Chemical Engineering Education

http://cee.che.ufl.edu/ ( Journal Site )
MISSING IMAGE

Material Information

Title:
Chemical Engineering Education
Alternate Title:
CEE
Abbreviated Title:
Chem. eng. educ.
Physical Description:
v. : ill. ; 22-28 cm.
Language:
English
Creator:
American Society for Engineering Education -- Chemical Engineering Division
Publisher:
Chemical Engineering Division, American Society for Engineering Education
Place of Publication:
Storrs, Conn
Publication Date:
Frequency:
quarterly[1962-]
annual[ former 1960-1961]
quarterly
regular

Subjects

Subjects / Keywords:
Chemical engineering -- Study and teaching -- Periodicals   ( lcsh )
Genre:
serial   ( sobekcm )
periodical   ( marcgt )

Notes

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

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 01151209
lccn - 70013732
issn - 0009-2479
Classification:
lcc - TP165 .C18
ddc - 660/.2/071
System ID:
AA00000383:00201

Full Text






















Feature Articles^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^















BUSINESS ADDRESS:
Chemical Engineering Education
5200 NW 43rd St., Suite 102-239
Gainesville, FL 32606
PHONE: 352-682-2622
FAX: 866-CEE-0576
e-mail: cee@che.ufl.edu

EDITOR
TimAnderson

ASSOCIATE EDITOR
Phillip C. Wankat


MANAGING EDITOR
Lynn Heasley

PROBLEM EDITOR
Daina Briedis, Michigan State

LEARNING IN INDUSTRY EDITOR
William J. Koros, Georgia Institute of Technology

-PUBLICATIONS BOARD--

CHAIR .
C. Stewart Slater
Rowan University
VICE CHAIR*
Jennifer Sinclair Curtis
University of Florida
MEMBERS
Pedro Arce
Tennessee Tech University
Lisa Bullard
North Carolina State
David DiBiasio
Worcester Polytechnic Institute
Stephanie Farrell
Rowan University
Richard Felder
North Carolina State
Tamara Floyd-Smith
Tuskegee University
Jim Henry
University of Tennessee, Chattanooga
Jason Keith
Mississippi State University
Milo Koretsky
Oregon State University
Suzanne Kresta
University of Alberta
Marcel Liauw
Aachen Technical University
David Silverstein
University of Kentucky
Margot Vigeant
Bucknell University
Donald Visco
University of Akron


CHEMICAL ENGINEERING EDUCATIONIISSN 0009-2479 (print); ISSN 2165-6428 (online)] 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, 5200 NW
43rd St., Suite 102-239, Gainesville, FL 32606. Copyright 2013 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 90 days of publication. Write for information on subscription costs and for back copy costs and availability.
POSTMASTER: Send address changes to Chemical Engineering Education, 5200 NW43rd St., Suite 102-239, Gainesvilie,
FL 32606. Periodicals Postage Paidat Gainesville, Florida, and additional post offices (USPS 101900). www.che.ufl.edulCEE


Vol. 47, No. 3, Summer 2013


Chemical Engineering Education
Volume 47 Number 3 Summer 2013




o DEPARTMENT
138 Chemical Engineering at Iowa State University
Chris Neary, Surya Mallapragada, and George Burnet

, CLASS AND HOME PROBLEMS
170 Continuous Feed and Bleed Ultrafiltration-a Demonstration of the
Advantages of the Modular Approach for Modeling Multi-Stage
Processes
Michael B. Cutlip and Mordechai Shacham

SURVEY
145 Chemical Engineering Students: A Distinct Group Among Engineers
Allison Godwin and Geoff Potvin

i CLASSROOM
154 Active Learning and Just-In-Time Teaching In a Material and
Energy Balances Course
Matthew W. Liberatore

> LABORATORY
161 Comparison Between Linear and Nonlinear Regression In a
Laboratory Heat Transfer Experiment
Carine Messias Goncalves, Marcio Schwaab, and
Jose Carlos Pinto

179 Remote Labs and Game-Based Learning for Process Control
Imran A. Zualkernan, Ghaleb A. Husseini, Kevin F. Loughlin,
Jamshaid G. Mohebzada, and Moataz El Ganml

m RANDOM THOUGHTS
178 Speaking of Everything-III
Richard M. Felder

P OTHER CONTENTS
177 In Memoriam: Don Woods










[M] department



Chemical Engineering at...


Iowa State University


CHRIS NEARLY,
SURYA MALLAPRAGADA
AND GEORGE BURNET
n many ways, Iowa State
chemical engineering
epitomizes the land-grant
philosophy its university lives
by-Iowa State University
was the nation's first land-
grant university. It is the
birthplace of the first digital
computer and is one of the
few universities to host a De-
partment of Energy national
laboratory, Ames Laboratory,
on its campus. Iowa State
ranks second among univer-
sities in R&D 100 awards,
given by R&D magazine for
top technologies. The Col-
lege of Engineering is the
10th largest in the nation. It


Sweeney Hall is home to the Department of Chemical and Biological Engineering
at Iowa State University.


is home to many interdisci-
plinary research centers of excellence including a National
Science Foundation Engineering Research Center (NSF-ERC)
focused on Biorenewable Chemicals. This research center,
commonly referred to as CBiRC, is led by Iowa State chemi-
cal engineering faculty members.
This year, the Department of Chemical and Biological
Engineering at Iowa State University, currently the ninth
largest in the nation, celebrates its centennial-2013 marks
100 years of education and research excellence in chemical
and biological engineering at Iowa State University. This
is a key milestone for the department as it looks forward to

Some parts of this text are summarized from "100 Years of Chemical
and Biological Engineering at Iowa State University," by George
Burnet and Steve Sullivan, set to be published in summer 2013. All
copyrights reserved.


continuing the long tradition of excellence and building on
its successes for the next 100 years.

CENTENNIAL HISTORY*
Chemical engineering courses at Iowa State were originally
offered in the Department of Mining Engineering, Ceramics,
and Chemical Engineering in 1913. Dr. O.R. Sweeney led
the Iowa State chemical engineering program and began pio-
neering work on utilization of agricultural wastes. Sweeney
became "Iowa's Edison" as he served on national panels with
the likes of General Motors Chairman Alfred Sloan, FBI
Director J. Edgar Hoover, and Canadian Prime Minister
W.L. McKenzie. Today, Iowa is home to a wealth of agricul-
tural raw materials that would otherwise be wasted. With the

Copyright ChE Division ofASEE 2013


CELEBRATING 100 YEARS --

Chemical Engineering Education










field of biorenewables experiencing resurgence, it is worth
looking at our departmental history for some of the earliest
successes that have shaped this field. For example, research on
furfural from agricultural byproducts took place at Iowa State
from the 1920s to 1940s. In the book, Creation of the Modern
Land-Grant University: Chemical Engineering, Agricultural
By-Products and the Reconceptualization of Iowa State Col-
lege, Alan Marcus notes that "the destructive distillation of
corncobs liberated substantial quantities of furfurall). Ames
engineers not only refined furfural production techniques and
designed the blueprints for commercial facilities, but also set
about to determine furfural's potential industrial uses."
In 1936, the Engineers' Council for Professional Develop-
ment (ECPD, now ABET) first accredited the Iowa State
chemical engineering program. The American Institute of
Chemical Engineers (AIChE), a founding society of ECPD,
recognized the Iowa State University chemical engineering
program in 1925. Because Iowa State was one of the first
schools to be recognized by AIChE, many schools around the
country patterned their courses after Iowa State. In fact, the
Iowa State program was the first to offer chemical engineer-
ing plant design as a course. In 1926, Juanito Mina was the
first female graduate of the Iowa State chemical engineering
program. One year later, the department moved into its own
building complete with a large unit operations laboratory,


classrooms, offices, and research labs. The U.S. Department
of Agriculture built its By-Products Laboratory west of the
Chemical Engineering Building in 1935, further strengthen-
ing work on the utilization of agricultural wastes. In 1964,
Sweeney received the ultimate honor (six years after his
passing) when a second chemical engineering building was
constructed and named Sweeney Hall. The new three-floor
building added dozens of faculty and graduate student offices
and new spaces for research. By the time Sweeney stepped
down as department head in 1947, Iowa State's chemical
engineering program was one of the largest and most highly
respected in the United States.
In 1947, chemical engineering research was significantly
enhanced with the creation on campus of the Ames Labora-
tory, an interdisciplinary research laboratory of the Atomic
Energy Commission (now the U.S. Department of Energy).
The Ames Laboratory became a "clearinghouse for nuclear
research on campus, a public resource for atomic energy con-
sultation, a liaison between Iowa State, the Argonne National
Laboratory, and its associated 25 Midwestern universities,
a mecca for graduate students, and an administrative hub
for processing federal and private funds as they become
available," according to Joanna Abel Goodman's National
Science in the Nation's Heartland: The Ames Laboratory
and Iowa State University, 1942-1965. Through a chemical


Chemical engineering students build a "Products of Corn "float for 1937 VEISHEA, the country's longest student-run
festival. Items displayed on this float include syrup, starch, rayon, and wallboard-all made in Iowa State labs from corn.
This demonstrated various ways chemical engineering was commercialized in the 1930s.
Vol. 47, No. 3, Summer 2013 139









engineering division at the Ames Laboratory, chemical en-
gineering faculty and graduate students were very involved
in the development of processes for the recovery of thorium,
rare earth minerals, and uranium from monazite sands, all
high-priority needs at that time.
In 1957, the department name changed from the Depart-
ment of Chemical and Mining Engineering to simply the
Department of Chemical Engineering. In 1959, the Iowa State
College of Agriculture and Mechanic Arts was renamed Iowa
State University of Science and Technology.
From 1961 to 1978, George Burnet led Iowa State chemical
engineering with engineering education as the significant focus.
Bumet held many positions nationally including as president
of American Society for Engineering Education (ASEE);
national president of Omega Chi Epsilon; U.S. representative
to the United Nations Committee on Education and Training;
member of the NSF Commission on Precollege Education in
Mathematics, Science, and Technology; and many more.
Ray Fahien, a faculty member in the department, also
echoed Burnet's passion for engineering education. From
1967 to 1995, he served as the editor of Chemical Engineer-
ing Education. The journal created an award after him, which
honors an educator who has shown evidence of vision and
contribution to chemical engineering education.
In 1968, Iowa State chemical engineering's first female
doctoral candidate, Idelle Peterson, graduated.
Maurice Larson became department chair of chemical
engineering in 1978. Larson and his crystallization work put
Iowa State University on the world map. In a 1988 letter,
Professor John Garside from The University of Manches-
ter said, "(Iowa State's) work on crystallization has enabled
crystallizer design methods to be put on a rational, quantita-
tive basis.... Virtually all chemical engineering crystallization
research groups throughout the world now base their devel-
opments on these methods." Professor Richard Seagrave
became department chair after Larson stepped down in 1983.
Seagrave picked up on Larson's vision to expand Sweeney
Hall facilities to accommodate the new research areas such
as biotechnology. Larson's vision for expanded facilities, and
the continued work of Seagrave and Terry King, who became
chair in 1990, culminated in the dedication of the Sweeney
Hall addition in 1994. The $8 million addition expanded
Sweeney Hall by 35,000 square feet. In 2005, the department
changed from the Department of Chemical Engineering to
the Department of Chemical and Biological Engineering to
better reflect the bio-based research and teaching under way.
L.K. Doraiswamy joined Iowa State faculty as Glenn Mur-
phy Visiting Professor in Engineering in 1989. At Iowa State
he helped create a research program in catalysis and reaction
engineering. In 1998, Iowa State and the National Chemical
Laboratory in India established the L.K. Doraiswamy Honor


Lectureship, a dual-lecture series where a distinguished leader
in chemical engineering lectures at both Iowa State in Ames,
Iowa, and at NCL in Pune, India. Among the highest honors
Doraiswamy received was election to the National Academy
of Engineering in 2010. Doraiswamy passed away in 2012.
Professor Charles "Chuck" Glatz was chair of chemical
engineering from 1997 to 2005. During this time the depart-
ment pursued further interests in biological engineering,
and several faculty were hired, mainly at the associate and
full professor levels. These included Brent Shanks, Jackie
Shanks, Rodney Fox, Andrew Hillier, and Vlasta Klima
Balloun Professor Balaji Narasimhan. In 1999, former chair
Seagrave served as interim provost of Iowa State University,
and later as interim president of the university from 2000 to
2001. Seagrave stepped into the national curriculum spotlight
as chairman of ABET from 1996 to 1997, then as ABET
president from 2005 to 2006.
Professor James Hill became chair of the department in
July 2005. For the next four years, the Iowa State chemical
engineering program expanded research in biological and
biomedical areas. Six professors joined the faculty during this
time. In 2007, ConocoPhillips established an eight-year, $22.5
million biofuels research program to develop technologies that
produce biorenewable fuels, led by Professor Robert Brown.
In 2010, the biorenewables faculty and work found a
new home at the Biorenewables Research Laboratory, built
directly west of Sweeney Hall. The $32 million Phase I of
Iowa State's $107 million Biorenewables Complex serves as
a "front door" to the university's diverse and broad-reaching
programs in biorenewables.
In 2009, Professor and Stanley Chair in Interdisciplinary
Engineering Surya Mallapragada-an Iowa State faculty
member since 1996-was named the first female chair of the
Department of Chemical and Biological Engineering. Mal-
lapragada's research in biomaterials and bioinspired materials
has led to several awards including a TR100 award. Under
her leadership, the CBE department has attracted consecutive
record enrollment, consecutive record department research ex-
penditures, and record diversity within the student population.
To keep up with this rapid growth, the department renovated
20 laboratories in Sweeney Hall through a $1.75 million
competitive grant from the National Science Foundation to
grow research in biomedical engineering and nanomaterials.
Several teaching spaces have also been renovated through
funding from alumni as well as the Carver Charitable Trust.

FACULTY
The department currently has 20 tenure-track/tenured fac-
ulty members who are engaged in the research and educational
mission of the department. Included among the faculty are
four Iowa State B.S. chemical engineering graduates: Assis-
tant Professor Eric Cochran (B.S.ChE '98; PhD., University


Chemical Engineering Education


































of Minnesota), Assistant Professor Ian Schneider (B.S.ChE
'00; Ph.D., North Carolina State University), Mike and Jean
Steffenson Professor Brent Shanks (B.S.ChE '83; Ph.D.,
California Institute of Technology) and Manley Hoppe Pro-
fessor Jacqueline Shanks (B.S.ChE '83; Ph.D., California
Institute of Technology).
Current faculty of the Iowa State CBE department have
garnered more than 35 national honors and awards, including
NSF CAREER Awards (Cochran, Mallapragada, and Hillier),
NSF Young Investigator Awards (Rodney Fox and Jacqueline
Shanks), and several spots on national research committees
and "top" lists. Such lists include MIT Technology Review's
Top 100 Young Innovators (Mallapragada and Narasimhan)
and Biofuels Digest's Top 100 People in Bioenergy (Robert
Brown). The department has five professors with the Iowa
State College of Engineering's highest faculty honor, Anson
Marston Distinguished Professor: Brown, Fox, Peter Reilly,
Bumet (emeritus), and Seagrave (emeritus). Three faculty
members have the University Professor honor: Charles Glatz,
James Hill, and Thomas Wheelock (emeritus).
The first female chemical engineering faculty member at
Iowa State, Carole Heath, was hired in 1993. Today the Iowa
State chemical and biological engineering department has
one of the country's highest percentages of female tenured/
tenure-track faculty. Since Heath, the department has hired
eight tenured and three non-tenured female faculty members;
today, the proportion is 42 percent female.
Professor Derrick Rollins was hired in 1990 as Iowa State
chemical engineering's first African-American faculty mem-
ber. He founded several Iowa State programs for recruitment
and retention of underrepresented minority students. His pas-


A Prof. Derrick Rollins works with
students in the undergraduate
teaching labs.



sion for student success has been
recognized nationally by several
awards, including the American
Association for the Advancement
of Science Mentor Award and Tau
Beta Pi McDonald Mentor Award.
In August 2007, Narasimhan was
named associate dean for Research
and Economic Development in the
S College of Engineering. Among
many other successes in this role,
S Narasimhan has overseen devel-
opment of the interdisciplinary
Dean's Research Initiatives, which
are designed to lead to the creation
of large center-scale grants.
Iowa State biorenewables research received a major boost
in 2008 with the creation of the National Science Foundation
Engineering Center for Biorenewable Chemicals (CBiRC).
NSF contributed $18.5 million for the first five years of devel-
opment. Headed by Brent Shanks, CBiRC is Iowa State's first
NSF Engineering Research Center and serves to transform the
chemical industry into a renewable resource-based industry.
In 2012, CBiRC was awarded an additional $12 million,
three-year, NSF grant to continue research and educational
activities. Professors Jacqueline Shanks (who also serves as a
Thrust Leader in CBiRC), Laura Jarboe, Rollins, and Reilly
have contributed to the development of CBiRC research and
education. Two new departmental hires, Jean-Philippe Tes-
sonier and Zengyi Shao, are also associated with CBiRC.
Chemical engineering facilities at Iowa State continue
to meet the department's leading research and teaching
demands. The W.M. Keck Laboratory for High Throughput
Atom-Scale Analysis opened in Sweeney Hall in 2007. The
1,600-square-foot space houses leading combinatorial science
and atom-scale materials research. A state-of-the-art local
electrode atom probe microscope, along with complementary
instrumentation, provides the most advanced analysis to date
of compositional mapping of materials. Professors Hillier
and Narasimhan from CBE led the development of the Keck
Laboratory.

UNDERGRADUATE PROGRAM
Over the past 100 years, Iowa State's chemical engi-
neering curriculum has transformed along with the many
advancements and diversifications that chemical engineer-
ing industries have experienced. Today students can take


Vol. 47, No. 3, Summer 2013









Prof. Jacqueline Shanks,
the microbial metabolic
engineering thrust leader
for CBiRC, is pictured here
with current students and
research staff.






either the general chemical
engineering track or biological
engineering track. As juniors,
undergraduates enroll in a CHE
325 (chemical engineering labo-
ratory I) / ENGL 314 (technical
communication) hybrid course
that teaches students both how
to operate a laboratory and
how to develop communica-
tion materials that accompany
chemical engineering practices. The program also provides
a freshmen learning community, which introduces students
to the chemical engineering profession, and provides career
planning and academic course support. Undergraduates learn
in a state-of-the-art two-story unit operations laboratory,
named the Herbert L. Stiles Teaching Laboratory after 1929
alumnus Herbert Stiles.
Enriching undergraduate experiences extend beyond the
Ames, Iowa, campus. In summer 2001, Reilly helped start
the International Summer Course in Chemical Engineering,
held every summer at the University of Oviedo, Spain. Select
undergraduate students participate in an intense, five-week unit
operations laboratory course. Iowa State partners in the course
with the University of Wisconsin-Madison and the University
of Oviedo. Professor Emeritus Ken Jolls has coordinated the
program since 2002. Students also take advantage of summer,
semester, and mini-term internships and co-op experiences at
companies in Iowa and throughout the United States. Since its
beginning, the Iowa State chemical engineering program has
awarded more than 4,700 baccalaureate degrees. More than 250
companies attract interns at the Engineering Career Fair every
semester at Iowa State University-site of one of the country's
largest collegiate career fairs in engineering-leading to very
high placement rates (90%+) for graduates of the college.
In 2006, Iowa State first received NSF funding for the
Biological Materials and Processes Research Experience
for Undergraduates (BioMaP REU). Every summer since, a
dozen or so undergraduate students from around the United
States have been matched with Iowa State CBE faculty to
conduct 11 weeks of research and present their work at a
public symposium on campus.

142


In addition to coursework and research, Iowa State chemi-
cal engineering faculty members have contributed to major
literature in the field. Doraiswamy published six top chemical
reaction engineering books: Chemical Reaction Engineer-
ing: Beyond the Fundamentals (2013); Organic Synthesis
Engineering (2001); Catalytic Reactions and Reactors
(1991); Analysis of Chemically Reacting Systems: A Sto-
chastic Approach (1987); Across Millenia: Some Thoughts
on Ancient and Contemporary Science and Engineering
(1987); and Heterogeneous Reactions: Analysis, Examples
and Reactor Design (1984). Brown has published three
books contributing to biofuels and biorenewable resources:
Why Are We Producing Biofuels? (2012); Thermochemical
Processing of Biomass: Conversion into Fuels, Chemicals,
and Power (2011); and Biorenewable Resources: Engineer-
ing New Products from Agriculture (2003). Fox published
three books on computational fluid dynamics: Computa-
tional Models for Polydisperse Particulate and Multiphase
Systems (2013); Multiphase Reacting Flows: Modeling and
Simulation (2007); and Computational Models for Turbulent
Reacting Flows (2003). Mallapragada and Narasimhan have
co-published three books: Combinatorial Materials Science
(2007); Handbook of Biodegradable Polymers and Their
Applications (2006); and Biomaterials for Drug Delivery
and Tissue Engineering (2001). These and other current Iowa
State chemical engineering faculty members have served on
many journal editorial boards: AIChE Journal, Industrial &
Engineering Chemistry Research, International Journal of
Multiphase Flow, Annual Review of Fluid Mechanics, The
Electrochemical Society's INTERFACE, Fluid Dynamics
Research, Journal of Nanoparticle Research, ISRN Nano-
technology, Biotechnology Progress, Metabolic Engineering,
Chemical Engineering Education








FACULTY GALLERY (In alphabetical order with Ph.D. institution and research area)


Kaitlin Bratlie, Uni-
versity of California-
Berkeley. Biomateri-
als, tissue engineer-
ing, imaging


Rebecca Cademartiri,
University of Pots-
dam. Interactions of
biological entities
with materials
Eric Cochran,
University of
Minnesota-Twin Cit-
ies. Self-assembled
polymers


Rodney Fox, Kansas
State University.
Computational fluid
dynamics, reaction
engineering
Charles Glatz, Uni-
versity of Wisconsin-
Madison. Bioprocess-
ing, bioseparations


-Kurt Hebert, Uni-
versity of Illinois at
Urbana-Champaign.
Corrosion, electro-
chemical engineering
Jennifer Heinen,
University of
Delaware. Mecha-

nism and kinetics of
controlled polymer-
izations in heteroge-
neous media
James Hill, University
of Washington. Turbu-
lence, computational
fluid dynamics


Andrew Hillier, Uni-r
versity of Minnesoa
Interfacial engine, i -
ing, electrochemi.,ti i
Vol. 47, No. 3, Summer 2013


Laura Jarboe, Uni-
versity of California
Los Angeles. Bio-
renewables produc-
tion by metabolic
engineering


Monica Lamm,
North Carolina State
University. Mo-
lecular simulation of
advanced materials


Stephanie Loveland,
Iowa State University.
Senior lecturer


kSurya Mallapragada,
Purdue University.
(Department Chair)
Tissue engineering,
gene delivery
Balaji Narasimhan,
Purdue University.
associate dean for
research for College
of Engineering; Bio-
materials, drug and
vaccine delivery
Peter Reilly, Uni-
versity of Pennsyl-
vania. Molecular
mechanics, molecular
dynamics, quantum
mechanics


Derrick Rollins, The
Ohio State Uni-
versity. Statistical
process control


Ian Schneider, North
Carolina State Univer-
sity. Cell migration,
mechanotransduction


Breni Shanks., Cali-
tirnia Inttiule ot
Tc'hnolog.. Hi itc.
hi / It'lft l. ll/tll\ e ,
J71, Jl t Ilt'tt'tlJ'/t; 3


Jacqueline
Shanks,
California
Institute
of Tech-
nology.
Metabolic
engineering, plant biotechnology
Zengyi
Shao,
University
of Illinois
at Urbana-
Champaign.
Biorenew-
ables production by metabolic
engineering


Jean-
Philippe
Tessonnier,
University
de Stras-
bourg. Het-
erogeneous
catalysis, biorenewables
R. Dennis
Vigil, Uni-
versity of

l. Michigan.
Transport
phenom-
Ill ena, reac-
tion engineering in multiphase
systems
Qun Wang,
University
of Kansas.
Drug deliv-
ery, nano-
It'Chith [o' Q1,
bt.l"h1M t'l I "










Current Opinion in Biotechnology, Electrochemical Society,
Biotechnology Letters, Starch, ASME Journal of Nanotech-
nology in Engineering and Medicine, and ASME Journal of
Fluids Engineering.
Several alumni from the department have had notable and
distinguished careers and received national/international
recognition. Iowa State University B.S. graduates such as
Allen Jacobson (retired CEO of 3M), James Katzer (retired
from Exxon), Paul Willhite (University of Kansas), Jerry
Schnoor (University of Iowa), and Lanny Robbins (retired
from Dow Chemical) have been elected to the National
Academy of Engineering. Several alumni with undergraduate
degrees from Iowa State University have established success-
ful careers in academia: Alumnus Tim Anderson is currently
dean of Engineering at the University of Massachusetts; other
successful alumni in academia include Mark Saltzman at
Yale and Edward McGinn at the University of Notre Dame.
Distinguished recognition have come to recent graduates
as well. For instance, in 2010, chemical engineering junior
Meredith Gibson was a guest speaker (the only college
student speaker) at the Fortune magazine Most Powerful
Women summit, held in Washington, D.C. There she shared
her Iowa State engineering experiences and involvement
with the National Math and Science Young Leaders Program.
Gibson graduated with a Bachelor's in chemical engineering
in December 2012.
On campus, undergraduate students are very active in
chapters ofAIChE, Omega Chi Epsilon (chemical engineering
honor society), and the National Organization for the Profes-
sional Advancement of Black Chemists and Chemical Engi-
neers. In 2010, the AIChE Iowa State student chapter hosted
the Mid-America AIChE Student Regional Conference.

GRADUATE PROGRAM
Iowa State offers the Doctor of Philosophy (Ph.D.), Master
of Science (M.S.) and Master of Engineering (M.Engr.) de-
grees in chemical engineering. While Ph.D. and M.S. require
a thesis, M.Engr. requires only coursework. M.S. students take
an average of two years to graduate, while Ph.D. candidates
complete their degrees in 4.5 years, on average. The Iowa
State Chemical Engineering Graduate Student Organization
(CEGSO) provides a venue for all Iowa State ChE graduate
students for special professional development opportunities
as well as for volunteering and social activities.
Graduate students become heavily involved in research
endeavors in chemical engineering concepts and technol-


ogy used in today's and tomorrow's energy, sustainability,
and health industries. Research areas include advanced and
nanostructured materials, biorenewables, catalysis and reac-
tion engineering, computational fluid dynamics, health care
technology and biomedical engineering, and renewable en-
ergy. Biorenewables research is quite popular given Iowa's
abundance of biomass as a biorenewables resource. Doc-
toral graduate Catie Brewer (2012) received the 2011 George
Washington Carver Award Scholarship Prize for Outstanding
Student Achievement in Biorenewables at the World Congress
on Industrial Biotechnology & Bioprocessing. Substantial
funding has recently been attributed to health care technology
and biomedical engineering, particularly in nanovaccine gene
delivery and neuroregenerative strategies. Research has been
published in high-impact journals such as Nature Materials
and Nature. Since the Iowa State program was founded, 624
Master's degrees and 450 doctoral degrees have been awarded.
Graduate alumni have established successful careers.
Recent alumni in academia include Ganesh Sriram at the
University of Maryland; Matthew Kipper at Colorado State
University; Russell Gorga at North Carolina State Univer-
sity; Erin Jablonski and Brandon Vogel at Bucknell; and
Venkat Raman at University of Texas at Austin, among
others. Umit Ozkan, now a distinguished professor at The
Ohio State University, has received many national honors
and awards for her teaching and research in heterogeneous
catalysis. Deniz Uner is the professor and chair of chemical
engineering at Middle East Technical University in Ankara,
Turkey, and recently co-authored a book with the late Do-
raiswamy called Chemical Reaction Engineering: Beyond
the Fundamentals (2013). Our alumni advance their Iowa
State chemical engineering research through such academic,
industrial, and entrepreneurial endeavors.

SUMMARY
The first century of Iowa State chemical engineering re-
search and education excellence culminates in 2013. The
department is proud of the academic, research, and profes-
sional networks it identifies with, and will continue to serve
as a premier resource for chemical engineering teaching and
development. Teaching and research facilities will expand
to meet the demands of chemical engineering excellence.
Faculty and students will continue to push the frontiers of
interdisciplinary research to improve the theory and practice
of chemical engineering. In 2013 Iowa State chemical and
biological engineering starts an exciting new chapter-its
second century. O


Chemical Engineering Education











[MM] survey


CHEMICAL ENGINEERING STUDENTS:


A DISTINCT GROUP AMONG ENGINEERS








ALLISON GODWIN, GEOFF POTVIN
Clemson University Clemson, S.C. 29634


n traditional analyses of students' career choice, students
in engineering majors are often treated as a monolithic
population rather than belonging to a constellation of relat-
ed disciplines. A few studies have documented different types
of students and cultures within engineering disciplines,"1-3'
but there is little discussion about how students' attitudes
and perceptions of engineering disciplines affect their choice
of major upon entrance into college and the characterization
of these students. Studies that have been published have
compared personality types through tests like Myers-Briggs,
which have validity problems, especially due to the lack of
relevant context.'4] Additionally, much of this work was con-
ducted more than 30 years ago, and there is little current work
on articulating students and cultures of different engineering
disciplines. One study that does illustrate the differences
between engineering groups examined the different attitudes
between industrial engineering students and more traditional
mechanical and electrical students at a single institution.'51
The conclusions from this study provide motivation to further
explore this under-researched area: "Instead of examining the
characteristics of persons gravitating toward engineering, we
should inquire into what types of persons select which types
of engineering."''5 Another study examined the differences
between chemical engineers and other engineering students,
science students, and non-science students. The data from
this study were limited to transcript information from the
Southeastern University and College Coalition for Engineer-
ing Education (SUCCEED). The authors did find differences
in chemical engineering students with higher SAT Math
and Verbal scores, higher high school GPAs, longer time to
graduation, higher cumulative college GPAs, fewer changes in
Vol. 47, No. 3, Summer 2013


declared major, and more semester hours than other students."6]
Because of its design, however, this study did not explore
students' career interests and attitudes, which is the focus of
our study. Further study of students' interests and attitudes
about engineering disciplines is vital to the recruitment and
retention of engineering students.
It has been shown that students will develop a strong attach-
ment to their chosen major when the perceived identity of a
practitioner agrees with a student's self-defined identity.[7-12]
Additionally, students who are more familiar with specific
engineering disciplines express a greater confidence in their
choice of major.031 These findings have important implications
for how students are recruited into particular programs as well

Allison Godwin is a Ph.D. student and NSF
graduate research fellow in engineering and
science education at Clemson University.
She completed her B.S. in chemical and
biomolecular engineering at Clemson
University. Her thesis focuses on identify-
ing predictive factors for increasing female
enrollment in engineering.
Geoff Potvin
has been an as-
sistant professor
in the Department of Engineering & Science
Education at Clemson University since 2008.
He has a background in theoretical physics
and STEM education and teaches under-
graduate mathematics and physics courses,
as well as graduate courses in STEM edu-
cation research. His research is focused on
sociocultural issues in the preparation of
physical scientists and representation issues
in the physical sciences.
Copyright ChE Division of ASEE 2013










as how these students are instructed. With little literature on
the differences between students who choose the spectrum
of engineering disciplines, there is a missed opportunity to
improve the recruitment, retention, and teaching practices for
the students who enter chemical engineering classrooms, as
well as other engineering majors.
Since high school students have not yet been fully exposed
to science practice, let alone engineering practice, the choice
of an engineering discipline upon entrance to college is only a
partly informed decision. Many students enter college know-
ing they want to pursue engineering with particular career
outcomes in mind, but they do not know which engineering
discipline fits those career aspirations. Additionally, the
coursework to prepare for a specific STEM career is often
undifferentiated in high school. Some studies have attempted
to classify different types of engineers by their career roles of
research, development, production, and sales through using
students' Strong Vocational Interest Blank (SVIB) scores. [14]
These roles in industry are outdated and do not directly
address the question of student attitudes and interest upon
entrance into college, but these findings do add to the body
of evidence that there are differences in engineers' specific
interests and, more importantly, a specialization within the
designation of a specific engineering career choice (e.g.,
chemical, mechanical, electrical, etc.). Students just entering
college may not be prepared to make a specialty choice in en-
gineering. Instead, students choose an engineering discipline
based on the perceived fit with their intentions and several
irrational factors .13] These findings add to the motivation to
explore the underlying differences in students who choose
different engineering disciplines.
Performance in math and science are not the primary reason
that students either leave engineering studies or do not enter
them in the first place.115'161 Students reported that a loss of
interest in their original major, pedagogical and curricular
issues, disenchantment with perceived future careers, inad-
equate advising, lost confidence due to low grades or poor
preparation, and-for females-covert or overt gender bias
within the discipline, caused them to leave their originally
declared major. In general, gaps between students' expecta-
tions and the perceived "fit" of a major result in students
leaving. In Talking About Leaving, 10.5 percent of students
initially declaring an engineering discipline as a major re-
settled within engineering, while 51.4 percent remained in
the originally declared major, with the remaining group of
students (38.1 percent), leaving engineering altogether.051 A
recent study of students switching from other engineering
majors into industrial engineering found that the same pushes
highlighted by Seymour and Hewitt continue to affect current
engineering students .[16) From transcript databases, research
has shown that engineering as a group does have one of the
highest rates of persistence in STEM and the lowest rate of
inward migration. Engineering students are also more likely


to graduate in their declared major.]7' Additionally, testimony
before the Subcommittee on Research of the Committee
on Science for the U.S. House of Representatives in 2006
indicated that little, if anything, has changed since Seymour
and Hewitt's findings almost 10 years prior. Current numbers
show that 50 to 60 percent of students initially declaring a
major in STEM eventually leave those studies. In view of the
current situation in STEM attrition, the President's Council
of Advisors in Science and Technology (PCAST) recently
called for 1 million new STEM graduates over the next 10
years.118 One way to address the need for more STEM gradu-
ates is through understanding which types of students choose
engineering and how to more effectively recruit them upon
entrance into college.
Understanding what factors (beliefs, attitudes, and goals)
lead students to choose specific engineering disciplines can
help address the need for new STEM graduates. By more
thoroughly understanding students in chemical engineer-
ing departments, chemical engineering educators can better
address their particular interests and needs. If, as expected,
these students are different from non-engineering students,
but are also different from their peers in other engineering
disciplines, departments will reap many benefits from an
improved understanding of their students.
In this paper, an exploration was conducted of pre-college
factors (including academic backgrounds, classroom ex-
periences, out-of-class experiences, attitudes, family influ-
ences, and demographic backgrounds) that impact students'
chemical engineering career intentions, as measured by their
self-identified likelihood of choosing a career in a specific
engineering discipline. The results illustrate the specific dif-
ferences in chemical engineering students identified in a
nationally representative sample of college freshmen, and
provide emphasis for the statement that "engineers should
not be lumped together into a single category."'141

METHODS
The data used in this study were drawn from a subsection
of the Sustainability and Gender in Engineering (SaGE)
survey (), a
large- scale study of students in introductory English courses
enrolled in colleges across the United States (NSF GSE
1036617). This methodology uses a cross-sectional approach
relying on the natural variation in students' experiences and
backgrounds across the United States. The SaGE project
used a representative, stratified, random sample taken from
a comprehensive list of four-year and two- year institutions.
A list of all colleges and universities in the United States was
obtained from the National Center for Education Statistics
(NCES) and was divided by institution type (two-year or
four-year) and by institution size (small, medium, or large)
into six lists. Each list was randomized and then recruiters
contacted schools on each list. The stratification accounted


Chemical Engineering Education













PC*-

Oregiin C
o r






*~~ ~ ~ Wkrmin


for the size of the institution and prevented over-sampling of
the smaller, but numerous, liberal arts colleges. In total, 50
schools agreed to participate in the survey. The survey was
administered in required freshman English courses to capture
a sample representative of both STEM and non-STEM ma-
jors. In all, 6,772 students completed the survey during the
administration period in the Fall of 2011. The survey instru-
ment focused on student backgrounds, pedagogical factors
in physical science classrooms, classroom achievement, and
student attitudes toward STEM and sustainability. Sustain-
ability is most commonly and broadly defined as meeting the
"needs of the present without compromising the ability of
future generations to meet their own needs."["] The intent of
the study was to focus on factors that increased enrollment in
engineering majors and to explore the connections between
engineering and sustainability-related topics in students'
experiences.
. Using this retrospective cohort methodology, substantial
natural variability in students' background and prior experi-
ences can be captured. Students reported that they came from
homes in at least 2,533 different ZIP codes across the United
States. A map of the engineering students' home ZIP codes in
the contiguous United States can be seen in Figure 1. This map
is included to illustrate the geographic representativeness of the
population which is reflective of the population of the United
States.1201 International students are included in the study as a
part of the cross- sectional sample gathered from the 50 institu-
tions surveyed. Of the total student population that completed
the demographic portion of the survey, 54.7% were female.
Of the 814 students who indicated the choice of any intended
engineering career, 19.8% of respondents were female.
The final version of the survey included 47 questions about
student career goals, high school science experiences, earlier
math, and science enrollment and achievement (including
types of courses taken, the level of courses, the year courses


were taken in high school, final grades, and AP test scores),
student attitudes about sustainability, and demographic infor-
mation. These questions consist of primarily Likert, Likert-
type, multiple-choice, and categorical items.
Multiple aspects of validity and reliability of the instrument
were assessed. An open-ended hypothesis-generation survey
was collected from 82 first-year engineering and 41 non-en-
gineering students, as well as 83 high school science teachers
(recruited via the listserv of the National Science Teachers'
Association). Lending to content validity, these hypotheses
were included in the survey. Questions were further refined
based on feedback from assessors and the results of pilot
testing in a first-year freshman engineering course. In-person
pilots of the survey and focus groups were conducted with
first-year freshmen engineering students. Thus, each item of
the survey was further examined for face and content validity.
One question used in this analysis asked students to "Please
rate the current likelihood of your choosing a career in the
following:". The various career options were "Mathematics,"
"Science/math teacher," "Environmental science," "Biol-
ogy," "Chemistry," "Physics," "Bioengineering," "Chemical
engineering," "Materials engineering," "Civil engineering,"
"Industrial/systems engineering," "Mechanical engineering,"
"Environmental engineering," and "Electrical/computer
engineering." Students were asked to rate the likelihood of
choosing a career in each discipline on a Likert-type scale
from 0 ("not at all likely") to 4 ("extremely likely"). In the
current analysis, students that responded as "extremely likely"
to choose a career in chemical engineering were grouped
together, and all other students that responded "extremely
likely" to choose at least one other engineering discipline were
grouped together for a comparative analysis. The reason for
this choice was to identify students with the most unambigu-
ous intentions of majoring in chemical engineering on the
one hand and all other engineering disciplines on the other.


Vol. 47, No. 3, Summer 2013


Figure 1.
Engineering
students'
hometowns in the
contiguous
United States
created with
BatchGeo./211


Gu, ol
Mexicoco


Heiana



























In biology asked questions, answered questions, or made comments (scale: 0-never;
4-daily)


3.22 0.08


2.85 0.30


In chemistry asked questions, answered questions, or made comments (scale: 3.39 0.23 2.74 0.43
0-never; 4-daily)
In physics asked questions, answered questions, or made comments (scale: 0-never; 3.19 0.04 2.84 0.33 *
4-daily)
Interest in understanding natural phenomena (scale: 0-not at all interested; 4-very 2.94 0.09 2.50 0.36 **
interested)
Interest in understanding science in everyday life (scale: 0-not at all interested; 3.11 0.12 2.67 0.34 ***
4-very interested)
Interest in explaining things with facts (scale: 0-not at all interested; 4-very inter- 3.24 0.08 2.88 0.29 ***
ested)
Interest in telling others about science concepts (scale: 0-not at all interested; 4-very 3.04 0.21 2.39 0.45 ***
interested)
Interest in making scientific observations (scale: 0-not at all interested; 4-very 3.04 0.12 2.55 0.38 ***
interested)
Confidence in designing an experiment to answer a scientific question (scale: 0-not 2.81 0.06 2.44 0.32 **
at all confident; 4-very confident)
Confidence in conducting an experiment on your own (scale: 0-not at all confident; 3.03 0.10 2.62 0.32 ***
4-very confident)
Confidence in interpreting experimental results (scale: 0-not at all confident; 4-very 2.99 0.10 2.59 0.32 ***
confident)
Confidence in writing a lab report/scientific paper (scale: 0-not at all confident; 2.90: 0.18 2.30 0.43 ***
4-very confident)
Confidence in applying science knowledge to an assignment or test (scale: 0-not at 3.04 0.09 2.63 0.33 ***
all confident; 4-very confident)
Confidence in explaining a science topic to someone else (scale: 0-not at all confi- 3.24 0.24 2.58 0.44 ***
dent; 4-very confident)
Confidence in getting good grades in science (scale: 0-not at all confident; 4-very 3.50 0.17 2.98 0.36 ***
confident)
Learning science will improve career prospects (scale: 0-strongly disagree; 3A45 0.11 3.04 0.30 ***
4-strongly agree)
Science is helpful in my everyday life (scale: 0-strongly disagree; 4-strongly agree) 3.23 0.08 2.87 0.29 ***
Science has helped me see opportunities for positive change (scale: 0-strongly 3.27 0.13 2.84 0.32 ***
disagree; 4-strongly agree)
Learning science has made me more critical in general (scale: 0-strongly disagree; 3.14 0.08 2.77 0.30 **
4-strongly agree)
I see myself as a physics person (scale: 0-strongly disagree; 4-strongly agree) 2.74 0.10 2.25 0.41 **





The level statistical significance is coded in the final column: represents a statistical significance less than 0.05 but greater than or equal to 0.01,
** represents a statistical significance less than 0.01 but greater than or equal to 0.001, and *** represents a statistical significance less than 0.001.

148 Chemical Engineering Education










According to the classification outlined above, 123 students
in the sample were categorized as chemical engineering
students (29.3% of which were female) and 691 students
were categorized as "other" engineering students (18.1%
of which were female). The chemical engineering students
were composed of 72% freshman, 21% sophomores, and 7%
upperclassman. Similarly, the "other" engineering students
were composed of 73% freshman, 20% sophomores, and 7%
upperclassman.
For the questions with linear responses, Welch's t-test was
used to compare the mean responses of chemical engineering
with other engineering students.[223 A chi-square test was used
for dichotomous variables to assess whether there is a statisti-
cally significant difference in the responses of the two groups.J23]
For all tests performed in this analysis, the maximum probability
of Type-I error (e.g., a false positive result) that was permitted was
5%. Note that only survey items pertaining to student preparation,
background, and attitudes were analyzed in this paper. All analy-
ses were conducted using the statistical software system R.241

RESULTS
The results of the various t-test and chi-square tests are
summarized in Tables 1 and 2. Only tests relating to the re-
search question that were statistically significant are reported;
in total, 26 linear and seven dichotomous variables showed
significant differences.
For each variable in Table 1, the mean and standard error are
given for both groups of students. The larger mean is listed in
bold. Similarly, Table 2 gives the results from the chi-square
tests. The percentages of each group answering affirmatively
to each factor are listed, followed by the statistical significance.
The higher percentage is listed in bold. Tests for related vari-
ables are grouped together in Table 1: first, career goals (in
gray); second, science identity variables (in white); third, high
school chemistry experiences (in gray). In Table 2 the questions
are also grouped together: first, sustainability factors for career
goals (in gray); second, family involvement (in white); third,
type of high school (in gray).


As indicated in Tables 1 and 2, chemical engineering stu-
dents show several substantial differences from students in
other engineering disciplines. To understand the uniqueness
of chemical engineering students and consider how to spe-
cifically design pedagogy for these students, it is instructive
to consider the meaning of the related blocks of factors that
were found to be significant.
In considering the demographic and prior educational expe-
riences of chemical engineering students, there were several
factors that were found to be not significantly different from
other engineers including: SAT/ACT scores, high school phys-
ical science classroom experiences, family background, num-
ber of AP credits, and math and science preparation factors.
This finding is perhaps not surprising since prior literature has
shown that engineering students in general who persist are
well prepared for their college courses.1151 The only overlap be-
tween the current work and the study by Zhang and colleagues
is students' SAT scores and high school GPA. This earlier
work found that chemical engineering students had higher
SAT scores and GPAs than other engineering studentsJ161
Some reasons for the differences in these findings are that the
transcript data collected by Zhang and colleagues range from
1988 to 2003, while the data in this study were collected from
students enrolled in the Fall of 2011. In 2005, between these
studies, the SAT assessment changed significantly.'251 In addi-
tion, Zhang and colleagues' sample is limited to Southeastern
schools with several listed as research universities with "high"
or "very high" research activity, which may have limited the
earlier sample to exceptional engineering students.1263 There
are a few indicators that students in chemical engineering
come from a somewhat higher socioeconomic background
than other engineering majors: students are more likely to
come from a foreign high school (p<0.05) and these students'
families are more likely to have arranged a tutor in math or
science in the past (p<0.01). Many of the high school class-
room practices and student attitudes were not found to be
different, as well as a number of variables related to students'
high school science course length, class sizes, frequency of
meetings and activities. Similarly, students were questioned


TABLE 2
Chi-square test outcomes for dichotomous variables.
Percent of Other Level of Significance
VilPercent of ChE Students Pgnern f Otr (*:p <0.05,
Variable Indicating (N=123) EgnrigSu p <0.01,
Indicating (N= 123) Indicating (N=691) ***: <0.001)
--- llq ,.- 4"

W^ stto. adj s liatg e chm.^~ -:~ cm -e .< -': ft% IL ; 4-^ **

lt u._ai! s-j.u v.supp i y- .c ... .-ae'r - ,. ~
_-_ _-__._-_' 1 ,% :- -.. .. ..--
Science was a diversion or hobby in my family 37% 26% *
My family arranged for tutoring in science 20% 10% **
Ak n.4edafo.eigNaSgh schol 90% 4%

Vol. 47, No. 3, Summer 2013 149










about their out-of-school science activities such as hobbies,
exposure to science-related media, and possible engineering
influences. These results suggest that, while such factors may
have a significant impact on the recruitment of students into
engineering, no differential effects could be found for chemi-
cal engineering majors.
Students' prior experiences in chemistry courses differed
between chemical engineers and other engineering majors.
Chemical engineering students more often take a higher-level
chemistry course (p<0.05) and have higher chemistry grades
than other engineering students (p<0.01). Additionally, these
students report higher levels of engagement with the material,
positive interactions with other students, and highly rated
chemistry teachers. Student engagement was measured by
how often the student asked questions, answered questions, or
made comments in his/her classes (p<0.001) as well as mea-
suring how interested the student was in his/her high school
chemistry classes (p<0.01). Students identified chemistry
topics as more relevant to their everyday lives (p<0.001).
Chemical engineers listed some career outcome expecta-
tions that were different from other engineering students.
A surprising finding was that chemical engineers reported
a stronger desire to apply math and science in their future
career over other engineers (p<0.001). Chemical engineers
also reported a stronger interest in solving societal prob-
lems (p<0.05) and making use of their talents and abilities
(p<0.05). Some specific ideas related to sustainability were
also highlighted as concerns that chemical engineers spe-
cifically hoped to address in their careers including: energy
(p<0.05), disease (p<0.001), climate change (p<0.05), and
water supply (p<0.001).
Chemical engineering students showed a strong interest
in science and understanding the world around them. They
indicated higher scores on their interest in understanding
natural phenomena (p<0.01), understanding science and ev-
eryday life (p<0.00 1), explaining things with facts (p<0.001),
telling others about science concepts (p<0.001), and making
scientific observations (p<0.001). Another set of questions
measures students' confidence in their scientific and math-
ematical abilities. Chemical engineers reported significantly
higher differences on their abilities to design an experiment
to answer a scientific question (p<0.01), conduct an experi-
ment on their own (p<0.001), interpret experimental results
(p<0.001), write a lab report or scientific paper (p<0.001),
apply science knowledge to an assignment or test (p<0.001),
explain a science topic to someone else (p<0.001), and get
good grades in science (p<0.001).
Identity
To clarify the questions addressing students' interest and
confidence discussed above, a composite measure of "science
identity" was constructed using several of these items. As a
construct, identity has been conceptualized as inherently re-


lated to individuals' self-beliefs. 101 In this study the particular
context of interest is that of an engineering discipline. The
importance of understanding identity is highlighted by Brick-
house and colleagues: If more students are to enter science
and engineering, they need to see themselves as the "kind of
people who would want to understand the world scientifi-
cally."'1' The construct of identity is based on four measur-
able dimensions of students' beliefs about their performance,
competence, recognition by others, and interest.?81 These four
dimensions richly capture an individual's self-perceptions
and can be used to study the development of an engineering
identity specifically in relation to critical events in students'
experiences, their perceptions of the world around them, and
the development of agency (i.e., beliefs about the ability to
act and enact change in one's world) in their lives and careers.
The study of identity has proven useful in understanding col-
lege persistence.'21 This framework for measuring identity
has been previously used in large-scale studies of physics
and mathematics.181
Of the four sub-constructs of identity, the recognition
component consists of beliefs about external recognition by
parents, teachers, other students, etc., of an individual person
as a good science student. Interest in the subject material also
plays a key role in the choice of an engineering major. In this
analysis, the questions used to construct an identity composite
are the interest and confidence questions in Table 1 (which
include students' perceptions of their performance and compe-
tence together) and the questions about family recognition and
involvement in Table 2, all of which were found to be highly
correlated with one another for each of the four sub-constructs
measured.18' These questions were averaged for each of the
sub-constructs of identity (interest, recognition, performance,
and competence) and used to compare chemical engineering
students with other engineering students. Performing a t-test
to compare chemical engineers and other engineers on this
composite shows that the former have a higher overall science
identity than the latter (p<0.001). Thus, chemical engineers
appear to be responsive in ways that are somewhat more akin
to traditional physical scientists than other engineers.

DISCUSSION
To a certain extent, these findings agree with prior work
investigating the differences between engineering disciplines.
Namely, this analysis found that there are, in fact, notable
differences between chemical engineering undergraduates
in career aspirations, perceived identities, and approaches to
learning. This work is a step towards clarifying some of the
differences between students who choose chemical engineer-
ing in college and others.
In utilizing a cross-sectional study design, the data gathered
have some strengths: large statistical power, national repre-
sentativeness in the sample, and the ability to test hypotheses
surrounding events that were introduced to students naturally


Chemical Engineering Education









rather than through an intervention. This study design also
has certain weaknesses, notably including the inability to
draw causal conclusions. Rather, results are correlational
in nature. The results do indicate substantial correlations
between student responses and students' choice of major, but
further work is necessary to indicate a causal direction to these
relationships. For example, students may see chemistry topics
as relevant to their everyday lives because of their choice of
chemical engineering as a major, or they may choose chemi-
cal engineering because of their prior view of chemistry as
relevant to their lives.
Students' experiences in their high school chemistry class-
rooms tell us how students engaged in chemistry classes may
develop a particular connection to the material and see a future
in chemical engineering. Chemical engineering students usu-
ally take a second course in chemistry and do better than other
future engineering students. The particular reasons why these
students choose chemical engineering over chemistry are not
yet clear, but may be rooted in better or more extensive math
preparation, a stronger connection to hands-on applications
of science, or other factors. The differences between chem-
ists and chemical engineers upon entrance into college are an
interesting topic that will be explored in the future.
A stronger interest in solving societal problems and address-
ing issues such as disease may point to chemical engineering
students being interested in industries such as pharmaceuticals
or possibly going on to a career in medicine. Connecting cur-
riculum to current issues facing our global community may
help to harness chemical engineering students' concerns re-
lated to the sustainability-related issues that were highlighted
in this analysis. The inclusion of emerging fields such as
nanotechnology and biomolecular engineering within tradi-
tional chemical engineering instruction is also suggested by
these findings; these topics have direct connections to future
solutions in human health and environmental applications.
Also, in engineering, the perceived lack of a connection to
societal problems is a substantial barrier to women entering
the field,P271 and the subject of sustainability can overcome
this barrier by explicitly connecting engineers' contributions
to solving problems such as resource depletion, catastrophic
climate destabilization, and social inequity. As STEM educa-
tors move toward the recruitment and education of 1 million
extra STEM graduates in the next 10 years, attracting more
women and students from other traditionally marginalized
groups into engineering is vital181 By reducing some of the
barriers to women relating to engineering through curricular
choices, some of these hindrances may be addressed by our
current chemical engineering faculty.
Perhaps unsurprisingly, chemical engineers have more
positive experiences in their high school chemistry courses
than other engineering students. Such findings could be
expected due to typical college admissions requirements
and the motivations of students who traditionally intend to


Students' experiences in their high

school chemistry classrooms tell us

how students engaged in chemistry

classes may develop a particular

connection to the material and see a

future in chemical engineering


enter chemical engineering. These positive experiences may
partially explain why students have a significantly stronger
science identity than their peers. Zhang and colleagues
found that chemical engineering students transferred more
frequently to physical science majors than other engineer-
ing students and that students leaving physical science and
entering engineering chose chemical engineering over other
engineering disciplines more frequently.6"' These results trian-
gulate the current paper's finding of a higher science identity
among chemical engineering students. These students may
also have a stronger connection with chemistry as it relates
to their everyday lives and see chemical engineering as a
way to positively affect the world around them, desires to
solve societal problems, and apply mathematics and science
in their careers. A strong science identity coupled with the
desire to apply math and science also has implications for
educators' curricular choices. Traditionally, students spend
much of their time in the first two years of college learning
basic theory (e.g., fluid dynamics, heat and mass transfer,
thermodynamics). This practice may hinder students' ability
to connect their choice of major with their career goals and
may reduce student persistence in the field or lead to a loss
of motivation and perceived relevance of their chosen field.
Additionally, students may be more engaged with the mate-
rial if the connections to "real-world applications" are made
explicit throughout their college studies rather than simply
giving a perfunctory nod to the importance of the material
for use in activities which may not appear until much later.
For example, students may be told that thermodynamics is
important because it allows them to predict system properties
that will be used later in their design courses and often are
expected to simply learn the principles first. However, the
lack of a connection in the present course may negatively
influence their perception of the material and its usefulness for
their future career. Additionally, students may have difficulty
applying abstract concepts and ideas to practical applications.
Being able to understand the physical meaning of equations
and manipulate those equations is an important engineering
skill. Many students have difficulty grasping and understand-
ing the abstract concepts of thermodynamics, making it one
of the most difficult courses in the undergraduate career.E28'91
Creating connections to real-world scenarios that chemical


Vol. 47, No. 3, Summer 2013










engineering students will implement in their careers may help
students see the importance of the material and grasp concepts
before the final year in senior design.
In previous work, engineering majors have been found to
have marginally lower socio-economic status, stronger math
skills, and less parental and teacher encouragement towards
science than science majors.t301 From the current work, it can
be seen that chemical engineering students are a demonstrably
different group from other engineers. Further investigation
of the specific pre-college influences and experiences that
cause students to choose chemical engineering over other
career choices is a topic for future study. The implications of
the current findings, however, are that students' experiences
in high school chemistry and a desire for deep understanding
of natural phenomena may predict entrance into chemical
engineering, and it may be possible to target students for re-
cruitment into chemical engineering through specific support
and encouragement. Additionally, a pedagogy that reflects
students' deep interest in why things work and the premise
behind particular chemical engineering theories may increase
student interest in chemical engineering coursework.

CONCLUSIONS
The findings in this work have implications for student
recruitment and/or matriculation into chemical engineering
and how to improve the relevance and effectiveness of college
instruction for these students.
To summarize the results of this paper in a useful way, we
have prepared a list of possible considerations that may lend
guidance to the recruitment, retention, and effective instruc-
tion of chemical engineering students:
Given the number of differences in the attitudes of
chemical engineering students identified in this study,
it may be less than optimal to the retention of these
majors to make over-generalizations about "engineer-
ing students" when designing curricula or pedagogy in
general.
As chemical engineering students have been found to
have particularly high expectations towards solving
societal problems in their careers including a more
frequent desire to address sustainability-related issues
(disease, climate change, energy and water supply), it
is likely to be beneficial (to their motivation, engage-
ment, and ultimate performance) to regularly address,
as part of the normal classroom activities, how and
why the content students are learning can be used to
address specific social issues.
Similarly, since chemical engineering students have
been found to put more weight on developing a deep
understanding (of natural phenomena, in everyday life,
using scientific questioning and evidence), attention
should be paid in the classroom to explaining physical
phenomena in more detail and to connecting these top-
ics explicitly to students'everyday lives.


It appears that chemical engineering majors would
benefit particularly from having increased opportuni-
ties to examine scientific evidence and gain experi-
ences providing explanations/argumentation towards
its interpretation. This recommendation is consistent
with the broad movement in STEM towards "active"
learning environments and the emphasis on inquiry
in the classroom; our work indicates that chemical
engineering students would respond especially well to
increased opportunities for this type of learning.
As we found that chemical engineering students were
particularly confident in their abilities to perform tasks
related to their scientific and course activities (write a
lab report, interpret experimental results, apply knowl-
edge to an assignment/test, get good grades), it may be
a waste of time to spend inordinate amounts of class or
laboratory time having students develop these meta-
cognitive skills; rather, putting more emphasis on other
things (as discussed above) may be more beneficial.
Lastly, our results indicate that students who choose
chemical engineering are from slightly higher socio-
economic backgrounds. In order to increase enrollment
and encourage diverse engineering perspectives, less
traditional students that may prove to be highly com-
petent engineers should be recruited.

While chemical engineering students do have clear differ-
ences in their career aspirations, understanding of engineering,
science identity, chemistry background, and family support
than other engineering students, it is important to keep in
mind that this group is nonetheless not homogeneous; there
are a variety of students that may choose to pursue chemi-
cal engineering as a major. Thus, these results should not be
over-interpreted to suggest that there is a "one-size-fits-all"
solution to the successful recruitment and preparation of the
next generation of chemical engineers.

ACKNOWLEDGMENTS
This work has been supported by a National Science Foun-
dation Graduate Research Fellowship (Grant No. 0751278)
and a Research on Gender in Science & Engineering Grant
(No. 1036617).

REFERENCES
1. Johnson, H., and A. Singh, "The Personality of Civil Engineers," J.
Manag. Eng., 14(4), 45 (1998)
2. Dee, K.C., E. Nauman, G. Livesay, and J. Rice, "Research Report:
Learning Styles of Biomedical Engineering Students,"Annals Biomed.
Eng., 30(8), 1100 (2002)
3. Godfrey, E., and L. Parker, "Mapping the Cultural Landscape in En-
gineering Education," J. Eng. Ed., 99(1), 5 (2010)
4. Hunsley,J., CAM. Lee, andJ .M. Wood, "Controversial and Questionable
Assessment Techniques," in Science and Pseudoscience in Contempo-
rary Clinical Psychology, Lilienfeld, S.O., J.M. Lohr, and SJ. Lynn,
eds., New York, Guilford Press (2004)
5. Izraeli, D., M. Krausz, and R. Garber, "Student Self-Selection for
Specializations in Engineering," J. Vocat. Behav., 15(1), 107 (1979)
6. Zhang, G., B. Thorndyke, and M. Ohland, "Demographic Factors and


Chemical Engineering Education











Academic Performance: How Do Chemical Engineering Students
Compare with Others?" in American Society for Engineering Educa-
tion Proceedings, Nashville, TN (2003)
7. Witt, P., and P. Handal, "Person-Environment Fit: Is Satisfaction
Predicted by Congruency, Environment, or Personality?" J. College
Student Personnel, 25(6), 503 (1984)
8. Hazari, Z., G. Sonnert, P.M. Sadler, and M.C. Shanahan, "Connecting
High School Physics Experiences, Outcome Expectations, Physics
identity, and Physics Career Choice: A Gender Study," J. Res. Sci.
Teach., 47(8), 978 (2010)
9. France, M.K., 0. Pierrakos, J. Russell, and R.D. Anderson, "Measuring
Achievement Goal Orientations of Freshman Engineering Students,"
in ASEE Southeastern Section Conference Proceedings, Blacksburg,
VA (2010)
10. Johnson, A., J. Brown, H. Carlone, and A. Cuevas, "Authoring Iden-
tity Amid the Treacherous Terrain of Science: A Multiracial Feminist
Examination of the Journeys of Three Women of Color in Science,"
J. Res. Sci. Teach., 48(4), 339 (2011)
11. Brickhouse, N.W., P. Lowery, and K. Schultz, "What Kind of a Girl
Does Science? The Construction of School Science Identities," J. Res.
Sci. Teach., 37(5), 441 (2000)
12. Carlone, H.B., and A. Johnson, "Understanding the Science Experi-
ences of Successful Women of Color: Science Identity as an Analytic
Lens," J. Res. Sci. Teach., 44(8), 1187 (2007)
13. Shivy, V., and T. Sullivan, "Engineering Students' Perceptions of
Engineering Specialties," J. Vocat. Behav., 67(1), 87 (2005)
14. Durmnnette, M.D., P. Wemrnimont, and N. Abrahams, "Further Research on
Vocational Interest Differences Among Several Types of Engineers,"
Pers. Guid. J., 42(5), 484 (1964)
15. Seymour, E., and N. Hewitt, Talking About Leaving: Why Undergradu-
ates Leave the Sciences, Boulder, CO, Westview Press (1997)
16. Walden, S., and C. Foor," 'What's To Keep You From Dropping Out?'
Student Immigration into and within Engineering," J. Eng. Ed., 97(2),
191 (2008)
17. Ohland, M.W., S.D. Sheppard,G. Lichtenstein, 0. Eris,D. Chachra, and
R.A. Layton, "Persistence Engagement and Migration in Engineering


Programs," J. Eng. Ed., 97(3), 259 (2008)
18. President's Council of Advisors on Science and Technology, Engage
to Excel: Producing One Million Additional College Graduates with
Degrees in Science, Technology, Engineering, and Mathematics (2012)
19. Bruntland, G.H., Our Common Future, New York, Oxford University
Press (1987)
20. Mackun, P., and S. Wilson, United States Census Bureau, (2010),
(accessed Nov. 10,
2012)
21. BatchGeo, (2012), (accessed Oct. 25,
2012)
22. Welch, B.L., "The Significance of the Difference Between Two Means
When the Population Variances are Unequal," Biometrika, 29(3/4), 350
(1938)
23. Ott, L.R., and M.T. Longnecker, An Introduction to Statistical Methods
and Data Analysis, 6th ed., Belmont, CA, Brooks/Cole Press (2008)
24. The Core Development Team. R: A language and environment for
statistical computing (accessed Aug. 21,2012) org>
25. College Board History of SAT Assessment, org/about-tests/history-of- the-tests>, (accessed Oct. 22,2012)
26. Carnegie Foundation Institution Classification, camegiefoundation.org/lookuplistings/standard.php> (accessed Nov.
2,2012)
27. Widnall, S., "Digits of Pi: Barriers and Enablers for Women in Engi-
neering," in SE Regional NAE Meeting Proceedings, Atlanta (2000)
28. Hadfield, L.C., and C.E. Wieman, "Student Interpretations of Equations
Related to the First Law of Thermodynamics," J. Chem. Ed., 87(7),
750 (2010)
29. Thomas, P.L., and R.W. Schwenz, "College Physical Chemistry Stu-
dents' Conceptions of Equilibrium and Fundamental Thermodynam-
ics," J. Res. Sci. Teach., 35(10), 1151 (1998)
30. Potvin, G.,R. Tai, and P. Sadler, "The Difference Between Engineering
and Science Students: Comparing Backgrounds and High School Expe-
riences," in American Society for Engineering Education Proceedings,
Austin, TX (2009) 0


Vol. 47, No. 3, Summer 2013










[L= classroom


ACTIVE LEARNING AND

JUST-IN-TIME TEACHING

In a Material and Energy Balances Course


MATTHEW W LIBERATORE
Colorado School of Mines Golden, CO 80401


Material and energy balances are used daily by most
practicing chemical engineers across a wide range
of job duties and industries. Due to the foundational
nature, the material and energy balances course is usually
delivered first in the chemical engineering curriculum. The
literature includes numerous papers on the importance of the
course, the difficulty of the course and its concepts, and high
fail rate (i.e., reputation as a "weed out" course).1-7' Here, 21st
century tools and techniques add to the established learning
tools and have led to improved outcomes for the course.
Felder and Rousseau's textbook181 is widely adopted for the
course and defines the structure and course topics covered.
While the concepts covered in the course have not dramati-
cally changed recently, how the course is delivered has been
altered by the availability of technology. A very recent sur-
veyE41 on how the course is taught elucidates numerous trends
for the course. One clear evolution of the course delivery is
the widespread use of software tools such as Excel (spread-
sheets), Matlab (advanced mathematics), and many others.
Although not covered in the survey, course-specific tools have
also been developed.
Online homework from Sapling Learning has supplemented
or replaced traditional problem sets out of the textbook for
some instructors of a material and energy balances course.16.91
Features of this tool include personalized problems (i.e.,
same problem statement with different numbers), multiple
attempts for the students to work until the problem is com-
pleted correctly, hints and tutorials available in real time, and
real-time grading and class statistics. In addition, the rolling
numbers on each problem make creating a solutions manual
for all variations difficult. Therefore, the online homework
dramatically decreases a common concern about the course,
namely cheating through the availability of downloadable
solutions manuals.14'101 Another tool designed to improve
students' problem-solving skills is open source educational
software called ChemProV.14,111 ChemProV is a chemical
process visualizer that helps students learn material bal-


ances through the construction of process flow diagrams.
This scaffolded software tool led to statistically significant
improvement in problem-solving accuracy when dynamic
feedback was built into the tool. Overall, a critical aspect of
the Sapling homework and ChemProV are the immediate
feedback mechanisms.
Leveraging technology to provide real-time feedback to
students, both inside and outside of class, has spurred an in-
structional approach called just-in-time teaching (JITT).112"141
The most widely used form of JITT centers on the use of
clickers.113'15-171 More than a decade ago, a group of physics
faculty created assignments due before every class to mini-
mize the ebb and flow or cramming throughout a semester.
Not only did the faculty have a large amount of data on the
students' learning and misconceptions, the faculty could
improvise within the current class period and address the
students' knowledge gaps. A more recent treatise covering
JITT across disciplines"121 presents a number of techniques
and settings to collect learning information from students'
responses. The common theme is to stop regularly (within
a class period or several times per week) so students and
the instructor can assess what has been learned. Numerous
platforms to interact and collect learning data exist (e.g.,
clickers, pen-based technologies, course-management sys-
tems, and concept warehouses and inventories).


Copyright ChE Division ofASEE 2013
Chemical Engineering Education


Matthew W. Liberatore is an associate
professor of chemical and biological engi-
neering at the Colorado School of Mines.
He earned a B.S.degree from the University
of Illinois at Chicago and M.S. and Ph.D.
degrees from the University of Illinois at
Urbana-Champaign, all in chemical engi-
neering. His current research involves the
rheology of complex fluids as well as active
and self-directed learning.










Overall, delivering courses to students who are digital na-
tives (e.g., References 18-20) can involve numerous active
-learning techniques and technology12[1 to keep the activity
level in the room high, independent of class size. Three main
sections of this work include teaching in a large class envi-
ronment (>100 students), homework and JITT response, and
assessment. Course surveys and grades provide two assess-
ment tools in evaluating the effectiveness of these various
techniques.

COURSE OVERVIEW
At the Colorado School of Mines (CSM), the chemical
engineering curricula (i.e., for accredited degrees in chemi-
cal engineering and chemical and biochemical engineering)
begin with a course in material and energy balances, which is
delivered in the spring of the sophomore semester. The place-
ment in the curriculum is one term later than many schools.141
About 80% of the students in the course have completed a
core sophomore-level thermodynamics course that covers
a number of energy balance concepts. The course format is
three 50-minute class meetings per week at 8 a.m. in a single,
large classroom, with an enrollment of more than 150 stu-
dents in 2011 and 2012 (Table 1). The course had been taught
with multiple sections and instructors (including the author)
during 2009 and 2010. A number of reasons for moving to
a single section are outlined in this manuscript (e.g., technol-
ogy such as online homework, creating a small class within
a large class). Larger sections are becoming more common
for this course in recent years,141 and a number of different
approaches can be employed without overwhelming the in-
structor or "weeding out" large numbers of students. While
traditional graduate student teaching assistants have not been
available for the single primary instructor setting, a group of
three or four senior undergraduates assist the instructor in
the classroom as well as in grading homework and quizzes.
Grade point averages are on a 4.0 scale and are consistent
with those reported earlier.131
The course's content follows the textbook by Felder and
Rousseau,181 which is used in ~85% of chemical engineering
programs.J4' To mitigate the course's cost to the students, the
textbook was a suggested resource in 2011 and 2012 (espe-
cially the ~$50 ebook version from Wiley compared to >$200

TABLE 1
Class statistics of material and energy balances class at the Coloi
School of Mines. Statistics do not include students withdrawing I
the course.
Year no. of Average %C or Sections Prin
students GPA better Instn
2012 142 2.39 80 1 1
2011 147 2.50 82 1 1
2010 156 2.38 79 3
2009 96 2.04 69 2

Vol. 47, No. 3, Summer 2013


hard cover book at the university bookstore). Once the text-
book's solutions manual is available, the utility of the book as
a whole decreases dramatically, in the author's opinion. Most
students, however, have access to a version of the textbook.
While no formal handouts or alternative textbook are used,
all notes written by the instructor during class are scanned
and posted. The primary "textbook" cost is for the Sapling's
online homework (~$35/student). The course can be divided
into three main areas, namely the classroom environment,
homework, and assessment of learning. All three components
play a critical role in the delivery of the course.

CLASSROOM ENVIRONMENT
An active-learning classroom is created using peer-to-
peer instruction, YouTube videos with course-related prob-
lems,122241 and JITT feedback from the previous assignment.
The majority of class time centers on activity by the students,
applying learning-by-doing to the course. In addition, work
implementing a small class within a large class was instituted
to engage a larger number of students during each class period.
Providing structure in peer-to-peer instruction exercises
improves focus and decreases the number of students "wait-
ing to be taught" by a lecturing professor.'17'251 The teacher-
centered instruction (i.e., lecture) is limited to two or three
5-10 minute blocks per 50-minute class period. Groups of
three students are formed at the beginning of the semester
and usually maintained throughout the term. Students have
self-selected their groups in recent years. Sitting in groups of
three provides a format to randomly assign three roles when
working on examples. The roles are leader, questioner, and
scribe. The leader takes the lead, does the talking to initiate
problem solving, and outlines the steps to complete the prob-
lem. The questioner listens to the leader and asks questions if
something is unclear or seems incorrect. The scribe's role is
to write key steps to the solution for the group and share the
solution with the group during or after class. The group work
time varies from 2 minutes for concept questions and simpler
tasks such as drawing and labeling process flow diagrams
to 10 minutes for writing and solving multiple balances. A
timer is projected to keep students on task for these periods,
however if student use of electronic devices with games or
text messages starts to increase, the time is cut off and refo-
cused on the next segment of the class. The roles
are randomly rotated by a set of cards used by the
rado instructor (e.g., tallest-leader, shortest-scribe).
om With groups working diligently during 40 to 75%
of the class time, the instructors use this time to
iary actively engage a number of groups. At least one
Actors
instructor for every 40 to 50 students is needed to
I engage the groups in a large class setting. Faculty,
I1 graduate students, or senior undergraduate students
3 can fill this role as secondary instructors during
2 group activities. Having a diversity of instructors,











1 Draw and label PFDI

2 Is a basis defined? OR choose a basis?

3 FDefine system of interest
S --I .



Write overall
and component
Mass balance
equations


Write overall
and component
Mole balance
equations


Write overall
and component
Atom balance
equations


1
7 Write given extra equations

8 Determine # of unknowns


9 Can the defined system be solved? 0N
Yes Yes
10 Are other systems needed? P-e
I No
"1 Solve for all unknowns

12 Check answers


Figure 1. A 12-step method for solving material balance
problems.
i.e., young/old, male/female, etc., can help build relationships
with the students and appropriately represents the future work
environment. The idea of using secondary instructors in large
classes is not new and has been implemented successfully
in 1,000 student sections with improved learning.1261 For the
material and energy balances course in 2011 and 2012, three
instructors alternated walking the front, middle, and back
of the room depending on the day of the class (Monday,
Wednesday, or Friday). Since students normally sit in the
same general areas, each instructor had the opportunity to
engage all of the students regularly.
The primary focus of the class is on providing basic tools
for problem solving related to chemical engineering prob-
lems. A basic framework for problem solving in the course is
summarized in a 12-step process (Figure 1). The utility of a
12-step method for energy and entropy problems (i.e., first and
second law) was recently summarized[271 and provided a linear
problem-solving scheme. For more complex material balance
problems, specifically multi-unit operations, required decisions
and loops are added to the framework. The 12-step process
is complementary to evaluating degrees of freedom, which
is a point of emphasis in the Felder and Rousseau textbook.
Feedback from students on the 12 steps (done with anonymous


notecards near the middle of the semester) finds the steps useful
but not critical to students learning the new material.
A final active-learning component used in the classroom is
YouTube Fridays. Several papers have been published on this
topic providing the details and feedback on this technique.122-241
Briefly, students select videos from the Internet and write a
course-related problem based on the events of the video (a
collection of videos is available'281). Most course topics have
been covered by these problems over the past few years
including multiple units, reaction-recycle systems, and vapor-
liquid equilibrium. Therefore, selecting the most interesting
and challenging YouTube problems to replace the "tried and
true" textbook examples increases the energy level in the
room. At this point, the examples are not restricted to Fridays
since the database of problems has grown steadily in recent
years. Groups of three students create one YouTube problem
as a project during the semester. Overall, the integration of
visuals is an established technique to increase learning, and
the sense of personalization of the course engages a large
number of digitally native students[118201

HOMEWORK
Three homework assignments per week instill hard work
and persistence. One assignment is due each Monday,
Wednesday, or Friday class period except for exam days (i.e.,
13 assignments of each type of homework are due over the
course of a semester). The delivery, length, and content of the
assignments vary by assignment type. Short multiple-choice
quizzes, personalized online homework, and a traditional
"textbook" homework are the three types whose utility will
be detailed here.
Instructor-written multiple-choice quizzes are delivered
within the course management software (i.e., Blackboard in
this case). The content was developed over one semester with
updating each semester to avoid the solutions being passed
down from the previous year's students. The quizzes ask five
to 10 questions per week covering vocabulary, basic calcula-
tions (e.g., stoichiometric coefficients, vapor pressure), and
concept questions. Adapting pieces of textbook examples or
homework is one type of problem. For adding "bio" content,
the BioEMB databaset29] contains a wealth of full-length prob-
lems that can be simplified for this format (Figure 2). Perform-
ing atom balances on non-integer stoichiometry (e.g., yeast
in Figure 2) emphasizes the universality of the atom balance
vs. balancing reaction stoichiometry by inspection. Overall,
these quizzes primarily cover material at the remember and
understand levels of Bloom's taxonomy.
For developing skills such as applying and analyzing (levels
3 and 4 of Bloom's taxonomy), personalized online homework
and handwritten homework fill the role. The initial experi-
ment with Sapling Learning's personalized online homework
was published previously.161 In summary, the students using
Sapling earned consistently and statistically significant higher
Chemical Engineering Education










quiz and exam scores, leading to a much
lower fail rate for the course compared
to students only completing textbook
homework. The improved student
achievement related to online home-
work led to adoption of this technology
for the two more recent offerings of the
course. Most of the online homework
problems are as rigorous as the text-
book (e.g., multiple units, multiple-part
problems). The personalization of the
problems comes from rolling numbers
within the problem statement. Thus,
the concepts and problem-solving skills
are the same from student to student


but the numerical answers are differ-
ent. Students were allowed to work in groups on the online
homework, but each student needed to apply the correct bal-
ances to his or her set of numbers. No data was collected to
quantify how many students worked in groups for any of the
homework types. Most of the Sapling problems include hints
to help students start or to correct errors. Also, some of the
problems include full tutorial problems covering the similar
concepts before attempting the problem for a grade.
Problem sets done with paper and pencil are the third type
of homework. Each year fewer problems are taken from the
textbook to minimize the amount of rote copying of the solu-
tions manual, which was discussed earlier. Alternate problems
and solutions exist without a huge time commitment by the
instructor or teaching assistants. Textbook problems with dif-
ferent numbers require work beyond copying the solutions
manual. Rolling numbers is trivial in simpler cases (e.g.,
non-reacting systems) and strongly constrained in others (e.g.,
vapor-liquid equilibrium). Other sources include problems from
other textbooks, the BioEMB database, and old quiz and exam
questions. Doing some problems with pencil and paper each
week is the best way to simulate quiz and exam situations for
the students. While final numeric answers are given on some
of the paper homework problems, focus in grading is placed
on the problem-solving technique and correct balances, which
is also how exams are graded.
While traditional textbook homework is graded within a
week of completion (by undergraduate graders in this case),
the short quizzes and online homework allow for just-in-time
feedback. Both Blackboard's course management software
and Sapling's online homework instantly tabulate individual
and aggregate grades for evaluation. Both systems tabulate
class averages for each problem while Sapling also produces
a matrix with varying colors to represent the number of at-
tempts the students needed on a specific problem. On Sapling,
the average score is not always the best representation of the
class's performance. Students who do not persist to the correct
answer receive no credit for the problem (a very small frac-
tion of the class). Distinguishing between the class needing
Vol. 47, No. 3, Summer 2013


A yeast (CHI .66N0. 1300.40) is growing aerobically on arabinose (C5H 1005)
and ammonium hydroxide (NH40H) with a respiratory quotient (ecb) of 1.4.
The reaction is:
a C5HOS1005 + b 02 + c NH4OH --> d CHI.66N0. 1300.40 + e C02 + fH20O

Assume 1 mole of yeast as the basis. What is e?

0 0.853
O 0.299
0 0.974
O 0.411

Figure 2. Example of a multiple-choice quiz problem based on content in the
P;-IAlfp


several attempts on average to complete a problem and a low
average skewed by a number of students giving up or not
attempting a problem is data available for the instructor's
professional judgment.
Two of the three class meetings begin by addressing one or
more sticking points from the most recent homework assign-
ment (due two hours before class begins). The JITT exercises
last from 2 to 10 minutes. For example, a short lecture reviews
and reinforces unclear concepts identified in the homework.
Alternatively, active problem solving has included re-doing
the most difficult problem in their groups, isolating one part
of a problem for discussion and resolution, or assigning
another problem covering the concept as the problem with
the low score. Overall, online tools provide feedback to the
instructor instantly that can help keep the students focused on
the most important topics in the course. The JITT exercises
need additional prep time for the instructor, which is not very
difficult if the instructor has taught the course before. The
assessment of the JITT exercises and homework is included
in the next section.

ASSESSMENT
Homework, quizzes, and exams contribute to the grades
earned by students in the class. In addition, formal and informal
student surveys provide a second perspective on the multiple
homework format and JITT. First, the grades for the three
types of homework are aggregated into a single portion of the
course graded (~ 15%). The average grades for homework are
generally high (~90%) for the students who complete all of the
problems. Next, in-class quizzes approximately 10- given
over the course of a semester provide a means to simulate the
exam environment with a problem similar to exam problems.
These quizzes take 10 minutes for vocabulary to 25 minutes
for longer problems such as reaction with recycle problems.
Some quizzes are announced while others are not, to encour-
age consistent studying of the course material (i.e., avoid
cramming before exams). On average,the students earn ~75%
on the quizzes. While the majority of the students' effort for









the class is on homework and the 10 quizzes, exams make up
the majority of the student's course grade.
The timing and frequency of major exams are especially
important in the material and energy balances course. As
pointed out previously,t2' the course starts out deceptively
simple (e.g., units, density) and quickly builds into multi-unit
problems that do not always have an obvious place to start
solving. During the previous four years, either two or three
preliminary exams preceded a cumulative final exam. In years
using the two-exam format, students covered the first four
chapters of Felder and Rousseau before the first preliminary
exam and the first eight chapters before the second prelimi-
nary exam. At the time, the logic of covering four chapters of
material and then giving an exam seemed correct and in line
with the previous deliveries of the course. In 2009, however,
more than 75% of the class earned less than 60 out of 100
and a number of students dropped the course as a result. The
main feedback from the students was that the difficulty of the
material, specifically reaction-recycle problems, was greater
than other sophomore-year courses (e.g., math or chemistry).
It was decided that overwhelming students in their first exam
in their chosen major is not the best way to encourage students
to enjoy the chemical engineering profession.
Further, the two-exam format with so many low scores
required a curve, and students thought their grades were
somewhat arbitrary. Thus, the next year (2010) the three-
exam format was adopted and the distribution of material
changed. The first preliminary exam covered the first three
chapters (i.e., no reacting systems), the second preliminary
exam emphasized reaction systems and vapor liquid equilib-
rium (Chapters 4 through 6), and the final preliminary exam
focused on energy balances (Chapters 7 through 9).
All exams are cumulative but emphasize the most recent
material. In 2010, it turned out that the first exam provided
a false sense of confidence, i.e., an exam without reacting
systems was trivial (over 91% average). The parsing of the
second and third exam materials gave sensible averages (mid
60s to mid 70s). Therefore, as a further refinement in 2011 the
additional material was covered before the first exam, namely
single-unit reacting systems (the first part of Chapter 4). The
results for 2011 and 2012 showed this new timing for the first
exam as optimal with averages of 78 and 74, respectively.
While a fraction of the class still earns a failing score on


the first exam, the exam is representative of the rigor of the
rest of the course and curriculum. As a side note, the ABET
continuous improvement forms were used as a way to build
this knowledge related to the exam scores.
Overall, course grades and the number of students earning
a C or higher in the course have improved in recent years
(Table 1). While the author's university teaching evaluations
have fallen below the university average for the large course
sections the last two years, student learning has improved by
another metric. The number of students failing chemical en-
gineering courses the next semester, namely thermodynamics
and fluid mechanics, decreased to a four-year low after the
Fall 2011 semester (the most recently available data). While
course grades are not a standardized metric for engineering
education researchers, trends can demonstrate the utility of
the teaching strategies discussed earlier.
Online homework was shown to have a significant impact
on student achievement when two control sections of the
course were compared to one using online homework from
Sapling Learning in addition to textbook homework.t61 The
success of the online homework in 2010 led to its universal
adoption during the past two offerings. Grades, although an
incomplete metric, show a measurable improvement since
the adoption of online homework for the course (Table 2). In
addition to the dramatic shrinking of students earning an F
grade in the course, the percentage of students withdrawing
from the course also decreased (i.e., from 7.5% to 6.5% of
the total enrollment). The results are statistically significant
(p<0.0001) and consistent with respect to a higher percentage
of students earning the C or higher grade needed to enroll in
the junior courses. In addition, student surveys beyond the
university course evaluations provide insights into which
techniques the students feel are helpful.
Three student surveys have been administered during the
last two offerings of the course, i.e., online homework, just-
in-time teaching, and YouTube Fridays. YouTube survey
results are covered elsewhere .t22-241 Surveys related to online
homework show a number of interesting trends. During its
first introduction in 2010, the students preferred the textbook
homework (Table 3), with respect to their perception of gain-
ing understanding and "liking." Online homework and Sapling
Learning were unfamiliar to most of the students in 2010,
outside of freshman physics (i.e., LON-capat301); however,


TABLE 2
Grades earned when using online homework or not during the last four years.
% students earning grade Average no. of %C or
Condition and Years' D course students better
A B C D F GPA students better
With Online Homework 23 29 30 12 5.5 2.52 345 82
Without Online Homework 17 20 33 12 17 2.08 196 70
1 With Online Homework occurred for some students in 2010 and all students in 2011 and 2012. Without Online Homework occurred for
all students in 2009 and some students in 2010.

58 Chemical Engineering Education










TABLE 3
Student survey responses to four questions related to online homework.
% Strongly Agree/Agree 2010 2011 2012
Online homework helps me understand the 85 96 95
course concepts and topics.
Textbook homework helps me understand the 92 84 92
course concepts and topics.
I like doing Online homework. 50 75 60
I like doing Textbook homework. 65 42 52
Note: n=52 students for 2010, 134 students for 2011, and 123 students for 2012


60 -




S40 -
Gi
W



20 -



0-


To maximize learning of the course material,
completing _____ homework is necessary


E2011


L* I K_
Online Textbook


Online +
Textbook


Online +
Textbook +
MC Quiz


Figure 3. Students' preferences on homework type(s)
over the last three years. The n-values for each year are
included with Table 3.

online homework is becoming a more standard tool with use
in organic chemistry, mechanics, and other courses across
campus during the last few years. In the two subsequent
years, students scored online homework higher than textbook
homework on both questions. The category "understanding
course concepts from online homework" received almost
unanimous response during 2011 and 2012.
Another survey question probed the homework type or types
that students perceived help them learn the course material.
Textbook homework as a singular homework type received a
majority of the responses in 2010, but has garnered only 3%
of the response in the two most recent offerings.
To summarize, both familiarity with online homework and a
smaller number of glitches with the online homework system
likely led to the very favorable survey results over the last
two years. Additionally, the vast majority (~80%) students in
2011 and 2012 believe that multiple types of homework help
maximize their learning.


The final student survey probed the students'
feedback on just-in-time teaching. As discussed
earlier, multiple-choice quizzes and online
homework provided immediate results to the
instructor, which were acted upon to adjust the
course content to the current group of students.
Responses from 2011 and 2012 were averaged
since the students responded with the same level
of agreement (i.e., within 3%). First, the im-
mediate feedback from the homework resonated
with the majority of the students (Figure 3). The


students agreed that the JITI process gave them
a means to be an active participant in class and
was an effective use of class time. Students clearly understand
that the instructor is aware of their strengths and weaknesses
as well as not just delivering the same lecture as every previ-
ous year. The instructor taking class time to address students'
concerns and deficiencies in real time (i.e., not just in the exam
review weeks later) is appreciated. Finally, more than 86% of
students liked reexamining difficult course material during the
JITT exercises (see Figure 4, next page). Focusing class time
on the most important material has always been an instructor's
prerogative, but now the instructor determines some of that
important material from the responses of the students via online
tools. The compromise on using class time for JITT exercises
has been removing some introductory lecture material from
class (e.g., definitions). Overall, implementing JITT should
become more common as more online tools are developed and
available to faculty.

CONCLUSION
A number of techniques for delivery of a material and en-
ergy balances course have been explored and several items
optimized over the last four years teaching the course. First,
student engagement is achieved even at large class sizes by
using multiple instructors-corroborating findings in other,
non-engineering disciplines. Active-learning techniques,
including short problem-solving periods in teams, problems
based on YouTube videos, and JITT exercises, keep students'
attention by varying the activity every 10 to 15 minutes. Next,
a move away from textbook-based homework was necessary
to avoid rote copying of the solutions manual that is available
via a simple web search.
A combination of homework types has proven successful in
engaging students several times per week in the course material.
The implementation of Sapling Learning's online homework
has allowed self-directed and personalized problem solving
as well as the ability to deliver just-in-time feedback to the
class (i.e., only hours after students complete the assignment).
Traditional paper and pencil homework and multiple-choice
quizzes round out the homework assignments each week, and
the quizzes also allow JITr feedback. Overall, JITT exercises
received positive feedback from student surveys.


Vol. 47, No. 3, Summer 2013











Individual exams and surveys provided as-
sessment of the changes to the course. Timing
of the first of three preliminary exams is criti-
cal to provide a fair assessment and minimize
the students withdrawing from the course and
likely changing majors. Student surveys show
a strong preference (~80%) for multiple types
of homework, especially online homework, to
maximize their learning. In total, more active
and self-directed tools with immediate feed-
back are needed to enhance the engineering
education community in the near future.

ACKNOWLEDGMENTS
The author thanks Theresa Nottoli for data
entry from the paper surveys. Partial support
from the National Science Foundation through
CBET-0968042 is acknowledged.

REFERENCES
1. Felder, R.M., "Knowledge Structure of the Stoichi-
ometry Course," Chem. Eng. Ed., 27,92 (1993)
2. Felder, RAV., "Stoichiometry Without Tears," Chem.
Eng. Ed.,24, 188 (1990)
3. Bullard,L.G., and R.M. Felder, "A Student-Centered
Approach to Teaching Material and Energy Balances
2. Course Delivery and Assessment," Chem. Eng.
Ed., 41,167 (2007)
4. Silverstein, D.L., L.G. Bullard, and MA. Vigeant, "H
Material and Energy Balances," Proceedings of the 20
ference 2012, AC 2012-3583
5. Keith, JM., D.L. Silverstein, and D.P. Visco Jr., "Ideas
New Chemical Engineering Educators: Part 1 (Courses
in the Curriculum)," Chem. Eng. Ed., 43,207 (2009)
6. Liberatore, M.W., "Improved StudentAchievement Usil
Online Homework for a Course in Material and Ene
Chem. Eng. Ed., 45,184 (2011)
7. Faraji, S., "The Enhancement of Students' Learning
Division and Upper Division Classes by a Quiz-Bas
Chem. Eng. Ed., 46,213 (2012)
8. Felder, R.M., and R.W. Rousseau, Elementary Principl
Processes, 3rd Ed., Wdey (2005)
9. Walton, S.P., and A.P. Malefyt, "Increasing the Spiral
and Energy Balances," Proceedings of the 2012 ASE
AC2012-3359
10. Choi, C., "The Pull of Integrity," Prism, 18,28 (2009)
11. Hundhausen, C., P. Agarwal, R. Zollars, and A. Carte
and Experimental Evaluation of a Scaffolded Software I
Improve Engineering Students' Disciplinary Problem-S
J. Eng. Ed., 100,574 (2011)
12. Just-In-Tlime Teaching: Across The Disciplines, Across
Simkins, S. and M. Maier, eds. Stylus Publishing: Sterl
13. Novak, G., A. Gavrin, W. Christian, and E. Pattersor
Teaching: Blending Active Learning with Web Techm
Hall, Upper Saddle River, NJ (1999)
14. Prince, MJ., and R.M. Felder, "Inductive Teaching
Methods: Definitions, Comparisons, and Research Base
95,123(2006)
15. Falconer, J.L., "Use of ConcepTests and Instant Feedba
dynamics," Chem. Eng. Ed., 38,64 (2004)
16. Falconer, JL., "Conceptests for a Chemical Engineeri
namics Course," Chem. Eng. Ed., 41,107 (2007)


I


100
n=252 students


80



60


40




g 20'



0

JlTr exercises helped JITT exercises are an I like the immediate
me feel like an active effective use of class refresher of difficult
participant during class time material using JITT


Figure 4. Percentage of students who responded "agree" or "strongly
agree" to three statements related to Just-In-Time Teaching exercises. Data
are averages from 2011 and 2012.

17. Crouch, C.H., J. Watkins,A.P. Fagen, and E. Mazur, "Peer Instruction:
[ow We Teach: Engaging Students One-on-One,All at Once," Research-BasedReform
12 ASEE Con- of University Physics, 1,40 (2007)
18. Digital Natives blog (accessed July 12,
to Consider for 2010)
Offered Earlier 19. Palfrey,J., and U. Gasser, Born Digital: Understand the First Genera-
tion of Digital Natives, Basic Books (2008)
ng Personalized 20. Tapscott, D., Grown Up Digital: How the Net Generation Is Changing
rgy Balances," Your World, McGraw-Hill (2009)
21. Koretsky, M. and B. Brooks, Student Attitudes in the Transition to an
in Both Lower Active Learning Technology," Chem. Eng. Ed., 46,289 (2012)
ed Approach," 22. Liberatore, M.W., "YouTube Fridays: Engaging the Net Generation in
5 Minutes a Week," Chem. Eng. Ed., 44,215 (2010)
les of Chemical 23. Liberatore, M.W., C.R. Vestal, and A.M. Herring, "YouTube Fridays:
Student-Led Development of Engineering Estimate Problems," Ad-
ity of Material vances in Eng. Ed., 3,1 (2012)
E Conference, 24. Liberatore, M.W., D. Marr, AM. Herring, and J.D. Way, "Student-
Created Homework Problems Based on YouTube Videos," Chem. Eng.
Ed., 46(2), 122 (2013)
r, "The Design 25. Prince,M.,"DoesActiveLearningWork? A Review oftheResearch"
Environmentto J. Eng. Ed., 93,223 (2004)
Solving Skills," 26. Prather, E., A. Rudolph, G. Brissenden, "Using Research to Bring In-
teractive Learning Strategies into General Education Mega-Courses,"
The Academy, Peer Review, 13 (2011)
ing,VA(2010) 27. Liberatore, M.W., "Problem Solving in 12 Steps For Introductory
n, Just-In-lime Thermodynamics," Chem. Eng. Ed., 45, 0 (2011)
ology, Prentice 28. Liberatore, M.W., Rheology of Complex Fluids Laboratory Home
Page, (accessed November 2012)
and Learning 29. Komives, C., M. Prince, E. Fernandez, and R. Balcarcel, "Integration
s," J. Eng. Ed., of Biological Applications into the Core Undergraduate Curriculum:
A Practical Strategy," Chem. Eng. Ed., 45,39 (2011)
ick in Thermo- 30. Kortemeyer, G., E. Kashy, W. Benenson, and W. Bauer, "Experiences
Using The Open-Source Learning Content Management and Assess-
ng Thennrmody- ment System LON-CAPA in Introductory Physics Courses," Amer. J.
Physics, 76,438 (2008)0

Chemical Engineering Education










L=] laboratory
----------------


COMPARISON BETWEEN

LINEAR AND NONLINEAR REGRESSION

In a Laboratory Heat Transfer Experiment


CARINE MESSIAS GONCALVES,1 MARCIO SCHWAAB,1 AND JosE CARLOS PINTO2
1 Universidade Federal de Santa Maria, Cidade Universitdria Santa Maria, RS, 97105-900, Brazil
2 Universidade Federal do Rio de Janeiro, Cidade Universitdria Rio de Janeiro, RJ, 21941-972, Brazil


he estimation of parameters from experimental data
is a very common practice in teaching and research
laboratories in chemical engineering and related fields.
Data analysis and estimation of model parameters are usually
performed without the required statistical accuracy, however,
making the interpretation of obtained results more difficult and
leading to erroneous and/or ambiguous parameter estimates.
As the mathematical models used to represent chemical pro-
cesses are generally nonlinear, estimation of model parameters
can only be performed with the help of numerical procedures.
For this reason, a very common practice consists of rewriting
the model in a linearized form through manipulation of the
original measured values in order to facilitate the mathemati-
cal treatment of the data and allow for analytical analysis.
In a laboratory lesson about transport phenomena, interpre-
tation of the data collected by students is usually performed
through a linearized version of nonlinear models. Also, for
the students this procedure had become the only way to per-
form this analysis. In some cases where linearization of the
model is not possible, sometimes the students believe that it
is not possible to perform the data analysis. In the educational
literature there are some works where nonlinear regression is
introduced to students, usually through the use of the com-
mercial spreadsheet softwares.11'21 In a paper by Fahidy,t31 the
use of statistical-based procedures in an undergraduate course
is presented. Its importance in many aspects and subjects of
chemical engineering education and research is discussed.
Also, the problem associated with the misuse of linearized
models instead of nonlinear ones is briefly addressed.
Unfortunately, even in the scientific literature the use of
linearized models is still very common. Although this type of
manipulation of model equations is not necessary nowadays,
given the impressive development and availability of com-
puter resources, this procedure still finds widespread use for
estimation of parameters of the Arrhenius[4-61 and Langmuir
equations,t7,1 among many others. Linearization of the original


nonlinear model leads to estimation of parameters that are
different from those obtained when the original nonlinear
nature of the model is preserved, however, as a consequence
of the statistically biased parameters that are estimated, as
pointed out by Fahidy.(3]
One of the criticisms concerning linearization of a nonlinear
model is related to the change in the error structure of the
dependent variable,t3'7] since if the variance of the original
variable is constant along the experimental conditions, the
variance transformed variable (usually the logarithm or re-
ciprocal of the original variable) is not constant, invalidating
the use of the least squares function. In fact, nonlinear models

Carine Messias Gongalves received her B.S.
in chemical engineering in 2011 at the Federal
University of Santa Maria, Brazil. Her interests
are mathematical modeling and parameter
estimation of chemical processes.

Marcio Schwaab
earned chemical
engineering de-
grees from State
University of Mar-
inga (B.S., 2002),
and Federal Uni- .J
versity of Rio de Janeiro (M.Sc. 2005, D.Sc.
2007). He is a professor in the Chemical Engi-
neering Department and Process Engineering
Graduate Program of the Federal University of
Santa Maria. His research interests are statisti-
cal analysis of experimental data, mathematical
modeling, catalysis, and reaction kinetics.
Josd Carlos Pinto earned degrees in chemi-
cal engineering from the Federal University of
Bahia (B.S., 1985) and the Federal University
of Rio de Janeiro (M.Sc. 1987 and Ph.D. 1991).
He is a professor in the Chemical Engineering
Graduate Program of Federal University of Rio
de Janeiro and is a permanent professor of the
Graduate Program in Chemistry of the Military
Engineering Institute. His focus is on general
chemical reactors, with particular emphasis in the area of modeling, simula-
tion, and control of polymerization systems.
Copyright ChE Division ofASEE 2013


Vol. 47, No. 3, Summer 2013









are linearized not because of the structure of the error, but to
avoid difficulties in numerically obtaining nonlinear least-
squares estimates, and to permit graphing a straight line, as
pointed out by Harrison and Katti.181
Furthermore, it is not only the heteroscedasticity of the
variance of the dependent variable that invalidates the use of
the least squares function. According to Bard,g91 in the 19th
Century Gauss had already observed that minimization of the
least squares function leads to maximization of the probability
of finding the experimental results, when the experimental
errors are normally distributed around the values predicted
by the mathematical model. That is, the least squares func-
tion should only be used when the experimental errors of the
measured data are distributed normally. This observation is
also in accordance with the maximum likelihood principle. [9, 10]
For this reason, it is important to determine the probability
function distribution of the experimental errors, in order to
guarantee the statistical basis of the parameter estimation pro-
cedure. Unfortunately, this does not constitute an easy task, as
it is usually necessary to obtain a large number of replicates,
making the experimentation time and cost very high.
In our opinion, in order to consolidate nonlinear regression
in the scientific literature, it is necessary to begin with the
insertion of statistically based procedures in the undergraduate
courses of chemical engineering, as pointed out by Fahidy.131
And as put forth by Harley,[111 chemical engineering and
statistics are sufficiently successful professions that they
can support further development of statistical procedures for
experimental data analysis.
To analyze and illustrate this important parameter estima-
tion problem, in this paper the estimation of the heat transfer
coefficient is performed in a simple heat transfer laboratory
experiment of the undergraduate course. The experiment
analyzed in the present work is simple, fast, and can be
performed without any significant costs, allowing for execu-
tion of the replicates and permitting a statistical evaluation
of the proposed parameter estimation procedures, based on
the linearized and nonlinear models. The estimation of the
convective heat transfer coefficient was performed with and
without the linearization of the original nonlinear mathemati-
cal model. Obtained results are analyzed and compared to
each other with the help of statistical tools in order to show
the very significant differences of final parameter estimates
and respective confidence regions obtained with the two
analyzed procedures.

METHODOLOGY
Experimental
In this section, the experimental procedure usually per-
formed by undergraduate students in the sixth semester of
chemical engineering is presented. The experimental proce-
dure consists of placing a cylindrical aluminum piece, initially


maintained at 33 C, inside a water bath with a volume equal
to 4 L, kept at 87 C, through a heated jacket. The cylindri-
cal piece has height of 5.0 cm and diameter of 3.0 cm. Some
relevant physical properties of aluminum are121: thermal
conductivity of 236 W.m-'.K-1; density of 2702 kg.m3; and
heat capacity of 896 J.kg-1.K-1. After immersion in the water
bath, the temperature of the metal piece is measured with a
K-type thermocouple at its center. As time goes on, the test-
piece temperature increases and eventually becomes equal to
the bath temperature. A chronometer is used to register the
time. The time length of the experiments was equal to 180 s,
although only the first 100 s were considered for parameter
estimation (which was sufficient for the test piece to reach
the bath temperature). The procedure was repeated 16 times
in order to provide enough data for the statistical analysis.
Since it is a simple experiment and many more replications
could be easily performed, the number of replications was
kept as low as possible while still allowing observation of
differences in the probability distributions of the nonlinear
and linear procedures, since executing a very high number of
replications is time and/or cost prohibitive in a general case.
Mathematical Development
Temperatures inside the test piece are assumed to be ho-
mogeneous. This assumption is supported by the fact that
heat transfer by conduction in a small metal piece is much
faster than heat transfer by convection from the liquid phase
to the metal surface. The validity of this assumption can be
formally verified with the Biot number, which represents the
ratio between the external heat transfer resistance (fluid) and
the internal heat transfer resistance (solid), as
h-L
Bi=- (1)
k
where Bi is the Biot number, h is the convective heat trans-
fer coefficient (fluid), k is the thermal conductivity of the
solid, and L is a characteristic dimension of the solid (for
instance, the ratio between the volume and surface area of the
cylindrical aluminum piece). According to the heat transfer
textbook of Kreith and Bohn,P'2J when Bi is smaller than 0.1,
the convective heat transport can be regarded as very small,
as compared to the thermal conduction in the solid, so that
the heat transfer is controlled by convection. In these cases,
it becomes reasonable to assume that the temperature profiles
are homogeneous inside the solid, as the heat transported
through the surface is quickly distributed throughout the solid
volume. It is important to notice that for all estimated values
of h the validity of this assumption was verified through Biot
number computation.
Further, due to high water bath volume, compared with
the volume of the cylindrical aluminum piece, the decrease
in the temperature of the water bath due to the insertion of a
cold piece inside it is lower than 0.1 C, and can be clearly
disregarded.


Chemical Engineering Education










90

80 o 0

70
0' 0
0
.60 6

50

40

30
0 20 40 60 80 100
t(s)



0
6 0 0
09


2 0
00
0-0
2 0


0 20 40 60 80 100
t (s)

Figure 1. Measured experimental T, values (A) and
computed "experimental" -ln(p ) values (B) as a function
of time.

Based on the previous remarks, the energy balance of the
solid can be described as:
dT
p.V.C. -==-h.A.(T-Ts ) (2)
dt
where p is the solid density, V is the solid volume, Cp is the
heat capacity and T. is the solid temperature, h is the heat
transfer coefficient, A is the external surface area of the solid
piece, and T is the fluid temperature far from the solid piece.
Solution of Eq. (2) is presented in Eq. (3), where Ts0 is the
initial solid temperature and a is defined in Eq. (4).
T, = T + (Ts0 T ). exp (-a. t) (3)
h.A
h=--- (4)


Eq. (3) can be converted into a linear form presented in Eq.
(5), where Vp is a dimensionless variable defined in Eq. (6).
-ln(W)= a.t (5)
V=(T -T)/(T.-T) (6)

The mathematical models described by Eqs. (3) and (5) are


exactly the same, but lead to different parameter estimation
problems. For the mathematical model written in the linear
form [Eq. (5)], the proposed parameter estimation procedure
is based on a simple linear regression that minimizes the least
squares function:

S = iln("V)-ln(W, )J (7)
i=1

where i indicates the experimental measurement at time t, N is
the number of measurements, "e" represents the experimental
values and "m" represents the values predicted by the model.
As the model is linear, minimization of Eq. (7) leads to an
analytical solution, and the value of a can be calculated as:
N
i-ti-ln(In)
= iN (8)
J(ti)2
i-i

This is the procedure commonly used to obtain the value of
the parameter h [which is calculated from the value of a, using
Eq. (4)]. Nevertheless, the minimization of the least squares
function defined in Eq. (7) takes into consideration the square
of the differences between the values of-ln(W), which is not
the real experimental measurement. The goal of the parameter
estimation procedure should be finding the parameter values
that allow for good prediction of the observed experimental
variables. In this problem the value observed experimentally
is the solid temperature T. and not -ln('p) So the least squares
function should be written as:
N
S= (T;,-sI ) ) (9)
i=l

and the nonlinear form of the model, as defined in Eq. (3),
should be used. In this case, it is not possible to obtain an
analytical solution for a (and consequently h) and the use of
a numerical procedure becomes necessary if one intends to
obtain the point of minimum of Eq. (9).
It must be noted that, as the relationship between Ts and -
ln(p) is nonlinear, the minimization of the sum of the squares
defined in Eq. (7) leads to a result that is different from the
result obtained when minimization of the sum of the squares
defined in Eq. (9) is performed. The selection between the
mathematical form of the model, linearized or nonlinear, and
thereafter, between the form of the least squares function, us-
ing Eq. (7) or Eq. (9), should be made on statistical grounds.

RESULTS
Statistical analysis of data
Figures 1A and 1B presents the experimental data obtained
for 16 replicates.
These data were easily obtained by an undergraduate


Vol. 47, No. 3, Summer 2013









student, following the experimental procedure usually em-
ployed during the laboratory lessons. As one can observe,
experimental measurements are scattered to some extent,
as always. Data points were collected every 10 seconds for
statistical analyses.
An initial contradiction appears when comparing the
experimental errors between the measured variable T, and
the transformed variable -ln(ip), as shown in Figures 1A
and IB. The real experimental data Ts seem to be more
scattered at the beginning of the experiment in Figure 1A;
as time increases, data fluctuation decreases significantly,
since the system achieves the thermal equilibrium and the
only source of fluctuation is the thermocouple measurement
noise, since fluctuations due the reproducibility of the heat
transfer experiment are only important in the transient phase
of the experiment. The more important error source in the
beginning of the experiment is related to disturbance of the
static bath due to insertion of the cylindrical body. Although
the body was always inserted carefully into the static bath, it
is virtually impossible to reproduce this procedure and it is for
this reason that the experimental data is more scattered at low
time values. Otherwise, the variable -ln(i) seems to be more
scattered at the end of the experiment, even when the system
achieves the thermal equilibrium, as shown in Figure lB. The
behavior of the experimental fluctuations can be explained
with the help of standard error propagation analysis. When
a variable y is calculated as a function of a second variable
x, which is subject to measurement errors, the variance of y
can be calculated as13'9'101:

SayG2 +(V2 (10)
,(x)= (ax

where the first term on the right-hand side accounts for the
variability ofx and the second term accounts for the variability
of the y measurement device. Considering the nonlinear model
defined in Eq. (3), which describes how Ts changes as a func-
tion of time t, the total variance of Ts can be calculated as:

a' (t)=[(Tso- -T)aexp(-at)T +o<, (11)

Eq. (11) makes clear that, as time increases, the variance
of Ts diminishes exponentially and approaches the variance
of the measurement device o0M as shown in Figure 1A. On
the other hand, in order to analyze the variance of-ln(ip), it
is necessary to acknowledge that:
-ln(v)=-ln[(T, -T) / (T0 -T)] (12)

so that:

-,()(Ts)=fTs _1T jy (13)


As o' approaches a constant value thermocouplee noise)
and TS approaches T when the time increases, the total vari-
and Ts approaches T when the time increases, the total vari-


TABLE 1
Results of Shapiro-Wilk normality test.
Time (s) p-value for -ln(VI) p-value for Ts
0 0.0646 0.9190
10 0.0319 0.6729
20 0.0019 0.0786
30 0.0026 0.1234
40 0.0217 0.2854
50 0.0009 0.7872
60 0.0020 0.3442
70 0.0000 0.3384
80 0.1352 0.3741
90 0.0970 0.2232
100 0.5706 0.5606

ance of -ln(p) increases continuously with time (and goes
to infinity!), as shown in Figure lB. Eqs. (11) and (13) are
very interesting because they explain the behavior observed
in Figure 1A and 1B and also prove that the original and
the transformed variables can present completely distinct
statistical behavior.
The available experimental data were then analyzed to
determine whether the normal distribution could be used to
represent the fluctuations of experimental measurements.
To do that, the normality test of Shapiro-Wilks1131 was used.
This normality test was easily performed with the help of
the Statitica software,1141 which is available for students of
the undergraduate course of the Federal University of Santa
Maria, Brazil.
This normality test consists of determining if the normal
distribution can be used to represent the distribution of data
points in a particular data set, where measurement conditions
are assumed to be the same and the data points represent the
same experimental system. Sixteen replicates were carried out
as similar experiments and both Ts and -ln(p) were submit-
ted to the Shapiro-Wilks normality test. Obtained results are
presented in Table 1 and Figures 2A to 2H. Table 1 reports the
p-values obtained when the Shapiro-Wilks test is performed.
When the p-value is lower than 0.05 (in the case of a 95%
confidence level) the test indicates that the data cannot be
represented by the normal distribution.


Figure 2, facing page. Histograms and
normal distribution fits
for-ln(p) and Ts
at: t =10 s for A and B;
t =20 s for C and D;
t =40 s for E and F;
t =70 s for G andH.


Chemical Engineering Education















U,/
C



O3
(2
0
3


E=I
0
.2

z
0

U1




0



03

0
z_


-In(y)


52 54 56 58 60 62 64 66 68 70 72 74
Ts (C)


1
0
71 72 73 74 75 76 77 78 79 80
Ts(C)

4 .


2.3 2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5
-In(ig)


82.2 82.6 83.0 83.4
Ts(C)


33.


8 84.2 84.6


3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0
-In(V)


U-

2
0
CU

.0
0

a)
"5



.Q
E
z
0


85.3 85.5 85.7 85.9 86.1 86.3 86.5 86.7
Ts (C)


Vol. 47, No. 3, Summer 2013 16


0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
-In(V)


12
CO
o 10

8
4)
0 6
Q6

-4

E 2
0
0


c



7



4
a
2
1


0










TABLE 2
Results of linear and nonlinear estimation of parameter h.
Parameter Estimates Deviation Average Devial
Exper- h (kW/nm2.K) (%) ofT,
iment Linear Non- Linear No
linear lin
1 0.8031 1.2040 33.29 2.135 1.
2 0.8641 1.1443 24.48 1.529 0.
3 0.8227 0.9669 14.92 1.068 0.
4 0.8077 1.1087 27.15 1.761 0.
5 0.7696 1.0952 29.73 2.002 1 .
6 0.7423 0.9594 22.62 1.712 0.
7 0.8302 1.0595 21.65 1.560 0.
8 0.8275 1.1753 29.59 1.882 0.
9 1.0671 1.6725 36.20 1.632 0.:
10 0.8359 1.0303 18.87 1.244 0.,
11 0.8629 0.9729 11.31 0.718 0.:
12 0.8620 0.9896 12.89 0.798 0.,
13 0.9187 1.0474 12.28 0.742 02.
14 0.7856 0.9084 13.51 0.950 0.:
15 0.8203 0.9059 9.45 0.694 0.:
16 0.9696 0.9699 0.03 0.784 0.'
Mean 0.8493 1.0756 19.87 1.326 0.


According to the p-values reported in Table 1 for variable
-ln('ip), only the data collected at sample times equal to 0,80,
90, and 100 s could be represented by the normal distribution.
In other words, the assumption of normal distribution of er-
ror measurements was not adequate for the majority of the
observed data. It is important to note that the points where the
normal assumption could be accepted were the ones placed in
the beginning of the experiment (time equal to 0 s), when the
heat transfer process had not started, and in the final part of
the experiment (times equal to 80,90, and 100 s), when the
heat transfer process had reached the equilibrium condition.
During most of the dynamic trajectory, the fluctuations of
the variable -ln(|p) could not be represented by the normal
distribution, since the p-values were lower than 0.05. On the
other hand, for variable Ts, fluctuations could be described by
the normal distribution during the whole dynamic trajectories,
since p-values of the Shapiro-Wilk test were always higher
than 0.05. This clearly shows that the nonlinear transformation
of Ts to- ln(Vp) changed the statistical significance of the ana-
lyzed measured data. Figures 2A to 2H show that the normal
distribution fits to the obtained experimental replicates, used
to perform the Shapiro-Wilk test. It can be seen that variable
transformation caused magnification of some deviations from
the average, also causing the formation of an asymmetric
tailored distribution that cannot be well represented by the
normal model.


Parameter estimation
According to the Shapiro-Wilk normality test,
ion measurement fluctuations of variable Ts can be
adequately described by the normal distribution,
n- validating the use of the proposed nonlinear least
ear squares function, as defined in Eq. (9). On the other
191 hand, the Shapiro-Wilk normality test indicated that
540 the least squares function should not be used for the
217 linear fitting. Despite that, both linear and nonlinear
approaches were used for estimation of parameter h
908 in order to allow for comparison of final results. It
D027 must be emphasized that normality tests are almost
598 never performed to support the use of least squares
585 procedures in real problems, which can be regarded
734 as a fundamental statistical weakness of most report-
920 ed parameter estimation works. (As a matter of fact,
--- as the variances of experimental measurements were
.440
not constant, more rigorous maximum-likelihood
316 procedures should be used to perform the estimation
t17 of model parameters.T7,1"11 This was not done here in
567 order to keep the presentation simple and because
397 least-squares estimators are used more frequently to
184 solve estimation problems.)
783 The linear parameter estimation procedure in-
5 volved the use of the linear model defined in Eq.
614
S (5) and the objective function defined in Eq. (7).
The nonlinear parameter estimation made use of the
nonlinear model defined in Eq. (3) and the objective
function defined in Eq. (9). Both linear and nonlinear regres-
sions were performed with the help of the Statistica software.
[141 Table 2 shows the estimated parameters for each experiment
for both linear and nonlinear procedures and also the relative
deviations between the two obtained values (using the nonlinear
result as reference). It can be observed that the average devia-
tion between the two estimated parameters was close to 20 %,
which can be regarded as a very significant difference. The
highest difference was close to 36 % in Experiment 9, although
in Experiment 16 the deviation between both parameter values
was close to zero. In order to analyze why differences can be
sometimes small and other times high, Figures 3A to 3D show
linear and nonlinear fits for data obtained in Experiments 9 and
16. Table 2 also shows the average deviation between measured
and calculated Ts values. In all experiments the average Ts
deviations were smaller when the nonlinear procedure was ad-
opted, showing the better prediction capability of the nonlinear
procedure, as Ts was the real measured variable.
Comparing the fits presented in Figures 3A to 3D, it can
be observed that for both linear and nonlinear approaches
the quality of the fit is better when Experiment 16 is consid-
ered. To show this more clearly, Figures 4A and 4b show the
residuals of the fits, obtained as the differences between the
experimental and predicted solid temperatures. In Figure 4A,
the fit obtained with the nonlinear procedure is much better


Chemical Engineering Education

























t (s)


0 20 40 60 80 100
t (s)


0 20 40 60 80 100 0 20 40 60 80 100
t (s) t (s)
Figure 3. Linear and nonlinear fits for Experiments 9 (A and B) and 16 (C and D).


Figure 4. Residuals between experimental and predicted temperatures for the linear fit (circles) and
nonlinear fit (crosses) in Experiments 9 (A) and 16 (B).


than the one obtained with the linear procedure. In Figure 4B,
however, both fits are very similar, and the residuals are much
smaller in Figure 4B than in Figure 4A. It can be concluded
that both linear and nonlinear procedures lead to similar
parameter estimates when the fits are very good; however, if
the data presents some larger deviations that the model cannot
predict with great precision, linear and nonlinear procedures
can lead to very different parameter estimates.
Vol. 47, No. 3, Summer 2013


It also must be pointed out that when undergraduate students
encounter a result such as the one obtained in Experiment 16
they conclude that it does not matter if a linear or nonlinear
regression is used, since the parameter estimate will be the
same. And as the linear regression can be performed easily, it
should be preferred. After showing the results for the sixteen
runs and the normality tests, however, the undergraduate
students can straightforwardly conclude that nonlinear regres-


4
0
9
6
0
3
+ 0
0--


p


20 40 60 80 100 0 20 40 60 80 100


t (s)


t(s)










sion must be used, avoiding model manipulation. At any rate,
with the purpose of establishing the importance of a proper
statistical analysis among undergraduate students, the follow-
ing section is devoted to the computation of the parameter
estimate deviations, showing students that together with its
value, the parameter estimate must be always presented with
its uncertainty, which assures the quality of the parameter
estimate values.
Uncertainties of parameter estimates
Besides the analysis of the parameter values, it is also
important to analyze the parameter uncertainties. As the
experimental data are subject to measurement noise, the pa-
rameter values are also uncertain to some extent.9 "'01 In order
to calculate the confidence intervals of the obtained parameter
estimates, the following equation can be used'10'15'161:

S '(h)=SH(1+N-N FI,- (14)

where S is the objective function [Eqs. (7) or (9)], h is the
estimated parameter value, NP is the number of estimated
parameters (equal to 1 in this case), N is the number of experi-
mental points (equal to 10 in this case), Fp. NNp is the Fisher
probability distribution function with NP and N-NP degrees
of freedom, and a is the confidence level. As a confidence
level of 95% was assumed, Fp NNP is equal to 4.9646 and
Eq. (14) can be rewritten for the analyzed problem as:

S(h)=l 1.49646.-S(hi) (15)

Determination of the confidence intervals of parameter h
requires determining the values of h that satisfy Eq. (15),
remembering different objective function and h values were
found for each particular data set. Table 3 presents the confi-
dence limits of parameter h for each experiment.
It can be observed that the lower limit of parameter h esti-
mated with the nonlinear procedure, except for Experiment
16, was always higher than the upper limit of parameter h
estimated with the linear procedure, confirming that nonlinear
and linear procedures generally lead to different parameter es-
timates. Further, although the estimated parameter values were
similar for linear and nonlinear procedures in Experiment 16,
the obtained confidence intervals were very different, showing
that the results obtained with both procedures were not the
same. It must also be noticed that the confidence intervals
were not symmetric in the nonlinear case, meaning that the
estimated parameter values were not placed at the center of
the confidence interval and indicating that the uncertainties of
the parameter estimates did not follow a normal distribution.

CONCLUSION
In this paper a comparison was made between linear and
nonlinear procedures used for regression of the convective
heat transfer coefficient in a heat transfer problem, where


TABLE 3
Confidence limits of the parameter h
for each experiment.
Linear Nonlinear
Exp -- N
ho. h pp hlow hpp
1 0.7484 0.8579 1.0587 1.3835
2 0.8137 0.9145 1.0829 1.2112
3 0.7784 0.8669 0.9493 0.9851
4 0.7600 0.8553 1.0112 1.2217
5 0.7225 0.8167 0.9871 1.2228
6 0.6916 0.7932 0.9151 1.0070
7 0.7596 0.9007 1.0007 1.1237
8 0.7751 0.8800 1.0887 1.2732
9 1.0056 1.1284 1.4676 1.9306
10 0.7992 0.8727 0.9901 1.0731
11 0.8423 0.8835 0.9470 1.0001
12 0.8428 0.8814 0.9515 1.0298
13 0.8908 0.9468 0.9874 1.1132
14 0.7610 0.8103 0.8773 0.9412
15 0.7917 0.8490 0.8923 0.9198
16 0.9355 1.0036 0.8750 1.0810
Mean 0.8074 0.8913 1.0587 1.3835

a metal piece, initially at temperature Ts0, was placed in a
thermostatic bath, kept at temperature T.. This procedure
consists of an experiment usually found in many laboratory
classes in the chemical engineering courses all over the world.
Application of the Shapiro-Wilk test showed that assump-
tion of normal fluctuation was valid for the measured solid
temperature values, Ts, but not for the transformed variable,
ln('p), showing that variable transformation can change the
statistical behavior of analyzed variables and make the use of
least squares estimation procedures inappropriate. Analyses
of 16 independent experiments showed that parameter values
could present bias of up to 36%, when estimates obtained
through regression of the transformed variable and of the
originally measured temperatures were compared to each
other. Further, analysis of model prediction residues showed
that the nonlinear procedure could lead to much better rep-
resentation of the measured Ts values. Particularly when the
model deviations were small, fits obtained with the linear
and nonlinear procedures were similar, although differences
became evident when deviations increased because of the
unavoidable measurement noise. Finally, it was also shown
that the uncertainties of the obtained parameter estimates were
sensitive to the adopted regression procedure, indicating that
variable transformation should not be adopted when rigorous
statistical analysis of obtained estimates is sought.
The proposed experiment, which can be easily performed
in most chemical engineering laboratories, illustrates very
Chemical Engineering Education










clearly the large number of problems that can arise when
linearization of model structures is introduced into the quan-
titative analysis of measured data. It can be concluded that
the use of linear regression procedures should be avoided
when they are not supported by rigorous statistical arguments,
since variable transformation can lead to biased and errone-
ous estimation of parameter values and respective parameter
accuracies, as described by the confidence regions of the
parameter estimates.
It is also important to report that this work has been used
in the preparatory lessons for the laboratory experiment and
assists new students in understanding the importance of the
statistically based analysis of data. Most importantly, they
can readily conclude that, despite the slightly higher compu-
tational effort, nonlinear regression must be preferred against
the use linearized version of the nonlinear models.
Finally, besides learning the concepts regarding heat trans-
fer process using a procedure described in this work, the
undergraduate students become readily familiar with some
statistical procedures and statistical software, which can help
their future investigations as engineers or researchers.

ACKNOWLEDGMENT
The authors thank FIPE/UFSM for providing scholarships
and for supporting this work.

REFERENCES
1. Machuca-Herrera, J.O., "Nonlinear curve fitting using spreadsheets,"


J. Chem. Ed., 74,448 (1997)
2. Denton, P., "Analysis of first-order kinetics using Microsoft Excel
Solver," J. Chem. Ed., 77,1524 (2000)
3. Fahidy, T.Z., "An undergraduate course in applied probability and
statistics," Chem. Eng. Ed., 36,170 (2002)
4. Chen, N.H., and R. Aris, "Determination of Arrhenius constants by
linear and nonlinear fitting," AIChE J., 38,626 (1992)
5. Sundberg, R., "Statistical aspects on fitting the Arrhenius equation,"
Chemom. Intell. Lab. Syst., 41,249 (1998)
6. Klicka, R., and L. KubAcek, "Statistical properties of linearization of
the Arrhenius equation via the logarithmic transformation," Chemom.
Intell. Lab. Syst., 39,69 (1997)
7. Kumar, K.B., and S. Sivanisan, "Prediction of optimum sorption
isotherm: Comparison of linear and non-linear method," J. Hazard.
Mater., 126,198 (2005)
8. Harrison, F., and S.K. Katti, "Hazards of linearization of Langmuir's
model," Chemom. Intell. Lab. Syst., 9,249 (1990)
9. Bard, Y., Nonlinear Parameter Estimation, Academic Press,New York
(1974)
10. Schwaab, M., and J.C. Pinto, Andlise de Dados Experimentais. Vol 1:
Fundamentos deEstatistica e Estimafao de Pardmetros; e-Papers, Rio
de Janeiro (2007)
11. Hartley, H.O.,"Statistics as a science and as a profession," J.Am. Stat.
Assoc., 75 1(1980)
12. Kreith, F., and M.S. Bohn, Principles of Heat Transfer, 5th ed.; PWS
Publishing Company; Boston (1997)
13. Thode Jr., H.C., Testing for Normality, Marcel Dekker Inc., New York,
(2002)
14. StatSoft, Inc., STATISTICA (data analysis software system), version
8.0. (2007)
15. Bates, D.M., and D.G. Watts, Nonlinear Regression Analysis and its
Applications, Wiley, New York (1988)
16. Schwaab, M., E.C. Biscaia Jr., J.L. Monteiro, and J.C. Pinto, "Nonlinear
parameter estimation through particle swarm optimization," Chem.
Eng. Sci., 63 1542 (2008) C


Vol. 47, No. 3, Summer 2013










] class and home problems j


CONTINUOUS FEED AND BLEED

ULTRAFILTRATION

A Demonstration of the Advantages of the

Modular Approach for Modeling Multi-Stage Processes




MICHAEL B. CUTLIP1 AND MORDECHAI SHACHAM2
1 University of Connecticut Storrs, CT 06269
2 Ben-Gurion University of the Negev Beer-Sheva 84105, Israel


Multiple stage operations are widely used in sepa-
ration processes. There are two commonly used
approaches for modeling of multi-stage processes.
The "simultaneous" approach involves modeling of the
complete process (which includes all the stages) as one large
set of equations. This approach is widely used for modeling
of distillation columns and its main advantage is the high
computational efficiency. In the "sequential modular" ap-
proach a model of a single stage (module) is prepared and
tested separately. The complete multi-stage process is then
constructed by tying together several modules by means of
the material and energy flows between them. The advantages
of this approach are that the model building is more straight-
forward; that the models are easier to construct, to follow, and
to debug; and that the computer code can actually serve as
problem documentation. Thus this approach is more adequate
for educational use than the simultaneous approach.
An example presented by Foley11] that concerns the design
of a multi-stage ultrafiltration unit operated in a feed and
bleed mode will be used to demonstrate the advantages of
the "sequential modular" approach in the solution of various


types of problems. This example is suitable for courses in
separation processes, introduction to modeling and computa-
tion, and process and product design.
Michael B. Cutlip is professor emeritus of
the Chemical, Materials, and Biomolecular
Engineering Department at the University of
Connecticut and has served as department
head and director of the university's Honors
Program. He has B.Ch.E. and M.S. degrees
from Ohio State and a Ph.D. from the Univer-
sity of Colorado. His current interests include
the development of general software for
numerical problem solving and application
to chemical and biochemical engineering.
Mordechai Shacham is professor emeritus of
the Department of Chemical Engineering at the
Ben-Gurion University of the Negev in Israel.
He received his B. Sc. and D.Sc. degrees from
tthe Technion, Israel, Institute of Technology. His
research interests include analysis, modeling
and regression of data, applied numerical
methods, and prediction and consistency
analysis of physical properties.


Copyright ChE Division ofASEE 2013
Chemical Engineering Education


The object of this column is to enhance our readers' collections of interesting and novel
problems in chemical engineering. We request problems that can be used to motivate student
learning by presenting a particular principle in a new light, can be assigned as novel home
problems, are suited for a collaborative learning environment, or demonstrate a cutting-edge
application or principle. Manuscripts should not exceed 14 double-spaced pages and should be
accompanied by the originals of any figures or photographs. Please submit them to Dr. Daina
Briedis (e-mail: briedis@egr.msu.edu), Department of Chemical Engineering and Materials
Science, Michigan State University, East Lansing, MI 48824-1226.












Figure 1. Schemat-
ic plot of the Multi-
stage Ultrafiltration
System.


PROBLEM BACKGROUND
Continuous concentration of a protein solution by
multi-stage feed and bleed ultrafiltration
A typical 3-stage feed-and-bleed ultrafiltration process is
shown in Figure 1. Fresh feed with a volumetric flow rate of
Q0 (m'/s) and solute concentration of co (g/L) is mixed with
retentate (product stream) recycle from the first stage and
enters the first membrane module. Some of the solvent (and
possibly some of the solute) passes the membrane and exits
the unit as permeate filtratee). The concentrated product stream
(with solid concentration of c, g/L) is partially recycled in
order to increase the flow rate into the unit to ensure well-
mixed conditions. The product (bleed) stream from the first
unit is fed into the second stage with a volumetric flow rate
of Q1 (m3/s).
Detailed discussion of the multi-stage ultrafiltration pro-
cess, including the associated design equations, is provided,
for example, by Seader and Henley.121 Here the assumptions
suggested by FoleyM" are used for simplification of the model.
These assumptions are that 1) no solute passes the membrane
(complete rejection), and 2) the gel polarization model ap-
plies. Thus the membrane is operating at the limiting flux (j,
m/s) given by

j=klncA (1)
Ci

where k is the mass transfer coefficient (m/s), cg is the limiting
or "gel" concentration (g/L), and ci is the solute concentration
in stage i. The solute balance on stage i (assuming complete
rejection) yields
Qici =Qiici_ (2)

The total balance on the feed, retentate, and permeate
streams of stage i can be written
f, =Q,.,-Q,-jA, (3)

where A. is the membrane area in stage i.
Eqs. (1), (2), and (3) represent a complete model of stage
i. At the solution f must vanish (f = 0). One of the variables
associated with stage i (say ci, A, etc.) can be selected as
unknown and values for the rest must be specified.


The sequential modular approach requires building the
model of a single unit so that it can calculate the "output"
variables (Q., c., and the permeate flow rate) if the input
variables (Q,- and c,_,) and the design parameters (k, cg, and
A,) are specified. One possibility to carry out this calculation
is to solve Eq. (2) for Q, yielding
Q =Qilci_ /ci (2A)

Introducing Eqs. (1) and (2A) into Eq. 3, and replacing
the input variables and design parameters by their numeri-
cal values, yields a single nonlinear algebraic equation that
can be solved for the unknown output variable cr. The other
output variable Q, can be consequently calculated directly by
solving Eq. (2A).
If all the input variables and design parameters are specified
and the output variables need to be calculated, the problem
is categorized as a "simulation problem." In a "design prob-
lem" some of the output variables are specified and the same
number of input variables and/or design parameters need to
be determined, so that the specification regarding the output
variables is met. Adding an objective function that contains
input variables and/or design parameters that need to be mini-
mized to meet an economic objective, subject to constraints
related to the output variables, constitutes an "optimization
problem."
The different parts of the following problem statement are
related to these three types of problems.

PROBLEM STATEMENT
The assignments and numerical data presented by Foley"'
are considered. A multistage feed-and-bleed ultrafiltration
unit is used for concentrating a protein solution. Fresh feed
enters the first stage at the rate of Qo = 1 L/min with solute
(protein) concentration of co = 10 g/L. Complete rejection can
be assumed for the protein. The membrane is operating at the
limiting flux with the mass transfer coefficient: k = 3.5.10.6
m/s and the gel concentration cg = 300 g/L.
a) Given that the membrane area in the first stage is A, =
2.7 m2, calculate the product's outlet concentration: c,
and flow rate: Qfrom this stage.
b) A 3-stage system with equal membrane areas of A =0.9
m2 is used to separate the protein solution. Calculate the


Vol. 47, No. 3, Summer 2013


Recycle 1 Recycle 2 Recycle 3 Final
-- f -- f i RePtentate
Aj i- 1 c, c c2 c3 33
Cfi C1 1 2^ C -2 C2 2 3^ C3&


Permeate 1 Permeate 2 Permeate 3 o
jlA1 jiA2 J_3A3










product's outlet concentrations and flow rates for the
three stages (a simulation problem).
c) Find the total membrane area of a three-stage ultrafiltra-
tion system that yields retentate concentration leaving
the third stage of 100 g/L assuming equal membrane
areas for the three stages (a design problem).
d) Repeat question (c) but this time allow for different
membrane areas for the different stages so as to obtain
the same final concentration but minimizing the total
membrane area (an optimization problem).

PROBLEM SOLUTION
Modeling a single stage using the sequential
modular approach
Modeling a single stage can proceed following the al-
gorithm outlined in the problem background section. Sev-
eral available software packages can be used for solving the
nonlinear algebraic equation of the single-stage model. The
POLYMATH13] software package was used for this purpose.
The input of part (a) of the problem into the POLYMATH
software package is shown in Table 1. The POLYMATH
program includes the code and comments (text that starts
with the "#" sign and ends with the end of the line). The row
numbers shown in Table 1 are not part of the program; they
were added as references for the explanations that follow.
In the POLYMATH program, the equations and data are
grouped into "model equations" [lines 1-4, Eqs. (1), (3), and
(2A)], problem-specific data, including units (lines 6-11,
including the input variable and design parameter values),
and initial estimates for the unknown exit concentration. The

TABLE 1
POLYMATH program for the solution of
Part (a) of the assignment
No. Equation/ # Comment
1 #Model equations
2 j=k*ln(cg/cl ) #Membrane Flux
3 Ql=c0*QO/cl #Complete Rejection of Protein
4 f(cl) = Q0-Ql-j*A#Overall Material Balance
5
6 # Problem specific data
7 Q0=l/(60* 1000) #mA3/s
8 k=3.5e-6 #m/s
9 cO=10 # g/L
10 A=2.7 # mA2
11 cg =300 # g/L
12
13 # Initial estimates
14 cl(min)=20
15 cl(max)=100


POLYMATH software automatically orders the equations
prior to solution. The solution with POLYMATH using this
equation set was c, = 66.93 g/L and Q, = 0.15 L/min.
The program shown in Table 1 provides a clear, precise, and
complete documentation of the problem and its mathemati-
cal model. Observe that there is actually no need to combine
the three basic equations into one equation and this provides
an easier-to-follow problem documentation. Combining the
equations can also be a source of errors.
Modeling the three-stage system using the
sequential modular approach
The model used for the first stage can be used for the sub-
sequent stages except that the outlet variables of the earlier
stage become the inlet variables of the subsequent one and
the design parameters need to be updated, if necessary. Thus,
modeling the operation of three consecutive stages of the ultra-
filtration system [Part (b) of the problem statement] involves
writing the model equations that were used for the first stage,


TABLE 2
POLYMATH program for the solution of Part (b) of the
assignment
No. Equation/ # Comment
1 #Model equations
2 #First Stage
3 jl=k*ln(cg/cl) #Membrane Flux
4 Ql =cO*QO/c 1 #Complete Rejection of Protein
5 f(c1) = Q0-Q1-j 1*A #Overall Material Balance
6 #Second Stage
7 j2=k*ln(cg/c2) #Membrane Flux
8 Q2=c *Q 1 /c2 #Complete Rejection of Protein
9 f(c2)=Ql-Q2-j2*A#Overall Material Balance
10 #Third Stage
11 j3=k*ln(cg/c3) #Membrane Flux
12 Q3=c2*Q2/c3 #Complete Rejection of Protein
13 f(c3)=Q2-Q3-j3*A#Overall Material Balance
14
15 #Problem specific data
16 Q0=l/(60* 1000)#mA3/s
17 k=3.5e-6 #m/s
18 A=0.9 # m^A2
19 c0=10 # g/L
20 cg =300 # g/L
21
22 # Initial estimates
23 cl(0)=20
24 c2(0)=50
25 c3(0)=170


Chemical Engineering Education










TABLE 3
Results summary for Parts (b), (c), and (d) of the
____________multi-stage ultrafiltration problem
Part (b) _____ Part (c) _____ Part (d)
Stage c Q, A, c, Q, A C Qi
No. (g/L) (m3/is) (m2) (g/L) (m3/s) (m2) (g/L) (i3/s)
1 20.34 0.492 0.712 17.4 0.575 0.947 22.14 0.452
2 56.07 0.178 0.712 37.7 0.265 0.505 49.73 0.201
3 163.45 0.061 0.712 100 0.100 0.567 100 0.100
Total 2.136 2.019


A B C D E F
1 POLYMATH NLE Migration Document
2 Variable Value PolymathEquation Comments
3 Explicit Eqs ji 9.1218E-06 j1l=k'ln(cg/cl) Membrane Flux
4 Q1 7.5266E-06 Q1=cO*QO/cl Complete Rejection of Protein
5 j2 6.2903E-06 12=k "1nicg c2i Membrane Flux
6 02 3.3516E-06 Q2=cl *Q1/c2 Complete Rejection of Protein
7 j3 3.8451 E-06 j3=k'ln(cg/c3) Membrane Flux
8 03 1.6667E-06 :3=c2Q*02/c3 Complete Rejection of Protein
9 At 2.01539045 At=A..IA2+A3 Objective function
10 00QO 1.6667E-05 Q0=1/(60"1000) mA3/s
11 k 0.0000035 k=3.5e-6 m/s
12 cO 10 c0=10 giL
13 c3 100 c3=o100 gAL
14 cg 300 cg=300 g/L
15 A1l 0.94676198 Al=0.7
16 'A2 0.50474175 A2=-O.7
17 Implicit Vars cl 22.1436941 c1(0)--20 Overall Material Balance
18 c2 49.7272272 c2(0)=50 Overall Meterial Balance
19 A3 0.56388672 A3(0)=1 Overall Material Balance
20 Implicit Eqs f(cl) 5.0389E-07 f(c1)=QO-Q--j1'A1
21 f(c2) 1E-06 f(c2)=Ql-Q2-j2*A2
22 f(A3) -4.833E-07 IA3i)02-Q3-j3*A3
23 Sum of Squares: I 1.4875E-12 F=f(cl)A2 +f(c2)A2+f(A3)A2


Figure 2. Excel Worksheet for the solution of Part (d) of the multi-stage ultrafiltration problem.


for the second and third stage while changing the indices of
the inlet and outlet streams and introducing equal A = 0.9
nm2 membrane areas as design parameters. The POLYMATH
program prepared according to these principles is shown in
Table 2 and the results are presented in Table 3. Observe
that (as was pointed out by Foley[U) the final concentrate in
this case is c3 = 163.45 g/L, which is much higher than the
concentration reached by a single stage unit of the same total
membrane area [in Part (a), c1 = 66.93 g/L].
The model shown in Table 2 can be easily modified to solve
Part (c) of the assignment. In this part the exit concentration
of the protein from the third stage (c3) is specified and it is
required to calculate the membrane areas in the three stages,
assuming equal areas (A). There is no need to modify the
model equations in Table 2 except to change the status of
c3 to a specified design parameter (instead of unknown) and
to include the unique A value as one of the unknowns. The
results of the computation for Part (c) are shown also in Table


3. Observe that the total area
required to reach c3 = 100
g/L is A, = 2.136 m2.
These examples demon-
strate the flexibility provid-
ed by the sequential-modu-
lar approach to investigate
various design alternatives.
Minimizing the total
membrane area of the
ultrafiltration system
In part (d), it is requested
to minimize the total area
of the multi-stage system
with a pre-specified c3 value
while allowing different
areas A, A2, and A3 for the
three stages. This prob-
lem can be formulated as a
constrained minimization
problem to minimize A,
= A1 + A2 + A3 subject to
the constraints f(c1) = 0,
f(c2) = 0,and f(A3) = 0. The
variables are Ap A2' A3 cl,
and c2.
Only minor changes need
to be introduced in the pro-
gram shown in Table 2
to accommodate this new
problem formulation. As the
POLYMATH package does
not solve constrained non-


linear optimization prob-
lems, the program should be solved with a software package
that includes tools for solving such problems. POLYMATH
can automatically export programs to MATLAB141 or Excel151
that can solve a constrained minimization problem. The Excel
solution will be shown here.
In Figure 2 the Excel worksheet- as obtained by exporting
the POLYMATH program to Excel-is shown. Note that the
Excel formulas of the various equations are in column C. The
additional information, generated by the POLYMATH export
routine, is textual and serves as documentation. The names
of the variables, as defined in the POLYMATH program, are
shown in column B. The original POLYMATH equations are
displayed in column E and the associated comments are given
in column F. The Excel "Solver" Add-In is used for finding the
minimal area. The following information is used to specify the
"Solver" parameters; the minimum is sought for "Target cell"
C9 (At) by changing cells C 15 through C19 (A1,A, 3, c, and
c2), subject to the constraints C20 = 0, C21 = 0, and C22 = 0


Vol. 47, No. 3, Summer 2013










[which forces the residuals of the nonlinear equations given by
f(cl), f(c2), and f(A3) to be zero]. The optimal solution found
is shown in Table 3 and Figure 2. At the minimum, At = 2.015
m2, which is slightly lower than the total area required (A, =
2.136 im2) when stages with equal areas are used to achieve
the same final concentration.
Solution approach presented by Foley
To highlight the educational advantages of the "sequential
modular" approach for solving problems that involve staged
processes, the solutions provided here can be compared with
the solution techniques used by Foley."l' According to his ap-
proach, the equations representing a single stage [Eqs. (1),
(2), and (3)] are brought into the form of a single nonlinear
algebraic equation containing two unknowns, xi.- and x,
where x. = c0/ci.

Q0(xi.-xi)-ln CS-lnxi =0 (4)
kA C0

This model is inconsistent with the "sequential modular"
approach as it includes the input variables of the first stage
(co and Q0) in the models of all the subsequent stages. While
the definition of the unknown x1 used in this equation was
essential for graphical solution of the ultrafiltration problems,
it is absolutely unnecessary for numerical solution. Keeping


TABLE 4
MATLAB function (model) for the solution of part (d) of the
ultrafiltration problem
No. Command % Comment
1 function fx = MNLEfun(x);
2 cl = x(1);
3 c2 = x(2);
4 A3 =x(3);
5 cg =300; %g/L
6 cO = 10; %g/L
7 k =.0000035; %m/s
8 QO = 1 / (60 1000); %mA3/s
9 jl = k log(cg / cl); %Membrane Flux
10 c3= 100; %g/L
11 A2 = .7 %mA2
12 Q1 = cO QO / cl; %mA3/s, Complete Rejection of Protein
13 j2 = k log(cg / c2); %Membrane Flux
14 Q2 = cl Q1 / c2; %mA3/s, Complete Rejection of Protein
15 Q3 = c2 Q2 / c3; %mA3/s, Complete Rejection of Protein
16 j3 = k log(cg / c3); %Membrane Flux
17 A1 = .7 %mA2
18 fx(1,1) = QO Q1 (jl A1); %Overall Material Balance
19 fx(2,1) = Ql Q2 (j2 A2); %Overall Material Balance
20 fx(3,1) = Q2 Q3 (j3 A3); %Overall Material Balance


the original variables associated with a particular stage makes
it much easier to understand, to interpret, and to modify the
model in order to fit the various problem types (design, op-
timization, etc.). Furthermore, the model equations cannot
provide full documentation of the problem as some of the
original variables (Q, and c) do not appear in them.

USING THE EXAMPLE TO DEMONSTRATE
GOOD PROGRAMMING PRACTICES
Software packages such as POLYMATH and Excel are very
convenient tools for problem solving, but there are more com-
plex tasks that may require programming. One such complex
assignment can be optimization of the membrane areas of the
multi-stage system for different final concentration: c3 values
parametricc runs). In this section the membrane area optimiza-
tion assignment (d) is carried out for various c3 values: c3 =
50,60 ... 150 g/L and the resultant optimal areas are plotted
vs. c3. Such parametric optimization runs can be carried out
with Excel by manually changing the parameter values. This
approach is inefficient and cumbersome, however, particu-
larly for problems where there are many parameters and a
wide range of parameter values to be considered. In such
cases, programming is required for repetitive solution of the
problem with the various parameter values. One option is to
carry out the parametric runs efficiently using MATLAB. The
development of the MATLAB program can serve for
demonstration of good programming practices.
The MATLAB function representing the operation
of the multistage ultrafiltration unit can be auto-
matically and efficiently generated by POLYMATH
(Table 4). The function is named MNLEfun. It accepts
C, C2, and A3 as input parameters and returns the val-
ues of f(c), f(c2), and f(A3) to the calling program.
Note that MATLAB requires input of the variable
values into the function in a single array (x, in this
case), and return of the function values in a single
array (fx, lines 18-20 in Table 4). The variable values
are put back into variables with the same names as
used in the POLYMATH model (lines 2-5) to make
the MATLAB code more meaningful. POLYMATH
orders the equations sequentially as required by
MATLAB and converts any needed intrinsic func-
tions and logical expressions.
The function in Table 4 contains several variables
(cg, CO, c3, Q0 A1, and A2) to which constant numeri-
cal values are assigned. Assigning numerical values
to variables inside functions is considered poor
programming practice as it limits the use of the func-
tion to one particular problem with just one set of
parameter values. To enable general use of the func-
tion the numerical values of these variables must be
passed to the function by the program that calls this
function. One possibility to pass variable values to a

Chemical Engineering Education










function is by defining these variables as "global" variables.
Unlike "local" variables, which are separate for the different
functions and the main program, a single copy of a "global"
variable is shared by all of them. The use of "global" variables
is not considered good programming practice, however, as
it overrides the hierarchical structure of the program. This
means that change of a global variable in a lower-hierarchy-
level function may cause unforeseen changes in higher-level
functions or in the main program. Good programming practice
requires the passing of the numerical values of constants as
input parameters to the function.
In Table 5 the revised form of the MNLEfun is shown.
Observe that in this version all the numerical values of the
variables are passed through one array, named parm. Good
programming practice requires introducing the numerical
values into the original variables (see lines 5 to 11 in Table 5)
so that the original forms of the equations (lines 12 through
20) can serve as clear documentation of the ultafiltration
system model.
The MATLAB multiple variable minimization function:
fmininsearch combined with the nonlinear equation solver func-
tion: fsolve can be used to find the minimal membrane area
configuration for various c3 values. Thefininsearch function
is called with the following parameters:
[Aopt,TArea,exitflag] = fminsearch(@minAAlA2,[],parm);
The input parameters are: minA is the name of the function
that calculates the objective function At = A1+ A2 + A3 value,
AIA2 is an array containing the current values ofAl and A2,
and parm is the same array of the parameter values that is
used in the MNLEfun function (Table 5). The output param-
eters are: Aopt is an array that contains the optimal values
of A, and A2 TArea contains the optimal value of the total
membrane area, and exioflag indicates whether a minimum has
been found (exitflag = 1) or the search has been terminated
for other reasons (exitflag # 1).
The function minA is shown in Table 6. This function
obtains the values of AIA2 and parm from the fininsearch
function and returns Asum (A1). The minA function passes the
current A 1 and A2 values to the MNLEfun function (through
the parm array) and uses the nonlinear equation solver func-
tionfsolve to solve the system of nonlinear equations: f(c1) =
0, f(c2) = 0, and f(A3) = 0. This is necessary in order to find
the A3 value that satisfies the constraints with the current set
of Al andA2 values.
The complete MATLAB program can be downloaded from
the ftp site: . The
calculated optimal areas are plotted vs. the exit concentration
c3 in Figure 3 (next page). As expected, higher outlet concen-
trations require larger membrane areas.
A similar example involving development of a MATLAB
program for modeling of imperfect mixing in a Chemostat
that involves the use of minimization and nonlinear equation

Vol. 47, No. 3, Summer 2013


TABLE 5
Generalized form of the MATLAB function (model) of
the ultrafiltration system
No. Command % Comment
1 function fx = MNLEfun(x,parm);
2 cl =x(l);
3 c2 = x(2);
4 A3 = x(3);
5 cg =parm(1); %g/L
6 cO = parm(2); %g/L
7 k = parm(3); % m/s
8 QO = parm(4); %mA3/s
9 c3 = parm(5); % m/s
10 Al = parm(9);%mA3
11 A2 = parm(10);% mA3
12 j I = k log(cg / c 1); %Membrane Flux
13 Q1 = cO QO / cl; %m^A3/s, Complete Rejection of
Protein
14 j2 = k log(cg / c2); %Membrane Flux
15 Q2 = cl Ql / c2;%mA3/s, Complete Rejection of
Protein
16 Q3 = c2 Q2 / c3; %m^3/s, Complete Rejection of
Protein
17 j3 = k log(cg / c3); %Membrane Flux
18 fx(l,1) = (QO Q1 (jl Al)); %Overall Material
Balance
19 fx(2,1) = Q1 Q2 (j2 A2); %Overall Material Bal-
ance
20 fx(3,I) = Q2 Q3 (j3 A3); %Overall Material Bal-
ance


TABLE 6
MATLAB Function for calculating the total membrane
area
No. Command % Comment
1 function Asum=minA(A 1A2,parm)
2 cl=parmnn(6); c2=parmnn(7); A3= parm(8);
3 xguess = [cl c2A3];
4 Al= A1A2(1);A2 =A1A2(2);
5 parm(9)=Al; parm(10)=A2;
6 options = optimset('Display',['off'],'TolFun',[le-
____ 14],'TolX',[le-14]);
7 xsolv=fsolve(@MNLEfunxguess,options,parm);
8 A3=xsolv(3);
9 Asum=Al+A2+A3;


solver functions can be found in Cutlip, et. al.J61 The example
presented there can also be used for demonstration of good
programming practices.











CONCLUSIONS
The example presented here provides an opportunity to
practice several aspects of modeling and design of multi-
stage processes
Using a consistent "sequential modular" approach for
modeling the single units.
Solving problems of increasing levels of difficulty-simu-
lation, design, and optimization-while selecting the
most effective software tool for numerical solution of the
problem at hand.

Building the model and the computer input so that they
can serve as clear and complete documentation of the
problem and its solution.
Using advanced tools available for solving nonlinear
algebraic equations and optimization problems.


REFERENCES
1. Foley, G., "Solution of Nonlinear Algebraic Equations in the Analysis,
Design, and Optimization of Continuous Ultrafiltration," Chem. Eng.
Ed., 45(1), 59 (2011)
2. Seader, J.D., and EJ. Henley, Separation Process Principles, 2nd Ed,
New York, Wiley (2006)
3. POLYMATH is a product of Polymath Software, polymath-software.com>


0.50

0.00


Exit concentration c3 (g/L)

Figure 3. Plot of membrane areas vs. required
exit concentration.



4. MATLAB is a product of MathWorks, Inc., comn>
5. Excel is a registered trademark of the Microsoft Corporation, www.microsoft.com>
6. Cutlip, M.B., N. Brauner, and M. Shacham, "Biokinetic Modeling of
Imperfect Mixing in a Chemostat -an Example of Multiscale Model-
ing," Chem. Eng. Ed., 43(3), 243 (2009) 0


Chemical Engineering Education













IN MEMORIAL



DONALD ROBERT WOODS
(April 17, 1935 April 26. 2013)
Don Woods. Professor Emeritus at MNcMaster University and former member of CEE's Pub-
licauons Board. died April 26, 2013. He was 78 sears old. Don's beloved fanuly includes his
% ife of 52 years. Diane: his children, Russell Glen (predeceased), Suzanna Lynn Peters (Denis
Dallaire i. and C ntihia Jane Veals (Scott); and five grandsons, Caleb, Marcus, and Andrew Veals
and Nicholas and Benjamin Peters. I!
Don was a chemical engineering professor at McMaster University from 1964-2000 where he
used innoatii e teaching methods and \von manN teaching aards (and three honorary doctorates,
from Queens, Guelph, and NlMcMNlasterl. In the words of Phil Wood, Associate Vice President (Student Affairs) & Dean
of Students. "'Don "as the greatest educator in MNcMaster's histon "
In 1986. Don %as named to the inaugural cohort of 3M National Teaching Fellow s. He is perhaps most % idely kno" n
as a pioneer of McMNaster's distinctive learning strategies: inquiry and problem-based learning, as well as a recognized
expert on teaching and learning within n the engineering academic community. He was author/coauthor of more than a
dozen books including Probht m-inhbased Learning and was on the editorial board of The lIntermtional Journal oj PBL and
The Journal of General Education. He edited the newsletter Problem Solving News for 20 years and wrote a column.
Deeloping Problem Solh ing Skills, in the Journal of College Science Teaching for 10 years.
During his career and well into retirement, Don gave more than 500 v, orkshops on effect% e teaching and process skill
development, problem-based learning, and motivating and rewarding teachers to improve student learning Befitting
his research focus, he was a regular and valued contributor to Chemical Engineering Education.
Ti's re'pori '1L crtnpihJ iri'i 't'lituinrv notices prepared b, tht Cociddihi S.t,:icr fir Teaching and Lt uriinv in Hi-ticr EJw,' tii ,i d b,r
thic tjiiii 'i : ..







A remembrancefromin CEE Associate Editor Phil Wankat .. .
Don I\ods., one of i'the great originals of engineering education, has passed on. My first imipri,.ssion of Don was of
eIergy-h'he itas a fierce of nature. But I learntcd thliat he was much more than that. Underneath the energy Donrwas a
I'ery caring. person \feho believed in his students. Everyone who attended one of Don's workshops was a student, and
Don did his absolute best to reach and teach every student whether there were four or four hundred. Because Don be-
lieved I wias a letter, more capable, teacher ithin I ihiongth- I was, I was able to become that teacher. He had theii oeIer
t tinprove people by believing in them-great teachers do that. He was interested in everYone tiildentl. piolessor, and
janitor. Probably because hlie was so helpful Don w''as able 1c cractioulv accept litlp and that o.vinuron, conti tuciiie
criticism. f'om others. Don received honors durtting his' career, butin not othlier's that hlie ouid have recciived. When I
ad.ked about one. it was clear that Don was liht' by this lack of recognition, bit I never heard liim say a mean or cruel
11 ord about an 1'0nC.
The morning I heard that Don had died I cried and did not want to have to withstand the pain. But Don's enduring
message is that the pain of reaching out, caring, and teaching is worth it. :


Vol. 47, No. 3, Summer 2013 177










Random Thoughts...


SPEAKING OF EVERYTHING III


RICHARD M. FIELDER
North Carolina State University


he intuitive mind is a sacred gift and the rational mind is
a faithful servant. We have created a society that honors
the servant and has forgotten the gift. Albert Einstein
Man is rated the highest animal, at least among all animals
that returned the questionnaire. Robert Brault
The only man I know who behaves sensibly is my tailor: he
takes my measurements anew each time he sees me. The rest
go on with their old measurements and expect me to fit them.
George Bernard Shaw
Fanaticism consists in redoubling your effort when you have
forgotten your aim. George Santayana
The cardiologist's diet: If it tastes good, spit it out.
Source unknown
I am, and ever will be, a white-socks, pocket-protector, nerdy
engineer, born under the second law of thermodynamics,
steeped in steam tables, in love with free-body diagrams,
transformed by Laplace and propelled by compressible flow.
Neil Armstrong
A conference is a gathering of important people who singly
can do nothing but together can decide that nothing can be
done. Fred Allen
Find a job you love and you'll never have to work a day in
your life. Confucius
Never ruin an apology with an excuse. Kimberly Johnson
Why do they put Braille on the drive-through bank machines?
George Carlin
It is a tremendous act of violence to begin anything. I am not
able to begin. I simply skip what should be the beginning.
Rainer Maria Rilke
A conclusion is the place where you got tired of thinking.
Steven Wright
As to the Seven Deadly Sins, I deplore Pride, Wrath, Lust,
Envy, and Greed. Gluttony and Sloth I pretty much plan my
day around. Robert Brault
Your assumptions are your windows on the world. Scrub
them off every once in a while, or the light won't come in.
Isaac Asimov
You cannot truly listen to anyone and do anything else at the
same time. M. Scott Peck


Thank you for sending me a copy of your book; I'll waste no
time reading it. Moses Hadas
It's hard to be religious when certain people are never inciner-
ated by bolts of lightning. Bill Watterson
One does not discover new lands without consenting to lose
sight of the shore for a very long time. Andre' Gide
I cannot see how to refute the arguments for the subjectivity
of ethical values, but I find myself incapable of believing that
all that is wrong with wanton cruelty is that I don't like it.
* Bertrand Russell
We should be careful to get out of an experience only the
wisdom that is in it-and stop there-lest we be like the cat
that sits down on a hot stove-lid. She will never sit down on
a hot stove-lid again-and that is well; but she will also never
sit down on a cold one anymore. Mark Twain
Seek simplicity, and distrust it. Alfred North Whitehead
How we spend our days is, of course, how we spend our lives.
* Annie Dillard
I have only three enemies. My favorite enemy, the one most
easily influenced for the better, is the British nation. My
second enemy, the Indian people, is far more difficult. But
my most formidable opponent is a man named Mohandas K.
Gandhi. With him, I seem to have very little influence.
* Mohandas Gandhi
Whether you think you can or think you can't, you're right.
* Henry Ford
Protons have mass? I didn't even know they were Catholic.
* Source unknown
Life is like driving a car at night. You never see further than
your headlights, but you can make the whole trip that way.
* E.L. Doctorow
No snowflake in an avalanche ever feels responsible.
* Stanislaus Jerzy Lee
His mother should have thrown him away and kept the stork.
* Mae West
I look back on my life like a good day's work. It was done
and I am satisfied with it. Grandma Moses
I knew if I stayed around long enough, something like this
would happen. George Bernard Shaw's epitaph


Copyright ChE Division of ASEE 2013


Chemical Engineering Education










[1M laboratory
------------------


REMOTE LABS AND GAME-BASED

LEARNING FOR PROCESS CONTROL


IMRAN A. ZUALKERNAN, GHALEB A. HUSSEINI, KEVIN F. LOUGHLIN, JAMSHAID G. MOHEBZADA, AND
MOATAZ EL GAML
American University of Sharjah Sharjah, UAE


emote laboratories appeared in higher education al-
most two decades ago. Since then, the infrastructure
for building remote laboratories has come a long way
and stand-alone and commercial tools such as MATLAB and
LabVIEW are easily integrated with off-the-shelf learning
management systems (LMS) to build comprehensive remote-
laboratory learning environments.11-31 Shared social network-
ing platforms such as Facebook have a potential to take remote
laboratories to yet another level. A recent survey shows that
a significant amount of research in remote laboratories has
focused on comparing remote laboratories against hands-on
and virtual laboratories.141 For example, Tzafestas et al.151
show a comparison between students trained in a traditional
way on robots as opposed to using a virtual laboratory or a
remote laboratory, and observed no statistical differences in
performance due to the modality of delivery. Each modality
tends to emphasize different educational objectives.1[41 In
addition to conceptual understanding, hands-on laboratories
have historically emphasized design, professional, and social
skills. Virtual laboratories, on the other hand, have focused
primarily on professional and conceptual skills while remote
laboratories have mostly addressed professional skills and
conceptual understanding.
In some sense, prior research on remote laboratories has
centered on the amplification and attenuation effects of intro-
ducing the remote laboratory technology.161 The amplification
effect is represented by positive impacts of this technology
such as round-the-clock remote access. For example, surveys
in specific fields such as RF have been conducted recently
to identify scarce laboratory equipment most suited for such
amplification.171 The attenuation effects, on the other hand,
represent negative impacts of introducing this technology.
For example, the slow response time is one such attenuation
effect mediated by using high-speed Internet to provide real-
time video from a remote laboratory.181 The predominance


of these two factors is also reflected in assessments models.
For example, Nickerson et al.91 present a detailed model for
comparing hands-on laboratories, remote laboratories, and
simulated laboratories. This model contains criteria such as
purpose of the experiment, the experimental and the coordi-
nation interface, and laboratory frame and technology. The
only pedagogically related criterion in this model, however,
is the individual differences between learners. This paper
takes the view that remote laboratories are more than simple
surrogates for real or virtual laboratories and can be used to
explore forward-looking learning designs such as game-based
learning. The rest of the paper is organized as follows. First,
a brief introduction to game-based and mobile learning (m-
learning) is presented. This is followed by a description of
a learning problem in chemical engineering. A game-based
remote laboratory to address this learning problem is pre-
sented next. This is followed by a pilot study to validate the
learning design.

Imran A Zualkernan is with the Department of Computer Science and
Engineering at the American University of Sharjah, Sharjah, UAE. Prof.
Zualkeman received a B.S.in 1983 anda Ph.D.in 1991 from the University
of Minnesota, Minneapolis. Research interests include advanced learning
technologies, IT services management, agile processes, and Six Sigma.
Ghaleb A. Husseini is with the Department of Chemical Engineering at
the American University of Sharjah, Sharjah, UAE. Prof. Husseini received
a B.S. in1995, an M. Eng. Mgmt. in 1997, and a Ph.D. in 2001, all from
Brigham Young University in Provo, UT Research Interests include drug
delivery and biomaterials and modeling of biological processes.
Kevin F Loughlin is with the Department of Chemical Engineering,
American University of Sharjah, Sharjah, UAE He received his B.E in
1965 and M.Eng.Sc. in 1970, both from the National University of Ireland
and his Ph.D.in 1978 from the University of New Brunswick, Fredericton,
Canada. His research interests include adsorption equilibria, kinetics
technology, and modeling of processes.
Jamshaid G. Mohebzada is with the Department of Computer Science
and Engineering, American University of Sharjah, Sharjah, UAE.
Moataz El Gaml is with the Department of Computer Science and En-
gineering, American University of Sharjah, Sharjah, UAE.


Copyright ChE Division ofASEE 2013


Vol. 47, No. 3, Summer 2013









GAME-BASED LEARNING
What constitutes a game has many definitions. For example,
PrenskyE101 defines game as organized play. After analyz-
ing many definitions of a game, Salen and Zimmerman(11
define a game to be "a system in which players engage in
artificial conflict, defined by rules, that results in a quantifi-
able outcome," (pp. 80). We extend this definition to define
instructional games to be a system where learners engage in
artificial conflict, defined by rules, that results in quantifiable
outcome and enhances learning.
While Aldricht121 and Gibson et al.t1131 have outlined various
principles of digital game-based learning, recent studies114,
have shown that game-based learning actually improved
student motivation as well as performance. There is also
emerging evidenceE151 that when using game-based learning as
opposed to self-learning, the students not only thought that the
game was the preferred method and was more enjoyable, but
also they showed willingness to continue to use this method
of learning. Tuzun et al.'161 show that not only did children do
better while using game-based learning, but playing games
also developed them as independent learners. Game-based
learning has also been used successfully in higher education.
For example, Ebner and Holzinger[171 show that performance
of students using game-based learning was better than the
control group in the context of civil engineering.
M-learning has also received much attention recently.18'191
For example, Akkerman et al.1201 show that students learned
history by walking around with their mobile phones and by
sharing information about what they saw in the city. There are
also efforts to move existing gaming platforms to m-learning
platforms'211 Mobile game-based learning has
also been used in the context of providing mass
learning in developing countries.1221 M-leam-
ing, to extend a traditional LMS, also received
a positive response from college students.?31
Podcasting was also found to be more effective
for revision of lectures than notes or other con-
ventional means.1241 M-learning has also been
shown to significantly increase environmental
awareness.1251 A framework for designing m-
learning games has been proposed.[261 Specific
criteria for assessing the quality of learning
games have also been proposed.J271 Issues
addressing the quality of m-learning are also
outlined in Gafni.J281
Alternative pedagogical approaches in-
cluding using competition games have been
recently used to teach robotics in remote labo-
ratories.J291 Using a gaming engine as an inter- s A-
face to a remote laboratory in a conventional
pedagogical context has also been explored.
[30] Such studies are more an exception than a
rule, however. The relative absence of game-


Figure 1. Block diagram for PI controller for a water tank.


Figure 2. General effect of K and K. on the setpoint.
p


Figure 3. Physical setup for the remote lab game.


Chemical Engineering Education


ErrorSignal (e) Actuation Signal (u)
Differencing / /
Amplifier / O/
Set-Pont + P1 /e Waterr Level)
(Desired Controller Tank
Water Level) -- --

Feedback Loop

PI Controller -- Kp
fe1 T- + u
e -^- ( --9-+


Under Damped


Over Damped

Increasing Kp










based learning in remote laboratories can be attributed to the
fact that infrastructure is a pre-requisite for experimentation
with novel pedagogical paradigms like game-based learning.
Since such infrastructure is now widely available or easy to
build, another explanation can perhaps be that most teachers
and researchers are digital immigrants who did not grow up
with the Internet and gaming technologies and consequently
hold a completely different viewpoint, while most students
are digital natives of the gaming generation.1101

THE LEARNING PROBLEM
The learning problem being addressed in this paper is how to
teach proportional integral (PI) controllers in an undergraduate
chemical engineering course on control systems. Using remote
laboratories to teach control applications is not new.13121 The
water tank equipment has been typically used to demonstrate a
PI controller. This equipment consists of a tank of water that is
connected to a water reservoir. There is a constant drainage of
water from the tank into the reservoir. The control problem is
to maintain a constant water level in the tank by continuously
pumping water back into the tank from the same reservoir.
This is done by providing a square-wave input to a controller
that sends a voltage signal to an electric pump to maintain a
particular level of water. Figure 1 shows a PI controller for
the water tank equipment. Setpoint represents a desired water
level in the water tank. Actuator signal is the voltage applied to
the water pump. The output process variable is the actual water
level. Difference between the desired and actual water level
is represented by an error signal. The error signal is fed into
the PI controller to continuously calculate the voltage being
applied to the pump. A PI controller does so by multiplying
the error with a constant proportional constant (Kp) and with
an integral term represented by the sum of error accumulated
so far multiplied with an integral constant (K1). The control
loop in Figure 1 can be described by a transfer function, and
a time-domain solution can be derived as shown in Eq. (1).


Figure 3a. Schematic for Proportional + Integral Control.


Sh'(t)= -e sin tI
c3 l1-; L 3!

*-[C-CS(t-tw)+CS(t-2tw)-CS(T-3tw)+-.. (1)

C and tw are the height and width of the input square wave,
respectively. Also,

K3 K (2)






wKKe (4)


where K is the controller gain and K1 is the integral or reset
time. K and K1 are constants that include the process gain, the
gain on the valve, and the gain on the measurement sensor.
The values of KP and K, determine an appropriate behavior for
the closed-loop system. As Figure 2 shows, an ideal behavior
for the water level is represented by the near critically damped
curve where the water level oscillates initially but settles down
to the setpoint quickly. An under-damped behavior means that
the water level will fluctuate around the desired level and take
a long time to settle down. Finally, as Figure 2 shows, when
the system is over-damped, it takes a long time to reach the
desired water level. The area under each curve on both sides of
the setpoint represents the total accumulated error. A smaller
error typically corresponds to a better controller.

THE REMOTE LABORATORY GAME
A key consideration in the design of any control system
is to quickly bring the system to the desired setpoint and to
maintain it there. This remote laboratory game has two learn-
ing goals or objectives. The first goal is to learn how to throw
a system into oscillations and the second goal is to stabilize
or tune the system and bring it back to the setpoint.
Architecture
Figure 3 shows the physical setup for the remote laboratory
game and Figure 3a shows the schematic of our process. The
water tank equipment (CE 105 from QT- Quasar Technologies,
Oslo, Norway) consists of two identical water tanks with the
corresponding water pumps and a reservoir. Each tank has a
flow valve that can adjust the amount of water draining into
the reservoir. The pump for each tank is controlled through
a voltage provided through the digital interface. The digital
interface is connected to a personal computer (PC) through a
serial port. Figure 4 (next page) shows the software architec-
ture. Ajava program (1180 lines of Java code) was written to
communicate with the microcontroller in the digital interface


Vol. 47, No. 3, Summer 2013

























Figure 4. Software architecture for the remote lab game.


Figure 5. The Mobile User Interface.


through the serial port and to implement the PI-controller on
the PC. In addition, the same Java program also communi-
cates with an Apache server-based PHP/MySQL application
(1487 lines ofPHP code) that implements a game interface by
communicating with the popular Facebook social networking
site through the Internet. The Java application is also able to
send SMS messages through a Siemens MC-35 SMS modem
connected to another serial port on the PC. The Java server
consists of various classes providing interfaces to each of the
hardware components. For example, CTHardwarelnterface
class implements the binary communication protocol between
the PC and the digital controller running over the serial port.
Similarly, the HTTPlnterface class is responsible for process-
ing new inputs from Facebook that come through the Apache
HTTP server and forwarding these control parameters to the
equipment using an instance of the CTHardwarelnterface
class. The HTTPlnterface class is also responsible for re-
trieving the current values of various tank parameters (such
as the water level) using the CTHardwarelnterface class and


updating these in the MySQL database. The SMSSender class
communicates with a Siemens MC35i SMS modem to send
alerts to the game players. A ControlPanel class provides a
user-friendly GUI for the server and allows a user to start
and stop the server and to display the current readings of the
equipment for diagnostic purposes. Instances of each class run
concurrently using Java's native multi-threading capabilities.
For example, instances of SMSSender and HTTPInterface
class each run in their own thread. Since the architecture uses
Apache server's native HTTP concurrency-handling mecha-
nism in addition to Java threads, the number of concurrent
players at one point is limited only by the memory and pro-
cessing limitations of the Apache server and the Java virtual
machine. The MySQL database records the complete history
of all the changes made by each team along with time stamps.
A learner who wishes to play the remote laboratory game
can access the game by logging into Facebook. It is possible
to create a new game tied to a particular water tank or to join
an existing game. Two roles are provided to a player. A player
can either join as an invader or as a defender. The objective
of a defender is to change the values of K, and K to bring the
water level in the tank to a particular setpoint. An invader, on
the other hand, tries to destabilize the system by throwing the
system into oscillations. The game continues until the person
who created the game decides to stop it.
Typically, two teams play the game. One team is charged
with defending one tank and attacking the other. Another team
does the converse; attacking the tank being defended by the
first team, and defending the tank being attacked by the first
team. Each player is allocated a fixed number of changes to
either of the two parameters (KI and K.) after which they are
not able to make any changes until the end of the game. Within
these limitations, each player is allowed to make changes at
any time. This means that players can request changes to the
parameters of any tank in a concurrent fashion; the requested
changes are queued and are actually applied to the system in
a first-come-first-served fashion. Each player can view the
behavior of the system the player is attacking or defending in
the form of a visual display. In addition, each player sees the
total error the player has accumulated on a particular system
indicating whether the player is winning or losing the game.
The computer connected to the water-column instrument runs
a customized Java-based web-server (using server sockets) that
allows the Facebook web application to connect as a socket
client. This server uses Java's native multi-threading and syn-
chronizing mechanism to ensure that only one client has access
to one water column at a time. A single thread is spawned for
each socket connection from the Facebook application, and the
request to make changes to each water column is queued on the
semaphore being used to access each water column. The change
request is automatically processed once the water column is
released by another client's request (if any).


Chemical Engineering Education










Each player can access the game using his or her mobile
phone or through a normal browser on the Internet. The user
interface (UI) for the game is specifically designed using the
Facebook mobile interface to accommodate small mobile de-
vices. The UI has been deliberately kept simple and functional.
For example, as Figure 5 shows, in addition to showing that
the team is supposed to invade Tank #1 as opposed to defend-
ing it, the UI also details the desired setpoint, the number
of changes allowed, a graphical representation of how the
water level has changed over time and the total system error
accumulated so far. Figure 5 shows how another team can use
the simple UI to change the K and the K, of a tank they are
defending or invading; they are only allowed to change one
parameter at a time. A team member can quickly scroll up
and down to view the same information about each tank the
team is either attacking or defending in a consistent manner.
If the user leaves the game and significant changes appear in
the behavior of either tank, they receive an SMS warning that
a change has occurred. One important aspect of the system is
its ability to capture and show the actions of any player in real
time. Such information can also be used in the post-analysis
of any game to pinpoint conceptual gaps of the students.
Finally, it should be noted that some parameter changes can
take a long time to manifest in the behavior of the system. This
is part of learning how a PI controller may behave in the real
world. Consequently, the game is designed to be played over
a few hours. A limited number of changes force the players
to be deliberate about their choices, however.
Game rule design
The main objective of the game is to teach chemical engi-
neering seniors the principles of proportional integral control
including the objective of feedback, the basic equations of
proportional and integral controllers, the key concepts behind
both proportional and integral control, the general purpose
behind the two controllers, how K and K respond to a process
disturbance, and in addition the disadvantages of proportional
and integral control.
The game rules and regulations were:
1) The period of the contest was 3 hours.
2) Each team consisted of 2 members.
3) 12 students constituted the "control" group.
4) 12 students constituted the "gaming" group.
5) The "control" group consisted of the students who did
not play the game but ran a simple level-control experi-
ment with one tank. The "control" group was then
asked to take the quiz before and after they ran their
experiment.
6) The "gaming" group was divided into 6 teams (two
members in each group). The teams were randomly
selected and named (Team A, Team B, Team C, Team D,
Team E, and Team F).


7) The teams were randomized to see who played who.
8) Each team member needed a mobile phone with SMS
capability.
9) The objective of the game was to minimize the Standard
Square of Error (SSE) between the setpoint and the
control variable over a 3-hour period.
10) The offensive and defensive teams had different objec-
tives. The offensive teams were trying to maximize the
SSE in their opponents'tank. The defensive teams were
trying to minimize the SSE for their own tanks.
11) The teams could control the level by changing K or K.
12) Each team could change K or K1 12 times in the 3-hour
period (6 changes to fix their tank and 6 changes to
disturb the other team's tank).
13) The 12 changes allowed the students to change either
K,or K
14) The minimum and maximum values for K were I and
10, respectively.
15) The minimum and maximum values for K, were 0 and
1.0, respectively.
16) The setpoint change was a square wave input.

Typically any game contains three levels of rules.111 Op-
erational rules are rules that come with a game as a set of
instructions. Constitutive rules define the underlying formal
structure below the surface of a game. These structures can be
mathematical or logical. Finally, implicit rules are unwritten
rules of the game and are concerned with etiquette and good
sportsmanship. These rules can change from game to game.
For example, a child playing chess may be allowed to take
back a move while an adult might not. Table 1 shows the three
types of rules for this game.


TABLE 1
Three Levels of Game Rules
Nature of Rules
Rules
Operational Each team acts as an invader for one tank
Rules and a defender for another.
Each member of the team gets a fixed
number of tries at changing the parameters
of a tank as a defender or an invader.
Only one parameter can be changed at
a time.
The team whose total error for the tank
they are invading is more than the one
they are defending wins.
The game ends in fixed amount of time.
Constitutive These rules are governed by the transfer
Rules function of a PI system which determines the
error that will be accumulated.
Implicit Rules The players will not abandon the game; the
players will not cheat or collaborate; etc.


Vol. 47, No. 3, Summer 2013










TABLE 2
Effectiveness of Game Design
Criteria Comment
Challenge: There should There are multiple ways to
be multiple ways to win the win the game by varying K
game, vary the difficulty of and K1 differently. Constant
the game, sufficient random- feedback in the form of the
ness, and constant feedback total accumulated error and
about performance, its profile is provided.
Curiosity: The activity Since the competing team
should offer sensory stimula- is constantly reacting, there
tion and novelty to stay in the is enough novelty and the
game. error curve provides sensory
stimulation.
Control: The player should This is clearly provided
feel control over the activity when every change to K or
and witness the effects of K1 results in an immediately
making choices, different response from the
system.
Fantasy: The player should Although the game does not
feel involved in the game. have a "surface story," since
there is a real chemical pro-
cess that needs attention, the
players should feel involved.
Interpersonal Motivation: This is a team-based game
The players meet and play and in addition pits students'
with others and earn respect knowledge of control systems
among peers for performance, against each other.


Effectiveness
Malone and Lepper"34I define five criteria for evaluating the
effectiveness of a game. Table 2 shows how each of these
criteria is incorporated in the game design. For example, the
curiosity criterion is satisfied in two ways. First, the team is
left to wonder which parameter the opposite team changed
to explain the current behavior. Secondly, the team is curious
about what impact changing a particular parameter will have
on the system. In either case, the response is novel because it
depends on the current as well as previous states of the system.
Similarly, Dondi and Morettit271 define four classes of criteria
to evaluate a learning game. Pedagogical and context criteria
include target groups, learning objectives, context of usage,
didactic strategy, communication and media, and evaluation
activities. Each of these criteria has been considered in de-
signing the game. For example, the target groups have been
clearly identified as the chemical engineering students taking
the control course. In addition, the two learning objectives
have been formally specified. The instructions for playing the
game are clear and the game is clearly related to the working
context because both system tuning and trying to determine
the causes of a system destabilization are important profes-
sional activities for control engineers. The didactic strategy
has clearly defined roles for attackers and defenders; the rules
are clear and there is a clear coherence between the rules and
the consequences of actions that a learner makes. The user
interface of the game has been kept minimal and simple and


leads to a good quality of interaction between the user and
the game. The evaluation is inherently built into the game in
term of the accumulated error. In other words, if a student
can minimize the error, this is a direct measure of their under-
standing of how to tune the system. Content criteria include
properties of content such as obsolescence and balance for
the target group. In this game, content only consists of game
instructions and as such does not play a major role. Technical
criteria include credits, conformance to standards, and tech-
nical quality issues. The game is robust because it has been
tested for many hours without any errors. In addition, it con-
forms to the Internet standards which mean that any standard
browser on a mobile phone or a laptop can be used to play the
game. The user interface is minimal and functional and the
images of the graph are clear even on small mobile screens.
Finally, for the information-produced category, the game uses
Facebook as the underlying platform, and all passwords and
user information are safely maintained through Facebook. In
addition, the game also saves all the activities including each
action of each player as a history. These reports can easily be
printed by the instructor as desired.

Game design and outcome criteria
It is also instructive to evaluate game design from a prepara-
tion for lifelong learning perspective. The ABET organization
provides one set of criteria that graduating chemical engineer-
ing students must meet on their graduation day. Table 3 shows
how each of these criteria is addressed by the game. As Table 3
shows, the game directly addresses criteria (a)-(e), (g), and (k).

EVALUATION
A pilot study was conducted to evaluate the remote labo-
ratory game. The pilot study had two objectives. The first
objective was to gain an insight into whether the students
would like playing the game. A second objective was to see
if playing the game would make a quantitative difference
in student performance. Unlike previous remote laboratory
studies that have compared a remote laboratory to a virtual or
hands-on laboratory, the objective here was to compare game-
based remote laboratories treatment against a control group
represented by students who had no exposure to this game.

Pilot study design
Students currently taking the CHE 421 Chemical Process
Dynamics and Control class at the American University of
Sharjah were recruited to evaluate the game. The students had
been exposed to the proportional and integral control loop in
classroom lectures before their participation. Twelve students
were chosen at random as the control group while another
12 volunteered to play. There was no statistical difference
between the mean GPA of the control and treatment groups.
The control group had seven women while the treatment group
had five women. The 12 volunteers were randomly divided
into six teams of two students each. Each of the six teams was
Chemical Engineering Education











TABLE 3
ABET Outcome Criteria and Game Design
ABET Criteria Game Design
(a) an ability to apply knowl- The game requires a math-
edge of mathematics, science, ematical interpretation of
and engineering the transfer functions.
(b) an ability to design and The students need to
conduct experiments, as well as conduct mini-experiments
to analyze and interpret data to verify their assumptions
about the parameters of the
system being attacked or
defended.
(c) an ability to design a Tuning or destabilizing
system, component, or process the system is a parameter
to meet desired needs within design problem.
realistic constraints such as
economic, environmental,
social, political, ethical, health
and safety, manufacturability,
and sustainability
(d) an ability to function on Since control is taught
multidisciplinary teams in many engineering
disciplines, the game can
be played by a multidisci-
plinary team.
(e) an ability to identify, System tuning is an
formulate, and solve engineer- engineering problem and is
ing problems in engineering regularly practiced in many
practice, engineering contexts.
(f) an understanding of profes- Not directly addressed in
sional and ethical responsibility the game.
(g) an ability to communicate Team members are playing
effectively from remote locations
and therefore require an
effective communication
strategy.
(h) the broad education neces- Not directly addressed in
sary to understand the impact the game.
of engineering solutions in a
global, economic, environmen-
tal, and societal context
(i) a recognition of the need Not directly addressed in
for, and an ability to engage in the game.
life-long learning
(j) a knowledge of contempo- Not directly addressed in
rary issues the game.
(k) an ability to use the To compete successfully,
techniques, skills, and modem the students need to use the
engineering tools necessary for mathematical knowledge
engineering practice, and tools often used in engi-
neering practice for control
engineers.


randomly assigned a competing team. This resulted in three
gaming contests each consisting of two teams of two students
each. Each contest was run for a total of three hours. All stu-
dents took a pre-quiz on PI controllers before the contests.
Similarly, all students took a post-quiz on PI control after the
contests. One day before the contests, the students playing the
Vol. 47, No. 3, Summer 2013


game were invited to the laboratory where they were shown
the laboratory equipment and given explanation of how it
worked. Twelve hours before the game, each team was e-
mailed a copy of game instructions that included a description
of the operational rules of the game. The students were given
the option of either playing the game on their mobile phone or
on their laptop computers using wireless LAN. After playing
the game, each student was asked to individually fill out a
survey to evaluate various aspects of the game.
Results
Post-game survey
Table 4 shows the post-game survey questions. Table 5 (next
page) shows the results from the survey including a weighted
average (WA) for each question, where 1 means strongly agree
and 5 means strongly disagree. Results in Table 5 indicate
that more than 10 out of 12 students either agreed or strongly
agreed that the game helped with the learning objectives (WA
= 1.33). Most students enjoyed playing the game (WA= 1.13)
and felt that the game was immersive (WA=1.27).
In addition, most students indicated that they would recom-
mend this game to a friend (WA = 1.27). Half the students
felt that the game motivated them to learn more about control
systems while the other half were not so sure (WA = 1.6). None
disagreed with the assertion, however. Out of 12 students,
four were not sure if the game was addictive (WA = 1.67). In
summary, the students generally thought that it helped them
learn the material and was enjoyable.
None of the students noticed the unusual behavior caused
by opening the flow valve. Many students carried the control
textbook with them. In addition, in two out of three contests,
students continued to play with the system even after they had
won the game. When asked about the reason, they typically
indicated that they were curious about how the system would
behave under certain conditions.

Performance
Playing the game once did not have an impact on perfor-
mance of the students. An analysis of variance to compare the
test results from before and after the game for both control

TABLE 4
Post-Game Survey Questions
No. Post-Game Survey Questions
Ql. Playing the game improved my understanding of how to
tune proportional-integral control systems.
Q2. Playing the game improved my understanding of how to
destabilize proportional-integral control systems.
Q3. I felt immersed in the game.
Q4. I enjoyed playing the game.
Q5. The game motivated me to learn about control systems.
Q6. I would recommend this game to a friend.
Q7. The game was addictive.










TABLE 5
Post-Game Survey Results
No. Strongly Agree Neither Disagree Strongly Weighed
Agree Disagree Average
Q1. 6 4 2 0 0 1.33
Q2. 4 7 1 0 0 1.40
Q3. 7 3 2 0 0 1.27
Q4. 9 2 0 1 0 1.13
Q5. 6 1 4 1 0 1.60
Q6. 7 3 2 0 0 1.27
Q7. 3 5 4 0 0 1.67


5.0 UJCL-5.00
X=4.33
2.o5 LC-L 3.66
1 6 31 46 6 26 91 16 131 1F6 15


1 166 331 496 661 826 991 1156 1321 1486 1651
Observation



1 I


4 !


6.0"
S4.5-

S3.0-
0
1.50
0n.


1486 1651


1 166


331 496 661


826 991 1156 1321
Observation


Figure 6. Performance of Tank A in the second

and treatment group showed no statistical difference F(3,44)
= 0.55; p = 0.625. This means that playing the game once did
not directly lead to improving the performance of the students.

Qualitative analysis
One important aspect of the remote laboratory game is the
ability for post-game analysis to determine the gaps in student
understanding. An illustrative analysis of the second game is
presented next. The second game was played between Team-A
and Team-B, both teams composed of two students. Team-A
was defending Tank A while attacking Tank B and Team-B
was defending Tank B and attacking Tank A.
Figure 6 shows the behavior of TankA over the three-hour
period of play. The top graph in Figure 6 shows the actual level
of a tank as a function of time. The bottom graph in Figure
6 shows the moving range of the water level for the last two
time intervals. The moving range shows the level of fluctua-
tion over time; spikes in this graph represent drastic changes


UCL=0.825
M253


in the water level over short periods
of time indicting turbulence or level
instability. The first half of the game
is characterized by the level going
out of control and then coming back
into control for brief periods of time
before it is out of control again.
Team-A was not very conscious of
how they used up the limited amount
of changes to K and Ki and hence
ran out of changes about 70% into
the game. This means that afterward
Team-B had a free reign over chang-
ing the parameters as they liked.
This is reflected in the uncontrolled
variation seen after about two-thirds
of the game. It is interesting to note
that towards the end, Team-B had
brought down the level of tank to
zero, which means that the tank
was accumulating maximum error.
They chose to play with the tank,
however, and made it saturate by
pushing it to the other direction.
This posed no advantage in terms
of the game. A post-interview with
the students indicated that they were
just "playing around" with the op-
ponent's system to see what would
happen. The invaders' median K
p
was 3 (with 95% confidence inter-
vals of [2,5] using Wilcoxon signed
rank test) while the defenders' me-
dian K was 5 (with a 95% interval
p
of [4,6]). Though not statistically
different, this means that invaders


were trying lower-range Kp values to over-dampen the system
and hence build up error while the defenders were trying the
opposite strategy. An optimal strategy, however, would be to
simply use a K of 1.0-indicating that the invaders perhaps
did not fully comprehend what K does. Similarly, the invad-
ers' estimated median for K was 0.75 (with 95% confidence
interval of [0.5,0.95]) while the estimated median for KI for
the defenders was 0.5 (with 95% confidence intervals of [0.45,
0.75]). Clearly, the invading team was trying to destabilize
the system by using high values of K. Why the defending
team was also using reasonably high values, however, tends
to suggest a misunderstanding of how K, works.
Figure 7 shows the behavior of Tank B. As the Figure shows,
Team-B was eventually able to find the values of parameters
to get the water level within control. Team-B was partially
helped by the fact that the Team-A ran out of turns. The invad-
ers' estimated median for K was 5.5 (with a 95% confidence
interval of [5,9.5]). The defenders' estimated median for Kr,
Chemical Engineering Education


10.0' I '-f

7.5-lV










on the other hand, was 7 (with a 95%
confidence interval of [5, 7]). The 10.0
invader's estimated mean for K, was 7 5 I
0.4 (with a 95% confidence interval _
of [0.2, 0.5]). The defender's esti- so -
mated mean for K was 0.25 (with | .
a 95% confidence interval of [0.15,
0.4]). In other words, the defending o.o
team kept trying the middle values 1 166 33
for K while the invading team tried

the defenders used a reasonably low
K, while the invaders also kept the 6
K, below 0.4. Again, this confirms I
Team-A's fundamental misunder-
standing of how K, impacts the 0
system response.

DISCUSSION 1 166 33
While playing the remote game
did not show a statistical increase in Figurt
the performance of students, this is
perhaps expected since the students only had one chance to
play the game. The performance could perhaps be improved if
they were allowed to play the game many times. The post-game
survey results, however, clearly indicated that they enjoyed the
game and that they would recommend it to their friends. This
was confirmed by the excitement shown by students while
playing the game as well. It is interesting to note that half the
students used the mobile phone to receive the SMS messages
but chose to use the laptop to play the game. The other half were
comfortable using a mobile device. The students also indicated
that three hours was too long and thought that the time for the
game should be reduced to one hour only.
A number of improvements can be made in the game. For
example, since the game creator can view the performance of
each team, he or she can easily use Facebook to communicate
with the students to either mentor or ask them why they are
using a particular strategy. Live video showing the tanks and
turbulence of the water would also add to the gaming expe-
rience. Since Facebook mobile currently does not support
video, however, this feature can be included only if the game
is played on a laptop.
One final observation is that despite dealing with off-the-
shelf hardware, currently available technology makes it very
easy to put together a remote laboratory. The challenge lies
in one's ability to utilize this technology in a pedagogically
sound and interesting manner.

CONCLUSION
This paper has shown how remote laboratories can be used
for game-based learning. A game specifically designed to
teach a particular control topic was developed using a sound
Vol. 47, No. 3, Summer 2013


_ UCL=5.86
X=4.97
LCL=4.08


1 496 661 826 991 1156 1321 146
Observation










1 496 661 826 991 1156 1321 1486
Observation


UCL=1.095
Mte 335


S7. Performance of Tank B in the second game.

pedagogical foundation. In addition, the game design was
evaluated against various criteria for what constitutes a good
game. Finally, a pilot study was conducted to further validate
the game design. The pilot study indicates that the game was
well received by the students who enjoyed the game. Pre- and
post-tests comparing this group to a control group, however,
did not show any significant statistical differences.

REFERENCES
1. Rapuano, S., and F. Zoino, "A learning management system includ-
ing laboratory experiments on measurement instrumentation," IEEE
Transactions on Instrumentation and Measurement, 55(5), 1757, (2006)
2. Cmuk, D.,M. Borsic,T. Mutapcic, and S. Rapuano,"A novel approach
to remote teaching: Multilanguage magnetic measurement experiment,"
IEEE Transactions on Instrumentation and Measurement, 57(4), 724,
(2008)
3. Gurkan, D.,A. Mickelson, and D. Benhaddou,"Remote laboratories for
optical circuits," IEEE Transactions on Education, 51(1), 53, (2008)
4. Ma, J., and J.V. Nickerson, "Hands-on, simulated, and remote laborato-
ries: A comparative literature review," ACM Computing Surveys, 38(3),
1,(2006)
5. Tzafestas, C.S., N. Palaiologou, and M. Alifragis, "Virtual and remote
robotic laboratory: Comparative experimental evaluation," IEEE
Transactions on Education, 49(3), 360, (2006)
6. Lindsay, E.D., and M.C. Good, "Effects of laboratory access modes
upon learning outcomes," IEEE Transactions on Education, 48(4),
619,(2005)
7. Cagiltay, N.E., E. Aydin, R. Oktem, A. Kara, M. Alexandru, and B.
Reiner, "Requirements for remote RF laboratory applications: An
educator's perspective," IEEE Transactions on Education, 52(1), 75,
(2009)
8. Kikuchi, T., S. Fukuda,A. Fukuzaki, K. Nagaoka, K. Tanaka,T. Kenjo,
and DA. Harris, "DVTS-based remote laboratory across the Pacific
over the gigabit network," IEEE Transactions on Education, 47(1),
26, (2004)
9. Nickerson, J.V., J.E. Corter, S.K. Esche, and C. Chassapis, "A model
for evaluating the effectiveness of remote engineering laboratories











and simulations in education," Computers & Education, 49,708-725,
(2007)
10. Prensky, M., Digital Game-BasedLearning, McGraw-Hill, New York
(2004)
11. Salen, K., and E. Zimmerman, Games of Play: Game Design Funda-
mentals, The MIT Press, Cambridge (2004)
12. Aldrich, C., Learning by Doing: A Comprehensive Guide to Simula-
tions, Computer Games, and Pedagogy in e-Learning and Other
Educational Experiences, Pfeiffer, San Francisco (2005)
13. Gibson, D., C. Aldrich, and M. Prensky, Eds., Games and Simulations
in Online Learning: Research & Development Frameworks, IGI Global,
(2006)
14. Papastergiou, M., "Digital Game-Based Learning in High School
Computer Science Education: Impact on Educational Effectiveness
and Student Motivation," Computers & Education,52(l1), 1-12, (2009)
15. Sward, K.A., S. Richardson, J. Kendrick, and C. Maloney, "Use of a
Web-Based Game to Teach Pediatric Content to Medical Students,"
Ambulatory Pediatrics, 8(6), 354, (2008)
16. Tuzun, H., M. Yilmaz-Soylu, T. Karakus, Y. Inal, and G. Kizilkaya,
"The effects of computer games on primary school students' achieve-
ment and motivation in geography learning," Computers & Education,
52(1), 68, (2009)
17. Ebner, M., and A. Holzinger, "Successful implementation of user-
centered game-based learning in higher education: An example from
civil engineering," Computers & Education, 49(3), 873, (2007)
18. Ryu, H., and D. Parsons, Innovative Mobile Learning: Techniques and
Technologies, Information Science Reference, Hershey, PA (2009)
19. Guy, R., (Ed.), R. Gafni, The Evolution of Mobile Teaching and Learn-
ing, Informing Science Press, Santa Rosa, CA (2009)
20. Akkermnan, S., W. Admiraal, and J. Huizenga, "Storification in His-
tory education: A mobile game in and about medieval Amsterdam,"
Computers & Education, 52(2), 449, (2009)
21. Lavifn-Mera, P., P. Moreno-Ger, and B. Femndez-Manj6n, "Develop-
ment of Educational Videogames in m-Learning Contexts," in Proc.
Second IEEE International Conference on Digital Game and Intelligent
Toy Enhanced Learning, pp. 44-51 (2008)
22. Kam, M., A. Agarwal, A. Kumar, S. Lal, A. Mathur, A. Tewari, and
J. Canny, "Designing e-learning games for rural children in India: a
format for balancing learning with fun," in DIS '08: Proceedings of


the 7th ACM conference on Designing interactive systems. : ACM,
New York (2008), pp. 58-67. [Online]. Available: org/10.1145/1394445.1394452>
23. Motiwalla, L.F., "Mobile learning: A framework and evaluation,"
Computers & Education, 49(3), 581, (2007)
24. Evans, C., "The effectiveness of m-learning in the form of podcast
revision lectures in higher education," Computers & Education, 50(2),
491(2008)
25. Uzunboylu, H., "Using mobile learning to increase environmental
awareness," Computers & Education, 52(2), 381, (2009)
26. Parsons, D., H. Ryu, and M. Cranshaw, "A design requirements frame-
work for mobile learning environments," J. Computers, 2(4), 1-8,
(2007)
27. Dondi, C., and M. Moretti, "A methodological proposal for learning
games selection and quality assessment," British J. Educational Tech-
nology, 38(3), 502, (2007)
28. Gafni, R., "Measuring Quality of m-Learning Information Systems,"
The Evolution of Mobile Teaching and Learning, R. Guy, Ed., Inform-
ing Science Press, Santa Rosa, CA (2009), pp. 211-247
29. Fernandez, J., R. Marin, and R. Wirz, "Online competitions: An open
space to improve the learning process," IEEE Transactions on Indus-
trial Electronics, 54(6), 3086, (2007)
30. Arango, F., G. Altuger, E. Aziz, C. Chassapis, and S. Esche, "Piloting
a game-based virtual learning environment," Computers in Education
J., 18(4), 82, (2008)
31. Casini, M., D. Prattichizzo, and A. Vicino, "The automatic control tele-
lab: A user-friendly interface for distance learning," IEEE Transactions
on Education, 46(2), 252, (2003)
32. Hassan, H., C. Dominguez, J. Martinez, A. Perles, and J. Albaladejo,
"Remote laboratory architecture for the validation of industrial control
applications," IEEE Transactions on Industrial Electronics, 54(6),
3094,(2007)
33. Zualkeman, I., "A framework and a methodology for developing
authentic constructivist e-Learning environments," Educational Tech-
nology & Society, 9(2), 198, (2006)
34. Malone, T.W., and MR. Lepper,"Making learning fun: A taxonomy of
intrinsic motivations for learning," in R.E. Snow and MJ. Farr (Eds.),
Aptitude, learning and instruction: III. Congnitive and affective process
analyses, Erlbaum, Hillsdale, NJ (1987), vol. 3, pp. 223-253 0


Chemical Engineering Education












Author Guidelines for the

LABORATORY

Feature

The laboratory experience in chemical engineering education has long been an integral part
of our curricula. CEE encourages the submission of manuscripts describing innovations in the
laboratory ranging from large-scale unit operations experiments to demonstrations appropriate
for the classroom. The following guidelines are offered to assist authors in the preparation of
manuscripts that are informative to our readership. These are only suggestions, based on the
comments of previous reviewers; authors should use their own judgment in presenting their
experiences. A set of general guidelines and advice to the author can be found at our Web site:
.

> Manuscripts should describe the results of original and laboratory-tested ideas.
The ideas should be broadly applicable and described in sufficient detail to
allow and motivate others to adapt the ideas to their own curricula. It is noted
that the readership of CEE is largely faculty and instructors. Manuscripts must
contain an abstract and often include an Introduction, Laboratory Description,
Data Analysis, Summary of Experiences, Conclusions, and References.
An Introduction should establish the context of the laboratory experi-
ence (e.g., relation to curriculum, review of literature), state the learning
objectives, and describe the rationale and approach.
The Laboratory Description section should describe the experiment in
sufficient detail to allow the reader to judge the scope of effort required
to implement a similar experiment on his or her campus. Schematic dia-
grams or photos, cost information, and references to previous publica-
tions and Web sites, etc., are usually of benefit. Issues related to safety
should be addressed as well as any special operating procedures.
If appropriate, a Data Analysis section should be included that concisely
describes the method of data analysis. Recognizing that the audience
is primarily faculty, the description of the underlying theory should be
referenced or brief. The purpose of this section is to communicate to the
reader specific student-learning opportunities (e.g., treatment of reac-
tion-rate data in a temperature range that includes two mechanisms).
The purpose of the Summary of Experiences section is to convey the
results of laboratory or classroom testing. The section can enumerate,
for example, best practices, pitfalls, student survey results, or anecdotal
material.
A concise statement of the Conclusions (as opposed to a summary) of
your experiences should be the last section of the paper prior to listing
References.


































gttp://cha uflIedu/CEE