be-
gins, p, remains essentially constant at p"n, but p (T)
decreases as T decreases. Condensation begins at the
(initial) dew-point temperature, Tnt (x1in,P), where P
is the local pressure. This temperature is determined
from the condition of equality of p, and p analogous
to the development of Eq. (4)

p* Tiit (x P= = I xnP -x npin (6)
Eq. (6) is a nonlinear relation used to calculate Tinit
from xTn, P pin, and a vapor pressure relation such as
Eq. (3).
As condensation proceeds, x, decreases, with con-
sequences for T. At any point between the point of ini-
tial condensation and the exit, T may be considered to
be a varying dew-point temperature at the local com-
position and pressure TDP(Xl,P). It is determined from
a relation analogous to Eq. (6)
p [TDp(xl,P)]= XP (7)
TDP can be shown to decrease as x decreases

aTDP dTDp ap* dTDP Fap+ ap ] dP
axI dpf axI dp; [axl P dxlj
dTDP [P+X1 P (8)
dp1 dx j
Thus, the temperature of condensation of a condens-
able species from a mixture with a noncondensable
species decreases as condensation proceeds. This is in
contrast to the condensation of a pure substance (not
in a mixture with an inert gas), which occurs at con-
stant temperature.
(Q3) In establishing an expression for flex, we have a choice
of system variables to use, but the number of indepen-
dent variables-that is, the number of degrees of free-
dom (F)-is dictated by the difference between the
number of these variables (including flex) and the num-
ber of independent relations among them.
In Eq. (1), which serves as a material balance for
component 1, fleis an intensive quantity, but it is ex-
pressed in terms of extensive flow quantities. To
obtain flex in terms of variables describing the intensive
state of the system, we introduce additional variables
and relations among them as follows:
lin = x in n (9)

iex =ex ex (10)
where xin and xFx are the mole fractions of compo-

nent 1 in the inlet and exit streams, respectively, and
in and fix are similarly the total molar flow rates
in =in +fin (11)

x = x + i~x (12)
where fi" and i'x are the molar flow rates of compo-
nent 2, as indicated in Figure 1.
From assumption (A5), a material balance on com-
ponent 2 is
hin .ex (13)
2 = (13)
Finally, according to assumption (A4), xf is constrained
by the equilibrium condition analogous to Eq. (4)
xXp(Tex/Pex (14)

Eqs. (1) and (9 to 14) are seven equations in 11 system
variables [and one constitutive parameter, pi (Te)],
which seems to suggest F = 11 7 = 4. If, however,
fle, an intensive quantity, is independent of flow rate,
F is reduced from 4 to 3. This can be established either
by assigning an arbitrary (control or basis) value to,
say, nfi, or by solving Eqs. (1) and (9 to 14) for flec so
as to eliminate all extensive variables and xfto give
pex (Tex)/ Ixn
flex = (15)
Ic pex Pl (Tex) (

Eq. (15) confirms that F = 4 1 = 3 and involves the
choice of xin, Tex, and Pex, all intensive, as indepen-
dent system variables.
The extensive state of the system, as given, for example,
by the value of hin, involves a scale factor that deter-
mines the size of the condenser (not treated here), but
does not affect the value of fle.
(Q4) For condensation to take place, flex (LL) < fex < fex (UL),
where the lower and upper limits may or may not equal
0 and 1, respectively, as might be thought at first. The
variables to be investigated are, from (Q3), (i) xin, (ii)
ex, and (iii) TeX. We consider the allowable range of
values of each of these in turn, for given (allowable)
values of the other two.
(i) xn: Setting fx(LL) = 0 in Eq. (15), we obtain
I Setting fle(L)=i Eq(15 ta
xn (LL) =p Tex/ pex(= xx) (16)

If we set flex (UL) = 1 in Eq. (15), the result can only be
satisfied by x1n= 1, which is not valid; we conclude
that xn (UL) is constrained by the result of (Q1)
xi'L(UL) = p;(Tin)/ in (17)
(ii) Pex: Setting fl (LL) = 0 in Eq. (15), we obtain
pex(LL)=p;(Tex)/x n (18)

If we set flx(UL) = 1 in Eq. (15), the result can only be

Chemical Engineering Education

satisfied by

pex (UL)>> p Tex) xi or pex(UL) -oo (19)

(iii) Tex: The value of Tex (LL) is set by the thermal
condition in Eq. (2)

Tex(LL) = Tol+Tappr (20)

From the result of (Q2), this sets the condition
for fle(UL). To obtain Tex(UL), we therefore
set fx (LL) = 0 in Eq. (15) to give Tex (UL) implicitly

p[Tex(UL)=x npex (21)

An explicit expression for Tex(UL) can be obtained by
using a vapor pressure equation, such as Eq. (3), to
eliminate p .
The above results for (Q4) are collected in Table 1,
columns 2 and 3.
(Q5) The results for flc (LL) and fic (UL) corresponding to
the limits of the system variables are obtained from
(Q4) where noted, and otherwise from Eq. (15)-in
the two cases at xn (LL) and TeX(LL). The results are
collected in Table 1, columns 4 and 5. The lower limit
is 0 in each case, and the upper limit is less than 1,
although it can approach 1 at very high values of Pex.
With respect to Te, the results in columns 2 and 5 em-
phasize that for the greatest efficiency in solvent re-
covery, the condenser should be designed so that TeX
approaches To ,.1[51

(Q6) From Eq. (15), fle" is an explicit function of the system
variables xln and Pex and also of the constitutive pa-
rameter p *(Tex). The sensitivity coefficients of fe with
respect to these quantities, a flex/Xlin, af x/fpex, and
SfeX/p,*(Te), can be obtained by direct differentia-
tion. Since flex is an implicit function of Tex, through
pl*(Tex), the sensitivity coefficient with respect to Tex
is obtained as

aflex aflex dp (Tex)
c Icp (22)
aTex ap*(Tex) dTex

The results for all four sensitivity coefficients are
listed in Table 2, column 2.
(Q7) One use of the sensitivity coefficients in Table 2 is the
information they provide about the direction of change
of fex for a specified direction of change of a system
variable or constitutive parameter. This information is
contained in Table 2, column 3, which shows the sign
of each coefficient, and column 4, which shows the
interpretation of the sign. Thus, for the system vari-
ables, flex increases as xin or Pe increases, and as Tex
decreases; conversely for decreasing f l".
Achieving favorable values of Pe above ambient or
the upstream value, by compression, and of Tex below
that obtained using cooling water, by refrigeration, is a
matter of economic analysis in conjunction with the
thermal design of the condenser. It may turn out that
even with refrigeration and/or compression, the sol-
vent concentration in the exit vapor stream may not be

Lower Limits (LL) and Upper Limits (UL)
Collected results for (Q4) and (Q5)

(Q4) SV (UL)

(Q5)f,," (LL)
at SV(LL or UL)

p (Tex) _

given (pex, Tex)


(given x n, Tex)



at xn (LL)


-4 oo

at Pex(LL)

pex (Tex)in
p (Tin
pex -p*(Te)
at x n(UL)


at pex (UL)

Tnoo + ATappr

p [Tex(UL)] = xnPex

(given xin, pex)

pex P(To +ATappr)
0 xl <1
pex (Tol + ATappr)
at Tex(UL) at Tex(LL)

System Variable

(Q4) SV (LL)

(Q5) f," (UL)
at SV(LL or UL)

Summer 2004

low enough to satisfy environmental requirements, or may not be
outside explosive limits, if applicable. This may lead to the con-
clusion that condensation in a particular case is best used as an
intermediate stage in a more comprehensive system.1']
(Q8) Another use of sensitivity coefficients is the information provided
about uncertainty in a dependent variable, here fle, caused by
uncertainty in a constitutive parameter, here the vapor pressure p ,
which is subject to experimental error. (Values of the indepen-
dent system variables are set by the investigator and are consid-
ered to be without error.) In this case, the uncertainty in fi, flc
arising from the uncertainty in pl, Spl, is given by
_fC= f 6p (23)
ap p
A representative value of 6pl is 0.2%.[9]


To illustrate use of the results obtained above, consider condensation
of n-hexane (component 1) from an inert gas such as nitrogen (compo-
nent 2) by means of a water-cooled condenser. The given conditions
and calculated results are summarized in Table 3. The vapor pressure,
p7, is calculated from the Antoine equation, Eq. (3), using values of the
constants provided by Dean.17] The calculated value of fex is 0.556 cor-
responding to xf = 0.100, which are modest results. [If a refrigerated
condenser with Tex = -25C is used,151 flex is increased to 0.975 and xf
is reduced to 0.006 (results not given in Table 3).] The uncertainty
in flx because of uncertainty in p (Tex) is negligible, based on Eq. (23),
with the value of Spl indicated there and that of
afix / ap (Tex from Table 3.


Financial assistance has been received from the Natural
Sciences and Engineering Research Council of Canada. The
paper was discussed initially with W.R. Smith.

1. Cooper, C.M., in Kirk-Othmer Encyclopedia of Chemical Technol-
ogy, 3rd ed., Wiley, New York, NY, Vol. 21, pp. 355-376 (1983)
2. Smith, W.R., and R.W. Missen, Chem. Eng. Ed., 37(3), 222; 37(4),
254 (2003)
3. McAdams, W.H., Heat Transmission, 3rd ed., McGraw-Hill, New
York, NY, pp. 355-356 (1954)
4. Rohsenow, W.M., J.P. Hartnett, and Y.I. Cho, eds., Handbook of Heat
Transfer, 3rd ed., McGraw-Hill, New York, NY, p. 17.123 (1998)
5. Frank, Otto, in Kirk-Othmer Encyclopedia of Chemical Technology,
4th ed., Wiley, New York, NY, Suppl. Vol., pp. 515-516 (1998)
6. Antoine, Ch., Compt. Rend. de l'Academie des Sciences, Paris, 107,
681,836 (1888)
7. Dean, J.A., Lange 's Handbook of Chemistry, 15th ed., McGraw-Hill,
New York, NY (1999)
8. SmithW.R., and R.W. Missen, Chemical Reaction EquilibriumAnaly-
sis, Wiley-Interscience, New York, NY (1982); Krieger, Malabar,
FL, pp. 50-51 (1991)
9. Linstrom, P.J., and W.G. Mallards, Eds., NIST Chemistry WebBook,
NIST Standard Reference Database Number 69, National Institute
of Standards and Technology, Gaithersburg, MD, 20899> (2003) 0

First-Order Sensitivity Coefficients (S.C.) of flc:
Collected Results for (Q6) and (Q7)

(Q6) expression (Q7) sign (Q7)
for S.C. of S.C. interp.


I (-xi")pT ) + pex
+ xin "

a ex \2

3f^ (L-xin)pex[dp;(T eX)/dx] ^ TeX ,
ex xin[Pex p(Tex)]

ex (I x )Pex dp Tex) dTex



x n[pex-p (Tex)]2

- p(Tex)

Example: Condensation of n-hexane (1) from Inert Gas

Yen eon ons -..-

xin =0.200 fex =0.556

Tin=50C xx =0.100=xin(LL)

pin =2.5bar x" (UL)=0.216

pex =2.2bar flex(UL)=0.597atxl (UL)

Tin 220 Tinit =47.8 C
Toola = ( DP = .r ( g

ATappr =5K pex (LL)=l.lbar (at given xr ,Tex)

pl(Tin = 0541barT

p;(Tex =27c)=0.220bar

Tex(UL)=44.4C (at given xin,Pex)

aflex /axin =2.78(atgiven conditions)

fex /apex = 0.224bar-1

afex/aT ex =-0.0213K-l
aflex/ap (Tex )_-2.245 bar-1

t In Eq.(3) for p,', the Antoine constants are 17, p 548: A=6.87601; B=
1171.17; C=224.41

Chemical Engineering Education

iG C diti

C~-~l 1.. -d-.i

Continued from page 215.
expression of GFP after transformation with pGLO plasmid,
the addition of IPTG to plates will prevent growth of the
B121(DE3) cells transformed with pET-GFPuv plasmid. The
inducer should only be added after the cells have been success-
fully transformed. Figure 2 shows the pET29a-GFPuv plasmid
cut with KpnI and XbaI. The insert runs close to 800 bases.
In the third-week experiments, it is helpful to note that the
bases cut from the ends of the PCR product cannot be seen
on the gel because they are too small to generate visible fluor-
escence. The failure of one or both of the enzymes to cut the
ends, therefore, cannot be identified from the PCR product.
It is instructive to check the activity of the enzymes by cut-
ting the plasmid with the enzymes individually in two sepa-
rate reactions. The single-cut plasmids should show bands of
identical length provided the enzymes both cut efficiently. If
this holds true, then the enzymes likely cut the PCR product
successfully as well. It is also useful to explain to the stu-
dents that as restriction enzymes are generally designed to
cut longer pieces of DNA, they may not cut effectively close
to the end of the gene, as with the PCR product. A useful
reference is the New England Biolabs catalog, which can
be viewed on-line at frame_cat.html>. The section of "Technical Reference Lit-
erature" includes a table with optimal end lengths for com-
monly used restriction enzymes.
The fifth week is for analytical measurements of the pro-
tein expression in both folded and unfolded forms. The fluo-
rescence emitted by the properly folded protein can be mea-
sured and normalized to the optical density. The difference
between the normalized fluorescence of both the pGLO bac-
teria and the pET-GFPuv bacteria can be expressed as a ratio
or percent increase, thus omitting the need for a GFP stan-
dard. The excitation light is ultraviolet light, and the amount of
emitted fluorescence is measured with a fluorometer.
This is a challenging experiment for the students because
they must dilute the culture until they are measuring in a lin-
ear range of both the fluorometer and the UV/vis spectro-
photometer. Since two different expression systems are com-
pared, ideally one would optimize the protein expression first
and then make the measurements of fluorescence-but this
has not been done. Instead, overnight cultures were prepared
and the inducers were added at the initiation of the culture.
When this approach is taken, the difference in fluorescence
between the two cultures is minimal, being on the order of
15% higher for the pET-GFPuv culture. This also gives the
students an opportunity to observe the difference in the fluo-
rescence using the "wrong" induction method as compared
with the "right" method that is used when the students run
the fermentation. At that time, the culture is induced at
the beginning of the lab period when the culture is in the

exponential growth phase.
There has been very positive feedback about the lab. The
enrollment has steadily increased, and there have also been
students from local industries who have joined the class. In-
dustry personnel who serve as advisors for the program or
who have phoned for student references have been particu-
larly positive about the content of the course. While they are
interested in the fermentation and protein purification por-
tions of the course, they have also praised the molecular bi-
ology portion. Because the chemical engineers often work
closely with biologists for strain improvements, having hands-
on experience with the DNA modifications enables them to
follow the work of the biologists with greater understanding.

This paper describes a set of laboratory experiments that
can be performed in five weeks for training undergraduate
chemical engineering students in the basic steps in subcloning
genes and molecular biology techniques. PCR, ligation, and
the use of restriction enzymes are introduced with the lab,
along with the use of gel electrophoresis to analyze changes
in DNA. The use of green fluorescent protein enables fast
determination of protein expression. Students who have per-
formed the experiments have a better understanding of mo-
lecular cloning methods than students who have learned about
it in the lecture course alone.

Support for the laboratory course was provided by NSF
(CCLI grant to C. Komives, M. McNeil, and S. Rech, Award
#0088653) in addition to funds from the California State
University Program in Education and Research in Biotech-
nology (CSUPERB), the California Workforce Initiative, and
the SJSU College of Engineering. The authors are grateful
for the assistance of Ludmilla Stoynova and Dr. Ramesh Nair
in the development of the experiment. Additional thanks to
Agilant for donating the 2100 Bioanalyzer and to Novagen
for contributing the pET Expression Kit.

2. Crameri, A., E.A. Whitehom, E. Tate, and W.P. Stemmer, "Improved
Green Fluorescent Protein by Molecular Evolution Using DNA Shuf-
fling," Nature Biotech., 14, 315 (1996)
3. NSF, Shaping the Future: New Expectationsfor Undergraduate Edu-
cation in Science, Mathematics, Engineering and Technology, Divi-
sion of Undergraduate Education, NSF-96139 (1996)
4. Studier, FW., and B.A. Moffatt, "Use of Bacteriophage T7 RNA Poly-
merase to Direct Selective High-Level Expression of Cloned Genes,"
J. Molec. Bio., 189, 113 (1986)
5. Rosenberg, A.H., B.N. Lade, D. Chui, S.W. Lin, J.J. Dunn, and F.W.
Studier, "Vectors for Selective Expression of Cloned DNAs by T7 RNA
Polymerase," Gene, 56, 125 (1987)
6. Studier, F.W., A.H. Rosenberg, J.J. Dunn, and J.W. Dubendorff, "Use
ofT7 RNA Polymerase to Direct Expression of Cloned Genes," Meth-
ods in Enzymology, 185, 60 (1990) C

Summer 2004

e 1 curriculum



At Rose-Hulman Institute of Technology

Rose-Hulman Institute of Technology Terre Haute, IN 47803

Freshman Design or Freshman Engineering has become an integral part of en-
gineering education, but the format of the course varies significantly from one
program to another. This paper discusses the Freshman Design course in Chemi-
cal Engineering at Rose-Hulman Institute of Technology (RHIT). While freshmen
attend a very general "Introduction to Engineering" course at some institutions, RHIT
students have already selected their major when enrolling in the freshman engineer-
ing class (Introduction to Design). Each section of the class is designated as being for
a given major. This provides two benefits in particular: 1) the "competition" between
engineering disciplines often associated with freshman engineering courses is elimi-
nated by this early declaration of their major, and 2) the group of students is focused
on a common interest in chemical engineering in particular rather than on engineer-
ing in general. This allows the instructors to directly introduce basic chemical engi-
neering and process design concepts at the freshman level to give students a better
overview of what lies ahead.

Since chemical engineering is a challenging area of endeavor, the freshman design
course is designed as a challenging experience. The primary student-learning goals
of the course include developing an understanding of how basic science and math-
ematics interface with engineering, fostering an awareness of the chemical engineer-
ing curriculum as well as careers available to chemical engineers, developing and
improving communication and teaming skills, and creating a realistic view of prac-
ticing chemical engineering (see Table 1). For a class that meets
two hours a week for ten weeks, this is a challenging list of
goals (for both faculty and students)! The project-based approach
is used to achieve the learning goals.
We will describe how a project, based on the preliminary de-
sign and economic analysis of a chemical plant, is used to achieve
four of the five goals (see Figure 1). The project is used to intro-
duce concepts such as the steps of process design and the con-
nectivity of math, science, and engineering, to promote a more
thorough awareness of the curriculum and the relevance of
each of the required courses, to further develop teaming and
other soft skills, and to encourage the proper documentation

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

Course Goals

* Expand the students' knowledge of
the various areas of endeavor
available to chemical engineers.
* Develop the students' understand-
ing of how basic science and
mathematics interface with
* Increase the students' awareness of
the chemical engineering curricu-
lum and core competencies.
* Develop, extend, and improve the
written and oral communication
and teaming skills of the students.
* Present the students with a realistic
view of practicing chemical

Sharon G. Sauer is Assistant Professor of
Chemical Engineering at Rose-Hulman Insti-
tute of Technology. She has a BS in Chemical
Engineering from Florida State University
(1993) and a PhD in Chemical Engineering
from Rice University (2001). Before joining the
faculty at Rose Hulman, she worked for three
years at Shell Oil Company in Houston, Texas.

of technical work.
In 1990, Professor Emeritus Carl Abegg created a con-
trolled-release nitrogen fertilizer plant-design problem and
the basic project content to remedy what he, and our fresh-
men, felt was an unsatisfactory experience in freshman de-
sign. He had been actively involved with the design of a pro-
duction facility based on the process in the disclosed patent"1
that is used in the course.
The department has continued using the project, making
modifications and varying emphasis to the process each year.
It has been taught by Professors Carl Abegg, Jerry Caskey,
Atanas Serbezov,
and myself. Profes-
0 sor Serbezov and I
compnh currently teach two
sections each of the
Course, with an av-
erage of approxi-

per section.

Figure 1. The project is the focal describes the results
point of the course. of the course's evo-

Figure 2. Simplified process schemati

lution and reflects the author's current style of instruction.

One challenge instructors face is to find a process that is
appropriate for freshmen. Very few process operations can
be presented with sufficient detail at an introductory level.
Since some of the primary operations in the process used in
our course are mixing and drying, the project is quite suit-
able for introduction at the freshman level. These concepts
are familiar to the students from their personal experiences-
for example, mixing water, tea, and sugar in a glass or using
a blender to make a milkshake are common "mixing" pro-
cesses. Moreover, blow drying hair or drying clothes in a
dryer (provided they are not still taking their laundry home
for Mom to do) are familiar "drying" processes. While these
concepts are readily grasped by a freshman, distillation or
gas absorption would be very difficult to present in sufficient
detail without background knowledge the students will at-
tain in the sophomore and junior core courses.
The simplified process used in the course is based on an
expired patent,"' which the students actually read. This gives
them an opportunity to experience the differences between
"textbook" writing and the writing in a formal patent. By
discussing the legal implications of actual design documents,
the students become more aware of the need to properly docu-
ment their work. It is also an opportunity to introduce the
career option of patent law.
The primary pieces of equip-
ment in the operation, depicted
in Figure 2, are reaction kettles,
mixers, a stirred reactor, a cur-
i And ing belt, and a dryer. After they
SHO) have read the patent, an in-class
brainstorming session is held
and the students are asked to in-
dicate what information is
needed to design the plant.
Questions such as "How much
do we want to produce?" and
"What is the amount of raw ma-
terials needed?" to "How much
Jacketed steam should be supplied?" and
Reactor "Do we need to maintain a spe-
cific pH or temperature?" are
generated by the students and
during Belt listed on the board. We then
map this information to the re-
quired courses in the chemical
engineering curriculum. A sam-
pling of students' answers and
the mapping is given in Figure
3. Although they may not yet re-
c. alize drying is a "mass transfer"

Summer 2004

operation or that the study of heat transfer is needed to fully
design a reaction kettle, these points are made during the class
discussion in addition to other connections that encompass
the entire core curriculum. This addresses part of the third
goal for the course.
At this time we further emphasize the complimentary na-
ture of the fields of chemistry and chemical engineers.
Through using examples such as new drug development or
the discovery of a novel polymer, the differences in the work
of a chemist conducting a bench-scale study and a chemical
engineer performing a pilot plant investigation and full pro-
duction are discussed.
The students need to realize the significance of basic chem-
istry knowledge to their success as chemical engineers. For
example, a one-to-one mole ratio reaction occurs in the first
reaction kettle. In the stirred reactor, an acid catalyzed poly-
condensation reaction is initiated. By adding acid to this re-
actor, a neutralization reaction also occurs with the caustic
that was added upstream to the first reaction kettle. By specifi-
cally including the reaction chemistry, the students immediately
grasp the relationship between the two fields and their true com-
plimentary nature. This addresses the second goal.

The preliminary design of the process allows us to address
a variety of core competencies (Goal 3), beginning with in-
troduction of the laws of conservation of mass and energy.
Material balances are performed to determine the amount of

Material Chemistry

Quantity Reaction
Produced Chemistry

Amounts of
Raw '
Materials Size of
Needed Equipment

Amount of
in Dryer

Figure 3.
Mapping of Kinetics &
student Mass Reactor
ideas to Transfer Design
idea the -Design

raw materials needed to produce a set amount of fertilizer of
a particular composition. Energy balances are used to deter-
mine the amount of steam at a specified pressure needed to
supply heat for the endothermic reaction that occurs in the
first kettle. We also ask the students to determine the amount
of natural gas needed to be supplied to the dryer for a given
efficiency of the dryer. Typically we ask them to size the re-
action kettle and either the curing belt or dryer based on den-
sity information and/or desired moisture content of the final
product. Some of the variables that can be modified from
year to year include the pieces of equipment sized, the pro-
duction rate, the fertilizer composition, and variations in the
emphasis on certain aspects of economics.
In addition to the use of material and energy balance con-
cepts, the students begin to understand the role of chemical
reactions and how it affects other principles. For example,
since the polycondensation and neutralization reactions both
produce water, this additional water must be taken into ac-
count when determining the amount of water that must be
evaporated in the dryer to achieve the desired moisture con-
tent and production rate.
We sometimes give students insufficient information, which
means they have to use steam tables or find information in
the literature. An added benefit of using the literature is that
it requires the students to actually visit the library to find the
needed information, and when reading the literature they are
forced to use their "engineering judgment" to determine the
most appropriate information. For example, as part of their

Chemical Engineering Education

economic analysis, we require that the students use the Chemi-
cal Market Reporter to find current cost data for the raw ma-
terials. Sulfuric acid, which is used in this process to catalyze
the polymerization reaction, is listed as 100% whereas our
process calls for 20 wt%. This requires the students to actu-
ally think about the proper way to use the information they
find, as opposed to the plug-n-chug approach that many stu-
dents entering college are accustomed to. As a result, they
discover that not all the information needed to solve a prob-
lem will be available in the "back of the book." Our hope is
that by requiring the use of the library to find relevant infor-
mation in current literature early in their educational devel-
opment, students will continue to use the technique through-
out their lives. (Goals 2, 3, 5)
After the actual process, as depicted in Figure 2, is pre-
sented to the students and a discussion is held regarding fer-
tilizers and how and why they are important to society, the
idea of design as a systematic process with explicit steps is
introduced. For example, all design ultimately begins with
identification of a need, definition of the problem, and a search
for relevant information.
The class members make a list of potential sources of in-
formation, ranging from published works to operators in the
plant. The iterative nature of the design process is noted. For
instance, once the constraints (such as a budget or available
land) are established and the product criteria is determined, a
need for additional research or refining the problem defini-
tion may be necessary.
Realizing that various solutions to the same problem are
possible gives rise to a discussion of the weighting of con-
straints and criteria as part of the decision-making process.
We reiterate the idea of design as a systematic process at vari-
ous stages, including near the end of the course when we
review what we've accomplished. (Goals 3, 5)

The economic analysis component of the project affords
the students opportunities to make decisions, using their en-
gineering judgment, re-
garding the cost of various
aspects such as raw mate-
rials and utilities, labor TT
The Classroom al
costs, supervisory costs,
and type of processing
(fluid, fluid-solid, solid), Multiple solutions 4 1 Engin
and to notice how these Role of the market -4 Econo
connect to the total prod- Working in teams 41 Systen
uct cost. The ratio factors W c
Written communication 11 Comi
based on delivered-equip-
Problem solving 11 Presen
ment cost are used to de-
termine the fixed and total Time management 41 Infom
capital investment.'21 The Project management ,4 Summ
students use a variety of

graphs and charts to size and cost the equipment and to deter-
mine the number of employee hours needed based on the num-
ber of processing steps. They are given cost information for
one capacity and apply the 0.6 power law in order to scale
the known information to the project conditions. They make
use of the up-to-date Marshall & Swift Cost Index as pub-
lished in Chemical Engineering to estimate current cost us-
ing older data. Throughout this project, we introduce the stu-
dents to many of the concepts needed for a preliminary de-
sign, which aids in their understanding of the real-world de-
cision process. (Goals 3, 5)

As the ideas needed to analyze the project progress, we
incorporate various additional "soft skill" items (Goal 4). For
example, we require the students to work in teams during the
last half of the quarter. Usually one week before they select
team members, we ask the students to silently brainstorm char-
acteristics of a "good" team (a team they would like to be-
long to) and those of a "bad" team (a team that they would
not want to belong to). The instructors and/or students then
list these characteristics on the board or on an overhead, and
we discuss the potential impact of these traits on selecting
team members and on team performance.
We introduce the students to the four stages of team dy-
namics: forming, storming, norming, and performing3' and
provide them with sample meeting agendas and minutes,
along with tips on building consensus. We then ask them to
formally submit at least one set of meeting agenda and min-
utes along with the documented outcomes, such as a timeline,
deployment chart, and/or a Gantt chart. By indicating how
they have divided their work and the timeline that they set to
complete the project, the students are introduced to more ef-
fective ways of operating within a team setting, time man-
agement, and project management (Goal 5). Since the report
is a team effort, it is important that the students have a com-
mon framework to which they can refer for timing and re-
Throughout the quarter,
we stress the importance
of proper documentation
LE 2 of technical work with the
e Practicing Engineer weekly problem sets. The
students must submit a
judgment formal written report de-
must be considered tailing the entire analysis
along with their conclu-
approach to problem solving along with their conclu-
sions. The report is a team
citing with peers and supervisors effort (Goal 4). Addi-
work in logical steps for others to follow tional optional activities
must be sought from new sources include the preparation
ig results in project report and presentation of a
poster or building either

Summer 2004

d th


a partially working or nonworking model of the process.
Students are also exposed to the role of the market for the
particular product. We discuss the effects of competitors in
terms of "What do they have that you don't?" or "Can you
offer a 'better' product than anything already on the mar-
ket?" We ask the students to make recommendations regard-
ing further pursuit of the project based on their preliminary
design and the market conditions. The effect of the competi-
tors' prices must be directly considered in this evaluation.
We introduce the concept of supply and demand in terms of

questions such as, "Is the market demand such that you can
sustain a long-term reasonable share in the market?" (Goal 5)
Some of the ways in which the classroom mimics the environ-
ment of a practicing engineer are summarized in Table 2.

While the project (see Figure 1) is the focal point for meet-
ing Goals 2-5, the first goal is approached in such a way as to
serve as the introduction to the project. On the first day of
class, we ask the students to brainstorm ideas of where in our

Select all that apply.
1. Consensus
b) is a unanimous or majority vote
O means that everyone agrees to
support the decision
) involves clarification and
e) means that everyone gets
everything they want

2. The steps of the design process
a. Direct charges
SIdentification of a need
Constraints, such as cost or time
d. Maintenance
e. Steam Tables

Fill in the blank.
3. A material balance is a statement (or an application) of the law of conservation
of mass.
4. Storming is a stage of team growth or team dynamics in which the group
members are in a state of panic, often characterized by each member working
independently rather than together.
Short Answer
5. List 3 advantages of working in a team.
Part II: Problems.
1. A plant is to be designed to produce a slow release fertilizer with analysis of 30-12-
10. MAP with an analysis of 13-52-0 is to be used as a phosphorous source and KCI
with analysis of 0-0-60 is to be used as a potassium source. Fertilizer grade urea with
analysis of 45-0-0 will be the primary nitrogen source. UFC with composition 60 wt%
formaldehyde (H2CO), 25 wt% urea and 15 wt% water will be the formaldehyde (H2CO)
source. A partial list of the raw material amounts to be used is shown in the table below:
Raw material Amount (a) What is the fertilizer production capacity (in
(Ib/hr) lb/hr) for this plant? A periodic table is
MAP 6,923 attached at the end of the handout. If you
KCI 5,000 cannot solve this part, use a value of 25,000
UFC 6,400 Ib/hr to continue the problem. (25,000 Ib/hr is
Solid Urea not the correct answer.)
Water 1,690 (b) What is the required amount of solid urea (in
Surfactant 112 lb/hr) for this plant?
H2SO4(20%) 1,250 (c) Find the amount (in lb/hr) of steam at 50 psi
NaOH(20%) 95 needed to supply the necessary heat for the
methylol urea reaction if it is known that the
efficiency of the heat exchanger is 75 % and the methylol urea reaction consumes
23,500 BTU per Ibmole of methylol urea produced. A steam table is attached at the
end of the handout.
(d) What is the amount of water (in Ib/hr) produced in the reactor as a result of the
chemical reactions? You do not need to write the reaction chemistry or to explain it
in detail.
Figure 4. Sample exam questions.

Chemical Engineering Education

society chemical engineers play a role or influence the soci-
ety in some way. A "cartoon" depicting items such as paint
cans, gasoline pumps, plastic containers, and space ships is
sometimes used as the springboard for this discussion. This
introduces some of the more traditional career paths for chemi-
cal engineers such as the process engineer at chemical, plas-
tics, or petrochemical plants. Students jump readily from tra-
ditional areas to some of the cross disciplines, such as envi-
ronmental concerns, giving us an opportunity to talk about
careers in environmental firms, OSHA, or other regulatory
agencies. Pharmaceuticals, an area of strong interest in our
area due to the nearby Eli Lilly and Pfizer plants, is always
mentioned by the students, allowing us to further bring into
the discussion the cross-disciplinary nature of the field. We
introduce the various minors and areas of concentrations avail-
able in the curriculum, including biomedical, biochemical,
environmental, chemistry, semiconductors, polymers and
plastics, and a certificate in engineering management. This
initiates thinking "outside the box" for potential careers and
provides an opportunity for the students to use foresight in
preparing their plan of study in order to incorporate specific
areas of interest.
One additional area of focus is the role of professional so-
cieties. We strongly encourage the students to become active
members of AIChE and at least one other professional soci-
ety, such as SWE, ISPE, NSBE, ACS, or AXE. We stress the
many benefits of membership such as networking, introduc-
tion to various companies through talks hosted by the societ-
ies, exposure to a wider range of ideas and applications
through the membership journals, etc.
One of the most important ideas presented to the students
during this course results directly from their introduction to
the professional societies-the engineering code of ethics.
Every year we present the engineering code of ethics, either
from NSPE or AIChE, to the students. We discuss some of
the ways in which the code applies in their life today, such as
honesty and integrity. We then ask them to submit as a home-
work assignment how they will apply each area of the code
of ethics during their undergraduate career.

The degrees to which the goals for the course are met are
assessed based on a set of student learning objectives that
describe the concept that is to be mastered and how that mas-
tery will be measured. The course assignments, ranging from
the weekly homework assignments to the final project and
course exam, are mapped to these learning objectives to verify
that each objective was assessed appropriately. A few
sample questions from a recent exam are presented in Fig-
ure 4. The types of questions range from "short answer"
and "fill in the blank" to challenging problems. Also, a
questionnaire for seniors is currently being developed to
assist in analyzing the effectiveness of the course in pro-

Summer 2004

moting learning in subsequent courses.

The course gives students a realistic taste of what it is like
to be a practicing chemical engineer and what it takes to be-
come one. Students are given a project that deals with the
preliminary design and economic analysis of a chemical plant.
Through various homework and class assignments they are
required to apply the steps in engineering design, to partici-
pate in team assignments, and to solve open-ended problems
where answers are obtained over weeks instead of minutes.
They work on the project individually and in groups under
the supervision of the faculty instructor. These open-ended
problems involve applying engineering constraints, making
approximations and engineering judgments, and using team-
ing skills as well as communicating the results through a for-
mal written report.
In their class activities, students emulate practicing engi-
neers, which keeps their interest and motivation high despite
the various challenges. At the end of the course, they have a
much clearer focus and understanding of what lies ahead of
them, not only in their undergraduate career but also in the
initial years as a practicing engineer.
By using the project-based approach with an actual chemi-
cal process design, students are better able to grasp the ba-
sics of chemical engineering and improve or develop team-
ing and communication skills. Most students find the project
challenging and develop a better understanding of the pro-
fession and the curriculum. They leave the course with a sense
of accomplishment and pride having completed a "real"

Professor Emeritus Carl Abegg created the controlled-re-
lease nitrogen fertilizer plant-design problem and the basic
project content in 1990. Professor Emeritus Jerry Caskey
continued to use the project after Carl's retirement. Professor
Abegg is thanked for his helpful discussions regarding the
course and its content. The many discussions with and input
from Professor Serbezov regarding the course are greatly ap-
preciated. Professor Robert Sauer and Mr. Samuel Bunch are
acknowledged for their suggestions to improve the manu-
script. Mrs. Mary Wade contributed to the design of Figure
3. Parts of this work were presented at the ASEE 2003 An-
nual Meeting under a similar title.

1. Richard H. Czurak and Robert M. Thompson, Foamed Fertilizers and
Combination Products, US Patent 3,705,794, December 12, 1972
2. Peters, M.S., K.D. Timmerhaus, and R.E. West, Plant Design and Eco-
nomics for Chemical Engineers, 5"h ed., McGraw Hill, Boston, MA
3. Scholtes, Peter R., B.L. Joiner, and B.J. Streibel, The Team Hand-
book, 2nd ed., Oriel Inc., Madison, WI (2001) 0

[1WR= laboratory

An Integrated



Universidade do Porto Porto, Portugal

Chemical reaction engineering (CRE) represents a fun-
damental topic in the undergraduate chemical engi-
neering curriculum.1" It is also, however, a complex
and multifaceted subject that students may fail to grasp as a
whole. Traditional teaching approaches are often confined to
the theoretical treatment of ideal systems, blurring the prac-
tical engineering implementations of CRE concepts. Issues
such as residence time distribution (RTD) characterization
techniques, treatment and implications of flow nonidealities,
or prediction of a reactor's performance from the combina-
tion of RTD and reaction kinetics data, need to be practiced
by the students in the lab in order to achieve a good level of
In our senior lab course, we have implemented a set of
three experiments (each to be performed in an individual
three-hour lab session) that integrate some fundamental con-
cepts of CRE. These experiments allow students to under-
stand the sequential procedure for characterizing a chemical
reaction system: a) determination of the kinetic parameters
for the reaction in question; b) characterization of the reactor's
flow pattern (resident time distribution); c) implementation
of a model for predicting the reactor's performance (conver-
sion), based on the information collected in the two previous
items. The model is validated by comparing its results to ex-
perimental data. The content of each lab session is summa-
rized below.
> In the first session, students determine the rate constants of a
second-order reaction at different temperatures, using a
batch reactor. The reaction in question is the ethyl acetate

CH3COOC2H5 + NaOH- -> CH3COO-Na+ + C2HOH (1)
We took the widely used saponification reaction, incorporat-
ing the acid base indicator indigo carmine into the reaction
medium. Indigo carmine reflects the change in the reaction's
medium pH with conversion as it undergoes a color change

from blue to yellow/green. This change allows the students to
visually observe the reaction's evolution as a function of
> In the second session, students characterize the RTDfor a
continuous-flow reactor-a tubular reactor packed with
glass beads. They discover the reactor may perform non-
ideally, a fact that becomes apparent to them during their
visual observation and data analysis. The realization that the
ideal plug flow model may inadequately describe the
reactor's flow pattern leads students to the need for a more
elaborate RTD model. They then perform two tracer
experiments: concentration step change and concentration
pulse. These experiments allow the students to compare the
two methods and to discover that both experiments lead to
equivalent results. The transparent reactor walls allow the
students to visually track the tracer's advance along the
reactor and to see the effects of axial dispersion, such as the
broadening to the tracer pulse.
I In the third session, students evaluate the reactor's perfor-
mance. They measure the conversion and compare it to the
theoretically predicted value, calculated using the kinetic

Adelio M. Mendes is Associate Professor of Chemical Engineering at the
University of Porto, where he also graduated in chemical engineering (1987)
and earned his PhD (1993). He teaches chemical engineering laborato-
ries, separation processes, and numerical methods. His main research in-
terests include membrane and sorption gas separations, catalytic mem-
brane reactors, and fuel cells.
Luis M. Madeira is Assistant Professor of Chemical Engineering at the
University of Porto. He graduated in Chemical Engineering (1993) and re-
ceived his PhD (1998) from the Technical University of Lisbon. He teaches
chemical engineering laboratories and chemical reaction engineering. His
main research interests are in heterogeneous catalysis, catalytic membrane
reactors, and wastewater oxidation.
Fernio D. Magalhies is Assistant Professor of Chemical Engineering at
the University of Porto where he graduated in chemical engineering (1989).
He received his PhD (1997) from the University of Massachusetts. He is
currently teaching chemical engineering laboratories and advanced calcu-
lus. His main research interests involve mass transport and sorption in
porous solids and membranes.
Jose M. Sousa is Professor Assistant in the Chemistry Department at the
University of Tras-os-Montes e Alto Douro. He is a PhD student t the Uni-
versity of Porto, where he received his degree in chemical engineering in
1988. His research interests include catalytic membrane reactors.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

and RTD data collected in the first two sessions. They
discuss the implications of the axial dispersion effects and
the validity of the ideal plug-flow model. Because the
students also collect transient conversion data, they can
discuss the theoretical predictions during the buildup of
steady-state conditions. Again, the pH indicator, present in
the reactant feed, allows observation of the reactor's axial
concentration gradient.
The fact that all experiments have a strong visual element
is quite relevant in terms of helping students to understand
some of the phenomena involved, particularly during the
tracer experiments (which they consider to be the most at-
tractive). In addition, the reactants are environmentally harm-
less and all experiments are intrinsically safe and inexpen-
sive. They also incorporate computer-assisted data acquisi-
tion, thanks to the serial communication interface that comes
with the measurement device (conductivity meter).
There is a web site that complements this paper and sup-
plies additional information as well as photographs of the
setup and experimental runs. It can be found at>.

The same setup (with small modifications) is used for all
three experiments. It consists of a peristaltic pump (from
Watson-Marlow, Model #5058), a microprocessor conduc-
tivity meter with temperature compensation (from EDT In-
struments, Model #RE 387 Tx) connected to a PC through an
RS-232 interface, a thermostatic bath with cooling and heat-
ing (Huber, Polystat ccl), and two conductivity electrodes,
one of them of the flow-through type. A program for data
acquisition and monitoring of the conductivity measure-
ments was developed in Labview (National Instruments).
Specific details of operation for the three experiments are
described below.

1) Determination of the reaction rate constant in a batch
reactor A glass-jacketed batch reactor, with a volume of
300 cm3 and equipped with magnetic stirring, was used for
this experiment. Temperature was measured with a mercury
thermometer, having a precision of 0.10C. The reactant so-
lutions employed were sodium hydroxide (0.2 M, 100 cm3)
and ethyl acetate (0.25 M, 100 cm3), this one containing a
small amount of indigo carmine (0.005% wt.), which is a pH
The reaction occurs too rapidly at temperatures above 30C
and too slowly below 10'C. Such high or low temperatures
also make it difficult to control the reaction medium tem-
perature since they differ significantly from room tempera-
ture. Finally, at high temperatures (above 300C), the evapo-
ration of ethyl acetate becomes significant. Therefore, stu-
dents measure the rate constant at three temperatures between
150C and 25C (15.0, 19.9, and 25.0 in the present case).

Typically, in a three-hour lab session, no more than three ex-
periments can be performed.
The concentration of the limiting reactant (NaOH) is mea-
sured by conductometry. The calibration procedure for this
method is described in Appendix A.

2) Flow pattern characterization in the packed-bed tubu-
lar reactor We use an acrylic tubular reactor of 101 cm in
length (L) and 3.6 cm in internal diameter. It is packed with
glass beads (d = 3 mm) and, since it has an effective volume
(V) of 372 cm3, the packing porosity can be assumed to be
0.36. The reactor is fixed to the workbench in the vertical
position. A static mixer is introduced at the reactor's inlet
(bottom) for homogenization of the two reactant streams. It
consists of a cylinder (about 1 cm3) filled with small glass
beads (d = 1 mm).
The tracer is a KC1 solution and the detection method is
again conductometry, using the flow-through-type electrode
at the reactor outlet. Indigo carmine is added to the tracer,
but this time its only purpose is to give color to the solution.
One of the experiments is performed as a negative step in-
put. For this, the reactor is initially filled with a 0.1 M tracer
solution (with 0.01% wt. indigo carmine). The step purge is
performed by feeding distilled water to the reactor at a volu-
metric flow rate (v) of 58.3 1.0 cm3min-' (this represents
the average of at least three measurements made during the
run, using a 50 cm3 graduated cylinder and a chronometer).
The other experiment is a pulse input, which is performed
using an injection valve connected to a 10 cm3 loop. The re-
actor is initially filled with distilled water. The KCl solution
we use as tracer is now 1.0 M (with 0.1% wt. indigo car-
mine). The volumetric flow rate of distilled water used dur-
ing operation is 66.6 1.0 cm3min-'. Considering the volume
of the injection loop, this implies that the tracer pulse has a
duration (At) of 9.0 s at the inlet.

3) Determination of reaction conversion in the packed-
bed tubular reactor The saponification reaction is per-
formed in the continuous-flow tubular reactor at room tem-
perature (17.4C in the present case). Both ethyl acetate (0.25
M containing 0.01% wt. of indigo carmine) and sodium hy-
droxide (0.2 M) solutions are fed to the bottom of the reactor
in a proportion M = CEAco /CNaOH = 1.25. The total volu-
metric flow rate is, for the data presented here, 64.1 1.0
cm3min-'. The NaOH concentration at the reactor outlet is
once again measured by conductometry, using the flow-
through type electrode and the calibration procedure described
in Appendix A. To evaluate the transient conversion, the mea-
sured conductivity must always be between KO and K--oth-
erwise the calibration method does not apply (see Eq. 13 in
Appendix A). This implies that the initial NaOH concentra-
tion inside the reactor must be CNaOH0, i.e., the same as in

Summer 2004

the feed stream. Therefore, the reactor is initially filled with
a 50% mixture of the 0.2 M NaOH solution and water.
Some other aspects related to the experimental setup were
not mentioned here. For instance, students must know the
dimension of the outlet tube between the reactor and the flow-
through electrode, since the out-flowing liquid continues to
react as it travels along it. This contribution to the overall
conversion must be deducted from the experimental data. For
simplicity, plug flow behavior can be assumed in the tube. In
our case, this correction is almost negligible, due to the small
space-time in the tube. For additional information regarding
the experimental assembly, the reader can go to>. Photographs of experimental
runs are also shown, illustrating some important issues that
are described below.


1) Determination of the reaction rate constant in a batch
reactor Figure 1 shows typical results that illustrate the
evolution of conductivity with time in the batch reactor at
different temperatures. As expected, conductivity decreases
along time, since one CHCOO ion forms for each OH- ion
consumed (see Eq. 1).
It is known that the ethyl acetate saponification reaction is
first-order in relation to each reactant.[31 The mass balance
for an overall second-order reaction, taking place in a con-
stant-volume (liquid phase) batch reactor, leads to the fol-
lowing result for the limiting reactant's (NaOH) conversion:[41

eCNaOHo(M-l)kt _1
XNaOH = M 1 (2)
MeCNaOHo(M-1)kt 1

with M = C EAc /C NaOHo 1 The kinetic constant, k, can thus
be obtained by applying the integral method to the experi-
mentally measured transient conversion, i.e., from the slope,
CNaOHO (M )k of the linear plot of


versus time. It should be stressed, however, that as conver-
sion approaches unity, even small errors in the calibration
procedure (and thus on XNaOH) will cause significant devia-
tions from linearity in such a plot. A simple numerical ex-
periment can be suggested to students to demonstrate this:
first, one computes XNaOH from Eq. (2) and plots it according
to the linearization strategy described above. Then, one re-
peats the plot, adding a fixed "error" to XNOH (for instance,
0.01). This second curve deviates significantly from linear-
ity, especially at high conversions. Students are therefore ad-
vised not to include data corresponding to high conversion
values (typically above 90%) in the linear fitting. A valid al-

ternative to this strategy consists of performing a nonlinear
fitting, using Eq. (2) directly (XNaOH versus time). This gener-
ally leads to a more accurate estimation of the reaction rate
Figure 2 shows the kinetic rate constants obtained at dif-
ferent temperatures. Arrhenius behavior is observed, with an
activation energy (Ea) of 39.9 kJ mol and a frequency factor
(ko) of 1.05x103 m3mol-'s-1. Students are strongly encouraged
to compare their results to literature values. This gives them
a reference point for analyzing the validity of their work and,
eventually, makes them feel confident about their own capa-
bilities. In the present case, a published paper13 reports a
kinetic constant at 230C that differs only 4% from the one
computed from the Ea and ko values reported above (k =
9.8 x 10-5 m3mol-'s-i).
The data shown in Figure 2 represent typical results ob-
tained by students in a lab session. Ideally, more data points
should be used for building an Arrhenius plot-as mentioned
before, however, due to class time-length restrictions (three

18 --- -- -


0 500 1000 1500 2000
t (s)
Figure 1. Conductivity of the reaction mixture in the
batch reactor, as a function of time, at
different temperatures.

Figure 2. Arrhenius plot of the reaction rate constants
for the second-order reaction between ethyl
acetate and sodium hydroxide.

Chemical Engineering Education



-9.25 -

-9.40 -


-9.70 L

0.00336 0.00340
1/T (K-1)

0 00344 0.00348

hours), kinetic measurements can only be performed at three
different temperatures.
As the reaction proceeds, sodium hydroxide is consumed,
therefore decreasing the pH of the medium. Thus, the acid-
base indicator indigo carmine (which has a color transition
from blue to green/yellow in the pH range 11.5 to 13.0) al-
lows for a visual evaluation of the reaction progress.
2) Flow pattern characterization in the packed-bed tubu-
lar reactor Figure 3A shows the response of the packed-
bed reactor to a negative step change in terms of normalized
concentrations, i.e., the so-called Danckwerts' P curve. The
slanted shape of the curve indicates that the flow dynamics

0 200 400 600 800


1 0.050

0025 -

0 200 400 600 800
t (s)
Figure 3. KCI normalized concentration data at the outlet
of the packed-bed tubular reactor for (A) a negative step,
and (B) a pulse input. The lines represent the fittings ob-
tained using Eqs. (4) and (9), respectively. The fitted pa-
rameters are shown in Table 1.

Fitted Parameters for the
Flow-Pattern Characterization Runs

SFitted Parameters
Tracer Experiment 7(s) Pe Obj.Fctn.
Negative step input 383.6 158.7 1.4 x 102
Pulse input 358.3 181.5 1.5 10-3

do not obey the ideal plug-flow pattern. Therefore, a more
complete model must be used to describe the data. It is known
that, for a semi-infinite axially dispersed plug-flow reactor,
the residence time distribution (RTD) function is given by1[67'

E(t) e 4t (3)
2 tt3

where r is the space-time and Pe is the Peclet number. Stu-
dents are informed that other equations can be found in the
literature for the E(t) of a plug-flow reactor with axial dis-
persion, depending on the boundary conditions used. For small
extents of dispersion (which is actually the case, as will be
discussed below), however, the shape of the curve is insensi-
tive to the boundary conditions imposed.14' The particular for-
mulation adopted here has to do with the fact that an analyti-
cal solution exists and that this has a relatively simple form.
The outlet tracer concentration can be obtained from the
E(t) by integration of Eq. (3), which provides the P(t) curve

oo t t e pe( -t)2
Pt= = 1- (t)dt= 1- e t dt (4)
C 0 0 2 it3

The parameters T and Pe can be determined by fitting the
model equation to the experimental data. This implies com-
bining numerical integration and nonlinear fitting. Students
can easily perform this task by using, for instance, the "Solver"
add-in in Microsoft Excel. An initial estimate of the Peclet
number for the fitting procedure can be obtained from avail-
able correlations, e.g., using the expression proposed by
Chung and Wen'81 for a packed column with inert nonporous
S 0.48
0.2 +0.011 pud, L0(5
Pe= -- (5)

where L is the length of the column, d is the average particle
diameter, E is the bed porosity, u is the superficial velocity,
and the other parameters refer to fluid properties. For this
experiment, Eq. (5) leads to Pe = 202.9. A simpler estimation
is based on the particle's Peclet number (Pe )

Pe=Pe L (6)

In the current case, the Reynolds number is lower than 100,
which implies that Pe varies in the range of 0.3 to 1.0.[91
Therefore, Eq. (6) yields an average value of 218.8 for Pe
(ranging between 101.0 and 336.7).
Table 1 shows the parameters obtained from the fitting.
The objective function minimized was the sum of the squares
of the residuals. Figure 3A clearly shows that the E(t) curve
obtained with these parameters reproduces the experimental

Summer 2004

data quite well.
Another very common tracer technique is the pulse input.
If one can overcome the drawbacks associated with the imple-
mentation of a pulse injection, this technique represents a
straightforward and less costly (less tracer is spent) way of
obtaining the RTD.[rOJ The concentration pulse can be math-
ematically formulated as
Co(t) = Co[H(t-0) H(t-At)]
where At is the duration of the perturbation and H(t) is the
Heaviside function. The response of a continuous reactor to
a pulse input in the feed stream is[71

C t t
C(t)- = E(tdt E(t Atdt (7)
Co 0 0
where C(t) is the Danckwerts' C curve. For a sufficiently small
pulse, i.e., At -> 0, Eq. (7) can be simplified to

t t (t
E(t)dt-jE(t- At)dt d jE(t)dtj
C(t)= At0 o At =At dt =At.E(t)
At dt

Thus, for an infinitesimal pulse, Eq. (8) will show a maxi-
mum at t = T. For a finite pulse, with length At, the maximum
in the concentration will come out at t = T+At/2. Extending
this to the RTD function given by Eq. (3), the C curve be-

C tPe 4T(t+At/2)
C(t)=- =At T e 4 (9)
CO 2 U(t+At/2)3

Figure 3B shows the results of the pulse input tracer ex-
periment. Equation (9) is fitted to the experimental data,
yielding the parameters indicated in Table 1. The good
agreement between the experimental and theoretical
curves in Figure 3B is particularly noteworthy, indicating
once more that the reactor is well described by the axially
dispersed plug flow model.
The otherwise colorless KCl tracer solution used in these
experiments contains indigo carmine dye. This allows stu-
dents to observe propagation of the concentration waves as
they travel along the reactor. It is interesting to note that, for
the pulse-input experiment, the tracer pulse is about 10 cm
in length at the beginning of the run, but it noticeably
spreads as it approaches the outlet, evidencing the exist-
ence of dispersion effects. Students find this experiment
particularly interesting.
It is also interesting to actually look at the form of the E(t)
functions obtained with the parameters computed from each
tracer experiment. This is more conveniently done if one uses
the normalized RTD function

-e Pe(1l-)2
E(e)= -E(t)= e 40
2 7te3

where 0 = t/T. For the Pe and T values obtained from the two
experiments, one obtains the plots shown in Figure 4. The
similarity between the two curves is quite remarkable, con-
sidering that they were obtained by two distinct methods and
under different operating conditions.
In a semi-infinite tubular reactor, the mean residence time,
defined as

t= ftE(t)dt

or = = JE(0)d9

is related to the space-time according to61'

ti=T +-L or = 1+- (11)
( Pe) Pe
From Eq. (11), as long as Pe is sufficiently large, as in the
present case, t must tend to T, or 6 to 1. The RTD curves in
Figure 4 show indeed that 0 is close to unity. This agreement
indicates the absence of mixing problems in the reactor, such
as short-circuiting or dead volume. The fact that Pe is rela-
tively large traditionally implies that dispersion effects
can be assumed as unimportant. That is also indicated by
the almost-Gaussian shape of the RTD curves in Figure
4.141 Does this mean that axial dispersion can be neglected
in this system (even though the presence of dispersion
effects is evident from the experiments)? We'll get back
to this question later.
3) Determination of conversion in the tubular reactor *
Sodium hydroxide and ethyl acetate solutions are fed to the
reactor until steady state is achieved, i.e., a constant conduc-
tivity is measured at the reactor outlet. It is particularly inter-
esting to observe the color gradient along the reactor-yel-

0.5 1.0

1.5 2.0

Figure 4. Normalized RTD function for the packed-bed
tubular reactor based on the negative step and pulse
experiments (Pe values are shown in Table 1).

Chemical Engineering Education

lowish at the entrance and blue at the exit. Students give par-
ticular attention to this fact and easily relate it to the concen-
tration gradient in the reactor, which they've studied in the
theoretical classes. At steady state, an average conductivity
of 11.86 mS/cm is attained, corresponding to a sodium hy-
droxide conversion of 77.5% (see Figure 5).
For a given RTD and reaction orders greater than unity, the
total segregation model and the maximum mixedness model
represent the upper and lower limits for conversion, respec-
tively."I0 The total segregation model assumes that fluid ele-
ments having the same age (residence time) "travel together"
in the reactor and do not mix with elements of different ages
until exiting the reactor. Because no mass interchange occurs
between fluid elements, each one acts as a batch reactor and
the mean steady-state conversion (X) in the real reactor is
given by10'

X= Xbatch(t)E(t)dt (12)
where X ba(t) is given by Eq. (2) and the RTD function
by Eq. (3), assuming that the axially dispersed plug-flow
model properly describes the flow pattern in the packed-
bed reactor.
Using the two Peclet numbers obtained from the two flow-
pattern characterization experiments (step- and pulse-inlet
perturbations) (see Table 1) and the space-time correspond-
ing to the present operating conditions, the total segregation
model leads to a theoretical steady-state conversion of 77.5%
(for Pe = 158.7) and 77.6% (for Pe = 181.5). Differences
between the two results are minimal and both agree very well
with the experimental result of 77.5%.
Axial dispersion is undoubtedly present in this system and,
as shown before, it must be taken into account in order to
accurately describe the flow pattern in the reactor. The in-



0 :m --Expenmental Data
Mass Balance
Segregation Model Plug Flow

0 00
0 150 300 450 600 750
t (s)

Figure 5. Sodium hydroxide conversion at the outlet of the
tubular reactor. The lines were obtained using the total seg-
regation model and the mass balance explained in Appen-
dix B (with Pe = 158.7 and T= 348.3 s).

Summer 2004

quisitive engineer, however, might ask if the simpler ideal
plug-flow model might not reasonably estimate the reactor's
conversion. Therefore, we ask the students to also compare
the experimental steady-state conversion with the ideal plug-
flow model prediction (i.e., using Eq. (2) with the time t re-
placed by the space-time T). Interestingly, this model leads to
a conversion of 77.7%, which is also remarkably close to the
experimental result (77.5%)! This comes as no surprise if
one considers the relatively low degree of axial dispersion
present, which can be recognized a priori if one uses Eq. (6)
for a first estimate of the Peclet number. The high L-to-dp
ratio (L/d = 337) leads to a relatively high Pe. In conclusion,
the ideal plug flow model is not completely worthless. In this
particular system, it is a straightforward and useful tool for
estimating the reactor's steady-state conversion. Students'
attention is drawn to this issue, which may save time and
effort in real-world engineering situations.
A more advanced topic that results from this experiment
has to do with the prediction of the reactor's start-up behav-
ior, i.e., the transient before the steady-state conversion is
attained. The segregation model can also be used for this
purpose, as long as the upper integration limit in Eq. (12) is
set to t. The result obtained is shown in Figure 5. The agree-
ment between the experimental and theoretical data is very
good. The slight time lag between the curves is due to the
difficulty in measuring the space-time rigorously. It must
be noted that the ideal plug-flow model is not able to de-
scribe this behavior.
An alternative way to predict the conversion is based on
the solution of the reactors' differential mass balance (see
Appendix B). These computations involve solving a system
of two partial differential equations (PDEs), which is far too
advanced at an undergraduate level. It may be interesting to
supply students with a software simulator that solves the bal-
ance equations, however, so they can analyze the results. The
transient conversion computed from this approach is also
shown in Figure 5 (see Appendix B for details). A steady-
state conversion of 77.2% is obtained (for any of the Peclet
numbers mentioned above), which is once again in very good
accordance with the experimental value. As expected, the
steady-state conversion computed from the mass balance is
lower than the one obtained from the segregation model-
77.5%-since the reaction order is greater than one. Differ-
ences are minimal, however, due to the high value of the Peclet
number, i.e., low dispersion.
We frequently suggest that students perform an additional
experiment, involving the removal of the static mixer at the
reactor's inlet, so that the two reactants enter through a simple
"Y" connection. This causes improper mixing, leading to
partial segregation and a decrease in the overall conversion.
This fact is actually visible during the experiment: two dif-
ferently colored symmetrical zones, each occupying about
half of the reactor's cross section, are perfectly visible along


its length. This can also be seen in the web site mentioned
previously in this article. Students should ponder the fact that
a simple flow-pattern characterization, such as the tracer ex-
periments discussed here, is unable to detect this problem
since fluid elements end up being mixed at the reactor outlet,
before reaching the detector.

The lab experiment described in this paper allows under-
graduate students to integrate concepts taught in conventional
chemical reaction engineering courses. In three independent
lab sessions (each being about three hours long), they apply
the fundamentals of: 1) determination of reaction rate con-
stants in a batch reactor by applying the integral method; 2)
characterization of the flow pattern (residence time distribu-
tion) in a tubular packed-bed reactor by trace experiments;
3) determination of the conversion in a continuous-flow re-
actor (both steady state and transient behavior). In the final
written report, the information collected in each session is
integrated to obtain a theoretical prediction of the conversion
in the tubular reactor. The good agreement obtained between
experimental and theoretical results helps students feel con-
fident about their capabilities and improves their self-reli-
ance regarding practical application of theoretical concepts.
The experiments are intrinsically safe and cost-moderate,
while the reactants and products involved are nontoxic and
environmentally nonaggressive or even biodegradable (at least
considering the low concentration levels used).
Using transparent reaction vessels and adding an acid-base
indicator to the reactant and tracer solutions introduces a con-
siderable pedagogical content. Some of the concepts involved
are more easily understood and grasped by students, such as
the progression of the reaction in the closed vessel, the exist-
ence of a concentration profile along the tubular reactor, and
the more difficult notion of axial dispersion effects in a
packed-bed reactor. The dispersion effects are more easily
understood by students during the tracer experiments, par-
ticularly in the case of a concentration pulse, which is par-
ticularly fascinating for them.

The authors gratefully acknowledge In8s Pantaledo and
Nuno Barbosa, senior undergraduate students, for obtaining
the experimental results and taking the photographs. We also
wish to thank the Department of Chemical Engineering (Fac-
ulty of Engineering, University of Porto) for providing fi-
nancial support for the set-up of the experiments.

C concentration of tracer, sodium hydroxide, or of ethyl
acetate (mol-m-3)
C(t) Danckwerts' C curve, dimensionless
d diameter of the particles (m)

E(t) residence-time distribution function (s')
E(6) normalized RTD function, dimensionless
E activation energy (Jmol 1)
H(t) Heaviside step function
K initial conductivity in the calibration procedure
(mS cm')
K" final conductivity in the calibration procedure (mS cm-')
k reaction rate constant of the second-order reaction
k frequency factor (m'mol 's-I)
L length of the column/reactor (m)
P(t) Danckwerts' P curve, dimensionless
Pe Peclet number for the tubular reactor, dimensionless
Pep Peclet number for the particles, dimensionless
t time (s)
t mean residence time (s)
T temperature (K)
u superficial velocity (m s-')
V volume of reactor (m3)
v volumetric flow rate (m's-')
X conversion, dimensionless
z length in the axial direction (m)
Z dimensionless length in the axial direction
EAc ethyl acetate
NaOH sodium hydroxide
o entering or initial conditions
Greek Symbols
At duration of the perturbation in a pulse input (s)
e bed porosity
0=t/T reduced time, dimensionless
p. fluid viscosity (N s m-2)
p fluid density (kg m-3)
r space-time (s)

1. Fogler, H.S., "An Appetizing Structure of Chemical Reaction Engi-
neering for Undergraduates," Chem. Eng. Ed., 27, 110 (1993)
2. Shalabi, M., M. Al-Saleh, J. Beltramini, and D. Al-Harbi, "Current
Trends in Chemical Reaction Engineering Education," Chem. Eng.
Ed., 30, 146 (1996)
3. Abu-Khalaf, A.M., "Mathematical Modeling of an Experimental Re-
action System," Chem. Eng. Ed., 28, 48 (1994)
4. Levenspiel, O., Chemical Reaction Engineering, 3rd ed., John Wiley
& Sons, New York, NY (1999)
5. Chen, N.H., and R. Aris, "Determination of Arrhenius Constants by
Linear and Nonlinear Fitting," AIChE J., 38, 626 (1992)
6. Wen, C.Y., and L.T. Fan, Models for Flow Systems and Chemical Re-
actors, Marcel Dekker, Inc., New York, NY (1975)
7. Rodrigues, A.E., "Theory of Residence Time Distributions," in
Multiphase Chemical Reactors, A.E. Rodrigues, J.M. Calo, and N.H.
Sweed, eds., NATO ASI Series, Sijthoff Noordhoff, No. 51, Vol. I,
225 (1981)
8. Chung, S.F., and C.Y. Wen, "Longitudinal Dispersion of Liquid Flow-
ing Through Fixed and Fluidized Beds," AIChE J., 14, 857 (1968)
9. Ruthven, D., Principles ofAdsorption and Adsorption Processes, John
Wiley & Sons, New York, NY (1984)
10. Fogler, H.S., Elements of Chemical Reaction Engineering, 3rd ed.,
Prentice Hall, New Jersey (1999)
11. Petzold, L.R., and A.C. Hindmarsh, LSODA Computing and Math-
ematics Research Division, Lawrence Livermore National Laboratory,
Livermore, CA (1997) 0

Chemical Engineering Education


Calibration Method for NaOH Concentration
During saponification (Eq. 1), one ion CH3COO- forms for each
OH ion consumed. These two species have very different molar
conductivities, allowing for the reaction progress to be followed by
conductometry. For that, students are asked to determine the molar
conductivity of the involved ions in concentrations close to those
employed in the runs. First, they must measure the conductivity (K)
of a sodium hydroxide solution having a concentration equal to the
start value used in the experiments (C NaOHo) (see Figure 6). Ko is
therefore associated with the presence of Na' and OH- ions, and for
CNaOHO=0.1 M it has a value typically around 22.6 mS/cm. The
conductivity of the reaction product eventually obtained when all
OH- is consumed (K"), is therefore associated with the presence of
only acetate and sodium ions and also has to be evaluated. An addi-
tional run is performed for this purpose, reacting a NaOH solution
with 10-20% molar excess of ethyl acetate, to ensure total conver-
sion, and measuring the final conductivity. For
M =CEAc0 /CNaOHO = 1.25 and CNaOH=0.1 M, K is typically
around 7.3 mS/cm. The calibration line, which provides the sodium
hydroxide concentration as a function of the conductivity of the re-
action mixture, is therefore given by (see Figure 6)

K-K (13)
CNaOH = CNaOHo (13)
Since the solutions are sufficiently diluted, linearity can be assumed.

Na + OH +
100 Na+OH+ CH3COO +

0.0- -
0.00 0.05 0.10
S,,o- CNaOH (M) Con

Figure 6. Calibration curve for the method of analysis-
conductivity of the reaction mixture versus the
sodium hydroxide concentration.

A lab technician calibrates the conductivity electrode before
classes so that students only need to normalize the conductivity data
by the conductivity of the tracer solution used. When working with
a relatively broad concentration range, however, nonlinearity may
be present and a correction factor must be considered. For that, stu-
dents use data provided in the operator's manual of the conductivity


Mass Balance in the Axially-Dispersed Plug-Flow
The tracer experiments have clearly shown that the continuous-
reactor is described by an axially dispersed plug-flow reactor. Thus,
its performance can be predicted from the solution of the mass bal-
ance equations.
The system of two PDEs, each one describing the unsteady be-
havior of each reactant, is

1 a2c. ac ac
_i C- CNaOHCEAc (i=NaOH,EAc) (14)
Pe aZ2 Z z at

where Z=z/L is the dimensionless length in the axial direction. The
boundary conditions, considering a semi-infinite reactor (open to
diffusion at the inlet) are

C =Cio at Z=0 (i= NaOH,EAc) (15)
C Co a Z-

0 at Z = 1

(i = NaOH, EAc)

The initial conditions are

CNaOH(t=0)=CNaOHo for VZ

CEAc(t=0)=0 for VZ

The spatial discretization along the axial coordinate was performed
with 51 points, which was found to be quite satisfactory. A three-
point central difference scheme was employed for calculating the
spatial derivatives. Routine LSODA"1 was used for time integra-
The outlet conversion obtained from the numerical solution of
the mass balance equations is shown in Figure 5. This model pre-
dicts quite well the steady-state conversion. Regarding the transient
behavior, one notices that the mean residence time estimated from
the mass balance simulated curve (338.8 s) is smaller than the ex-
perimental space-time (T = V/v = 348.3 s). Indeed, this may be ex-
pected since the decreasing concentration profile in the reactor pro-
vides, in the mass-balance model, an additional driving force for
the reactant's diffusion toward the outlet (and thus a lower mean
residence time).
On the other hand, the total segregation model assumes no inter-
action between fluid elements of different ages and therefore the
mean residence time is dependent on the RTD function alone, inde-
pendently of the effect that the presence of chemical reaction might
have. Thus, the mean residence time is practically equal to 7, as
previously explained. The slight shift of the experimental data with
respect to this model is due to the error associated with the space-
time measurement.

Summer 2004

M 9Ff laboratory



University of Waterloo Waterloo, Ontario, Canada N2L 3G1

As the world population continues to escalate, the de-
mand and consumption of energy also increases.
While conventional means of energy production may
meet current demand in the short term, there are environ-
mental implications as energy demand increases. Pollutants
such as photochemical smog, NOxs, SOxs, and CO2 contrib-
ute to the rising problems of urban-air quality, acid rain, and
global warming. For this reason, fuel-cell technology is
emerging as a preferred method of power generation.
Fuel cells represent a class of electrochemical power sources
that can be continuously refueled, have higher efficiency (~40-
60%) and reliability, allow for modular design, and can be
used in cogeneration systems. They represent an ideal prod-
uct for distributed generation systems that hold the promise
of higher electrical reliability and greater energy efficiency.
Potential applications for fuel cells range from micro cells
for cell phones and pagers to portable fuel cells for backup
and portable power to stationary systems for residential and
large-scale power generation. The most promising applica-
tion for 'polymer electrolyte membrane' (PEM) fuel cells is
in the transportation sector. Many prototype buses and cars
are already in service, and most major vehicle manufactures
have a prototype on the road. PEM fuel cells have been un-
der development for many years. When compared to other
fuel cell technology, their major advantages are their solid
electrolyte and low operating temperature (25 to 90C).
There are numerous text books and other information avail-
able with respect to fuel cells, but this paper will recommend
only two literature sources-they provide a good overview
of the theory and technology as well as being readily avail-
able to undergraduate students on the internet. Sharon and
Zalbowitz'l provide a brief overview of the technology suit-
able for providing an introduction to the field, and the Fuel
Cell Handbook[21 provides more in-depth information as well
as the electrochemistry and thermodynamic basics with
sample calculations.

Since PEM fuel cells are an emerging technology that will
dominate the power generation sector of the future, it is im-
portant that graduating chemical engineers have a good un-
derstanding of the fundamental principles of fuel cell opera-
tion and the supporting electrochemical principles. A fuel cell
laboratory experiment is a good way to achieve this educa-
tional objective. This paper will describe the design and con-
struction of a PEM fuel cell and a fuel-cell test station for an
undergraduate laboratory experiment, a project was under-
taken by a group of five undergraduate students at the Uni-
versity of Waterloo.

A PEM fuel cell uses hydrogen as a fuel and air as an oxi-
dant, which are supplied at the anode and cathode, respec-
tively-a schematic of a PEM fuel cell is shown in Figure 1.
The basic operation of a PEM involves a hydrogen-rich fuel
stream that flows across the porous anode electrode where
the following half-cell reaction occurs:
H2 -> 2H+ = 2e (1)
The separated electron flows through an external circuit re-
turning to the cathode. The proton, solvated in water, dif-
fuses through the membrane to the cathode. An oxidant, usu-
ally air, oxygen, or helium/oxygen, flows across the cathode.
The following is the half-cell reaction that occurs at the cath-

1 02 +2H++2e- H20 (2)

Michael Fowler is on the faculty in the Department of Chemical Engineer-
ing at the University of Waterloo. He has degrees from the Royal Military
College of Canada and Queens University. His current research includes
materials degradation in fuel cells and fuel-cell engineering.>
Alfred Lam obtained his Bachelor of Applied Science in Chemical Engi-
neering at the University of Waterloo in 2003. He is currently in pursuit of
his MASc in Chemical Engineering, with research interests in the material
development and reliability of PEM fuel cell components.

Copyright ChE Division of ASEE 2004

Chemical Engineering Education

The proton and returning electron combine with oxygen to
form water. The net reaction is

H2 + 02 H20 (3)

The membrane electrode assembly (MEA) consists of a
polymer electrolyte membrane, electrodes, and a gas-diffu-
sion layer. Within a fuel-cell stack, the MEA is compressed
between two bipolar plates. Therefore, the key components
of a PEM fuel cell are
A polymer electrolyte membrane
An electrode
A gas diffusion layer (GDL)
In designing a PEM fuel cell, the objective was to select an
appropriate "scale" for the active area so the students would
be able to generate a reasonable current, could vary the vari-
ous parameters, and could observe the operational variabil-
ity. A further constraint was a very limited budget for the fuel
cell and test station.
Ultimately, the design group developed a test station for
the operation of a PEM fuel cell with an active area of 50 cm2
(a 100-cm2 cell was also designed and constructed by stu-
dents). Note that a "stack" of a number of cells, or a larger


'--- 1


rPI "(from air)

______________ \


Anod t. Membrane-
GDL Catalyst

Figure 1. Operational schematic of PEMFC.

Gik Gaset


Graphbe flow-chane block ad plte Bolt olso

Figure 2. PEMFC component schematic.13j

cell, would also be interesting, but would be cost prohibitive
in capital and operating costs for most undergraduate labs.
The PEM fuel cell design consists of the following major
components: a polymer electrolyte membrane, a gas diffu-
sion electrode, a gasket, graphite plates, current collectors,
end caps, and bolts. A schematic is shown in Figure 2.
The polymer electrolyte membrane is the most critical com-
ponent of a PEM fuel cell. It conducts hydrogen protons from
the anode to the cathode and acts as an electrical insulator to
ensure that electrons travel through an external circuit. In the
test set-up, an integrated membrane electrode assembly
(MEA) with a perfluorosulfonate ionomer was selected.
The catalyst layer was composed of a carbon powder
(Vulcan XC-72, Acetylene Black), precious metal (Pt, Pt-
Ru), and Nafion mixture.
Since PEM fuel cells operate at low temperatures, a cata-
lyst is required to achieve sufficient reaction rates for hydro-
gen oxidation at the anode and oxygen reduction at the cath-
ode. The gas-diffusion layer is composed of porous carbon
paper or carbon cloth that is treated with polytetra-
fluoroethylene. The GDL ensures that reactant gases are dis-
tributed evenly and effectively over the catalyst layer, pro-
vides effective liquid water removal at the cathode, pro-
vides for uniform load distribution on the electrolyte, and
provides for electrical conduction between the graphite
plates and the catalyst layer.
The gasket is composed of Teflon sheeting. It is positioned
around the membrane electrode assembly to ensure complete
sealing upon compression. The seal prevents reactant gases
from escaping into the environment or crossing over from
the anode to the cathode. Gas-flow channels are machined
into the plates to provide an inlet and outlet point for reactant
gases. The depth and width at which each channel is ma-
chined can affect gas distribution over the MEA and wa-
ter removal at the cathode. The plates also serve as elec-
trical conductors.

Copper current collectors with ports for electrical connec-
tions are used. Aluminum end caps and bolts are used to form
mechanical compression and connection of the fuel-cell com-
ponents. This is essential for complete sealing
Sand the reduction of contact resistances.

In designing the test station, our objective
was to allow easy monitoring and manipula-
tion of the appropriate variables so calculations
associated with different electrochemical prin-
cipals could be conducted. The station provides
an environment in which a PEM fuel cell can
operate. It will also enable the testing of fu-
ture undergraduate design projects that involve
the construction of a fuel cell. The test-station
environment consists of the following systems:

Summer 2004

Current collector

Gas diffulin layer

gas-flow control, water management, electrical, data ac-
quisition, and safety. A schematic of the test station is
shown in Figure 3.
The gas-flow system comprises two mass-flow meters and
two mass-flow controllers. The role of these components is
to monitor and regulate reactant gas flow, control systems
pressure and to aid in fuel consumption calculations. The wa-
ter-management system comprises two hydrators, a heated
water bath, and two knock-out drums. The role of the hydra-
tors is to add moisture to the reactant gases prior to entering
the PEM fuel cell because hydration of the membrane is criti-
cal to optimal performance.
Double-pipe Nafion hydrators are used because of their
transfer efficiency over a small surface area and volume. The
role of the heated water bath is to control hydration levels
and to heat and circulate water to the hydrators. The role of
the knock-out drums is to condense the moisture in the exit
streams as the water may damage the mass-flow controllers.
The electrical system comprises a switch, a shunt, and a
resistor. The switch is used to open the circuit, while the shunt
allows accurate measurement of high current and the resistor
enables the application of variable resistance. There is a cur-
rent undergraduate project that involves designing a variable
electrical load for the system.
The data acquisition system enables fuel-cell performance
data to be recorded. LabView software is used for real-time
monitoring and control of the following parameters: resis-
tance, pressure, reactant gas flow, temperatures, power mea-
surements, and reactant gas stoichiometric ratio. This soft-
ware allows maximum flexibility for future students to de-
sign experimental protocols. Figure 4 is a screen shot of the
LabView virtual instrument.
The safety system enables detection and removal of explo-
sive reactant gases (i.e., hydrogen) from
the test station. A hydrogen detector is
installed to monitor any leakage that
may occur during operation. An emer- Flow Mi
agency stop button is available on the
Lab View interface and upon activation,
the three-way solenoid valve switches Air
and purges the station with nitrogen.
Exhaust gases that exit the fuel cell are
routed into a fume hood.

Objective H2 3 Wi
The objective of the laboratory ex-
periment was to examine fuel cell tech-
nology and its associated electrochemi-
cal properties, as well as to examine the
effect of performance characteristics.

The PEM fuel-cell test station will be integrated into the ex-
perimental portion of a third-year chemical engineering
course, "Inorganic Process Principles 2," and as a demon-
stration experiment in "Chemical Environmental Engineer-
ing Concepts 2." Used as a demonstration, students will be
able to conduct basic mass balances, as well as humidifica-
tion calculations, on the gas streams. In the process principles
course, more advanced electrochemical concepts are covered,
including polarization and efficiency. Operation of the fuel
cell provides a practical application of polarization in an elec-
trochemical cell. In addition, fourth-year students from chemi-
cal, mechanical, or electrical engineering disciplines can use
the fuel cell and test station for related design projects.
Basic Fuel-Cell Thermodynamics
Fuel cells directly convert chemical energy into electricity
and are thus not bound by the Carnot law as in combustion
reactions. Heat engines cannot completely convert all the
available heat energy into mechanical energy due to some of
the heat being rejected.
The reversible energy efficiency of the electrochemical cell
can be found using
Trev = 1-T-- (4)
The direct conversion of chemical energy of a fuel and
oxidant into electrical energy can be described in terms of
electrical current output and cell potential. A maximum cell
potential, also known as the reversible cell potential, is
achieved when a fuel cell is operated under a thermodynami-
cally reversible condition. This can be calculated using

E A- (5)
The effect of temperature changes on cell potential can be

Chemical Engineering Education

calculated by
aE 1 (ifAG AS (6)
j-Tp nF JT nF

The effect of pressure changes on cell potential can be cal-
culated by
aE 1 AG AV (7)
= =T n(7)F
[P nF aP J nF

The Nernst equation for a hydrogen oxygen system is de-
fined by

E = E 2FIn P + 2, F nFPo2 (8)
2F) / PH20 F

As hydrogen and oxygen are consumed, the partial pres-
sures of each decrease respectively. This results in a decrease
of overall potential, as stated by Eq. (6). In order to reduce
these losses, excess stoichiometric amounts of reactant gases
are supplied. The excess also addresses water management
issues. The stoichiometric ratio is

St = lNi= NNin (9)
N consumed = in Nout

Figure 4. LabView screen shot of the fuel-cell test station
virtual instrument.
Theoretical EMF or Ideal Volta2
I ------- ---------I---^ | --- | --- 20
1.2 Higher Efficiency
Larer Cell Power
R egin of Activtion Polari ton 15
actionn Rate Los) /

R ^.o/ .... ... L
/ Reglio of Ohom lc Pol arliztloo
4 / (Resstfnce Loss)
/ Regio of Co..ncentr.ation
0./2 Polerleilo
/ (Mass Transport LoSt)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Current Density Amp cih
Figure 5. Typical polarization curve and power curve.

Typically for a PEM fuel cell, S for hydrogen is 1.1 to 1.2,
and St for oxygen is 2. Fuel utilization is the inverse of the
stoichiometric ratio

1 N. N
U in ou (10)
St Nin

Electrode Kinetics

In the practical application of fuel cells, the reversible cell
potential can never be achieved due to irreversible activa-
tion, ohmic, and diffusion overpotentials. The activation
overpotential is the loss associated with the activation barrier
that the reactant species must overcome. The ohmic
overpotential is the loss associated with the resistance of cell
components. The increased mass transfer resistance of reac-
tant gases through the electrode structure at high current den-
sities causes diffusion overpotentials. Figure 5 is a typical
polarization curve showing the effect of the losses on the re-
versible cell potential.
The following relations describe the activation

-2.3RT 2.3RT
nact =a+blogi (11) a= logio (12) b= (13)
anF anF
Ohmic overpotentials are described by

rlohmic = iROhm (14)
where R is the resistance (ohm).
The concentration overpotential is described by

RT i'"[ d o5)
con= Tn 1-- (15)
c nF i-

The objective of this experiment is to examine the effect of
performance characteristics, fuel-cell technology, and the
associated electrochemical losses. The associated class cov-
ers electrochemical concepts such as activation overpotential,
ohmic losses or overpotential, and concentration
overpotential, with the objective of reinforcing these concepts
with the student while providing an introduction to the tech-
nology. This lab also provides a basic engineering design ex-
ercise as well as reviewing physical chemistry calculations.
The undergraduate experiment involves development of a
polarization curve (voltage vs. current density) at various tem-
peratures (kept low so as to not stress the cell, 20-700C), vari-
ous cathode and anode pressures (once again, kept low to
avoid stressing the membrane, 5-15 psig). The apparatus also
allows the students to explore operation at various stoichio-
metric ratios for hydrogen and oxygen. Changes in perfor-
mance against a baseline polarization curve are examined and
analyzed. Actual student data of the effect of variations in
temperature and pressure can be seen in Figures 6 and 7, re-
spectively. With advancement in technology and as "com-

Summer 2004


mercial" single cells and membranes become more widely
available, better performance of the cell can be expected, and
students can discuss how their performance compares to pub-
lished results. Post-experimental tasks include: examination
of the overpotential regions that the polarization points are
within; determination of the maximum power output; calcu-
lation of voltage and overall efficiencies; stack sizing for a
75-kW and 42-volt system; and hydrogen storage tank sizing
for a vehicle running at peak power (75 kW) for two hours.
Students are then asked to take the collected data and design
a cell for a 1-kW application (e.g., simple calculation of the
number of cells, stacks, and active area of the cell). These last
two calculations provide simple "engineering design tasks" that
are a critical component to educational programs.

The addition of a PEM fuel-cell test station and laboratory
experiment into an undergraduate chemical engineering cur-
riculum can have the following benefits:
Students are exposed to significant new technologies that are
being developed and produced. By providing laboratory
experiments that are relevant to operational technology, we
are able to teach pertinent concepts and skills. Students also
use software and research tools that are widely used in
Students are introduced to the emerging hydrogen economy
and the use of hydrogen as an energy vector


....4 ..

Figure 6. Effect of pressure on performance
(student-collected data).

1 ,-,-, 8
0.9 7
0.6 5
S0.5 4
0.4 -3
: 0.3 2In...ul T. mpl. ture. 2
0 .1 .-.--.......... ... -. ....... 1
0 0
0 0.1 0.2 0.3 0.4
Current Density (Amp cit)
Figure 7. Effect of temperature on performance
(student-collected data).

Students gain a working knowledge of PEMfuel-cell
operation, optimization, and equipment sizing.
Students gain a better understanding of electrochemical
properties and the effect of performance characteristics on a
fuel cell.
Students are able to examine the materials and functionality
of PEM fuel-cell components.
Students gain an understanding of the various systems that
enable a fuel cell to operate.
The test station can facilitate future projects relating to PEM
fuel cells at both the graduate and undergraduate levels.
Co-operative education and graduate employment are
assisted by providing students with education and experience
in a growing industry sector.
Should the reader wish to construct a similar lab, more de-
tailed information and specifications can be obtained from
the authors.

AG change in Gibbs free energy
AH change in enthalpy
AS change in entropy
AV change in volume
a electron transfer coefficient
E reversible cell potential
F Faradays Constant (96,487 Coulomb/mol electron)
i actual current density (A)
iL limiting current density
io exchange current density
n mols of electrons/mol of fuel
Nin molar flow rate of reactants supplied


molar flow rate of reactants exiting.
hydrogen partial pressure
water partial pressure
oxygen partial pressure
universal gas constant
resistance (ohm)
temperature (K)

Funding for construction of the PEM fuel-cell test station
was provided in part by the Waterloo Engineering Endow-
ment Fund and the Department of Chemical Engineering at
the University of Waterloo. We also acknowledge the remain-
ing members of the group that designed and built the station:
Erik Wilhelm, Rob McArthur, Dave Chen, Dara Jahani, Jon
Shui, and Sumit Kundu.

1. Sharon, T., and M. Zalbowitz, Fuel Cells: Green Power Los Alamos
National Laboratory (1999) GreenPower.pdf>
2. EG&G Services Parsons, Inc., Science Applications International Cor-
poration, Fuel Cell Handbook, 5th ed., U.S. Department of Energy
(2000) 1

Chemical Engineering Education


This guide is offered to aid authors in preparing manuscripts for Chemical Engineering Education (CEE),
a quarterly journal published by the Chemical Engineering Division of the American Society for Engineering
Education (ASEE).
CEE publishes papers in the broad field of chemical engineering education. Papers generally describe a
course, a laboratory, a ChE curriculum, research program, machine computation, special instructional programs,
or give views and opinions on various topics of interest to the profession. (Note: Articles for the special series on
outstanding ChE departments and ChE educators are invited articles.)

Specific suggestions on preparing papers *

TITLE Use specific and informative titles. They should be as brief as possible, consistent with the need for
defining the subject area covered by the paper.

AUTHORSHIP Be consistent in authorship designation. Use first name, second initial, and surname. Give
complete mailing address of place where work was conducted. If current address is different, include it in a footnote
on title page.

ABSTRACT: KEY WORDS Include an abstract of less than seventy-five words and a list (5 or less) of keywords

TEXT We request that manuscripts not exceed twelve double-spaced typewritten pages in length. Longer
manuscripts may be returned to the authors) for revision/shortening before being reviewed. Assume your reader
is not a novice in the field. Include only as much history as is needed to provide background for the particular
material covered in your paper. Sectionalize the article and insert brief appropriate headings.

TABLES Avoid tables and graphs that involve duplication or superfluous data. If you can use a graph, do not
include a table. If the reader needs the table, omit the graph. Substitute a few typical results for lengthy tables when
practical. Avoid computer printouts.

NOMENCLATURE Follow nomenclature style of Chemical Abstracts; avoid trivial names. If trade names are
used, define at point of first use. Trade names should carry an initial capital only, with no accompanying footnote.
Use consistent units of measurement and give dimensions for all terms. Write all equations and formulas clearly,
and number important equations consecutively.

ACKNOWLEDGMENT Include in acknowledgment only such credits as are essential.

LITERATURE CITED References should be numbered and listed on a separate sheet in the order occurring
in the text.

COPY REQUIREMENTS Submit the manuscript electronically as a pdf, Word, or tif file that includes all
graphical material as well as tables and diagrams. Send an additional copy of the manuscript on standard letter-
size paper through regular mail channels and include original drawings (or clear prints) of graphs and diagrams
on separate sheets of paper. Label ordinates and abscissas of graphs along the axes and outside the graph proper.
Figure captions and legends will be set in type and need not be lettered on the drawings. Number all illustrations
consecutively. Supply all captions and legends typed on a separate page. State in cover letter if drawings or
photographs are to be returned. Authors should also include brief biographical sketches and recent photographs
(preferably head shots) with the manuscript.

Send your electronic manuscript to
and your hard copy to
Chemical Engineering Education, c/o Chemical Engineering Department
University of Florida, Gainesville, FL 32611-6005


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 EFL4LLLOT_02EA46 INGEST_TIME 2012-03-09T00:30:53Z PACKAGE AA00000383_00159