V O LU M E F O U R N U M B E R O N E J A N UA RY - J U N E 2 0 0 4
P L AT F O R M
P L AT F O R M
Volume 4 Number 1
Jan - Jun 2004
SPECIAL INTEREST
Technology Innovation and Role of Research University
Sung-Kee Chung
3
Technology Cluster: OIL AND GAS
Technology Platform: Reservoir Engineering
11
A Holistic Approach to IOR
Brian GD Smart
18
The Development of An Optimal Grid Coarsening Scheme:
Interplay of Fluid Forces and Higher Moments of Fine-Scale Flow Data
N. H. Darman, G. E. Pickup and K. S. Sorbie
26
Surfactant Systems for Various Fields of EOR: Drilling Fluids,
Microemulsions Control of Viscosity, Breaking of Emulsions
Heinz Hoffmann
36
Technology Platform: Oilfield Gas Treatment and Utilization
Selective Fischer Tropsch Wax Hydrocracking – Opportunity for
Improvement of Overall Gas to Liquid Processing
Jack CQ Fletcher, Walter Böhringer and Athanasios Kotsiopoulos
46
Theory of Autothermal Reforming for Syngas Production from
Natural Gas
Kunio Hirotani
54
Development of Defect-Free and High Performance Asymmetric
Membrane for Gas Separation Processes
Ahmad Fauzi Ismail, Ng Be Cheer, Hasrinah Hasbullah and
Mohd. Sohaimi Abdullah
68
Technology Platform: System Optimization
Formulation Engineering and Product-Process Interface
J P K Seville and P J Fryer
84
Advancements in Tension Leg Platform Technology
John W. Chianis
91
Materials and Manufacturing and Design Issue Associated with
Composite Products for Gas and Petrochemical Industry
John P. Coulter, Joachim L. Grenestedt and Raymond A. Pearson
106
V O LU M E F O U R N U M B E R O N E J A N UA RY - J U N E 2 0 0 4
Air Injection-Based IOR for Light Oil Reservoirs
R. G. Moore, S. A. Mehta and M. G. Ursenbach
ISSN 1511-6794
P L AT F OR M
Contents
January-June 2004
Advisor:
Dr. Rosti Saruwono
UTP Publication Committee
Chairman:
Assoc. Prof. Ir. Dr. Ahmad Fadzil Mohamad Hani
SPECIAL INTEREST
Members:
Assoc. Prof. Ir. Dr. Ibrahim Kamaruddin
Assoc. Prof. Dr. Mohamed Ibrahim Abdul Mutalib
Assoc. Prof. Dr. Mohd. Noh Karsiti
Assoc. Prof. Dr. Fakhruldin Mohd. Hashim
Assoc. Prof. Dr. Madzlan Napiah
Dr. Hilmi Mukhtar
Dr. Abas M. Said
Dr. Noor Asmawati M. Zabidi
Hasbullah Abdul Wahab
Hasbullah Haji Ihsan
Mohamad Zahir Abdul Khalid
Technology Innovation and Role of Research University
Sung-Kee Chung
3
Air Injection-Based IOR for Light Oil Reservoirs
R. G. Moore, S. A. Mehta and M. G. Ursenbach
11
Secretary:
Raja Yasmin Raja Yusof
A Holistic Approach to IOR
Brian GD Smart
18
The Development of An Optimal Grid Coarsening Scheme:
Interplay of Fluid Forces and Higher Moments of Fine-Scale Flow Data
N. H. Darman, G. E. Pickup and K. S. Sorbie
26
Surfactant Systems for Various Fields of EOR: Drilling Fluids,
Microemulsions Control of Viscosity, Breaking of Emulsions
Heinz Hoffmann
36
Technology Cluster: OIL AND GAS
Technology Platform: Reservoir Engineering
University Editorial
Editor-in-Chief:
Mohamad Zahir Abdul Khalid
Chief Editor, PLATFORM:
Ir. Dr. Kamarul Ariffin Amminudin
Editor, UTP Quarterly:
Raja Yasmin Raja Yusof
Representative, IRC:
Rabiatul Ahya Mohd. Sharif
Secretary:
Raja Yasmin Raja Yusof
PLATFORM Editorial
Chief Editor:
Ir. Dr. Kamarul Ariffin Amminudin
Co-Editors:
Prof. Dr. V. R. Radhakrishnan
Assoc. Prof. Dr. Varun Jeoti Jagadish
Assoc. Prof. Dr. Norani Muti Mohamed
Dr. Nasir Shafiq
Dr. Abdul Rashid Abdul Aziz
Jafreezal Jaafar
Chong Su Li
Hairulzila Idrus
Address:
Chief Editor, PLATFORM
Universiti Teknologi PETRONAS
Bandar Seri Iskandar, 31750 Tronoh
Perak Darul Ridzuan, Malaysia
ht t p : / / w w w. u t p. e d u. my
ka m a r u l a ri f f i n _ a m m i n u d i n
@ pe t ro n a s . co m . my
Telephone
+ ( 6 0 ) 5 368 8687
Facsimile
+ ( 6 0 ) 5 365 4090
Copyright © 2004
Universiti Teknologi PETRONAS
Technology Platform: Oilfield Gas Treatment and Utilization
Selective Fischer Tropsch Wax Hydrocracking – Opportunity for
Improvement of Overall Gas to Liquid Processing
Jack CQ Fletcher, Walter Böhringer and Athanasios Kotsiopoulos
46
Theory of Autothermal Reforming for Syngas Production from
Natural Gas
Kunio Hirotani
54
Development of Defect-Free and High Performance Asymmetric
Membrane for Gas Separation Processes
Ahmad Fauzi Ismail, Ng Be Cheer, Hasrinah Hasbullah and
Mohd. Sohaimi Abdullah
68
Technology Platform: System Optimization
Formulation Engineering and Product-Process Interface
J P K Seville and P J Fryer
84
Advancements in Tension Leg Platform Technology
John W. Chianis
91
Materials and Manufacturing and Design Issue Associated with
Composite Products for Gas and Petrochemical Industry
John P. Coulter, Joachim L. Grenestedt and Raymond A. Pearson
106
ISSN 1511-6794
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
1
Editorial
This special edition of PLATFORM features a selection of papers which were presented
during the First International R&D Forum on Oil, Gas and Petrochemicals. The Forum was
organized by Universiti Teknologi PETRONAS (UTP) and PETRONAS Research and Scientific
Services in Kuala Lumpur from April 5th to April 6th 2004. The objectives of the forum are to
provide an avenue for university researchers and industrial practitioners to exchange ideas,
learn about the latest developments in the oil, gas and petrochemical industries and develop
strategies relevant to their areas of interest through collaboration.
Even though the Forum reflects UTP R&D technology cluster of Oil and Gas, this special edition
of PLATFORM begins with a special interest paper on the subject of technology innovation and
the role of university. This paper highlights the growing importance of innovation in our
economic growth today, and the role of universities in contributing to this innovation system.
The next ten papers in this special issue of PLATFORM cover three UTP’s R&D technology
platforms under the technology cluster of Oil and Gas; namely, reservoir engineering, oilfield
gas treatment and utilization, and system optimization. Four papers on reservoir engineering
technology platform are presented in this issue. Two papers highlight on Improved Oil Recovery
(IOR) in which a paper describes a technique using air injection for IOR and another paper
presents a holistic approach to IOR. The third paper is on the development of an optimal grid
coarsening scheme which relates to the improvement of the reservoir characterization; and
the fourth paper is on the surfactant system for Enhanced Oil Recovery (EOR).
Under the oilfield gas treatment and utilization technology platform, three papers are presented.
The first highlights opportunities to improve overall gas to liquid processing using selective
Fischer Tropsch wax hydrocracking; the second is on autothermal reforming for syngas
production from natural gas; and the third one covers the development of high performance
asymmetric membrane for gas separation processes.
Finally, three papers under the system optimization technology platform are presented. The
first paper expounds the concept of formulation engineering and product-process interface.
The second relates to system optimization on the offshore structure of tension leg platform;
and the final paper presents issues associated with composite products for gas and
petrochemical industry.
It is hope that these papers which were presented by the renowned authors in their respected
areas will help UTP’s ongoing research activities under this technology cluster to remain on the
right course.
Ir. Dr. Kamarul Ariffin Amminudin
Chief Editor
2
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
SPECIAL INTEREST
TECHNOLOGY INNOVATION
AND ROLE OF RESEARCH UNIVERSITY
Sung-Kee Chung
Department of Chemistry, Pohang University of Science & Technology,
San 31, Hyojadong, Namgu, Pohang, Kyungbuk 790-784, Korea.
[email protected]
ABSTRACT
Over the last 250 years of industrial economic history, it is a generally-accepted proposition that innovation is
the principal engine of the economic development and a primary means of wealth creation of society. More
recently it is also generally accepted that a knowledge-based economy will continue to be the prevailing mode
in the current century. Hence, the production, distribution and use of knowledge in various forms account for a
major portion of the productivity and the gross domestic product (GDP) of a country. A nation’s economic
strength no longer depends as heavily as before on low-cost labor, cheap raw materials or capital availability,
but on a continually operating innovation system. A key to this system appears to be an efficient and productive
alliance of industry, universities, government, and labor. The innovation system is expected to generate new
high value-added knowledge and technologies, to educate a highly-skilled work force to take advantage of
these new technologies, and to produce the next generation of innovators to carry on this process of everincreasing efficiency and productivity. This combination of new knowledge, skilled workers, and innovators
mixes with capital investment to yield a continuous stream of new products and services, which, in turn, advance
the economy and improve the quality of life. Because this innovation system is so central to any society’s economic
future, the role of each partner and the relationships among partners need to be redefined on a continual basis.
Keywords: university-industry cooperation, technology innovation, knowledge-based economy
TECHNOLOGY INNOVATION AND ECONOMIC
DEVELOPMENT
Prior to the first industrial revolution, more than 90%
of world population were dependent upon
agriculturally based economy, but now the agricultural
population in advanced countries stands less than 5%.
Around mid-18th century a fundamentally important
transformation in this agriculturally based economic
situation began to take place. What the first industrial
revolution and the steam engine in particular achieved
was to convert heat into work through the use of
machines, thus enabling mankind to exploit vast new
sources of energy. The impact of this change on textile
industry for example was simply stupendous; by 1780s
power-operated looms typically produced twenty
times the output of a hand worker, while a power
driven spinning machine had two hundred times the
capacity of a spinning wheel. The massive increase in
productivity of the textile industry generated a
cascade effect, thus stimulating a higher demands for
more machines, more raw materials, more iron, more
shipping, better communication and so on. This
virtuous cycles of economic expansion both in
national wealth and people’s purchasing power
constantly outpaced the increase of population; while
the Britain’s population rose from 10.5 million in 1801
to 41.8 million in 1911, its national product rose much
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
3
SPECIAL INTEREST
faster, perhaps as much as fourteen-fold. The
industrialization processes gradually spread out region
by region in the Western world. By the late 19th
century the new science-based chemical industry in
Germany and electrical industries in the United States
have been developed and stimulated great advances
in other areas as well. In the 20th Century, scientists
and engineers had become key players in laying the
foundations of technology-based society.
The Second World War marked an important turning
point in the evolution of science and technology and
of university-industry interaction. The high public
confidence in scientific research based on successful
wartime scientific projects and the subsequent ‘Cold
War’ situation stimulated a rapid expansion of large
corporate research laboratories and funding increase
of university-based research activities, which resulted
in major advances in industries like electronics,
computers, advanced materials, communications, and
biotechnology, thus providing the foundation for
economic growth in the late 20th Century. During this
period the major emphasis of public research funding
was given to science rather than technology, and
relatively little attention was paid to the sciencetechnology interactions and their impact on industrial
innovation. However, in the face of the economic
challenge from Japan’s manufacturing innovation in
the 1980s coupled with mounting trade deficits, the
university-industry partnership began to take a
dramatic turn in the United States. This change,
generally referred to as ‘the competitive approach’
added a new role to the traditional missions of
universities, namely education and research. Now
university was supposed to play a more active role in
the technology development and the
commercialization of the university-generated
research results in a direct way that leads to the
economic development of the society.
In the 1990s the changing nature of global economic
competition and the enhanced race for advanced
technologies forced national governments to adopt
policies which encouraged closer partnership
between the academia and business. Under this
4
environment, the university’s role in the production
and dissemination of knowledge and their utilization
for improving technological and innovative
capabilities has become much more important and
commonplace. The enactment of the 1980 Bayh-Dole
Patent and Trademark Act gave universities additional
financial incentives. These changes stimulated a
growing commercialization of academic research
results with emergence of entrepreneurial universities.
The shifting mode of the university-industry
relationship may be considered as a reflection of the
changing societal expectation of the roles of university.
This series of events was called as the second academic
revolution as opposed to the first academic revolution,
when Wilhelm von Humboldt enunciated the unity of
teaching and research as the core missions of
university, thus begetting research universities.
BankBoston’s Special Report issued in March of 1997
entitled “MIT: The Impact of Innovation,” tells the
following story. At the time of the report, MIT
graduates and faculty were credited with founding
4,000 companies, employing over 1 million people
nationwide. MIT-related companies with more than
10,000 employees numbered 43 and represented
more than 70% of the employment associated with
the 4,000 MIT-related companies. However, the 1997
rate of MIT-related companies being formed was about
150 companies per year, yielding much promise for the
future. While many of the founders had college
degrees from institutions other than MIT, the culture
of MIT had contributed significantly to national
technological innovation by its community. The MIT
story emphasizes that technology transfer is done by
people, the people with the know-how and show-how,
as well as the ability, passion and interest in
technological innovation. Like MIT, many research
universities can and should prepare their students as
future agents for technology transfer and as
contributors to technological innovation and impact
through the companies they form. The study of MIT
also revealed something else of importance: impact
in the area of technological innovation does not come
quickly. None of the companies employing more than
10,000 employees were founded more than 10,000
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
SPECIAL INTEREST
employees were founded later than 1968 and many
were formed much earlier.
ROLE OF HIGHER EDUCATION IN THE ECONOMIC
DEVELOPMENT OF KOREA
The current atmosphere in Korea on its economic state
and prospect is charged with senses of anxiety and
frustration, and the Korean economy is clearly at a
transition. The economy grew at a crawling 5.2%
during 1996-2002 and an anemic 3% growth is
projected for 2003. The causes of this lackluster
economic performance could be many, and many
discussions and debates are currently on-going
concerning how to revitalize and accelerate the
economic growth for the next decade. In the
meantime, nation’s universities are targets of much
criticism for their lack of global competitiveness in
education as well as research productivity. The Korea’s
economic development in the last 30 years prior to
the Asian monetary crisis in 1997-1999 was nothing
short of remarkable. Korea has literally undergone
industrial revolutions in a very condensed manner. The
GDP increased by an average of more than 7 percent
annually, from about $2.5 billion in 1965 to more than
$400 billion in 1999, ranking Korea as the 13th largest
country in GDP size. Per capita GDP grew from less
than $400 to close to $10,000 in the same period.
When the economic development began in the 1960s,
neither endogenous technological capability nor
technical manpower was in existence. For example, at
the end of World War ll, there were only a dozen or so
university graduates who had B.S. or higher degrees
in physics. The situation in the 1950s was not much
better due to the Korean War and subsequent social
confusion. The only resources Korea had were
desperate but disciplined workers with only some
general and basic education. Thus, the economic
development plan was based on the “export-oriented
strategy,” and it relied heavily on foreign capital,
borrowed technology, and imported raw materials, and
as well as on cheap labor.
Initially, an emphasis was given to light industries, such
as textiles, plywood, cement, fertilizer, and electric
power generation. As the development proceeded,
technological support capabilities became necessary.
In order to aid private sectors’ need for technology and
build the S&T base for the future, the government
established Korea Institute of Science & Technology
(KIST ) in 1969. KIST concentrated on providing
technological support for the industries in adopting,
modifying, and improving the imported foreign
technologies, and it served as a model for all
government-supported research institutes
subsequently established. In the early 1970s, advanced
training for scientists and engineers was envisioned
in preparation for competitive indigenous
technologies. In mid-1960s, there were about 46,000
undergraduates and 900 graduate students in all
science and engineering fields. There were several
leading universities, including Seoul National
University, Yonsei University, and Korea University (the
latter two being the oldest private institutions), with
relatively high academic standards and good
reputations. However, their research capabilities were
deemed as substantially below global benchmarks.
The Korean government initially wanted to reform and
upgrade the existing universities to research
institutions, but quickly realized that reforms would
be more difficult and slower than starting fresh as
Machiavelli opined more than 450 years earlier: “he
who innovates will have his enemies all those who are
well off under the existing order of things, and only
lukewarm support in those who might be better off
under the new.”
Thus, in 1972 the government established Korea
Advanced Institute of Science (KAIS, later renamed as
KAIST, adding “technology” to the name). KAIS, a new
graduate school of engineering and applied sciences,
was expected to meet the needs for highly trained
manpower and high-level research. With exceptionally
high salaries and excellent research facility for its
faculties, full financial support and exemption from
compulsory military service for students, and
significant autonomy and little red tape for the
administration, KAIST has been successful in carrying
out serious research and producing in substantial
numbers well-trained M. S. and Ph. D. graduates. These
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
5
SPECIAL INTEREST
graduates have subsequently played key roles for the
development of manufacturing industries. In the
middle of the 1970s the industrial development plan
began to shift toward heavy and chemical industries,
such as steel, shipbuilding, machinery, mining,
petroleum refining, and electronics. The shift was
designed to upgrade industrial structure and capacity,
but it also had implications in the national defense
strategy. During this phase Korean industries
experienced an increasingly acute need for advanced
indigenous technology, which could not be readily
obtained from abroad. Furthermore, numerous
government-supported research institutes were
established. A series of subsequent policies also
encouraged the establishment of private research
institutes and R&D consortia, resulting in a significant
expansion of civilian investment in technology
development. In addition, a system of collaboration
was established among industry, academia, and
research institutes - especially in the designated core
areas: machine parts, new materials, semiconductors,
computer network, communication technology,
energy, nuclear fuel, pharmaceuticals, and
biotechnology. Investments in S&T showed an average
annual increase of 16% - from 0.7% of the GDP in 1980
to 1.87% in 1987. The major source of investment in
S&T also changed from the government (64%) to the
private sector (75%).
By mid-1980s, Korean S&T areas needed additional
stimulus and a fresh perspective. In 1986, the Pohang
University of Science & Technology (POSTECH) was
established as the first private, research oriented
institution of higher learning by POSCO, then a
government-owned, one of the largest iron and steel
producing companies in the world. From its inception,
POSTECH has strongly emphasized excellence in
research and education, the synergy between research
and education, and cooperative activity between the
university and industry. With well-defined goals, solid
financial means, and the state-of-the-art facilities,
POSTECH has been quite successful in the given
missions, as will be described later in much more
details.
6
Over the last 30 years, private colleges and universities
in Korea have proliferated because of the anticipated
industrial need and a Confucian culture, which values
education and attaches prestige to academic
credentials. In the year 2000, more than 1.7 million
undergraduates and 230,000 graduate students were
enrolled in almost 200 colleges and universities, of
which more than 150 institutions were private schools;
undergraduate and graduate enrollments in S&T
totaled 727,000 and 71,000, respectively. The number
of college graduates per every 10,000 of the general
population was 495, one of the highest in the world.
Mass higher education appears to have been very
successful in terms of quantity, and it has contributed
to the development of the manufacturing industries
and the democratic political process. However, in
terms of quality, the financial means of Korean
universities are grossly inadequate to run high quality
programs, especially in the S &T areas. The average
quality of higher education has suffered greatly from
the policy of increasing enrollment without careful
regard to maintaining high standards. Our policy
makers need to be reminded that in the era of the
knowledge-based economy, it is the quality of
knowledge, such as creativity and originality, which is
the source of the high value addition rather than the
general knowledge of the educated masses. The road
to democracy, especially since 1988, has been
accompanied by severe labor disputes, wage increases
beyond productivity, and lower savings rates, thus
leading to the declines in global economic
competitiveness. In addition, the opening up of the
domestic markets, coupled with the concurrent
protectionist trend of developed countries in
technology transfer, further weakened the exportoriented economic development.
During the period of the rapid economic growth in
Korea, the industrial sector clearly served as the main
engine for creating more per capita income and
national wealth. However, in the face of rapidly
changing world economic scenes with emerging new
trends including the knowledge and information
revolution, globally linked economy, and heightened
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
SPECIAL INTEREST
environmental awareness, and unsettled domestic
environment including lack of transparency in
business activity and high labor cost, the Korean
economy lately has not successfully responded to the
new situation. It has been experiencing the nutcracker
effect, namely getting squeezed between the highly
innovative economy and newly developing economy.
Recent data show that unemployment has been
creeping up rather rapidly with the current number at
ca. 3.5 %, and even the increasing exports are limited
to only a few industrial items such as automobile,
semiconductor, mobile phone, computer, and
shipbuilding. More recently, industry-academygovernment got together and identified ten most
promising industries/products as the next growth
engine of Korean economy. However, it is so obvious
that these expectations cannot be met without a wellfunctioning technology innovation system in place.
It is quite evident that the old economic development
strategies have become nonviable for the future
economic progress of Korea, and that the necessary
growth and value addition must come from
indigenous R&D innovation and entrepreneurship
together with global business alliances. Korean
research universities are expected to play bigger roles,
since a bulk of Ph.D. level researchers are in academia.
Various funding programs to shore up capabilities in
research and development, and linkages between
science, technology and commercialization have been
initiated by the government. For the year 2001, the
government R&D investment was increased to $4.27
billion (4.3 % of the total budget), and the budget for
education to $21.6 billion (4.2% of GDP), with notable
additional amounts to boost the quality of graduate
education in S&T. Lately, research papers and outputs
began to show substantial increases not only in
quantity but also in quality. For the 5 year period
between 1997 and 2001, more than 55,000 SCI papers
were published with increasing impact factors, which
ranks Korea as 16th in the world, and for year 2002
15,643 SCI papers were published with 14th global
ranking. However, universities are still experiencing
great difficulties in implementing the necessary
reforms toward higher research productivity and more
creative education, because of faculty resistance to the
proposed measures incorporating competitive and
selective processes, such as a strategic funding and
merit-based pay system. Major tasks facing universities
and government alike are how to mobilize enough
financial resources to rapidly and selectively upgrade
the quality of higher education against the prevailing
sentiment of masses demanding equality in every
aspect. How to achieve global competitiveness in
research productivity and quality education through
systematic innovation processes appears to be the key
issue that Korean society needs to address in order to
revitalize the slacking economic progress.
UNIVERSITY-INDUSTRY COOPERATION
The advent of economic globalization, proceeding at
an ever-increasing pace in recent years, has put private
industry under increasing pressure to pay more
attention to short-term profitability. Subsequently,
industry has pushed their R&D activities much more
into projects that show immediate commercialization
potential. Hence, industry is forced to look for
opportunities in university laboratories to capture new
and longer-term ideas. On the other hand, as public
research support is becoming limited and more
competitive, and as accountability requirements
become more stringent, universities also are forced to
look to industry for financial support. Thus the
connection between the university and industry has
become both a necessity and a trend. Technology
transfer, as pointed out by Donald Kennedy, President
Emeritus of Stanford University, has become the
newest addition to the traditional roster of academic
duties, i.e. teaching and research. Not only has the
public come to expect the university as knowledge
producer to share what it discovers for the benefit of
society, but it has also become necessary for the
university to apply its intellectual properties into
commercial services for its own financial gain. Charles
Vest, the current president of MIT, summed up the
potential benefits to universities from cooperation
with industry as follows: “Over the longer term,
collaborations can have a transforming effect on the
ability of institutions to attract high quality faculty, to
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
7
SPECIAL INTEREST
encourage faculty and their students to interact more
closely with industry, and to design curricula and
academic programs better attuned to the needs of
industry and challenges we face as a nation.” With
regard to this relationship, it is interesting to note the
observation made by late Clark Kerr, Former President
of University of California: “..…in America, the land
grant university in the 19th century has evolved to the
federal research grant university in the 1960s, and is
now in the process of transforming to the private grant
university.”
The university-industry cooperation can take many
different forms ranging from simple research grants
and contracts, research collaborations and strategic
partnerships, to technology transfer arrangements.
Technology transfer may include licensing, venture
incubation, investment through venture capital
operation, as well as participation in science/techno
park project, among others. The relationship between
university-industry is regarded as particularly
important in Korea, since the academic sector has
more human resources and research capabilities while
industry possesses more financial resources; ca. 75 %
of all Ph.D. level researchers in Korea are employed in
academia, whereas industrial R/D budgets account for
more than 70 % of Korea’s national total. Currently
POSTECH has 1,200 undergraduate and 1,600
graduate students, 220 faculty, 500 staff, and 600
researchers. Since 1994 POSTECH also operates the
Pohang Light Source - 2.5 GeV synchrotron facilities,
the only one in Korea. With a firm, established criteria
and vision in mind, POSTECH has been highly selective
in its recruitment of both faculty and students.
Limiting freshmen enrollment to a maximum of 300,
POSTECH has managed to continuously recruit the top
1-2 percentile of graduating high school students in
the country. In a sense, by keeping the size small it
strives to grow bigger. It has also been playing a
catalytic role in the nation’s attempts to improve the
quality of the S&T research and graduate education.
POSTECH’s current annual budget is US$ 140 million
for the academic units and US$ 20 million for the
Pohang Accelerator Laboratory. POSTECH is now
highly regarded as a premier research university in not
8
only Korea but throughout Asia. How did it come to
be so well regarded? One of the reasons is certainly
through constant efforts to develop and maintain the
synergy and cooperative relationship between
POSTECH and industry.
With the changing global and national trends in S&T
education, POSTECH has continuously been striving
to become actively engaged in university-industry
cooperation in a variety of forms. The statistical
numbers on the research activity at POSTECH over the
last 10 years demonstrate interesting trends. The total
number of regular, full-time professors has slightly
increased from 187 in 1991 to 210 in the year 2001.
The size of research fund per faculty has increased
about 6-fold to US$ 328K, while total research volume
shows more than 7-fold increase to US$ 69 million.
POSTECH faculty has been very successful in securing
research support from national funding agencies in
highly competitive programs, such as Excellence
Centers (SRC/ERC), Creative Research Initiatives, Brain
Korea 21, and National Laboratory Programs. On the
other side, the largest portion of POSTECH’s industrial
support came in the form of strategic partnerships
with companies like POSCO, Samsung, LG, and Hyundai.
The impact of such partnerships on the research
support has been particularly noticeable over the last
four years.
The partnership between university and industry can
potentially pose a number of challenging problems.
And a number of problems have been identified even
in the United States, where such partnership is
considered to be most active and successful. It is with
these problems in mind that the U. S. Congressional
Report,‘Unlocking Our Future. Toward a New National
Science Policy,’ gave the following recommendations:
“If the collaboration is to be effective, 1) universities
must not lose sight of their ultimate aims of teaching
students and performing basic scientific and
engineering research inquiry. Unless universities retain
their culture, base of fundamental research, and
educational mission, they will not have a value to bring
to the partnership; 2) university researchers should not
be discouraged from publishing or otherwise
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
SPECIAL INTEREST
disseminating their research results, due to proprietary
claims to these results made by their industry partners;
3) private sector entities that partner with universities
should not view their university partners as fullfledged substitutes for their own research programs. ”
We, at POSTECH, have been very mindful of these
potential and real problems, and fortunately, we have
yet to encounter any major problems that are so
serious that we have had to consider withdrawal from
a partnership. The true spirit of the university-industry
cooperation has a common thread with the so-called
Jeffersonian research mode. It has been argued that
the confining dichotomy of basic research in academe
(Newtonian mode) versus applied research in industry
(Baconian mode) may be reconciled through a third
mode (Jeffersonian mode), which suggests research
activity that is driven by practical, societal needs but
carried out under conditions of imagination, flexibility,
and competition that are usually associated with the
traditional basic research.
At POSTECH, we encourage our faculty to file patent
applications through generous incentives, such as
splitting the royalty income between the university,
their respective department, and the principal
investigator (50:10:40). The yearly number of patent
applications has been steadily increasing from 63 in
1998 to 242 in the year 2002, and the number of
technology transfer cases has averaged around 17 over
the last 5 years. We are quite pleased to observe that
progresses in this area was made without any
noticeable negative impact on academic performance;
the total number of scholarly publications has
increased from 531 (76 in SCI journals) in 1991 to 1,219
(790 in SCI) in 2002, which corresponds to 5.6 papers
(3.8 in SCI) per faculty.
In order to facilitate technology transfer POSTECH
started in 1997 its Venture Business Incubator (VBI) and
the Venture Capital Corporation (VC). The Incubator,
together with the VC, supports the commercialization
of new ideas or new technology - generated by
primarily the faculty, students and research staff - by
providing laboratory space and facilities, funds, and
advice on various aspects of business management
such as marketing, financing, as well as tax and legal
matters. Eleven start-up companies have graduated
from our incubation program and have gone on to
become independent business operations, and 20
entities are currently under incubation, mostly in the
areas of information, bio-, and environmental
technology. The Postech Venture Capital Corporation
with a total capitalization of ca. US$ 60 million, so far
has invested in more than 70 start-up companies
(about US$ 44 million) mostly at their early stages that
have business activities in the areas of communication,
electronics, the Internet, computer software,
biotechnology, and environmental technology. We will
have to wait a few more years before we can assess
the success of these operations in terms of both the
university mission and the business standpoint.
POSTECH is also actively involved in the Pohang
Technopark Project, which is jointly managed by the
local governments, the University and POSCO. The
project got started with the ultimate objective of
transforming the city of Pohang (which is dominated
by heavy industry, such as iron and steel and related
manufacturing) into a much more diversified hightech industrial city. The Pohang Technopark is also a
member of the National Technopark Association under
the jurisdiction of the Ministry of Commerce, Industry
and Energy. The project calls for a total investment of
US$ 250 million over the 15-year period for the
development of a park in the vicinity of the POSTECH
campus, and it is to accommodate R&D, business
incubation, high-tech manufacturing, and educational
and training activities. The Pohang Technopark
Foundation was incorporated in 2000, a US$ 72 million
fund was raised, and a strategic master plan was drawn
up to ensure its future success. Two buildings have
already been constructed and occupied by 35
companies with business interests in the areas of IT,
BT, NT and ET. The role that POSTECH plays in the
Technopark Project is a pivotal one, as it is expected
to make major contributions toward providing highly
skilled manpower, technologies, and able
entrepreneurs.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
9
SPECIAL INTEREST
I have attempted to briefly review the why, what, and
how of the academe-industry cooperation with some
limited examples and experiences we have had at
POSTECH. We feel that we have been making good
progress, although the jury is still out as far as the final
assessment of our successes and failures are
concerned. We are hopeful and optimistic in the sense
that we, as a research university, have made a strong
commitment to this area, not as a choice, but rather as
an added academic responsibility, but we have had to
stay on guard so that any number of well-argued
negative consequences will not become a nasty reality.
The prevailing current that the science should be made
more like the technology in order to reap shorter-term
economic benefits, and a more proactive attitude
should be taken to enhance the university-industry
relationship, needs to be counterbalanced by the
concern that such an attitude is likely to be found
short-sighted in the long run. In this connection,
society as well as university in general might be well
served by remembering the caveat offered by Derek
Bok, former president of Harvard. “ … Making money
in the world of commerce comes with a Faustian
bargain in which universities have to compromise their
basic values - and thereby risk their very souls - in order
to enjoy the rewards of the marketplace. …By
compromising basic academic principles, universities
tamper with ideals that give meaning to scholarly
community and win respect from the public. The
common values are the glue that binds together an
institution already fragmented by a host of separate
disciplines, research centers, teaching programs and
personal ambitions. They keep the faculty focused on
the work of discovery, scholarship, and learning despite
the manifold temptations of the outside world. They
help maintain high standards of student admissions
and faculty appointments. They sustain the belief of
scientists and scholars in the worth of what they are
doing. They make academic careers a calling rather
than just another way to earn a living. …Before moving
further down this path, university leaders should recall
the history of intercollegiate athletics and ponder the
lessons it teaches.”
10
REFERENCES
[1] Committee Print 105-B, Unlocking our Future: Toward A New
National Science Policy, Committee on Science, U.S. House of
Representatives, 105th Congress, 1998.
[2] Business-Higher Education Forum, Working Together, Creating
Knowledge, The University-Industry Research Collaboration
Initiative, 2001.
[3] M. S. Wrighton, Economic Development and the Research
University: An American Perspective, Presidents Forum,
Shandong University, 2001.
[4] D. Kennedy, Acadmic Duty, Harvard University Press, 1997.
[5] C. Kerr, The Uses of the University, 4th Edition, Harvard
University Press, 1994.
[6] D. Bok, Universities in the Marketplace. The Commercialization
of Higher Education, Princeton University Press, 2003.
[7] G. Holton and G. Sonnert, A Vision of Jeffersonian Science,
Issues in Science and Technology, Fall, 1999.
[8] L. M. Branscomb, The False Dichotomy: Scientific Creativity and
Utility, Issues in Science and Technology, Fall, 1999.
[9] L. M. Branscomb and Y. -H. Choi, ed. Korea at the Turning Point:
Innovation-based Strategies for Development, Praeger
Publishers. 1996.
Sung-Kee Chung is a Professor in the
Department of Chemistry at Pohang
University of Science & Technology
(POSTECH). Currently, he is a Board
Member of Korea Institute of S & T
Evaluation and Planning, Samsung
Scholarship Foundation and Korean
Chemical Society. He is also a Guest
Professor for University of Oxford, ETHZurich, University of Science & Technology of China and University
of Tokushima. He holds a BSc (Chemistry) from Yonsei University,
PhD (Organic Chemistry) from University of Illinois at Urbana and
Postdoctoral (Bioorganic Chemistry) from Yale University. He has
published more than 150 refereed papers, 14 patents and books
in the area of natural products, mechanistic and synthetic
bioorganic chemistry, and medicinal chemistry.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
AIR INJECTION-BASED IOR FOR LIGHT OIL RESERVOIRS
R. G. Moore1, S. A. Mehta and M. G. Ursenbach
In Situ Combustion Research Group, Department of Chemical and Petroleum Engineering,
University of Calgary, 2500 University Drive, Calgary, Alberta, Canada, T2N 1N4.
1
[email protected]
ABSTRACT
Air injection is currently being evaluated as an IOR process for a wide variety of reservoirs. When applied in
deep, high pressure, low permeability reservoirs containing high gravity crudes, the process is generally known
as High Pressure Air Injection (HPAI). This paper will review the key factors that should be understood when
considering the application of HPAI and it will describe the laboratory tests that are used to evaluate the “burning”
characteristics of a candidate reservoir.
Keywords: High Pressure Air Injection (HPAI), In Situ Combustion, Thermal Calorimetry, IOR, Oxidation Kinetics
INTRODUCTION
Interest in the application of high pressure air injection
was generated by the pioneering projects of Koch
Exploration Company in the Williston Basin of South
and North Dakota which are described by Erickson et
al. [1] and Kumar et al. [2]. Continental Resources, Inc.
purchased the Koch projects and they have continued
to expand their air injection operations to other
reservoirs within both the Dakotas and Montana.
Amoco Production Research Company (now BP) had
the vision to develop a high pressure air injection
laboratory and to put together an industrial
consortium to evaluate prospective HPAI candidate
reservoirs and it is a result of the donation of the
Amoco laboratory to the University of Calgary which
allowed the combustion research group to be actively
involved in the development of HPAI technology.
Reviews of HPAI projects have been provided by
Sarathi and Olsen [3] and by Moore et al. [4], hence the
reader interested in further information on HPAI is
referred to these publications. Discussions on the
laboratory screening tests which are performed to
evaluate potential reservoirs for application of HPAI are
provided by Sarma et al. [5], Juan et al. [6], and
Takabayashi et al. [7].
PROCESS OVERVIEW
High Pressure Air Injection (HPAI) is loosely defined as
an IOR process in which compressed air is injected into
a high gravity, high pressure oil reservoir, with the
expectation that the oxygen in the injected air will
react with a fraction of the reservoir oil at an elevated
temperature and produce carbon dioxide. The
resulting flue gas mixture, which is primarily nitrogen
and carbon dioxide provides the mobilizing force to
the oil downstream of the reaction region, sweeping
it to production wells. The gas-oil mixture may be
immiscible, or partly or completely miscible. In some
situations, the elevated temperature reaction zone
itself may provide a critical part of the sweep
mechanism in terms of incremental recovery.
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
11
Technology Platform: RESERVOIR ENGINEERING
In its simplest implementation, the process is initiated
simply by injecting air, which will spontaneously ignite
the in-place oil due to the high temperature and
pressure conditions in the reservoir. In situations
where spontaneous ignition of the native reservoir oil
won’t likely occur – this can be determined by
laboratory testing – the ignition must be aided with
the injection of a chemical mixture which is capable
of spontaneous ignition at the reservoir conditions or
by an initial input of energy, usually provided by a
down-hole heater or burner. Air injection is achieved
using compressors that are specifically designed for
air at the pressure levels that are required to inject the
desired volumes.
Why Consider HPAI?
HPAI has been shown by Continental Resources Inc. to
be an effective recovery process in light oil reservoirs
which are not considered to be good candidates for
water-flooding or which have reached the end of their
economic life under water-flood. Air injection can also
be operated in conjunction with an active water-flood
as a means for increasing the vertical conformance of
the combined flood.
Unlike water, whose use for injection in oil field
reservoirs is being restricted by environmental
regulations, air is available everywhere. Air injectivity
into tight reservoirs is higher than that of water, hence,
for a given injector it is possible to inject greater
volumes (measured at reservoir conditions) of air than
water. Pumping requirements are reduced as
produced water volumes are significantly less for air
injection projects and reservoir re-pressurization by
the injected gas will often allow the producers to
operate on flow.
All air injection processes benefit from the fact that
the reservoir is an excellent scrubber for the sulfur
compounds formed during the oxidation reactions
and the operating temperatures are generally below
those which result in the generation of nitrogen oxides.
Sequestration of the produced carbon dioxide can be
achieved by re-injection of the product gas either by
12
Technology Cluster: OIL AND GAS
co-injection with the air or injection into dedicated flue
gas injectors.
Oxidation Kinetics
The critical factor for success of HPAI is the reaction
between the oxygen and the hydrocarbons. Two
distinct reaction pathways exist. The first is referred
to as “bond scission” reactions and represents the
reactions where the oxygen breaks up the
hydrocarbon molecules to produce carbon oxides and
water. These are the traditional combustion reactions.
In many of the high gravity, light oil reservoirs, bond
scission reactions occur in the range from 230°C to
300°C, while in heavier oils, bond scission reactions will
not be the dominant oxidation reactions at
temperatures below about 450°C. Bond scission
reactions are critical to the success of HPAI as well as
all other air injection-based IOR processes.
The second possible set of reactions between oxygen
and hydrocarbons are called “oxygen addition”
reactions. These reactions constitute the dominant
oxidation reactions at temperatures below about
150°C for light oils and below approximately 300°C for
heavy oils. Oxygen atoms are chemically bound into
the molecular structure of the liquid hydrocarbons,
producing immobile oxygenated compounds such as
hydroperoxides, aldehydes, ketones and acids. The
compounds may undergo bond scission reactions if
they are exposed to elevated temperatures or they
may undergo polymerization, forming heavier, less
desirable oil fractions.
Oxygen addition products such as hydroperoxides
may undergo auto-decomposition reactions which
can damage well bores. They are also not effective at
mobilizing oil ahead of the elevated temperature zone
as the reactions are consuming oxygen but not
generating gas phase products. This causes the pore
pressure where the addition reactions occur to drop
below the flowing gas pressure which promotes
trapping of the oil phase. Oxygen addition reactions
also result in shrinkage in the volume of flue gas
available for displacement downstream of the elevated
temperature zone.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
The key to the long term success in any air injection
process is to design the operating strategy and well
layouts to ensure operation in the bond scission mode.
Fortunately, for many high gravity, high pressure, light
oils, the bond scission reactions are the favored
reaction path and air injection processes operate in
the bond scission mode over a wide range of operating
conditions. While oxygen addition reactions are not
desirable during the steady state operation of an air
injection process, these are the reactions which are
responsible for spontaneous ignition. They also enable
oxygen consumption within the un-heated portion of
the reservoir. This is beneficial from the viewpoint of
un-reacted oxygen reaching the production wells.
Air Injection Processes: In Situ Combustion
vs. High Pressure Air Injection
Air injection based oil recovery processes were
historically referred to as “in situ combustion”. For the
sake of definition,“in situ combustion” or “fireflooding”
is assumed in this paper to refer to air injection based
processes applied in heavy oil reservoirs that require
operation in the high temperature range (450°C plus)
for effective displacement of the oil by the oxidation
zone. “High Pressure Air Injection or HPAI” implies air
injection into deep, light oil reservoirs, for which bond
scission or combustion reactions are dominant in the
range from 150°C to 300°C (low temperature range).
Many candidate reservoirs for air injection processes
show oxidation characteristics that fall between those
described above and it is for this reason that laboratory
testing of the oxidation characteristics is a highly
recommended step in the design of a new air injection
process.
capacity is primarily dependent on the volume of
reservoir to be serviced (based on a minimum air flux
concept), and the desired oil production rates. If
existing wells are to be used, the air injectors should
be equipped with new tubing and monitoring of the
annulus should be conducted to ensure that the air is
not entering and corroding the casing. Adequate
drying of the injected air is required if HPAI is to be
operated in a high humidity environment.
The product gas composition should, as a minimum,
be analyzed for oxygen, nitrogen, carbon dioxide, and
carbon monoxide. A gas chromatograph using a
helium gas carrier is recommended for periodic
measurements of the composition of the product
gases. Carbon dioxide and nitrogen are the key
components with respect to interpreting the state of
the oxidation reactions. Continuous oxygen analyzers
on the production wells are recommended during the
early life of an air injection process as the produced
gas may contain a significant concentration of stripped
solution gas. Some air injection projects produce
significant quantities of condensate, hence periodic
analysis of the product gas for hydrocarbon
components is also recommended.
Field Screening of HPAI Candidates
There are a number of screening guides published for
in situ combustion projects in heavy oil reservoirs and
they have very limited applicability for an HPAI project
in a deep, light oil reservoir. The parameters which are
considered to have the greatest significance with
regard to the implementation of a HPAI process
include:
DESIGN CONSIDERATIONS
Oil Reactivity
One of the main advantages of air injection as a
secondary or tertiary process is that the existing
infrastructure can be utilized. The main addition to the
existing infrastructure is the air compressors. They
should be designed for output pressures which are just
below the fracture pressure, and should utilize
synthetic lubricants. Selection of the air injection
Oil reactivity is an extremely important as the
effectiveness and ease of operation of an air injection
project depends on the oxygen being reacted through
bond scission or combustion reactions to form carbon
oxides. It is not desirable to distribute un-reacted
oxygen throughout the non-heated portion of a
reservoir as the oxygen addition reactions can result
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
13
Technology Platform: RESERVOIR ENGINEERING
in undesirable compositional changes to the native oil
and may lead to safety considerations associated with
the formation of hydroperoxides or the entry of unreacted oxygen into the production wells and facilities.
Pressure has a significant effect on the oxygen uptake
versus temperature behavior of an oil, hence the oil
reactivity should be measured at pressures
corresponding to field operation.
Experience has shown that some crude oils do not
exhibit the oxidation characteristics which are required
for a successful air injection process. Experience has
also shown if an oil does not exhibit stable burning
characteristics in the laboratory, it will almost certainly
not perform well in the field.
Reservoir Geology
Considerations on the reservoir geology are essentially
the same for an air injection process as they would be
for any gas injection project. While reservoir geology
has often been blamed for the failure of in situ
combustion projects in heavy oil reservoirs, many of
these failures were a result of improper ignition of the
combustion zone.
Reservoir Temperature
Reservoir temperature is a key parameter in terms of
the ease of ignition of a HPAI process and in terms of
the re-ignition of a HPAI process where miscibility is
achieved. In this case, the displacement efficiency of
the miscible bank can reduce the residual hydrocarbon
concentrations to levels that will starve the reaction
zone of fuel. Re-establishment of a new reaction zone
at a downstream location will only be achieved if the
combination of residual saturation, temperature, and
air flux is sufficient for spontaneous re-ignition. As a
rule of thumb, an initial reservoir temperature in excess
of 85°C is desirable for the application of spontaneous
ignition.
14
Technology Cluster: OIL AND GAS
Oil in Place at Start of Air Injection
As is the case for all enhanced oil recovery processes,
the oil in place at the start of air injection (reflected by
the product of porosity and initial oil saturation) is a
key economic parameter as it has a direct impact on
the incremental and cumulative injected air/produced
oil ratios. This is especially true for projects where
displacement by the thermal front is significant and
less so if the HPAI is designed as a re-pressurizationflue gas flood.
Phase Behavior
For light oils, the oil reactivity at a given temperature
is strongly impacted by the phase behavior. Bond
scission reactions which generate carbon oxides, and
which show maximum rates at approximately 280°C,
take place in the vapor phase. In order to initiate this
vapor phase combustion, the hydrocarbon
concentration in the vapor phase must fall within the
flammable range when the reaction temperature is
sufficiently high to enable ignition. This interaction
between oil reactivity and phase behavior explains
why oxidation behavior observed at low pressures
appears to be significantly different from that seen at
elevated pressures.
Phase behavior of the oil in the non-heated portion of
the reservoir is equally important due to its impact on
the effectiveness of the flue gas flood. Phase behavior
data are required to tune the compositional simulators
which are often used to predict the oil recovery
performance associated with the flue gas flood in the
non-heated portions of the reservoir.
Existing Infrastructure
If HPAI is be used as a tertiary recovery process, the oil
in place at the time of initiation of HPAI will often not
warrant the installation of new wells. While this is not
the case if air injection is to be used as a secondary
process in preference to a water-flood, the economics
of any air injection project are sensitive to the number
of new wells which must be drilled.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Laboratory Screening of Candidate Reservoirs
Design of both facilities and operating strategies for
HPAI project can be significantly enhanced through a
comparatively minor investment in laboratory studies
and field simulations in the early stages of the
planning. A number of laboratory screening tests have
been developed to access the potential of reservoirs
as candidates for air injection. Based on three decades
of air injection research and field design experience
at the University of Calgary, three separate tests are
considered to be essential in order to provide
fundamental screening and economic parameters and
to provide data for simulation studies as well as field
implementation and monitoring. These tests, detailed
below, are normally referred to in terms of the
apparatus utilized: Combustion Tube (CT), Accelerating
Rate Calorimeter (ARC) or Pressurized Differential
Scanning Calorimeter (PDSC) depending on the
anticipated reservoir pressure, and Ramped
Temperature Oxidation (RTO). A variety of other
testing, also described below, may be warranted to
address specific reservoir situations.
Combustion Tube (CT)
This is the prime device which provides an overview
of the process when viewed in a one-dimensional
elemental model. The historic application of
combustion tube tests was to measure the air required
and oil consumed as the combustion zone swept a unit
volume of reservoir. These design numbers are still
important, but the tests also provide information on
the burn stability, oil and water production rates as a
function of the location of the combustion front,
compositions of the produced gases, properties
(density, viscosity) of the produced oils, pH of produced
water (corrosion considerations), specific ion content
of produced water (sulfate concentration is a good
indicator of the approach of the reaction zone to the
exit of the core). While the experiments are onedimensional and not scaled, experience has shown
that the product gas compositions during stable
combustion tube tests are very similar to those
observed in the field if the oxidation reactions in the
field project are in the bond scission mode. This does
not mean that all combustion tube tests are successful,
as some oils will not sustain a propagating reaction
zone.
Accelerating Rate Calorimeter (ARC)
The accelerating rate calorimeter measures the
intensity (self heating rate) of the oxidation reactions
as a function of temperature for a given crude oil under
elevated pressure conditions. Two ARC units are in
operation. One is set up for non-flow testing while the
second unit has been modified to allow operation at
constant pressure. The operating pressure and
temperature limit of the ARC units is 41.4 MPa and
500°C respectively.
Conventional Differential Thermal Analyzers
Conventional thermal analysis apparatus including
thermogravimetric (TGA), differential thermal (DTA),
differential scanning calorimeter (DSC) and
pressurized differential scanning calorimeter (PDSC)
are all used to fingerprint oils. These techniques can
be used over a wider temperature range than is
possible with the ARC units, but the upper pressure
limit for the PDSC is limited to 10.3 MPa.
A very important observation which has arisen from
the ARC and thermal analyzer research is the existence
of both liquid and vapor phase reactions over the
nominal temperature range from 180°C to 300°C. As
mentioned previously, this is often refered to as the
low temperature range and it is at temperature near
280°C that light oils exhibit maximum energy
generation and carbon oxide production.
Ramped Temperature Oxidation (RTO) Apparatus
The ramped temperature oxidation (RTO) apparatus
utilizes a one dimensional core, so it is possible to
conduct tests on re-constituted cores under conditions
of changing air injection rate. The unit is equipped for
evolved gas analysis; hence the RTO enables direct
determination of the oxygen uptake rate as a function
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
15
Technology Platform: RESERVOIR ENGINEERING
of temperature starting at the actual reservoir
temperature. The low temperature oxidation rate
versus temperature data yield the Arrhenius
parameters required to predict spontaneous ignition.
RTO tests may be operated at different air injection
rates; hence it is possible to focus a given run on either
the low or high temperature range. Tests are often
terminated at a pre-selected heater temperature
which enables post test core analysis of the
composition and mass of residual hydrocarbon. Much
of the past work in our laboratory has been conducted
using a low pressure apparatus (8.3 MPa), but a new
system is nearing construction which will enable
operation to 41.5 MPa. This unit will be equipped with
a core flood type collection system to allow for
observation of the nature of the hydrocarbon involved
in the vapor phase reaction.
The four apparatus described above are routinely used
to evaluate the expected “burn” performance for an
air injection process in a specific reservoir. Light oil
reservoirs are often further evaluated using
conventional tests that would normally be conducted
to evaluate flue gas or miscible solvent injection
candidate reservoirs. This reflects the fact that oil
recovery from the unheated portion of the reservoir is
much more significant for light mobile oils than for
heavy oils or bitumens. The non-thermal experimental
capabilities which are utilized in our laboratories
include:
Bomb Calorimeter
Numerical simulators used for predicting the field
performance of air injection processes require heating
values for the hydrocarbon groups or pseudocomponents which are involved in the assumed
oxidation/combustion models. Heating values are
measured on whole crude oils and their fractions
(solubility separated pseudo-components like
maltenes and asphaltenes and/or on SARA fractions
saturates, aromatics, resins, and asphaltenes as well as
on the coke deposited by the oil under tests. Heating
value determinations are also performed on oils which
have been pre-oxidized under prescribed conditions
16
Technology Cluster: OIL AND GAS
as well as on the modified oil pseudo-components or
SARA fractions.
High Pressure Core Flood
Conventional high pressure core flood apparatus
which is currently equipped with a 2.4 m consolidated
Berea core. This apparatus is currently being utilized
to determine the pore level displacement efficiency
of flue gases which would be generated during an HPAI
operation. Core flood combined with combustion
tube tests allow for the estimation of the relative
contributions to oil production which are due to the
growth of the elevated temperature zone as compared
to that associated with the flue gas flood in the
unheated portion of the reservoir. The core flood unit
can use stacked core plugs as is equipped for density
and viscosity measurements of the produced fluids at
the operating pressure of the test.
Slim Tube Apparatus
The slim tube apparatus is used to determine the
minimum miscibility pressure. This information is
required to predict if the generated flue gas will
achieve miscibility at the temperature and pressure
conditions expected in the unheated portion of the
reservoir. The question of miscibility is important as it
impacts the amount of flue gas which must be
generated (i.e. the size of the elevated temperature
region) and it controls the amount of residual
hydrocarbon available in the region where miscibility
is achieved. If the flue gas achieves miscibility, there
may not be sufficient residual hydrocarbon to sustain
the oxidation reactions. If this occurs, oil reactivity at
the reservoir temperature is very important, as a new
reaction zone must be established by spontaneous
ignition near the downstream extent of the miscible
flood front.
Phase Equilibria Cells
Visual phase cells have been constructed for operation
at pressures up to 138 MPa. These units were primarily
designed to support CO2 flood or sequestration
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
projects, however, as mentioned previously, phase
behavior has a significant effect on the effectiveness
of an HPAI project.
CONCLUSIONS
Air Injection is an effective IOR method that can be
applied as a secondary or tertiary recovery process in
on and off shore reservoirs. When applied in tight,
deep, light oil reservoirs, the terminology used is “High
Pressure Air Injection” which is the focus of this paper.
HPAI can be operated as an in situ flue gas generator
or it can be designed to benefit from the increased
displacement efficiency of the propagating
combustion zone. The key to success of any air
injection process, HPAI not withstanding, is to ensure
that the crude oil, under the conditions imposed by
the reservoir and air injection strategy, will sustain
bond scission or combustion reactions. Laboratory
tests are available to screen potential reservoirs and
to provide the parameters required for the design of
the facilities and operating strategies for a successful
HPAI project.
REFERENCES
[1] Erikson, A., Legerski, J.R., and Steece, F.V.,“An Appraisal of High
Pressure Air Injection (HPAI) on the In Situ Combustion Results
from Deep, High Temperature, High Gravity Oil Reservoirs,”
Presented at the 50th Anniversary Field Conference of the
Wyoming Geological Association Guidebook, 1994.
[2] Kumar, V.K., Fassihi, M.R., and Yannimaras, D.V., “Case History
and Appraisal of the Medicine Pole Hills Unit Air Injection
Project, SPE/DOE 27702, presented at 1994 SPE/DOE Ninth
Symposium on Improved Oil Recovery, Tulsa, OK, April 1994.
[3] Sarathi, P., and Olsen, D., ‘Field Application of In Situ
Combustion-Past Performance /Future Application”, NIPER/
BDM-0086, CONF-940450 (DE95000116), Bartlesville Project
Office, U.S. Department of Energy, Bartlesville, Oklahoma.
[5] Sarma, H.K., Yazawa, N, Moore, R.G., Mehta, S.A., Okazawa, N.E.,
Ferguson, H.A., and Ursenbach, M.G.,“Screening of Three LightOil Reservoirs for Application of Air Injection Process by
Accelerating Rate Calorimetric and TG/PDSC Tests”, Journal of
Canadian Petroleum Technology (JCPT), Vol. 41, No.3, pp.5061, March 2002.
[6] Juan, E.S., Sanchez, A., del Monte, A., Moore, R.G., Ursenbach,
M.G., and Mehta, S.A., “Laboratory Screening for Air InjectionBased IOR in Two Waterflooded Light Oil Reservoirs,” Paper
No. 2003 - 215, Proceedings of the Canadian International
Petroleum Conference (CIPC) of the Petroleum Society, Calgary,
Alberta, 10 - 12 June 2003.
[7] Takabayashi, K., Yazawa, N., Moore, R.G., Mehta, S.A., and
Ursenbach, M.G., ”Evaluation of Light Oil Air Injection Process
by Combustion Tube Tests,” Proceedings of 24th Annual
Workshop & Symposium, Collaborative Project on Enhanced
Oil Recovery, International Energy Agency (IEA), Regina,
Saskatchewan, 7 - 10 September 2003.
R. Gordon Moore is a professor and head
of department of chemical and petroleum
engineering at the University of Calgary. He
graduated from the University of Alberta
with a BSc (1965) and a PhD (1971) in
chemical engineering. During his years at
the University of Calgary, he has built a close
alliance with Alberta’s petroleum industry,
and has spent sabbatical years working with
Imperial Oil Resources, Mobil Oil Canada, and Hycal Energy
Research Laboratories. He has balanced his research
commitments with his teaching responsibilities and has received
numerous teaching awards from the University of Calgary Student
Council and the Engineering Students Society. Prof. Moore’s prime
research area has been the improved recovery of conventional
and heavy oils and bitumens, with particular emphasis on oil
recovery by in situ combustion. He is head of the In Situ
Combustion Research Group at the University of Calgary, and
continues to provide consulting services worldwide in the area
of in situ combustion and other thermal recovery processes. He
was recognized by the Society of Petroleum Engineers
International (SPE) as a Distinguished Member for his
contributions to IOR and was SPE Distinguished Lecturer in 2001.
He is an active member of the Petroleum Society, SPE, APEGGA,
AIChE, CSChE, and CHOA, and is a registered professional engineer
in the province of Alberta.
[4] Moore, R.G., Mehta, S.A., and Ursenbach, M.G., “An Engineer’s
Guide to High Pressure Air Injection (HPAI) Based Oil Recovery,”
SPE Paper No. 75207, Proceedings of the SPE/DOE 13th
Symposium on Improved Oil Recovery, Tulsa, OK, 13-17 April
2002.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
17
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
A HOLISTIC APPROACH TO IOR
Brian GD Smart
Institute of Petroleum Engineering,
Heriot-Watt University, Edinburgh, Scotland, UK.
[email protected]
ABSTRACT
IOR in its purest sense is really about improving recovery factors, and has now probably gone beyond the search
for a “silver bullet” that will make step-change improvements to recovery, with the exception of heavy oils. The
most appropriate and credible approach to IOR is to integrate all the best practice available, while staying
connected to the research community that continues to make improvements to the state-of-the-art. Hence IOR
is now really an exercise in multi-disciplinary knowledge acquisition and application, as this paper attempts to
demonstrate by introducing contributions made by the Institute of Petroleum Engineering, Heriot-Watt University,
to reservoir characterisation, modelling and management.
Keywords: IOR, reservoir characterisation, modelling, reservoir management
A HOLISTIC APPROACH TO IOR
IOR is achieved by improving the recovery factor of a
reservoir. Ultimately the best way to achieve this must
be to use a holistic approach that properly integrates
the contributions that all disciplines involved in
Petroleum Engineering can make, melding best
practice and, where appropriate, innovations derived
from research. A generic workflow for this process is
shown in Figure 1, together with topics that the
Institute of Petroleum Engineering at Heriot-Watt
University has made significant contributions to. The
fundamental principle that underpins the steps in the
process is that the best decisions are based on a
realistic appreciation of all factors influencing that
decision. For example, in otherwise relatively
homogenous reservoirs, thermally-induced stresses
created by the injection of cold water and the
anisotropic stress-sensitive response of the reservoir
will influence flood front migration and sweep
efficiency. Early water breakthrough may occur if this
phenomenon is not recognised and allowed for in
reservoir development and management.
Examples of work coming out of Heriot-Watt that
contribute new knowledge and methodology for
reservoir characterisation and reservoir modelling are
given below. As well as serving the needs of bringing
each task up to state-of-the-art capability, the power
of integrating that capability should also be obvious.
And so consider the following steps in the integrated
process – reservoir characterisation for sedimentary
processes and stress sensitivity, model building using
upscaling techniques, coupled modelling, simulation
combined with 3D visualisation to assist in
interpretation and decision making, enhanced
recovery using WAG, and reservoir monitoring using
well testing.
Reservoir Facies Modelling based on Sequence
Stratigraphy (Genetic Sedimentary Units) and
other Data derived from Reservoir Analogue
Outcrop Studies
This work was started at Heriot-Watt in 1992, to address
the need identified by the Petroleum Industry for
quantitative data on sedimentary systems, and for use
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
18
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 1: The Holisitic Process of Improving Recovery Factor (* Topics that Heriot-Watt is researching)
in supplementing well data in the building of
stochastic reservoir models for flow simulation. The
quantitative data collected from reservoir analogue
outcrops, using specially-developed photographic and
image processing methods, is used by BP, BG,
ChevronTexaco, ConocoPhillips, Exxon-Mobil,
Petrobras, Shell, Statoil, Unocal, and EnCana amongst
others. As well as providing data to the oil companies,
the work has more recently been extended to model
building and simulation as shown in Figure 2, the
supplementary well data required being provided by
oil companies. These models are being used to address
production-related issues such as the optimisation of
reservoir management.
Upscaling of Facies Permeabilities for
Model Building
In a number of related projects carried out over the
last 12 years, the effects of reservoir heterogeneity and
flow processes upon oil recovery have been examined.
In this work, the types of heterogeneity that matter
under particular recovery schemes have been
determined. Lithological heterogeneities are present
at a variety of length scales and, during the initial phase
of the work, the work concentrated on examining the
effects of small-scale structures, such as lamination,
which occur in many clastic rocks at scales of mm-cm
as shown in Figure 3. Through numerical simulations
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
19
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 2: The collection, analysis and use of data from reservoir outcrop studies
Figure 3: The Upscaling Challenge – How is the real reservoir, with small scale attributes that influence fluid flow, represented at the
simulator scale?
20
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
The Geomechanical Appraisal of Reservoirs
Figure 4: Upscaling from Plug Scale to Reservoir Simulation
Model Scale
and experiment, it was verified that such structures
may have a significant effect on recovery, due to the
combination of high permeability contrasts between
the laminae and to capillary forces which are dominant
over small length scales. Subsequently upscaling
methods were developed so that the effects of smallscale structures can be included in larger-scale
simulations.
In the second phase of the research, and probably
more appropriate to IOR, Heriot-Watt extended
upscaling, to greater length-scales. The Geopseudo
Toolkit was developed, providing a software package
to help sponsors analyse data, create permeability
models and perform scale-up through the ranges
shown in Figure 4. Worked examples are contained in
a GeoFlow Atlas. Practical guidelines for upscaling
have also been produced as an aid to transferring this
capability to Industry.
The oil companies using this approach to upscaling
include BG, Conoco, Shell, Statoil, Petrobras and
PETRONAS.
The background knowledge and methodology
required to create a geomechanical model of the
reservoir has been developed progressively over the
last 15 years. The approach is based on a conceptual
model of the reservoir involving intact rock and
discontinuities, loaded with an anisotropic stress field.
A comprehensive workflow managing the acquisition
of all the data for stress-states, intact rock properties
and discontinuity properties has been devised and
applied, most notably to one of the world’s largest
reservoirs. The workflow is economic with regard to
the effort required, as correlations are sought between
normal petrophysical properties and key rock
mechanical properties. Optimum use is made of
wireline logs, selective sampling of core and laboratory
testing. The core sampling for testing is driven by rocktype and porosity variations revealed by the logs,
effectively dividing the rocks into mechanical units. In
order to produce a necessary and complete data set
for some stress-sensitive phenomena however, the
overburden and the under burden must also be
characterised. Creating the Geomechanical Appraisal
is the first step in examining the stress-sensitivity of
the reservoir using coupled modelling, or in using
stress-analysis modelling in the simulation of the
tectonic processes applied to the reservoir. Reservoir
stress-sensitivity is recognised by ExxonMobil and BP
among others.
The Coupled Modelling of Reservoirs
In coupled modelling, stress-sensitive permeabilities,
fault transmissibilities and/or acoustic impedances are
accommodated by combining the capabilities of a
conventional fluid flow simulator with a stress-analysis
simulator using proprietary interfaces. Production of
hydrocarbons can alter the stress state in the reservoir,
either by effective stress changes related to pore
pressure changes, or thermal stresses related to
injection of water. These changes in stress state in turn
trigger other effects such as permeability changes in
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
21
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 5: Coupled Reservoir Modelling – Deformations caused by production
(a) The predicted vertical displacements of overburden layer prior to production (the middle layer of Paleocene formation) and the
directions of faulting (the black rectangles). The simulated movements matched the obverted faults in terms of location and
direction.
(b) Pressure change in reservoir layer due to depletion.
(c) Compaction in reservoir layer due to depletion. The compaction was related to the pressure changes and the presence of faults
(d) The predicted shear strain of overburden layer due to depletion (the middle layer of Paleocene formation), which indicated the
re-activation of faults.
the matrix, or large deformations in the overburden.
Given the appropriate data set – the Geomechanical
Appraisal – Coupled stress/strain analysis and fluid
flow models can predict the impact of the productioninduced stress changes on reservoir performance by
feeding mechanically-induced controls on fluid flow
into the fluid flow model. Figure 5 shows the results
22
for such a coupled model, the software being used is
VISAGE TM . The impact of depletion on reservoir
pressure distribution and fault reactivation is
demonstrated in this model. This approach to adding
realism to reservoir modelling has been adopted by
ExxonMobil and BP.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 6: Micromodel studies revealing the effect of alternating water and gas injection on oil displacement from the pore spaces
Numerical Well Test Analysis,
Linked to Realistic Reservoir Models
This capability, developed in the first instance
specifically for fluvial and turbidite reservoirs, and
sandstone channels, uses a simulation model of the
reservoir being studied to produce synthetic well test
response curves for comparison with the real well test
data, providing another method for “ground truthing”
the reservoir model. This technology is used for
examining production log data and for history
matching selected wells, and possibly also as a
justification for some well tests specially
commissioned in clastic reservoir intervals. The
companies adopting this approach to well test analysis
include Shell, Phillips, Wintershall and Yukos.
Secondary Recovery Using WAG
The use of WAG (water alternating gas) injection can
potentially lead to improved oil recovery compared
to injection of either gas or water alone, however the
physical process is not well understood, and this is
where unique glass micromodels capable of operating
at high pressure have yielded knowledge of the porescale fundamentals involved. The micromodels, 2.5D
representations of the pores and grains of a reservoir
rock etched in glass and driven by a specially designed
fluid displacement and pressure control system, yield
stunningly clear images of the fluid phases and the
displacement processes that occur during the various
stages of a WAG process. Experiments can be
conducted using either water-wet or oil-wet
micromodels. The results presented here in Figure 6,
with an analysis in Figure 7, are for a water-wet
micromodel. The micromodels were initially fully
saturated with water and then displaced with oil to
establish the connate water saturation. They were then
flooded with alternate cycles of gas and water injection
to observe the three-phase flow/displacement process
and the associated oil recovery. The experiments were
performed within the capillary dominated flow regime,
with the flow processes being filmed and electronically
stored using high resolution imaging. Oil recovery at
the end of each flooding cycle were measured using
image analysis techniques.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
23
Technology Platform: RESERVOIR ENGINEERING
Figure 7: Oil recovery at each stage of the experiment as
percentage of initial waterflood trapped oil
The results highlighted the importance of corner
filament flow of water in the recovery process, with the
initial waterflood residual oil saturation being trapped
mainly in the centre of the majority of the pore space,
surrounded by films of water coating the water-wet
pore walls. The successive WAG cycles redistributed
the oil saturation in a way which resulted in improved
oil recovery, with clear evidence that the alternate use
of the water and gas injection mobilised oil that would
not have been mobilised under either gas or water
injection alone. The micromodel revealed that a
limited number of WAG cycles were required to
approach a maximum achievable oil recovery, as
shown in Figure 7.
Visualisation Laboratory
The Visualisation Laboratory being assembled in the
Institute of Petroleum Engineering combines super
computing “number crunching” power (ultimately a
256 node cluster) for conventional and coupled
reservoir simulation with the specialist graphics
computational power, e.g. that supplied by SGI,
required for 3D visualisation. This facility will enable
reservoir models to be run and visualised at the same
time, allowing “what if” type scenarios to be run and
examined in “real” simulation time. Thus multiple
scenarios and scenario variation can be handled, with
the impact of changes being immediately revealed in
the 3D images. New types of images, or other
24
Technology Cluster: OIL AND GAS
complimentary means of communication and data
pesentation will also be derived – for example, what is
the best way to view a reservoir attribute jointly with
the uncertainty assigned to that attribute? This facility
will be a powerful means of exploiting the full power
of the more realistic, detailed models being built –
fundamental to IOR! In addition, since rising to the
challenges and opportunities presented, the
effectiveness of multi-disciplinary team working will
be of key importance. Hence research into multidisciplinary team working will also be conducted using
the laboratory.
CONCLUSIONS
A holistic multi-disciplinary approach to petroleum
engineering has been suggested as a credible,
attainable means of achieving IOR. Using examples
from the research output of the Institute of Petroleum
Engineering, it has been further demonstrated how the
multi-disciplinary approach can be facilitated and kept
abreast of the most recent developments.
ACKNOWLEDGEMENT
The author gratefully acknowledges the work of
colleagues within the Institute of Petroleum
Engineering described in this paper. In particular, Dr
Andy Gardiner, Dr Gillian Pickup, Dr Jim Somerville, Prof
Ali Danesh, Prof Dabir Tehrani, Prof Patrick Corbett, Dr
Shiyi Zheng and their respective research groups. The
supply of coupled modelling output from VIPS Ltd is
also acknowledged.
REFERENCES
[1] Readers are referred to the Institute of Petroleum Engineering
web site (www.pet.hw.ac) and its pages devoted to research
for any further information required.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Brian Smart is Deputy Principal (Academic
Development) for Heriot-Watt University,
and Director, External Affairs, for the
Institute of Petroleum Engineering. Brian
Smart began work as an apprentice mine
surveyor, before being awarded an NCB
scholarship to study mining engineering at
Strathclyde University. He gained a 1st, and
continued to do a PhD.
After earning his PhD degree in 1973, Dr. Brian Smart was
appointed as lecturer in the Department of Mineral Resources
Engineering, University College Cardiff. In 1982, he returned as
lecturer to the Department of Mining and Petroleum Engineering
at the University of Strathclyde, where he developed a Rock
Mechanics Research Group comprising typically 9 personnel
working on coalface powered support and tunnel support
specification and design.
At Strathclyde, Dr. Brian Smart was promoted to senior lecturer
and professor. Professor Smart transferred with much of the
research group, including one academic colleague, and lab to the
Department of Petroleum Engineering at Heriot-Watt University
in 1990. In August 1996, Prof. Smart was appointed Head of the
Department of Petroleum Engineering at Heriot-Watt University
for three years, and was re-appointed for another three years. He
then held the same appointment for another year until August
2003, covering the transition of the Department to an Institute.
He has worked with his colleagues to maintain the reputation of
the Department, to further internationalise the Department, and
to introduce significant new research topics, including Hydrates
and Geophysics. Ongoing research projects address rock testing
technology, coupled rock mechanical/fluid flow modelling, and
poroelastic constants and petrophysical properties in anisotropic
stress fields. The work has been funded by the DTI, EPSRC, NERC,
EEC, (i.e. by Government) and industry, and has been extensively
published in reports and papers.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
25
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
THE DEVELOPMENT OF AN OPTIMAL GRID COARSENING
SCHEME: INTERPLAY OF FLUID FORCES AND HIGHER
MOMENTS OF FINE-SCALE FLOW DATA
N. H. Darman1, G. E. Pickup2 and K. S. Sorbie2
1
Petronas Research and Scientific Services Sdn. Bhd., Selangor Darul Ehsan, Malaysia.
2
Heriot Watt University, Edinburgh, Scotland, UK.
1
[email protected]
ABSTRACT
The accuracy of upscaling procedures can be improved by using non-uniform grid cells at the coarse scale level.
Several researchers have investigated objective methods of selecting the optimal coarse grid configuration for
a particular fine grid model 1-2 . However, all of these works neglect the effect of fluid force balances and always
assumed the model to be in viscous dominated flow. This paper describes a new optimal grid-coarsening scheme
for two-phase flow in porous media based on the quantitative use of fine-scale simulation data. The main idea
of this approach is to use fine-grid fluctuating moments to guide the choice of the coarse grid structure. These
quantities are derived from the volume average saturation equation for different fluid force balances i.e. viscous,
gravity and capillary. It is shown that this approach results in a more accurate prediction of quantities such as
total oil recovery and fluid production ratio in coarse grid models. Here, we explore the relationship between
the sub-grid properties described above, and coarse scale numerical simulations for several synthetic model
problems. To do this, we first introduce capillary terms into the volume averaging equations [3] to allow us to
assess the coarse scale simulations when capillary pressure is important in the fine grid models. Then we proceed
to the two-stage testing of the proposed technique to study the relationship of sub-grid variability to the error
in the coarse scale simulation results. In the first stage testing, we consider many different aggregations of 2D,
20-layer fine grid systems to equivalent 2-layer coarse grid systems and in the second stage testing, we then
apply our findings to more heterogeneous interbedded sand cases where capillary forces are more important
and consider coarsening the fine grid model to more than 2 layers.
Keywords: Upscaling, balance of fluid forces, simulation, optimal grid coarsening scheme
INTRODUCTION
In modern reservoir characterization, the spatial
resolution that may be incorporated into geological
models often exceeds the computational capabilities
of fluid flow simulators by a significant margin.
Therefore, some level of upscaling must be applied to
the fine scale geological models before they can be
used for practical flow calculations. This upscaling may
be a simple block averaging of the single-phase
permeability or it may involve the application of a
complex upscaling procedure.
In this paper, we presented a novel PETRONAS
coarsening procedure which utilizes balance of fluid
forces as its framework. This method differs from
conventional coarsening schemes in that it takes into
account not only the static properties of the fine grid
models but also the dynamic properties of the fine grid
simulation runs. This in turn can lead to more accurate
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
26
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
predictions of important quantities such as the
recovery factor, water-cut, pressure and gas-oil ratio.
To apply this coarsening method, we identify regions
of the fine grid models where there is low variability
of certain fluctuating fine grid parameters, and take
such regions as the corresponding coarse grid blocks.
Consequently, the final coarse grid may include both
finely and coarsely gridded regions (non-uniform
coarsening).
In this paper, we show that different fluctuating subgrid quantities lead to different coarsening strategies
depending on the force balance in the system. For a
viscous dominated flow regime, (i) velocity-saturation
covariance (v‘S‘) and (ii) variance of saturation (S‘S‘)
most strongly affect the coarse grid structure. In
gravity dominated systems, the key sub-grid moments
are (i) (S ‘S ‘) and (ii) covariance of permeabilitysaturation (k‘S‘). In the capillary dominated regime,
∂S‘
the appropriate sub-grid properties are (i) S‘
∂x
∂S‘
(ii) k‘
and (iii) S‘S‘. In this paper, we focus on cases
∂x
where viscous and capillary forces are predominant
as we have previously described cases where only
viscous and gravity forces are present [3].
For illustrative discussion, linear water floods were
simulated to show the influence of fluid forces to the
fluid flow in porous media. For this reason, we
constructed a simplified layered model of alternating
high and low permeability layers (permeability
contrast of 100). In addition, we assigned high capillary
pressure table at the low permeability layers and low
capillary pressure table to the high permeability layers.
Figures 1 and 2 show the water saturation distribution
predicted by the simulator when viscous or capillary
pressure dominates the flow respectively. To mimic
the required changes in fluid forces inside the model
i.e. from viscous force to capillary force, we reduce the
injection flow rate by a factor of 20. In Figure 1, it is
shown that when viscous force dominates the
displacement (high injection rate), water channels
through the high permeability layers, leaving much of
the oil in the low permeability layers.
If the flow dominates by capillary forces as shown in
Figure 2, the flow characteristic changes completely.
Water is now imbibed into low permeability layers and
leaving much oil in the high permeability layers. Due
to effect of capillary pressure, water also advances
BALANCE OF FLUID FORCES
Immiscible fluid flow in porous media is controlled by
three types of forces i.e. (i) viscous, (ii) capillary and (iii)
gravity. This force may act by its own or combine with
one or the other two forces simultaneously. The
existence and interplay of these fluid forces depend
on several phenomena such as scale of the model,
thickness of the layers, density difference of the in-situ
and displacing fluids etc. At small scale model e.g.
lamina scale of 1-50 cm, both viscous and capillary
forces may dominate the displacement. At larger scale
e.g. simulation grid of 10-100 m, viscous and gravity
forces may affect the flow significantly. Obviously, cross
over regimes of these forces must exist at same scale
as we progress our study from small scale to a larger
scale model (or vice versa).
Figure 1: Saturation distribution for viscous dominated flow
Figure 2: Saturation distribution for capillary dominated flow
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
27
Technology Platform: RESERVOIR ENGINEERING
faster in low permeability layers than in high
permeability layers. In this case, initially near the
injector, the difference in pressure due to viscous force
makes the water to enter the high permeability layers
more than the low permeability layers. However, the
undisturbed oil in the low permeability layers are then
contacted with water zones in the high permeability
layers. Pressure differential created at the interface of
the two fluids, due to difference in capillary pressure,
then force the water to imbibe into the low
permeability layers. As the results, premature water
breakthrough happens, and leaves much of the oil in
the high permeability layers.
As shown in these two examples, fluid forces may
significantly differ fluid flow characteristic of one
particular simulation model. Depending on fluid
displacement rate as well as other parameters that
control fluid force regime calculation, flow behavior
and thus oil recovery and fluid production ratio can
vary tremendously although when we use the same
model. As such, it is very important to capture the right
balance of fluid forces and their interplay that actually
happen in the fine grid model, and transfer these
information accurately into the coarse grid using an
appropriate upscaling technique.
Technology Cluster: OIL AND GAS
As shown in Equations (1) and (2), parameters that
affect Ncv are (i) vertical permeability, (ii) length (iii)
capillary pressure derivatives with respect to
saturation (iv) fluid total velocity (v) thickness and (vi)
oil viscosity.
On the other hand, parameters that affect Ngv are i)
vertical permeability, (ii) density difference between
the two fluids (iii) length (iv) fluid total velocity (v) oil
viscosity and (vi) thickness.
Zhou et al. discussed these two terms to have
directional effect relating the Ncv and Ngv to the
balance of viscous forces in the horizontal direction
to capillary and gravity forces, respectively. As we can
see in these two equations, as we lower the flow rate
in the model, the importance of viscous forces will
disappear favoring gravity or capillary forces
depending on the situation.
AVERAGED SATURATION EQUATION
We previously derived the averaged saturation
equation with gravity included as follows3:
(3)
SCALING GROUPS
The relative influence of the three forces on fluid flow
phenomenon is usually characterized in terms of
gravity-to-viscous ratio Ngv and capillary-to-viscous
ratio, Ncv. There are many techniques or formulation
available in the industry to group these forces [4-6].
For the purpose of this paper, we discuss the scaling
group recommended by Zhou et al [4].
Zhou et al. derived the capillary to viscous ratio as:
(1)
and gravity to viscous ratio as:
(2)
28
where Sg = S is the gas saturation, ˜(S) is the gas
fractional flow, η is kro.f , t is time, v is the total Darcy
velocity, ∆ρ = ρg - ρo (densities), g is gravitational
acceleration (z-dir), µo is oil viscosity, k is the local
permeability tensor, iz is the z-unit vector, kro and krg
are oil and gas relative permeabilities and fs, ηs, fss and
ηss are the 1st and 2nd derivatives of f and η with
respect to S.
The fine scale functions f and η appear explicitly and
without modification in the coarse scale equation; i.e.
“rock curves” are applied directly in the coarse grid
model, although the use of pseudo functions can
further improve accuracy at high coarsening levels
[3,7]. Also, the coarse scale equation contains three
additional terms involving higher moments that
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
account for the sub-grid effects, viz. S‘S‘, v‘S‘ and k‘S‘,
and these terms are always multiplied by derivatives
of f or η.
The volume averaged saturation equation with
capillary pressure (P c) is derived as before [3]. To
proceed, we write fine grid quantities as the sum of a
volume averaged coarse grid component (overbar)
and a fluctuating component (prime). Thus, any
_ fine
grid variable, F , can be written as Φ(x,z) = (x,z) Φ+Φ’ =
(x,z). The coarse scale saturation equation is then
derived by:
1. inserting these expressions into the fine grid
equation, volume averaging, and retaining terms
with products of two fluctuating quantities,
2. expanding terms in relative permeability and
fractional flow (with
Pc terms) around the average
_
gas saturation S ; and
3. retaining only first order terms.
when Pc dominates. From Eq. (3), we can determine
the form of the error that will result from using rock
curves (neglecting the higher moment terms) on the
coarse scale. Since such a scheme neglects all higher
moments, the error can be expected to correlate with
the magnitudes of the neglected terms.
NUMERICAL TESTING WHEN COARSENING
TO 2-LAYER COARSE GRID MODELS
Three cross-sectional models are used to determine
which higher moment is appropriate under different
fluid force balances. The first two models have 100x20
(x/z) fine grid blocks and are coarsening upwards (Case
I) or downwards (Case II) as shown in Figure 3 (k varies
from 10 mD to 1462 mD). The third model is a 100x80
(x/z) randomly populated layered model (Case III) with
two rock facies of 10 md and 1000 md as shown in
Figure 4.
This volume averaged saturation equation includes
viscous, gravity and capillary terms. Capillary terms
lead to 2 additional moments in the equation, which
may affect our coarsening strategy when P c is
important. Table 1 gives the appropriate moments
when different fluid forces act on one particular model.
Note that S‘S‘ appears in all the three force regimes;
v‘S‘ appears only in the viscous dominated case; k‘S‘
____
____
∂S‘
∂S‘
in the gravity dominated flow; and S‘.
and k‘.
∂x
∂x
Figure 3: Perm. distribution for Case I and II.
Tabl e 1: Appropriate fluctuating moments for each of the fluid
force balance regimes
Forces
Fluctuating
Moments
Comment
Viscous
S‘S‘ and v‘S‘
Have been presented
in SPE 69674.
Gravity
S‘S‘ and k‘S‘
Have been presented
in SPE 69674.
Capillary
____
∂S‘
S‘S‘, S‘.
∂x
____
∂S‘
and k‘.
∂x
Proposed in this
work
Figure 4: Perm. distribution for Case III.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
29
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
(7)
____
∂S‘
The S‘
term was calculated locally using finite
∂x
differences as follows:
(8)
Figure 5: Input J-function for fine scale models
Capillary pressure is included and scaled using the Jfunction shown in Figure 5 [8]. Note that the input Jfunction is high for a gas-oil system but this is done to
increase the effect of capillary pressure in our models.
The other fine grid properties are: ¿ = 0.2; ∆x = ∆y = ∆z
=7.63m; µo = 10 cp, µg = 0.1 cp; ρo = 700 kg/m3, ρg =
1.0kg/m3. The fluids are immiscible and the rock
relative permeabilities are as given by Guzman et al.
[9]:
and
(4)
where Swc = 0.15 (connate water), Sgc = 0.05 (trapped
gas) and Sorg = 0.1 (residual oil to gas).
and S ij (or k ij ) for the grid block saturation
(permeability). Since we have no values at the edge
of the model, these derivatives were ignored in this
region. Error is computed as the normalized root mean
square (rms) difference in the oil recovery factor
between the coarse grid and the fine grid results [10].
For each sub-grid moment, we find the coefficient of
determination (R2) between that quantity and the
resulting error. Hence, the bigger the R2 (using linear
regression) the better the results will be in terms of
the ability of the sub-grid quantity to be used as an
upscaling “error predictor”.
To simulate different balances of forces in Case I and II,
three scenarios were made for each model as
summarized in Table 2 (scenarios A, B and C). Since
results for scenarios B and C will be “mirror images” for
CALCULATION AND COARSENING PROCEDURES
Table 2: Description of the force balance scenarios
First note that we subsequently refer to v‘S‘, k‘S‘, S‘S‘,
____
____
∂S‘ and ∂S‘ as σ , σ , σ2 , σ
S‘.
k‘.
S S.(dS/dx) and σk.(dS/dx)
vS kS
∂x
∂x
A
∆ρ
(kg/m3)
700
Inj. velocity
(m/day)
0.30
B
1E-6
0.30
C
1E-6
0.06
D
700
0.30
E
1E-6
0.30
Scenarios
respectively. We focus on the last three terms since
they are thought to be important when significant Pc
effects exist in the fine grid models. The following
equations were used to calculate the three higher
moments (averaged over all the calculated time steps):
(5)
(6); and
30
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Force
balance
Yes Gravity,
capillary
and viscous
forces.
Yes Capillary
and viscous
forces.
Yes Mostly
capillary
forces - some
viscous.
No Gravity and
viscous forces
No Viscous
dominated
Pc
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
these 2 models, only Cases IB and IC are presented
(Case IIB and IIC are identical). Typically, we first
simulate the fine grid model and then generate
equivalent 2-layer coarse grid models. A 20-layer fine
grid model has 19 different coarsening combinations
that give 2-layer models (plus one 2D→1D, upscaling
option). For efficiency, only coarsening options
involving amalgamation of odd numbers of fine grid
layers into one coarse grid layer were considered.
In the x-direction, we uniformly coarsen the 100 fine
grid blocks to 20. Hence, our coarse grid models are
then 20x2. Before each of these options was run, the
values of the three sub-grid quantities were calculated
based on the saturation, grid block size and
permeability distributions of the fine grid model. Each
of the 2-layer coarse models was then run and the error
quantified as described above. As suggested by the
average saturation equation, rock curves are used
directly in the coarse models. The results are discussed
below.
TWO-LAYER COARSE GRID MODELS RESULTS
We first present results for the 20→2-layer upscaling
for Cases I and II under force balance scenarios A, B
and C (Table 2). The results in Table 3 summarize the
performance of each coarsening method for test cases
IA, IB, IC and IIA. The gradients of S and absolute
permeability are directional in nature and we therefore
calculated these quantities in the two directions
separately to find which direction is more appropriate
Table 3: R2 for the correlation of the moments vs. the coarse grid
normalized rms error in oil recovery factor for Cases IA, IB, IC and
IIA
____
∂S‘
S‘
∂x
____
∂S‘
S‘
∂z
____
∂S‘
kx
∂x
____
∂S‘
kz
∂z
S‘S‘
Case I A
0.7992
0.4564
0.7873
0.5400
0.9072
Case I B
0.5856
0.3143
0.9688
0.7074
0.9279
Case I C
0.7877
0.2394
0.9495
0.7065
0.8387
Case II A 0.8901
0.6893
0.7923
0.7772
0.8605
Cases
Average 0.7657 0.4246 0.8745 0.6828 0.8836
for application. Results in Table 3 show that the
fluctuating moments in the x-direction provides much
better R2 when plotted against the normalized rms
error compared to the z-direction. That is, the
moments in the x-direction act better as a better “error
predictor” (cf. the z-direction properties) possibly
because x-directional flow dominates in the system.
In all 4 cases, crossflow due to capillary forces and/or
gravity is relatively small compared to the flow in the
x-direction. Even for Case II A, where the fluid crossflow
is expected to be maximum, the ratio of fluid flowing
in the x-direction is over 20 times higher than the zdirection flow. Furthermore, we believe that the results
___
were mainly contributed by the ∂S‘ term, which is
∂x
“active” only at the flood front, which is generally
flowing in the x-direction. Results in Table 3 show that
____
____
∂S‘
∂S‘
the mean R2 for S‘
, kx
and S‘S‘ vs. the
∂x
∂x
normalized rms error for the 4 cases are excellent (0.77
to 0.88) showing these fluctuating higher moments
to be good error predictors when Pc is significant in
the fine grid models.
APPLICATION OF THE COARSENING SCHEME IN
MULTI-LAYERED 2D → 2D UPSCALING (CASE III)
We now consider 2D → 2D coarsening of the 80-layer
model of Case III under the various force balances in
Table 2 to determine if our coarsening approach
applies to multi-layer systems where Pc is large. In this
case, we consider scenarios A, D and E. In Case IIIA all
three forces act simultaneously, Case IIID has viscous
and gravity forces and Case IIIE has only viscous forces.
The oil recovery profiles for the three cases are shown
in Figure 6. Comparison of Cases IIIA and IIID highlights
the effect of Pc in this model. In Case IIIA, capillary
crossflow helps to improve the sweep efficiency.
Lowest recovery is seen for the viscous dominated
Case IIIE, where the helpful crossflow effects of both
Pc and gravity are absent. Figures 7, 8 and 9 show Sdistributions after 5pv injection for these 3 cases.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
31
Technology Platform: RESERVOIR ENGINEERING
Figure 6: Recovery factor versus pore volume injected for Cases
IIIA, IIID and IIIE
Technology Cluster: OIL AND GAS
Figure 10: Fluctuating moments versus no. of coarse grid layers
(S‘S‘)
Figure 7: Saturation distribution for Case IIIA (at 5 PVI)
Figure 11: Fluctuating moments versus no. of coarse grid layers
(σS.(dS/dx) and σk.(dS/dx))
Figure 8: Saturation distribution for Case IIID (at 5 PVI)
Case IIIA: We now examine the effect of 2D → 2D
coarsening in these models. Figures 10 and 11 show
the relations between the three higher moments (σ2S,
σS.(dS/dx) and σk.(dS/dx)) vs. the number of coarse grid
layers. The decreasing trend of the fluctuating
moment as the number of coarse layers increases is as
expected. However, we find that different fluctuating
moments suggested different configurations of the
coarse grid model (for same number of coarse layers).
As a result, each coarse grid model will produce
different simulation results and we must establish
which moment produces a coarse grid model that best
reproduces the fine grid results. For this purpose, we
simulate the 1, 10, 20, 30 and 40 coarse grid layers as
suggested by each of the above 3 higher moments.
Figure 9: Saturation distribution for Case IIID (at 5 PVI)
32
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 12 shows oil recovery profiles for 40-layer coarse
grid models showing that σ2S produces the best result
followed by σ k.(dS/dx), σ S.(dS/dx) and lastly, uniform
coarsening (every coarse layer contains two fine grid
layers).
Figure 12: Recovery factor vs. pore volume injected (40-layer
coarse grid, Case IIIA)
Figure 13: Fluctuating moments versus normalized rms error
(S‘S‘, Case IIIA)
Figure 14: Fluctuating moments normalized rms error ( σS.(dS/dx)
and σk.(dS/dx) , Case IIIA)
Figure 15: Normalized rms error versus no. of coarse grid layers
(Case IIIA)
Plotting the fluctuating moments versus normalized
rms error produces an interesting observation. As we
can see in Figures 13 and 14, all of the fluctuating
moments have an R 2 close to unity. For σ 2S, it is
observed that the resulting regression lines follow
exactly the same trend as the moments itself i.e. having
2 sloping lines, with cut-off around 25 coarse grid layers
as shown in Figure 10. The reason why all the three
fluctuating moments produce a good R2 versus error
perhaps is due to the fact that the coarse grid
configuration was designed by systematically
reducing the fluctuating higher moments in the fine
grid models. Although that particular moment might
not suggest the “optimal answer” for that coarse grid
configuration, it will nevertheless reduce the error
systematically as the number of coarse grid layers
increases. As a result, it makes the regression line
almost perfect when plotted against these two
parameters.
The results of Case IIIA are summarized in Figure 15,
which shows normalized rms error vs. number of
coarse grid layers for the three fluctuating moments.
Two interesting observations are: In all the three subgrid moments the error decreases (proportional to the
reduction in the average value of that moment) as we
increase the number of coarse grid layers. Secondly,
the best of the fluctuating moments is σ2S, followed
by σk.(dS/dx) and σS.(dS/dx) (consistent with our previous
finding using 2-layer coarse grid models). Overall, it
appears that all three moments may be used in
designing the coarse grid configuration as suggested
by the average saturation equation.
Case IIID: For this case, capillary pressure is ignored.
Again, we simulate the 1, 5, 10, 15, 20, 30 and 40 coarse
grid layers to study the relationship between the
resulting error in coarse grid model with the higher
moments. The overall results for Case IIID are very
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
33
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
similar to the one in Case IIIA i.e. the R2 ≈ 1 for all three
moments vs. normalized rms error.
Figure 16: Normalized rms error versus no. of coarse grid layers
(Case IIID)
Figure 17: Recovery factor versus Pore volume injected (40-layer
coarse grid, Case IIID)
Figure 18: Normalized rms error versus no. of coarse grid layers
(Case IIIE)
However, Figure 16, which shows normalized rms error
vs. number of coarse grid layers, highlights the
interesting observation that only σ2S can be used in
designing the coarse grid layer configuration i.e. using
σk.(dS/dx) and σS.(dS/dx) will lead to a very large error. This
is confirmed in Figure 17 which shows an example
result for the oil recovery factor when the 80-layer fine
grid model is coarsen into 40-layer coarse grid model
for the configuration suggested by the three moments.
This finding confirms the result as suggested by the
average saturation equation. As stated in Table 1,
σ k.(dS/dx) and σ S.(dS/dx) terms do not appear when
capillary pressure is zero as simulated in Case IIID, and
using them as coarsening guides will lead to significant
errors.
Case IIIE: This case has viscous dominated flow (g = 0,
Pc = 0). As before, the three fluctuating moments vs.
normalized rms error show a very good R2. For this
case, σ2S produces a single continuous slope. Figure
18 shows that only σ2S and σk.(dS/dx) can be used in
designing the coarse grid layer configuration i.e. using
σS.(dS/dx) will lead to a large error. Figure 19 shows the
trend for the oil recovery factor when the 80-layer fine
grid model is coarsened to 40-layer model for the
configuration suggested by the three moments. The
accuracy of the σ2S term reducing the error in the
coarse grid model is expected as shown in Table 1, i.e.
the term explicitly appears when flow is viscous
dominated as in this model. On the other hand, the
accuracy of the σk.(dS/dx) term, for this specific case, is
perhaps due to its similarity with the v‘S‘ term. In such
viscous dominated zero-crossflow layered models,
permeability and velocity are directly proportional. In
addition, the variation in the S-gradient and S will make
not much difference between them because flow is
totally in one direction in each of the fine grid layers.
The S-gradient term is non-zero only in the coarse grid
blocks that the flood front is passing through.
Figure 19: Recovery factor versus Pore volume injected (40-layer
coarse grid, Case IIIE)
34
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
SUMMARY AND CONCLUSIONS
In this paper, we studied the relationship between subgrid effects and the accuracy of coarse scale simulation
models for layered systems when capillary pressure
exists in the models. The actual sub-grid measures
considered represent higher moments of fine scale
variables (σ2S, σS.(dS/dx) and σk.(dS/dx)) when capillary
pressure is important and were derived from a volume
averaging of the fine grid saturation equation. The
coarse scale simulation results demonstrate that the
volume averaging procedure can be used to model
the form of the coarse grid configuration. Further, we
showed that the coarse grid error could be minimized
by designing the coarse grid models such that certain
of these higher moments are minimized. The specific
sub-grid measure that best correlated with the coarse
grid error was shown to vary depending on balance
of fluid forces as suggested by the average saturation
equation.
REFERENCES
[1] Durlosky, L. J., Jones, R. C. and Milliken, W. J., “A Non-uniform
Coarsening Approach for the Scale Up of Displacement
Processes in Heterogeneous Porous Media”, Adv. In Water
Resour. (1997) 20, 335.
[7] Wallstrom, T. C., Hou, S., Christie, M. A., Durlofsky, L.J. and Sharp,
D. H.: “Application of a New Two-Phase Upscaling Technique
to Realistic Reservoir Cross Sections”, SPE 51939, Proceedings
of the 15th SPE Reservoir Simulation Symposium, Houston,
TX, 14-17 February 1999.
[8] Eclipse 100 Technical Manual, Version 2000A, Geoquest,
Schlumberger.
[9] Guzman, R. E., Giordano, D., Fayers, J., Godi, A., Aziz, K., 1994.
“The Use of Dynamic Pseudo Functions in Reservoir
Simulation”, Presented at the 5th International Forum on
Reservoir Simulation, Muscat, Oman, December 10-14, 1994.
[10] Darman, N.H., Sorbie, K.S. and Pickup, G.E.: “The Development
of Pseudo Functions for Gravity-Dominated Immiscible Gas
Displacements”, SPE 51941, Proceedings of the 15th SPE
Reservoir Simulation Symposium, Houston, TX, 14-17 February
1999.
Nasir Haji Darman joined PETRONAS in
1991 and currently is a Principal Reservoir
Engineer with Petronas Research and
Scientific Services Sdn. Bhd. (PRSS). At the
moment, he is also the Initiative Champion
on Enhanced Oil Recovery initiatives at the
PETRONAS Integrated Transition Programs
(ITP). He has a first degree from Texas Tech
University and PhD from Heriot-Watt
University, Edinburgh, Scotland. Both of the degrees are in
Petroleum Engineering. His research interests include upscaling
of fluid flow in porous media, EOR and reservoir simulation. Dr.
Nasir has published several technical papers in local and
international journal.
[2] Stern, D. and Dawson, A.G., 1999,“A Technique for Generating
Reservoir Simulation Grids to Preserve Geologic
Heterogeneity”, SPE 51942, Proceedings of the 15th SPE
Reservoir Simulation Symposium, Houston, TX, 14-17 February
1999.
[3] Darman, N. H., Durlofsky, L.J., Sorbie, K. S. and Pickup, G.E.:
“Upscaling Immiscible Gas Displacements: Quantitative Use
of Fine-Grid Flow Data in Grid-Coarsening Schemes”, SPEJ, 6
(1), 47 - 56, March 2001.
[4] Zhou, D., Fayers, F. J. and Orr Jr., F. M. 1993.“Scaling of Multiphase
Flow in Simple Heterogeneous Porous Media”, SPE/DOE 27833,
presented at the ASME Winter Meeting, New Orleans,
November 28 - December 3, 1993.
[5] Rappoport L. A.,“Scaling lows for use in design and operation
of water-oil flow models” 1954, Trans AIME 204, 143-150.
[6] Shook, M., D. and Lake, L. W.,”Scaling immiscible flow through
permeable media by inspectional analysis”, 1992, In-Situ 16,
311-349.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
35
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
SURFACTANT SYSTEMS FOR VARIOUS FIELDS OF EOR:
DRILLING FLUIDS, MICROEMULSIONS, CONTROL OF VISCOSITY,
BREAKING OF EMULSIONS
Heinz Hoffmann
University of Bayreuth, BZKG, 95440 Bayreuth, Germany.
[email protected]
ABSTRACT
Different areas for the use of surfactants in tertiary oil recovery and in oil industry will be discussed. The basic
principles of the surfactants in these processes will be explained.
In EOR the most important factors are the viscosity and the interfacial tension of the surfactant solution. The
viscosity of the pushing fluid has to be matched to the viscosity of the reservoir oil and the interfacial tension of
the oil/water interface should have a very low value of around 1.10-3 mN/m. It is shown that the viscosity of the
fluid with about 1% surfactant can be adjusted between the viscosity of water and values of 100 Pas. However,
the high viscosities break down when the fluids are in contact with oil and solubilize oil. Under these conditions
always very low viscosities will be reached. It is therefore necessary to adjust the viscosity with water soluble
polymers.
Drilling fluids are another area where surfactants can be used. They usually contain surfactants, clays and water
soluble polymers. It is demonstrated that in such formulations all components can interact with each other and
these systems can form weak hydrogels with a strong shear-thinning behavior.
While surfactants or amphiphilic compounds are used to prepare emulsions and stable dispersions, surfactants
can also be used to break natural emulsions that are produced during the oil recovery process. Finally, it will be
discussed that surfactant systems can be used for drag reduction during the transport of oil in pipelines or for
the transport of waste water.
Keywords: surfactant systems, EOR, drilling fluids, microemulsions
INTRODUCTION
Surfactants are as their name expresses surface active
molecules. The surfactant molecules consist of two
parts, one which is oil soluble and insoluble in water
and one which is polar and water soluble and not oil
soluble. As a consequence of this amphiphilic nature
of the compound the molecules can do amazing
things. There is hardly any large scale industrial process
where surfactants are not involved. They play the key
role in the cleaning process of textiles and surfaces [1],
they stabilize emulsions and dispersions [2], but all the
same they can also flocculate a stable dispersion. They
speed up dyeing process [3], they lower friction and
can accelerate the flow in turbulant pipe flow [4]. In
nanotechnology they are used to structure the
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
36
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
porosity of mesoporous materials [5] and to
compartibilize nanoparticles like clay minerals with
bulk polymers [6].
•
In water or aqueous phases, surfactants selfassemble into micelles and at higher
concentration liquid crystalline phases [12].
As a consequence of their importance in industrial
processes, a large number of surfactants are produced
and are commercially available. Several thousand
different surfactants are used and produced
worldwide. The molecules differ in chainlength,
headgroups, branching ethoxylation degree and so on.
•
The aggregates do modify the flow behavior of the
aqueous phases. The fluids can become
shearthinning and shearthickening; and assume
properties of soft matter with a yield stress or form
highly viscoelastic fluids [13, 14, 15].
In spite of their large variety and different structures,
all of them have several effects in common on which
the many applications are based. Theoretically it
probably would be possible to generate all the
properties that can be made by surfactants with a few
dozens of compounds. The large number of used
compounds has to do with the availability of raw
materials, price, tradition, laws and so on.
The main effects on which their applications are based
are few and most of the actions add up to a dozen or
so.
• Surfactants are surface active molecules and when
dissolved in water form a monolayer on the water
surface or at an oil water interface [7].
•
As a consequence of this adsorption, the surface
tension of water and the interfacial tension of the
O/W interface is lowered [8].
•
Surfactants adsorb also on solid/liquid interfaces
and by doing so change the wetting behavior of
the fluid with the substrate [9].
•
Surfactants can also adsorb on other molecules
that are dissolved in the aqueous phase like on
polymers or on dyes or phospholipid structures
[10].
•
The adsorption therefore leads to a change of the
interaction between the molecules or the
molecules cluster themselves [11].
All of these remarkable properties are achieved at
rather small concentrations of less than 1 %, often
already at 0.1 or 0.01 %. These amazing properties of
surfactants can also be used in the petroleum industry,
in particular in enhanced oil recovery (EOR) [16].
Surfactants can be used for the formulation of
optimized drilling fluids [17], for the improvement of
the oil recovery [18], for the breaking of O/W and W/O
emulsions [19] and in particular for the speed up of
the flow in pipelines [20].
In this paper, the fundamentals for the various
applications will be described. It will become clear that
the optimized fluids for the various applications always
depend on the particular situation. It will not be
possible to come up with general formulations which
are good for all oil wells. Each oil well has its own
salinity in water, its special rock and porosity, its
temperature and its special composition of oil. All of
these parameters are important for the formulation of
the optimum fluid because phase diagrams of
surfactants and properties of micellar solutions and
in particular the interfacial tension depend on the
salinity of bulk water and temperature [21].
SURFACTANT FLOODING
It is a general knowledge that an oil well can be
considered to be exhausted when most of the original
oil is still trapped in the mesoporous system of
capillaries of the rock formation. But the water that
can be pushed into the reservoir to dislodge the oil
from the rock can find its own route to the pumping
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
37
Technology Platform: RESERVOIR ENGINEERING
hole and does not displace the oil. It may not have the
right viscosity and not the right surface activity to lose
the oil from the rock.
The natural oil maybe wetting the rock. The reason
for the wetting on a molecular scale can be many fold.
One often encountered situation occurs in oil
reservoirs in which the rock is limestone. Limestone
can have a positive zeta-potential and the crude oil
can contain surfactant-like compounds with
negatively charged groups which stick to the
limestone and thus make a good connection of the
oil to the rock. The oil may also be much more viscous
than water. Under these conditions it is easy to realize
that it is difficult for water as a flooding fluid to
dislodge the oil from the rock.
The situation is very similar to the washing of laundry
where greasy dirt is sitting on textile fibre and is hold
back by the capillary systems of the textile network.
We know that in this situation we have to adjust the
surface activity of the surfactant in the aqueous phase
to such a level that the aqueous phase will wet the
textile fiber much better than the greasy oil and
therefore dislodge the oil from the fibre.
The fundamentals of this process are given by Young
equation:
σ →sw = σ →so + σ →ow . cos (1)
In the washing process the amount of the oily phase
is miniscule in comparison to the amount of the
aqueous phase. In most practical situations, the
Technology Cluster: OIL AND GAS
amount of oil is even smaller in comparison with the
amount of surfactant in the aqueous phases. It is
therefore likely that in such situations the oily grease
is completely solubilized by the surfactant. In the oil
recovery process, solubilisation cannot be set as the
objective of the process. This strategy would be too
expensive because even under optimized conditions
like the ones which are encountered in
microemulsions [22], we would need significant
amount of surfactants. The goal therefore must be to
dislodge the crude oil from the wall of the capillaries
and to form emulsion droplets from the oil. This brings
up a second problem in the recovery process. The oil
droplets must be able to pass easily the capillary
system of the rock, as shown in Figure 1. If the emulsion
droplets have a larger diameter than the diameter of
the capillaries, the transport would be difficult because
the droplets would have to be deformed. This process
costs much energy. To get around this problem lies
the use of surfactants which give an interfacial tension
of the droplets. From our experience with immiscible
liquids we know that droplets of oil in water assume a
spherical shape. Such an emulsion would not pass
easily through a glass capillary in which its radius is
smaller than the radius of the droplets. A deformed
droplet has however a longer surface than a spherical
one. The interfacial tension of a normal emulsion that
is stabilized by a surfactant film may have an interfacial
tension of the order of 1 mN/m. This interfacial tension
may however be lowered to values that are 1000 times
smaller and are in the range of 10-3 mN/m.
Under these conditions the droplets can easily be
deformed because the energy of deformation which
Figure 1: Oil droplet with fluctuating shape when interfacial tension is extremely low. It then can easily pass capillaries in the rock
38
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
is given by the product of increase of area of droplet
times the interfacial tension is small and in the area of
kT. The situation is then similar to the flow of blood
through the fine arteri-system in the human body. The
erythrosites can easily be deformed during flow.
After oil is dislodged from the wall, it is hoped that the
emulsified oil can coalesce and form slugs of oil which
can be pushed to the central hole by the flooding fluid.
Now the viscosity of the oil slug becomes of
importance for the whole process: its viscosity could
be much higher than the viscosity of the aqueous
phase, a situation that could give rise to fingering [23].
In this situation we could think of using viscoelastic
surfactant solution as flooding fluids. Viscoelastic
surfactant solutions consist of wormlike micelles and
their zero-shear viscosities can be as high as 100 Pas
for 1 % by weight surfactant solutions [24]. Figure 2
Figure 2: Different surfactants with high viscosities at small
concentrations
shows zero shear viscosities for different surfactant
systems as a function of the concentration. The results
make it clear that the viscosities can easily be adjusted
within a wide range. In particular aqueous surfactant
systems with high viscosities can be obtained by
mixing surfactants or surfactants with co-surfactants.
Wormlike micelles can now be made visible with the
cryo-TEM method. A micrograph of such micelles is
shown in Figure 3. However, when such a surfactant
solution comes into contact with oil and some oil is
solubilized into the threadlike micelles, these long
entangled micelles are transformed to microemulsion
droplets and the viscosity of the fluid breaks down and
assumes viscosities of water [25]. The transformation
of long entangled micelles into small microemulsion
droplets is a very general one. It is entropy driven and
can not be suppressed. For this reason the use of
viscoelastic surfactant solutions as flooding fluids is
probably not practical. If the viscosity has to be
adjusted it is better done by water-soluble polymers
that do not bind to surfactants.
Figure 3: Cryo-TEM micrograph of entangled threadlike micelles.
Note the branching of the cylindrical micelles at the white arrow.
The block arrow marks the position where two micelles are on
top of each other (by courtesy of I. Talmon)
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
39
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
Figure 5: Viscosities of cationically modified hydroxyethylcellulose
(cat-HEC) for increasing surfactant concentration. Note the two
phase region around charge neutrality.
coils and therefore reduces the viscosity of the polymer
solution. In the following figures some of the
phenomena are illuminated that are encountered in
polymer solutions.
Figure 4: Laser light scattering results about the rod sphere
transition by solubilisation of hydrocarbon. Note the sharpness
of the transition
Some light scattering results to demonstrate the
breakdown of entangled threadlike micelles are given
in Figure 4. It is clear from the results that the
transformation of the rods to the microemulsion
droplets occurs rather abruptly.
WATER-SOLUBLE POLYMERS AND SURFACTANTS
Water-soluble polymers are known to be able to build
up viscosities in aqueous solutions [26]. Some of them
are known as thickeners in various applications. Their
ability to build up high viscosities depends on the
molecular weight of the polymer. On the flexibility or
the persistence length and in particular on the
hydrophobic and ionic substitution of the polymer.
Many water-soluble polymers are also surface active
and they interact therefore with normal surfactants.
The viscosity of ionic modified polymer solutions
depends very much on the salinity of the flooding fluid.
In general, excess salt leads to the collapse of swollen
40
In Figure 5, the viscosity of hydrophobically and
cationically modified hydroxyethylcellulose is shown
when increasing amounts of ionic surfactants are
added [27]. With increasing surfactants, we note first
an increase of the viscosity which actually can lead to
the formation of a gel-like system within a small
concentration region. This increase of viscosity and
formation of a yield stress in the gel is based on the
physical cross-linking of the hydrophobic groups by
the surfactants. Beyond a critical surfactant
concentration the system collapses and undergoes
phase separation into polymer and surfactant rich
phase and a polymer poor phase. This process is
accompanied by the breakdown of the viscosity of the
system and is caused by a complete charge
compensation of the ionic charges by the ionic charges
of the surfactants. As a consequence the repulsive
forces on the polymer are overcompensated by the
attractive forces and the systems collapse. If the
surfactant concentration is increased further, more
surfactants will bind to the polymer-surfactant
complex, recharge the systems and it becomes soluble
again with high viscosity.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
For HM-polymer systems that carry negative ions the
phase separation with increasing anionic surfactants
does not occur. In particular, the strong maximum of
the viscosity occurs and the viscosity after the
maximum is lower than the viscosity in the surfactant
free system. Some copolymers are available which give
rather high viscosities at small concentrations and the
viscosity is rather insensitive to the salt and the
surfactant concentration [28]. Figure 6 shows some
rheological data for such a system.
Polymer Synthesis
A micelle polymerization technique was used to
prepare the hydrophobically modified polyelectrolytes.
The presented data make it clear that it is generally
not possible to develop a formulation which can be
used in all oil-wells. Usually each well requires the
development of its own optimized formulation.
Rheological properties
Ionic strenght effect
Surfactant effect (SDS, CTAB, C14DMAO)
Temperature effect
Shear rate Effect
Viscoelasticity
AFM microscope
DRILLING MUDS
Drilling muds are fluids with a complicated rheological
behavior. They contain several components and they
usually have to fullfill several tasks. Very often they
contain clays, water-soluble polymers and surfactants;
and water. All of these components usually interact
with each other.
(6a)
Drilling fluids have to cool the drill. They have to
transport the drilled-up material to the surface and to
stabilize the drilled hole. This fluid should seal the wall
of the hole to prevent drilling fluid from escaping into
the rock formation. Finally, the drilling fluid should
prevent sand and other soils from falling and building
a sediment in the drilled hole when the drills have to
be exchanged what requires the taking out and in of
hundreds of meters of drilling pipes.
(6b)
Figure 6: Viscosities for different copolymers from polyacrylamide
and polyacrylate which are weakly hydrophobically modified at
different salinities. The polymer concentration is 0.5 %
Finally, so many surfactants are bound on the
hydrophobic groups that the network points open up
and the viscosity breaks down to values which are
below the viscosity of the surfactant free system.
Surfactants and water-soluble polymers are surface
active components and they will bind to clay surfaces
[29, 30]. In Figures 7 and 8, some results of the
adsorption of the surfactants and polymers are shown.
The results make it clear that clays can adsorb their
own weight of surfactant and their many fold weight
of polymers. Surfactants of course as we have
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
41
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
1.25 wt% PVA 1 + 1.5% SKS-20
1.25 wt% PVA 1 + 1.5% SKS-20
Figure 7: Surface tension measurements of
alkyldimethylaminoxide surfactants with and without the
presence of 1 % of clay mineral
Figure 9: Soft hydrogel from polyvinylalcohol and clay mineral
Figure 10: A rheogram of the hydrogel from Figure 9
Figure 8: Surface tension measurements of polyvinylalcohol
without and with the presence of different clay mineral
concentrations
discussed in the section on flooding do also adsorb
on polymers and by doing so modify the binding of
polymer to the clay. Under optimized conditions,
systems can be formulated to behave like soft gels in
the unstirred system but are shear-thinning under
shear and become water-viscous under high shear
[31].
Obviously, the yield stress of the soft gels is the
property that is necessary to prevent soil from
sedimenting into the hole when drilling is at stand still.
The viscoelastic properties of the fluid under pumping
conditions guarantee the transport of the drilled up
rock and the clay are the component that can seal the
wall of the hole. In Figure 9 the gel-like behavior of a
42
system that consists of clays, polymer and surfactant
is shown and in Figure 10, the shear thinning behavior
of the fluid is presented.
THE BREAKING OF EMULSIONS
During the flow through the porous rock the crude oil
is exposed to shearing forces. When it is at the same
time in contact with the flooding fluid emulsions will
be formed. In oil wells which are close to depletion
and where the aqueous phase is in excess it is likely
that O/W emulsions will be formed while for new wells
it is likely that W/O emulsions may be formed. Both
emulsions can typically have 50 % of water and can
be very stable. As a consequence of the dense packing
of the droplets they can be highly viscous and may
not even up-creame what they normally should. The
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
droplets in the continuous phase are usually stabilized
by rather thick films (skins) of surface active
compounds like asphaltenes, acids and alcohols that
are present in the natural oil and may actually make
up a considerable fraction of the oil (a few percent)
[32]. In order to save transport costs of the crude oil to
the refinery these emulsions have to be broken as close
to the oil well as possible and as quickly as possible in
order to avoid large storage facilites. The coalescence
of the droplets and hence the quick separation of the
oil and the aqueous phase can be accelerated by
adding small amounts of additives. Even in
concentrations of small amounts (~100ppm) such
additives can quickly break the emulsions [33].
However, a general solution for the problem does not
exist because in each case the stabilizing film of the
droplets is of different polarity and needs an optimized
molecule for the coalescence process to occur.
The question may be asked what is the underlying
molecular process for the coalescence of the droplets.
The problem is similar to the case of foam destruction
by antifoam additives [34]. It is clear from the small
amounts of additives which are added to break the
emulsions that these additives cannot replace the
large amounts of natural additives from the interface
of the droplets. A different mechanism must do the
job.
Some of the most active additives are copolymers with
hydrophilic and hydrophobic groups [35]. It is
therefore imaginable that these compounds are
soluble both in the aqueous and in the oil phase. In
solutions they will form a coil. Such a coil, if large
enough, can crosslink two neighbouring droplets and
thus form a link between the droplets through which
the fluid from one droplet can flow to the other. Since
one droplet will, on the average, have a different size
than the other and hence a different LaPlace pressure
the outcome of such a situation will be clear, i.e. the
little droplet will be quickly swallowed by the bigger
one. As a strategy for the breaking of emulsions we
therefore have to design large molecules or cluster of
molecules which are able to penetrate quickly oil/
water interfaces.
DRAG-REDUCTION
Newtonian fluids have a shear-rate independent
viscosity. For small pressure difference in pipe flow the
fluids show a laminar flow behavior with a parabolic
velocity profile. With increasing pressure the laminar
flow is replaced by a turbulent flow at a critical
Reynolds number. For turbulent flows the friction
coefficent of the fluid can be reduced by the addition
of small amounts of additives like polymers or
surfactant. This phenomenon is of great practical
significance because it means that for the same
pressure difference the fluid can flow faster or a
required flow rate can be reached with a lower
pressure. The practical consequences are that less
energy is needed for the transport of a fluid or for the
cycling of a fluid in a closed circuit. The effect is
demonstrated in Figure 11 where the waterflow from
two glass tubings, is shown in which the glass tubings
have the same diameter. If both tubings are filled with
water the flow velocity of the water when it leaves the
tubes will be the same and will therefore show the
same parabola. In the demonstration set-up in Figure
11 the water in one system contains 1000 ppm of a
special surfactant. The result is a faster flow rate at the
tube exit.
Figure 11: Demonstration of drag reduction in water by
surfactants. The blue water has a higher outflow velocity than
the yellow water even though the pressure difference and the
diameters of the glass tubings are the same
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
43
Technology Platform: RESERVOIR ENGINEERING
The molecular mechanism for this phenomenon is very
complicated and has been studied in detail [37]. In
the solution at rest, the surfactant molecules in the
aqueous phase form small rodlike micelles with an
axial ratio of about 10. With a diameter of about 25 ≈
and a length of 250 ≈ these rods undergo Brownian
diffusion and rotation processes. With their
dimensions they have a rotation time constant of the
order of one msec. This means that for small shear rates
the Peclet number τ . γ is much smaller than one what
means that the rods are not aligned at such flow-rate
and the fluid remains isotropic. However, even at small
shear rates the rods collide and make collisions. The
interface of the rods is such that they stick to each
other and form long necklace type of strings. These
long aggregates align in the flow and laminarize the
turbulent flow. The consequence is a reduction of the
friction in the turbulent flow region. The flow of oil in
the pipelines can be effected in the same way. It is
possible the additives act in the same way.
REFERENCES
[1]
Detergency, Ed. By W. G. Cutler and R. C. Davis, Vol. 5, Marcel
Dekker, Inc. N. Y. 1972.
[2]
Emulsions and Emulsion Technology Ed. By K. J. Lissant, Marcel
Dekker, Vol. 6, 1974.
[3]
Piero Savarino, Viscardi Guido, Quagliotto, Pierluigi, Barni
Ermanno, Friberg Stig E., J. of Dispersion Sci. and Techn. (1993),
14 (1), 17-33.
[4]
D. Ohlendorf, W. Interthal, H. Hoffmann Rheologica Acta 25
(1986), 468-486.
[5]
N. H¸sing, U. Schubert, Angew. Chem. 1998, 110, 22-47.
[6]
S. H. Tolbert, T. E. Sch ffer, J. Feng, P. K. Hansma, G. D. Stucky,
Chem. Mater. 1997, 9, 1962- 1967.
[7]
[8]
[9]
Adsorption and aggregation of surfactants in solution, Ed.
By K. L. Mittal and D. O. Shah Vol. 109, Marcel Dekker 2003.
Microemulsions and related systems Ed. by M. Bourrel and R.
S. Schechter, Vol. 30 Marcel Dekker, Vol. 30.
R. Lipowsky, Interface Sci. 9, 105 – 115, 2001
[10] E. D. Goddard and K. P. Annanthapadmanabhan Interactions
of surfactants with polymer and protein, CRC Press, Boca
Raten, 1993.
44
Technology Cluster: OIL AND GAS
[11] Dispersions, Ed. By E. Kissa Vol. 84, Marcel Dekker 1999.
[12] G. J. Tiddy, Physics reports 57:1, 1980.
[13] H. Hoffmann, ACS-Symposium Series 578, 1 – 31 (1994), Ed. C.
A. Herb and R. K. Prud’homme.
[14] Qi Yunying, Kawaguchi Yasuo, Christensen Richard N., Zakin
Jacques L., Intern. Journal of Heat and Mass Transfer (2003),
46 (26), 5161-5173.
[15] H. Hoffmann, C. Thunig, P. Schmiedel, U. Munkert Faraday
Discuss. 1995, 101, 319-333.
[16] Surface Chemistry in the Petroleum Industry, Chapter II J. R.
Kanicky, J. C. Lopez-Montilla, S. Parnacy and D. O. Shah
Handbook of Applied Surface and Colloid Chem. Ed. By K.
Holmberg, John Wiley 2001.
[17] J. S. Robinson, Corrosion Inhibitors-Recent developments,
Noyes Data Corporation, Park Ridye N. Y. 1979.
[18] Jr. L. A. Willson, Physicochemical environment of petroleum
reservoirs in relation to recovery systems Eds. D. O. Shah and
R. S. Schechter, Academic Press N. Y. 1977.
[19] P. D. Berger, C. Hsu, J. P. Arendell SPE Production Engineering,
522, 1988.
[20] L. Broniarz-Press, J. Rozanski, S. Dryjer, S. Woziwodzki, Intern.
Journal of Applied Mechanics and Eng. (2003), 8 (spec. issue),
135-139.
[21] M. Kahlweit, R. Strey, J. Phys. Chem. 1986, 90, 5239.
[22] U. Olsson, K. nagai, H. Wennerstrˆm J. Phys. Chem. Vol. 92, 1988,
6677.
[23] Afsar-Siddiqui, Abia B., Luckham Paul F., Matar Omar K.
Advandes in Colloid and Interf. Sci. (2003), 106, 183-236.
[24] H. Rehage, H. Hoffmann, J. Phys. Chem. 92 (1988), 4712.
[25] O. Bayer, H. Hoffmann, W. Ulbricht, H. Thurn, Advances in Coll.
And Interf. Sci. 26 (1986), 177-203.
[26] E. Ruckenstein, G. Huber, H. Hoffmann Langmuir 3, 1987, 382.
[27] U. K‰stner, H. Hoffmann, R. Dˆnges, R. Ehrler Colloids Surf. A.
1996, 112 (2-3), 211-227.
[28] L. Ye, R. Huang, H. Hoffmann, J. of Applied Polym. Sci. Vol. 89,
2664 (2003).
[29] Y. Yamaguchi, H. Hoffmann, Colloids and Surfaces A:
Physicochemical and Engineering aspects 121 (1997) 67-80.
[30] S. Holzheu, H. Hoffmann Progr. Of Coll. And Polym. Sci. 115,
265-269 (2000).
[31] J. Liu, H. Hoffmann, Colloid and Polym. Sci. in press April 2004.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: RESERVOIR ENGINEERING
Technology Cluster: OIL AND GAS
[32] Golovko A. K., Kam’yanov V. F., Braun A. E., Popov N. V.
Chemistry and Technology of Fuels and oils (Translation of
Khimiya I Tekhnologiya Topliv I Masel) (1999), 35 (4), 247-255.
Bunger James W., Russell Christopher P., Cogswell Donald E.,
American Chem. Soc. Division of Petroleum Chemistry (2001),
46 (4), 355-360.
[33] Sjoblom Johan, Kallevik Harald, Aske Narve, Auflem Inge
Harald, Havre Trond Erik, Saether Oystein, Orr Robert,
American Institute of Chem. Engineers, (Spring National
Meeting), New Orleans, LA, United States, Mar. 11-14, 2002
(2002), 1697-1704.
Auflem Inge Harald, Westvik Arild, Sjoblom Johan, J. of
Dispersion Sci. and Techn. (2003), 24 (1), 103-112.
[34] Garrett P. R. Surfactant Sci. Series Vol. 45, marcel Dekker Inc.,
New York (1993).
Heinz Hoffmann is a Professor at University
of Bayreuth, Germany. He holds a master
degree in Chemistry (1959) and PhD (1962)
from TH Karlsruhe University. He did his
Postdoctoral studies at Case Western
Reserve University, Cleveland Ohio (1963).
He was the Founding Chairman of European
Colloid and Interface Society (ECIS) in 1986,
the General Secretary of ECIS (1987 - 2000),
a Visiting Scientist at the Du Pont Company in Wilmington,
Delaware (1984-1985) and a Visiting Professor at Tokyo Science
Technology in 1989.
He has won several awards at national and international level.
Among them are the "Nernst" award of the "Deutsche
Bunsengesellschaft" in 1976, the Wolfgang-Ostwald-Award of the
"Kolloidgesellschaft" in 1995 and the Lectureship award from the
Chemical Society of Japan in 1998. He has published more than
300 publications in the field of colloid and surfactant science.
[35] Stroem-Kristiansen Tove, Lewis Alun, Daling Per S., Nordvik
Atle B., Spill Sci. & Techn. Bulletin (1995), 2 (2/3), 133-141.
Abendroth P., Koch B., Meerbote M., Sczekalla B., Tenside,
Surfactants, Detergents (1993), 30 (2), 122-7.
[36] Nowak M., Experiments in Fluids (2003), 34 (3), 397-402.
[37] Sugawara Hitoshi, Yamauchi Makoto, Wakui Futao, Usui
Hiromoto, Suzuki Hiroshi, Communications (2002), 189 (12),
1671-1683.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
45
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
SELECTIVE FISCHER-TROPSCH WAX HYDROCRACKING –
OPPORTUNITY FOR IMPROVEMENT OF OVERALL
GAS-TO-LIQUIDS PROCESSING
Jack CQ Fletcher1, Walter Böhringer and Athanasios Kotsiopoulos
Catalysis Research Unit, University of Cape Town, Private Bag, Rondebosch, 7701, South Africa.
1
[email protected]
ABSTRACT
Growing demand for high quality diesel is being driven principally by increasingly stringent legislation governing
the composition of liquid transportation fuels and, amongst the various alternatives to meeting this demand,
Fischer-Tropsch (F-T) based Gas-to-Liquids (GTL) processing is recognized as an industrially proven and
economically competitive route. Furthermore, it is generally accepted that for this purpose, GTL processing is
most effective when comprising an F-T synthesis driven to wax production, followed by hydrocracking to produce
the middle-distillate products. By means of comparison between refinery and F-T GTL hydrocracking feedstocks,
and demonstration of the poor applicability, to F-T wax processing, of catalysts optimized for refinery feedstocks,
this paper argues that a significant opportunity exists for optimization of the wax hydrocracking step and also
for an overall improvement of F-T based GTL processing.
Keywords: Gas-to-Liquids, Fischer-Tropsch, Hydrocracking, Wax, Catalyst
INTRODUCTION
Since the 1970’s, environmental legislation regulating
air quality has increasingly played an important role
in determining automotive emission standards.
Initially limited to the introduction of unleaded fuel
and catalytic converters for gasoline vehicles, more
recent regulation has focused on addressing emission
aspects directly linked to fuel composition. With the
increasing demand for diesel fuels, principally due to
a combination of the inherently greater efficiency of
high compression diesel engines (and associated
lower CO2 emissions per km traveled), attention has
turned to the regulation of diesel properties most
adversely affecting air quality. This notably reduced
sulphur, aromatics, fuel density and distillation
endpoint, and increased cetane number as provided
in European, American and Japanese specifications [1].
Of the various alternative fuels possible for use in diesel
engines (including biodiesel, dimethylether, LPG,
methanol and CNG), synthetic diesel from FischerTropsch (F-T) based gas-to-liquids (GTL) processing is
the most promising in terms of all of, potential to
produce quality and engine compatibility.
Although long recognized as a route for the conversion
of coal and natural gas to liquids, the Fischer-Tropsch
conversion is notoriously unselective, producing a
wide product carbon-number distribution. Despite
decades of research, efforts to improve the intrinsic
selectivity of the F-T synthesis towards desired
fractions has met with only limited success. With
increasing global interest in F-T processing as a route
to both associated and remote gas utilization, and the
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
46
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
production of high quality, clean distillate fuels,
attention is focused on routes to maximize overall
middle-distillate selectivity. Consequently, it is
generally accepted that the most effective route
currently available for enhanced overall distillate
selectivity in F-T based GTL processing, is an approach
which makes use of high α-value F-T catalysts
generating long-chain, wax, products in the F-T step,
followed by cracking of the wax back into the distillate
product range [2].
Hydrocracking is a well established process for the
conversion of heavy feedstock into lighter fractions.
With origins based initially on coal liquefaction and
later on the conversion of heavy gas oils, hydrocracking
came into its own only during the 1960’s when it was
developed as the process of choice for the processing
of refractory crude and cycle oils, the latter increasing
with the then large expansion in refinery fluid catalytic
cracking capacity [3]. Despite its now much wider and
more versatile application, hydrocracking process and
catalyst developments have almost exclusively been
limited to the processing of crude and crude-derived
fractions within traditional oil-refining complexes.
Moreover, within the US market where this
commercialization drive took place, the desired
product was principally gasoline. Consequently,
process and catalysts have been optimized for such
feedstocks, rich in heteroatoms (sulphur, nitrogen and
metals) and generally highly unsaturated and
aromatic. These feedstocks contrast greatly with that
to be processed in a Fischer-Tropsch GTL environment,
where the F-T wax comprises almost exclusively linear,
paraffinic hydrocarbons free of sulphur, nitrogen and
metals. Additionally, the only desired products of F-T
wax hydrocracking are middle-distillate fuels.
consequently, it is pertinent to review and optimize
the F-T wax hydrocracking process.
EXISTING AND PROPOSED F-T COAL AND GAS
TO LIQUIDS CAPACITY
Both coal and natural gas based F-T plants are
operating commercially today. In South Africa, Sasol
operates 3 coal-based plants [5] with a total product
capacity of roughly 5 500 x 103 tpa (Sasolburg and
Secunda). The smaller, original plant in Sasolburg,
dating from the early 1950’s, is currently in the process
of conversion to natural gas feed. Two natural gas
plants of approximately 1 200 x 103 tpa (Mossel Bay,
South Africa) and 640 x 103 tpa (Bintulu, Malaysia) total
product capacity have been in operation by PetroSA
and Shell, respectively, for some ten years [6]. In the
above plants, all of fixed- (Sasolburg and Bintulu),
fluidized- (Secunda), entrained- (Mossel Bay) and
slurry-bed- (Sasolburg) reactor configurations are
employed, as are coal gasification (Secunda), combined
primary / autothermal reforming (Mossel Bay) and
partial oxidation (Bintulu) syngas generation
technologies.
Recently, some 12 plants based on natural gas have
been proposed, totaling in the order of 33 000 x 103
tpa capacity [6]. Although several of the proposed
plants may be considered competitive bids, it is
reasonable to expect that at least 3 commercially
viable projects will be implemented still this decade
(more than doubling the current installed global
capacity), viz. plants of 1 300 x 103 tpa in Nigeria (Sasol /
Chevron Texaco [6]) and plants of 1 300 x 103 tpa (Sasol
/ QPC [6]) and 5 500 x 103 tpa (Shell / QPC [7]) in Qatar.
REFINERY HYDROCRACKING
Given that the capital cost of the GTL plant contributes
a substantial portion to the cost of production and that
the refining step, including hydrocracking, comprises
only 10 % of the overall plant capital investment [4], it
is clear that the feedstock to the GTL hydrocracking
step is a valuable (expensive) material. In light of the
above, it is imperative to ensure optimal yield of the
desired distillate product from hydrocracking and,
The subject of refinery hydrocracking has been
reviewed in detail elsewhere [3]. It is applied for the
conversion of a range of ’heavy’ fractions (typically rich
in all of sulphur, nitrogen, metals and polyaromatics)
to lighter products, mostly diesel, jet fuel and gasoline
(the latter mostly in USA). A variety of catalysts have
been developed, depending on feedstock and product
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
47
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
demands. Hydrogen transfer is mediated via either
noble metals (e.g. Pt, Pd) or combinations of various
group VIA (Mo, W) and group VIIIA (Co, Ni) metals. An
acidic carrier, typically comprising amorphous silicaalumina and zeolites, either alone or in combination,
may provide an isomerisation and cracking function.
When non-noble metals are used these are present in
the form of metal-sulphides in the working catalyst.
CoMo-type catalysts may be considered a typical nonnoble metals selection, with Ni and W being
introduced to provide increased hydrogenation and
hydro-denitrification activity as required. Although
other formulations are employed in special cases (e.g.
residue upgrading, lube oil dewaxing, etc.), the above
formulations are typical for the hydrocracking of
vacuum gas oils (VGO) and FCC cycle oils.
In the case of maximum distillate yield processing,
sulphided, non-noble metal formulations on silicaalumina, are preferred. Mild hydrocracking catalysts
are similar but often employ an even milder acid
function by carrier dilution or replacement with
alumina. Zeolites (almost exclusively zeolite Y) are
employed principally to achieve ‘severe’ hydrocracking
as is the case when high gasoline selectivity is desired
and / or the feedstock is highly refractive. In this case,
noble metals, especially platinum, are applied so as to
increase the catalyst hydrogen transfer activity and so
provide an appropriate cracking / hydrogenation
balance for the intended application.
F-T WAX HYDROCRACKING
Considering published findings, it appears as if only
very limited research and development has been
committed to the special case of distillate production
via F-T wax hydrocracking, e.g. [2, 5, 8, 9]. Results, for
total wax conversion, reported by Sasol [5] and UOP
[9], suggest that middle-distillate yields of the order
of 80 wt% may be achieved. These results are
consistent with calculated values based on F-T wax
with α -values of approximately 0.95 [2] (hydrocracking
distillate extracted from data presented for the overall
Shell Middle Distillate Synthesis process and direct
F-T synthesis product composition). On the one hand,
48
Technology Cluster: OIL AND GAS
these results suggest a significant opportunity for
improved middle-distillate selectivity in the wax
hydrocracking process, especially considering that the
non-selective products are poor quality gasoline /
naphtha (15 wt%) and C1 – C4 gas (5 wt%) [5]. This is
all the more pertinent when one considers the cost to
produce the wax (inclusive of 90 % of the overall GTL
plant capital investment prior to hydrocracking [4]),
such that even modest improvements in the
hydrocracking middle distillate yield are likely to
significantly impact on process economics. On the
other hand, if indeed these distillate selectivities are
essentially representative of the kinetically expected
distribution, then the opportunity for improved
middle-distillate production may be limited.
However, given the nature of the F-T wax feedstock
and the ever more stringent distillate fuel
specifications, targets for process optimization,
although strongly influenced by overall yield, are not
limited only to issues of boiling range. A further
opportunity for process optimization involves the
judicious selection of reaction conditions
(temperature, pressure and hydrogen / hydrocarbon
ratio) so as to promote the transfer of primary middledistillate product into the vapour phase with a view
to reducing its residence time in the catalyst bed and,
consequently, preventing distillate loss via secondary
cracking. This approach to improved overall distillate
selectivity has previously been referred to by Eilers et
al. [2] and provides the incentive to optimize catalyst
performance within the constraints of the operating
window so defined.
COMPARISON OF REFINERY AND F-T WAX
HYDROCRACKING
Whereas a wide choice of catalysts exists for
hydrocracking, current commercial catalyst
formulations have been developed and optimized for
use in typical refinery settings. Consequently, the
choice of metal function is driven not only by
considerations of cost but also by technical constraints
due to the conditions applicable. As an example, noble
metal catalysts are effective only when sulphur levels
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
are below approximately 500 ppm. Hence, such
catalysts are only applicable in two-stage
hydrocracking processes with interstage removal of
H 2 S – they are not appropriate to single-stage
hydrocrackers or two-stage units without intermediate
H2S removal [3]. Such limitations will not apply to F-T
wax processing.
Indeed, in the case of F-T wax hydrocracking the
feedstock contains negligible sulphur and nitrogen,
and is almost completely saturated, comprising
essentially linear paraffins (only in the case of Federived F-T wax may low levels of olefins and
oxygenates be present) [10]. Under these conditions
both metal (in the case of noble-metals) and acid
functions are un-poisoned by sulphur and nitrogen,
respectively, and catalyst activity is likely to be
significantly higher. Also, whereas medium-pore
zeolites, e.g. MFI-types (ZSM-5), have only limited
application in refinery hydrocracking (e.g. selective dewaxing applications), due to their inability to process
large, ‘bulky’ molecules, not only are these zeolites
applicable to the linear paraffinic nature of F-T wax but
may even be of advantage in limiting the extent of
branching possible and so serve to maintain molecular
linearity and its associated high cetane properties.
Performance of Classical Refinery Distillate
Catalyst in F-T Wax Duty
As a demonstration of the unsuitability of optimized
refinery-based maximum distillate yield hydrocracking
catalysts for F-T wax duty, a simple study is presented
below wherein such a catalyst was applied to nparaffin hydrocracking in the absence of sulphur.
Experimental
The catalyst employed was a commercial distillate
hydrocracking catalyst comprising cobalt and
molybdenum oxides on an amorphous silica-alumina
support, and applied in the form of 1/16le extrudates.
Hydrocracking tests were conducted in an isothermal
trickle-bed reactor of 20 mm inside diameter. The
catalyst bed, containing 3 g of catalyst extrudate, was
diluted with nominally 0.8 mm SiC granulate in the
volume ratio 2:1 (diluent:catalyst) and was both
preceded by and supported on beds of pure SiC
granulate which acted as preheat and post reaction
temperature trim zones, respectively.
The catalyst was activated by drying under flowing
nitrogen at 350°C followed by reduction in flowing
hydrogen at the same temperature and 80 bar for 5
hours. Pure n-tetradecane feed and hydrogen gas
were fed by metering pump and gas mass flow
controller, respectively, and mixed at the entrance to
the reactor pre-heat zone. Upon leaving the reactor,
the entire effluent stream was sufficiently heated,
diluted with hydrogen and progressively reduced to
atmospheric pressure so as to ensure complete
vaporization of all the hydrocarbon components prior
to on-line gas chromatographic analysis.
Reaction conditions were 300 – 350°C, 20 – 80 bar,
0.2 – 1.3 n-C14 LHSV and 10 – 116 H2/n-C14 molar feed
ratio.
Results
Initial findings for two series of experiments, varying
temperature and pressure, respectively, are presented
in Figures 1 and 2. It is notable that branched products
are present only to a very limited degree, despite
conversions approaching 100% in some cases.
With increasing temperature and, consequently,
increasing conversion, light gas (C1 – C3) selectivity is
observed to increase, especially methane selectivity
above 330°C (Figure 1). Notable, also, is the high
selectivity to the C13-fragment at low and medium
conversion levels. The selectivity of mid-carbon
number components (C 4 – C 10 ) shifts to lighter
fractions (C 3 – C 7 ) with increasing temperature,
indicative of secondary cracking at high conversions.
The same overall pattern is observed for increasing
conversion as a result of decreasing pressure (Figure
2).
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
49
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 1: C14 Hydrocracking: Effect of Temperature (P = 80 bar,
C14 LHSV = 0.2 h-1 , H2/C14 = 116 mol/mol)
50
Technology Cluster: OIL AND GAS
Figure 2: C14 Hydrocracking: Effect of Pressure (T = 330° C, C14
LHSV = 1.3 h-1 , H2/C14 = 10 mol/mol)
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 3: Hydrocracking Pathways and Theoretical Product Carbon Number Distributions for n-C14 Feed
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
51
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Discussion
Although the term hydrocracking is generally applied
to processes involving a reduction in feedstock
molecular mass in the presence of hydrogen transfer
reactions, in practice three principal cracking pathways
may be involved in the case of n-paraffins, viz. ‘true
hydrocracking’, hydrogenolysis and a variant of the
latter, ‘methanolysis’. These pathways are portrayed
graphically in Figure 3 by means of their associated
mechanisms and expected product carbon number
distributions.
‘True hydrocracking’ is mediated over dual-functional
metal / acid catalysts and proceeds via adsorbed
carbenium ion intermediates where cleavage is most
likely on central C-C bonds and, from the fourth Cposition, occurs with almost equal probability [8, 11,
12]. Moreover, due to the carbenium ion mechanism,
the intermediate to cracked products is typically an
isomerised carbenium ion with the result that products
are generally highly branched. As a consequence of
the aforementioned, the theoretical product carbon
number distribution is as presented in Figure 3(a)
where almost no methane is produced as are only low
amounts of C2 and C3 (and likewise for the associated
C11 – C13 fragments), while intermediate length carbon
fragments are formed with approximately equal
selectivity. Only with severe conditions, where
secondary cracking of the primary fragments occurs,
does the product carbon number distribution shift
towards lighter fragments. ‘True hydrocracking’
requires the presence of strong acid and
hydrogenation sites, the latter in order to minimize
secondary cracking.
Hydrogenolysis, on the other hand, proceeds via
adsorbed hydrocarbon radical intermediates where CH bonds show higher reactivity than C-C bonds.
Consequently, chemisorbed hydrogen-deficient
hydrocarbon intermediates undergo C-C scission and
the probability of such scission is almost identical for
all C-C bonds internal to the second carbon position
on the hydrocarbon chain. As a result, the
hydrogenolysis pathway produces a product carbon
52
Technology Cluster: OIL AND GAS
number distribution comprising low methane
selectivity but essentially equal selectivities of carbon
fragments from C2 and higher [13], as shown in Figure
3(b). In contrast to true hydrocracking, the adsorbed
radical intermediate mechanism results in low
isomerisation activity and a product molecular
structure which is essentially unbranched. Again, only
with more severe processing, does secondary
hydrogenolysis of the primary fragments shift the
overall product carbon number distribution to lighter
fractions. The hydrogenolysis reaction is typically
catalysed by base-metal oxides and sulphides and, to
a lesser extent, by some metals like Pt [14]. Additionally,
certain metals, notably Ni [14] but also Co [13] show a
tendency to promote successive demethylation
(‘methanolysis’) such that the overall product carbon
number distribution is dominated by the presence of
methane and the corresponding higher primary
fragments, as presented in Figure 3(c).
Returning to the findings for n-C14 conversion over the
CoMo/silica-alumina catalyst of this study (Figures 1
and 2): the observed lack of isomerisation, together
with the uniform distribution of carbon number
selectivities in the C2 – C12 range, as well as the high
methane and C13 selectivities, are findings which are
not consistent with true dual-functional hydrocracking
reactions. Rather, the observed product distribution
is typical of hydrogenolysis on the metal / metal-oxide
component of the catalyst, including a significant
contribution from ‘methanolysis’. Considering the
catalyst composition employed in this study and
recognizing that the catalyst is not present in the form
of metal sulphides (as is the case in typical refinery
hydrocracking applications), it may be reasonable to
ascribe the high degree of methanolysis observed to
the presence of free Co metal, in much the same way
as has been reported for free Co-sulphide islands in
the case of sulphided CoMo-type hydroprocessing
catalysts [15].
Although the above findings demonstrate the
unsuitability of applying typical maximum distillate
type catalysts to the case of normal paraffin (F-T wax)
hydrocracking in the absence of sulphur, the high
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
yields of linear paraffins observed in the mid-carbon
number range (C 4 – C 10), together with minimal
isomerisation, would be of interest to high cetane
distillate fuel production from F-T wax. The challenge,
therefore, may be to reformulate the catalyst so as to
avoid the high methane and light gas yields which
correspond to fuel gas and poor quality gasoline
(naphtha) in the case of wax feedstock.
CONCLUSIONS
With increasingly stringent legislation in respect of
transportation fuel specifications, a growing demand
for high quality diesel fuel is likely to be met by
synthetic production via proposed GTL plants based
on F-T synthesis of heavy hydrocarbon wax followed
by wax hydrocracking. As this feedstock differs
substantially from traditional refinery feedstock,
existing hydrocracking catalysts and processes are not
necessarily optimized for use in the GTL environment.
Consequently, and recognizing the high value of wax
feedstock to the GTL hydrocracking stage, selective
wax hydrocracking presents a significant opportunity
for improving overall GTL performance.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge financial and
technical support from Akzo Nobel Catalysts bv,
PetroSA (Pty) Ltd, the South African National Research
Foundation (NRF GUN 2053385), the South African
Department of Trade and Industry THRIP Programme
(PID 2445) and the University of Cape Town (URC Fund
457021).
REFERENCES
[1] Diesel Fuel – Specifications and Demand for the 21 st Century.
UOP LLC Publications. 1998. Citing internet sources URL
http://www.uop.com/solutions_and_innovation/Is sues%20&
%20Solutions/UOPDieselFuel.pdf
[2] Eilers, J., Posthuma, S. A. and Sie, S. T. 1990. The Shell Middle
Distillate Synthesis Process (SMDS). Catalysis Letters 7 253-270.
[3] Scherzer, J. and Gruia, A. J. 1996. Hydrocracking Science and
Technology. New York. Marcel Dekker Inc.
[4] Dry, M. E. 2001. High quality diesel via the Fischer-Tropsch
process – a review. Journal of Chemical Technology and
Biotechnology 77 43-50.
[5] Dry, M. E. 2002. The Fischer-Tropsch Process: 1950-2000.
Catalysis Today 71 227 – 241.
[6] The Catalyst Review Newsletter. 5 December 2002. Spring
House, Pa. USA. Published by the Catalyst Group - Resources.
[7] Watts, P. and Fabricius, N. 2003. The Qatar Shell Gas to Liquids
Project. 3rd GTL Commercialisation Conference. Doha, Qatar.
20 October 2003.
[8] Sie, S.T., Senden, M.M.G. and van Wechem, H. M. H. 1991.
Conversion of Natural Gas to Transportation Fuels via the Shell
Middle Distillate Synthesis Process (SMDS). Catalysis Today 8
371-394.
[9] Shah, P. P., Sturtevant, G. C., Gregor, J. H., Humbach, M. J., Padrta,
F. G. and Steigleder, K. Z. 1988. Fischer-Tropsch Wax
Characterization and Upgrading. Final Report for U. S.
Department of Energy. DOE/PC/80017-T1 (DE88014638).
[8] Dry, M.E. 2003. Fischer-Tropsch Synthesis – Industrial.
Encyclopedia of Catalysis. Horvath I. T. ed, Vol. 3, 347-403, New
York, John Wiley and Sons.
[11] Weitkamp, J., Jacobs, P. A. and Martens, J. A. 1983. Isomerisation
and Hydrocracking of C9 – C16 n-Alkanes on Platinum/HZSM5 zeolite. Applied Catalysis 8 123 - 141.
[12] Martens, J. A., Jacobs, P. A. and Weitkamp, J. 1986. Attempts to
Rationalize the Distribution of Hydrocracked products. I.
Qualitative Description of the Primary Hydrocracking Modes
of Long Chain Paraffins in Open Zeolites. Applied Catalysis
20 239 - 281.
[13] Sinfelt, J. H. 1973. Specificity in Catalytic Hydrogenolysis by
Metals. Advances in Catalysis 23 91 - 119.
[14] Gates, B. C., Katzer, J. R. and Schuit, G. C. A. 1979. Chemistry of
Catalytic Processes. New York. McGraw-Hill. p 274.
[15] Topsøe, H., Clausen, B. S. and Massoth, F. E. 1996. Hydrotreating
Catalysis. Catalysis – Science and Technology. Anderson, J. R.
and Boudart, M. eds, Vol. 11, 29 – 33, Berlin, Springer-Verlag.
Jack Fletcher is Director of the Catalysis
Research Unit in Chemical Engineering at
the University of Cape Town, South Africa,
the country’s premier Industry/University
Cooperative Research Centre in
heterogeneous catalysis (specifically as
applied to phenols and derivatives) and
catalysis by gold. He has held this position
since 1999. Prior to joining academia, he
was Director Research and Development for the Catalyst Division
based in Germany. He has supervised more than 12 PhD and MSc
theses and published some 20 papers in international journals
and conference proceedings. He is the current chairman of the
Catalysis Society of South Africa.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
53
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
THEORY OF AUTOTHERMAL REFORMING FOR
SYNGAS PRODUCTION FROM NATURAL GAS
Kunio Hirotani
Licensing and Patents Department, Toyo Engineering Corporation,
2-8-1 Akanehama, Narashino-shi, Chiba 275-0024, Japan.
[email protected]
ABSTRACT
Autothermal Reforming (ATR) is believed to be the most economically viable technology for large-scale conversion
of natural gas to synthesis gas (Syngas) used as make-up gas for synthesis of GTL (Gas to Liquid) products or
methanol or dimethylether (DME). ATR is, however, sensitive to carbon formation and mechanical design due to
the elevated temperature appeared thereupon. It is important to select proper design parameters for ATR
considering chemical reaction equilibrium with respect to thermal cracking of methane resulting in carbon
formation and to make proper design of mixing between reactant and oxygen in order to avoid inadequate
oxidation which may lead to more elevated temperature. To ensure the design geometry and material selection
properly, the method of Computational Fluid Dynamics (CFD) is commonly used. The CFD solves the equations
of momentum, heat and mass transports with chemical reactions at the same time under fully turbulent flow
field. The suitable models for mixing and turbulence have to be used for the CFD analysis. This paper reviews
the reaction phenomena as well as fundamental equations for the design of ATR and discusses the model
parameters and, furthermore, it suggests what should be kept in mind with respect to risks due employing and
applying such design parameters to actual plant design.
Keywords: ATR, Oxygen Nozzle, Carbon Formation, Turbulent Model, Incomplete Combustion
INTRODUCTION
Natural gas is known as cleaner energy than oil and
coal. Methane has to be liquefied in the temperature
lower than -162°C for transportation at atmospheric
pressure from resource country to consumer country.
Keeping the liquefied state refrigeration energy is
required at all the places of handling such as shipping
yard, transportation vessel and receiving terminal.
Associated gas co-produced in the oil exploration field
is usually burnt and vent to atmosphere since there is
no measure to store it economically. It seems a good
idea to convert natural gas to transportable liquid.
Several projects to construct GTL plant for the above
purpose are currently in consideration in this planet
especially in the areas where cheap natural gas or
associated gas is available. Historically, Syngas
consisting of H2 and CO obtained by coal gasification
was converted to liquid fuel (Synfuel) by way of
Fischer-Tropsch (FT) Synthesis in Germany or in Japan
before World War II, or in the Republic of South Africa
since the fifties until today, especially during the era
when the countries were isolated from the world
econo-politically. While Sasol of the Republic of South
Africa is still operating the Coal - to-Synfuel Plants, the
company is also said to be considering change of the
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
54
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
feedstock from coal to natural gas or challenging to
invest in GTL Plants in the Middle East or Nigeria and
so on. A 34,000 bpsd GTL project by a joint venture of
Sasol and Qatar General Petroleum Co. (QGPC) has
been in progress since May 2003 and it is the first one
of such investment by Sasol. For Sasol, this project will
be the first plant to produce Synfuel from natural gas
and the plant employs ATR for Syngas generation. It is
also well known that Shell is operating a GTL plant in
Bintulu, Malaysia with the capacity of 12,500 bpsd
processing natural gas since 1988 and that Petro SA
of the Republic of South Africa (ex Mossgas) has put a
GTL plant on stream in Mosel Bay since 1992 also
processing natural gas with a capacity of 34,000 bpsd.
Shell used Partial Oxidation (POx) and Petro SA applied
a variant Combined Reforming (CR) technology for the
Syngas generation from natural gas for each GTL plant.
All the above three plants use pure oxygen. The scale
of the Syngas Generation Section in a 34,000 bpsd GTL
plant is almost equivalent to that of a 10,000 t/d
methanol plant which is three to four times larger than
the world scale methanol plant in commercial
operation today and parallel units for the Syngas
generation are actually installed in both the Shell and
the Petro SA plants. In addition, methanol plant itself
is also desired today to be scaled-up to a single train
for 10,000 t/d production as feedstock to DME or
methanol-to-olefin (MTO) production. In general, it is
believed that the ATR configuration using oxygen is
more economical than any other configurations for
reforming natural gas to achieve such scale of Syngas
generation.
This paper reviews the theory of ATR for reforming
natural gas and discusses the issues currently existing
for designing such a reforming system and equipment.
SYNGAS GENERATION REACTIONS
Methanol industry applies steam reforming
technology to produce methanol. Natural gas is
converted to Syngas consisting of H 2 and CO in
accordance with the steam reforming reaction (1) as
the first step.
CmHn + m H2O = m CO + (m+n/2) H2
(1)
The reaction (1) is catalytic and endothermic, and the
reaction heat is provided externally by radiation of fuel
combustion. The composition of reformed gas is
determined by the chemical reaction equilibriums
both for Steam Methane Reforming (SMR), i.e. m = 1
and n = 4 in the reaction (1), and the following CO shift
reaction (2) at the operating pressure and temperature.
CO + H2O = CO2 + H2
(2)
The reformed gas composing of H2, CO and CO2 can
be make-up gas for methanol synthesis usually
without any adjustment of composition because of the
following reactions as the second step.
CO + 2 H2 = CH3OH
(3)
CO2 + 3 H2 = CH3OH + H2O
(4)
Methanol is a product transportable at atmospheric
pressure and temperature in liquid form. Thus
methanol synthesis from natural gas is a sort of the
purposed GTL technology but consisting of two-step
conversions. Although methanol might be produced
in accordance with the selective oxidation reaction (5),
it has not accomplished commercial production yet
with sufficient efficiency of selectivity for the product,
for which an excellent catalyst has to be found out.
CH4 + 1/2 O2 = CH3OH
(5)
Today, the two-step conversion process in accordance
with the reactions (1) (2) (3) and (4) is still the scheme
of methanol production from natural gas. It is
apparent from reactions (3) and (4) that the ratio R
defined by the equation (6) below equals ideally to 2
(two) when the relationship between the molar
components of Syngas, i.e. H 2 , CO and CO 2 , is
stoichiometric for methanol synthesis.
R = ([H2] - [CO2]) / ([CO] + [CO2])
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
(6)
55
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Syngas is also converted to Synfuel via FT synthesis
(7) & (8), and GTL usually means to obtain the FT
products from natural gas.
n CO + (2n+1) H2 = C nH2 n+2 + n H2O
(7)
n CO + 2n H2 = C nH2 n + n H2O
(8)
The difference of the Syngas requirement for FT
synthesis from that of methanol synthesis is the
stoichiometric ratio of Syngas compositions. This ratio
of [H2]/ [CO] should be 2 while CO2 remains inert in
this synthesis.
DME is a possible alternative fuel having similar
physical properties of LPG and produced from
methanol by the dehydration reaction (9) as follows
2 CH3OH = CH3OCH3 + H2O
(9)
or is converted directly from Syngas by the following
reaction (10) or (10a)
3 CO + 3 H2 = CH3OCH3 + CO2
(10)
2 CO + 4 H2 = CH3OCH3 + H2O
(10a)
In ammonia industry, most producers in the world have
gradually changed the feedsock from coal in the old
days or from naphtha in 1960s - 1990s to cheap natural
gas. Natural gas is converted to Syngas in accordance
with the steam reforming reaction (1) as the first step
and, different from GTL or methanol synthesis, CO and
CO2 have to be completely removed from the makeup stream of Syngas to ammonia synthesis loop. Thus,
shift conversion of CO under the reaction (2) is
executed at lower temperature as low as possible using
very active catalyst, and further CO2 is removed from
shift-converted gas by wet scrubbing process. The sopurified hydrogen reacts with nitrogen to form
ammonia with the reaction (11) as the final step.
N2 + 3 H2 = 2 NH3
56
Technology Cluster: OIL AND GAS
Because of the equilibrium of the SMR reaction (1) and
the shift reaction (2) in the fired heated steam reformer
(the primary reformer), certain methane remains unreformed. In ammonia plant, therefore, the residual
methane is further reformed at the elevated
temperature of ca. 1000°C, where the endothermic
reaction heat for the residual methane is provided by
combustion of a part of hydrogen contained in the
reformed gas using air in an adiabatic packed bed
reactor (the secondary reformer). Nitrogen in the
combustion air should be in stoichiometric balance to
react with hydrogen to produce ammonia. The exit
temperature of the primary reformer is determined to
meet thermal balance of this sequence. In ammonia
plant, as above, the CR consisting of the steam
reforming primary reformer and the air blown
secondary reformer, is essential as process
configuration for Syngas generation and is successfully
proven in commercial plants over the world.
The concept of ATR is to execute this two-step
reforming in a single fixed bed reactor, however,
without any external supply of the reforming reaction
heat. In the process configuration of ammonia plant,
it is thermodynamically impossible by any means to
configure it without considerable excess N2 (air) over
H2 stoichiometrically required for ammonia synthesis.
On the other hand, for methanol plant, it seems to be
possible to configure it thermodynamically. For
example assuming, for simplicity, the feedstock natural
gas to be 98.5% methane, 1.0% ethane and 0.5%
propane, then the following overall reaction (12) can
be accomplished at 2 MPa and 1,000°C of the ATR exit
conditions as shown in Figure 1.
CH3.96 + 0.55 O2 + 0.62 H2O → 2.02 H2 + 0.86 CO +
(11)
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
0.14 CO2 + 0.59 H2O
(12)
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 1: ATR Mass Balance Example
In accordance with the equation (6), R equals to 1.88.
Therefore, it is necessary to adjust to get the
stoichiometric 2.0 prior to the methanol synthesis by
removing small amount CO2 or adding small amount
H2 recovered from purge gas from methanol synthesis
loop.
For the make-up gas of GTL product, however, it is
thermodynamically difficult to produce the Syngas of
[H2] / [CO] = 2.0 in the ATR. Figure 2 illustrates this view
with a parameter, the Steam-to-Carbon (S/C) ratio.
It is said that ConocoPhillips started a 400 bpsd GTL
pilot plant in June, 2003, with so-called Catalytic Partial
Oxidation (CPOx) technology of its own for the part of
Syngas generation, which claims to produce Syngas
of the ratio [H2]/ [CO] = 2. If their approach is same
following the above evaluation, the CPOx technology
seems to be equivalent to ATR with almost nil steam
injection. Figure 2 also indicates that the most
significant indicator for economics, the total Syngas
yield, ([H2]+ [CO]) / [NG] , decreases dramatically when
S/C is close to zero. It may be also economically of
question to develop an excellent catalyst to reform
natural gas to Syngas without steam injection due to
the increase of natural gas consumption.
OXIDATION REACTIONS
It is important to understand how oxygen reacts with
hydrocarbons, H2 and CO respectively in ATR. Figure 3
illustrates combustion of CH4 and how excess oxygen
and fuel CH4 concentrations decrease in contrast as to
how combustion products, H2O and CO2, increase in
the flame distance [2].
Figure 2: H2/O2 Ratio vs. S/C
The increase of steam leads to increase of H2 and CO2
in accordance with the equilibrium of shift reaction.
Thus, while the R-Value for methanol synthesis does
not differ too much, the [H 2]/ [CO] ratio changes
considerably depending upon the S/C ratio.
It is also remarkable to be able to recognize that the
ratio [H2]/ [CO] = 2 for GTL product is not obtained
unless steam injection is eliminated for Syngas from
natural gas by ATR.
Figure 3: CH4-O2 Flame Concentration Profile
CO and CO2 are increasing remarkably with the same
rate until almost complete consumption of CH4. Even
in such an environment of excess oxygen, CO is an
equivalent oxidation product to CO2 in a sense of the
reaction rate of methane with oxygen. This suggests
that the complete combustion consists of the
following two-step reactions (13) and (14). That is,
hydrocarbon is firstly oxidized with oxygen
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
57
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
incompletely and CO is produced together with water,
and the product CO is further oxidized with excess
oxygen as the second step to complete the
combustion.
CmHn+(m/2+n/4) O 2 = m CO +( n /2) H2O
(13)
m CO + (m/2) O 2 = m CO2
(14)
Kamenetskii used the word ‘After Burning’ for the
equation (14) [3]. Even with excess oxygen it is in fact
CO which proves to be the determining factor. The
combustion of dry CO is said to be very difficult. The
incomplete combustion reaction (13) is the same as
the POx Reaction (15) followed by very rapid hydrogen
combustion reaction (16) below.
CmHn + (m/2) O 2 = m CO + (n/2) H2
(15)
(n/2) H 2 + (n/4) O 2 = (n/2) H2O
(16)
Figure 4 presents the concept of such a mechanism as
the combustion of hydrocarbon around the fuel
burner. The jet fuel stream causes good mixing of fuel
and oxygen as well as intermediate oxidation products
such as H2 and CO, and relative difference of reaction
rate among this mixture is demonstrated in the
illustration for the combustion.
Where existing excess oxygen does not exist but rather
sub-stoichiometric oxygen like in the ATR
environment, the After Burning (14) is not likely to
occur.
Figure 4: Open Air Combustion Mechanism
58
Technology Cluster: OIL AND GAS
In ATR, in which oxidation occurs differently from
combustion, oxygen is fed into hydrocarbon feedstock
through an Oxygen Mixing Nozzle (not ‘Burner’ as this
device does not aim to combust fuel completely, but
to oxidize it incompletely). At the tip of the Oxygen
Mixing Nozzle, the hydrocarbon feedstock, completely
mixed by turbulence of the jet stream from the Oxygen
Mixing Nozzle and approaching to oxygen, reacts with
oxygen under excess oxygen condition. Mixing is
important in this step for hydrocarbons in order to
contact with oxygen steadily. In accordance with the
reactions (13) or more precisely (15) and (16), and
before commencing the slow reaction (14) all the
remained oxygen is consumed by H2, the intermediate
product, in accordance with the reaction (16).
Finally the oxidized gas composition is determined in
accordance with the chemical equilibrium of shift
reaction under sufficiently elevated temperature
enough to approach to the equilibrium
simultaneously. As illustrated in Figure 6, the adiabatic
temperature of this reaction zone reaches ca. 2000 °C.
If the mixing is performed badly, depending upon the
concentration of the intermediately produced H2 at
each instance the oxygen boundary would be broaden
and shortened repeatedly. This may cause non-steady
state of oxidation in ATR.
This temperature is more than enough for thermal
cracking of surrounded methane at the oxidized
boundary and to form carbon in accordance with the
Figure 5: Incomplete Combustion in ATR
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 6: Thermal Cracking of Hydrocarbon
Figure 7: O2 Blown Secondary Reforming
equilibrium reaction (17). The deposit carbon may
cause to plugging of micro pores of the catalyst loaded
in ATR leading to its short life.
POTENTIAL OF CARBON FORMATION
CH4 = 2 H2 + C
(17)
The oxygen blown secondary reforming of CR is
different from ATR since oxygen is mostly consumed
by the hydrogen combustion, the reaction (16), in the
secondary reformer as illustrated in Figure 7.
The difference of oxidation reaction between ATR with
reaction (13) and the O2 blown secondary reforming
with reaction (16) is obvious from the illustrated
comparison of Figures 6 and 7. It is explicit that
existence of H2 is significant to avoid danger of carbon
formation and/or higher adiabatic temperature at the
oxidized boundary as a fact that ammonia plant is
actually operated with air blown secondary reformer
with ca. 40% H2 and ca. 7% CH4 composition in the feed
gas to the oxidation reactor.
Some people are saying that the design of ATR and
the O2 blown secondary reformer are similar except
for the difference of volumetric ratios between feed
gas and oxidant flow. But the oxidation reactions are
considerably different. If one did not recognize this
difference for his design, one would not understand
the reason why problems occur related to carbon
formation
In this section, the potential of carbon formation in ATR
is examined. As illustrated in Figure 6, thermodynamics
predicts the maximum adiabatic temperature within
the oxidation boundary in ATR to be ca. 2000°C under
complete mixing.
In such a temperature range, it is well known as a result
of the research work for flame structure specifically
performed for internal combustion engine that the
actual adiabatic flame temperature is lower than the
computed value somewhat by about 10%. This is due
to radicals creating plasma state at such temperature.
In this connection, Westbrook and Dryer [4] describe
that the radicals cannot react with CO and H2 when
fuel or other hydrocarbon species still remain. Radical
species levels are kept very small until the
hydrocarbons are consumed, whereupon CO and H2
oxidation can begin. As a result, there is no means of
keeping the radicals concentration in regions in which
fuel remains. This insight would be applicable to the
environment for ATR and the O2 blown secondary
reforming.
On the other hand, some people say that there are
hydrocarbon radicals being precursors of carbon soot
in the high temperature flame in ATR or the O2 blown
secondary reformer. They also say that only intensive
mixing is essential in order to avoid soot formation.
However, it would be extremely difficult in
measurement of the concentration with respect to
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
59
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
such radicals and resultant soot quantity in the
oxidation region of ATR.
Thus, in order to examine the potential of the carbon
formation, the effect of possible radicals is omitted in
accordance with the above insight of Westbrook and
Dryer, i.e. simple thermodynamic heat balance is
assumed to decide the adiabatic temperature at which
the potential of carbon formation is discussed.
Figure 8: Equilibrium Line for Carbon Formation
According to the thermodynamics, the equilibrium line
for the carbon formation is given as a C-H-O ternary
diagram. The upper side of the equilibrium line of the
ternary diagram of Figure 8 is the area of the carbon
formation at 2 MPa and 1,000°C. The marked points
are corresponding to H2 (C=0%, H=100%, O=0%),
methane (C=20%, H=80%, O=0), CO (C=50%, H=0%,
O=50%), CO2 (C=33.3%, H=0%, O=66.7%) and water
(C=0%, H=66.7%, O=33.3%) respectively. Methane and
CO are potentially decomposed to carbon at 2 MPa
and 1000°C.
Figure 9 presents the effect of temperature on the
equilibrium line for the carbon formation. For the
temperature above 1200°C, the equilibrium curve
converges with the straight line connecting H2 and CO
in the ternary diagram.
Figure 9: Temp vs. Carbon Formation
Similarly it is also confirmed that the change of
pressure does not affect the curve of the equilibrium
for the carbon formation as far as the temperature is
above 1200°C.
For S/C=0.01, 0.6, 1.0 and 3.0, the adiabatic temperature
at the oxidation boundary inside ATR are estimated as
2058°C, 1988°C, 1,868°C and 1,505°C respectively for
the feed NG in Figure 1. As the temperature above
1200°C gives an almost identical equilibrium curve for
carbon formation, the influence of S/C is expressed in
Figure 10 plotted with the line for 1400°C. The point
for S/C=0.6 is located just below the boundary of
carbon formation.
Figure 10: Temp vs. S/C for Carbon Formation
60
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 11: Operating Condition of Air Blown Secondary Reformer
of Ammonia Plant
Figure 11 presents an actual design of the secondary
reformer in an ammonia plant commercially in
operation without problem at the capacity 1,650 t/d
under S/C=3 for the primary reformer. Figure 12
introduces a criteria of the temperature at the oxidized
boundary (1357°C) to be almost same as that of the
ammonia plant (1306°C) with the amount of oxygen
determining the outlet temperature of the catalyst bed
to be 1000°C under S/C=2.5 at the inlet of the primary
reformer. The R-Value is adjusted to be 2.0 by the exit
temperature at the outlet of the primary reformer, but
the design parameters for the oxygen blown
secondary reformer shall be essentially based on the
analogy for the above ammonia plant configured with
a CR Syngas generation. The operating points are
marginally below the carbon formation boundary.
The above ternary diagrams are, however, at the state
of chemical equilibrium and there is no reaction path
identified. This is equivalent to the carbon, produced
according to the reaction (17), which may immediately
disappear because of the following reaction (18) or the
reverse Boudouard Reaction (19). However, these
endothermic reactions should be catalytic reactions
and is unlikely to take place within such a short contact
time without catalyst in the subject oxidation zone.
Figure 12: Recommendable Design of O 2 Blown Secondary
Reformer of CR Methanol Plant
Tablel 1: Comparison of ATR and O2 Blown Secondary Reformer
of CR
Feed Gas to ATR
NH3
S/C
2.2
Water 42.18
N2
0.33
Ar
0.00
CO
5.68
CO2
6.36
H2
38.17
C1
7.29
C2
0.00
C3
0.00
Total
100.00
Press bar
Temp ºC
37
823
Intermidiate Gas at Oxidized Boundary
MeOH ATR
NH3
MeOH
1.6
0.6
S/C
2.9
2.1
34.03 37.97 Water 45.25 44.43
0.12
0.00
N2
16.84
0.14
0.00
0.00
Ar
0.20
0.09
7.48
0.00
CO
4.60
7.48
6.01
0.00
CO2
5.16
6.01
44.45 0.00
H2
22.03 33.94
7.90 61.10 C1
5.91
7.90
0.00
0.62 Carbon 0.00
0.00
0.00
0.31
100.00 100.00 Total
100.00 100.00
20
785
20
550
Oxidated
Temp ºC 1,302
1,357
ATR
2.6
32.20
0.00
0.00
4.95
5.51
38.87
1.80
16.68
100.00
1988
C + H2O = CO + H2
(18)
C + CO2 = 2 CO
(19)
Table 1 compares ATR of S/C=0.6 with the above
ammonia and methanol plants for the terms of feed
gas composition to ATR or the O2 blown secondary
reformer together with the estimated gas composition
at the oxidation boundary based upon the oxidation
models of Figure 6 and Figure 7 respectively where
carbon formation by thermal cracking is also
considered for the Figure 7 model. To incorporate the
kinetics of the thermal cracking reaction (17) into the
mass balance for the composition, the technique using
the same constant approach temperature (500°C as
an example) for the reaction (17) to the equilibrium is
applied. The reactions (18) and (19) are assumed not
to occur hereunder.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
61
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
The mass balance explicitly confirms that, if a certain
reaction path is selected intentionally, the carbon
formation is potentially possible by thermal cracking
of methane to hydrogen and carbon due to the
considerably high temperature of 1988°C for the case
of ATR.
The carbon formed by the reaction (17) is very fine and
usually most of it pass through the catalyst bed and is
detected in the process condensate in the down
stream of the process while certain carbon starts
blocking the micro pore of the catalyst resulting in
shortage of the catalyst life. However, some
researchers or developers of the ATR technology seem
to be reporting that carbon was not detected both in
the process condensate and in the catalyst bed
opened for inspection after the test runs. The author
believes that such case would be reported on the test
result with inefficient continuous run with respect to
the catalyst life. While the catalyst is fresh and active
for steam reforming before blockage with carbon, the
carbon is gasified to H2 and CO in accordance with the
reaction (18). Thus, a test run only with short period
does not prove that carbon is not formed in the
oxidation boundary in ATR and people realize it when
it was found that the catalyst life was abnormally short
in an industrial scale operation.
Technology Cluster: OIL AND GAS
complete combustion tends to occur, which is quite
different from the case of combustion with excess air
and this would cause thermal cracking of
hydrocarbons to carbon more favorably in accordance
with the reaction (17) due to certain local spots of
extremely high temperature, probably of 3000°C –
4000°C.
Therefore, the design of the mixing nozzle for ATR is
very important. Today, CFD is commonly used in
designing ATR.
Fundamental Equations
The flow behavior under fully turbulent conditions in
ATR can be identified as solutions of the equation for
conservation of mass (20) and the Navier-Stokes
equation of motion complied with the term of the
Raynolds stress (21). The compressibility of fluid is
neglected in this discussion.
(20)
(21)
COMPUTATIONAL FLUID DYNAMICS (CFD)
where, for averaging with respect of time
In the sub-stoichiometric condition for oxygen in ATR,
as above, the complete combustion of natural gas is
not likely to occur if oxygen and the feed gas are mixed
sufficiently. The role of so-called ‘oxygen burner’ is to
ensure mixing not for complete combustion with the
reactions (13) and (14), but for the incomplete
combustion based on the reaction (13) only to
broaden the oxidation boundary. In this sense, the
words ‘burner’ or ‘flame’, which is used for ‘combustion’,
suitably should not be used for the description of the
phenomena in this zone. Although a temperature
profile similar to the flame is observed there, the feed
gas is not burning like flare. This is why the author
uses the words ‘mixing nozzle’ or ‘iroxidation’
preferably. If the mixing is not sufficient, locally
62
with T large compared with the time scale of turbulent
motion. τki is momentum flux due to viscosity µ of the
direction k to the direction i given by equation (22)
below.
(22)
gi is gravitational constant to the direction i and δki is
the Klonecker’s delta.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Similar equation for the conservation of a
transportable scalar quantity is derived in a turbulent
flow, and equations (23) and (24) are respectively for
the mass fraction Y of chemical specie j to the direction
k and the enthalpy h to the direction k.
transportation and energy transportation by the
component of turbulent fluctuation respectively. It
seems to be a reasonable hypothesis that
transportation quantity by the turbulent component
is proportional to the gradient of mean quantity of the
transportation per unit volume, i.e.
(23)
(25)
(26)
(24)
jjk is mass diffusion flux of the chemical specie j,
almost equivalent to
, where Djm is multi-
component effective diffusion coefficient of the specie
—
j. Rj,f is the production rate of the chemical specie j by
chemical reaction f. qk is heat flux to the direction k
relative to mean mass velocity almost equivalent to
, where λ is thermal conductivity, and Qk is
heat input to the direction k to increase the enthalpy
h. In ATR Qk is, therefore, zero if heat loss is neglected.
For the steady state the differential term with respect
to time is also zero. If the components are all gasses,
the gravity term of the equation (21) is also neglected.
In order to resolve the above equations to obtain —
uk ,
—
—
Y j and h as a function of the location xk , the correlation
terms of turbulent fluctuation ρu’k u’i (Raynolds Stress),
ρu’k Y’j and ρu’k h’ are new terms to have to resolve at
the same time. The equations group does not
therefore close by themselves without addition of
some other equations relating to these new terms.
Each of the above three correlation terms is
corresponding to momentum transportation, mass
(27)
where,
is the turbulent energy and µt , Djt
and λt are turbulent transportation coefficients for
momentum, chemical specie j and enthalpy
respectively, namely so-called Effective Eddy Viscosity,
Effective Eddy Diffusion Coefficient and Effective Eddy
Conductivity. They are not real material properties and
differ depending upon flow velocity and other
conditions of turbulent field. Thus, the problem to
resolve the equations boils down to how to estimate
the above three turbulent coefficients.
Experience is telling that the Effective Schmidt
Number and the Effective Prandtle Number are known
to be close to unity. Thus the turbulent correlation
terms for mass transportation and enthalpy
transportation can be approximated by the Effective
Eddy Viscosity.
(28)
The κ – ε model is one of the measures for turbulence
preferably being used in many CFD programs to give
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
63
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
the Effective Eddy Viscosity by so-called PrandtleKolomogorov Model with the equation (29).
(29)
where, k is the turbulent energy and is energy
dissipation rate both to be derived, for example, from
the equation for momentum conservation (details to
be referred to the books of Rotta [5] etc.), which results
are as follows.
(30)
Technology Cluster: OIL AND GAS
experimental constants and they are not known
whether they have solution under mathematical
uniqueness. The turbulent models currently available
for CFD for the ATR design should be recognized
having the above issues and should be used just for
reference only for the intended design.
The κ – ε model provides with all the terms relating to
—
the correlation of turbulent fluctuations and only Rfj,
in the equation (23) remains unknown. We have also
to introduce a certain model of chemical reaction rate
in the turbulent flow field.
Based upon the well-known Arrhenius kinetics rate
expression, the net production rate of the reaction f
for the chemical specie j is given as the equation (32)
with the absolute temperature T.
(32)
(31)
In the above equations, CD, C1, C2, σk and σε are all
experimental constants based on the said model. Kent
and Bilger [6] suggested using the values CD = 0.09,
C1 = 1.45, C2=1.92, σk = 0.7 and σε = 1.3 respectively
for the combustion field. However, nobody seems to
have yet confirmed experimentally that these
constants are also applicable to the turbulent field in
ATR where the chemical reactions occurring are
considerably different from those of the complete and
excess oxygen combustion.
The two equation model such as the above κ – ε model
has performed excellently to resolve certain turbulent
problems until today, however it would not be able to
go beyond the bounds inevitably for such a flow field
sacrificed with sudden change with large turbulent
fluctuation against main flow as far as the model uses
the concept of the traditional and visceral Eddy
Viscosity for the term of the Reynolds stress.
Although there are attempts to use so-called ‘stress
equations model’ starting with putting away the
assumption of the Eddy Viscosity, the group equations
are of strong non-linearity with much more
64
CA,f , βf , Ef and αf are experimental constants for the
reaction£f and available in various books and papers
[2] [3] [4]. If the flow field is laminar, this expression is
applicable. It is generally not accurate for the highly
turbulent flow.
Magnussen and Hjiertager [7] introduced a model for
the rate of reaction to be relating to the rate of eddy
dissipation considering turbulence. The so-called Eddy
Dissipation Model is currently more commonly used
for the CFD analysis for combustion. In this model, the
net production rate of the chemical specie j to be
oxidized in accordance with the specific oxidation
reaction f is described as follows.
(33)
In the equation (33), rs is the stoichiometric mass ratio
of oxygen over the chemical specie j for the oxidation
reaction f. This model refers to a concept that the
oxidation occurs when the gaseous mass of chemical
specie j and oxygen mass are dissipated from each
other and reach to the micro scale mixing situation.
Equation (33) indicates that the dissipation of oxygen
mass will control the reaction rate if oxygen is
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
insufficient in mass for the intended oxidation such as
the case of ATR. ε and k can be calculated by the
equations (30) and (31) respectively and CB,f , is an
experimental constant which is said commonly 4.0 for
usual combustion of any kind fuels.
It might be true that most fuel combustion reactions
are very fast compared with the time scale of mixing,
then once the reactants are mixed, the reaction may
be considered to be completed. However, as a general
form to consider both the chemical reaction kinetics
and turbulent mixing for the rate of reaction f, the
following hybrid equation (34) is useful.
Figure 13: Velocity Profile
(34)
It should be in consideration especially when the
oxidation rate seems to be slow and doubtful against
the time scale of mixing in question. It may limit the
production rate.
Example of CFD Analysis
Figures 13 – 17 illustrate some examples of the result
using CFD on an O2 blown secondary reformer which
are presented in the recent report of the author [1].
Figure 14: Detailed Profiles around O2 Mixing Nozzle
A simple pipe nozzle, without any swirl, as oxygen
mixer into feed gas is provided in the top of the reactor
and feed gas is charged from the side of the neck of
the reactor at a certain location considered for the
desirable mixing of the oxygen and the feed gas [8].
Figure 13 shows velocity profile at the center
symmetrical plane. At the mixing point the
momentum of feed gas is pushing the high velocity
of oxygen stream to the opposite side.
As it can be seen more explicitly in Figure 14, the
incoming feed gas stream actually pushes the jet
formed by the oxygen stream to the side opposite the
inlet duct of the feed gas. Namely, the cold bulk of the
feed gas is redirected after colliding with the opposite
wall and dividing into two directions through wall
curvature. This divided flow and the wall induced
recirculation contribute to mixing of the two inlet jet
streams, which results in localizing higher temperature
zone to be far from the reactor wall. This flow pattern
also maintains the stream of oxygen rich gas to be
flowed down toward the catalyst bed keeping
sufficient distance from the wall of the neck part of
the reactor.
Figure 15 presents the temperature profile. The
geometry of the reactor is determined and designed
so that the temperature profile on the wall of the
reactor can be with minimum difference through the
whole wall allowable for mechanical design of the
reactor in accordance with trials of the CFD analysis.
In order to ensure complete mixing of feed gas and
oxidant, considerable effort to develop an oxidant
burner with different slots or swirls has been
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
65
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
the feed gas stream rather pushes the bulk of oxygen
toward the top of the neck of the reactor. Then, the
extremely hot zone due to the complete combustion
with excess oxygen appears there.
Figure 17 is a CFD example for the carbon formation.
In this example, certain amount of carbon (max. 1.2
ppm mol) is observed being produced at the area of
the highest temperature, which is near the oxidized
boundary of Figure 6 due to thermal cracking of
methane.
Figure 15: Temperature Profile
Figure 16: Temperature Profile with Slotted O2
In the above example it is assumed that the rate of
methane thermal cracking is instantaneous and the
result corresponds to the severest situation. As the
above the first term of the equation (34) plays a very
important part if the reaction rate is slow compared
with the second term. More essentially, the reaction
models, identifying which kind of reactions under
which exponential experienced numbers for the
reaction rates influence the CFD analysis in question
dramatically. Careful and sufficient experimental data
should be collected for each specific configuration for
each ATR on a case-to-case basis.
CONCLUSION
In ATR, different from and on the contrary to ordinary
combustion phenomena, complete combustion does
not occur if oxygen and feed gas are mixed
instantaneously and sufficiently. The role of oxygen
feeding device is mixing for achieving incomplete
combustion as uniformly as possible. In this sense, the
word “Oxygen Mixing Nozzle” is more suitable than
“Oxygen Burner”.
Figure 17: Carbon Formation Profile
accounted historically in the field of the secondary
reformer of ammonia plant. Figure 16 illustrates an
example of CFD result using certain slotted O2 tip in
the oxygen mixing nozzle.
The result is quite obvious in this example. The slots
disturb oxygen to collide with feed gas directly and
66
It is further understood that, if the mixing is not
sufficient, the complete combustion does tend to
occur locally, where excess oxygen meets less methane
causing much higher temperature spot locally. The
extremely high temperature leads to thermal cracking
of methane to carbon.
Thus, the Oxygen Mixing Nozzle for ATR is very
important and has, therefore, to be specially designed
with CFD analysis. Not only to detect the allowance of
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
wall temperature for the reactor shell or the entrance
of the catalyst bed in ATR but also to minimize the
possibility of carbon formation the correct models for
reaction rate of the incomplete combustion as well as
for turbulence are required in CFD.
In the event that there is no experimental data
available in one’s hands as the model parameters for
the intended ATR design, one might use those for a
turbulent combustion model, probably by applying
default values of a CFD simulator for ordinary
combustion. The reaction models and parameters
currently used for CFD in evaluation for the design of
ATR seem to be still all collected and based upon
complete combustion and under ambient
temperature. Furthermore, no kinetics for carbon
formation is taken into consideration.
In the CFD analysis for the design of ATR it is important
to identify whether or not the models for reaction rate
and turbulent mixing are exactly applicable for the
situation of ATR, which is sub-stoichiometric,
considerably far from complete combustion and under
pressure. Careful examination is required whether or
not the applied CFD models and parameters are
reliable or acceptable. The information with respect
to carbon formation must be endorsed at least in a
format of chemical equilibrium in such an examination.
It is, for example, suggested to try to change and
examine the sensitivity of the parameters one by one
and should assess at first the applicability of the fixed
value for the complete combustion to the present
issue for incomplete combustion in ATR.
ATR is regrettably not reliable at present against
carbon formation compared with O2 blown secondary
reformer in CR with primary H2 production, in which
H2 consumes all the supplied O2. Improvement of the
mixing design would not resolve this issue but it would
be only possible under the approach considering
chemical reaction equilibrium for the carbon
formation together with finding out the real reaction
rate under the designated S/C ratio and/or hydrogen
ratio in the feed gas.
ACKNOWLEDGEMENT
I wish to thank Dr. Masakazu Sasaki and Dr. Katsunori
Yagoh, both of Toyo Engineering Corporation, Chiba,
Japan. Dr. Sasaki assisted me with discussion on
chemical reaction equilibrium and Dr. Yagoh provided
me with specific examples on CFD modelling. Both
have continued their support throughout my
preparation of this paper.
REFERENCES
[1] Hirotani, K. 2003. What is the Problem in Autothermal
Reforming for Syngas Generation: Presentation at EFI Member
Conference ioAPEC Energy ΠNow and in the Futureln,
Yokohama, Japan.
[2] Fristrom, R. M.; and Westenberg, A. A. 1965. Flame Structure,
McGraw-Hill Book Company.
[3] Frank-Kamenetskii, D. A. 1969. Diffusion and Heat Transfer in
Chemical Kinetics, Second enlarged and revised edition
(Translated from Russian) Prenum Press.
[4] Westbrook, C. K. and Dryer, F. L. 1981. Simplified Reaction
Mechanisms for the Oxidation of Hydrocarbon Fuels in Flames:
Combustion Science and Technology, Vol. 27, pp 31-43.
[5] Rotta, J. C. 1972. Turbulence Strömungen: B. G. Teubner,
Stuttgart.
[6] Kent, J. H. and Bilger, R. W. 1976: The prediction of turbulent
diffusion flame field and nitric oxide formation. Proceeding of
16th Symposium on Combustion, pp 1643-1655. The
Combustion Institute.
[7] Magnussen, B. F. and Hjiertager B. H. 1976: On Mathematical
Modelling of Turbulent Combussion with Special Emphasis on
Soot Formation and Combustion. Proceeding of 16 th
Symposium on Combustion, pp 719-729. The Combustion
Institute.
[8] Patent Application No. JP2002-236447.
Kunio Hirotani joined Toyo Engineering
Corporation ( TEC) in 1976 as a process
engineer and had been involved in process
design of ammonia, methanol and
hydrogen production processes as well as
a new process development for methanol
production until 1983. Then in 1984, he
moved to London as Technical Manager of
Europe Branch to acquire technologies of
European licensors for Toyo's in-licensing purposes to its
customers worldwide. In 1990, he returned to Tokyo as Licensing
Manager of the TEC headquarter and is now General Manager of
Licensing and Patents Department. On his achievement in
Licensing Department, he is proud of the technology behind the
MRF-Z Methanol Synthesis Reactor, which is claimed to be the
only reactor in the world at present capable of producing 5,000
t/d more production in a single vessel.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
67
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
DEVELOPMENT OF DEFECT-FREE AND HIGH PERFORMANCE
ASYMMETRIC MEMBRANE FOR GAS SEPARATION PROCESSES
Ahmad Fauzi Ismail1, Ng Be Cheer, Hasrinah Hasbullah and Mohd. Sohaimi Abdullah
Membrane Research Unit, Faculty of Chemical and Natural Resourses Engineering, Universiti Teknologi Malaysia,
81310 UTM Skudai, Johor, Malaysia.
1
[email protected]
ABSTRACT
Gas separation by membranes is a rapidly growing field. It has been recognized as one of the most recent and
advanced unit operations. Most commercial gas separation membranes in use today are based on the defectfree and high performance asymmetric membranes. Since the membrane is the heart of the gas separation
processes, research and development in this field has been continued extensively. This paper reviews the route
that has been taken in developing defect-free and high performance asymmetric membranes for gas separation
processes through the combine effects of primary phase inversion process and rheological factors. The former
controls general morphology of membrane whereas the latter further affects molecular orientation in membrane.
The development of these aspects represents a major breakthrough in membrane technology.
Keyword: gas separation, asymmetric membranes, molecular orientation, rheology, phase separation
INTRODUCTION
Membranes have become an established technology
for gas separation processes since their first
introduction for commercial application in the early
1980s. Gas mixtures separation such as the recovery
of hydrogen from ammonia purge gas stream and
removal of CO 2 from natural gas uses membrane
separation system because it offers numerous
advantages over the conventional technology. Among
the advantages offer by membrane are less energy
requirement, small footprint and modular and
compact system design.
The scientific and commercial progress in membrane
science and technology had accelerated over the last
twenty years through a number of routes either the
development of novel membrane materials and
fabrication processes or development of robust
process and s ystem design and simulation.
The membranes were fabricated through a simple drywet phase inversion process. This process is widely
employed in developing an integrally skinned
asymmetric membrane for almost all membrane
processes. An integrally-skinned asymmetric
membrane consists of a very thin and dense skin layer
(0.1 µm to 1µ m) overlaying on a thick and highly
porous sub-layer (100µm to 200µm with an average
void size ranging from 0.01 µm to 1 µm), where both
layers are composed of the same material and formed
in a single operation [1-3]. The skin represents the
active layer that possesses selective properties, while
the substructure provides a mechanical support for the
skin, with negligible effects on separation.
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
68
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
In asymmetric membrane, permeability of the
membrane does not depend on entire membrane
thickness, instead it is inversely proportional to
thickness of skin layer [2,4]. Hence, a high value of
permeability can be achieved in asymmetric
membrane with very thin skin layer. Generally,
asymmetric membrane with effective skin thickness
of approximately 1000 - 5000 angstroms (Å) is
classified as ultrathin-skinned membrane; whereas
hyperthin-skinned asymmetric membrane possesses
a skin layer with thickness much less than that for
ultrathins [5]. Both of them are specifically designed
to achieve practical permeability.
However, these membranes would result in defects or
pinholes on skin surface due to irregular packing of
kinked polymer chains and incomplete coalescence
of polymer molecules in skin layer [7,8]. Solutiondiffusion mechanism of permeation through
membrane materials is very slow compared to
Knudsen diffusion or viscous flow that occurs through
nonselective pores in membrane; thus a defective area
plague on membrane surface can substantially cause
a dramatic loss in selectivity and prevent the intrinsic
selectivities (i.e., selectivities measured on dense
homogeneous films) from being achieved in very thin
membranes [6].
Consequently, for any given separation, there is usually
a trade-off between permeability (skin thickness) and
selectivity (skin integrity), where both parameters tend
to exhibit a contradictory relation, representing a
major problem in productions and applications of
asymmetric membranes for gas separation process. At
present, new challenges are directed towards
development of defect-free and ultrathin-skinned
asymmetric membranes for gas separation process, in
which further insight and significant progress are
being made.
Although a considerable amount of literature is
available on gas separation, this paper mainly
describes the route taken in developing a defect-free
and high performance asymmetric membranes for gas
separation processes through the manipulation of
rheological factors. This presents a novel approach in
membrane science and technology.
MEMBRANE FORMATION FOR GAS SEPARATION
Membranes are fabricated by a process known as
phase inversion [7]. At present, phase inversion has
been universally accepted as a standard technique for
fabricating commercial membranes. Phase inversion
is referred to as a process whereby a homogeneous
polymer solution is transformed or inverted in a
controlled condition into a gel comprising a polymer
rich phase where this phase will solidify to form a solid
membrane structure and a liquid polymer poor phase
forming the voids. Thus, decomposition of a
homogeneous polymer solution into a two-phase
system is made by bringing an initially
thermodynamically stable polymer solution to an
unstable state. The thermodynamic state of a system
consisting of multicomponents can be described in
terms of Gibbs free energy of mixing. Under certain
conditions of temperature, composition or pressure, a
decrease occurs in the free energy of mixing of the
solution and this will result in phase separation of the
initially stable solution into two or more phases with
different compositions. The resulting membrane
structure is determined by a spatial distribution of the
resulting phases at the point of solidification of the
polymer rich phase. This is a very versatile technique
allowing all types of membrane morphologies to be
tailor-made according to the separation processes of
interest [7].
The phase inversion technique can be further divided
into four different techniques namely, thermal
precipitation, air casting of dope solution, precipitation
from the vapour phase and immersion precipitation
[8]. The differences between these techniques are
based on differences in the desolvation mechanisms.
Among these techniques, immersion precipitation is
widely used in producing commercial gas separation
membranes and other membrane-based separations
that are presently available. In this paper, the
immersion precipitation technique is discussed in
detail whereas for other techniques, readers are
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
69
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
encouraged to refer to the relevant book by Mulder
[9] and the literature [8].
It has been said that the phase inversion process
generally possesses the following characteristics [10].
1. A ternary system. The process consists of at least
one for each polymer, solvent and nonsolvent
component. The solvent and nonsolvent must be
miscible.
2. Mass transfer. This process starts at the interface
of the polymer film and the coagulation medium
resulting in increasing nonsolvent concentration
in the film. Compositions changes are governed
by diffusion. No mass transfer takes place in the
case of thermally induced phase separation.
3. Precipitation. As a result of the increase of
nonsolvent content, the polymer solution
becomes thermodynamically unstable and phase
separation will occur. Phase inversion is therefore
related to the demixing phenomenon and phase
equilibria and the kinetics of phase separation are
important due to the dynamic nature of
membrane formation.
In immersion precipitation, two types of phase
separation or liquid-liquid demixing mechanisms have
been distinguished resulting in two different types of
membrane morphology [9,11].
1. Instantaneous phase separation and
2. Delayed phase separation
In instantaneous phase separation, the membrane is
formed immediately after immersion in the nonsolvent
bath. The resulting membrane structure typically
consists of a very thin but microporous skin layer and
an open-cell finger or sponge substructure, which is
suitable for microfiltration and ultrafiltration
applications [9]. The process that is responsible for this
rapid phase separation is called spinodal
decomposition. This will be addressed explicitly when
discussing the types of phase inversion processes in
70
Technology Cluster: OIL AND GAS
the following sections. However, if delayed phase
separation occurs, the resulting membrane structure
often possesses a relatively dense, non-porous and
thick skin layer, supported by a closed-cell sponge-like
substructure [9].
Immersion precipitation can also be classified as a wet
or a dry process, which can further affect membrane
structure. The distinction refers to whether the casting
knife or spinneret is exposed to air or is submerged
directly in a nonsolvent coagulation bath. Another
subtlety in the process of immersion precipitation is
the possibility of a nucleation and growth mechanism
due to binodal decomposition. This can form either a
latex or a closed cell structure depending on the
conditions.
If a substantial time is allowed in the dry stage for
partial evaporation of solvent from the cast or spun
membrane prior to immersion into the coagulation
bath then the process is called dry/wet phase
separation.
The dry/wet phase separation process requires more
attention to dope solution formulation than the other
methods. Recently, advancements in dope formulation
and fabrication technique have evolved [12-15]. The
following sections describe in more detail the
formation of integrally skinned asymmetric
membranes by different phase inversion processes for
the production of gas separation membranes. The
following techniques are applicable to different
membrane geometries since the basic principles
applied in the development of asymmetric
membranes are the same for both flat sheet and
hollow fibers. However, fiber spinning involves
additional complexities due to simultaneous
precipitation from both inner and outer coagulants.
Wet Phase Inversion Process
Membrane formation by a wet phase inversion process
involves three main steps namely, 1) casting of dope
solution 2) immediate immersion of the cast dope into
a liquid nonsolvent in a coagulation bath and 3) post-
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
induced at a point denoted by * anywhere within the
two phase boundary. The corresponding tie line
dictates the composition of the resulting two
equilibrium phases and the relative proportions are
given by the position of the point * on the tie line.
In the wet phase inversion process, when a nonsolvent
is introduced into a homogeneous polymer solution
resulting in an unstable mixture, the free enthalpy of
the system is decreased when the solution splits into
two phases. This split can occur by one of the following
mechanisms: by the nucleation and growth of the less
prevalent phase or by spinodal decomposition [18].
Figure 1: Schematic diagram of an isothermal ternary phase
diagram showing the equilibrium tie lines connecting equilibrium
compositions on the binodal curve having polymer rich and
polymer poor compositions indicated as PR and PP respectively
[16,17]
treatment of the membrane. Generally, wet techniques
require the cast dope solution to be in lasting contact
with the liquid nonsolvent in the coagulation bath. The
solvent/nonsolvent exchange brings the solution to
thermodynamic instability. The membrane structure
is formed by solidification of the casting dope solution.
Wet phase inversion for a typical polymer such as
polysulfone can be illustrated with the aid of ternary
phase diagram such as that shown in Figure 1 [16,17].
The phase diagram of the ternary mixture shows a
miscibility gap over a large range of compositions. The
corners of the triangle represent pure components i.e.
polymer, solvent and nonsolvent and a points within
the triangle represent a mixture of the three
components. In this region the binodal boundary (or
the locus of the precipitation points) and spinodal
boundary can be observed. The division of these
regions occurs at the critical point, CP. The tie lines
connect points on the binodal boundary that are in
equilibrium. The composition within this two-phase
region splits into two equilibrium phases along the tie
line and are known as the polymer rich, PR and the
polymer poor, PP phases - each on opposite sides of
the critical point. The phase separation process is
If the mixture composition * lies between the binodal
and spinodal lines then nucleation and growth of the
PP or PR phase occurs producing, initially, either a
closed-cell structure or a latex respectively depending
on whether the original mixture lies above or below
CP [11,16]. Either of these two nucleation and growth
structures is thought to be unsuitable for gas
separation membranes due to low permeability and
low mechanical integrity respectively. The closed-cell
mechanism tends to produce a non-porous but
excessively thick skin. However, the phase separation
must not be allowed to linger as this would allow
coalescence of the PP phase, which would cause
defects. The latex mechanism, on the other hand, tends
to produce a thin weak porous skin but this could in
fact be densified in the subsequent processing steps.
It has also been suggested that skin formation in wet
phase inversion may occur by rapid outflow of solvent
through a gelation process. The high viscosity of the
gel layer suppresses phase separation due to kinetic
limitations on nucleation and growth. A skin layer that
is both ultrathin and defect-free is very difficult to
obtain for asymmetric membranes made by the wet
phase inversion process. Process variables that favour
the formation of an ultrathin skin layer are 1) low
polymer concentration in the dope solution, 2) short
evaporation period, 3) addition of nonsolvent to the
dope solution and 4) rapid precipitation, that is,
instantaneous phase separation in the quenched
membrane. However, these process variables are
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
71
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
contrary to those that favour the formation of defectfree skin layers. Therefore, optimization of the
formation variables is necessary to yield optimum
membrane properties.
If the point * located within the spinodal boundary,
instantaneous phase separation occurs without the
need to form nuclei. This phase separation process is
called spinodal decomposition. Spinodal
decomposition involves propagation of critical
concentration perturbations having a characteristic
dimension that tends to be reproduced
instantaneously throughout the entire macroscopic
region undergoing the phase separation [11,16]. This
process leads to the formation of very thin but
microporous skin layer if the morphology can be
trapped by sufficiently rapid solvent removal to vitrify
the polymer rich phase. However, at low polymer
concentration, nucleation and growth and spinodal
decomposition tend to produce a similar membrane.
Generally, membranes formed by instantaneous phase
separation exhibit a solution diffusion mechanism in
gas separation, only when cast or spun from high
concentration polymer solution and coated with
silicone rubber to form multicomponent membranes.
In the case of delayed phase separation, the membrane
solidifies after a certain time lag. Initially, only solvent
is lost from the dope solution and the binodal line is
approached ‘gently’ from the outside. This results in a
single gel phase polymer solidification mechanism,
which produces the nonporous skin. Ultimately,
delayed phase separation, will produce a membrane
with a dense nonporous but rather thick skin layer
supported by a closed-cell, sponge-like substructure.
This type of membrane exhibits the solution diffusion
separation mechanism and is potentially useful for gas
separation processes. However, the fluxes associated
with such membranes are rather low as the closed-cell
support tends to add resistance to gas flow.
Recently, a unique modification of the coagulation
bath in wet phase inversion has resulted in a controlled
delayed phase separation system for the formation of
defect-free integrally skinned asymmetric gas
72
Technology Cluster: OIL AND GAS
separation membranes. This technique is called the
dual-bath coagulation method [19] where membranes
are formed within two nonsolvent baths in series. The
first bath contains a weak nonsolvent causing delayed
phase separation (gel) resulting in the formation of a
nonporous skin layer. The second bath contains a
strong nonsolvent that causes instantaneous phase
separation (spinodal) in the remaining membrane
structure. Using the modified dual bath coagulation
technique, Li et al. [20] reported higher than intrinsic
selectivities for polyethersulfone after solvent
exchange drying. The defect-free skin layer thickness
reported was around 0.3 µm. The dual bath technique
is therefore highly promising in terms of industrial
exploitation.
Dry/Wet Phase Inversion Process
Integrally skinned asymmetric membranes can also be
fabricated by a process called dry/wet phase inversion
[13,16,21,22]. In this process, phase separation in the
outermost region is induced by solvent evaporation
and the remaining membrane structure is
subsequently formed by solvent-nonsolvent exchange
during a quench step. This concept is known as forced
convective evaporation-induced phase separation.
The physical events occurring during the evaporation
step have been acknowledged as the most important
aspect of dry/wet phase inversion for the production
of asymmetric membranes for gas separation [17].
Recently, Pinnau and Koros [12,22] produced an
ultrathin defect-free skin layer by using a dry/wet
phase inversion process. The skin layer achieved was
of the order of 200 Å. They also developed some
empirical rules for the formation of optimised
membranes using this technique [17].
1. The casting solution must consist of at least three
components: the polymer, a solvent and a
nonsolvent. The solvent must have a higher vapour
pressure than the nonsolvent component, so that
phase separation can be induced during an
evaporation step.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
2. The composition of the dope solution should be
as close as possible to the thermodynamic
instability limit (binodal composition).
3. The evaporation should be carried out by forcing
a gas stream across the cast membrane to induce
phase separation in the outermost region of the
cast film (dry phase inversion).
4.
The quench step (wet phase inversion) should be
carried out in a thermodynamically strong
nonsolvent for the membrane forming polymer.
The quench medium must be miscible with the
solvent and nonsolvent of the dope solution.
The evaporation process for a typical ternary dope
solution used for the preparation of membranes made
by dry/wet phase inversion is schematically
represented in Figure 2 [16,20]. Point A represents the
initially stable dope solution. It is located near the
binodal boundary. During the evaporation induced
phase separation step, a gas stream is blow over a cast
or spun dope to encourage solvent outflow and thus
bring a few microns of the nascent membrane into a
spinodally decomposed structure with an average
composition denoted as “A”. This instantaneous process
can be witnessed by the onset of turbidity in the
surface of the nascent membrane [16]. If the spinodal
structure in the outermost membrane region vitrifies
instantaneously during the subsequent quench step
(wet phase inversion) without the material having
undergone further structural changes during the initial
evaporation process, it is expected that the skin layer
of the quenched membrane would be microporous.
These pores would result from the interstitial spaces
of the polymer-poor phase present in the outermost
region of the quenched membrane at the point of
vitrification. However, recent studies have reported
that an ultrathin nonporous skin layer is achievable by
using the dry/wet phase inversion technique.
Therefore, it has been suggested that an additional
physical process leads to the coalescence of the
polymer rich regions, referred to as nodules, producing
a homogeneous nonporous skin layer during the
evaporation step. Kesting [7] also suggested this
physical process, however no detailed explanation was
given.
In the development of this theory, Pinnau and Koros
[14] reconciled the physical events occurring during
the drying of a latex to produce a nonporous film with
the events in the evaporation step that produce a
nonporous membrane skin. They proposed that
capillary pressures resulting from the gas-liquid
interface in the polymer poor region cause the nodular
regions of the polymer rich phase to coalesce. The
coalescence is faster in membrane production than in
latex drying because of the presence of considerable
amounts of plasticising solvent and nonsolvent
components in the polymer rich phase, which reduce
the rigidity modulus of the nodule.
Figure 2: Schematic representation of diffusion paths initiating
near the binodal boundary and potentially penetrating to the
metastable (nucleation and growth) region (A’), the unstable
(spinodal decomposition) region (A”) or the solidus tie line” (A”’)
where the morphology is vitrified immediately upon phase
separation and unable to evolve [16,17]
The condition for nonporous skin formation requires
that the force resulting from capillary pressure exceed
the resistance of the polymer nodule to deformation.
For the spinodal network, the two interpenetrating
continuous media are composed of the polymer-lean
phase pervading the void space between the polymers
rich-phase. Only the outermost region of the
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
73
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Technology Cluster: OIL AND GAS
This approach that was put forward by Pinnau and
Koros [16,17] provides an understanding of the
operating parameters that are likely to be important
in achieving a defect-free selective skin layer by dry/
wet phase inversion.
If the composition of the polymer rich nodules exceeds
the solidus tie line during the evaporation step then
the glassy nodules will fail to coalesce due to their high
rigidity (high G) and a microporous skin layer will be
obtained. Excessively high polymer concentrations in
the casting/spinning dope should therefore be
avoided.
Figure 3: Contracting forces resulting from the capillary pressure
in the interstitial spaces between polymer rich nodules [6,17]
membrane is exposed to the high capillary pressures
arising from the curvature of the gas-liquid interface
between nodules of the polymer-rich phase. The
capillary forces tend to deform the nodules of the
solvent-swollen polymer-rich phase, promoting
elimination of interstitial surface voids of the polymer
poor phase. Thus a defect-free gel-like surface layer is
generated, as schematically depicted in Figure 3.
The capillary pressure, Pc, can be estimated by the
Young-Laplace equation for perfect wetting
conditions (contact angle equal to zero) [16,17]:
Pc = 2γ/r
(1)
where γ is the surface tension of the interstitial fluid
and r is the effective radius of the throat that exists in
the plane passing through the centres of the three
touching spherical nodules. Compressive stresses are
estimated to be greater than 100 atm during the
evaporation step thus causing the coalescence of the
nodules. The conditions required for nodule
coalescence can be related to the shear modulus, G, of
the polymer rich nodule. The following expression was
suggested:
G < 35γ/R
(2)
The dry evaporation step described above results in a
compact, but minimally coalesced homogeneous skin
resting on top of a secondary layer of uncoagulated
dope. The subsequent quench step instantaneously
transforms the highly plasticised, but homogeneous
skin layer into essentially a solvent-free glass
containing only an equilibrium amount of the quench
medium. As the quench medium diffuses across the
homogeneous skin layer, solvent and nonsolvent
contained in the underlying secondary layer are able
to move by counter-diffusion into the miscible quench
bath thus causing the wet phase inversion of the
secondary layer/substructure. Vitrification of the
secondary layer should occur as rapidly as possible, to
avoid any loss of the interconnectivity and porosity.
The use of thermodynamically strong nonsolvents
such as water and methanol in the quench bath allows
the bicontinous secondary layer to solidify almost
instantaneously [12,22] thus causing an open-cell,
sponge-like substructure [9,16].
One of the major problems confronting the use of
membrane based gas separation processes in a wide
range of applications is the lack of membranes with
high flux and high selectivity. During fabrication,
membrane formation process plays an important role
and certain factors need proper attention in order to
produce a good gas separation membrane. Currently,
gas separation membranes technologies are
challenged to maintain their favorable economics
while improving their gas selectivity, flux and
where R is the nodule radius.
74
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
durability. The improved membranes would be
attractive in large potential markets such as CO2/CH4,
hydrocarbon/H 2 and olefin/paraffin separations
[23,24]. Though extensive research and development
are being devoted to develop high performance
membrane, most of the works done were still based
on manipulation of the phase inversion factors. Phase
inversion factors control general morphology of
membrane and hence determine basic capability of
membrane for gas separation. Phase inversion factors,
including forced-convective evaporation time,
polymer concentration and solvent ratio, have been
identified as dominating fabrication factors in
controlling skin thickness and skin integrity, and thus
have substantial effects on separation properties of
resultant membranes. However, this section explores
potential route in improving the separation
characteristics of membranes through rheological
approach, which has demonstrated significant
improvement in the separation performance.
Using this technique, essentially defect-free skin layers
membranes has been produced. Even without surface
coating, the selectivities obtained were similar to those
in dense nonporous polysulfone films. This technique
has been successfully employed for a variety of glassy
polymers either in flat sheets [14,21,22] or in hollow
fiber geometries [15]. The wet and dry/wet phase
inversion processes are the most widely used for
membrane fabrication. In the present discussion, dry/
wet phase inversion process is employed for the
fabrication of defect-free and high performance
hollow fiber membranes.
IMPROVING THE SEPARATION CHARACTERISTICS
THROUGH RHEOLOGICAL APPROACH
properties and separation performance of asymmetric
membrane. Recently, rheological factors in membrane
formation processes have been extensively
investigated by Ismail, Shilton and Chung [25-33].
Membrane formation process involves casting of a
homogeneous multicomponent solution. Behavior of
solution behavior under shear is described by
response to tangential stress, which is related to
molecular orientation or preferential alignment of
randomly coiled chain molecules [25]. If the
relationship is linear then the fluid is said to be
Newtonian;
.
τ = ηγ
(3)
where t is the shear stress, h is the viscosity and g&
is the shear rate. If the relationship is non-linear, the
fluid is said to be non-Newtonian;
.
τ = kγ n
(4)
where k and n are constants for the particular fluid. k
is a measure of the consistency of the fluid, a high value
of k indicates a more viscous fluid. n is a measure of
the degree of non-Newtonian behaviour; if n > 1 the
fluid is said to be shear thickening whereas if n < 1 the
fluid is shear thinning. This empirical relationship is
known as the power law proposed by de Waele and
Ostward [25]. Polymer solution is often shear-thinning
suggesting a progressive alignment of polymer
molecules under shear in flow direction [25,26].
Rheological tests were carried out using a rheometer,
and a typical behavior of a shear thinning solution is
shown in Figure 4 [25]. The dashed line represents
Rheological Factors
Formation of asymmetric membranes through
manipulation of rheological factors is a novel approach
in membrane technology, which provides a potential
platform to develop defect-free and ultrathin-skinned
asymmetric membranes for gas separation.
Rheological factors would affect morphology, physical
Figure 4: Shear stress versus shear rate curve for shear thinning
fluid
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
75
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 5: Ideal shear creep and creep recovery from small shear
stresses
Newtonian behavior and the solid line represents nonNewtonian behavior (shear-thinning) under shear.
Stretching of polymer chains between entanglement
junctions leads to a decrease in solution viscosity [25].
A substance is viscoelastic if it exhibits both energy
dissipation and energy storage in its mechanical
behavior. A viscoelastic fluid will exhibit a Trouton ratio
(the ratio of shear viscosity to elongational viscosity)
of greater than 3 [25]. A higher value of Trouton ratio
indicates greater viscoelasticity. Viscoelasticity relates
to the relaxation time of the solution, which is the
characteristic time of the exponential stress decay
curve. The relaxation time can be determined by
conducting a creep test whereby a viscoelastic fluid
will respond in the manner depicted in Figure 5 [25].
After an instantaneously imposed stress, the strain of
the sample is monitored over time. The stress is then
released and the relaxation, reflected in negative strain
or recoil, is observed as a function of time. After any
instantaneous yield, the non-linear region of the curve,
C1, relates to strain growth, which is controlled by
combined viscous and elastic effects. This region is
called the retardation period. After all elastic yield is
exhausted pure viscous flow prevails. This is reflected
in the linear (constant shear rate) region depicted by
curve C2 which corresponds to a steady shear viscosity.
When the stress is released, point C3, elastic recovery
will occur. Combined viscous and elastic effects again
control the rate of recovery. This is the relaxation
period and corresponds to curve C3 to C4. In this period
the fluid recoils due to polymer chain relaxation and
76
Technology Cluster: OIL AND GAS
eventually reaches equilibrium. The strain is not
recovered completely because of the viscous flow in
region C2. Relaxation relates to a number of separate
strain increments each decaying exponentially with
time. Each increment is governed by elastic and
viscous influences working together analogous to a
spring and dashpot system. The total relaxation time
of the fluid is the summation of the characteristic
decay times of each strain increment. Analysis of the
relaxation region of the creep curve, C3 to C4, allows
the relaxation time to be calculated.
The relaxation time is significant in membrane
fabrication. As phase separation progresses, the
relaxation time will actually increase due to the onset
of solidification. As long as phase separation occurs
quickly, then the material has no chance of relaxing
and hence the orientation will be frozen into the
membrane. In fact, relaxation times become
progressively longer at lower levels of residual strain
and thus some level of orientation would still be likely
even if phase inversion is delayed. The relaxation time
indicates the level of viscoelasticity in polymer
solutions and hence the level of molecular orientation
induced under shear.
During casting a shear thinning and viscoelastic
solution, polymer molecules are maintained in an
oriented (partially) conformation by castline
deformation. After casting, polymer molecules would
relax to some preferred state. However, they recover
only a portion of their total deformation [34]. The ascast membrane is then going through forcedconvective evaporation. Dry phase separation
progresses instantaneously and limits conformational
and configurational rearrangement especially in
nascent skin region. Polymeric material has no chance
of relaxing and therefore shear-induced molecular
orientation will be frozen into nascent skin layer of
membrane [17,25,26,35]. Besides that, dope
formulation is tailored to be close to thermodynamic
instability limit and approaching cloud point curve.
Nonsolvent is added into polymer solution in order to
speed up preprecipitation of phase-separated
structures and reduce relaxation effects on molecular
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
orientation [31]. In addition, shear fields (over a certain
value) can distort phase diagram and hence alter
phase stability, demixing and precipitation kinetics of
membrane formation process [32]. The nascent
membrane is then immersed in a coagulation bath for
wet phase separation. The nascent skin layer with
sufficiently rigid structures is solidified immediately to
form a well-defined skin layer with enhanced
molecular orientation [27].
By phase inversion, Serkov and Khanchih [36]
examined the development of molecular orientation
during the precipitation process of polymer solution.
They found that preorientation was developed during
precipitation, which significantly determined the
subsequent stretchability, structure and physical
properties of the spun fibers. Molecular orientation
by ionotropic coagulation is only applied to polyelectrolyte precipitation. Due to this limitation, the
mechanism cannot be adopted to explain the
structural alignment in polymeric solutions generally.
Serkov and Khanchih suggested that as the front
moves through the solidifying medium, orientation is
developed. They also postulated that as the polymer
rich phase solidifies, spherulites could form if the
energy of interaction with the solvent is high. This
process can lead to highly ordered structure at the
molecular level. Referring to shear flow, Takuechi et al.
[37]. studied the morphological development in a
copolyester fiber with increasing shear rate. They
observed a skin-core structure in the resultant
extrudate and characterized this in terms of band
pattern, which were monitored through a polarizing
microscope. Bousmina and Muller [38] also proposed
a mechanism responsible for the development of
ordered structure in an extruded filament according
the influence of shear. They described a process which
particles become aligned in the shear direction as
shown in Figure 5. Figure 5 illustrated the mechanism
considering the non-Newtonian velocity profile across
the extrudate and described a process by which
particles become aligned in the shear direction. Serkov
and Khanchih [36] also postulated that shear-induced
orientation was frozen into the skin of a fiber during
formation, while the molecular orientation was
Figure 5: Illustration of mechanism responsible for the
development of ordered structure in an extrudate [37]
induced by elongation flow. Perepelkin [39] outlined
the principle aspects in the structural reorganization
of polymer solutions that are; 1) a change in the
orientational and three dimensional order of the
structure, 2) an increase in the degree of orientation
of the crystalline and amorphous sections of the
structure, 3) a change in the conformational
arrangement of the molecule chain and 4) a change
in the molecular structure.
In gas separation membrane, it has been
acknowledged that molecular orientation in the active
layer will affect the selectivity. In recent years, besides
the phase inversion process being the primary factor
to determine the separation performance of gas
separation membrane, more researches have been
focusing on the rheological aspect. For that reason,
rheological conditions during manufacturing will
affect membrane performance by altering molecular
orientation. The shear during casting and spinning is
one of the important rheological factors. It was shown
that molecular orientation was intensified with the
shear rate increment during casting and spinning and
that there is a favorable effect on selectivity. The shear
rate for flat sheet membrane varies with drag time. On
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
77
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
the other hand, the hollow fiber membrane varies with
dope extrusion rate.
Shilton et al. [29] found that permeability and
selectivity rose with increasing dope extrusion rate for
coated fibers due to the combined effect of enhanced
polymer selectivity and increased in surface porosity.
However, elongation experienced during spinning was
found to be detrimental to membrane performance.
Permeability increased but selectivity decreased with
increasing jet stretch ratio or elongation. This was due
to increase in porosity in the skin layer and unfavorable
polymer structure. To explain their findings, Shilton et
al. [29] suggested that shear and elongation affected
the selectivity of the solid polymer in the membrane
by altering molecular orientation. The actual rise in
selectivity of the coated membranes with increasing
dope extrusion rate could be explained by observing
the increase in the selectivity of the polymer itself. This
was related to the rheological behavior of the spinning
dope under shear.
Aptel et al. [40] spun polysulfone ultrafiltration hollow
fibers and found that the performance of the
membrane depended on the extrusion rates of the
polymer solution and on the bore liquid. According
to them, shear forces in the spinneret caused
orientation of polymer molecules which in turn
affecting pore shape. If gelation of the solution due
to contact with the coagulant liquid is faster than the
relaxation time of the polymer solution, the polymer
chain alignment is frozen into the membrane wall.
Ismail et al. [41] indicated that increasing molecular
orientation occurred in the high-shear membranes.
The gas permeation results are shown in Table 1.
Furthermore, the selectivity of these membranes was
heightened and even surpassed the recognized
intrinsic selectivity of the membrane polymer. The
results suggested that increased shear during spinning
would increase molecular orientation and, in turn,
enhanced selectivity. The selectivities were much
greater for high shear membrane [35,41].
78
Technology Cluster: OIL AND GAS
Table 1: The effect of casting shear rate on the gas permeation
properties of Polysulfone hollow fiber membrane [41]
Dope Extrution Rate
(cm3/min)
PCO 2
(cm3 (STP)s-1 cm2cmHg-1)
CO2
αCH
4
1.0
2.5
13.7
38.5
37.5
83.1
Takuechi et al. [37] studied the morphological
development in a copolyester fiber with increasing
shear rate. They observed a skin-core structure in the
resultant extrudate and characterized this in term of
band pattern. Takuechi et al. [38] also studied the
texture of the extrudate by birefringence and showed
that with shear rate induced, molecular orientation was
concentrated only in the skin layer. Both Takuechi et
al. [37] and Bousmina and Muller [38] suggested that,
understanding and quantifying, whenever possible,
the relationship between extrudate morphology and
rheology was the outmost importance in optimizing
processing conditions.
Spectroscopy has become an important tool to
investigate the morphological characteristics of a
membrane at the molecular level. Diffuse reflectance
infrared Fourier transform spectroscopy (DRIFTS) has
been used to characterize the surface structure, in
particular the molecular orientation, in wide variety of
polymer films, fibers and glass fibers. In polymer films,
DRIFTS was capable of providing good quality spectra
and structural information of polymer surfaces up to
nanolevel thickness [42]. Reflectioninfrared
spectroscopy has the advantage of being able to
analyze the surface layers of opaque films or fibers.
Recently, the technique has been developed to
determine orientation of polymer molecules in
asymmetric hollow fiber and flat sheet membranes for
gas separation. By using FTIR, plane polarized infrared
absorbed more strongly when the plane polarization
is parallel. There is a difference in absorption between
parallel polarized and perpendicularly polarized
radiation and this phenomenon is known as linear
dichroism [28].
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Khulbe et al. [43] employed electron spin resonance
and Raman spectroscopy techniques to study the
different morphologies in integral-asymmetric and
symmetric membranes of poly(phenylene oxide) (PPO)
and polyamide (PA). They found that molecular
orientation existed in the skin layer of the asymmetric
membranes and this was responsible for greater
selectivities. This orientation was subsequently related
to molecular vibration in specific bonds in the
membranes. Pronounced infrared dichroism (the
difference in absorption between parallel and
perpendicularly polarized light) indicates that the
alignment of molecules, with the absence of dichroism
showed that a sample had been randomly orienting
the molecules. Polarized IR spectra of hollow fiber
membranes had been recorded by IR reflection from
samples of the fibers wound several time round a KBr
plate of rectangular cross section. In this study the
membrane were cast at low and high shear rate. Both
membranes exhibit dichroism in the infrared, but the
effect is more pronounced in the high shear
membranes suggesting greater molecular orientation
[35].
Molecular orientation in the active layer of these
membranes was directly and quantitatively measured
using plane-polarized reflectance infrared
spectroscopy as shown in Figure 6. This technique
could reveal anisotropy on the molecular level within
a sample by pronounced infrared [35,41,44].
Shear and elongation during spinning have been
proven to affect the permeation performance of
polysulfone hollow fiber membranes [28,40] and this
was attributed to molecular orientation in the active
Figure 6: View of the polarized IR beam path through single
bounce ATR version
layer. If special skin formation conditions prevail,
increased shear can create an oriented and highly
ordered membrane active layer which can exhibit
selectivities significantly greater than the recognized
intrinsic value for the isotropic polymer. In addition,
the membranes have been tested recently using
positron annihilation lifetime spectroscopy (PALS) to
further investigate the conformation of the active layer
at the molecular level [35].
Wang et al. [45] and Chung et al. [46] suggested that
hollow fiber membranes spun with high shear rates
apparently have a thicker dense skin layer, indicating
increasing gas transport resistance because of shearinduced chain orientation and packing. This result
might also imply that a high shear rate may yield a
hollow fiber with a “denser” selective layer with a lower
permeability. At low shear rates, the permeances of
non-polar molecules such as H 2 , O 2, N 2 , and CH 4
decreases, while their relative selectivities increase
with an increase in shear rates. Once a certain shear
rate is reached, all permeances increase, while their
selectivities decrease with an increase in shear rates.
In low shear rate regions, the decrease in permeance
or increase in selectivity with increasing shear rates
arises from the better molecular orientation and chain
packing induced by shear. With increasing shear in
high shear rate regions, the increase in permeance or
decrease in selectivity is mainly attributed to relatively
porous skin structures induced by the low viscosity
nature of a power-law spinning fluid at high shear
rates, fracture, and modified thermodynamics and
kinetics of phase inversion process. This work
suggested that there might exist an optimum shear
rate to yield optimal membrane morphology for gas
separation. An increase in CO 2 permeance with
increasing shear rates are possibly due to enhanced
coupling effect between CO2 and the highly oriented
and closely packed fluoropolyimide molecular chains
induced by shear.
Ismail et al. [47] clearly illustrated the effects of shear
rate and forced convection residence time on
asymmetric polysulfone membrane structure and gas
separation performance. The rheologically induced
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
79
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Figure 7: The normalized difference spectrum among the
different shear rate (220s -1 , 265s -1 and 367s -1 ) of polysulfone
flat sheet asymmetric membrane measured directly using plane
polarized reflectance infrared spectroscopy technique [47].
molecular orientation in membranes during casting
was measured directly using plane polarized
reflectance infrared spectroscopy technique. Figure 7
shows the normalized different spectrum obtained
from the high shear membrane (367s-1). Pronounced
positive and highest dichroism (absorption parallel to
the shear direction > absorption perpendicular) were
exhibited in their samples. This indicates that the
polymer molecular chain was aligning in the shear
direction. The highly sheared asymmetric membranes
tend to exhibit greater molecular orientation in the
skin layer; thus, a high pressure-normalized flux and
selectivity were obtained. The mean pressurenormalized fluxes of O2 and CO2 were about 5.05 and
11.41 GPU, respectively. The selectivity of O2/N 2 and
CO2/CH4 were approximately 6.72 and 32.63,
respectively, at shear rate of 367s-1.
The best membrane performance obtained based on
the trade-off between pressure-normalized flux and
selectivity was observed at forced convection
residence time of 20s and at 367s-1 shear rate.
In addition, Ismail et al. [48,49] found that the
combination of an optimizing dope formulation and
casting condition produced defect-free, highly
selective and ultrathin skin membranes without any
post-treatment. The dry/wet phase inversion process
with forced convection has provided a method of
preparing defect-free and ultrathin interally-skinned
80
Technology Cluster: OIL AND GAS
asymmetric membranes. The skin layer thicknesses
produced were less than 0.5 µm. Both permeability and
selectivity of membrane were found to increase with
high-sheared casting, which were reflected in a
reduction of skin thickness as well as enhancement of
molecular orientation in skin layer. These suggest that
the rheologically induced molecular orientation plays
a significant role to heighten membrane separation
performances. As a conclusion, correlation of
rheological aspects with primary phase inversion
conditions had successfully developed defect-free and
ultrathin-skinned asymmetric membranes for gas
separation.
At present, rheologically induced molecular
orientation has received considerable attention from
a number of investigators since it has been recognized
as the factor that significantly enhances membrane
separation characteristics. Therefore, systematic study
on these factors with regard to gas separation
membranes is widely carried out. Further studies on
the fundamental aspects of rheologically induced
molecular orientation using others spectroscopy
techniques are foreseen to generate a complimentary
results which can be used to enrich the understanding
of the molecular orientation phenomenon. Modelling
of the gas transport phenomena in oriented skin
membranes and determination of the interaction
between the permeance and the membrane materials
will be considered as one of the future research
directions in membrane science and technology in
view of producing a practical and high performance
membrane for gas separation.
CONCLUSIONS
Increasing number of research in the formation of gas
separation membrane indicates that membranes
technology is fast growing area and becoming another
alternative for industrial gas separation processes. The
fabrication of asymmetric polymer consisting a defectfree and ultrathin-skinned layer has become the strong
subject of research interest because it stresses on the
importance of having high permeability and selectivity
gas separation membranes. Development of defect-
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
free and high performance asymmetric membranes
for gas separation processes with practically
acceptable separation properties often exhibit defectfree skins and negligible substructure resistance.
Though a lot of efforts are being devoted to membrane
formation for gas separation in a number of promising
areas, more research and development are still needed.
Rheologically induced molecular orientation is found
to be one of the most important factors if membrane
performance is to be heighten beyond the generally
recognized intrinsic selectivity. The aspect will be
further developed since it has now confirmed that
formation of super selective synthetic polymeric
membranes are only possible with the combine effect
of phase inversion and rheological factors during
manufacturing process. Systematic and brilliant
strategies are still needed in order to exploit the full
potential of this technology especially in producing
high performance membrane for gas separation in the
next decade.
REFERENCES
[1] A.Bungay, P. M.; Lonsdale, H. K.; De Pinho, M. N. “Synthetic
Membranes: Science, Engineering and Applications”. D. Riedel
Publishers.:Dordrecht (1986) 1-38.
[2] Chung, T. S.; Hu, X. D. “Effect of Air-gab Distance on the
Morphology and Thermal Properties of Polyethersulfone
Hollow Fiber”. J. Appl. Polym. Sci. 66, (1997) 1067-1077.
[3] Wang, I. F.; Minhas, B. S. (1991) U.S. Patent 5,067,970.
[4] Kesting, R. E.“Synthetic Polymeric Membranes”. (1971) McGrawHill: USA, 117-157.
[5] Wang, M. T.; Zhu, X. G.; Zhang, L. D. “Hole Structure and Its
Formation in Thin Films of Hydrolyzed Poly(styrene maleic
anhydride) Alternating Copolymers”. J. Appl. Polym. Sci., 75
(2000) 267-274.
[6] Prasad, R.; Shaner, R. L.; Doshi, K. J. in“Polymeric Gas Separation
Membranes”; Paul, D. R.; Yampol’ skii, Y. P. (1994) CRC Press: Boca
Raton, Chapter 11.
[7] Kesting, R.E. “Synthetic Polymeric Membranes: A Structural
Persperspective”, 2nd Ed. (1985) McGraw Hill, New York.
[8] Van de Witte, P., Dijkstra, P. J., Van den Berg, J. W. A., Feijen, J.
“Phase Separation Processes in Polymer Solution in Relation
to Membrane Formation”. J. Membrane Sci., 117 (1996) 1-31.
[9] Mulder, M.“Basic Principle of Mem brane Technology”, 2 nd Ed.,
Kluwer Academic Publishers, 1996.
[10] Wijmans, J. G., and Smolders, C. A.“Preparation of Asymmetric
Membranes by The Phase Inversion Process” in Bungay, P. M.,
Lonsdale, H. K., and de Pinho M. N., (Eds),“Synthetic Membranes:
Science, Engineering and Applications”. NATO ASI Series, Series
C: Mathematical and Physical Sciences, D. Reidel Publishing
Co., Vol. 181, (1983) 39-56.
[11] Koros, W. J. and Fleming, G. K., “Membrane-based Gas
Separation”. J. Membrane Sci., 83 (1993) 1-80.
[12] Pinnau, I. “Skin Formation of Integral Asymmetric Gas
Separation Membranes Made by Dry-wet Phase Inversion”. PhD
Thesis, University of Texas at Austin USA, 1991.
[13] Pinnau, I., and Koros, W. J. “Defect-free Ultrahigh Flux
Asymmetric Membranes”. (1990) US Patent 4,902,422.
[14] Pesek, S. C., and Koros, W. J. “Aqueous Quenched Asymmetric
Polysulfone Membranes Prepared by Dry/wet Phase
Separation”. J. Membrane Sci., 81 (1993) 71-88.
[15] Pesek, S. C. and Koros, W. J. “Aqueous Quenched Asymmetric
Polysulfone Hollow Fibers Prepared by Dry/wet Phase
Separation”. J. Membrane Sci., 88 (1994) 1-19.
[16] Pinnau, I. and Koros, W. J. “Membrane Formation for Gas
Separation Processes” in Paul, D. R. and Yampols’kii, Y. (Eds),
Polymeric Membrane for Gas Separation, CRC Press Inc., 1994,
Chapter 5.
[17] Pinnau, I. and Koros, W. J. “A Qualitative Skin Layer Formation
Mechanism for Membranes Made by Dry/wet Phase Inversion”.
J. Polym. Sci.: Part B: Polym. Physics, 31 (1993) 419-427.
[18] Broens, L., Altena, F. W., Smolders, C. A. and Koenhen, D. M.
“Asymmetric Membrane Structures As A Result of Phase
Separation Phenomena”. Desalination, 32 (1980) 33-45.
[19] Van’t Hof, J. A., Reuvers, A. J., Boom, R. M., Rolevink, H. H. M. and
Smolders, C. A. “Preparation of Asymmetric Gas Separation
Membranes With High Selectivity by A Dual-bath Coagulation
Method”. J. Membrane Sci., 70 (1992) 17-30.
[20] Li, S.-G., Koops, G.H., Mulder, M.H.V., Van den Boomgaard, T. and
Smolders, C.A. “Wet Spinning of Integrally Skinned Hollow
Fiber Membranes By A Modified Dual-bath Coagulation
Method Using A Triple Orifice Spinneret”. J. Membrane Sci., 94
(1994) 329-340.
[21] Pinnau, I., Wind, J. and Peinemann, K. V. “Ultrathin
Multicomponent Poly(ether sulfone) Membranes for Gas
Separation Made By Dry/wet Phase Inversion”. Ind. Eng. Chem.
Res., 29 (1990) 2028-2032.
[22] Pinnau, I. and Koros, W. J.“Influence of Quench Medium on The
Structures and Gas Permeation Properties of Polysulfone
Membranes Made By Wet and Dry/wet Phase Inversion”. J.
Membrane Sci., 71 (1992) 81-96.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
81
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
[23] Koenhen, D. M., Mulder, M.H.V. and Smolders, C.A. “Phase
Separation During The Formation of Asymmetric Membranes”.
J. Appl. Polym. Sci., 21(1977)199. 24. Kawakami, H., Mikawa, M.,
and Nagaoka, S. “Gas Permeability and Selectivity Through
Asymmetric Polyimide Membranes”. J. Appl. Polym. Sci., 62
(1996) 965-971.
[25] Ismail, A. F., Novel Studies of Molecular Orientation In Synthetic
Polymeric Membranes For Gas Separation, Ph.D. Dissertation;
University of Strathclyde: Scotland, 1997.
[26] Shilton, S. J., Ahmad Fauzi bin Ismail, Gough, P. J., Dunkin, I. R.
and Gallivan, S. L.,“Molecular Orientation and the Performance
of Synthetic Polymeric Membranes for Gas Separation.”
Polymer. 38. (1997) 2215-2220.
[27] Wang, D. L., Li, K. and Teo, W. K. “Relationship Between Mass
Ratio of Nonsolvent-Additive to Solvent in Membrane Casting
Solution and its Coagulation Value.” J. Membr. Sci. 98 (1995)
233-240.
[28] Shilton, S. J., Bell, G. and Ferguson, J., “The Deduction of Fine
Structural Details of Gas Separation Hollow Fiber Membranes
Using Resistance Modelling of Gas Permeation.” Polymer. 37
(1996) 485-492.
[29] Shilton, S. J., Bell, G. and Ferguson, J., “The Rheology of Fiber
Spinning and the Properties of Hollow-Fiber Membranes for
Gas Separation.” Polymer. 35. (1994) 5327-5335.
[30] Chung, T. S., Lin, W. H. and Vora, R. H. (2000).“The Effect of Shear
Rates on Gas Separation Performance of 6FDA-Durene
Polyimide Hollow Fibers.” J. Membr. Sci. 167. 55-66.
[31] Chung, T. S., Qin, J. J. and Gu, J., Effects of Shear Rate Within the
Spinneret on Morphology, Separation Performance and
Mechanical Properties of Ultrafiltration Polyethersulfone
Hollow Fiber Membranes., Chem. Eng. Sci. 55 (2000) 1077-1091.
[32] Sharpe, I. D., Ahmad Fauzi bin Ismail and Shilton, S. J., A Study
of Extrusion Shear and Forced Convection Residence Time in
the Spinning of Polysulfone Hollow Fiber Membranes for Gas
Separation.” Sep. & Purif. Tech. 17 (1999) 101-109.
[33] Ismail, A. F., Dunkin, I. R., Gallivan, S. L. and Shilton, S. J. (1999).
“Production of Super Selective Polysulfone Hollow Fiber
Membranes for Gas Separation.” Polymer. 40. 6499-6506.
[34] Hagler, G. E. “Qualitative Prediction of The Effects of Changes
In Spinning Conditions On Spun Fiber Orientation.” Polym. Eng.
Sci. 21 (1981) 121-123.
[35] Ismail, A. F., Shilton, S. J., Dunkin, I. R. and Gallivan, S. L., “Direct
Measurement of Rheologically Induced Molecular Orientation
in Gas Separation Hollow Fiber Membranes and Effects on
Selectivity.” J. Membr. Sci. 126. (1997) 133-137.
[36] Serkov A. T and Khanchic O. A. “Formation of oriented
structures in the precipitation of polymers from concentrated
solutions”. Khimicheskie Volokna 4 (1977) 12-16.
82
Technology Cluster: OIL AND GAS
[37] Takeuchi, Y., Shuto, Y. and Yamamoto, F. “Band Pattern
Development In Shear-Oriented Thermotropic Copolyester
Extrudates”. Polymer, 29 (1988) 605-612.
[38] Bousmina, M. and Muller, R., “Rheology/Morphology/Flow
Conditions Relationships For Polymethylmethacrylate/Rubber
Blend”. Rheol. Acta, 35 (1996) 369-381.
[39] Perepelkin, K.E. “The Structural Factor in the Orientation
Process of Fibres and Films of Flexible and Rigid-chain
Polymers”. Khimicheskie Volokna, 4 (1977) 7-12.
[40] Aptel, P., Abidine, N., Ivaldi, F. and Lafaille, J. P., “Polysulfone
Hollow Fibers - Effect of Spinning Conditions on Ultrafiltration
Properties”. J. Membrane Sci., 22 (1985) 199-215.
[41] Ismail, A. F., Shilton, S. J., ”Polysulfone Gas Separation Holow
Fiber Membranes With Enhanced Selectivity.” J. Membr. Sci.
139. (1998) 285-286.
[42] Jansen, J. A. J. and Haas, W. E. “Applications of Diffuse
Reflectance Optics for The Characterization of Polymer
Surfaces By Fourier Transform Infrared Spectroscopy”, Polymer
Commun, 29 (1988) 77-80.
[43] Khulbe, K. C., Gagne’, S., Tabe Mohammadi, A., Matsuura, T. and
Lamarche, A. M. “Investigation of Polymer Morphology of
Integral-asymmetric Membranes By ESR and Raman
Spectroscopy and Its Comparison With Homogeneous Films”.
J. Membrane Sci., 98 (1995) 201-208
[44] Lai, P. Y. “Development of Defect-free and Ultrathin-skinned
Asymmetric Membranes for Gas Separation”. MSc. Dissertation
Universiti Teknolofi Malaysia, 2002.
[45] Wang, R. and Chung, T. S. “Determination of Pore Sizes and
Surface Porosity and the Effect of Shear Stress Within a
Spinneret on Asymmetric Hollow Fiber Membranes.” J. Membr.
Sci. 188 (2001) 29-37.
[46] Chung, T. S., Lin, W. H. and Vora, R. H. “The Effect of Shear Rates
on Gas Separation Performance of 6FDA-Durene Polyimide
Hollow Fibers.” J. Membr. Sci. 167 (2000) 55-66.
[47] Ismail, A.F., Ng, B.C., Abdul Rahman, W.A.W., “Effects of Shear
Rate and Forced Convection Residence Time on Asymmetric
Polysulfone Membranes Structure and Gas Separation
Performance.” Separation and Purification Technology, Vol. 33
Issue 2 (2003) 255-272.
[48] Ismail, A. F., Lai, P. Y.,“Effects of Phase Inversion and Rheological
Factors on Formation of Defect-Free and Ultrathin-Skinned
Asymmetric Polysulfone Membranes for Gas Separation”,
Separation and Purification Technology, Vol. 33 Issue 2 (2003)
127-143.
[49] Ismail, A. F., Lai, P. Y., “Effect of Shear Rate on Morphology and
Gas Separation Performance of Asymmetric Polysulfone
Membranes”, ASEAN Journal of Chemical Engineering, Vol. 2,
No. 1, (2002) 67-74.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Cluster: OIL AND GAS
Technology Platform: OILFIELD GAS TREATMENT & UTILIZATION
Ahmad Fauzi Ismail is Head of Membrane
Research Unit in the Universiti Teknologi
Malaysia. He holds a BSc and MSc in
chemical engineering from Universiti
Teknologi Malaysia and PhD in membrane
technology from University of Strathclyde.
He has participated in numerous national
and international exhibition. He also has
won several awards at national and
international level. Among them are Malaysia National Young
Scientist Award (2000), Universiti Teknologi Malaysia Researcher
of The Year (2000) and Bronze Medal Award in the 30th Geneva
Exhibition: New Techniques and Products (2002). He was listed
in Marquis Who’s Who in Science & Engineering for the year 2003.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
83
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
FORMULATION ENGINEERING AND
THE PRODUCT-PROCESS INTERFACE
J P K Seville1 and P J Fryer
Centre for Formulation Engineering, School of Engineering, University of Birmingham,
Birmingham B15 2TT, UK.
1
[email protected]
INTRODUCTION
All of the products which sustain us in our daily lives –
foods, pharmaceuticals, cosmetics, detergents, fibres
for clothes and furnishings, paints and coatings,
building materials, electronic components – are
structured at a range of scales. At the smallest level
are molecular structures, and it is largely the job of the
chemist to devise the appropriate molecule to fulfil
the need. However, selecting the molecule is not
enough. At the next upper level, these molecules must
be incorporated into larger structures – with other
components - which will deliver them in the right way
to the target environment in which they have to
function. This is the area of product formulation. In
most of the examples above, it is the nano- and microstructure which controls the physical and chemical
properties which are essential to the product function.
For example, the structure of a food controls its texture
and the way in which taste chemicals are released and
perceived; the structure of a pharmaceutical may
control how a powder dissolves or how an active
chemical is released in the body. To produce these
materials efficiently requires combined understanding
of their chemistry, processing and materials science this is a very interdisciplinary subject.
Industries making use of Formulation Engineering
include some of the largest and most profitable in the
world. A good example is Unilever, which has annual
sales of £30 billion, and spends £0.5 billion per year
on R & D, primarily aimed at formulating and reformulating products in order to maintain a
competitive edge. Its various businesses, which are
summarised in Table 1, employ approximately one
third of a million people worldwide.
Table 1: Unilever main business areas
Unilever Main Business Areas
Product examples
Foods
Ice cream, margarine, spreads, tea
Detergents
Wash powders, soap bars
Personal Products
Creams and lotions, toothpaste
In each of these product areas, development is a
multidisciplinary activity involving fundamental
science, process engineering and “a good portion of
inventiveness”. Product formulation is an area which
has received little attention from university-based
researchers in the past.
HOW IS FORMULATION ENGINEERING DIFFERENT
FROM “TRADITIONAL” CHEMICAL ENGINEERING?
All branches of Engineering must reflect the changes
in the industries they serve, or else, will become
obsolete. The last three decades have seen
unprecedented and irreversible changes in the
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
84
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Table 2: Global changes in the chemical and process industries
• “Maturity” in the petrochemical industries –
petrochemical plants can now be designed reliably by
computational methods and operated with minimal
workforces
Table 3: Changes in the products of the chemical and process
industries 1970-2000 (after Oliver, 2003)
1970
homogeneous
commodity
cost competition
macro/meso scale
(large-scale continuous process)
• Progressive migration of the petrochemical industry
from North America and Europe to the Middle and Far
East
• The rapid development of speciality chemical
businesses serving the emerging high value industries
in electronics, pharmaceuticals, biotechnology, etc.
chemical and process industries.
summarised in Table 2.
These are
Formulation Engineering differs from traditional
process engineering in that it is much more strongly
consumer-led. The products of Formulation
Engineering are specified in terms of a set of consumer
desirables or effects, rather than simple chemical
formulae, for example. Since Formulation Engineering
is driven by customer demand, which can change
rapidly, processes must be designed to be flexible,
enabling rapid introduction of new products. This is
one of many differences from traditional process
engineering. Others include the fact that processes
are often operated on a much smaller scale, making
high-value specialities rather than high-volume
commodities and with competition on the basis of
quality and innovation rather than price. A single batch
of pharmaceutical active ingredient can be worth
some millions of dollars.
Of course these changes are seen differently in
different parts of the world, but it is already apparent
that South East Asia will be able to follow a twin-track
strategy of developing “traditional” petrochemicals
and “new” speciality businesses simultaneously.
These global-scale industrial developments have led
to more specific changes in the way chemical and
process engineering is practised, particularly an
emphasis on development of products at the expense
of processes. Some changes in the nature of the
products of the chemical and process industries are
summarised in Table 3.
2000
composite, structured
speciality
quality competition
micro/nano scale
(smaller scale, flexible process)
Figure 1: The place of formulation or product engineering in
relation to the chemical and process industries [after 1]
Figure 1 gives some examples of the industries which
make use of formulation or “product engineering”,
according to an early pioneer in this field, Hans
Wesselingh [1]. The examples of industries here are
arranged according to the scale of operation. From
the top downwards, process engineering becomes
more important, as does the contribution of
equipment to the overall operational costs. The
“science” component of the operation increases in the
opposite direction, while formulation or product
engineering is important over the entire range. Clearly
there is room for argument about the precise shape
of the diagram!
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
85
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Finally, it is now an imperative for product producers
to understand the life-cycle of their products and
hence their overall effect on the environment.
Consumer resistance to products which are perceived
to be environmentally damaging is a major factor in
marketing and Formulation Engineers have a major
role to play in “green product” design. A recent
example of reformulation, in response to these
pressures, is the production of effective aerosol
products without the use of environmentallyunacceptable compressed gases.
Figure 2: Chemical product engineering and some relationships
to science and process engineering [after 1]
This diagram puts product engineering at the interface
between process engineering and science. Another
way of looking at it is that it incorporates aspects of
both. Wesselingh sees the subject sitting at the centre
of a group of scientific and engineering disciplines
(Figure 2), including, most importantly, the life sciences.
Some consideration of the relationship between
products and processes is necessary here. Clearly they
are always inextricably linked, and this is even more
the case for speciality products than for commodities.
To take two contrasting examples, a commodity such
as ethanol has no “memory” of its processing steps, but
a speciality such as yoghurt certainly has [2]! The
simple act of pumping can degrade its microstructure
and destroy its value. Development of “formulated
products” therefore, requires detailed understanding
of the product-process relationships.
A further interesting difference between this work and
“traditional” chemical and process engineering is that
formulated products are often deliberately designed
to break down in use. For example, the structure of a
tablet is designed to dissolve; the structure of a paint
is designed to maintain its anti-drip properties but to
allow shear and spreading into a thin smooth film. The
study of the processes by which the product breaks
down in use is very much part of Formulation
Engineering.
86
PRODUCT TYPES
The range of products of possible interest to the
Formulation Engineer is vast. Table 4 gives some ideas
of a possible classification.
The product areas are here broken down into foods,
pharmaceuticals and specialities. There will necessarily
be much overlap between these – the rapidly growing
area of neutraceuticals is one example.
The physical principles that govern structure creation
and use are common across these diverse industrial
sectors, as shown in the Table. An important aspect of
formulation engineering is that it reinforces the
generic technology links across the industrial sectors,
bridging between product specialists. It is clear that a
common factor between the product types is
structured at a micro- and possibly nano-scale. The
skills of particle technology are important here.
The remainder of this section covers some brief
examples of product formulation in the three major
areas identified above.
FOOD
Foods are structured materials: structure may be in the
solid phase (for example crystals in chocolate and icecream; cells in meat and plant tissue) or in liquids
(network or particulate gels, based on starches or milk
proteins, in sauces; emulsions in spreads and creams).
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Table 4: Skills areas, industry sectors and products in Formulation Engineering
Speciality Products
Skills Areas
Foods
Pharmaceuticals &
bioproducts
Personal
and home
care
Coatings
and fibres
Agrochemicals
Energy
products
Structured
liquids
Emulsions/
dispersions
Emulsions/
creams/
dispersions
Creams and
foams
Paints and
film
coatings
Dispersions
Fuels &
lubricating
oils
Soft solids
Crystals
+ spreads
Solid/liquid
dispersions
Crystals,
creams and
foams
Pigments +
paints
–
–
Particulate
processes
Granules +
extrudates
Granules +
Granules +
encapsulation powders
+
supercritical
route
powders
Pigments +
paints +
fibres
Granules +
extrudates
Fuel cell
structures,
catalysts
Biodelivery
Structures to
promote
flavour and
nutrition
Structures for Products
controlled
with
drug release biological
action
Sutures and
medical
fibres
Granules
–
Many of the processes of the industry have been
developed empirically; new techniques of
manufacture must generate products which have the
same quality as the old. Some novel techniques and
products are available, such as the production of lowfat alternatives to ‘classical’ foods which have the same
attractiveness. The challenge to the processor is the
efficient production of foods which are safe and
attractive to the consumer.
• diffusion and reaction in food solids: foods are
complex structures through which diffusion is
highly complex and non-Fickian. Diffusion controls
many food properties: for example, moisture
content is important, as is the motion of flavours
and nutrients. It is necessary to understand the
relationship between the structure of the material
and the transport properties of the system: for
example in plant and animal tissue to identify
drying rates and associated quality loss [4].
• the structure and formation of food granules. There
is as yet no reliable way of ensuring that food
granules can be constructed to ensure that they (i)
are strong enough to be conveyed through the
food chain, (ii) are structured so that they hydrate
or dissolve rapidly, (iii) have the required chemical
structure and properties - for example, the surface
composition often has no relationship with the bulk
composition because drying of the granules has
brought surface active species to the edge [3].
• structuring of liquids and soft solids: structured
fluids are common in food processing. Examples
include (i) in oil/aqueous dispersions (such as
spreads and ice-creams), where the kinetics of fluid
processes, such as drop breakage and coalescence,
are a function of interfacial conditions, and are in
practice set by additives such as monoglycerides
and emulsifiers [5]; (ii) aqueous/ aqueous
dispersions (such as low fat spread formulations);
(iii) soft solids (such as gels and chocolates), where
shear affects the final product structure and
properties.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
87
Technology Platform: SYSTEM OPTIMIZATION
BIOPRODUCT PROCESSING
The effective delivery of such bioproducts as enzymes,
probiotics, gene therapy vectors and particulate
vaccines depends on maintenance of the product
characteristics during their manufacture, packaging,
storage and end-exploitation. Such characteristics
may be easily degraded as a result of a complex array
of degradative side-reactions or system antagonisms
[6]. Degradation processes may be reduced or
suppressed, and end use enhanced (site targeting,
biological efficacy, operational longevity, etc.) by giving
attention to the details of product formulation. Product
formulation (defined here as those processes which
advance the efficacy, longevity and targeting of
bioproducts) must be studied in tandem with the
processes of manufacture, selective isolation and
polishing which define the end specification of
products [7].
An area of current interest is the formulation of
probiotic assemblies. The assembly of probiotics
involves the encapsulation or entrapment of the
preserved micro-organism(s) into a particle, which
permits the ready oral delivery, targeting and
controlled release of the agent(s). For example, the
targets for Aspergillus oryzae (which enhances
cellulose digestion in humans and animals) are the
small and large intestines, so that resistance to passage
through the acidic stomach is essential. Processes of
encapsulation based upon aqueous-based
polymerisation reactions must therefore, be
augmented with enteric coating procedures which
impart resistance to acidic attack in the stomach and
a sensitivity to alkaline solubility and controlledrelease in the intestine. Studies must include not only
the impact of encapsulation on the original cell
viability, but also its reproducibility in scaled-up
manufacturing operations.
SPECIALITY PRODUCTS
In common with food and bioproducts, the
functionality of speciality products is usually derived
from their complex structure, which is produced by
88
Technology Cluster: OIL AND GAS
dispersion of one phase in another – pigment in paint,
for example – and the creation, maintenance and (in
some cases) controlled break-down of the structure is
the key to product form. For example, paints and
coatings must incorporate dispersed particulate
matter of precisely controlled particle size and shape,
and in the required concentration, to produce the
desired optical properties and resistance to
degradation in use.
The dispersed structures of speciality products may
have continuous gas, solid or liquid structures, such
as:
• multi-phase dispersions produced by mechanical
and mechano-chemical methods (for diameters of
the dispersed phase entity above and below 5µm
respectively)
• pastes
• agglomerated particles (for the formulation of
which see [7])
• foams and other porous structures
Most pharmaceuticals and many other products such
as agrochemicals require gradual or localised delivery
of active species, so that structures for controlled
release are required.
Current particle-formation processes rely heavily on
spray drying, binder agglomeration and conventional
crystallisation. During the next decades, conventional
methods will give way to a new generation of
techniques relying on more imaginative manipulation
of the phase diagram and non-equilibrium
thermodynamics. Examples include:
• supercritical fluid processing for fast reactions,
giving high purity products
• cryoprocessing,
incorporating
ultrafast
solidification/sublimation to produce very open,
active structures
• sonic, electrical and microwave techniques for
supplying energy in a controlled way to promote
reaction, phase transformation and other structuremodifying changes.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Development of effective structured materials requires
novel characterisation methods. Since most of the
systems of interest are opaque, tomographic
techniques are particulary useful. For equilibrium
structures, x-ray microtomography provides the
possibility to interrogate structures with a resolution
better than 5µm. By applying suitable thresholds on
the basis of x-ray opacity, different features of the
structure may be separated. For dynamic systems,
including flow imaging within metal-walled vessels,
Positron Emission Particle Tracking (PEPT) can be used
[8].
CONCLUDING REMARKS:
TEACHING FORMULATION ENGINEERING
The changes in Chemical Engineering described above
have stimulated new methods of teaching a productbased subject [9]; [10]. “Traditional” Chemical
Engineering was about processes, often
petrochemical processes, and the culmination of the
undergraduate course was and remains the “Design
Project”, in which students design a real process from
the ground up, using all their accumulated knowledge
to do so. Chemical Product Engineering draws on the
same basic scientific framework, but it emphasises the
importance of the product. The emphasis in teaching
therefore, changes to attempting to achieve an
understanding of how the process variables influence
the final product properties.
At the University of Birmingham, for example, students
undertake a combined product and process design
project. The pilot project concerns the manufacture
of detergent granules and has been developed in
conjunction with Unilever Research [11]. In the first
week, the team members are given some Unilever
product granules and are sent to purchase some rival
products! They are then encouraged to think about
which features of these products are most desirable
and undesirable and how these might be measured
in a scientific way. This is achieved by analysing the
interaction of the product with the customer and its
environment during its useful life and breaking this
into stages which can be analysed from a Chemical
Engineering perspective using fundamental principles
(Figure 3).
Figure 3: Product properties at various stages of the detergent powder life cycle
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
89
Technology Platform: SYSTEM OPTIMIZATION
Starting from a basic formulation, and using the
information they have gleaned from patents and
published data, the team members are then set to
produce a detergent granule for themselves,
modifying the formulation and process conditions to
produce the desirable features they have identified
earlier. Finally, they compare the desirable features of
their own product with those of Unilever’s own and
the competitor products.
Following the practical project, the students design a
process for producing 100,000 tonnes per annum of
formulated granulated detergent powders. The brief
varies from year to year and various routes may be
chosen. The students must consider the problems
likely to be encountered with scaling-up the process
they are using, such as heat transfer and rates of
reaction. Through the interaction with Unilever’s
research laboratory, they have access to or are able to
measure themselves the key data necessary for the
design of their process.
The design project is designed to allow the student to
apply the engineering skills and principles learnt
during the course. The analysis of a product function
requires fundamental chemical engineering to be
applied in a very different but equally valid and
rigorous way, using some imagination and a full
understanding of the ideas behind the theories.
So far, the students who have undertaken this
integrated project seem to have relished the
experience. It introduces them for the first time to the
Chemical Engineer’s role in product development, a
role which is becoming increasingly important in most
high-tech industries. Further projects along these lines
are planned, in conjunction with companies in the food
industry, pharmaceutical industry and speciality
chemical processing.
REFERENCES
[1] Wesselingh JAM (1984), in “Training Chemical Engineers for
Plant and Product Design of Speciality Chemicals”, European
Federation of Chemical Engineering, IChemE, Rugby UK.
90
Technology Cluster: OIL AND GAS
[2] Fangary YS, Barigou M and Seville JPK (1999), “Simulation of
yoghurt flow and prediction of its end-of-process properties
using rheological measurements”, Trans. IChemE, part C, (Food
and Bioproducts Processing), 77, 33-39.
[3] Lillford PJ and Fryer PJ (1998),“Food particles and the problems
of hydration”, Trans. IChemE A, 76, 797-802.
[4] Stapley AGF, Gladden LF and Fryer PJ (1998) “Diffusion and
reaction in whole wheat grains during boiling”, AIChEJ, 44,
1777-1789.
[5] Pacek AW, Man CC, and Nienow AW (1997).“Coalescence rates
in water-in-oil and oil-in-water dispersions”, 9 th European Conf.
on Mixing, 263-170.
[6] Braas G, Searle PF, Slater NKH and Lyddiatt A (1996),“Strategies
for the isolation and purification of retroviral vectors for gene
therapy”. Bioseparation, 6, 211-228.
[7] Lyddiatt A and O’Sullivan DA (1998),“Biochemical recovery and
purification of gene therapy vectors”, Current Opinion in
Biotechnology, 9, 177-185.
[8] Knight PC (2001), “Structuring agglomerated products for
improved performance”, Powder Technology 119, 14-25.
[9] Stein M, Martin TW, Seville JPK, McNeil PA and Parker DJ (1997),
“Positron Emission Particle Tracking: particle velocities in gas
fluidised beds, mixers and other applications”, in “Non-Invasive
Monitoring of Multiphase Flows”, J Chaouki, F Larachi and MP
Dudukovic (eds.), Elsevier Science 309-333.
[10] Wesselingh JAM (2001),“Structuring of products and education
of product engineers”, Powder Technology 119, 2-8.
[11] Cussler EL and Moggridge GD (2001), “Chemical product
design”, Cambridge University Press.
[12] Seville JPK, McCormack A, Yuregir K and Instone T (2000),
“Teaching chemical product engineering”, The Chemical
Engineer, 21 September 2000.
Oliver R (2003), personal communication Villadsen J (1997),
“Putting Structure into Chemical Engineering”, Chem. Eng. Sci.
52, 2857-2864 Lillford P and Fryer PJ (1988), “Food particles
and the problems of hydration”, Trans. IchemE, part C, (Food
and Bioproducts Processing), 76, 797-802.
Jonathan Seville (MEng, MA (Cambridge),
PhD (Surrey), CEng, FIChemE) is Head of
Chemical Engineering at the University of
Birmingham and Director of Research for
the new School of Engineering. He is also
the first Head of the new Centre for
Formulation Engineering, which was
established in August 2001, with a £3 million
grant from the Universities’ Joint
Infrastructure Fund. Before moving to Birmingham, he was the
staff of the University of Surrey (1981-94) and at Courtaulds Ltd
(1979-81); he was a Visiting Professor at the University of British
Columbia (1989-90) and the Technical University of Denmark
(Copenhagen) (1997-2000). Professor Seville has over 100
publications, including 3 books, the most recent of which is J P K
Seville, U Tüzün and R Clift, “Processing of Particulate Solids”,
Chapman and Hall, London (1997).
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
ADVANCEMENTS IN TENSION LEG PLATFORM TECHNOLOGY
John W. Chianis
Deepwater Technology and Engineering,
ABB Floating Production Systems, Houston, Texas.
[email protected]
ABSTRACT
Nearly twenty years have passed since the first Tension Leg Platform (TLP) was installed in the North Sea in
approximately 150 m of water. By contrast, the eighteenth TLP will be installed in the Gulf of Mexico mid 2004 in
nearly 1500 m of water. This ten-fold increase in water depth has proven that the TLP concept has evolved at an
incredible rate.
The TLP has proven itself to be the system of choice for deepwater oil and gas developments. This is demonstrated
by its current worldwide application in the North Sea, Gulf of Mexico, West Africa and SE Asia. The later is proven
to be an ideal as well as exciting opportunity for the application of TLP technology. Ongoing developments in
SE Asia are well suited to this Operator-friendly system. Also, due to the compact nature of its hull form, fabrication
and integration of the system can be performed locally. The TLP can become a conduit that ties Malaysia to the
global deepwater market. In addition, there are some local applications of the technology that could further
develop TLPs in the quest to support Operators as they go deeper.
As an introduction, this paper first addresses the basic principals of the TLP such as its superior motion
characteristics and associated operational advantages, mooring system, compact hull form, and more. The paper
will then present the development of the new Extended Tension Leg Platform (ETLP) technology and its
application on the two most recent TLP projects including extension of water depth, high top tensioned riser
counts, quayside integration, etc. Lastly and most importantly, other technology developments pertinent to
specific near-term applications in SE Asia, are presented. These include local fabrication and integration, tender
assisted drilling, hull structural efficiency, and others.
Keywords: Tension Leg Platform, TLP, Extended Tension Leg Platform, ETLP, Deepwater, SE Asia, Malaysia
INTRODUCTION OF TLP CONCEPT
System Functions
There are large numbers of available platform options
for deepwater development. These include, but are
not limited to the Tension Leg Platform,
Semisubmersible, Spar, Single Column Floater (SCF),
etc. To the eye, these hull forms appear very differently
as shown in Figure 1. However, they can all be grouped
into two families – TLPs and Floaters.
A TLP system can support any necessary functions in
deepwater. However, due to both its superior motion
characteristics and its associated operational
advantages as compared to Floaters, the TLP is the
preferred dry completions solution. In a dry
completion role, the TLP can function as a full
production, drilling and quarters facility (PDQ). Or it
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
91
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Semisubmersible
Single Column
Floater (SCF)
Tension Leg
Platform (TLP)
Spar
Figure 1: General Hull Forms for Deepwater
can be utilized as a surface wellhead platform (SWHP),
i.e. drilling and quarters are supported by the TLP with
remaining functions provided by an adjacent Floating
Production, Storage and Offloading facility (FPSO). In
its most streamlined wellhead form, drilling systems
and quarters are efficiently provided from a tender
assist vessel in the more benign metocean
environments.
The TLP is also a viable wet tree solution utilizing
subsea tiebacks. The favorable motions of a TLP result
in fewer fatigue-related concerns for Steel Catenary
Risers (SCRs).
Figure 2: 6 DOF Motions for a TLP
Defining Characteristics
Contrary to Floaters (such as the Semisubmersible, SCF
and Spar), the mooring system of a TLP is vertically
oriented and consists of tubular steel members called
tendons. The tendons are highly tensioned using
excess buoyancy of the platform hull. The highly
tensioned tendon system limits horizontal offsets to a
very small percent of water depth. The high tendon
stiffness also reduces the system’s vertical natural
periods to a level well below that of the dominant wave
energy. As a result, dynamic amplification of vertical
motion is nearly non-existent and the platform has
limited heave, roll and pitch motions. Essentially, the
92
vertical motions correspond to the stretch in the
tubular steel tendons. For even the largest Gulf of
Mexico hurricane waves of 25 m, the corresponding
vertical motions of the platform are limited to a few
centimeters.
The TLP’s six-degree of freedom (DOF) rigid body
motions are defined in Figure 2. For those degrees of
freedom in the vertical plane (heave, roll and pitch),
the motions are effectively eliminated by the axial
stiffness of the tendon system. As such, the sensitivity
of critical operational aspects including drilling and the
support of Top Tensioned Risers (TTRs) to vertical
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
History
Seventeen TLPs have been installed to date worldwide.
The first, Conoco’s Hutton platform in the UK North Sea,
was installed in 1984 in approximately 150 m of water.
Today, ABB has designed a TLP for nearly 1,500 m of
water for Conoco’s Magnolia project in the Gulf of
Mexico. This record-breaking platform will be installed
in the third quarter of 2004.
Figure 3: Definition of TLP Offset and Setdown
The accelerating rate of industry acceptance of the TLP
is very much linked to the concept’s rapid growth in
technology.
For example, specific design
enhancements, such as the development of the
Extended Tension Leg Platform (ETLP), have allowed
the concept to carry heavier payloads into deeper
water. This is effectively demonstrated in Figure 4.
DESCRIPTION OF A CONVENTIONAL TLP
Figure 4: TLP History – Year Installed vs. Water Depth
motion is eliminated. The motions in the horizontal
plane (surge, sway and yaw) are controlled by tendon
tension.
Offset with setdown is another very interesting aspect
of TLP behavior. This is illustrated in Figure 3. The TLP
experiences offset when acted upon by a horizontal
force. These forces arise from wind, waves, current and
other sources such as SCRs and Fluid Transfer Lines
(FTLs). Due to the high axial stiffness of the tendon
system, offset will be accompanied by an associated
increase in TLP draft. This is called setdown. If a TLP is
offset to a given position and released, the horizontal
tendon tension component due to offset and the
increased vertical tendon tension component due to
setdown act together to restore the platform to its
vertical position.
From an operational perspective, a beneficial outcome
of this offset-setdown behavior is a greatly reduced
TTR stroke as compared to floater concepts.
The design of a TLP hull form, and particularly its
distinctive mooring system, is very much linked to
payload and water depth. Despite its weight-sensitive
nature, the TLP has proven to be the Industry’s favored
deepwater hull form worldwide for dry tree
applications because of its operationally-superior
motion characteristics. This has been demonstrated
on a global scale with TLPs now located in the North
Sea, Gulf of Mexico, offshore West Africa and SE Asia.
A brief overview of TLP configuration, motion
characteristics and associated operational advantages
are presented below.
Typical Configuration
A typical TLP hull configuration consists of four vertical
columns that can be square or cylindrical in cross
section. Rectangular pontoons connect the columns
below the water surface. On top of the columns, and
integral to the hull, is the structural deck that supports
the topsides production facilities, drilling system,
production risers, living quarters, etc. If a TLP has the
capabilities for vertical access to wells, i.e. drilling or
well work-over, then it is deemed to be a “dry tree” unit.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
93
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Figure 7a: Surge, Sway and Heave RAOs
Figure 5: Major Components of a Typical PDQ TLP
Figure 7b: Roll, Pitch and Yaw RAOs
Figure 6: Typical Natural Periods of TLP Motion
Figure 5 identifies the major system components of a
typical PDQ TLP.
Motion Performance
The key element that enables the TLP to respond
favorably to waves is the separation of the six DOF
natural periods from the dominant wave energy. This
is shown in Figure 6 where the TLP’s six DOF natural
periods are superimposed over a typical 100-year
storm wave spectrum (hurricane event) in the Gulf of
Mexico. Notice that the heave, roll and pitch motions
94
(vertical plane) exhibit natural periods at 4 sec or less
and the surge, sway and yaw motions (horizontal
plane) have natural periods of motion greater than 100
sec. Since both of these natural period regions fall
outside the range of dominant wave energy, the
dynamic response of the TLP system to waves is small.
A typical set of Response Amplitude Operators (RAOs)
for the six DOF motions of a TLP are shown in Figures
7a and 7b. These RAOs represent a typical TLP located
in 1,800 m water depth offshore Malaysia. An RAO can
be defined as the linear relationship between the
motion of a TLP (in a single degree of freedom) and a
unit amplitude regular wave. For the translational
motions (surge, sway and heave), the units of the RAO
are ft/ft, or feet of motion per unit wave amplitude.
For the rotational motions (roll, pitch and yaw), the
units of the RAO are deg/ft, or degrees of motion per
unit wave amplitude.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Some of the more important tangible benefits are
presented and discussed below:
• Support of dry trees – The superior motion
characteristics of the TLP (due primarily to the
tendon mooring system) enables safe and efficient
support of TTRs. The platform is able to remain
operational in more severe metocean conditions.
Because of the offset and setdown behavior of the
TLP, riser stroke is greatly reduced compared to
other competing Floater concepts. SCRs (import,
export, or both) are also efficiently supported by
the TLP.
Figure 7c: Tendon Tension RAOs
In addition, typical tendon tension RAOs are shown in
Figure 7c. The units of the tendon tension RAO are
kips/ft, or kips of tension per unit wave amplitude. A
typical offshore Malaysia 100-year event wave
spectrum has been superimposed on this plot for
comparison. It can be seen that this TLP system has
been optimized such that the tendon tension RAO in
the range of maximum wave energy is minimized.
This set of RAOs represents the fundamental motion
characteristics of the platform. During a typical design
process, many variables such as hull dimensions, draft,
tendon cross-sectional area, tendon tension, etc. will
be varied in order to obtain favorable motions. This
highly important work, often referred to as Global
Performance analysis, is performed over a wide range
of metocean conditions, operational phases, intact and
damaged conditions, and both inplace and pre-service
conditions. The pre-service conditions capture all of
the construction and delivery phases prior to being
installed. These often include deck/hull quayside
integration, dry transport, wet tow, installation, etc. The
final configuration and sizing of the TLP (and therefore
its motion characteristics) for a particular application
can only be established once all of the above
conditions have been considered.
Operational Advantages
Utilization of TLP technology for a deepwater
development results in operational advantages not
necessarily found with other competing hull forms.
• Wellbay configuration – A conventional and proven
wellbay is inherent to the TLP. TTRs are supported
within a grid-like wellbay centered between the TLP
columns. The risers are untouched until their
connection at the seabed. In comparison with other
TLP concepts, riser keel joints and a centerwell (with
potential hydrocarbon safety hazards) are not
necessary. Also, the contiguous wellbay
configuration of the conventional TLP results in
reduced trim ballast requirements compared to
other TLP options with split wellbays.
• Drilling and workover – The TLP can support drilling
either in a self-contained mode or via tender assist.
The drilling tender vessel can be moored integral
to the TLP. Significant loads can also be eliminated
from the TLP by physically moving drilling systems
and platform accommodations to the tender vessel.
The size of the TLP can then be reduced. The global
behavior of the integrated TLP/tender vessel system
is well understood and proven.
• Functional versatility – TLPs to date have performed
many different functions and combinations. These
have included:
- Self-contained production, quarters and drilling
platform
- Wellhead platform with adjacent FPSO
- Self-contained or off-TLP production with
drilling tender assist
• Small footprint – Because the tendon mooring
system is vertical, the footprint on the seafloor is
small. Contrary to other Floater hull forms utilizing
catenary (spread moored) mooring systems, the TLP
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
95
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
system will not interfere with adjacent facilities such
as FPSOs, production barges, drilling tender vessels,
pipelines, etc.
• Installation of completed system – The compact
nature of the TLP hull (i.e. small height, width and
draft) permits quayside lift of the deck and the
associated integration activities. Near full
commissioning of the system is also performed at
quayside. Installation of a fully integrated and nearfully commissioned facility results in a shorter
schedule, lower cost and reduced risk compared to
equivalent operations performed offshore.
Figure 8: Comparison of Conventional TLP and ETLP Hull
Configurations
TECHNOLOGY BREAKTHROUGH –
THE ETLP™ PLATFORM
Extensive experience and lessons learned from ABB’s
major involvement in TLP projects over the last 25
years have been applied to the development of the
ETLP™ platform. The ETLP™ system has proven to be
a significant technology breakthrough. An overview
of the ETLP™ system development history and its
advantages are presented below.
Development History
Figure 9: Rendering of 3-Column ETLP
For several years, ABB has aggressively pursued a
technology development effort to revolutionize the
conventional TLP. A multi-million dollar investment has
been made by ABB in the development of the ETLP™
concept for a variety of topsides payloads, riser counts,
water depths and regions of the world. Extensive
engineering, design and physical model testing has
been performed.
Previously, tendons were connected to the lowermost
outboard portion of the hull on the columns. For the
ETLP™ concept, the columns have been moved
inboard allowing a more favorable support condition
for the deck and its associated riser and drilling related
loads. Pontoon extensions on the outboard edge of
the column are used as tendon connection points.
Figure 8 shows a plan view of the conventional TLP
and ETLP™ platform to illustrate the differences
between the two hull forms.
ABB has performed a large number of paid studies and
design competitions for Clients whose deepwater
development options are favorable for an ETLP™
platform. The ETLP™ concept has shown to be
technically sound as it is comprised of conventional
and well-proven systems.
The primary difference between a conventional TLP
and the new ETLP™ platform is the hull form.
96
For developments where the production riser count
is low, a three-column ETLP™ platform can be utilized.
This smaller version of the ETLP™ platform has
undergone the same extensive development work as
the four-column ETLP™ concept. The 3-column ETLP™
platform is shown in Figure 9.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Advantages
RECENT APPLICATIONS OF THE ETLP” CONCEPT
The most salient features of the ETLP™ system are
related to weight reduction and the efficiencies gained
in project execution through quayside integration.
Two of the Industry’s most recent deepwater projects
have been technically challenging for different
reasons. ExxonMobil’s Kizomba “A” field offshore
Angola required a SWHP to support 36 TTRs, and
ConocoPhillip’s Magnolia platform will be installed in
nearly 1,500 m of water in the Gulf of Mexico. In both
cases, the ETLP™ platform was selected. These two
projects for which this breakthrough technology has
been applied are presented below.
Compared to a conventional TLP, the hull and deck
steel weight savings of an ETLP™ platform is
approximately 40%. This has been demonstrated via
controlled shadow studies on recent ETLP™ system
designs.
A measure of the effectiveness of the ETLP™ platform
to carry topsides payload can be described by a ratio
called structural weight efficiency. This ratio quantifies
the weight of payload that can be supported by a
given weight of hull and deck platform. The structural
weight efficiency for an ETLP™ platform is high
compared to a conventional TLP. For future SE Asia
applications, it is interesting to note that the structural
weight efficiency increases dramatically as the design
metocean conditions become more benign. This is
discussed in more detail in a later section.
The other significant advantage of the ETLP™ platform
is its ability to be installed as a completely integrated
and commissioned unit. As such, the installation
operation is less complex as compared to other
floating hull forms and is therefore, less likely to be
adversely affected by weather.
Kizomba “A” Project
The Kizomba “A” deepwater development located in
1,178 m water depth offshore Angola (block 15) has
an estimated 1 billion barrel recoverable reserves. To
further enhance reservoir recovery, ExxonMobil
selected a dry tree development plan that called for a
SWHP tied to a FPSO with product export provided by
an offtake buoy. The overall development layout for
Kizomba “A” is illustrated in Figure 10.
The ETLP™ system was selected for the Surface
Wellhead Platform. Functions of the SWHP include
drilling, well intervention, and support of the dry tree
manifold. All processing and storage is provided by
the adjacent FPSO. Utilities for the SWHP’s drilling
operations are also provided by the FPSO.
The advantages of the ETLP™ system can be
summarized as follows:
• Comprised of safe, conventional and well-proven
systems,
• Greater than 40% hull and deck steel weight savings
when compared to the conventional TLP,
• Integration and commissioning of topsides and hull
is done quayside,
• Minimal risk exposure for installation in remote
regions, and
• A large crane vessel is not required for installation.
Figure 10: Field Layout for Kizomba “A”
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
97
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Development details for Kizomba “A” are as follows.
Kizomba “A” SWHP
Draft
m
34
No. of TTRs
---
36
No. of Tendons
---
8, 4x2 pattern
Total Payload
mt
23,000
Displacement
mt
53,033
Kizomba “A” FPSO
Capacity
BPD
250,000
Storage
MMB
2.2
LxBxH
m
285x63x32
Selection of the ETLP™ system for Kizomba “A” has
demonstrated that:
• Benign metocean conditions allow TLPs to carry
larger payloads in deeper water
• TLP can be designed to have 100% operational
uptime year round
Figure 11: Magnolia ETLP Illustration
Development details for Magnolia are as follows.
Magnolia Facility
Draft
m
25
No. of TTRs
--8
No. of Tendons
--8, 4x2 pattern
Total Payload
mt
13,814
Displacement
mt
34,280
• ETLP” platform can be installed fully integrated and
commissioned
• Using a TLP, separation distance between the SWHP
and FPSO is greatly reduced as compared to other
Floater hull forms
Magnolia Project
The Magnolia deepwater development is located in
the Gulf of Mexico in nearly 1,500 m water depth.
ConocoPhillips selected the ETLP™ platform for this
dry tree development.
Figure 12: Magnolia ETLP Deck Lift at Quayside
The ETLP™ platform supports full production,
workover rig, 4 subsea tiebacks and 2 umbilicals. Gas
and oil are transported to shore via export SCRs and
pipelines. The Magnolia ETLP™ platform is illustrated
in Figure 11.
Selection of the ETLP™ system for Magnolia has
demonstrated that a full TLP is viable in nearly 1,500
m of water in the Gulf of Mexico and the platform can
be fully integrated and commissioned prior to
installation.
A recent photograph showing the Magnolia ETLP™
platform deck lift and integration at quayside is
provided in Figure 12.
98
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
NEAR-TERM TECHNOLOGY APPLICATIONS
FOR SE ASIA
The ETLP™ platform is ideally suited for ongoing and
future applications in SE Asia. This new technology is
Operator-friendly and is comprised of conventional
and well-proven systems. Also, because of the
compact nature of the hull form, fabrication and
integration of the system can be performed locally.
Local Fabrication and Integration
Details regarding fabrication of the ETLP™ hull and
deck, and deck-hull integration are outlined below.
Comments relating to local construction of the ETLP™
platform are found at the end of this section.
Figure 13: Kizomba “A” ETLP Hull in Drydock
ETLP™ Hull
Fabrication of the ETLP™ hull can be tailored to suit a
number of construction methods depending upon the
capabilities and strengths of a given Fabricator due to
its compact form. For example, ABB has demonstrated
that land-based fabrication at quayside can be equally
as efficient as drydock fabrication.
For either of these methods, hull fabrication is a very
systematic and disciplined process, and one that is fully
integrated with engineering. The fabrication sequence
begins with sub-assemblies, which form blocks, which
in turn form super-blocks. Super-blocks are the largest
of the hull components and are lifted into the drydock
or rolled into the land-based erection area. The superblocks are pre-outfitted to the maximum extent with
equipment, piping, pipe supports, penetrations, cable
trays, ladders, stairs, etc. Due to parallel designfabrication activities, pre-outfitting is often limited by
engineering supply of drawings. Painting is the last
activity prior to final lifting of the super-block into
place within the structure.
A total of 24 super-blocks were used for the Kizomba
“A” ETLP™ hull. Because of this efficient procedure,
complete erection of the hull in the drydock was
performed in only six weeks. A photograph of the
Figure 14: Kizomba “A” ETLP Hull Float-On
Kizomba “A” hull in drydock (Daewoo Shipbuilding and
Marine Engineering - Korea) is shown in Figure 13.
Upon leaving the drydock, the ETLP™ hull is moved to
a nearby quayside location where final mechanical
completion and pre-commissioning activities are
performed. Systems entirely contained in the hull are
fully commissioned. The hull is then readied for
transport. Figure 14 shows the float-on operation of
the Kizomba “A” ETLP™ hull.
If the hull is fabricated at a quayside land-based
location, all final mechanical completion and precommissioning activities are performed prior to direct
loadout onto the transportation vessel. A photograph
showing land-based fabrication (Belleli - Italy) of the
Auger TLP hull is shown in Figure 15.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
99
Technology Platform: SYSTEM OPTIMIZATION
Figure 15: Land Based Fabrication of Auger TLP Hull
Technology Cluster: OIL AND GAS
Figure 17: Kizomba “A” ETLP Deck Module
Because of its single piece construction, the deck will
generally be more structurally efficient (less redundant
steel) than an equivalent modular-type deck. Due to
the float-over operation, a sheltered deepwater area
will be required. The hull will experience a slight
growth in weight due to the increased draft
(hydrostatic design head) associated with this
operation.
Figure 16: Dry Transport of Magnolia ETLP Hull
Once inplaced on the transport vessel and secured, the
ETLP™ hull is then transported to the deck-hull
integration site. A photograph showing the arrival of
the Magnolia ETLP™ hull at the Gulf of Mexico
integration site is presented in Figure 16.
ETLP™ Deck
Depending upon its size and deck-hull integration
method, the deck can be fabricated in a single
integrated piece or in modules.
A single integrated deck will in most cases exceed a
Fabricator’s lifting capacity. As such, this option will
involve a float-over deck-hull integration operation.
100
In most cases, TLP decks have been fabricated in
modules. As a general rule, fewer modules will benefit
subsequent module-to-module integration activities.
The Kizomba “A” ETLP™ deck was fabricated in two
major modules, where the Mars TLP and Ram/Powell
TLP decks were fabricated in five modules. Lifting
capacity was the major factor in determining the
number of deck modules for these cases.
The deck’s major truss rows are fabricated first. Deck
pancake sections are then placed between the rolledup truss rows. Deck pancake sections span between
the truss rows and contain deck beams, plating and
all self-contained equipment.
For the Kizomba “A” ETLP™ platform, each module was
fabricated under cover in a shop. The weight of each
module was kept below 6,000 mt for lifting purposes.
Roll-out of the North deck module is shown in Figure
17.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Figure 18: Kizomba “A” ETLP Deck Lift at Quayside
Figure 20: Dry Transport Kizomba “A” Platform to Site
fully commissioned at quayside. This is shown in Figure
19. The fully integrated and commissioned ETLP™
system can be wet-towed to site, or loaded onto a
heavy-lift submersible vessel for dry transport as
shown in Figure 20.
Local Construction
A significant level of local content is achievable given
the compact form of the ETLP™ platform and the wellproven nature of its associated key systems.
Figure 19: Commissioning of Kizomba “A” Platform at Quayside
ETLP™ Platform Deck-Hull Integration
Upon arrival at the deck-hull integration site, the hull
is brought to quayside and readied for the deck lifting
program. For Kizomba “A”, the lifting sequence of the
deck and other key packages are as follows:
• North module,
• South module,
• Living quarters,
• Drilling module, and
• Drill rig including skid base, drill floor and derrick.
Figure 18 shows the lifting operation for the second,
or South, module.
After all of the above lifts and associated integration
activities are performed, the complete topsides are
Construction of the ETLP™ deck and hull is possible in
SE Asia and would be encouraged in order to shorten
the project schedule via the elimination of lengthy
transports. As discussed earlier in this section, the
uncomplicated design of the ETLP™ hull allows
flexibility in construction methods. Since drydock and
land-based construction methods are possible,
fabrication yard options can be expanded. Platform
appurtenances such as boat landings, riser tensioner
cassette frames and wellhead access platforms can
also be fabricated locally.
Tendon piles are excellent candidates for local
fabrication, and with proper training, the tendon
strings themselves can also be fabricated locally.
Methods have also been developed by ABB to permit
safe pre-installation of the tendon system without the
use of a costly heavy lift vessel.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
101
Technology Platform: SYSTEM OPTIMIZATION
The platform-based drilling system is typically
fabricated in several modules and then transported to
the integration site for assembly directly on the ETLP™
platform. Drilling system modules can be fabricated
locally. Prior to final sail-away of the ETLP™ system,
the completed drilling system is fully commissioned.
The top-tensioned drilling and production risers would
likely be fabricated elsewhere and transported to site
for direct installation on the in-place ETLP™ platform.
Pre-installation of all the riser tensioner cassette frames
while at the deck-hull integration site permits simple
installation of the risers after the ETLP™ platform has
been installed.
Technology Cluster: OIL AND GAS
• No requirement for an internal centerwell
(moonpool) to trap hydrocarbons
• Riser top tensioning systems are less complex due
to limited motions and offsets, and are installation
friendly. Buoyancy cans on risers are unnecessary.
• Reduced relative motions between tree and BOP
results in riser strokes in the range of 1 _ m as
compared to 9 m for competing Floater concepts
• Riser tensioning equipment can be eliminated in
benign metocean regions like SE Asia
• A TAD approach significantly reduces weight
management concerns
• Utilization of guidelines allows positive control to
access subsea wellheads
• Close proximity (coupled) mooring with TAD vessel
Drilling Optimization
Elimination of Heavy Lift Vessels
A TLP–based system permits a higher level of drilling
optimization than other competing Floater concepts.
For example, Tender Assisted Drilling (TAD) reduces
payload requirements on the TLP by allowing storage
of drilling fluids, tubulars and other equipment on the
tender vessel. Accommodations can also be founded
on the tender vessel. The relocation of these loads to
the tender vessel results in a smaller TLP with an
associated reduction in capital costs.
The superior motion characteristics of the TLP permit
simple coupling with the tender vessel, thereby,
enhancing personnel and equipment transfer safety.
Drilling operations are also maximized due to the
reduction in weather related rig downtime.
The favorable motions of the TLP also eliminate the
need for motion compensation on the drill string, an
active mooring system and complex telescoping joints.
In benign metocean regions like SE Asia, direct
connection (hang-off ) of production risers is possible,
thus, eliminating riser tensioning equipment.
A brief summary of the drilling related advantages
from a conventional TLP or ETLP™ platform is given
below:
• Simplified wellbay because it is open and
unobstructed
102
Heavy lift vessels are typically utilized for specific TLP
installation related tasks. These have included: offshore
deck lift and integration (with associated weather and
equipment related risk), installation of foundations and
tendons, and platform installation.
Unfortunately, heavy lift vessels attract very high day
rates. Also, since existing heavy lift vessels are located
primarily in the N Sea and Gulf of Mexico, mob and
de-mob costs are typically high. These construction
vessels also must be contracted early often resulting
in premature definition and agreement of complex
installation campaigns and windows.
Alternatively, it is possible for heavy lift vessels to be
eliminated altogether in SE Asia. This is summarized
in the sections below.
Deck Lift and Integration
The conventional TLP and ETLP™ platform are
integrated at quayside. The platform can be near fully
commissioned prior to wet tow to site. Lengthy risk
prone offshore campaigns inherent to other
competing TLP concepts and some Floaters are not
necessary.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Foundation and Tendon Installation
Structural Weight Efficiency
Workable and proven alternatives exist for foundation
and tendon installation. For example, suction piles
(providing soil conditions are appropriate) can be used
instead of driven piles. Suction piles can be reliably
deployed in SE Asia from a locally obtained subsea
construction vessel. Entire tendon strings can be
prefabricated at a local waterside location and wet
towed by tugs to site. Upon arrival, each tendon is then
upended and stabbed into its bottom connector
located on the suction pile. The tendon will be held
vertically by a Temporary Buoyancy Module (TBM). The
TBMs are removed once the TLP is connected to the
tendons.
One important consideration with any floating
deepwater development option is its structural weight
efficiency. A more structurally efficient TLP system will
result in an increased topsides payload capacity and/
or the ability to be applied to deeper water depths.
The more structurally efficient TLP hull and deck will
also typically require fewer tendons due to the
reduction in system dynamics. The ETLP™ platform
has a much broader range of application, because of
its approximately 40% steel weight savings over a
conventional TLP.
Platform Installation
Structural weight efficiency can be quantified by
comparing the total topsides payload of a TLP to its
total weight of hull and deck steel. This is illustrated
by the following equation:
It has been customary for heavy lift vessels to be onsite
during platform installation, i.e. ballasting the TLP
down for connection to the preinstalled tendons. In
these cases, the dynamically positioned heavy lift
vessel performs no real activities other than to provide
additional accommodations during the installation
phase. Alternatively, readily available tugs can be used
to hold the TLP in position over the preinstalled
tendons. The time taken for ballasting process in order
to engage the tendons and a reach storm-safe tension
level is approximately 12 hours.
Local Solutions
An execution plan utilizing quayside deck lift,
integration and commissioning of the completed TLP
facility is well proven and is the preferred solution over
the more expensive and higher risk offshore
alternative inherent to competing TLP concepts.
Eliminating the dependence on a heavy lift vessel will
encourage local solutions and thereby increase local
content. The resulting impact on schedule, and
flexibility of the installation window, will be positive.
It is estimated that the installation related costs of a
TLP in SE Asia can be reduced by half simply by
eliminating of the heavy lift vessel.
Metric for Structural Weight Efficiency
Total Topsides Payload
Structural Weight Efficiency = ________________________
Total Hull and Deck Weight
The following important definitions are provided:
Total Topsides Payload – Weight of all deck
equipment and facilities including quarters,
drilling systems, etc. Also includes TTR loads, SCR
loads and secondary deck steel. Topsides
equipment or facilities carried in the hull, and hull
ballast earmarked for future expansion are also
included.
Total Hull and Deck Weight – Structural steel
weight of hull, hull marine systems, hull
appurtenances and outfitting, and trim ballast.
Also includes deck primary structural steel weight.
Based on the above definitions, the structural weight
efficiency of an ETLP™ platform can be expressed as a
ratio between Total Topsides Payload and Total Hull
and Deck Weight. This ratio quantifies the amount of
Total Topsides Payload that can be carried by a unit
Total Hull and Deck Weight. Resulting from a recent
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
103
Technology Platform: SYSTEM OPTIMIZATION
ETLP™ system design study for a SE Asia application,
the computed structural weight efficiency was 1.44.
This conceptual-level efficiency can be compared with
the following as-built platforms.
Hull Form / Location
TLPs in GoM
ETLP™ Platform in GoM
ETLP™ Platform W Africa
ETLP™ Platform in SE Asia
Maturity
As-Built
As-Built
As-Built
Conceptual
Ratio
0.6 – 0.8
1.1 – 1.2
1.3 – 1.4
1.4 – 1.5
The above efficiency ratios cannot be directly
compared, as different locations (regions having
significantly different metocean conditions) are
represented. Other aspects that contribute to a non
like-for-like comparison include differences in water
depth, payload and design approach. The structural
weight efficiency ratio is also affected by the maturity
of the design (e.g. conceptual level design vs. as-built).
As such, the 1.44 ratio for the conceptual ETLP™
platform for SE Asia would be expected to increase as
the design becomes more developed.
The structural weight efficiency ratio of an ETLP™
platform improves dramatically as metocean
conditions become more benign. This would be the
case in SE Asia. As a demonstration of the value of the
ETLP™ system, notice the improvement in structural
weight efficiency ratio between the conventional TLP
and the ETLP™ platform, both in the Gulf of Mexico.
CONCLUSION
The development of deepwater fields in Southeast
Asia is generally complex requiring the installation of
several major elements for a successful development.
Among these are subsea infrastructure, dry-tree unit,
and if so determined, FPSO and export shuttle tankers.
This paper has demonstrated the breakthrough
technology associated with the ETLP™ platform and
how it can become a conduit that ties Malaysia to the
global deepwater market. Many systems of the ETLP™
104
Technology Cluster: OIL AND GAS
platform lend themselves to fabrication in a variety of
countries, allowing domestic fabrication where
desirable. Developed from the TLP, the ETLP™ concept
has evolved into a very safe and cost effective solution.
A low risk installation operation that minimizes
offshore exposure time is also possible with the ETLP™
platform.
There are significant benefits to be gained from the
use of the ETLP™ concept for deepwater
developments in SE Asia:
• Due to the benign metocean conditions in the SE
Asia region and the inherent efficiency of the
concept, the ETLP™ platform can carry larger
topsides payloads and extend into deeper water
depths. At this time, nearly 2,000 m water depths
are possible with moderate sized payloads.
• Current technology developments are ongoing to
further extend water depths and payload limits.
• With surface completions, a much higher level of
reservoir management and productivity is possible
given the ease of intervention.
• Tender assisted drilling from a TLP is safe and
provides an opportunity to reduce the size and cost
of the platform.
• The ETLP™ system has very high operational uptime for drilling and related activities due to its
ability to maintain a very small watch circle even
for the most severe metocean conditions in SE Asia
• Structural weight efficiency of the concept is high
for SE Asia. The ETLP™ platform can carry 1.44 mt
of topsides payload for each 1.0 mt of hull and deck
steel.
• Since the ETLP™ system can be installed fully
integrated and commissioned, significant
reductions in offshore installation time and risk are
possible.
• When the ETLP™ platform is used in conjunction
with a FPSO, the separation distance between the
two vessels can be greatly reduced compared to
other Floaters such as Spar. Flow assurance issues
and fluid transfer line loads acting on the ETLP™
platform are minimized. Significant cost savings are
achieved with shorter fluid transfer lines.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
• Local content is facilitated by the ETLP™ concept.
Significant parts of the facility can be fabricated in
domestic SE Asia fabrication yards.
• Due to the compact form of the ETLP™ platform,
fabrication of the hull can be tailored to suit a
number of construction methods depending upon
the capabilities and strengths of a given Fabricator.
• When the economic life of a development has been
reached, the ETLP™ platform can more easily be
decommissioned, modified and redeployed to a
new field compared to other competing TLPs and
Floaters.
ACKNOWLEDGEMENT
The author would like to thank all those who helped
contribute to this paper and for ABB’s continued and
generous support of deepwater technology
development.
John W. Chianis is currently Vice President,
Deepwater Technology and Engineering at
ABB Floating Production Systems in
Houston, Texas. He has worked in the
deepwater engineering field since 1978. His
specific areas of technical competence are
structural design and analysis, Naval
Architecture, hydrodynamics and software
development. Although he has worked on
numerous types of floating hull forms, his prime area of interest
over the last 24 years has been on Tension Leg Platforms (TLPs).
In addition to countless TLP studies and related development
efforts, he has participated in the detailed design and delivery of
the following TLPs – Jolliet, Auger, Mars, Ram/Powell, Ursa, Marlin,
Brutus, Kizomba “A”, Kizomba “B” and most recently, the Magnolia
Extended Tension Leg Platform (ETLP) detailed design for 4,700
ft of water in the Gulf of Mexico.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
105
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
MATERIALS, DESIGN, AND MANUFACTURING ISSUES
ASSOCIATED WITH COMPOSITE PRODUCTS
FOR THE GAS AND PETROCHEMICAL INDUSTRY
John P. Coulter1, Joachim L. Grenestedt2 and Raymond A. Pearson3
Department of Mechanical Engineering and Mechanics,
Lehigh University, Bethlehem, PA 18015, USA.
1
[email protected], [email protected]
3
Materials Science & Engineering Department / Center for Polymer Science and Engineering,
Lehigh University, 5 East Packer Avenue, Bethlehem, PA 18015, USA.
[email protected]
ABSTRACT
The composite materials and petrochemical industries have been integrally linked for decades. In the case of
polymer-based composites, the relationship is truly circular. Polymer composites are based on petrochemical
constituents, and products made from them are used throughout the gas and petrochemical industry. The
effective development of any composite product requires the fully integrated consideration of materials, design,
and manufacturing issues. The proposed presentation will include an identification of the critical issues that
need to be addressed, and the relevant research work involving understanding of material behavior, design
issues, and intelligent manufacturing science development that is currently being pursued at Lehigh University.
Keywords: Fatigue Crack Propagation, Composite, Design, Neural Network.
1.0
INTRODUCTION
This paper addresses the material, design, and
manufacturing issues relevant to composite
fabrication. For this purpose, three important research
areas, which are currently being pursued at Lehigh
University, namely, fatigue crack propagation
resistance composite materials, design issues, and
intelligent manufacturing utilizing neural network
methods are highlighted in this paper.
1.1
Material Behavior
The fatigue crack propagation resistance (FCP) of
polymer-matrix composites is of paramount
importance when such materials are used for
engineering components that are subjected to cyclic
loading. The FCP behavior of model epoxy-matrix
composites has been studied in the past [1 - 3] and
several toughening mechanisms have been proposed.
Such mechanisms include matrix micro-cracking,
matrix shear banding, crack path deflection, and fiber
bridging.
The focus of the study presented in this paper is to
examine the FCP resistance of a series of filled epoxies
where the matrix-filler adhesion has either been
promoted or reduced using organosilane compounds.
Moreover, the FCP behavior will be examined as a
function of filler type (sphere vs. fiber) and after
moisture exposure.
This paper was presented at the International R&D Forum on Oil, Gas and Petrochemicals, Kuala Lumpur, Malaysia, 5-6 April 2004.
106
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
1.2
Composite Structural Design Challenges
Composite structures are in general not too difficult
to design, manufacture and analyze, except for when
it comes to features such as joints, edges and local load
introductions. At present, numerical tools such as
Finite Elements do not have reliable predictive
capabilities for these features. Composite structures
should in general be designed with as few joints and
irregularities as possible. However, there are often a
number of such features that still need to be
incorporated. The question is then how to best
accommodate these, without adding too much weight
and still gain the strength and stiffness advantages
expected from composites. A few relatively new
concepts for joining in particular steel and composites
are reviewed below. The focus of the work presented
was marine structures, although many of the concepts
are also applicable to other areas.
The joints discussed in this paper are mainly of the type
where a steel edge was molded into a composite at
the time the composite was made. Vacuum infusion
was employed for most specimens. The materials used
were stainless steel and glass or carbon fiber reinforced
polymers. The investigations performed may be
classified in two general types: adhesion and
geometry related, and are discussed separately in the
current manuscript.
1.3
Neural Network Based Intelligent Control of
Autoclave Cure Process
The autoclave cure process is used in the production
of advanced polymer composites, primarily for the
automotive, marine, and aerospace industries.
Historically, autoclave cure has been used to produce
composite parts of high quality. One major hindrance
to the production of composites through autoclave
cure is that the process is largely trial-and-error, leading
to much waste and parts with less-than-optimum
properties.
Recently, much effort has been put into controlling the
autoclave process both off-line and on-line. Off-line
Figure 1: Autoclave Components
controllers determine processing conditions prior to
cure, not allowing the variation of processing
parameters during cure. Consequently, unexpected
occurrences, variation in prepreg properties, and
imperfect controllers lead to poor quality and much
waste. On-line controllers avoid these problems by
maintaining product quality in real-time (adjusting
processing parameters when necessary). During this
study, an on-line controller was developed utilizing the
predictive capabilities of a neural network: a form of
artificial intelligence designed to model the human
brain’s ability to learn and devise relationships.
An autoclave is, in essence, an insulated pressure vessel
that allows heat, pressure, and vacuum to be applied
to a laminate (Figure 1). The laminate is laid-up by hand
onto a tool plate prior to cure and surrounded by a
lay-up assembly composed of numerous parts
including bleeder material which absorbs excess resin,
and a vacuum bag which renders the system air-tight.
Each layer of the laminate is a section of prepreg tape,
composed of a thermosetting resin matrix (commonly
epoxy) and fibrous inclusions (including glass, carbon,
and aramid fibers).
The time-dependent variation of temperature,
pressure, and vacuum during the autoclave process is
referred to as the cure cycle. Prepreg manufacturers
provide estimates of the proper cure cycle, estimates
developed through experience and trial-and-error.
Such manufacturer-defined cure cycles do not take
part thickness and shape, unexpected occurrences
during the process, or variations in prepreg properties
(such as resin content, molecular properties, and slight
inclusions) into account. While manufacturer-defined
cure cycles are widely used during autoclave cure, an
improved method of determining cure cycles is
necessary to ensure high product quality.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
107
Technology Platform: SYSTEM OPTIMIZATION
1.3.1
Numerical Model Based Control
One recently developed method of determining cure
cycles involves the use of numerical simulations.
Numerical simulations are hampered by the
complexity of the autoclave process. The process
involves heat transfer between the lay-up assembly
and the environment, between the assembly and the
prepreg, and between the prepreg and the tool plate.
The process also involves chemical reactions, void
formation and transport, resin flow through fibrous
media, fiber compaction, and much more. The
modeling of such a system is clearly an overwhelming
task. In all cases, simplifying assumptions must be
made to reduce the complexity of the problem. Even
when simplifying assumptions are included, numerical
models of the autoclave process require excessive
computer time. In total, excessive run time nullifies
numerical model applicability to on-line control, while
inaccuracies reduce the effectiveness of numerical
models for off-line control.
1.3.2
Knowledge Based Control
A solution to the problems of unexpected occurrences
and variation in prepreg properties is real-time control.
Real-time control has been accomplished with the use
of three aides: expert systems, fuzzy logic, and neural
networks. Expert system and fuzzy logic (knowledge
based) controllers rely on the availability of a human
expert. The human expert devises rules based on
available information, rules that must be sufficient to
provide control during the entire process. Formulation
of this complete set of rules is difficult and timeconsuming.
Some success has been achieved with knowledgebased controllers. This is partially due to the fact that
rules may be devised which depend only on readily
measurable quantities such as air temperature,
composite temperature, autoclave pressure, and
laminate compaction. In addition, knowledge based
controllers need not know the shape, thickness, or
components of a composite to provide adequate
control. Ciriscioli and Springer [4] developed one such
expert system controller. The controller provides good
108
Technology Cluster: OIL AND GAS
real-time control, allowing the creation of parts with
high quality when prepreg manufacturer-defined cure
cycles are less adequate or not provided.
Improvements over expert system controllers are
possible. Expert system controllers raise and lower
processing parameters only when rules are broken.
Therefore, an incomplete or errant set of rules does
not allow sufficient, continual control. Further, the
magnitude of change in processing parameters is
often not clearly defined.
1.3.3
Neural Network Based Control
Neural networks can provide rapid on-line prediction
of cure parameters, leading to the possibility of
continual, well-defined, real-time control. Joseph and
Hanratty [5] demonstrated the theoretical applicability
of neural network based control to the autoclave
process. They applied a neural network based
controller to a simulation of the autoclave process.
Results from this simulated control process showed
that a neural network based controller could
compensate in real-time for unexpected occurrences
and prepreg property variation. Results also showed
that composites of the same or higher quality than
from conventional cycles could be consistently
produced with the neural network controller, even
when unexpected occurrences and prepreg property
variation are not included.
Although Joseph and Hanratty’s controller
demonstrates the theoretical applicability of neural
network based control, improvements upon their
control procedure may be made. First, the controller
was developed under the assumption that the
autoclave cure cycle can be divided into four phases
(based on autoclave temperature): a first ramping
phase, a first holding phase, a second ramping phase,
a second holding phase. While this assumption is
consistent with traditional cure cycles, it may not allow
for optimum variation in temperature. Second, the
controller optimizes two important aspects of final
part quality, void size and composite thickness, but
does not inherently optimize other product quality
attributes such as cure extent and cycle time. Third, as
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
of yet, the controller has not been deployed on an
actual autoclave: all results were obtained from a
numerical simulation.
2.0
APPROACHES AND METHODS
2.1
FCP Study
The model epoxy system consisted of a bisphenol A
based epoxy cured with piperidine. Additional
bisphenol A resin was added to reduce the crosslink
density and promote toughness.
Two types of fillers were used: glass spheres with a
mean diameter of 42 µm and milled glass fibers with
average dimensions of 1.5 mm long and 16 µm in
diameter. Both types of fillers were washed with
methanol. Some fillers were treated with an
aminopropyl silane (APS) to increase matrix-filler
adhesion while other fillers were treated with a butyl
silane (nBS) to reduce adhesion.
The epoxy resin and 24 phr bisphenol-A (phr: part per
hundred parts resin by weight) were mixed at 180°
under vacuum and cooled to 80°. Then, 10 volume%
of reinforcement was added and mixed under vacuum.
Next, 5 phr of piperidine was added and mixed for
about 10 minutes under vacuum. The mixture was
poured into a mold, that had been preheated at 160°.
Curing was performed at 160° for 6 hrs. Specimens
were cut from the plaques prepared by this procedure.
2.1.1
FCP Testing Procedure
Fatigue crack propagation tests were performed
according to ASTM D647 [6] guidelines using compact
tension specimens (specimen size was 13.75 mm and
the thickness was 6 mm). The specimens were cut from
plaques that were prepared as described above.
Sinusoidal loading was applied to the specimens at a
frequency of 10Hz with a load ratio of 0.1 (minimum
load / maximum load). Crack opening displacements
were measured with a clip gauge. Crack lengths were
calculated automatically using the compliance of the
specimen.
2.1.2
Fractography of FCP Specimens
Fracture surfaces were observed by a JEOL 6300F or
Philips ESEM XL30 scanning electron microscopy (SEM)
with an acceleration voltage of 5kV. The fracture
surface was covered with sputtered Au-Pd before
examination. For optical microscopy, specimens were
mounted in a room temperature-cured epoxy and
ground perpendicular to the fracture surface and
along the width of the specimen using standard
petrographic techniques. These thin sections were
inspected using an Olympus BH2 transmission optical
microscope.
2.2
Structural Composite Design
Adhesion is affected by surface preparation, such as
grit blasting, solvent cleaning, priming, etc., as well as
by resin additives. Several researchers have
experimented with different surface preparations and
additives. For a particular glass/vinyl ester system of
interest for naval applications, the following
preparations were used:
- basic, which consisted of grit blasting and
trichloroethylene cleaning
- AF 163-2K.06 structural adhesive film from 3M
- priming system from PolyFiber consisting of C-2200
metal surface cleaner, EP-420 epoxy primer, and EP430 catalyst - EC-3901 structural adhesive primer
from 3M
- Loctite Hysol 9430 epoxy adhesive
- KZ 55 and NZ 97 zirconate coupling agents from
Kenrich Petrochemicals, Inc., added to the vinyl ester
resin before infusion
- HX0603-4 hydrochloric acid (HCl) etch.
The joints presently discussed were mainly of the type
where a steel edge was molded into a composite at
the time the composite was made. Vacuum infusion
was employed for most specimens. The materials used
were stainless steel and glass or carbon fiber reinforced
polymers. The investigations performed may be
classified in two general types: adhesion and geometry
related.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
109
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Figure 2: The FCP Behavior of Epoxy Reinforced with 10 vol.%
Glass Spheres is not very sensitive to epoxy-particle adhesion
Figure 4: The FCP behavior of epoxy reinforced with 10 vol.%
glass spheres is VERY sensitive to epoxy-particle adhesion after
exposure to moisture
Figure 3: The FCP behavior of epoxy reinforced with 10 vol.%
milled glass fibers is not very sensitive to epoxy-particle adhesion
Figure 5: The FCP behavior of epoxy reinforced with 10 vol.%
milled glass fibers is also VERY sensitive to epoxy-particle adhesion
after exposure to moisture
3.0
RESULTS AND DISCUSSION
3.1
FCP Studies
towards the reinforcements. Therefore, the amount of
adhesion between the reinforcements and the matrix
does not affect the mechanical properties under dry
conditions.
Figures 2 and 3 contain graphs that quantify the
fatigue crack propagation behavior of the neat epoxy
and the two types epoxy composites before moisture
exposure. In both figures, the x and y-axes denote the
applied ∆K at the crack tip and the resulting crack
propagation rate, respectively.
In all cases, the glass-reinforced epoxies exhibited
significant improvement in fatigue crack propagation
resistance than that of neat epoxy. However, the use
of different surface treatments on these
reinforcements did not significantly alter the fatigue
crack propagation behavior under dry conditions.
Residual stresses around the glass reinforcements due
to the coefficient of thermal expansion mismatch
between epoxy and glass provide compressive forces
110
Fractography studies revealed that pinning and microcracking occurred when the epoxy matrix is dry. More
detailed discussions on this study have been
submitted for publication [6, 7].
The fatigue crack propagation behavior of the neat
epoxy and glass-reinforced epoxies after moisture
exposure are shown in Figures 4 and 5. As is the case
before moisture exposure, improved fatigue crack
propagation resistance is observed for all of the
composites compared to that of the neat epoxy. In
contrast to the results before moisture exposure, the
fatigue crack propagation behavior after moisture
exposure strongly depends on the surface treatment
of the reinforcements. Glass fillers treated with nBS, in
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
which only weak van der Waals interaction between
adhesion promoter and matrix resin is expected,
exhibited improved fatigue crack propagation
resistance especially at high ∆K levels.
Subsequent fractography studies revealed a relatively
large and visually observable damage zone ahead of
the crack tip when the matrix-filler adhesion was poor
and the ∆K levels are high. A shift in threshold behavior
at low ∆K levels suggests that fiber bridging or crack
deflection mechanisms are occurring and indeed such
mechanisms were observed.
3.2
stainless steel plate with fiberglass skin infused on top
and bottom. These specimens were thermally cycled
in an environmental chamber, using various humidity
levels. The crack driving force was due mainly to the
difference in coefficients of thermal expansion of the
steel and the composite. It is interesting, although
discouraging, to note that there was no appreciable
debonding (interfacial cracking) when the discs were
cycled in dry air, but very extensive debonding when
cycling in humid air. The worst condition appeared to
be cycling in high humidity and in a freeze-thaw cycle.
In Figure 7, the debonded areas of a specimen is shown
after different numbers of cycles.
Structural Composite Design
3.2.2
3.2.1
Geometry of Simple Joints
Adhesion of Vinyl Ester Composite to Steel
Figure 6 shows the average transverse strength for the
spectrum of composite structure under study. The
strongest was to precure the film onto the stainless
sheet under a vacuum bag for 5 minutes, then vacuum
infuse the vinyl ester resin, and at last postcure the film
(and composite) for one hour. The preferred technique
was to use the PolyFiber primer, which yielded almost
the same strength as the AF 163 film but with
considerably less effort. All tests were performed
under dry conditions. The KZ 55 and NZ 97 zirconate
additives lead to reduced strengths under these
conditions. However, it is expected that they would
improve strength under hot-wet conditions.
Regarding joint shape, a number of different concepts
have been investigated. For a simple tensile loaded
joint, feasible designs include single lap (shear) joints,
scarf joints, and step lap joints. These may be designed
in different ways. A major challenge joining dissimilar
materials is that elastic mismatch leads to high stress
concentrations. When the two adherends have
reasonably similar compliances, the scarf and step lap
joints may be designed such that the effective stiffness
Grenestedt and Cao [8] recently performed
environmental tests on disc specimens consisting of a
Figure 6: Adhesion as measured by transverse tensile testing of
a fiberglass reinforced vinyl ester composite / stainless steel joint
Figure 7: Debonding propagating with the number of cycles. The
four photos were taken after 0, 12, 36, and 84 cycles (left top to
right bottom). The debonded areas have been highlighted for
clarity
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
111
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
is constant along the adherends as well as over the
whole joint. The average axial strain would then be
constant, and desirably also the local strain would be
fairly uniform. For materials with vastly different
stiffnesses but similar strengths, such as stainless steel
and fiberglass, this approach leads to very bulky joints.
For example, if the steel were 10 times stiffer than the
fiberglass but had the same strength, then the
fiberglass would have to be 10 times thicker than what
would be required for strength. This is certainly not an
optimal configuration.
Other ways to decrease the elastic mismatch include
to perforate the stiffer adherend, to taper it, to make
tongues and grooves, or to make the overlap area
wavy. All these schemes may be gradual, such that the
stiffness of the steel is progressively decreased. These
concepts are dealt with individually below.
3.2.2.1
Perforated Joints
Undén and Ridder [9] introduced the graded
perforation approach, and Melograna and Grenestedt
[10] used this technique for the presently considered
materials. Figure 8 shows two different perforation
patterns, which were investigated: circular and
triangular holes. The specimens were made by placing
the steel adherends between glass fiber fabrics under
a vacuum bag and vacuum infusing vinyl ester resin.
The strongest joint configurations were those that
failed by debonding at both ends of the joint more or
less simultaneously, as depicted in Figure 9. This could
be accomplished by increasing or decreasing the
number of rows of perforations in the steel adherend.
There was no major difference in joint strength
between the two perforation schemes. The best
perforation patterns investigated lead to joints which
were 30% stronger than their non-perforated
counterparts (when no primers were used).
3.2.2.2
Tapered Joints
Figure 10 shows the edge view of the two different
taper joint configurations tested. Tapering the steel
adherends would be expected to reduce the elastic
112
Figure 8: Photo of the steel adherends showing the two different
perforation patterns employed: round (left) and triangular (right)
graded perforations
Debonds began at both
the butt end of the steel
and the end of the
composite. Both grew
towards each other until
failure.
Figure 9: The strongest joints were perforated such that
debonding simultaneously would start from both the tip of the
steel and the tip of the composite adherends
Figure 10: Edge view of the two different taper configurations
tested. All dimensions in mm.
mismatch, at the cost of increasing through-thethickness stresses. Such joints have been investigated
by many researchers, including Melograna and
Grenestedt [11] who used the presently considered
materials and fabrication technique (stainless steel and
vacuum infused glass fiber reinforced vinyl ester). A
problem with the tapered steel adherends in
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
conjunction with the vacuum infusion technique was
that fabrication quality decreased. The taper would
lead either to waviness, or regions with less compacted
fibers. The latter lead to race tracking of the resin
during the infusion and areas with non-proper
infusion. The result of this study was that overall, the
tapered specimens were neither stronger nor weaker
than their plain counterparts. If the infusions could
be improved, the tapered joints would probably be
superior to the plain joints.
3.2.2.3
Tongue-and-Groove Joints
Tongue-and-groove joints are commonly used for
joining wood. Dvorak [12] used this concept to join
composites and steel. Melograna, Grenestedt and
Maroun [13] used it to join carbon fiber composite and
stainless steel. In the latter study, nine geometrically
different tongue-and-groove joint types were
manufactured and tested, Figure 11. The composite
was a vacuum and oven cured prepreg T700 carbon
fiber / epoxy. Conventional single lap joints were also
manufactured using the same materials and adhesive
and with a 25.4 mm overlap. The results of the tests
are presented in Figure 12. The “joint efficiency”,
presently defined as the strength of the joint divided
by the yield strength of the steel adherend, was
approximately 43% for the single lap joint and 60%
for the best tongue and groove joints. The carbon fiber
composite was 40% thinner than the steel, although
still stronger. With similar thickness of the adherends,
the joint efficiency may be expected to approach unity
for the best tongue-and-groove joints.
3.2.2.3
Figure 11: Specimens with one or
two steel The widths of the tongues
of the first three specimens were
different. Specimen 5 had a lollipop
shaped steel tongue. A specimen
with six tapered steel tongues and
another with eight straight tongues
are shown to the right.
Figure 12: Results from tests on tongue-and-groove joints
Sun, had the opposite trend: the wavy joints were
substantially weaker than their plain counterparts.
Grenestedt and Melograna [16] performed a classic
Finite Element based shape optimization of a wavy
joint between stainless steel and a carbon fiber
composite. Such joints were manufactured and tested
and proved to be approximately 15% stronger than
single lap joints between the same materials.
3.2.2.4
Wavy Joints
Wavy single lap joints between identical carbon fiber
adherends were proposed by Zeng and Sun [14], who
reported a substantial increase in tensile strength.
Melograna and Grenestedt [15] reproduced these
joints using different materials but the same geometry.
They used the same carbon fiber composite as used
for their tongue-and-groove specimens (see above).
The specimens of Melograna and Grenestedt, which
were considerably stronger than those of Zeng and
Topside Joints
Various schemes for joining a composite topside
structure to a steel hull have been proposed. A concept
of primary interest is to attach a steel edge to the
composite topside, and then weld the whole topside
to the steel deck using this edge. Such concepts have
been studied extensively by Hildebrand and Hentinen
[17], Bohlmann and Fogarty [18] who designed and
tested a bonded-bolted joint, and Clifford et al. [19]
who fabricated and tested a vacuum infused joint. The
latter joint was recently re-designed, manufactured
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
113
Technology Platform: SYSTEM OPTIMIZATION
Figure 13: Vacuum infused joint between a steel edge (left in
figure) and a glass reinforced vinyl ester skin / balsa core sandwich.
The specimen was tested in bending, which resulted in yielding
of the steel rather than composite failure
and tested by Cao and Grenestedt [8]. The re-designed
joint was both stronger and lighter. The general shape
of this joint is seen in Figure 13.
3.3
Neural Network Based Intelligent Control of
Autoclave Cure Process
During the present study, two real-time, intelligent
controllers were developed to guide the autoclave
cure process at discrete time intervals. The controllers
were designed to optimize many aspects of final part
quality: 1) minimum void size, 2) complete compaction,
3) minimum cure cycle time, 4) complete cure, and 5)
consistent molecular properties (avoiding degradation
and thermal runaway). Additionally, fewer
assumptions about the autoclave process were made
than with previous controllers, allowing more free
control of processing parameters.
The present controllers make use of the rapid
predicting capabilities of backpropagation neural
networks. Neural networks for each controller were
trained with data generated by a numerical model
developed by Telikicherla, et. al. [20, 21]. This model
incorporates many aspects of autoclave cure, including
heat transfer within the autoclave environment, giving
a good estimate of the total cure process.
Unfortunately, the CPU time (on a Cray YMP) necessary
to predict the outcome of one cure cycle is
approximately 8 hours [21]. A training process allows
the neural network to learn relationships between
input data and output data. (For a discussion of the
training process and capabilities of neural networks,
see Neural Computing [22] or any other of many books
on the subject.) In essence, the neural network learns
114
Technology Cluster: OIL AND GAS
to act like the numerical model running “forward” or
even “backward”. The application of neural networks
for intelligent control is beneficial for two reasons. First,
neural network prediction is much faster than
numerical model prediction. Neural networks can give
a prediction of autoclave cure in less than a second,
while such a prediction would take hours from
numerical models. Second, neural networks can learn
inverse relationships: Given a desired outcome, neural
networks can provide a direct estimate of the proper
processing parameters to achieve that goal.
Both controllers utilize descriptive indicators of the
state of a composite during processing: degree of cure,
degree of compaction, and viscosity. These can be
monitored through ultrasonic techniques [23]. Degree
of cure may be described as the proportion of
crosslinking a thermosetting resin has undergone.
Crosslinking reactions are exothermic, often leading
to thermal runaway (temperatures increasing above
the heater temperature to a point when degradation
of the polymer occurs). Accordingly, degree of cure
may also be thought of as the quotient of heat given
off during cure up to a certain time over the total heat
of reaction. Degree of compaction refers to the
thickness of the composite. Prepreg tape always
contains excess resin that must be forced out during
autoclave cure: the less resin, the more compacted the
composite, the better the composite properties (more
resembling that of the fibers). Viscosity is an important
parameter because the bulk of compaction occurs
when viscosity is low and resin flows more readily. Final
void content of the composite is also related to
viscosity. The three quantities of degree of cure, degree
of compaction, and resin viscosity describe the state
of the composite at all times, and are very useful for
control purposes.
Both of the developed controllers focus on specifying
autoclave temperature and pressure during
processing. Since no simple relationship between
autoclave temperature and final part quality has been
developed, and since autoclave temperature may vary
continually during processing, autoclave temperature
is controlled through the use of neural networks.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Olivier et. al. [24] imply a straight-forward relationship
between autoclave pressure and final part quality of
carbon/epoxy composites. Through experimentation,
they found that the final void content of produced
composites decreases asymptotically as autoclave
holding pressure is increased. Additionally, they found
that minimum final void content is obtained when
pressure is applied just after resin viscosity reaches a
minimum. Accordingly, the control of pressure during
this study was accomplished through a simple rule:
Pressure is applied at the level determined by the
prepreg manufacturer after resin viscosity reaches a
minimum value. The magnitude of pressure
application determined by the prepreg manufacturer
is deemed sufficient to provide adequate compaction
and void dissolution.
3.3.1 “Forward” Controller
One controller utilizes a “forward” neural network,
trained to predict exactly what the numerical model
predicts. Input to the neural network includes
indicators of the current state of the autoclave and
composite (Figure 14): 1) heater temperature, 2)
average resin degree of cure, 3) average resin viscosity,
4) degree of compaction, 5) average temperature
within the composite, 6) rate of change of temperature
within the composite, and 7) cure rate. Given this
information along with 8) the heater temperature
adjustment to be applied over the next time step, the
neural network predicts the cure indicators at the end
of the next time step: 1) average degree of cure, 2)
average resin viscosity, and 3) degree of compaction.
The learning process for the neural network is timeconsuming. Initially, the numerical model generates
numerous sets of data from several process
simulations. Care is taken to ensure that the generated
data represents the variety of possible values the
controller will see during an actual autoclave process.
The neural network is shown the data until it
successfully associates inputs with outputs. The neural
network is then tested on an unseen data set to assess
its performance. When its performance is optimized,
learning is complete. The neural network may then
be applied to the actual process.
A diagram of the iterative control process which
determines the appropriate HTA over the next time
step is shown in Figure 15. This control algorithm
requires the determination of weights representing
the relative importance of each of the monitored
parameters during all stages of cure. Since the
importance of each parameter varies continually
during cure, this is not a straight-forward
determination. Inaccuracy of the determined weights
reduces the accuracy of the “forward” controller.
3.3.2 “Inverse” Controller
The other controller exploits the ability of neural
networks to learn inverse relationships. In the “inverse”
Figure 15: Control Procedure Utilizing a “Forward” Neural
Network
Figure 14: Forward Neural Network Structure
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
115
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
Necessary to both control algorithms is a desired set
of data giving desired numerical values for degree of
cure, resin viscosity, and degree of compaction at all
times. This data set may be determined in the
following manner: First, run the numerical model with
the manufacturer-defined cure cycle and plot the
monitored variables. Then, modify the generated plots
so as to produce composites of high quality in
minimum cycle time.
3.3.3 Performance Testing
Figure 16: Control Procedure Utilizing an “Inverse” Neural
Network
Figure 17: “Inverse” Neural Network Structure
control algorithm (Figure 16), a neural network (Figure
17) is trained to give a direct estimate of the
appropriate HTA. This avoids the iterative process
necessary with the “forward” controller.
The “inverse” neural network is trained on the same
data as the “forward” neural network, but in a different
format. The neural network learns to perform like a
numerical model running backward. It should be
noted that prediction for the “inverse” neural network
is more complicated than for the “forward” neural
network, for there is a less direct relationship between
input and output. Thus, the accuracy of the “inverse”
neural network is expected to be less than the accuracy
of the “forward” neural network.
116
Back-propagation neural networks for each controller
were trained to yield their optimum predictive
capabilities. Then, both neural networks were tested
on data generated by the numerical model. The test
consisted of feeding the neural network the required
inputs and noting how closely the neural network
predicted the correct output (as given by the
numerical model).
The “forward” neural network was found to predict
values for degree of cure, degree of compaction, and
the natural log of viscosity very well during the entire
cure cycle. Root-mean-squared (RMS) errors were
calculated from neural network testing on ten trial cure
cycles and were found to be low: 0.0093 for degree of
cure, 0.016 for degree of compaction, and 0.13 ln(Pa.s)
for viscosity, the magnitude of error not varying
appreciably with extent of compaction. According to
the results of preliminary testing, the accuracy of the
“forward” neural network appeared adequate for
control purposes.
The “inverse” network produced more error than the
“forward” network. HTA was allowed to take on integer
values between -5 and 5 °C/min. As such, the neural
network output was rounded to the nearest integer
for control purposes. The “inverse” neural network was
tested on ten trial cure cycles. RMS error from the
neural network throughout the entire cure cycle was
1.00 °C/min. Neural network predictions were
concluded to be much more accurate before complete
compaction than at complete compaction. The RMS
error at degrees of compaction less than 1.0 was
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
0.40 °C/min, while the RMS error at complete
compaction was 1.37 °C/min. Even though diminished
accuracy was obtained at complete compaction, the
neural network provided a good estimate of the
proper HTA for any given time step.
The “forward” controller utilizes a more complicated
control algorithm, requiring an iterative process to
determine the proper HTA. The iterative process
requires the specification of weights representing the
relative importance of the three monitored variables
throughout cure. A C program was coded to
determine numerical weighting constants, utilizing a
factorial design of experiments approach. Weights
determined by the search program are shown in Table
1. Results obtained from ten trial cure cycles with these
weights yielded a total RMS error of 1.67 °C/min, an
RMS error before complete compaction of 1.34 °C/min,
and an RMS error at complete compaction of 2.36 °C/
min. Although the “forward” neural network was found
to be more accurate than the “inverse” neural network,
the performance of the “forward” controller was found
to be worse than that of the “inverse” controller.
Table 1: Weights Representing the Relative Importance of
Parameter
Stage Number
Parameter
Stage
1
(δ=0)
Stage
2
(0<δ<1)
Stage
3
(δ=1)
degree of cure (α)
0.50
1.00
1.30
degree of
compaction (δ)
0.10
1.20
4.50
ln(viscosity) (lnµ)
1.50
1.10
1.10
4.0
occurred in the fiber filled system. Matrix-filler
adhesion did not significantly affect FCP behavior
under dry conditions; however, adhesion was an
important factor when the epoxy matrix was saturated
with moisture. Microscopy studies indicate that more
matrix shear yielding occurred at crack tips when the
matrix-filler adhesion was poor. These results suggest
that maximum toughness is achieved when poorly
bonded fibers are used to reinforced an epoxy resin
that is plasticized with moisture.
Secondly, composite design procedures were
investigated, and were discussed in detail with respect
to two general types, namely adhesion and geometry
related design aspects.
Finally, in the current study, two neural network based
control algorithms and explored their applicability to
the autoclave process were developed. The developed
controllers differ from previous controllers in that
fewer assumptions about temperature variation
during the autoclave process were included, while
more aspects of final part quality were considered.
Therefore, the present controllers are more broad and
complete than previous controllers.
Both control algorithms accomplish the same goals:
to provide an estimate of the proper heater
temperature adjustment at discrete time steps, and to
control autoclave pressure based on the monitoring
of resin viscosity. The ultimate goal of this study is the
development of a controller, which guides the
autoclave to consistently produce high-quality
composites. Considering this goal, the “inverse”
controller is the most promising for two reasons: 1)
prediction speed is greater and 2) controller accuracy
is much greater.
SUMMARY AND CONCLUSIONS
First, fatigue crack propagation behavior of epoxymatrix composites reinforced with either glass spheres
or short glass fibers was investigated using linear
elastic fracture mechanics. FCP resistance improved
with the addition of either type of filler, however, short
glass fibers were more effective than glass spheres.
Microscopy studies indicate that more bridging
Future efforts are being directed toward applying the
developed controllers to actual experimental
autoclave processes. Products will be created utilizing
off-line and on-line control, and total quality will be
evaluated, allowing assessment of the applicability of
the developed controllers to the autoclave process.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
117
Technology Platform: SYSTEM OPTIMIZATION
5.0
ACKNOWLEDGEMENT
The FCP study was performed by Takafumi Kawaguchi
and was supported by the Osaka Gas Co. Ltd. in Japan.
Secondly, the composite design work was supported
in part by ONR Grant N00014-01-1-0956, in part by the
Pennsylvania Infrastructure Technology Alliance (PITA)
Grant PIT-190-00 and in part by the Department of
Mechanical Engineering and Mechanics, Lehigh
University.
Technology Cluster: OIL AND GAS
[7] Kawaguchi, T. and Pearson, R.A. 2004. The Moisture Effect on
the Fatigue Crack Growth of Glass Particle and Fibre Reinforced
Epoxies with Strong and Weak Bonding Conditions: Part 2 A
Microscopic Study of Toughening Mechanisms. (Submitted).
[8] Cao, J., Grenestedt, J.L., 2002. Test of a redesigned glass-fiber
reinforced vinyl ester to steel joint for use between a naval
GRP superstructure and a steel hull (Submitted).
[9] Undén, H., Ridder, S.-O., 1985. Load-Introducing Armature as
Component Part of a Laminated Structural Element. United
States Patent 4, 673, 606.
[10] Melograna, J.D., Grenestedt, J.L., 2002. Improving Joints
Between Composites and Steel Using Perforations,
Composites: Part A, 33: 1253-1261.
Finally, the controller design was based on research
conducted by Demirci [25]. NeuralWorks Professional
II/Plus v. 5.0, from NeuralWare, Inc. was applied for the
creation of neural networks throughout this study, and
the work was funded by the Presidential Faculty
Fellowship Program of the National Science
Foundation (grant number DDM-9350209). The
support of Drs. Kesh Narayanan and Bruce Kramer at
the National Science Foundation is greatly
appreciated.
[11] Melograna, J.D., Grenestedt, J.L., 2002, Various Adhesive Joints
Between Steel and Glass Fiber Reinforced Vinyl Ester,” in
Joseph D. Melograna’s Master Thesis, Department of
Mechanical Engineering and Mechanics, Lehigh University,
Bethlehem, PA 18015, USA.
REFERENCES
[14] Zeng, Q.-G., Sun, C.T., 2001. Novel Design of a Bonded Lab Joint,”
AIAA J., 39(10): 1991-1996.
[1] Azimi HR, Pearson RA, Hertzberg RW. 1995. Role of crack tip
shielding mechanisms in fatigue of hybrid epoxy composites
containing rubber and solid glass spheres. J. Appl. Polym. Sci.
58:449.
[15] Melograna, J.D., Grenestedt, J.L., 2001, Revisiting a Wavy
Bonded Single Lap Joint, (Submitted).
[2] Urbaczewski-Espuche E, Gerard JF, Pascault JP, Sautereau H.
1993. Toughness Improvement of an epoxy/anhydride matrix.
Influence on processing and fatigue properties of
unidirectional glass-fiber composites. J. Appl. Polym. Sci.
47:991.
[3] Sautereau H., Maazouz A., Gerard J.F., Trotignon JP. 1995.
Fatigue behavior of glass bead filled epoxy. J. Mater. Sci.
30:1715.
[4] Ciriscioli, P. R., and G. S. Springer, Journal of Composite Materials,
1991, vol. 25, n. 12, pp. 1542-1587.
[5] Joseph, B., and F. W. Hanratty, Industrial and Engineering
Chemistry Research, 1993, vol. 32, iss. 9, pp. 1951-1961.
[6] Kawaguchi, T. and Pearson, R.A. 2004. The Moisture Effect on
the Fatigue Crack Growth of Glass Particle and Fibre Reinforced
Epoxies with Strong and Weak Bonding Conditions: Part 1
Macroscopic Fatigue Crack Propagation Behavior. (Submitted).
118
[12] Dvorak, G.J., Zhang, J., Canyurt, O., 2001. Adhesive tongue-andgroove joints for thick composite laminates,” Composites
Science and Technology, 61(8): 1123-1142.
[13] Melograna, J.D., Grenestedt, J.L., Maroun, W.J., 2002. Adhesive
Tongue-and-Groove Joints Between Thin Carbon Fiber
Laminates and Steel, (Accepted, to appear in Composites Part
A).
[16] Grenestedt, J.L., Melograna, 2002. Design Optimization of a
Hybrid Single Lap Joint, in Joseph D. Melograna’s Master Thesis,
Department of Mechanical Engineering and Mechanics,
Lehigh University, Bethlehem, PA 18015, USA.
[17] Hildebrand, M., Hentinen, M., 1998. Efficient solutions for joints
between large FRP-sandwich and metal structures, 19th
International SAMPE Europe Conference. Paris, 22 - 24.
[18] Bohlmann, R.E., Fogarty, J.H., “Demonstration of a Composite
to Steel Deck Joint on a Navy Destroyer,” 2001. Proc.
International Conference on Marine Applications of
Composite Materials (MACM), Melbourne, Florida, USA, March
19-21.
[19] Clifford, S.M., Manger, C.I.C., Clyne, T.W., 2002. Characterisation
of a glass-fiber reinforced vinyl ester to steel joint for use
between a naval GRP superstructure and a steel hull,
Composite Structures, 57: 59-66.
[20] Telikicherla, M. K., M. C. Altan and F. C. Lai, International
Communications in Heat and Mass Transfer, Nov. - Dec. 1994,
21(6): 785-797.
PLATFORM VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004
Technology Platform: SYSTEM OPTIMIZATION
Technology Cluster: OIL AND GAS
[21] Telikicherla, M. K., X. Li, M. C. Altan and F. C. Lai, Thermal
Processing of Materials: Thermo-Mechanics, Controls, and
Composites, ASME Heat Transfer Division, 1994, (Publication)
HTD 289: 213-221.
[22] NeuralWare, Inc., Neural Computing: A Technology Handbook
for Professional II/Plus and NeuralWorks Explorer, 1995,
NeuralWare, Inc. Technical Publications Group, Pittsburgh,
Pennsylvania.
[23] Kline, R. A., and M. C. Altan, 1993. Ceramic Matrix Composites
and Other Systems, Vol. II, Proceedings of the Ninth
International Conference on Composite Materials, Madrid,
Spain, July 1993, 441-448.
[24] Olivier, P., J. P. Cottu, and B. Ferret, July 1995, Composites, 26:
509-515.
[25] Demirci, H. H.,“An Investigation of Real-Time Intelligent Control
of Molding Processes,” Ph. D. Dissertation, Lehigh University,
1994.
APPENDICES
Nomenclature
α – degree of cure
µ – viscosity (Pa.s)
δ – degree of compaction
T – temperature ( °C)
∆( ) – change in quantity per minute
( )avg – quantity averaged throughout composite
( )i – quantity at present time step
( )i+1 – quantity at succeeding time step
∆Tposs – possible heater temperature adjustment
(°C/min)
John P. Coulter is Professor and Associate
Dean at the P. C. Rossin College of
Engineering and Applied Science of Lehigh
University. He is formally a member of the
Department of Mechanical Engineering and
Mechanics where he is involved in material
processing, manufacturing science, and
intelligent material systems research. He
received his bachelor's, master's and
doctoral degrees in mechanical and aerospace engineering from
the University of Delaware in 1983, 1985, and 1988 respectively.
His graduate work was in the areas of thermal and fluid sciences,
composite materials, and manufacturing. Prior to joining the
faculty at Lehigh in 1990, Dr. Coulter served as Senior Research
Engineer and project leader of the adaptive structures research
effort at Lord Corporation. He is currently responsible for all
undergraduate and graduate manufacturing science components
of the Lehigh mechanical engineering curriculum. As an Associate
Dean, he is also responsible for all graduate study and research
programs in the College of Engineering.
Dr. Coulter holds three patents, and has authored over 100
professional technical publications. He has also organized several
international workshops and symposia related to intelligent
material processing, and serves on the editorial board of The
Journal of Material Processing and Manufacturing Science. He
has received numerous research and teaching awards while at
Lehigh, including both National Young Investigator (NYI) and
Presidential Faculty Fellowship (PFF) awards from The National
Science Foundation.
He has also received a Future Technology Award from the Society
of Plastics Engineers and two Innovative Curriculum Awards from
ASME. His manufacturing research efforts at Lehigh have been
consistently funded by industrial sponsors as well as federal and
state agencies for well over a decade.
VOLUME FOUR NUMBER ONE JANUARY - JUNE 2004 PLATFORM
119
PLATFORM is a biannual, peer-reviewed
journal of Universiti Teknologi
PETRONAS. It serves as a medium for
faculty members, students and industry
professionals to share their knowledge,
views, experiences and discoveries in
their areas of interest and expertise. It
comprises collections of, but not limited
to, papers presented by the academic
staff of the University at various local and
international conferences, conventions
and seminars.
The entries range from opinions and
views on engineering, technology and
social issues to deliberations on the
progress and outcomes of academic
research.
Opinions expressed in this journal need
not necessarily reflect the official views
of the University.
All materials are copyright of Universiti
Teknologi PETRONAS. Reproduction in
whole or in part is not permitted without
the written permission from the
University.
NOTES FOR CONTRIBUTORS
Instructions to Authors
Authors of articles that fit the aims, scopes and policies of
this journal are invited to submit soft and hard copies to the
editor. Paper should be written in English. Authors are
encouraged to obtain assistance in the writing and editing
of their papers prior to submission. For papers presented or
published elsewhere, authors should include the details of
the conference or seminar.
Manuscript should be prepared in accordance with the
following:
1. The text should be preceded by a short abstract of 50100 words and four or so keywords.
2. The manuscript must be typed on one side of the paper,
double-spaced throughout with wide margins not
exceeding 3,500 words although exceptions will be
made.
3. Figures and tables have to be labelled and should be
included in the text. Authors are advised to refer to recent
issues of the journals to obtain the format for references.
4. Footnotes should be kept to a minimum and be as brief
as possible; they must be numbered consecutively.
5. Special care should be given to the preparation of the
drawings for the figures and diagrams. Except for a
reduction in size, they will appear in the final printing in
exactly the same form as submitted by the author.
6. Reference should be indicated by the authors’ last names
and year of publications.
Chief Editor, PLATFORM
Universiti Teknologi PETRONAS
Bandar Seri Iskandar
31750 Tronoh
Perak Darul Ridzuan
MALAYSIA
V O LU M E F O U R N U M B E R O N E J A N UA RY - J U N E 2 0 0 4
P L AT F O R M
P L AT F O R M
Volume 4 Number 1
Jan - Jun 2004
SPECIAL INTEREST
Technology Innovation and Role of Research University
Sung-Kee Chung
3
Technology Cluster: OIL AND GAS
Technology Platform: Reservoir Engineering
11
A Holistic Approach to IOR
Brian GD Smart
18
The Development of An Optimal Grid Coarsening Scheme:
Interplay of Fluid Forces and Higher Moments of Fine-Scale Flow Data
N. H. Darman, G. E. Pickup and K. S. Sorbie
26
Surfactant Systems for Various Fields of EOR: Drilling Fluids,
Microemulsions Control of Viscosity, Breaking of Emulsions
Heinz Hoffmann
36
Technology Platform: Oilfield Gas Treatment and Utilization
Selective Fischer Tropsch Wax Hydrocracking – Opportunity for
Improvement of Overall Gas to Liquid Processing
Jack CQ Fletcher, Walter Böhringer and Athanasios Kotsiopoulos
46
Theory of Autothermal Reforming for Syngas Production from
Natural Gas
Kunio Hirotani
54
Development of Defect-Free and High Performance Asymmetric
Membrane for Gas Separation Processes
Ahmad Fauzi Ismail, Ng Be Cheer, Hasrinah Hasbullah and
Mohd. Sohaimi Abdullah
68
Technology Platform: System Optimization
Formulation Engineering and Product-Process Interface
J P K Seville and P J Fryer
84
Advancements in Tension Leg Platform Technology
John W. Chianis
91
Materials and Manufacturing and Design Issue Associated with
Composite Products for Gas and Petrochemical Industry
John P. Coulter, Joachim L. Grenestedt and Raymond A. Pearson
106
V O LU M E F O U R N U M B E R O N E J A N UA RY - J U N E 2 0 0 4
Air Injection-Based IOR for Light Oil Reservoirs
R. G. Moore, S. A. Mehta and M. G. Ursenbach
ISSN 1511-6794