HP_2010_12 by androsov.info

DECEMBER 2010

HPIMPACT

SPECIALREPORT

BONUSREPORT

European downstream safety performance

PLANT DESIGN AND ENGINEERING

HEAT TRANSFER

Advanced field diagnostics

Innovative methods facilitate project success

Energy conservation and reliability key goals

www.HydrocarbonProcessing.com

Select 74 at www.HydrocarbonProcessing.com/RS

DECEMBER 2010 • VOL. 89 NO. 12 www.HydrocarbonProcessing.com

SPECIAL REPORT: PLANT DESIGN AND ENGINEERING

Maximize return on capital projects

Best practices in treating liquefied petroleum gas are defined

How to achieve all of your major capital investments objectives on budget and schedule J. Humphries

Advantages of removing H2S before the LPG unit and energy optimization of the LPG splitter B. Ardalan, M. KHORSAND Movaghar and M. Maleki

Consider new coatings for maintenance turnaround

Operator-driven reliability: Training and implementation

Improve engineering via whole-plant design optimization

Epoxy system provides speed of application as well as environmental benefits for large storage tanks C. Karner and B. Toews Follow these guidelines to help execute the process T. Hanlon and T. McDougal

New simulation methods identify cost-effective advantages early H. Martín Rodríguez, A. Cano and M. Matzopoulos

BONUS REPORT: HEAT TRANSFER

51 53 57

Prevent flow-induced vibration in heat exchangers New technology can be used in existing tube bundles to enable higher flowrates without risk of tube damage A. S. Wanni and Z. F. Ruzek

Optimize heat exchanger installations

17 European downstream oil industry safety performance 17 Fieldbus Foundation registers devices with advanced field diagnostics 20 Ethanol to hit refining profitability

Convert waste heat into eco-friendly energy New developments, such as the organic Rankine cycle, help operations go ‘green’ A. Bourji, J. Barnhart, J. Winningham and A. Winstead

Benefits exceeding the guarantee were realized through APC by simultaneously stabilizing and optimizing the plant in the presence of significant feed density variations that impact the entire plant operation P. Banerjee, K. V. Siva Rama Brahmam, S. Al-Azmi, L. Nayfeh and K. Al-Azmi

Critical concepts in fieldbus system design differ from a conventional DCS Here are some key topics calling for a conceptual change from the traditional DCS D. Majumder

ROTATING EQUIPMENT/STANDARDS What is new in API 610 11th Ed. (ISO 13709 2nd Ed.)?

HPIMPACT

This case study proves that a systematic approach for thermal design can identify substantial savings in projects M. Mandal

PROCESS CONTROL AND INFORMATION SYSTEMS Implementing advanced process control on ammonia plants

Cover Fluor and its joint venture partner provided engineering, procurement and construction management services to Tengizchevroil for its recent production expansion projects at the giant Tengiz field in Kazakhstan, copyright © 2010 Tengizchevroil. Courtesy: Flour.

Updates to global specifications address pump reliability and much more F. Korkowski, R. L. Jones and J. D. Sanders

INSTRUMENTS AND NETWORKS Development of support vector regression-based soft sensor

Application was used in a commercial ethylene glycol plant S. K. Lahiri, S. Sawke, N. Khalfe and J. Al Ghamdi

DEPARTMENTS 7 HPIN BRIEF • 23 HPIN CONSTRUCTION 29 HPI CONSTRUCTION BOXSCORE UPDATE 82 HPI MARKETPLACE • 85 ADVERTISER INDEX

COLUMNS 9 HPIN RELIABILITY Finding answers to library and training dilemmas 11 HPINTEGRATION STRATEGIES 3D-design tools help EPCs and owneroperators alike 13 HPIN ASSOCIATIONS Growing the fuel pool 15 HPI VIEWPOINT Advanced process control: Optimization or control 86 HPIN CONTROL Process control practice renewal— consequences

www.HydrocarbonProcessing.com Houston Office: 2 Greenway Plaza, Suite 1020, Houston, Texas, 77046 USA Mailing Address: P. O. Box 2608, Houston, Texas 77252-2608, USA Phone: +1 (713) 529-4301, Fax: +1 (713) 520-4433 E-mail: editorial@HydrocarbonProcessing.com www.HydrocarbonProcessing.com Publisher Bill Wageneck bill.wageneck@gulfpub.com EDITORIAL Executive Editor Stephany Romanow Process Editor Tricia Crossey Reliability/Equipment Editor Heinz P. Bloch News Editor Billy Thinnes Associate Editor Helen Meche European Editor Tim Lloyd Wright Contributing Editor Loraine A. Huchler Contributing Editor William M. Goble Contributing Editor Y. Zak Friedman Contributing Editor ARC Advisory Group (various)

Because Hydrocarbon Processing is edited specifically to be of greatest value to people working in this specialized business, subscriptions are restricted to those engaged in the hydrocarbon processing industry, or service and supply company personnel connected thereto. Hydrocarbon Processing is indexed by Applied Science & Technology Index, by Chemical Abstracts and by Engineering Index Inc. Microfilm copies available through University Microfilms, International, Ann Arbor, Mich. The full text of Hydrocarbon Processing is also available in electronic versions of the Business Periodicals Index. ARTICLE REPRINTS If you would like to have a recent article reprinted for an upcoming conference or for use as a marketing tool, contact Foster Printing Company for a price quote. Articles are reprinted on quality stock with advertisements removed; options are available for covers and turnaround times. Our minimum order is a quantity of 100. For more information about article reprints, call Rhonda Brown with Foster Printing Company at +1 (866) 879-9144 ext 194 or e-mail rhondab@FosterPrinting.com.

MAGAZINE PRODUCTION Director—Editorial Production Sheryl Stone Manager—Editorial Production Angela Bathe Artist/Illustrator David Weeks Manager—Advertising Production Cheryl Willis

HYDROCARBON PROCESSING (ISSN 0018-8190) is published monthly by Gulf Publishing Co., 2 Greenway Plaza, Suite 1020, Houston, Texas 77046. Periodicals postage paid at Houston, Texas, and at additional mailing office. POSTMASTER: Send address changes to Hydrocarbon Processing, P.O. Box 2608, Houston, Texas 77252. Copyright © 2010 by Gulf Publishing Co. All rights reserved. Permission is granted by the copyright owner to libraries and others registered with the Copyright Clearance Center (CCC) to photocopy any articles herein for the base fee of $3 per copy per page. Payment should be sent directly to the CCC, 21 Congress St., Salem, Mass. 01970. Copying for other than personal or internal reference use without express permission is prohibited. Requests for special permission or bulk orders should be addressed to the Editor. ISSN 0018-8190/01. www.HydrocarbonProcessing.com

ADVERTISING SALES See Sales Offices page 84. CIRCULATION +1 (713) 520-4440 Director—Circulation Suzanne McGehee E-mail: circulation@gulfpub.com

GULF PUBLISHING COMPANY

SUBSCRIPTIONS Subscription price (includes both print and digital versions): United States and Canada, one year $199, two years $349, three years $469. Outside USA and Canada, one year $239, two years $407, three years $530, digital format one year $140. Airmail rate outside North America $175 additional a year. Single copies $25, prepaid.

John Royall, President/CEO Ron Higgins, Vice President Pamela Harvey, Business Finance Manager Part of Euromoney Institutional Investor PLC. Other energy group titles include: World Oil® Petroleum Economist Publication Agreement Number 40034765

Printed in U.S.A

PERFORMANCE

NOT PROMISES BEGEMANN®

Centrifugal Process Pumps

Precision Engineered for API 610 Delivering flawless performance in the demanding petrochemical environment requires a product that you can trust to deliver reliable and safe performance under the toughest conditions. With 40 years of experience doing exactly that, the Begemann range of centrifugal pumps delivers where it matters.

Weir Minerals France - Europarc du Chêne - 11 rue Pascal 69500 BRON France - T: 04.72.81.72.72 - F: 04.72.81.76.43 www.weirminerals.com - begemann@weirminerals.com

I DECEMBER 2010 HYDROCARBON PROCESSING

Select 151 at www.HydrocarbonProcessing.com/RS

Excellent Minerals Solutions

YOUR GLOBAL GASKET PROVIDER

AFTER

450°C

IT’S ADVANTAGE THERMICULITE . ®

Graphite oxidizes at high temps. So gaskets made with graphite ®

deteriorate as well. Thermiculite , the revolutionary sealing material from Flexitallic maintains its integrity up to 982°C. Preventing leakage and the loss of bolt load that can be so costly—and ultimately dangerous. Replace your graphite gaskets. Because when the heats on, graphite can’t serve. Visit: www.ﬂexitallic.com, or call us at USA: 1.281.604.2400; UK: +44(0) 1274 851273.

Select 93 at www.HydrocarbonProcessing.com/RS

Süd-Chemie Defining the Future V Conference May 23-25, 2011 – Beijing, China Industry executives, technical experts and analysts will be discussing the latest innovations and breakthroughs in catalysis, process engineering, battery materials and water treatment technologies applied in these fields at the “Defining the Future V Conference.” Through open exchange and in-depth discussion, it is our goal to offer valuable insight into opportunities and challenges the industries are facing, and to support the formulation of strategies that can be developed to address them. This conference will be comprised of parallel sessions on:

• Reﬁning Industries • Chemical Industries • Petrochemical Industries & Polyoleﬁn Plants • Coal-To-Chemicals (Methanol & Derivatives) • Environmental Technologies • Battery Materials • Water Treatment Technologies Please mark your calendar and join us in Beijing!

SÜD-CHEMIE patricia.hesse@sud-chemie.com www.sud-chemie.com Select 90 at www.HydrocarbonProcessing.com/RS

HPIN BRIEF BILLY THINNES, NEWS EDITOR

BT@HydrocarbonProcessing.com

Cobalt Technologies has an agreement with the US Navy to develop technology for the conversion of biobutanol into full performance jet and diesel fuels. Under the agreement, n-biobutanol produced by Cobalt will be converted to biojet and biodiesel fuels using technology developed at the US Naval Air Warfare Center Weapons Division (NAWCWD) in China Lake, California. The result will be a complete substitute for military and civilian jet fuel, meeting all applicable specifications. In addition, Cobalt will have an option to obtain an exclusive license to commercialize process improvements for the production of all military and civilian transportation fuels.

Shell has agreed to sell the majority of its refining and marketing businesses in Finland and Sweden to Keele Oy. Keele Oy is the major shareholder of St1 Holding Oy, whose businesses include fuel retail networks in Finland, Sweden, Norway and Poland. The terms of the transaction, which are subject to regulatory approvals, include Shell’s retail business, including some 340 service stations in Sweden and some 225 in Finland, as well as its commercial road transport (CRT) in both markets. All service stations, together with the CRT business, will remain Shell-branded in both markets under a licensing agreement. Also included is Shell’s 87,000-bpd Gothenburg refinery, Shell’s bulk fuels business in both markets and the Shell marine business in Sweden. The businesses will be sold as going concerns and Shell will receive a total cash payment of $640 million.

Repsol and KUO Group signed an agreement in Madrid that creates a joint company called Kuosol dedicated to the development of bioenergy from the cultivation of the jatropha curcas, an oilseed with a high content of nonedible oil. Both Repsol and the KUO Group will have a 50% stake in Kuosol; its headquarters will be in Mexico and the total investment is estimated at $80 million. Its activities range from farming to industrial installations, and its main objective is the use of integrated biomass plantations of jatropha curcas oil to generate biofuels and bioenergy in a sustainable manner. It is estimated that agricultural development will be completed in the next three years, allowing industrial production to start in 2013.

OSHA issued its annual inspection plan under the Site-Specific Targeting 2010 (SST-10) program to help the agency direct enforcement resources to high-hazard workplaces where the highest rates of injuries and illnesses occur. The SST program is OSHA’s main programmed inspection plan for non-construction workplaces that have 40 or more workers. This inspection plan is based on workrelated injury and illness data collected from a 2009 OSHA Data Initiative survey from 80,000 larger establishments in selected high-hazard industries. Establishments are randomly selected for inspection from an initial list of 4,100 manufacturing, nonmanufacturing, and nursing and personal-care facilities. The plan focuses on several variables, such as the number of injury and illness cases and the number of days a worker has to stay away from work, or the number of workers who received job transfers or work restrictions due to injury or illness.

Air Products has completed further expansion of its Asia Technology Center in the Zhangjiang Hi-Tech Park in Shanghai, China. The expansion will accelerate the development of industrial gas applications and solutions for general industries to support increasing demands in these high growth markets in China and across Asia. This expansion, built on the existing capabilities serving high growth performance materials segments including polyurethane chemicals, and epoxy and specialty additives for coatings, inks, adhesives and related markets, adds standard laboratories and high-bay space for industrial gas applications, product and process R&D, and analytical and testing capabilities for other key growth markets. These include metals processing, electronics packaging and assembly, industrial cryogenics, water treatment and energy applications. HP

■ Converting crude ethanol into bio-alkylate Exelus, Inc., has developed a process that converts low-cost, bio-derived, hydrous ethanol into “bio-alkylate,” a hydrocarbon fuel chemically identical to a conventional refinery blendstock called alkylate. This technology would bridge the production of petroleumderived fuels and biofuels, effectively incorporating the energy and carbon content of bio-ethanol into conventional gasoline. The Environmental Protection Agency has recently agreed to let refiners add as much as 15% ethanol to a new blend, up from the current 10%. Conventional vehicles cannot safely use fuels with significantly higher concentrations of ethanol without retrofits to replace gaskets and metal components susceptible to chemical attack by ethanol. Few gas pumps are certified as safe to use E15. Bio-alkylation offers a simple way around this problem while also addressing the concerns of Congress that higher ethanol blends could damage engines. The bifurcation of the gasoline supply into E15+, only used by newer vehicles, and E10, used by all other vehicles and motorized equipment, is also avoided. This process allows the renewable content of gasoline to reach 50% without any changes to the nation’s engine and fuel infrastructure. Most of the energy consumed in a conventional bio-ethanol plant is used to distill ethanol from water. The Exelus process uses wet ethanol as the feed, providing a 32% energy savings compared to anhydrous ethanol production. The fuel produced can be blended with gasoline in any proportion without altering the physical or combustion characteristics of the gasoline. This technology removes many of the technical barriers to the increased use of renewable fuels in the gasoline pool. The bio-alkylation process utilizes research funded partly by a $1 million grant from the US Department of Energy’s ARPA-E program. HP

HYDROCARBON PROCESSING DECEMBER 2010

GE Oil & Gas

Heavy duty refinery and petrochemical applications require heavy duty solutions. Our DDHF pumps are specifically designed to handle high pressure, elevated temperatures and low specific gravity fluids, delivering exceptional performance in extreme environments where pressures could reach up 500 bar and temperatures hit 450°C. Essential features include internal volute casing and a back to back impeller configuration, simplifying maintenance, reducing downtime and guaranteeing pump reliability. Widely used in engineered applications such as hydro cracking processes, reactor feeding, ammonia and urea and boiler feed services, our pumps set the standard for robust and durable solutions. geoilandgas.com

Extreme environments: reliable performance

This is Innovation Now

Select 64 at www.HydrocarbonProcessing.com/RS

HPIN RELIABILITY HEINZ P. BLOCH, RELIABILITY/EQUIPMENT EDITOR HB@HydrocarbonProcessing.com

Finding answers to library and training dilemmas We don’t mind receiving and answering an occasional question. Questions allow us to assess the state of knowledge in industry. Sometimes, though, these questions indicate that the questioners have not availed themselves of all the tools at their disposal. For reliability professionals to not insist on the conscientious pursuit of doing the right things and doing things right is a negative reflection on their diligence and competence. Their tolerance for repeat failures can boggle the mind. We have even seen claims that a particular equipment problem is unique and has never been experienced elsewhere. The statistical improbability of a petrochemical plant or refinery experiencing repeat failures due to phenomena never experienced elsewhere is staggering. It follows that troubleshooters or reliability professionals will find the solution to their problems in the literature to which we all have access. Literature? What literature? The principal engineer of a major design contracting firm at a US West Coast location commented on literature issues in a recent letter. He regretted the lack of people’s efforts to find answers by exerting themselves. When engineers have questions, they should understand and research an issue before depending totally on input from others. This principal engineer observed that, when we leave an engineering school, we have—hopefully—a grasp of the basics of certain engineering fundamentals. Initially, none of us have the experience, equipment or industry-specific knowledge to do much useful work. Being able to teach ourselves as needed is the key element for doing useful work. But he points out that teaching ourselves requires the attitude, motivation and willingness to work. Study is work. Study also qualifies for receiving support— mentoring—in the workplace. Moreover, study encompasses giving support and acting as a mentor. Current social and management trends cut away at both ends of the spectrum. Social trends make sustained work on a single subject less likely. If a subject can’t be broken down into 30-second segments it’s often viewed as too hard. It’s difficult to get attitudes reset and, with some people, one never succeeds in resetting attitudes. But the employer has to make a conscious effort to nurture changing attitudes and must be willing to subsidize such small steps as attending local ASME meetings, or by simply requesting a free listing of our Essential Reliability Library from hpbloch@ mchsi.com. Comments on management trends. While social trends

are disturbing to some readers, management trends are distressing to others. One reader noted that it had been more than 20 years since he had visited a plant site or major engineering office with more than a mediocre collection of reference material that covered basic engineering needs. In fact, most sites he visited no longer had a reference library. This particular reader contended that if it

could not be found on the Internet, information was treated as if it didn’t exist. Some companies avail themselves of a “pay-for-use” service. A pay-for-use resource can heighten an employees’ job satisfaction because it makes retrieving data less tedious and engineers can get to the problem-solving stage (the fun part) more quickly. A few of the more enlightened companies purchase subscriptions to engineering libraries such as Knovel. But while using an online library is a legitimate cost-saving method, companies using it are in the minority. Of course, we want to give full credit to individuals in different industries who managed to collect excellent personal libraries at their own expense. Indeed, having one’s own collection of reference texts shows dedication to technical excellence. Assembling such personal reference libraries sometimes becomes the way we work because we are not given other choices. We are of the opinion that employers would do well to create alternatives to pure self-teaching and insistence on personal reference libraries. Reasonable alternatives involve access to other resources and development through training. Training is far more important than we think. Training is an easy budget item to cut in hard times, and the refining and chemical processing industries are still experiencing hard times. Both training seminars and conference participation are being cut across the industry. This saves a little money now, but costs much more as time progresses. It should be recognized that there still are zero-cost, highly effective training methods and these should never be abandoned. Again, one reader advised that creative and determined individuals can do a lot to help themselves. All that’s required is the right attitude, and finding small, local opportunities such as attending local ASME or AIChE chapter meetings does not cost much. Look to networking with other people in similar situations, organizing inhouse shirt-sleeve seminars, mandating the quick perusal of certain no-cost trade journals and working on keeping up-to-date. Lack of self-help because an individual persists in harboring the wrong attitude is relatively rare. When it does occur, it is —hopefully—confined to an individual. But management failure is more insidious and, ultimately, much more serious. Management failure will inevitably include both lack of effort in improving attitudes across the board as well as starving the organization of resources to support those with good attitudes. That’s when an entire organization is on the path to failure. HP The author is Hydrocarbon Processing’s Equipment/Reliability Editor. He is the author of 17 textbooks and over 470 papers or articles, and he advises process plants worldwide on reliability improvement and maintenance cost-reduction opportunities. He acknowledges Andrew Sloley for contributing his views here. Mr. Bloch can be contacted at hpbloch@mchsi.com. HYDROCARBON PROCESSING DECEMBER 2010

GE Power & Water Water & Process Technologies

Meeting your challenges head on As the global economy slowly recovers, refiners can expect to see a positive shift in the demand for finished petroleum products in mature markets and growing, developing regions. Now more than ever, GE’s advanced treatment technologies, monitoring tools, and unmatched domain expertise for the hydrocarbon process industries provide a variety of solutions to help our customers find new ways to solve their toughest challenges. For more information, contact your local GE representative or visit www.gewater.com

Phase Separation

Corrosion Inhibitors

Embreak* demulsiﬁer technology improves desalter performance and efﬁciency, maximize crude throughput and reduce fuel gas consumption.

LoSALT* and pHilmPLUS* minimize corrosion in critical production units and extend equipment life, maintaining throughput and increase operational ﬂexibility.

Antifoulants

Finished Fuels

Our Thermoﬂo* chemistry, engineering expertise, and Heat-Rate Pro software provide a state-of the-art antifoulant treatment program.

ProSweet*, SpecAid* and ActNow* products help ensure refined fuels and other hydrocarbons meet required specifications and improve final product quality.

* Denotes a trademark of General Electric Company.

Select 72 at www.HydrocarbonProcessing.com/RS

HPINTEGRATION STRATEGIES DICK SLANSKY, CONTRIBUTING EDITOR dslansky@arcweb.com

3D-design tools help EPCs and owner-operators alike The wide diversity of energy market requirements, potential power sources and enabling technologies drives strong demand for design/build/operate solutions for projects in the hydrocarbon processing industry (HPI) and other energy sectors. These projects require engineering design solutions that span multiple engineering disciplines and applications, from plant and facility design, process simulation, equipment design, mechanical, electrical and controls design, collaborative engineering, data model management and version control, to document and drawing management. As projects become larger and more complex, engineering, procurement and construction (EPC) companies have become much more dependent upon their 3D-design tools. EPC companies use a wide variety of engineering design and data management applications from design/build solution providers. This very heterogeneous environment presents significant challenges for the EPCs using such a range of engineering design tools, with multiple formats, models and configuration management frameworks. Additionally, since the lion’s share of today’s EPC business is conducted on a global scale with a variety of stakeholders ranging from owner/operators to extended supply chains for equipment mandates, a company’s design engineering solutions must be both open and highly collaborative. 3D-design tools are essential to EPC projects. Present

design solutions offer a broad range of design/build capabilities. For some time, suppliers have based design applications on 3D modeling. These provide the EPCs with the ability to model in 3D space. In HPI projects, 3D modeling is a critical aspect of design work, which has become essential for structural design, and for configuring all the equipment and infrastructure internal to the facility. 3D modeling provides the most effective means to do space control, and most EPC tools provide automation features that reduce the labor required for laying out, detailing and revising infrastructure such as pipe and raceway. Additionally, these 3D-design tools perform clash analysis to determine interferences between structures, equipment and infrastructure. Designing in 3D space provides the capability to visualize and integrate multiple engineering disciplines, plus the flexibility to make configuration and structural changes throughout the design process. As projects grow in size and complexity, especially in energy-related projects, EPCs have become dependent upon their 3D-design tools. Moreover, data and model management, project management, engineering change management and design collaboration across multi-discipline engineering organizations have become essential to the overall design/build process. Managing the mass of information generated by design/build requirements for major projects represents a primary challenge. Better data management and collaboration. Infor-

mation management and interoperability of models and data are

critical to doing business and managing large projects efficiently and successfully. EPCs agree that interoperability should be standards based, as in ISO 15926. Major plant design software providers support this standard for data integration, sharing, exchange and hand-over between computer systems, is supported by major plant design software providers. While the current design solution providers offer robust building information modeling/management (BIM) systems, EPCs would like to see better data management and design collaboration capabilities across all their various engineering organizations and equipment providers. This is why most EPC design/build processes still involve a mixture of commercial software and inhouse applications, which can include data management. While EPC engineering organizations recognize the need for improved data management and collaboration, they are reluctant to rely on their design software providers to handle all the specialized in-house applications, data formats and hand over information that their customers demand. Owner-operators have similar data management issues. Information management is as important to the owner-

operators as it is to the EPCs. Most owner-operators consider the handover process (where all of the plant/facility drawings, layouts, equipment and infrastructure information are handed over to the operating organization) one of the most critical aspects of the project. Not only is this engineering information essential to the operation of the facility, it must be managed and organized properly so that it can be readily accessed and available to the operations personnel during plant life. The owner-operators clearly have the need for the same information management and collaborative engineering platforms that the EPCs obtain from their engineering design software suppliers. Currently, most owner-operators do not demand a 3D-design model from the EPC at project completion. Nearly all want the asset information turned over in a form that they can import into operations/asset management/maintenance management systems that they use to operate and maintain the facility. This generally takes the form of 2D drawings and other equipment/asset da-tabase populations that operations personnel are able to use without the specialized training needed for a 3D-design environment. Owner-operators are beginning to show interest in 3D-virtual simulation, which they can use for training and safety incident response and mitigation. Virtual simulation tools are also beginning to be Larry O’Brien is part of the automation at ARC covering the applied to the pre-construction phaseconsulting of new team projects where simuprocess industries, and an HP contributing editor. He is responsible for tracking the lation can detect interferences and streamline erection activities. HP market for process automation systems (PASs) and has authored the PAS market studies forauthor, ARC sincea 1998. O’Brien has Advisory also authored many other research, The seniorMr. analyst at ARC Group, focuses onmarket PLM and digital strategy and custom research including process fieldbus, collaborative manufacturing. He holds a BSreports degreeonintopics mechanical engineering from the University partnerships, total automation trends and others. He hasPacific been with ARC since of Kansas and a BS degree in market computer science from Seattle University. He January 1993, and started career withproject marketmanagement research in the fieldthe instrumentation also holds a certificate in his client/server from University of markets. Washington in Seattle. HYDROCARBON PROCESSING DECEMBER 2010

I 11

We put the best hands in the business to work on your project. When it comes to turnarounds, no one can beat the loyalty, dedication to quality and pure craftsmanship of our “hands” in the field. Through the years AltairStrickland has developed a following of skilled workers. Many have worked on the same projects together for a decade or more. Their familiarity with one another’s abilities, talents and experience means a more cohesive effort and efficient project execution for you.

ADDRESSING YOUR NEEDS… By combining the proficiency of our hands with the expertise of our field supervisors and management, we can bring you a team of professionals that will work hand-in-glove with you to competently address the changing scope, problems and pressures inherent in every turnaround.

EMPLOYEE BENEFITS… We also provide a variety of insurance programs and retirement benefits, such as 401-K plans, to all our eligible employees and offer many opportunities for training and advancement.

SAFETY…

AltairStrickland 1605 South Battleground Road La Porte, TX 77571 Call 281-478-6200 ■ 1-800-478-6206 www.altairstrickland.com

And, when it comes to safety, it is our job to help our hands protect theirs. Our safety teams review the planned work scope to assure that safety is integral to the process. All AltairStrickland employees receive extensive safety training. We utilize third party safety services to assure objective safety observance and reporting.

Call us for your next turnaround, revamp, upgrade, repair or emergency; we’ll put the best hands (and minds) in the business to work for you.

Select 51 at www.HydrocarbonProcessing.com/RS

HPIN ASSOCIATIONS BILLY THINNES, NEWS EDITOR

BT@HydrocarbonProcessing.com

Growing the fuel pool Whether refiners like it or not, the drumbeat for renewable fuels grows daily. The building momentum for such fuels is boosted by government mandates and regulations. Unfortunately, the tidal wave for renewables is barely hindered by rational objections pointing out that these new fuels have low BTU content and questionable greenhouse gas (GHG) emissions savings when compared to petroleum. Still, renewables are the future and it is important for the refining industry to understand them and read the tea leaves governing where renewable fuels standards will be taking us in the future. The National Petrochemical and Refiners Association (NPRA) grappled with these and other topics during its annual October gathering, the Q&A and Technology Forum. Hosted in Baltimore, Maryland, this year, the conference analyzed many technical matters in great detail. One of the keynote speakers, Dr. Jennifer Holmgren, CEO of Lanzatech Ltd., offered an in depth overview of renewable fuels and the need for growing the fuel pool. “By 2030, 30% of the fuel pool needs to be at zero-carbon fuel,” Dr. Holmgren said. “When you look at the driver for low-carbon fuel, how do you get there from here?” Several drivers for alternative fuels include energy security, the desire to mitigate GHGs and rural job creation (because it is necessary to plant, grow and harvest these fuels, jobs in rural America are created). The Energy Independence and Security Act (EISA), signed into law in December 2007, significantly increased required volumes of renewable fuel. “We went from 7 billion by 2012 to 36 billion gallons by 2022,” she said. The legislation separates the volume requirements into four separate categories of renewable fuel: cellulosic biofuel, biomass-based diesel, advanced biofuels and total renewable fuel. Out of these categories, there are different production requirements for each. Biomass-based diesel is

expected to achieve 1 billion gallons by 2012 and beyond. Biodiesel is considered “renewable diesel” if not co-processed with petroleum. Meanwhile, cellulosic biofuel is looking at a required 16 billion gallons by 2022. This second edition of renewable fuel standards also mandates that 21 billion gallons must be totally advanced biofuel by 2022. The ethanol blend wall. Previous analyses were based on an assumption of 36 billion gallons of ethanol by 2022, to be used as E10 and E85 (E85 would be in very small volumes). By 2013, the EPA expects that all gasoline in the US will be E10, with 14–14.5 billion gallons of ethanol being used in E10 form. Basically, ethanol demand is maxed out at the current 10% blend rate and production has hit a ceiling. So, unless either gas demand increases or the blend rate goes up, there’s just no need for any more ethanol at the pump. However, some at the USDA are advocating for increasing the blend rate and are trying to convince the EPA to up the rate to 15%. GHG emissions. Lifecycle GHG analysis is integral to the new renewable fuel standards. This analysis is used to categorize fuels, not to value them. “The term life-cycle GHG emissions means the aggregate quantity of GHG emissions as determined by the administration, related to the full fuel life-cycle,” Dr. Holmgren said. “The mass values used to account for all GHGs are adjusted to account for their relative global warming potential.” The life-cycle GHG emissions classification is a big issue for plant-based renewables that require a lot of fertilizers. Under this school of thought, nitrogen is more harmful to the atmosphere than CO2 and, therefore, fuels that utilize nitrogen fertilizer to create them are weighted more heavily. The EPA is currently mulling over how best to do these analyses. “Forests are the lungs of the planet,” Dr. Holmgren said. “If you end up chang-

ing the use of the land, like by planting soybeans on what was once a forest, you are losing lots and lots of carbon-reduction possibility.” Emphasizing her point, Dr. Holmgren noted that, depending on the impact of global deforestation, it will take up to 40 years of biofuels to make up the ground that has already theoretically been lost. The whole forests-to-soybeans discussion is an example of what is called indirect land use change. But indirect land use change is a tricky concept. For instance, if one is using a soy field to make a biofuel, the classification might be okay because the field in question was always a soy field. The problem is if the soy was exported and used as food and it is now being used as fuel, food consumers are affected by indirect land change. Production efficiencies still need to go up considerably for biofuels to be viable. One example that Dr. Holmgren offered was that one acre of land produces 80 gallons a year of soybean-based diesel. Ethanol numbers are a little higher, but still, you are talking about 600 gallons per acre and when you adjust for the BTU content you could end up about the same. So, the productivities are not huge. “You’ll end up having to plant entire continents to meet the mandate,” she said. HP

Dr. Jennifer Holmgren, CEO of Lanzatech Ltd., was a keynote speaker during the NPRA Q&A and Technology Forum in Baltimore.

HYDROCARBON PROCESSING DECEMBER 2010

I 13

The Best Compressors for Productivity… and the Environment When you need the optimal solution for your gas compression application, look to Kobelco. We offer all types of compressors, so we can custom-engineer the best possible combination of reliability, efficiency, economy and environmental benefits. Whether you need the operating cost savings of a screw compressor, the large volume of a centrifugal compressor or the high efficiency of a reciprocating compressor – we’re the ones to call.

Ask Kobelco! The Best Solution for Any Gas Compression. Kobelco EDTI Compressors, Inc. Tokyo +81-3-5739-6771 Munich +49-89-242-18424 www.kobelco.co.jp/compressor

Houston, Texas +1-713-655-0015 rotating@kobelcoedti.com www.kobelcoedti.com

Select 80 at www.HydrocarbonProcessing.com/RS

HPI VIEWPOINT ALLAN KERN, GUEST COLUMNIST Allan.Kern@yahoo.com

Advanced process control: Optimization or control After 20 years of advanced process control (APC) praise, a reckoning is now taking shape, with some challenging questions, such as: What happened to the original vision of “true real-time optimization”? Why is sustained performance proving so intractable? What are the implications when a major APC supplier suggests there are more poor-performing APCs in industry than well-performing ones?1 Continued APC progress demands a clearer industry consensus on the role APC can or should play in process control and optimization applications. It’s an appropriate time to evaluate the lessons learned and make a course correction, before moving ahead with renewed purpose.

weekly meetings, and regular plans and schedules, with phone calls, e-mails and conversations adding necessary fluidity. Within this process, equipment health is monitored and, in conjunction with production and maintenance planning, “optimum” operating points are arrived at, with a keen eye to meeting production targets and protecting process up-time. The lion’s share of optimization value resides in this cloud, centered around planning, and for this process to work, the plants need to operate reliably and reject disturbances intelligently. That is where process control can add important value, rather than by trying single-handedly to wrest another percentage point from the process.

Why has optimization proved unsuitable in APC?

What are the lessons learned so far? If answers to APC shortfalls are in the real nature of industrial plant operations, rather than in quality of support, that raises questions about the emerging strategy of engaging skilled consultants and deploying performance monitoring software to address the shortfalls. It might be throwing good money after bad if the lesson learned is that realism factors limit APC success, not lack of monitoring and support. Instead, the lessons suggest that changes are needed in the upfront APC selection and design process, to avoid poorly performing and unnecessary applications in the first place. Two likely lessons stand out at this juncture. One is that APC applications need to be more selective and smaller (in matrix size) to reflect “real” multivariable control opportunities. The practice of putting a big matrix on every process on the assumption that it will find the benefits, it turns out, is decidedly not best practice. A second lesson is that much of what APC does realistically accomplish can be done more easily at the distributive control systems (DCS) level, without the costs, risks and headaches of APC. Constraint overrides, inferential control, and smart flexible control strategies are the heart of DCS capabilities. The “outsidethe-box” manager, wrestling with limited budgets, might try putting APC aside and refocusing on the DCS. And they might be pleasantly surprised. HP

Zak Friedman writes that it’s due to unmanageability of fast changing economics.2 That is one factor. But a more fundamental cause of APC under pressure is that, while elegant and pleasing on an academic or simulation level, APC runs into many limitations in a real-world process control. For example, in an actual plant operation, it is usually only practical to use each manipulated variable (MV) to control one, occasionally two, or very rarely three controllable variables (CVs), regardless of the number of models in the matrix row. Where an APC theorist sees a large matrix with many degrees of freedom, in practice many of these models are unsuitable for control and soon become defeated by “clamping” limits. Similarly, APC designers tend to include many constraints for which operations only wants (and already has) alarms, not control action. Many heater constraints fall into this category—when skin temperature or burner pressure is high, operations wants an alarm to check burners and flames, not to automatically reduce process temperature. A third example is that operations personnel often quietly resist moving key MVs. This is often mistaken for lack of boldness or understanding on their parts, but it is more often a lack of understanding on the APC side. Usually, the equipment is already running at one or more limits, having been optimized by the more inclusive “cloud” process long before APC ever arrived on the scene. Realism factors like these are typical and commonly reduce a large matrix, in practice, to a small handful of control overrides. This goes far in explaining under-performance, low MV utilization, the haphazard(ous) practice of frequent limit adjustment and the operational confusion that results (in part) from big matrix controllers with more variables defeated (one way or another) than in action. If not in APC, then where? To borrow an Internet term,

optimization takes place offline, in the refinery “cloud.” It is a collaborative process including engineers, planners and operations. It includes formal and informal work processes, daily and

1 2

LITERATURE CITED Jubien, G., “Successful APC: Design and Maintain for Long-Term Benefits,” Honeywell Process Solutions, ISA 2009. Friedman, Y. Z., “APC Application Ownership,” Hydrocarbon Processing, September 2010.

Tim Lloyd Wright is HP’s European Editorexperience and has been as a reporter The author has 30 years of process control and active has authored many

and conference chairprocess in thecontrol, European downstream industry since 1997,systems, before papers on advanced expert systems and decision support which he was aonfeature writer and reporter for the UK broadsheet pressMr. and BBC with emphasis operation and practical process control effectiveness. Kern is radio. Mr. Wright lives in aSweden is founder a local climate and sustainability a professional engineer, senior and member of ISA,ofand a graduate of the University initiative. of Wyoming.

HYDROCARBON PROCESSING DECEMBER 2010

I 15

A fruitful partnership. To help secure future food supplies, Uhdeâ&#x20AC;&#x2DC;s engineers develop large-scale plants for the fertiliser industry. As a leading EPC contractor, we also have a proprietary portfolio of technologies. And we network intelligently within the Uhde group based on our business philosophy Engineering with ideas. Select 70 at www.HydrocarbonProcessing.com/RS

Uhde www.uhde.eu

HPIMPACT BILLY THINNES, NEWS EDITOR

BT@HydrocarbonProcessing.com

European downstream oil industry safety performance

Fieldbus Foundation registers devices with advanced field diagnostics

CONCAWE has released a report that parses statistics on work-related personal injuries for the European downstream oil industry’s own employees, as well as contractors, for the year 2009. Data were received from 33 companies representing more than 97% of the European refining capacity. CONCAWE has been studying this subject for 17 years, dating back to 1993. The most important finding of the report is that 11 fatalities occurred in 2009. Following a steady downward trend during the 1990s, fatality numbers began to increase in the first year of this decade. CONCAWE analyzed statistics from 2004–2006 and they revealed a reversal of the upswing and the fatality numbers have shown little variation since then. Marketing contractors continue to be the most vulnerable work group. The fatal accident rate (FAR) continues to be at a level similar to that observed in the late 1990s. CONCAWE’s research suggests that all injury frequency (AIF) peaked around 1996–1997, but this is likely the result of improved reporting standards. The group contends that the trend is definitely on a downward slope and AIF figures have improved for all categories. Road traffic accidents were clearly reduced compared to the early years, but the rate appears to have now reached a plateau. These accidents essentially occur in the refined-products marketing activity where the bulk of the driving takes place. Fig. 1 details the causes for the 11 fatalities in 2009. Fig. 2 shows the percentage of the main causes over the last five years. Fig. 3 provides an overview of information on this subject compiled since 1998. In 2009, three fatalities occurred in one incident due to confined space operation, three were due to road accidents and two were as a result of falling from height. Of the remaining three, one was a result of construction/maintenance activities, one from burning/electrocution and the last from other industrial activities. For the last five-year period, construction/ maintenance/operations activities and road accidents remain the principal causes of fatalities.

The Fieldbus Foundation has registered the first FOUNDATION fieldbus devices incorporating advanced field diagnostics technology. The new registration requirements help to standardize how fieldbus devices communicate their diagnostic data to the host and asset management tools within a plant automation system. Advancements in field diagnostics support a structured approach to asset management, which simplifies plant operaRoad accidents 60%

40% Oper./maint./const. 20%

Others

Third party action

FIG. 2

Causes of fatalities from 2004–2009.

Road accidents 60%

40%

Road accident 3

Others

2 Conﬁned space

Burn/ﬁre/explosion

Source: CONCAWE

Oper./maint./const.

20%

Falls 1

Burn/electrical

Source: CONCAWE

FIG. 1

Construction/ maintenance

Others

Causes of fatalities in 2009.

Manufacturing Marketing

Third party action

Burn/Fire/explosion

Source: CONCAWE

FIG. 3

Causes of fatalities from 1998–2009.

HYDROCARBON PROCESSING DECEMBER 2010

I 17

HPIMPACT tors’ tasks and increases their confidence in utilizing equipment diagnostics and asset software. This, in turn, will enable improved process performance, greater reliability, increased uptime and lower operating costs. Yokogawa (field indicator) and Fluid Components International (thermal mass flowmeter) are the first FOUNDATION fieldbus H1 (31.25 kbit/s) device suppliers to pass the field diagnostics registration process. The Fieldbus Foundation developed profile specifications enhancing the organization and integration of device diagnostics within FOUNDATION fieldbus systems. The new diagnostic

profile includes a standard and open interface for reporting all device alarm conditions, and provides a means of categorizing alert conditions by severity. The technology facilitates routing of alerts to appropriate consoles based on user-selectable severity categories. In addition, it provides recommended corrective actions and detailed help, as well as an indication of the overall health of the device. The FOUNDATION fieldbus diagnostics profile specification (FF-912) was defined to allow any electronic device description (EDD)-based system to access and configure the diagnostics in fieldbus devices. The field diagnostics profile makes no changes to the existing FOUNDATION fieldbus stack specifications. However, the profile does introduce a new field diagnostic alert type. System updates will provide more extensive integration capabilities (such as Wizards for configuration) that will enhance diagnostics performance. Rather than introduce significant changes to the current FOUNDATION protocol, the new diagnostic profile specification builds upon the existing, powerful diagnostic capabilities of FOUNDATION fieldbus equipment, and at the same time, adds a greater degree of organization so field instruments can represent their diagnostics in a more consistent way. FOUNDATION fieldbus devices submitted for field diagnostics registration must pass interoperability test kit (ITK) test cases, which exercise the bit alerts generated for fail alarms, check alarms, off-specification alarms, and maintenance alarms. Devices also must support multi-bit alert reporting, as well as the new alert object designed for field diagnostic alarms. In addition, they must support new field diagnostics parameters in the resource block. “In the FOUNDATION fieldbus automation infrastructure, field diagnostics is a way of standardizing how all fieldbus devices communicate their diagnostic data to the host and asset management system —regardless of the vendor. This technology streamlines the way data is presented in order to take advantage of the rich diagThis bench top analyzer tops all others in its price range for nostic information available in FOUNfeatures and performance. It’s equipped with an intuitive user interface, full-color touch screen and on-board Windows XP DATION fieldbus devices,” said Stephen computer. Ethernet electronics that permit remote access for Mitschke, Fieldbus Foundation’s manager calibration, diagnostics or service support. Plus, the Phoenix II of products. “For end users, the largest has a large sample compartment that accommodates spinners benefit is that advanced field diagnostics and special holders yet requires little or no sample preparation. enables role-based diagnostics, meaning the It all adds up to the lowest cost of ownership, backed by right information is sent to the appropriAMETEK’s reputation for reliability and world class customer ate person when they need it. FOUNDAsupport. Visit: ametekpi.com TION technology has always utilized push diagnostics allowing the user to receive alerts much quicker, instead of the traditional method of requesting diagnostic information from devices. Field diagnostics technology will now enhance user control

Select 152 at www.HydrocarbonProcessing.com/RS

Results

Seven (7) ethane cracking furnaces supplied by Selas Fluid to Saudi Ethylene and Polyethylene Company (SEPC) on an EPC basis. Completed ahead of schedule and contributed over 1 million accident-free man hours to the entire project.

Selas Fluid has supplied innovative heater, furnace, and oxidation technologies to reļneries and petrochemical plants worldwide for more than ļve decades. As an approved supplier of reľnery heaters, petrochemical furnaces and oxidation/incineration technologies to major oil and gas companies, Selas Fluid is prepared to surpass your expectations. We have built a history of proven results, with thousands of installations, by providing continuous innovation, quality products and reliable customer support. Selas Fluid is a full-service partner to the reľning, petrochemical, and chemical industries worldwide. • • • •

Equipment supply Process design and engineering Modular construction Erection services

• • • •

Revamp, retroľt and upgrade Heater WellnessSM Program Start-up and operator training Spare parts and technical services

Select 96 at www.HydrocarbonProcessing.com/RS

Selas Fluid Subsidiary of The Linde Group

Headquarters: Five Sentry Parkway East • Blue Bell, PA 19422 USA • Tel: 610-832-8797 • Fax: 610-834-0473 Texas Ofļce: 16225 Park Ten Place • Suite 250 • Houston, TX 77084 USA • Tel: 281-717-9090 • Fax: 281-717-9091

www.selasĽuid.com sales@selasĽuid.com

HPIMPACT and distribution of messages between field devices and host/asset management systems. This will allow for faster response times as each message is presorted according to criticality, whether it is a process alarm or a maintenance alarm.” Yokogawa’s registered field indicator offers not only the standard functions of a field indicator, but also PID function block, link master and software download capabilities. It enables users to switch and display up to 16 indicated values for FOUNDATION fieldbus devices. No complex operation is needed in the field in order to observe the indicated values. A self-diagnostic function based on the NAMUR NE107 standard detects failures in the

ambient temperature limit, communications, and hardware such as the LCD and amplifier assembly. Fluid Components International’s registered thermal mass gas flowmeter is industrial process and plant-grade suitable for all air and gas flow measurement applications. It provides direct gas mass flow measurements, including flow rate, totalized flow and temperature; specialized versions also include pressure measurement. The meter has no moving parts to clean or maintain, and it is offered in a variety of process connections. The electronics/ transmitter can be integrally mounted with the flow sensor or remote mounted up to 1,000 ft from the sensor element. A complete list of registered Foundation fieldbus products is available on the Fieldbus Foundation’s website (www.fieldbus.org).

Wouldn’t it be great if everything was FRACTURE RESISTANT?

CeraComp™ Components

A single pump failure can disrupt your plant for days – bringing production to a halt and impacting your bottom line. Replacing silicon-carbide with CeraComp™ dramatically reduces your risk of pump failures! This new material offers the chemical and thermal resistance you require with the increased toughness and fracture resistance you’ve been looking for. CeraComp delivers increased reliability and MTBR (mean time between repair), eliminating catastrophic failure and keeping your plant running efficiently. Contact Greene, Tweed today to learn more about this exciting new material!

45678 Greene, Tweed & Co. | PetroChem & Power | Tel: +1.281.765.4500 Fax: +1.281.821.2696 | www.gtweed.com 11/10-GT AD-US-PP-004

Select 153 at www.HydrocarbonProcessing.com/RS

Ethanol to hit refining profitability Surging ethanol blending will undermine wholesale gasoline prices in the Atlantic Basin, starting in 2012, according to ESAI’s newly updated two-year outlook. The outlook says that ethanol will add supply to the gasoline market, especially in the US. As this additional volume meets organic demand growth, it will weaken gasoline and widen the disparity between gasoline and diesel prices on both sides of the Atlantic. ESAI believes that a new Atlantic Basin transport fuels market is emerging. Along with more refining capacity, ethanol blending in the US will increasingly expand domestic gasoline supply and reduce the region’s fuel import requirements. An import requirement that was above 1 million b/d in 2008 will collapse to below 400,000 b/d by 2012. At that point, gasoline will decline relative to diesel and stay that way. ESAI expects that New York Harbor diesel spreads to WTI will average $9/ barrel in 2012, whereas gasoline’s relationship to crude will decline to $7.70/barrel. While diesel-centered production will benefit, “gasoline-heavy producers will take a hit,” notes ESAI Principal Sander Cohan. “Europe will have a substantial gasoline surplus, but won’t be able to send it to usual markets.” This will likely encourage refiners there to trim back runs and reduce product output overall, making that region even more reliant on imports from Russia and elsewhere to meet demand. The influence of biofuels in Atlantic Basin fuel market dynamics is a relatively recent phenomenon, but is expected to accelerate and deepen in the coming years across the region. Consequently, refiners will come under continued pressure to adapt to this new market dynamic, especially as gasoline and diesel diverge. HP

AD: www.graficadueprint.com ÂŠ 2010 Costacurta S.p.A.-VICO

SINCE 1921... AND WE STILL LOVE IT For more than eighty years, we at Costacurta have been constantly and resolutely committed to the development and manufacture of special steel wire and plate components used in many different industrial processes. Every day at Costacurta, we work to improve the quality of our products and services and the safety of all our collaborators, paying ever-greater attention to the protection of the environment. Within the wide range of Costacurta products you will also find some, described below, that are used specifically in the oil, petrochemical and chemical industries: - RADIAL FLOW AND DOWN FLOW REACTOR INTERNALS; - GAS-LIQUID AND LIQUID-LIQUID SEPARATORS; - ARMOURING OF REFRACTORY, ANTI-ABRASIVE AND ANTI-CORROSIVE LININGS . For more information visit our website or contact the division 'C' components for the oil, petrochemical and chemical industries at tcrc@costacurta.it.

Costacurta S.p.A.-VICO via Grazioli, 30 20161 Milano, Italy tel. +39 02.66.20.20.66 fax: +39 02.66.20.20.99

Armouring of refractory, anti-abrasive and anti-corrosive linings

Management systems certified by LRQA: ISO 9001:2008 ISO 14001:2004 OHSAS 18001:2007 Select 57 at www.HydrocarbonProcessing.com/RS

www.costacurta.it

Serving you along the entire LNG Value Chain. Rely on our expertise in process engineering as well as in the planning, project development and execution of turnkey process plants. Proﬁt from our competence along the entire LNG Value Chain: – Process plants BUhifU` [Ug `]eiYZUWh]cb BUhifU` [Ug dfY!hfYUhaYbh B;@ YlhfUWh]cb B]hfc[Yb fY^YWh]cb <Y`]ia fYWcjYfm UbX `]eiYZUWh]cb

– Coil-wound and plate-ﬁn heat exchangers – Coldboxes ¶ @B; ghcfU[Y hUb_g

– – – – – – – ¶

Small and mid-scale LNG receiving terminals LNG loading facilities for ship, truck and rail LCNG fuelling stations LNG fuelling systems Local storage and re-gasiﬁcation LNG tanker trailers for road and rail Submerged combustion vaporizers @B; WUff]Yf Vc]`!cZZ [Ug fY!`]eiYZUWh]cb gmghYag

Designing Processes - Constructing Plants. Linde AG Engineering Division Dr.-Carl-von-Linde-Straße 6-14, 82049 Pullach, Germany Tel. +49.89.7445-3784, Fax +49.89.7445-4928 E-Mail: natural-gas-plants@linde-le.com, www.linde.com Select 89 at www.HydrocarbonProcessing.com/RS

HPIN CONSTRUCTION HELEN MECHE, ASSOCIATE EDITOR HM@HydrocarbonProcessing.com

North America C&I Engineering has completed the startup of what is said to be the world’s only dimethyl sulfide (DMS) plant, which is a cooperative venture between Hunt Refining Co. and Gaylord Chemical Co. LLC. C&I Engineering developed the process technology and provided pilot-plant design and testing, detailed design, procurement and startup services. Construction was completed on the 50 million-lbs/yr plant in June 2010. DMS production began in August 2010, and its quality reportedly exceeds available reagent-grade material. Enbridge Pipelines Inc. (EPI) will expand tankage of its mainline terminal at Edmonton, Alberta, Canada, by a million bbls at an estimated cost of $260 million, subject to regulatory approval. The expansion is targeted for completion by late 2012. The expansion is required to accommodate growing oil-sands production receipts both from Enbridge’s Waupisoo Pipeline and other non-Enbridge pipelines. It will be undertaken under the terms of the 2010 Incentive Tolling Settlement between EPI and the Canadian Association of Petroleum Producers (CAPP). Enbridge has received CAPP’s letter of support for the project. The project will be done in two phases. It involves constructing four tanks in the North Terminal and installing a short segment of pipeline and related infrastructure. Subject to regulatory approval, construction will commence early in 2011. Phase I is expected to be in service in the third quarter of 2012 and Phase II in the fourth quarter of 2012. ExxonMobil has completed the commissioning of new units to produce ultralow-sulfur diesel (ULSD) at its Baytown, Texas, and Baton Rouge, Louisiana, refineries. The units will enable ExxonMobil to increase ULSD supply by over 3 million gpd from the refineries and allow for reduced emissions from diesel consumption when used in modern engines. In December 2008, ExxonMobil announced plans to invest over $1 billion in

three refineries to increase ULSD supplies. The projects, located in the US and Belgium, required construction of new hydrotreater units at each facility, as well as modification to the existing facilities. In the US, the two projects provided more than 3,000 construction jobs and hundreds of millions of dollars of economic impact in the Baytown and Baton Rouge communities. Completion of commissioning activities in Antwerp, Belgium, is expected later this year. SABIC Innovative Plastics is adding specialty polypropylene (PP) compounding to its Bay St. Louis, Mississippi, site in the first quarter of 2011. The site will use its existing infrastructure and adopt new processes to produce SABIC PP compounds and SABIC STAMAX long-glass fiber-reinforced PP composites. Adding specialty PP to the site’s compounding capabilities enables the company to provide a broader range of highperformance materials to its North America customers. It also helps satisfy demand for specialty compounds. Investment in this process supports SABIC Innovative Plastics’ commitment to meeting both automotive OEM needs for lightweight, highperformance materials that can significantly reduce fuel consumption and emissions vs. competitive products, and nonautomotive needs in key segments. Chevron Phillips Chemical Co. LP plans to build a 1-hexene plant capable of producing in excess of 200,000 metric tpy at its Cedar Bayou Chemical Complex in Baytown, Texas. The project is approved to begin engineering design work and to develop engineering, procurement and construction deliverables. The plant would start up during the first quarter of 2014. With worldwide supply capabilities, the 1-hexene unit will realize significant advantages in infrastructure, feedstock availability and operational knowledge at the existing Cedar Bayou Chemical Complex. Upon construction, the new plant will be the third such plant to utilize Chevron Phillips Chemical’s proprietary selective 1-hexene technology, which produces comonomergrade 1-hexene from ethylene with exceptional product purity.

South America Ecopetrol has successfully started up two new units for producing ultra-low-sulfur diesel(Prime-D) and ultra-low-sulfur gasoline (Prime-G+), as part of its hydrotreating (HDT) project at its Barrancabermeja refinery, located in Santander, Colombia. Axens scope of work included license, optimization studies, basic process engineering and consulting services during the engineering, procurement and construction phase for a complex with seven process units and their interconnections. The HDT complex incorporates a diesel hydrotreater (Prime-D), a gasoline hydrotreater (Prime-G+), a sulfur-recovery unit, a tail-gas treater (Clauspol), a new amine regeneration unit, a new sour-water stripper and a new hydrogen-production unit. The complex brings the Colombian fuels production compliant to the latest environmental regulations. Excelerate Energy L.P. has signed a term sheet with a consortium formed by Argentina’s ENARSA S.A. and YPF S.A. to develop a second Argentinean liquefied natural gas (LNG) import facility in Escobar, approximately 30 miles north of Buenos Aires. The Escobar LNG facility, complementary to the existing GasPort operation

Trend analysis forecasting Hydrocarbon Processing maintains an extensive database of historical HPI project information. The Boxscore Database is a 35-year compilation of projects by type, operating company, licensor, engineering/constructor, location, etc. Many companies use the historical data for trending or sales forecasting. The historical information is available in comma-delimited or Excel® and can be custom sorted to suit your needs. The cost depends on the size and complexity of the sort requested and whether a customized program must be written. You can focus on a narrow request such as the history of a particular type of project or you can obtain the entire 35-year Boxscore database or portions thereof. Simply send a clear description of the data you need and you will receive a prompt cost quotation. Contact: Drew Combs P.O. Box 2608, Houston, Texas, 77252-2608 Phone: 713-520-4409 e-mail: Drew.Combs@GulfPub.com HYDROCARBON PROCESSING DECEMBER 2010

I 23

HPIN CONSTRUCTION south of Buenos Aires in Bahia Blanca, will allow delivery of up to 500 million cfd of natural gas to Argentina’s market. Due to its favored location, the Escobar GasPort facility will have a significant impact on the natural gas supply to Buenos Aires and North Argentina, adding flexibility and fast response to these growing markets. The Escobar facility, scheduled for completion in May 2011, will be fully integrated with Argentina’s existing gas transmission utilities. The year-round facility will accommodate an Excelerate Energy 150,900-m3 Energy Bridge regasification vessel receiving and regasifying LNG cargos via conventional LNG carriers utilizing Excelerate Energy’s proven ship-toship transfer process alongside the jetty. In addition to providing basic engineering and design support, Excelerate will provide modular-equipment and processcontrol components based on its proven GasPort technology. Stamicarbon has a license agreement with Tierra del Fuego Energia Y Quimica S.A. concerning a urea-synthesis plant and a urea-granulation plant with a capacity of

2,700 metric tpd to be built in southern Argentina. The urea-synthesis plant will use Stamicarbon’s Avancore technology, which was introduced in 2008. Taking full advantage of the benefits of utilizing Safurex in the HP synthesis section, the Avancore technology features zerooxygen intake, and minimum equipment and plant height. The granulation plant comprises Stamicarbon’s latest granulationtechnology innovations. Stamicarbon will deliver the processdesign package, and the proprietary HP equipment and associated services for both plants. China Chengda Engineering Co., Ltd., will build the plants, which will be located near the city of Rio Grande in Tierra del Fuego, Argentina. Startup is planned in 2012.

Europe Burckhardt Compression will deliver three process gas compressors to LUKOIL for its refinery-upgrade project in Volgograd, Russia. The compressors will be used as hydrogen-makeup compressors in the diesel-fuel hydrofining unit. This specific process requires dry-running technology

knowledge and experience with the operating conditions, as well as high reliability and availability of the equipment offered. The company’s process gas compressors, which are manufactured according to API 618 5th edition and include the latest stateof-the-art technology, will be delivered in mid-2011. They are scheduled to be in operation by the end of 2011. Statoil is exercising a one-year option to extend its contract with Aker Solutions at the Mongstad refinery in Norway. The estimated contract value for 2011 totals approximately NOK 100 million, and the work will occupy about 120 employees from Aker Solutions in Bergen, Norway. The oil refinery at Mongstad is a modern highly upgraded plant with an annual capacity of 10 million tons of crude. It is the largest facility of its kind in Norway. Aker Solutions’ scope of work mainly comprises minor modifications to increase robustness and generally improve the facility. The option period commences Jan. 1, 2011. In 2005, Statoil awarded Aker Solutions a modifications framework agreement for the Mongstad refinery. Remaining contract

We design it as if we had to run it From crude units to cokers, your ISBL resource C&I ENGINEERING

From Concept

502.451.4977 www.cieng.com Crude-Vacuum, HydrotreaƟng, Hydrocracking, Fluid CatalyƟc Cracking, Delayed Coking, IsomerizaƟon, Reforming, Amine, Sulfur Recovery, Sour Water Stripping, AlkylaƟon 24

I DECEMBER 2010 HYDROCARBON PROCESSING

Select 154 at www.HydrocarbonProcessing.com/RS

To CompleƟon

How come the weather is the only nasty thing at this gas field?

E50001-E440-F140-X-4A00

Innovative compressor trains from Siemens boost production and preserve the environment. After 50 years of operation, the Groningen gas field in the Netherlands is now, and also for the next decades, fit to secure the supply of its clients. The facilities are fully modernized. One key success factor was the long-term relationship of the operating company NAM and its contractors. Siemens has advanced the compression and variable speed drive technologies to ensure the adaptation of the gas supply to fluctuating demand, to slash maintenance requirements, and to maximize environmental performance. Highest availability and low power consumption of all units are the best basis for an eco-friendly and successful operation. Learn more: www.siemens.com/energy

Answers for energy. Select 63 at www.HydrocarbonProcessing.com/RS

HPIN CONSTRUCTION options include a one-year option followed by a two-year option. The contract parties are Aker Solutions’ subsidiary Aker Offshore Partner AS and Statoil ASA. Praxair Volgograd has a long-term contract to supply oxygen, nitrogen and compressed air to Plastkard, a division of the Nikochem Group. Praxair will build a new 350-tpd, energy-efficient air-separation plant in Volgograd, southern Russia.

The new plant, scheduled to start up in late 2011, will replace older air-separation plants and produce liquid products for the region’s merchant market. Plastkard is Russia’s third largest producer of polyvinyl chloride, a widely used plastic. INEOS Technologies has licensed its Innovene S process for manufacturing high-density polyethylene (HDPE) and linear-low-density polyethylene (LLDPE)

Let’s talk numbers

Air Liquide has a long-term agreement with RusVinyl to supply oxygen, as well as nitrogen and compressed dry air, to RusVinyl’s new worldscale polyvinyl chloride plant being built in Kstovo, in Russia’s Nizhny Novgorod region. Air Liquide will invest, build and operate a new state-of-the-art air-separation unit with an oxygen capacity of more than 350 tpd in Kstovo, which is scheduled to start up at the end of 2012. In addition, the company will produce a large quantity of liquid-air gases (oxygen, nitrogen and argon) to meet the needs of the region’s industrial customers. Air Liquide’s investment amounts to €60 million for the production facilities and supply chain. With growing requirements for highperformance, specialty polypropylene (PP) products, LyondellBasell plans to extend production and technology capabilities at its Spherizone PP process technology plant in Brindisi, Italy. Scheduled for completion in 2012, an upgraded process design and capacity expansion will include using additional comonomers, such as hexene, to manufacture products with the properties required by pipe, sophisticated film and healthcare applications. Plant capacity is expected to be increased by 50 kilotons, extending total capacity to 235 kilo tpy.

Select 155 at www.HydrocarbonProcessing.com/RS

PPI00181EN

Prize performance, capacity gains Packinox heat exchangers pack up to 16 000 m2 of heat transfer surface area into one single unit. That makes them the largest plate heat exchangers in the world. The performance benefits of the Packinox design include closer temperature approach, which gives rise to lower fuel consumption, and reduced emissions, plus a lower pressure drop. It all adds up to gigantic savings on your infrastructure and installation costs as well as your operating costs. Those kinds of numbers really make you a winner.

to ROSNEFT’s JSC Angarsk polymer plant in Angarsk, Russia. The 345,000-tpy plant will produce a full range of Ziegler and chrome monomodal and bimodal products. It will be ideally positioned to deliver specialty and commodity polyethylene (PE) products for the Russian and Chinese markets using INEOS Technologies’ slurry PE technology. The JSC Angarsk polymer plant will provide a broad product range, including bimodal pressure pipe, ensuring a competitive advantage for ROSNEFT customers, both in domestic and global markets. The companies have begun the project’s engineering phase.

Weyland has started production of second-generation bioethanol at its pilot plant in Bergen, Norway. The plant has the capacity to produce 200,000 lpy of bioethanol through an innovative method that promises to be profitable and environmentally friendly. The Weyland process is based on concentrated acid hydrolysis, with the com-

HPIN CONSTRUCTION pany’s core technology being a method (patent pending) for recovering acid consumed in the process. The process has a high-bioethanol yield and can use a variety of feedstocks. The company’s ambition is that the pilot plant will pave the way for a larger production plant in Norway, with an annual production capacity in the order of 25–30 million liters. Weyland is currently evaluating such a project in cooperation with the industrial company Elkem.

Middle East Foster Wheeler USA Corp. has been awarded a project-management consultancy (PMC) contract by Carbon Holdings for a petrochemical complex in Ain Sokhna, Egypt, with a nameplate capacity to produce 1.35 million tpy of polyethylene. Contract terms were not disclosed. Foster Wheeler will provide technical support and consulting services to Carbon Holdings through the project’s financial close, estimated to be late 2011. Following this, Foster Wheeler will book the full scope of the PMC award, which includes overseeing the activities of the selected engineering, procurement and construction contractor and subcontractors. The world-scale facility is expected to come onstream in 2015. Techint Engineering and Construction will build a complete transportation and storage system for post-refinery sulfur and petroleum coke in Yanbu, Saudi Arabia, for Saudi Aramco. The system is part of the Yanbu Export Refinery Project for constructing a new full-conversion refinery to process 400,000 bpd of oil. Techint E&C will provide turnkey engineering, procurement and construction activities: transportation and storage of petroleum coke, from the delayed coker to the industrial port of King Fahd; desulfurization process loading and stocking; and storage and loading of petroleum coke and solid-sulfur materials onto ships. The hand-over is scheduled for the first quarter of 2014.

Asia-Pacific The Linde Group and Haldor Topsøe will supply the syngas treatment and methanation unit for POSCO’s synthetic natural gas (SNG) plant. The new plant will be erected in Gwangyang, South Korea, producing SNG from coal and/or petcoke. With a nominal capacity of 500,000 metric tpy of pipeline-ready SNG, this reported

first SNG plant in South Korea will be operational by the end of 2013. The Linde Group’s Engineering Division will supply the complete technology chain of syngas treatment and conditioning, including sour shift, acid-gas removal—featuring Linde‘s Rectisol sulfur recovery process technology. Haldor Topsøe will supply the complete methanation technology, TREMP, including product gas conditioning, to deliver SNG with a methane purity of 98+%. The project is owned by POSCO of South Korea and executed by POSCO Engineering and Construction.The plant, which will feature ConocoPhillips’ E-gas gasification technology, will be adjacent to the steel works of Gwangyang, where site preparation is already initiated. Solvay will build a specialty polymers compounding plant at its site in Changshu in the province of Jiangsu, China. The plant, which requires a €21 million investment, is expected to start up in the last quarter of 2012. The compounding plant will serve China’s fast-growing markets for electronics, automotive, consumer and industrial applications, and will initially start producing compounds of Amodel polyphthalamide, Ixef polyarylamide and Kalix. It will be fully adaptable for future expansion for both overall capacity and for other highperformance and fluorinated polymers. Alfa Laval has an order for its Packinox heat exchangers to be used in an Indian refinery. The order value is about SEK 110 million and delivery will be finalized during 2011. The heat exchangers will be used in a catalytic chemical process for mixed xylene, which, among other things, can be used for producing PET bottles. Albemarle has completed the research and development (R&D) laboratory facilities and has begun construction on its Yeosu, South Korea, manufacturing facility. Albemarle staff, representatives of the local construction firm, Safety First, and City of Yeosu officials, including Mayor Kim Chung-seog, joined to mark the occasion. Completion of the Yeosu site R&D lab facility enables production of metallocene polyolefin catalyst samples incorporating Albemarle’s breakthrough ActivCat technology for qualification trials with local customers. Pilot-plant facilities will be completed in 2010. Intermediate commercial

operations will begin in mid-2011, with the commercial facility being fully operational in 2012. The site will produce finished catalysts, activators, such as methylaluminoxane and metallocene components, and high-purity metal organics. ExxonMobil Yugen Kaisha has announced that Japan Butyl Co. Ltd. has completed a major expansion to increase butyl-rubber production capacity at its plant in Kawasaki, Japan. The expansion adds 18,000-tpy production capacity, bringing the plant’s total capacity to 98,000 tpy. The expansion is part of the company’s commitment to help meet growing demand for butyl rubber and showcases ExxonMobil’s recent advances in process technology. For example, these new proprietary technology advancements allow for the butyl-polymerization reaction that normally occurs at –95°C, to operate at –75°C, creating significant energy savings and capital investment reductions. CB&I has been awarded two contracts valued in excess of $50 million by Woodside for additional work on the Pluto liquefied natural gas (LNG) project in Western Australia. This is a third-quarter award. Earlier this year, CB&I completed two full-containment LNG tanks and three condensate tanks on the project. Invensys Operations Management has a five-year, multimillion-dollar contract to deliver a comprehensive refinery-wide optimization solution for Bangkok-based Thai Oil Public Co., Ltd. Under agreement terms, Invensys will implement its SimSci-Esscor ROMeo optimization software to improve the real-time performance of Thai Oil’s refinery operations. An integrated software solution for simulation, data reconciliation and operations decision support, the ROMeo software solution will be a critical part of Thai Oil’s five-year master plan to optimize all its major distillation and conversion units, as well as its plant-wide energy and hydrogen systems, supporting its vision of becoming a fully integrated high-performance refiner. Invensys will also provide turnkey design, engineering, commissioning and consulting services for the refinery-wide optimization solutions. JGC Corp. has been awarded the Hachinohe liquefied natural gas (LNG) HYDROCARBON PROCESSING DECEMBER 2010

I 27

HPIN CONSTRUCTION terminal project in the Aomori district of Japan by JX Nippon Oil & Energy Corp. JGC will perform engineering, procurement, construction (EPC) and assistance services for commissioning work under a lump-sum turnkey scheme. Contract value was not disclosed. The project encompasses the construction of two LNG tanks with a capacity of 140,000 m3 each, unloading facilities for LNG tankers, loading facilities for coastal

tankers, regasification facilities, loading facilities for tank cars, etc. JGC will take charge of all EPC work, excepting the LNG tank erection and part of the marine civil work. The terminal will start operations in April 2015. Samsung Engineering has a $770 million Sabah oil and gas terminal project contract with Petronas Garigal Sdn. The plant will be built in the Sabah region of

Designed specifically to meet the requirement of API 610, the API Maxum Series is available in 35 sizes to handle flows up to 9,900 GPM and 720 feet of head. Standard materials include S-4, S-6, C-6 and D-1. A wide range of options makes this the API 610 pump for you!

Creating Value. Carver Pump Company 2415 Park Avenue Muscatine, IA 52761 563.263.3410 Fax: 563.262.0510 www.carverpump.com

Select 156 at www.HydrocarbonProcessing.com/RS

Malaysia, with a capacity of 300,000 bpsd of oil and 1,250 million scfd of gas, which will account for 40% of Malaysia’s crudeoil production. Samsung Engineering will lead engineering, procurement, construction and commissioning (EPCC) work with local partner NCSB Engineering and will form an EPCC Alliance Integrated Team with the client. The plant plans to be mechanically complete by December 2013. Larsen and Toubro Ltd. has two projects from ONGC to set up additional processing units at its gas-processing complexes at Hazira and Uran. The project at the Uran complex will enhance the complex’s gasprocessing capacity by 5 million scmd. New facilities to be set up include a gas-sweetening unit (GSU), an LPG recovery unit, a condensate fractionation unit, a condensate handling unit and other utilities. The project for the Hazira complex is for augmenting the complex’s gas-processing capacity by 5.6 million scmd. New facilities include a GSU, a gas dehydration unit, a dew-point depression unit, and offsite and utilities. The contract scope includes project management, residual basic design, planning and monitoring, residual process engineering, detailed engineering, procurement, supply, fabrication, manufacturing, inspection, transportation, storage, construction, installation, testing, mechanical completion, precommissioning, commissioning, performance-guarantee run testing and handing over of new process units, offsite and utilities to the owner. Indian Oil Corp. Ltd.‘s Bongaigaon refinery (Assam) has selected AVEVA Group plc’s AVEVA plant solutions for design and build on two projects: a new diesel hydrotreater and a motor-spirit (MS) quality upgrader. The new diesel hydrotreater aims to improve the quality of high-speed diesel to conform to Euro III-equivalent norms. The project will also improve the smoke point of raw kerosine and enhance production of superior kerosine oil and automatic transmission fluid. The MS quality-upgrade project will boost MS quality to conform to Euro III-equivalent norms. Indian Oil’s Bongaigaon refinery comprises two crude-distillation units, two delayed-cokers and a coke-calcination unit with a processing capacity of 2.35 million tpy of crude oil. HP

HPI CONSTRUCTION BOXSCORE UPDATE Company

City

Plant Site

Project

SAMIR Chevron Nigeria Ltd PetroSA Undefined

Mohammedia Escravos Port Elizabeth Hoima

Mohammedia Escravos Coega Hoima

Hydrocracker Hydrocracker Refinery Refinery

Capacity Unit Cost Status Yr Cmpl Licensor

Engineering

Constructor

AFRICA Morocco Nigeria Repub S Africa Uganda

40 34 400 200

Mbpd Mbpd Mbpd bpd

11000 46000

C U P P

2010 2011 2015 2015

CLG CLG

ASIA/PACIFIC Bangladesh

BCIC

Fenchuganj

Urea

China China China India Malaysia

PetroChina KNPC KNPC Brahmaputra Cracker and Polymer Petronas

Shenzhen Zhangjiagang Zhangjiagang Lepetkata Sabah

Liquefied Nat Gas Ethylene Refinery Pygas Terminal, Gas

Total Total Total Alma Petroli SpA Alma Petroli SpA Yamal LNG LLC Gazprom

Gonfreville Gonfreville Gonfreville Ravenna Ravenna Yamal Pancevo

Distillation, Crude Hydrocracker Hydrotreater, Gas Oil Naphta Vacuum Revamp LNG Liquefaction Plant Cracker, Catalytic

La Paz La Paz La Paz

Diesel Gasoline LPG

1760 Mtpd

565

2014

9900 9000 9000 1140 770

P P P U E

2011 2013 2013 2012 2013

EX 205 Mbpd 950 EX 48 Mbpd 950 364 Mbpd 950 85 t/a 1.1 TO 100 tph 1.5 15 MMtpy 20000 RE 22 bbl 20

H H H E S P M

2013 2013 2013 2011 2013 2016 2011

3 1 300 55 300

m-tpy m-tpy BNm3/y Mtpy bpsd

KBR|Stamicarbon Hofung Technology

Chengda Eng |Complant

Samsung Eng

EUROPE France France France Italy Italy Russian Federation Serbia

FW Technip Technip Conser Conser CB&I

LATIN AMERICA Bolivia Bolivia Bolivia

Yacimientos Petr Fiscales Yacimientos Petr Fiscales Yacimientos Petr Fiscales

17 bpd 7070 bpd 164 m-tpd

1000 1000 1000

P P P

2015 2015 2015

2.5 2.2 4163

U P E

2011 2014 2013

bpd t/a Mtpy tpy

12800 1100 0.5 2.5

U P U U

2013 2014 2011 2011

2 MMgpy 40 Mbpd 1 Bcf

1000 15 1200

C C E

2010 2010 2015

MIDDLE EAST Iran Iran Kuwait

Ehtemam Jam Co Undisclosed KNPC

Assaluyeh Assaluyeh Formaldehyde Yasuj Yasuj Refinery Mina Al Ahmadi Mina Al Ahmadi Clean Fuels

40 Mtpy 150 bpd TO 50 Mbpd

Saudi Arabia Saudi Arabia Turkey Turkey

Saudi Aramco\Total JV Al Rajhi Petrochemical Gentas Kimya Kastamonu Entegre

Jubail Yanbu Gebze Kastamonu

Jubail Ind City Yanbu Gebze Kastamonu

Refinery Silicones Formaldehyde Formaldehyde

400 6 RE 30 BY50000

Baton Rouge Artesia Coos Bay

Baton Rouge Navajo Rfy Coos Bay

Diesel, Low Sulfur Distiller, Crude LNG Terminal

RE BY

Alder

Staff

Axens|Haldor Topsøe| CLG Fluor SGS |Lummus Technology Alder Alder

Alder Alder

Staff Staff

KP Engineering, LP Black & Veatch

KP Engineering, LP ENTREPOSE|Kiewit Energy|Vinci Construction

UNITED STATES Louisiana New Mexico Oregon

ExxonMobil Holly Corp Jordan Cove Energy

See http://www.HydrocarbonProcessing.com/bxsymbols for licensor, engineering and construction companies’ abbreviations, along with the complete update of the HPI Construction Boxscore.

BOXSCORE DATABASE

ONLINE

THE GLOBAL SOURCE FOR TRACKING HPI CONSTRUCTION ACTIVITY For more than 50 years, Hydrocarbon Processing magazine remains the only source that collects and maintains data specifically for the HPI community, publishing up-to-the-minute construction projects from around the globe with our online product, Boxscore Database. Updated weekly, our database helps engineers, contractors and marketing personnel identify active HPI construction projects around the world to: • Generate leads • Market research • Track trend analysis • And, decide future budget planning. Now, we’ve made our best product even better! Enhancements include: • Exporting your search results to Excel so you can compile your research • Delivering the latest updated projects directly to your inbox each week • Designing customized construction reports for your company using our 50 years of archived projects. For a Free 2 -Week Trial, contact Lee Nichols at +1 (713) 525-4626, Lee.Nichols@GulfPub.com, or visit www.ConstructionBoxscore.com

Select 157 at www.HydrocarbonProcessing.com/RS HYDROCARBON PROCESSING DECEMBER 2010

I 29

PROCESS INSIGHT Optimizing CO2 Capture, Dehydration and Compression Facilities The removal of CO2 by liquid absorbents is widely implemented in the field of gas processing, chemical production, and coal gasification. Many power plants are looking at post-combustion CO2 recovery to meet environmental regulations and to produce CO2 for enhanced oil recovery applications. The figure below illustrates actual data of fuel consumption in 2005 and an estimate of energy demand for various fuels from 2010 to 2030. The world energy demand will likely increase at rates of 10–15% every 10 years. This increase could raise the CO2 emissions by about 50% by 2030 as compared with the current level of CO2 emissions. The industrial countries (North America, Western Europe and OECD Pacific) contribute to this jump in emissions by 70% compared to the rest of the world, and more than 60% of these emissions will come from power generation and industrial sectors.

formulated solvent without implementing any split flow configurations. This is much less than the reported steam usage for the MEA solvent. The design of a facility to capture 90% of the CO2 from the flue gas of a coal fired power plant is based on the specified flue gas conditions, CO2 product specifications, and constraints. Using the ProMax® process simulation software from Bryan Research & Engineering, CO2 capture units can be designed and optimized for the required CO2 recovery using a variety of amine solvents. The following figure represents a simplified process flow diagram for the proposed CO2 Capture Plant.

Despite the strong recommendations from certain governments, there are very few actual investments in CO2 capture facilities geared toward reducing greenhouse gas emissions mainly because of the high cost of CO2 recovery from flue gas. CO2 capture costs can be minimized, however, by designing an energy efficient gas absorption process. Based on the findings of recent conceptual engineering studies, HTC Purenergy estimated the production cost to be US$ 49/ton CO2 (US$ 54/ tonne CO2) for 90% CO2 recovery of 4 mole% CO2 content in the flue gas of NGCC power plants. A separate study showed the cost for 90% CO2 recovery of 12 mole% CO2 from a coal fired power plant to be US$ 30/ton CO2 (US$ 33/tonne CO2). The cost of CO2 recovery from coal power plant flue gas is substantially less than that of NGCC power plant flue gas due to the higher CO2 content in the feed. The energy efficiency of a CO2 capture plant depends primarily on the performance of the solvent and optimization of the plant. In traditional flue gas plant designs, MEA was the primary solvent and was limited to 20 wt% to minimize equipment corrosion. Recent developments in controlling corrosion and degradation has allowed an increase in the solvent concentration to about 30 wt% thus decreasing the required circulation and subsequent steam demand. A recent DOE study shows the steam consumption for an existing CO2 plant using 18 wt% MEA (Kerr McGee Process) is 3.45 lb of steam per lb of CO2 for amine regeneration. A modern process that uses 30 wt% MEA is expected to use 1.67 lb of steam per lb of CO2 for amine regeneration. The HTC formulated solvent is a proprietary blend of amines and has a lower steam usage than the conventional MEA solvent. Based on the material and energy balances for the plant designed in the recent study, the reboiler steam consumption is estimated at about 1.47 lb steam/lb CO2 using the proposed

The table below presents the main findings for CO2 capture from the coal fired power plant and the NGCC power plant, each designed to produce about 3307 ton per day (3,000 TPD metric). To produce the same capacity of CO2, only one train with smaller column diameters is required in the case of the coal power plant and two trains with larger column diameters are required in the NGCC Power Plant case. This is mainly due to processing a larger flue gas with lower CO2 content in the NGCC power plant. Consequently, a substantial reduction in the capital and production cost was reported for the coal fired power plant CO2 recovery facility.

For more information about this study, see the full article at www.bre.com/support/technical-articles/gas-treating.aspx.

Bryan Research & Engineering, Inc. P.O. Box 4747 • Bryan, Texas USA • 77805 979-776-5220 • www.bre.com • sales@bre.com Select 113 at www.HydrocarbonProcessing.com/RS

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

Maximize return on capital projects How to achieve all of your major capital investments objectives on budget and schedule J. HUMPHRIES, Fluor, Greenville, South Carolina

ith today’s tough economic conditions, competitive global markets and tight capital, the need to optimize net present value and internal rates of return from capital investments has never been greater. To better address this issue, one option applies a set of best practices packaged via a holistic approach to maximizing returns from major projects.

Getting started. First, we will examine the mathematical drivers for optimizing capital project returns. As suggested by the fishbone diagram in Fig. 1, maximizing capital return means generating the highest revenues with the lowest operating cost from the smallest investment. To generate the highest revenues, we maximize production rates, plant availabilities, and the quality and value of our products, as shown in Fig. 2. To achieve lowest operating costs, we blend high plant productivity, maximum production yields and competitive labor, material and utility costs with low inventories. For low capital costs, we design and build flexible facilities efficiently with competitive costing for the engineers, craftsmen, construction materials and production equipment. In addition, we secure inexpensive, timely financing to enhance procurement leverage without excessively burdening the cash flows.

compromises among the drivers. Therefore, the key to maximizing return is to address all drivers in optimal fashion. How is this done? A structured suite of tools and processes with the expertise and discipline to effectively apply them is critical. Experience indicates that there are five toolkits that are essential: • Stage-gated capacity planning is one of the more critical sets of tools and processes for optimizing capital project returns. As shown in Fig. 3, customized for the specific industry and company strategy, a stage gate process is the principal means for managing a capital program portfolio. This means identification, justification, and approval of capital projects to promote incremental investment and risk management subject to preestablished criteria between development stages. • Project execution efficiency is a second obvious and critical tool set. Refined repeatable processes must address estimating, cost and schedule controls, project management, engineering practices, material and contracts management, and safety. Such

Availability Production Rates Yields Quality

Money flow. The velocity of the cash flows is also a critical

issue for maximum net present value and optimal rates of return. We must not only generate significant gross margins from the delta between revenues and operating costs. We must also generate these margins quickly to compound early returns. The logic outlined focuses on maximizing returns from a single, specific project. In reality, we need to optimize inter-related returns from a portfolio of capital projects while managing risks. From this discussion, it should be clear that there are many value drivers. The biggest challenge is in optimizing trade-offs and

Project portfolio

FIG. 1

Flexible Lean Competitive EPC Costs Financing

FIG. 2

Drivers impacting returns for major projects.

Capital costs

Stage-gate capacity planning

Maximum return on investment

Operating costs

Productivity Labor Material Utilities Yield Inventories

Operating revenues

Cost factors in major capital project.

FEL 1

FEL 2

FEL 3

Implement

Identify

Select

Deﬁne

Execute

Needs Objectives Options Strategies

Technology Location Flows

Components Arrangement Speciﬁcations

Design details Procurement Construction Commissioning

FIG. 3

Stage-gate capacity planning.

HYDROCARBON PROCESSING DECEMBER 2010

I 31

SPECIALREPORT

PLANT DESIGN AND ENGINEERING

Value improving practices Value engineering Risk management Energy reduction Lean production Reliability Engineering Standards and specs Waste minimization Condition monitoring

FIG. 4

Process simpliﬁcation Performance simulation Readiness planning Life cycle valuation Technology landscaping Benchmarking Classes of facilities

Value improving practices.

Operational readiness Master planning Life-cycle design support

Vertical launch

Maintenance readiness Capacity planning

Operations readiness Systems readiness Organizational readiness Support readiness Supply chain readiness

Owner’s risk and readiness reviews Concept and front-end Detailed Procurement engineering design engineering and construction

FIG. 5

Commissioning and ramp-up

Operational readiness stages in major projects.

tools also include software for engineering and design, cost and critical path management, quality control, commissioning, risk management, logistics and construction. • Value-improving practices, such as those listed in Fig. 4, are a third set of critically important tools for maximizing capital project returns. These tools include processes for value engineering, performance simulation, risk management, energy reduction, operational readiness planning, lean production, process simplification and life-cycle value assessment. They assure that a project is well scoped and focused on financial objectives rather than merely on technical objectives. • Operational readiness processes ensure that the personnel, systems, practices, tools and materials needed to operate and maintain a new facility are in place at startup. Fig. 5 summarizes the tools found to be critical when optimizing the readiness of their global mega capital projects. • Operational excellence involves tools that enable facility owners to produce optimal profit while achieving employee safety and plant integrity in a sustainable green environment. As suggested in Fig. 6, operational excellence addresses the company culture, adaptability and competitive aptitude in addition to the human resources, work processes, and systems in a facility. Clearly, without operational excellence, the best designed, most modern and potentially most efficient industrial facilities provide only mediocre returns. When operational readiness measures taken before startup are weak, the time required to achieve operational excellence after startup is delayed to the extent that return on capital investment is discounted and the potential net present value achieved by the project is far from optimal. In summary, the drivers for maximizing financial returns from capital projects are fairly well-known and understood. Yet, the means to optimize conflicting drivers are less understood and deployed. The key is to build an orchestra of expert teams with the tools and track record to optimize the critical value drivers. HP

Jim Humphries is the vice president of Performance Technology in Fluor’s Global Services business group. HIs team provides consulting and technical services to start new facilities and for step-change performance improvement initiatives. The scope of this support varies from capital program policies, life-cycle design optimization, and value improving practices to maintenance, reliability, lean production, purchasing, contracts management, quality certifications, engineering and human performance.

Skilled motivated personnel

EMERGENCY SERVICE 800-231-0077

ACS Industries can set you free.

Best practices application

Systematic improvement culture

Operational excellence UNHAPPY with long lead times for response and product delivery? Don’t trap yourself into thinking only one source is able to handle your requirements. ACS can replace almost any existing tray, regardless of original manufacturer. With 70 years’ experience, we use advanced 3-D modeling and CAD/CAM to design and make a wide variety of trays and internals.

YOU ARE FREE to choose the highest quality and best price, delivery, and engineering support. Call ACS Industries for all your trays and internals.

Highperforming physical assets

FIG. 6 Select 158 at www.HydrocarbonProcessing.com/RS 32

Superior management systems

Challenging accountabilities

Factors involved in operational excellence.

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

Best practices in treating liquefied petroleum gas are defined Advantages of removing H2S before the LPG unit and energy optimization of the LPG splitter B. ARDALAN, M. KHORSAND MOVAGHAR and M. MALEKI, Energy Industries Engineering and Design Consultant Company (EIED), Tehran, Iran

his article presents the basic engineering design of a liquefied petroleum gas (LPG) recovery unit at the Khuzestan oil refinery, Iran. This refinery, like other oil refineries, consists of one LPG recovery unit. The feed to the LPG recovery unit comes from three different sources. The first feed is from a catalytic cracking reforming (CCR) unit that is completely sweet with no H2S content. The second feed is from a hydrocracking unit that is sour and has H2S content. The third feed comes from a stabilizer in the crude distillation unit (CDU) which is also sour and has H2S and mercaptan contents. The LPG recovery unit in this project consists of two main separation sections including deethanizer and depropanizer columns. A main product from this unit is propane from the top of the depropanizer section. This is routed to propane storage spherical tanks. The other main product from this unit is butane. Butane is routed from the depropanizer bottom to butane product spherical tanks. These products are blended at different ratios in different seasons depending on the demand. The byproduct from the top section of the deethanizer is fuel gas and it is routed to a fuel gas unit. A study was performed on the effects of H2S content of the feed on the deethanizer design parameters. In addition, operating pressure effects of the depropanizer column with regards to the depropanizer’s condenser utility consumption were considered. The recommended scheme for treating the LPG streams before recovery is illustrated in Fig. 1. Deethanizer. LPG feed from three sources is mixed in the feed

surge drum and pumped to the deethanizer splitter. The column’s function, operating at 26 barg at the top, is to remove ethane as an overhead vapor stream and yield a bottom product containing propane and heavier hydrocarbon products. The deethanizer column is designed to meet the specification of maximum C3 loss in the fuel gas at 7 wt%. The column overhead is cooled to 60°C in the overhead air cooler, then lowered to 40°C in the overhead condenser. Finally, the mixed phase is routed to the reflux drum. In the separator, the vapor phase includes lighter hydrocarbons (C2-) which are sent to the fuel gas unit as a byproduct. The liquid phase is sent back to the top of the column with a reflux pump. The liquid hydrocarbons recovered from the deethanizer bottom (C3+) are sent under level control, resetting a flow control valve to the depropanizer section for further processing. Fig. 2 shows the flow diagram for the deethanizer section of the Khuzestan oil refinery’s LPG recovery unit.

Case studies. Process simulations were developed for the following cases to study the effect of H2S content in the feed to the deethanizer column on the column design parameters, including condenser heat duty, reboiler, column diameter, etc. • Case 1. Sour feed is routed to the LPG recovery unit with the deethanizer column playing the role of H2S removal. In this case, the design is performed based on 5 ppm of H2S content in the deethanizer bottom. • Case 2. Sour feed is routed to the deethanizer column and the design is accomplished on the basis of 300 ppm H2S content in the deethanizer bottom. • Case 3. Treated feed enters the deethanizer column with no H2S content. Process simulation. A simulation for Cases 1–3 was done

for this study in the LPG recovery unit, using commercially available software.1 Table 1 summarizes the simulation results of the deethanizer column design parameters for Cases 1–3. Assumed specifications in simulation include:

• Condenser temperature is fixed at 40°C for all case studies due to refrigeration limitations in the plant. • Column optimum operating pressure is 27 bara in the top section of the splitter to meet all specifications—a pressure drop of 0.5 bar is considered between the top and bottom of the column. • C2/C3 ratio in the bottom of the deethanizer is set to a maximum of 0.2 as one of the main tower design specifications. • Column trays are commercial sieve trays. LPG from stabilizer CDV/DU unit

LPG treating Unit: section 1

Treated LPG

H2S and mercaptan removal

LPG from HCR unit

LPG treating Unit: section 2

Fuel gas

LPG recovery unit

C3 product C4 product

Treated LPG

Only H2S removal LPG from CCR unit

FIG. 1

Block flow diagram of an LPG unit.

HYDROCARBON PROCESSING DECEMBER 2010

I 33

SPECIALREPORT

PLANT DESIGN AND ENGINEERING

Discussion for C3 loss. C3 loss is defined as the ratio of pro-

pane mass flowrate in the distillate of the deethanizer column to propane mass flowrate of the feed entering the column. (C3 loss = propane flowrate in the top stream of the deethanizer column/ propane flowrate in the feed to the deethanizer column). The main objective of this unit is to produce propane and butane and send them to spherical tanks to mix at required ratios. Therefore, C3 loss is a critical factor in unit efficiency. Referring to Table 1, C3 loss in Cases 1 and 2 is much greater than that of Case 3, which is sweet. Condenser and reboiler heat duty. As illustrated in Table

1, larger quantities of heat duties are required for the condenser and reboiler when running the deethanizer with sour feed in Cases 1 and 2, as compared with the sweet feed in Case 3. Reflux rate (ton/hr). As shown in Table 1, the reflux rate

is increased from 11.7 in Case 3 to 27.7 in Case 2. Moreover, this rate is further increased to 81.7 in Case 1, which is poorly designed. It is worth noting that by increasing the reflux rate, the column diameter must also increase. C2- to fuel gas unit

NNF NNF

CW From LPG treating unit (LPG from HCR)

LPG from CCR

To Sour water unit

Split range

From LPG treating unit (LPG from stabilizer)

TC FC

LPS LC LC

NNF

To ﬂare header

LPC

To depropanizer column

Outlook. The major advantage of H2S removal before the LPG

Process flow diagram for the deethanizer section.

FIG. 2

Number of trays. An evaluation, as part of the C2 splitter process design, was done to consider the effect of varying the number of trays on the column heat duties. Results are illustrated in Table 2, which includes a comparison of changing tray numbers (increasing the number of trays in four steps) against the condenser and reboiler heat duties for sour cases. As shown in Table 2, increasing the number of trays results in a decrease in the column heat duties. The reduction rate for Case 1, up to tray number 34, will be near 40% and more than 34 will be less than 20%. While for Case 2 the reduction rate is not considerable, and up to tray number 30 (only in the first step) would be less than 20%. However, by increasing the tray number to 34 trays for sour cases, especially for Case 1, the heat duties were significantly different from those of Case 3. If the tray number increases to 40, there will be a slight difference in the condenser and reboiler duties but it never approaches exactly to the parameters of the sweet case. Result: No added benefit when increasing the number of trays over 34. From past experience, when the column diameter was between 2 m to 4 m, tray spacing needed to be at least 400 mm. Hence, by increasing the tray number from 26 to 34, a corresponding height of 3.2 m was added to the column height. In other words, increasing the tray numbers requires a considerable rise in column height and an increase in cost, which is not an economical and optimum design. Therefore, the tray numbers, based on the sweet case, were set at 26 as an assumption for the simulation design. Cost impact was due to: • Sour service. In Case 2, the whole unit material would have to be resistant to sour service because there was 300 ppm H2S in the LPG feed entering the C3 splitter. In Case 1, the deethanizer bottom product only had 5 ppm H2S content. If any malfunction in the deethanizer operation occurred, the unit material would be suitable for sour service. Thus, sour service requirements were considered for Cases 1 and 2. As a result, economically, Case 3 is the optimum case because it does not require materials suitable for sour service. • Larger sizes. Referring to Table 1, the deethanizer column diameter is much larger in Cases 1 and 2 compared to Case 3 in which the feed enters the column with no H2S content. In addition, as shown in Table 1, in Cases 1 and 2, the column has three stages. Case 3 only has two stages. The additional stage of Cases 1 and 2 causes additional manufacturing and maintenance cost. Thus, the smaller tower with fewer stages is the result of treating the LPG before entering the LPG recovery unit.

recovery unit is decreasing the heat duties of the deethanizer TABLE 1. Deethanizer column design parameters for Cases 1–3 Tray sizing Stages Diameter, m

Condenser duty Qc, MW

Reboiler duty Qr, MW

C3 loss, %

Reflux rate, ton/hr

H2S in bottom, ppm

Case 1 (sour case)

7.2

12.8

9.8

81.7

1 2–20 21–26

2.28 3.35 3.5

Case 2 (sour case)

2.5

8.1

7.3

27.7

300

1 2–22 23–26

1.372 3.0 3.2

Case 3 (sweet case)

1.1

6.7

3.6

11.7

0.0

1 2–26

1.06 2.90

I DECEMBER 2010 HYDROCARBON PROCESSING

PLANT DESIGN AND ENGINEERING column’s reboiler and condenser, which reduces the utility consumption. Another advantage of H2S removal is that the maximum diameter of the deethanizer splitter—a main factor of column design—can be reduced by 0.6 m. Therefore, the condenser, reboiler and not needing sour service material results in a considerable reduction in cost and lower utilities. Depropanizer. The LPG stream from the bottom of the

deethanizer column is fed to the depropanizer, under flow control cascades with level control, at an operating pressure of 26.5 barg. The column produces a propane stream as an overhead liquid product and a bottom stream containing butane. Butane is first air-cooled in a butane product air-cooler to 60°C then cooled to 40°C in the butane product trim cooler. The liquid butane product is routed to butane spherical storage tanks. The overhead propane gas is cooled in the depropanizer overhead condenser and totally condensed. The liquid propane product is routed to the depropanizer reflux drum. Some liquid (as reflux ratio) is returned by the depropanizer reflux pump as reflux to the top tray of the depropanizer. This is done under flow control cascaded with the temperature control located on the tower. The rest of the propane liquid is sent under level control resetting flow control by the propane transfer pumps to the propane spherical storage unit at 40°C.

SPECIALREPORT

tanks was 40°C (plant specification), two different cases were studied to achieve this temperature. • Case 1. Propane is routed to spherical tanks from the reflux drum with an operating temperature of 40°C. The related optimized operating pressure was achieved at 13 barg (at top) to meet the depropanizer design specifications. Fig. 3 illustrates the process flow diagram related to this case (the basic design was closed using this design code). • Case 2. No temperature limitation was considered for the depropanizer reflux drum. As a result, propane from the reflux drum with the higher temperature firstly entered a water cooler (cooled TABLE 2. Effect of tray number on condenser and reboiler duties Deethanizer Case 1 (sour case) Number of trays

Condenser duty Qc, MW

7.2

5.1

3.9

3.2

12.8

10.7

9.6

8.8

8.6

Reboiler duty Qr, MW

Case 2 (sour case) Number of trays

Condenser duty Qc, MW

2.5

2.1

1.9

1.8

1.7

Reboiler duty Qr, MW

8.1

7.7

7.5

7.4

7.3

Case 3 (sweet case)

Case studies. Process simulation was developed for the fol-

lowing cases to study the effect of operating pressure of a column on the cooling water consumption of the condenser. Since the temperature of the propane product routed to the spherical

NNF

Number of trays

Condenser duty Qc, MW

1.1

Reboiler duty Qr, MW

6.7

NNF

To ﬂare header

To fuel gas sys.

NNF

To fuel gas sys.

NNF

NNF PDC

PDC

Equalizing line

LC LC

To ﬂare header

From deethanizer column

NNF

FC TC

Propane to storage

To ﬂare header

From deethanizer column

NNF

FC TC FC

Propane to storage

LPS LC

LPC

LPC FC

FC TC

Butane to storage

Process flow diagram for depropanizer in Case 1.

Butane to storage CW

CW FIG. 3

LPS

FIG. 4

Process flow diagram for depropanizer in Case 2.

HYDROCARBON PROCESSING DECEMBER 2010

I 35

SPECIALREPORT

PLANT DESIGN AND ENGINEERING

down to 40°C) and eventually routed to spherical tanks with a temperature of 40°C. Hence, in this case, the maximum related operating pressure was set at 20.4 barg (at top). Fig. 4 illustrates the process flow diagram of this case. This case was further developed after the project was finished as an R&D internal project. Process simulation. A simulation for the previously mentioned cases was done. For these cases, a study was conducted for two different values of column operating pressure. The assumed parameters for the depropanizer design were: • Propane vol % in the bottom stream was 0.15% • Butane vol % in the top stream was 3%. Discussion. It was noted that the temperature of the propane routed to the spherical tanks was set at 40°C as a specification in the plant. After a simulation was carried out for Case 1, the temperature of the overhead stream of the depropanizer column was 42°C. Simulation results showed that, in Case 1, the condenser heat duty was 14.5 MW to cool the overhead stream from 42°C to 40°C. The air cooler cannot operate as the condenser to achieve the outlet temperature of 40°C, because the maximum ambient temperature of the plant is 53°C. Hence, a water cooler is the only option for the condenser where cooling water supply temperature is 35°C and the cooling water return is 45°C. Therefore, in this case, the cooling water consumption would be 2,425,000 kg/hr—a huge utility consumption amount. It should be noted that the corresponding condenser would consist of four shells in parallel. After performing a simulation

for Case 2, with higher operating pressure, the temperature of the overhead stream of the depropanizer column was 62°C, which was cooled via an air-cooler to 60°C. A great portion of this stream was refluxed to the column, while just one tenth of this stream was cooled to 40°C, as the unit’s propane product. The heat duty of the propane product cooler was 0.36 MW in this case. The cooling water consumption rate, which is in accordance with this amount of heat duty, was 60,410 kg/hr. This cooling water value exhibits an optimum rate of utility consumption compared to Case 1 in this unit. Outlook. The main advantage of increasing operating pressure of the depropanizer distillation column is reducing the cooling water consumption rate and replacing the water cooler condenser with air coolers (Fig. 4). The bottom product’s operating pressure of the deethanizer was 27.5 bara—no extra energy is required to operate the depropanizer at a higher pressure. As a result, in Case 2, the C3 splitter column operates at a higher pressure easily with a lower pressure loss of the control valve located in the feed line to the depropanizer. Also, when running the depropanizer at a higher pressure, a smaller condenser size with lower costs and convenience in manufacturing, can be achieved (from four parallel shells in Case 1 as opposed to only one shell in Case 2). Furthermore, there is no need to use a pump to transfer the propane product to the tankage area— another advantage of Case 2. Whereas, in Case 1, a transfer pump is required for this purpose. HP ACKNOWLEDGEMENT The authors thank S. J. Mousavi and B. Golsazi of NOIEC and N. Ashouri and A. A. Farahnak of EIED for their expert advice during design works.

LITERATURE CITED Hysys Process, “Simulation basis,” Hyprotech, LTD, 2002.

Bahareh Ardalan has been a senior process engineer working in the process department at Energy Industries Engineering and Design Company in Tehran, Iran, since 2008. She has participated in the basic design of the LPG recovery and cooling water and caustic dissolving units at the Khuzestan refinery as the deputy of the process lead engineer. Ms. Ardalan holds a BS degree in chemical engineering from Tehran University at Tehran, Iran. She has over 10 years of experience as a process engineer and piping designer in several oil and gas, and petrochemical projects.

Mohammad Reza Khorsand Movaghar has worked in the process department at Energy Industries Engineering and Design Company in Tehran, Iran, since 2008. He holds a BS degree in petrochemical engineering and a PhD in chemical engineering from Tehran Polytechnic University, Iran. Dr. Khorsand also received an MS degree from the University of Science and Technology in Tehran, Iran. He has participated in basic design projects for LPG recovery, crude/vacuum distillation units and fuel gas/oil units of heavy crude oil at the Khuzestan Refinery as a process (simulator) engineer. Dr. Khorsand has over four years of experience as a process engineer and in process simulation in several oil and gas projects.

Mohammad Maleki is currently the process, utility and safety department manager at Energy Industries Engineering and Design Company. He coordinated the Khuzestan refinery’s basic engineering design and development as the project manager. Mr. Maleki received a BS degree in chemical engineering from the University of Texas in Austin. He has over 30 years of experience as a project engineering manager, process and HSE manager, and principal process engineer in several oil and gas, and petrochemical projects. Select 159 at www.HydrocarbonProcessing.com/RS 36

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

Consider new coatings for maintenance turnaround Epoxy system provides speed of application as well as environmental benefits for large storage tanks C. KARNER, SemCrude, Cushing Oklahoma; and B. TOEWS, Sherwin-Williams, Dallas, Texas

hen to coat was the dilemma SemCrude faced as the company was adding new oil storage tanks to its Cushing, Oklahoma, tank farm (Fig. 1). Futures oil prices were moving forward by as much as $1.60/bbl a month. Customers needed tanks to store oil to take advantage of the price increases. Bringing new tanks online as fast as possible was key to both SemCrude and customer profits. Pricing condition for storage.

SemCrude builds and owns tanks that store oil sold on the New York Mercantile Exchange. Monthly futures price fluctuations affect the value of oil stored in the tanks. When prices of a commodity such as oil for future delivery rise, i.e., higher than the current spot price, it is known as a “contango” market. That means higher profits are possible if the oil is sold later rather than sooner. An empty 250,000 bbl oil storage tank in Cushing can store that oil until it is sold—and that makes the tank more valuable as well. Financial dilemma. An empty tank in the tank farm at Cushing can be a hot commodity, depending on the market. So, during a recent contango market, SemCrude was working feverishly to bring new tanks into service. Getting the tanks online as quickly as possible meant immediate cash flow for SemCrude. But the company had to make a decision as to whether to take the time to coat the bottom two-foot section and floor of the tanks’ interiors, or not coat them until a market downturn. The tanks need protection from corrosion; the crude oil to be stored contained a small percentage of salt water that can be sourced from oil wells or from the ship tanks that use seawater to displace the oil as

it is pumped into pipelines. Once in storage, salt water settles out of the light, sweet crude oil to the bottom of the tanks—typically, the bottom two feet, which is coated to prevent corrosion. The rest of the interior remains bare metal. The coating process SemCrude generally used to protect tanks was an epoxy novalac phenolic coating. The two-coat process involved a schedule that could take up to two weeks for the coating to set up, cure and find and repair holidays. The process takes longer in the winter months due to low temperatures. In the market environment SemCrude faced, two weeks was forever. New technology addresses needs with speed. As SemCrude was wres-

tling with the issue of coating, a new

FIG. 1

edge-retentive, ultra-high solids (UHS) epoxy amine protective coating technology could provide the solution. This new coating cut the application and curing time by 12 days. Substituting the traditional coating system for the new one meant that the tanks could be put back into service within two days of coating rather than two weeks. Given the option to save nearly two weeks and provide protective interior tank coating, SemCrude opted to give the new system a try. The new coating technology was specifically designed for immersion service in petroleum storage tanks, fuel/seawater ballast tanks and seawater ballast tanks. It provides single-coat protection, high-build properties, low volatile organic compounds (VOCs), low odor and fast return to ser-

Aerial view of SemCrude’s Cushing, Oklahoma, tank farm operation.

HYDROCARBON PROCESSING DECEMBER 2010

I 37

PLANT DESIGN

MERICHEM COMPANY

vice. The coating dries to “walk-on” levels within four hours. Application is achieved by using a plural spray system. Two pots of epoxy components are heated and pumped into a spray gun, which mixes them at the spray head. The contractor applied the coating to one side of the interior of a 194-ft-diameter tank. The coating set up so quickly that it could be walked on by the time the other side of the tank was finished. Case history. Eight new tanks were

Sweet Solutions.™

Announcing… next generation hydrocarbon-treating technology Problem: Mercaptan Odor Removal. Solution: It’s modular, simple and cost effective, too. Before your crude hits the pipeline, the light mercaptans must go. In remote areas, that’s especially tough. MERICAT™ C uses reliable FIBER-FILM® patented technology to sweeten mercaptan odors, even where access is limited.

brought online utilizing the new coating technology. After the first tank was coated, it was thoroughly inspected to detect holidays. The first inspection revealed only nine holidays, which were quickly repaired. In subsequent tank coatings, there were as few as six holidays. As coating progressed, SemCrude was able to bring new tanks online in less than two days. Another advantage of the new coating system was its low-odor and low-VOC properties. Personnel noted that the coating didn’t have the strong odor of the traditional coating, causing no disturbances. It made everyone feel better about the safety aspects for the applicators. Ontime delivery played key role.

The coating was delivered to the site on a just-in-time basis as it was stocked locally, eliminating the need for SemCrude to store coatings onsite. The contractor ordered what was needed at the end of a work day and the product was delivered the next morning, warmed up and ready for application. The supplier also provided onsite technical personnel to monitor temperatures and to ensure that conditions were correct for coating. HP ACKNOWLEDGMENT The authors thank Guthrie Industrial Coating, Inc., Stillwater, Oklahoma, for their contributions to this article.

Finding the right treating application for hydrocarbon streams is challenging. Merichem’s decades of experience and commitment to innovation means treating gaseous and liquid hydrocarbons is efficient, economical and clean. Learn how sweet it is at www.merichem.com/MERICATC

Carl Karner, senior engineer, SemCrude, L.P., has worked in the pipeline industry since 1972. He has been with SemCrude, L.P. in Oklahoma City for the past five years after working for Texaco Pipeline and Shell Pipeline. He earned BS and MS degrees in engineering from the University of Oklahoma, and has Professional Engineering licenses in both Kansas and Colorado. Bruce Toews is market director, petrochemical and

Merichem: A global provider of focused technology, chemical and service solutions.

P: 713.428.5000 | E: mptsales@merichem.com | www.merichem.com Select 160 at www.HydrocarbonProcessing.com/RS 38

offshore, for Sherwin-Williams Protective & Marine Coatings. He has been with the company 24 years, of which 19 have been in the protective and marine business. Prior to his current position, he served as director of sales, vice president of international, and vice president of global accounts. Mr. Toews is a NACE-certified Coatings Inspector Level III.

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

Operator-driven reliability: Training and implementation Follow these guidelines to help execute the process T. HANLON and T. MCDOUGAL, PeopleSoft, Inc.; Woolwich Township, New Jersey

he benefits of operator-driven reliability programs have been well documented. The operations personnel at a facility are with the equipment 24/7. It seems logical that they are an untapped resource for improving facility economics. At the same time, there are a lot of failed reliability programs out there. How can you make yours successful? O p e r a t o r- d r i v e n r e l i a b i l i t y defined. We all have something in mind

when the term operator-driven reliability (ODR) is used. For this purpose, ODR is the process that augments operator’s traditional duties with monitoring, analysis and care activities to improve plant reliability. It is important to note that ODR is a critical component of an overall maintenance and reliability strategy, and should be integrated with these work streams. Collectively, ODR working within a maintenance and reliability strategy can: • Increase the mean time between failure (MTBF) of the plant’s equipment, • Decrease maintenance costs • Decrease unplanned downtime • Increase total uptime. Operators can impact these key performance metrics in three ways: • Performing predictive monitoring tasks like measuring bearing temperatures or vibration monitoring • Performing simplified root-cause analysis and translating that analysis into effective work orders • Performing ‘“active care” activities like maintaining lube-oil levels or back flushing heat exchangers. The specific tasks that could be performed at your facility in each of the three areas will be unique, although you are probably already thinking of some “low hanging fruit” in the areas of lubrication, steam

conservation and machine condition monitoring. The key to long-term success is to sustain the long-term performance of the tasks identified. High-tech hardware and software can certainly be employed to help the operator perform the previously mentioned tasks. However, it is evident that none can be accomplished without good procedures and good training. Moreover, significant improvement in the performance metrics can be obtained simply by the application of effective training materials and the performance of good procedures—without the investment in high-tech equipment and software. Training materials and procedures. Training materials and procedures

are not the same, although they are often mixed together at many facilities. The difference is that training materials should describe for an operator and include: • How the equipment operates • Why certain tasks must be performed • Basis for troubleshooting. Additional information may describe the use of hand-held equipment, data analysis and the overall work process—although these topics will vary depending on the program design. Procedures should describe the steps required to perform a task. Procedures with action steps broken up by several long notes are an example of training materials mixed with the procedure. Information organized in this manner results in ineffective training and procedures that are difficult to follow. A better approach is to cleanly separate the two. Result: Shorter procedures that are much easier to follow. At a minimum, ODR training materials should address equipment operation, moni-

toring, and care. Target the materials to the audience and separate information into the nice to know and need to know categories. Most nice to know information should be excluded from the training materials. For example, training materials often include equipment’s metallurgy. However, the information is not very useful to the operator. The operator can’t change the metallurgy and must assume that the engineering staff made a good choice for the process conditions. Including this type of nice to know information simply adds volume to the training materials without adding value. On the other hand, it is very important to provide the operator with the normal operating envelope for the equipment. For example, reliability engineers know that operating a centrifugal pump outside of its normal operating envelope can result in a failed seal. If the operator knows the operating envelope and knows that operation outside the envelope can result in seal failure, the operator can increase the MTBF for the pump without any other intervention. This is done simply by ensuring that the pump performance variables stay within the envelope. The operating envelope for auxiliary systems (e.g., lube systems) should also be included. Predictive monitoring parameters such as measurement of vibration or bearing temperature should also be established. As the monitoring plan for an equipment item is developed, determine if failure to monitor a variable will impact one of the performance metrics. If there is no impact, consider removing it from the list of variables to be monitored. This will ensure that the monitoring program is a useful exercise, and not just “busy work.” Proper monitoring follows the establishment of a good operating envelope when the operating envelope is underHYDROCARBON PROCESSING DECEMBER 2010

I 39

SPECIALREPORT

PLANT DESIGN AND ENGINEERING

stood. Training materials should describe how and where to monitor, and tie monitoring information into descriptions of why monitoring is important. In addition, training materials should contain information that will help the operator to troubleshoot equipment. For example, if operators understand the relationship between steam use and vacuum on a condensing turbine, they will likely look for a problem in the vacuum system when steam use goes up. Without knowledge of those fundamen-

tal relationships, the increase in steam use would be a complete mystery. Just as the normal operating envelope establishes the monitoring plan, the monitoring plan establishes most of the plan for operator care—the operator must take actions to keep the equipment item in the normal operating envelope. These actions constitute a large part of the operator care program. For example, if the pump’s bearing box oil bulb should be half full and the operator’s monitoring shows that the bulb is

MORE THAN JUST SHARING YOUR VISION TOGETHER, WE CAN COMPLETE IT.

Agriculture Agri-food Chemicals and Petroleum Environment Facilities and Operations Maintenance Industrial and Manufacturing Infrastructure Mining and Metallurgy Pharmaceuticals

Implementation-getting the program off the ground. That is the

Power Telecommunications

SNC-Lavalin designs, develops and delivers leading engineering, construction, infrastructure and ownership solutions worldwide. We listen carefully to you, and the communities you serve, while striving for excellence in our commitment to health, safety and the environment. We have the global versatility and technical expertise to meet your expectations and complete your vision. www.snclavalin.com

SNC-Lavalin Engineers & Constructors Inc. 9009 West Loop South, Suite 800 • Houston, Texas 77096 • USA • 713-667-9162 • sncl@sncl.us North America

Latin America

Europe

Africa

Eurasia

Asia

Select 161 at www.HydrocarbonProcessing.com/RS 40

Middle East

below half, then the care program included in the training materials probably indicates that the operator should replenish the oil. Operator care in the training materials is fundamentally tied to procedures. When a program is prescribed for operator care, the operator is being asked to take an action. Actions should be defined by procedures. In many cases, actions are considered too simple or too mundane to be covered by a procedure—“It’s just part of the job. We don’t need a procedure.” However, there is no other formal method for operators to learn the actions they must take. If procedures aren’t established for the actions operators perform, new operators must somehow pick up the “tribal knowledge” necessary to perform the action correctly. Are you comfortable with that? Particularly when instituting new practices, which must be implemented to make the ODR program different than status quo, procedures are vitally important. At one refinery, the operators were asked to begin changing the oil in the bearing boxes on pumps. The maintenance personnel who had done this job before didn’t have an established procedure for the task. They probably did fine without one, because they had been performing the task for many years. However, the task was brand new to the operators. Detailed procedures were required to get the program off the ground.

Oceania

difficult part. World class training materials and procedures can be developed without changing the performance metrics one bit. Benefits will not accrue until the knowledge in the materials is applied on the job. How can you make that happen? The first element of a successful application is to establish buy-in on the

part of all those affected. Most facilities are completely saturated with initiatives, so developing initiatives with all the usual trappings (steering committees, etc.) is unlikely to be effective. A better approach is to visit with all the affected parties on a one-on-one basis. Lay out the goals and explain the benefits. Listen to their concerns and modify the plan as necessary. Make sure to include everyone who may be affected. Keep in mind that the benefits are a little different for each group. Additionally, ensure that the individuals responsible for labor relations in the facility are included in the early discussions. Some practices to implement in the ODR

PLANT DESIGN AND ENGINEERING program may be prohibited by the labor agreement. The key is to personalize the activity by talking to the stakeholders at the job site and listening to any feedback. Once the materials are prepared and the program has been communicated, it is time to begin implementation – this begins with training. Identify “trainers” that can also serve in a coaching capacity. Note that “trainer” does not have to be a formal title, and “trainers” can be operators, supervisors or others with ODR expertise and training abilities. Select areas of the facility that have been impacted by lacking good ODR practices. In these areas, the program can have a big impact early, giving the program momentum. Conduct training with operators using the training materials previously described. Take the training to the job site, discuss and demonstrate the actions, and listen to feedback. Feedback from training typically identifies obstacles that prevent sustained implementation of the ODR program. These issues should be addressed through further training, coaching or program modifications. Work with the operators and their supervisors to coach the activities that were described to them in the training. Since the activities are likely to be a change from the normal routine, time should be spent assisting the operators to become accustomed to the new activities and confident in their ability to perform them. It is also important to ensure that front-line supervisors understand the program. They may have some tasks in mind that could be added to the ODR program to make it more effective. Additionally, supervisors giving proper direction to the operators can make the program sustainable. So it is important that supervisors are included in the coaching to the point where they grow accustomed to the program. As the program develops, implementation may be more effective if some job aids were developed. These aids can take many forms. One facility color coded the oil bulbs on pump bearing boxes and matched that with color coded oil containers to ensure that the correct oil was used. There may be a realization that posting written instructions or instructions in a computer-based-training presentation are needed. Hand-held devices are the ultimate job aid, because they can combine monitoring and recording tools with instructions. Operations personnel must also be equipped with the right tools, if they are necessary to do the job. Evaluation. As ODR implementation ramps up, the program must be evaluated

on several levels. First, the operations personnel must be evaluated to ensure that they understand the new training materials and procedures. These evaluations should include both written tests and field exercises. Second, trainers, front-line supervisors and reliability personnel should observe whether operations personnel are performing the ODR on the job. It is important to provide feedback, both good and bad, to all the parties involved—at every level. In the program’s infancy, it is particularly important to catch the operators doing something right and provide positive feedback. Finally, the program’s success must be evaluated. The metrics discussed previously (or others that you feel will measure the success of your particular program) should be measured before and after implementation in each affected area. If the metrics don’t improve, the program needs to be immediately reevaluated. It’s important to understand to continue to do the things that work. It is just as important to stop doing things that don’t work. Summary. Implementation requires a

lot of hard work and attention, applied for a sustained time period. However, the beauty of the approach outlined is that it is solidified by the training materials and procedures. New personnel will always learn the ODR program as they train up in each position. In a few years, your ODR program will be the way you have always done it. HP Ted Hanlon is the director of projects for PeopleCore, Inc. He received a BS degree in chemical engineering from the University of Wyoming. Mr. Hanlon has 30 years of experience in the hydrocarbon and chemical processing industry as an engineer, consultant, and manager in operations and production. He has worked with several facilities in the areas of operations consulting, PSM, training, procedure development and organizational effectiveness.

Tom McDougal received a BS degree in mechanical engineering from Marquette University and has 20 years of experience in the manufacturing, hydrocarbon and chemical processing industries. Mr. McDougal has worked with several facilities in the areas of PSM, training and organizational effectiveness, and previously held management positions at Sunoco Refining and Supply in training and human resources areas, and vice president at a consulting firm. He is the president of PeopleCore, Inc., which provides workforce performance solutions to the hydrocarbon and chemical industries in the areas of organizational effectiveness, operations, and technical training systems, operator-driven reliability, procedure development and implementation and labor relations. Select 162 at www.HydrocarbonProcessing.com/RS

AVEVA Plant Integrated plant engineering and design technology Engineering IT has come of age. The days of inconsistent, disconnected 2D drawings, incompatible CAD formats and ‘over the wall’ project handover are being consigned to the history books. Today, a powerful, integrated and collaborative IT environment supports every stage of project execution – AVEVA.

Whether on complex new-build projects or in-service revamps, the smallest inefficiencies or delays cost real money. AVEVA Plant enables maximum productivity at every stage, reducing costs and timescales, eliminating the causes of errors, waste and rework, and removing limitations on project scale and global collaboration. And it’s the no-risk solution, proven on tens of thousands of projects by many of the world’s most successful engineering businesses.

Find out how AVEVA Plant can make your business more competitive Visit www.aveva.com to learn about the AVEVA Plant solutions, or www.aveva.com/events for opportunities to see them in action.

Head office: High Cross, Madingley Road Cambridge CB3 0HB UK marketing.contact@aveva.com Tel +44 (0)1223 556655 Select 54 at www.HydrocarbonProcessing.com/RS

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

Improve engineering via whole-plant design optimization New simulation methods identify cost-effective advantages early H. MARTÍN RODRÍGUEZ, Repsol (Technology Unit), Madrid, Spain; A. CANO, Process Systems Enterprise, Inc., Cedar Knolls, New Jersey; and M. MATZOPOULOS, Process Systems Enterprise Ltd, London, UK

rocess design requires consideration of many trade-offs. Some equipment decisions may improve the economics of the units under design but negatively impact the project’s total economics, as well as the operability of the plant as a whole. By considering the entire plant simultaneously during design, it is possible to balance decisions that result in the overall best economics. This article describes the design of a propylene oxide (PO) process for Repsol. This facility design included a complex multi-tubular reactor (MTR) and a separation section with many distillation columns (two involving reaction and one using azeotropic distillation), plus large recycle flows. Through model-based engineering (MBE) and optimization techniques, substantial economic advantages to the process development were discovered by the design engineering team. Background. Many companies, including Repsol, Lyondell,

Shell and others, use styrene monomer (SM)/PO technologies to co-produce SM and PO. As shown in Fig. 1a, ethylbenzene (EB) is treated with oxygen to form ethylbenzene hydroperoxide, which is then used to oxidize propylene to PO. The resulting 1-phenylethanol is dehydrated to produce styrene. This is a mature technology, involving a complex process and large capital investment. A major disadvantage for PO producers is the strong dependence of its economics on the variable SM market. Repsol’s policy of diversification led the company to initiate a research project to explore new technologies for PO production via the epoxidation of propylene using hydrogen peroxide (H2O2), as shown in Fig. 1b. Success of this research project led to the development of a commercial-scale process for the production of several hundred thousands tpy of PO. The hydrogen peroxide to PO (HPPO) process has a number of advantages. In particular, it is simpler, requires a lower capital investment and yields less byproducts. Approach to process development. Repsol began the process development project and decided to take a MBE approach to improve process understanding, develop insights, accelerate development and ensure discovering the best possible design with optimal process economics. Development involved a significant experimental phase, including the construction and operation of dedicated pilot plants. Experimental data was then used in conjunction with design and optimization models, combined

with the best engineering practices and commercially available modeling platforms. Model-based engineering. The MBE approach combines

first-principles engineering mathematical models of process physics and chemistry with experimental data to produce high-fidelity predictive models of the key process operations. Once a model of sufficient accuracy is established, the model—rather than the experimental data—is used to optimize process design and operation. This approach allows a much more comprehensive, effective and faster exploration of the process design space than can be achieved by experimentation alone. Also, it allows technology risks to be quantified and addressed systematically. Key to MBE is model-targeted model-based experimentation, where experimentation is focused on maximizing the predictive accuracy of the model rather than addressing individual aspects of process design. The models are used to design further experiments aimed at yielding the maximum amount of information, thereby reducing time and cost involved by the experimental program. Fig. 2 illustrates a typical approach. Applying MBE to new process development has many benefits. The key advantage is the ability to determine the optimal equipment design, operating conditions, overall process design and even optimal operating procedures. Alternatives can be rapidly screened and poor designs eliminated, with only the most promising alternatives going to pilot or demonstration plant testing for verification. The existence of a predictive model makes it easy to accommodate changing requirements and specifications during process design and to re-optimize for new conditions. Fig 1(a) SM/PO: Ethylbenzene to SM and PO C6H5CH2CH3 + O2ûC6H5CH2CH2O2H C6H5CH2CH2O2H + CH3 CH=CH2ûC6H5CH2CH2OH + CH3CHCH2O C6H5CH2CH2OHûC6H5CH=CH2 + H2O Fig 1(b) Hydrogen peroxide route to PO CH3CH = CH2 + H2O2ûCH3 – CH – CH2 + H2O O FIG. 1

Process reaction for producing PO.

HYDROCARBON PROCESSING DECEMBER 2010

I 43

SPECIALREPORT

PLANT DESIGN AND ENGINEERING

Another major advantage is that the combination of models and experimental data can be used to improve the effectiveness of the whole experimentation process. Model-based data analysis provides accurate parameter values as well as estimates of parameter reliability—information that can be used in formal risk assessment and to provide a quantitative assessment of where further experimental R&D effort should be focused. The models capture corporate knowledge in such a way that it can easily be transferred between different groups within the organization, facilitating an integrated design approach. The process. The process is shown schematically in Fig. 3. The

reacting mixture enters an MTR where an exothermic catalytic reaction yields PO and several byproducts. The reaction products are fed to a separation section, where the product (PO) is purified to the required quality. Solvent and unused reactants are recycled to the reactor.

profile from one tube to another. Uniformity of temperature profiles across tubes facilitates optimization of operating conditions to achieve better conversion and fewer side reactions, thus increasing plant throughput and reducing separation costs. It also enhances catalyst life by reducing the likelihood of hotspot formation and the consequent progressive catalyst burnout that results in poor conversion and controllability with early catalyst replacement. Separation section. The separation section is an equally

complex part of the HPPO process. The original design included many distillation columns (one involving azeotropic distillation and two involving chemical reactions). The model of the separation section involved 25 components, including reactants, solvent, product, byproducts and impurities. Beyond the intrinsic complexities of the MTR and separation section, a further layer of complication was introduced at a flowsheet level by the existence of a number of significant recycles, as well as possibilities for heat integration that needed to be considered.

Multi-tubular reactors. MTRs are widely used throughout

the petrochemical/chemical and refining industries for exothermic catalytic reactions. However, their behavior is highly complex, with good design and consistent operation difficult to achieve. An MTR contains a large number—sometimes 20,000 or more—catalyst-filled tubes within a shell through which cooling fluid passes. Often, each tube is divided into several sections, each packed with a different catalyst and/or inert material, and designed to promote different reactions. Heat generated by the reactions in the tube catalyst beds is typically removed by cooling liquid—often molten salt, but, in this case, cooling water—flowing through the shell side. There is strong interaction between shell and tubes: the temperature of the cooling fluid outside a tube influences the rate of reaction inside it, which, in turn, affects the amount of heat produced and transferred to the cooling medium at that point; and different tubes indirectly affect each other via their interactions with the cooling medium. A good design is achieved by adjusting key aspects of the shell and tube bundle geometry to minimize variations in temperature

FIG. 2

Model-based experimentation processes.

I DECEMBER 2010 HYDROCARBON PROCESSING

Process development: project phases. The process

development was done in three major phases: Phase 1: Experimentation. Extensive pilot-plant experimentation was undertaken to identify chemical kinetics and determine heat transfer characteristics. It was guided and supported by models of the experimental equipment. Phase 2: Process design and optimization. Detailed equipment and process models incorporating the physical and chemical phenomena identified during Phase 1 were used to construct the whole-plant flowsheet model. This model was then used for the evolutionary design of the commercial-scale process. Phase 3: Heat integration. The Phase 2 design was further improved by considering heat integration options in the optimization. The initial plan for the development of the new process envisaged the reactor and separation section being designed and optimized sequentially. However, it was quickly realized that, because of the close coupling of reaction and separation stages, it was essential to optimize the design of both sections simultaneously. This is a general characteristic of processes involving reaction and

FIG. 3

The new PO process schematic.

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

separation.1 Optimizing the reactor separately from the separation section can result in large recycle flows, which will increase the capital and operating costs of the separation section. Conversely, maximizing conversion in the reactor in an attempt to reduce separation costs will typically result in a reactor that is too large. The optimal level of conversion (and, therefore, the optimal reactor size) can be found only by considering reaction and separation costs simultaneously. Phase 1—Pilot-plant experiments.

Measurements predicted by model

This phase was to accurately identify chemical kinetics and determine heat transfer characteristics. Experimentation was done on a single-tube pilot plant. Over 60 pilot-plant experiments were conducted with various catalysts. The experimentation phase folFIG. 4 Phenomena considered in the catalyst pellet and fixed-bed model. lowed the classic step-by-step model-targeted experimentation approach shown in Fig. 2: Step 1—Build first-principles model of experimental apparatus. A detailed model of the tubular reactor 12 experimental rig was developed using off-the-shelf models that took account of all relevant heat and mass transfer phenomena, as shown 10 in Fig. 4. This included limitations in the mass-transfer rates to the pellets and, more importantly, for liquid-phase systems—where the diffusion coefficients are very low and the intra-particle transport 8 limitations are severe—in the catalyst pores within the pellets. The catalytic reaction was thus modeled using a detailed catalyst pellet model with radially distributed intra-pellet variations. The 6 catalyst-filled tube was represented by an axially and radially distributed model. A rigorous approach was used for the modeling of the thermodynamic and transport properties of the reaction mixture. 4 Step 2—Estimate parameters using model of experimental apparatus from Step 1. The model from Step 1 was used to estimate reaction kinetic and heat transfer parameters from experi2 mental data. Rigorous mathematical techniques for simultaneous estimation of multiple parameters in nonlinear models using 0 multiple experimental data sets were used for this purpose. 8 10 12 0 2 4 6 Step 3—Perform model-based data analysis. The confidence Laboratory measurements (uncertainty) information generated for each parameter by the FIG. 5 Predicted component lab values (y-axis) against the actual Step 2 parameter estimation was used to determine where the measurements (x-axis) for 38 experiments. most significant areas of risk inherent in the data were, and thus where subsequent experiments should be focused. Step 4—Design additional experiments if necessary. Using spheric conditions, and the second the increase in temperature the model constructed in Step 1, it was possible to propose several to ambient during gas chromatograph injection. Modeling the new experiments that would maximize information content in sensors is an important requirement for correct mass balance critical areas—for example, by replacing propylene by PO in the reconciliation and interpretation of lab data. The model used for reactor feed, and co-feeding PO and propylene. data analysis should reflect the physical reality of the experimental Steps 2 to 4 were repeated until the estimated values of key setup as closely as possible. parameters (primarily the main reaction kinetic parameters) were considered sufficiently accurate. Fig. 5 shows the accuracy of fit for Feedback to experimentation. Following initial use of concentration of one of the reaction products. the model to analyze data gathered earlier, the project team was able to make numerous recommendations with regard to the Pilot-plant model. Fig. 6 shows the flowsheet constructed in experimental setup. These included replacing cooling coils with Step 1 to represent the pilot-plant reactor. The reactor tube comprised a cooling jacket operated in the turbulent regime, improving a central section (about a meter long) filled with catalyst pellets, with control, adding thermocouples for temperature measurements inert bed sections placed at the beginning and end of the tube. at locations determined by the analysis, and modifying sampling The two flash units simulate the sampling procedure. The approaches. It was also possible to make recommendations on the first vessel simulates depressurization from high pressure to atmocatalyst pellet design to minimize side reactions. HYDROCARBON PROCESSING DECEMBER 2010

I 45

SPECIALREPORT

FIG. 6

PLANT DESIGN AND ENGINEERING

Pilot-plant model showing inert sections, catalyst bed and cooling jacket.

1-D

Catalyst particle 1D model Catalytic reaction Multicomponent diffusion Heat transfer

2-D axial

radial

the core of the commercial reactor model for process design. As mentioned earlier, it was important to consider the trade-off between the reaction and separation economics. As a consequence: • All work during the design phase considered the reaction and separation sections (including recycles) simultaneously. • Due to the large number of decision variables involved, the design was posed formally as an optimization problem and solved using rigorous mathematical techniques, in preference to attempting a trial-and-error analysis based on repeated simulations involving the variation of a few variables at a time.

Modeling the reactor. The optimal design of the MTR, involving complex phenomena occurring within a large number of tubes as well as complex cooling fluid behavior on the shell side, presents a significant challenge in itself—even before integration with the separation section is considered. Model-based techniques can help to ensure an axial tube-side temperature profile that favors the desired reactions. To achieve this, it is necessary to accurately quantify heat transfer at all points throughout the reactor, taking into account the fluid dynamics on the shell side and reactions occurring in the tubes simultaneously. The model of the complex liquid-phase MTR to be used in the full-scale new PO process was constructed using commercially available catalyst and fixed-bed reactor component models (Fig. 7). As the shell side comprised a number of compartments separated by baffles, lumped cooling models were used, with each compartment being assumed to be perfectly mixed. Modeling of separation section. The base case separation

Catalyst-ﬁlled tube 2-D (axial, radial) Bed and wall heat transfer Mass transfer

Shell Bafﬂed compartments Detailed geometry Heat transfer Hydraulics

FIG. 7

Reactor sub-models: Key phenomena and relationships shown.

Interestingly, the initial results also identified and quantified biases in thermocouple readings, effectively using the predictive model as a calibration instrument to correct pilot-plant measurements. This allowed more precise estimation of heat transfer and kinetic parameters to be obtained in subsequent experiments. The ability to use information from the model-based data analysis to identify improvements to experimental setups helps to maximize the quality of information collected. This is one key advantage of the MBE approach. Phase 2—Process design and optimization. Once

acceptable parameter accuracy was attained through experimentation, the validated model of the key phenomena could be used as 46

I DECEMBER 2010 HYDROCARBON PROCESSING

section design was taken from a steady-state design obtained using standard flowsheeting tools. This flowsheeting model was implemented within the design modeling environment using similar steady-state equilibrium distillation column models. On initial analysis, it quickly became apparent that the calculated performance and economic of the process were heavily dependent on the accuracy of the thermodynamic model. To improve its accuracy, binary interaction parameters in the activity coefficient model were fitted for 21 binary pairs using data from commercial databanks, as well as from targeted vapor-liquid equilibrium (VLE) experiments carried out in-house or commissioned to external laboratories. The result was a set of highly accurate parameters for the composition ranges of interest. Model implementation—overall. The models described

above were implemented in commercially available process design software capable of handling the physical and chemical complexity of the MTR, the convergence challenges of the recycles, and numerical complexities of the optimization. The flowsheet was constructed from standard model libraries. A custom-cost module was added for the optimization calculation. Reactor model and separation models were constructed and initially solved separately. The models were then integrated and recycle loops closed. Once an initial solution was reached, the team had a model that was capable of being used for optimization for the whole plant design.

PLANT DESIGN AND ENGINEERING

SPECIALREPORT

TABLE. 1. Optimization decisions regarding reactor design 1. Tube pitch (relative to tube outside diameter) 2. Minimum average coolant velocity within tube bank

Reactor diameter Tube length

3. Mass fraction of H2O2 in the feed stream (a measure of the amount of solvent used)

4. Molar ratio of PP to H2O2 in the feed 5. Concentration of H2O2 in the H2O2 raw material stream 6. Number of active reactors 7. Number of baffles in the reactor shell

Bafﬂe window size

8. Tube length

9. Cooling water inlet temperature

Bafﬂe span

10. Space time (reciprocal of kmol H2O2 /hr per kg catalyst) 11. Reactor process stream inlet temperature.

Outer tube limit Inner tube limit

Optimization. The optimal design of the new PO process was

Tube size

determined via an economic-based optimization calculation that maximized the annual profit,1 i.e., the difference between the value of the products created and the total annualized cost (TAC). Any such design must satisfy a number of key constraints and this is obtained by resolving a large number of design and operational decisions that are available to the designer. These are discussed later. Design objective function. TAC brings together recurring operating costs (€/yr) with capital costs (€/yr), by applying an annualization factor (AF), which has units of year-1 and represents the trade-off between capital investment costs and operating costs. High values of the AF bias designs in favor of smaller equipment and larger consumption of raw materials and utilities. In its simplest form, the AF is equal to the interest rate used to evaluate capital investments. This would assume infinite equipment life without the need to replace it. It has been shown that net present value (NPV) formulae can be translated into the TAC framework by a suitable definition of the annualization factor.2 AFs used can be as high as 1/yr. For the current project, a value of AF = 0.4/yr was selected. Total annualized profit (TAP) maximization was chosen as the objective function for the design optimization. As the production rate was fixed as a design criterion, maximizing TAP gives the same results as minimizing TAC. The objective function was formulated as: TAP = Annual operating margin – Total equipment cost ⫻ AF where: Annual operating margin = Annual revenue – Annual cost of raw materials – Annual cost of utilities – Annual cost of catalyst Constraints. Any valid design and operating conditions must

satisfy a number of key constraints, including: • Quality constraints on product, for example, maximum ppm levels of several impurities • Safety constraints, for example, maximum oxygen concentrations—arising from decomposition of unreacted H2O2—in potentially explosive streams • Coolant supply and return temperature constraints • Effluent concentration constraints determined by environmental considerations • Reactor shell-side minimum and maximum velocity constraints required to prevent tube fouling and tube erosion, respectively

Arrangement, pitch

Coolant ﬂowrate

Coolant temp

9 3 FIG. 8

5 11

MTR key design variables.

• Constraints on reactor and column dimensions, due to road transport considerations for equipment manufactured offsite. Optimization decision variables. These are the design

parameters—equipment dimensions, operating conditions and so on—that are varied during the optimization to maximize the value of the objective function. In the final design stages, there were 49 decision variables in total. Most of these were continuous variables—for example, reflux ratios that could be varied between a minimum and maximum value during optimization. However, the optimization included some discrete (or integer) decisions, for example, feed tray location or “on/off ” variables representing stream routing decisions. The optimization decisions relating to the reactor design included, as shown in Fig. 8 and Table 1. The optimization variables relating to the separation section design were divided into three categories, as summarized in Table 2. Different optimization variables were used for each column, as shown in Figs. 9 and 10, depending on the key 1

The term “process optimization” is often used loosely to mean any form of process improvement. The optimization referred to here is a true mathematical optimization, i.e., determining the mathematically-best solution given the data provided. The validity of the optimal solution of course depends highly on the predictive accuracy of the underlying model. HYDROCARBON PROCESSING DECEMBER 2010

I 47

SPECIALREPORT

PLANT DESIGN AND ENGINEERING 2

3 5

Pressure

Reﬂux ratio

Product specs

$£ € ¥ Calculator

Product specs Bypass

Reﬂux ratio

$£ € ¥ Calculator

10 Candidate feed locations

Annualized capital + operating cost

6 Reboil ratio 7

Reboil ratio

10 Product 11 FIG. 9

specs

Typical distillation column key design variables— operating conditions and specifications.

TABLE 2. Optimization variables for separation section of new PO design

Stream for heating

14 FIG. 10

Product specs Typical distillation column key design variables—product goals and configuration.

design criteria for the column; between two and four variables were used per column.

Category 1—Operating conditions 1. Condenser reflux ratio 2. Condenser temperature 3. Condenser pressure 4. Condenser temperature approach 5. Column top pressure 6. Reboiler boil ratio

Optimization methodology. The optimization phase

helped develop a deep understanding of the process. More than 10 optimization runs were done, each providing new insights into the process and its design, and leading to the consideration of more design alternatives via introducing new decision variables. Thus, each successive optimization resulted in improved process economics.

7. Reboiler temperature

Verifying design via hybrid process/CFD modeling.

Category 2—Product specifications*

The MTR model applied for the optimization-based design made use of velocity and pressure-drop correlations for modeling the shell side. Based on empirical data, such correlations allow a reasonable account of reactor behavior to be incorporated within a whole-plant optimization framework. However, once the final design is obtained, it is highly desirable to perform a full hydrodynamic calculation to provide an extra check on whether all relevant constraints are met. In the case of the new PO process, one important consideration related to the minimum velocities of the coolant throughout the shell. Because changing conditions within the shell also affect the tubeside behavior, the entire reactor design needed to be validated. Accordingly, detailed computational fluid dynamics (CFD) models of the shell side were implemented and then combined with the tube-reaction models, using a proprietary hybrid modeling interface (Fig. 11) to produce a fully coupled hybrid model of the MTR.3

8. Concentration of component x in the overhead distillate stream 9. Fraction of component x entering the column that is recovered in the vapor distillate stream 10. Mass fraction of component x in the bottoms stream 11. Mole fraction of component x in the bottoms stream Category 3—Column or flowsheet configuration (Fig. 10) 12. Fraction of feed stream not bypassing the column. This was used to allow the optimizer to make structural decisions on flowsheet alternatives (for example, to allow for potential removal of some columns in the original flowsheet). 13. Feed tray location. The optimizer was allowed to consider a number of alternatives, selecting the one that gave the economic maximum. 14. Heat integration options (see Phase 3). *These

were applied to the intermediate streams between columns (the final product stream qualities are constraints rather than optimization variables).

I DECEMBER 2010 HYDROCARBON PROCESSING

PLANT DESIGN AND ENGINEERING The hybrid model was also used to decide between configuration alternatives for baffle type (disc-and-doughnut vs. singlesegmental configurations) and distributor design (homogeneously distributed inlets vs. distributors comprising a discrete number of slits) and showed that the chosen options were indeed viable in terms of the detailed fluid flow characteristics. Fig. 12 lists selected results.

SPECIALREPORT

CFD model of shell side

Phase 3—Heat integration. In the final stage of the design, pinch analysis was carried out. Heat integration routes were introduced where there was any potential benefit, with the optimizer allowed to select process streams instead of utilities in certain reboilers to reduce the total steam cost.

Model of catalytic reaction in tube

Heat ﬂux distribution throughout shell Wall heat-transfer coefﬁcient; coolant temperature

Optimal new PO process design. The optimal design

represented large savings in operating and capital cost with respect to the initial base case. Two columns were eliminated entirely from the separation section. One was found to have negligible separation effect; the other, used for treating a waste stream, was found to have a steam cost that exceeded the cost of sending the stream directly to wastewater treatment. The final design for the MTR ensured that tube behavior and performance were practically uniform across the entire tube bundle. The heat integration yielded significant operating cost savings with attractive return on investment; payback was less than four months.

FIG. 11

Combining shell-side CFD and tube catalytic reaction models for hybrid simulation via proprietary software interface.

FIG. 12

Selected results from the CFD verification step for the MTR coolant side: from left to right, profiles for coolant velocity, coolant pressure drop and PO conversion.

Conclusion. During the course of the design optimization

described in this article, the value of the TAP increased by several tens of millions of €/yr, with approximately €5 million/yr saved by eliminating just one distillation column. The optimization resulted in an economically viable and robust process. At the outset of the project, the practicality of applying this level of design optimization to an MTR or separation section separately using detailed first-principles models was virtually unknown. Even less clear was the feasibility of optimizing the combined process. In addition to delivering an economicallyoptimal design, the project proved that: • Whole plant optimal design is now a viable technology and capable of significant savings—both in terms of capital expenditure (CAPEX) and operating expenditure (OPEX). • Models are sufficiently accurate to be able to reflect the effects of small changes in physical configuration, for example, tube pitch or tube diameter in the MTR. • Complex configuration decisions, such as the location of distillation feed trays, can be taken into account. • Optimization technology is suitably powerful and robust, and capable of large optimizations that include combinations of many continuous and discrete decisions. • Model-based data analysis helps to make the experimentation phase shorter and more effective. • MBE can be applied from the experimental stage onwards. Indeed, the earlier model-based techniques are brought to bear during process development and design, the better. The methods used in this design are suitably general and can be applied to any process plant and they are available via commercially-available simulation and modeling tools. Although the case presented here involved process design, the methodology applies to both design and operations. In fact, the rigorous models used for conceptual design are also suitable for detailed engineering and operation stages of the project. HP

1 2

LITERATURE CITED Douglas, J. M., Conceptual Design of Chemical Processes, McGraw-Hill, 1988. Cano-Ruiz, J. A. and G. J. McRae, “Environmentally conscious chemical process design,” Annual Review of Energy and the Environment, Vol. 23, November 1998, pp. 499–536. Shin, S. B., et al., “Optimize terephthaldehyde reactor operations,” Hydrocarbon Processing, April 2007, pp. 83–90.

Hilario Martín Rodríguez is a senior technologist at Repsol, Madrid, Spain. He has 13 years of experience in process design, modeling, optimization and applied mathematics in the oil & gas, refining and chemical industries. He is also Associate Lecturer at the Chemical Engineering Department at Valladolid University, Spain. He has a M.Sc. degree in mining engineering from the Madrid Polytechnic University, Spain. Alejandro Cano is a principal consultant at Process Systems Enterprise Inc., Cedar Knolls, New Jersey. He has 10 years of experience in the application of highfidelity modeling to complex processes involving chemical reactions and separations across many different industries. He has a Ph.D. in chemical engineering from the Massachusetts Institute of Technology.

Mark Matzopoulos is chief operating officer at Process Systems Enterprise Ltd, London, UK. He has 30 years of experience in the development and application of process simulation and modeling tools in steady-state and dynamic simulation and plant optimization in the oil & gas, refining and chemical industries. He has a chemical engineering degree from the University of Cape Town, South Africa. HYDROCARBON PROCESSING DECEMBER 2010

I 49

Anytime, Anywhere

Anchor-Loc 3 insulating fiber

modules keep your project on schedule with consistent, reliable performance anywhere in the world. ®

Anchor-Loc 3 insulating fiber modules are specifically designed and manufactured for the global market, providing a winning combination of performance, reliability and consistency you can count on anywhere in the world. This new generation ® Insulating Fiber Modules of Anchor-Loc modules is designed to meet a wide range of application requirements in a variety of heat processing vessels. They provide continuous S-folded blanket construction for improved thermal performance in high temperature applications and are available in various fiber chemistries, temperature grades and densities to meet the most demanding requirements. Anchor-Loc 3 modules offer:

Consistent design & quality assurance Anchor-Loc 3 module design features construction from a continuous fold of spun blanket, stainless steel alloy hardware and center mount attachment. The design allows for consistent furnace layout, ease of installation and dependable service life. All components meet or exceed established industry standards assuring the same high quality worldwide.

Fast, cost-effective delivery The Unifrax sales team provides design recommendations, engineering layout and product sourcing options, ensuring a costeffective furnace lining solution wherever you’re located in the global market.

Universal specifications

For more information and a list of our worldwide manufacturing locations, visit the Unifrax website, call 716-278-3800 or email anchorloc3anytime@unifrax.com.

Anchor-Loc 3 modules are produced in each of our global manufacturing centers using the same raw material specifications, dimensional tolerances and assembly procedures, providing product uniformity and consistency worldwide.

www.unifrax.com

Select 68 at www.HydrocarbonProcessing.com/RS

HEAT TRANSFER

BONUSREPORT

Prevent flow-induced vibration in heat exchangers New technology can be used in existing tube bundles to enable higher flowrates without risk of tube damage A. S. WANNI and Z. F. RUZEK, ExxonMobil Research and Engineering Co., Fairfax, Virginia

ptimization of an existing process plant sometimes referred to as revamp or debottlenecking, can result in increased flowrates and higher velocities through the existing facilities, including shell-and-tube heat exchangers. While a higher velocity results in an elevated pressure drop, research and plant operations data have shown that it often has the advantage of reducing heat exchanger fouling. However, on the negative side, higher velocity on the shell side of the exchanger increases the potential for tube vibration. Problems within. Potentially damaging flow conditions

may result in flow-induced vibration (FIV) tube failures due to fluidelastic instability and/or vortex shedding. Tube failures are usually very expensive and lead to loss of production, contamination of products, additional energy usage, and high heat exchanger repair or replacement cost.

then the â&#x20AC;&#x153;plug-and-runâ&#x20AC;? approach is ineffective and leads to further failure incidents over the short time. This is because, often, a whole region of the bundle is vulnerable to FIV. Therefore, when an exchanger tube fails, it is always prudent to carry out a root-cause-failure analysis to identify if and why FIV is occurring and what corrective action is warranted to improve the reliability of the exchanger. Often the solution is a new, bigger exchanger which might also require modifications to the foundation and piping, or a new bundle with additional baffles which result in an increased shell side pressure drop. Alternatively, in the cases of predicted vibration, the bundle can often be removed; the anti-vibration technology installed in less than a day; and the bundle reinstalled into the same shell.

Solutions. New anti-vibration technol-

ogy (AVT) solutions can be used to address existing or potential vibration problems. These technologies have been applied and successfully used in many applications worldwide. AVT can be applied in retrofit applications or new designs. In retrofits, AVTs can improve reliability of an existing bundle that has already suffered some vibration damage, provide tube vibration mitigation to an existing bundle predicted to have vibration problems at a future increased throughput, or be applied to a replacement bundle thus allowing for re-use of the existing shell. In new design applications, these technologies can be used to reduce the heat exchanger size, or to reduce pressure drop. For either of these cases, the AVT provides a very cost-effective solution. Failure situations. Very often, when heat exchanger tubes fail, the failed tubes are plugged, and the exchanger is put back in operation. However, if flow-induced vibration (FIV) is the cause of tube failure,

FIG. 1

Dimpled tube support strips inserted in U-bend tube bundle.

HYDROCARBON PROCESSING DECEMBER 2010

I 51

BONUSREPORT

HEAT TRANSFER

Dimpled tube support (DTS) technology. Dimpled

and corrugated straight-metal strips that are inserted into heatexchanger bundles to stiffen them and to eliminate tube chatter. Both dimpled and corrugated regions act as a wedge and slightly deflect the tubes thereby stiffening the bundle and avoiding tube chatter. The dimples also provide a locking mechanism to hold the strip in place. These strips are inserted into alternating lanes of the bundle in regions with a high risk of damage by FIV. The strips can be used in all common tube layouts and, because of the locking feature, can be applied in the U-bend areas and vertical bundles as well (Fig. 1). The strips have been successfully fabricated from many different metals including SS304, SS316, titanium, duplex, carbon steel, Monel and brass. Saddled tube support (STS) technology. Corrugated pairs of identical thin metal strips are welded together and inserted into alternate tube lanes in a heat-exchanger bundle. A “saddle” portion of the tube support technology provides a wide contact area against the tubes instead of a line contact. This tube-support technology is especially beneficial for application with integral-fin tubes made of softer metals such as carbon steel and brass; the saddles contact multiple fins and prevent any fin damage. This technology can only be applied 90° and 45° tube layouts. Case histories. Here are several examples in which AVT solu-

tions were successfully applied: Case History 1: Repair of existing exchanger. Soon after the startup of a new liquefied natural gas (LNG) plant, several kettle reboilers suffered vibration-induced tube failures owing to insufficient tube support. Inspection of the exchanger revealed that TABLE 1. Summary of design with new anti-vibration technology for optimum heat transfer conditions Exchanger: TEMA AES condensing service

Conventional design, with no vibration

Optimum design with new technology

44 in. x 192 in. Plain tubes, NTIW/Seg, 2p/1s

36 in. x 192 in. Plain tubes, Double seg, 2p/1s

Duty, MBtu/hr

17.7

Shell-side ⌬p, psi

1.8

1.4

more than 10% of the tubes had significant tube fretting at the baffle-hole locations. Some tubes were completely holed through. In this case, the heavily fretted tubes were plugged and the DTS strips were installed to stabilize the remaining tubes. A very high probability of fluidelastic instability at the top 12 rows of the bundle was identified during analysis which also confirmed why failures had occurred in this section of the bundles. The fluidelastic instability ratio (FIR) was calculated to be greater than 4.0 for the current conditions; generally, a value of less than 0.8 is recommended for reliable operation. Detailed vibration analysis determined the locations and number of DTS strips required and the use of DTS strips decreased FIR to 0.2. Since the modification, the exchangers have operated successfully for more than five years. This plant realized significant savings by preventing unplanned downtime and avoiding replacement cost for the exchanger bundles. Case history 2: New exchanger design. Sometimes the optimum design (i.e., the smallest exchanger size) for a specified heat transfer, subject to the allowable pressure drop limit, will not be acceptable owing to potential vibration problems. Often, the conventional solution is to reduce the shell-side velocity by changing baffle type, or spacing, or by increasing the shell size. This will result in an off-optimum design (lower heat transfer performance and a larger and more expensive exchanger). In this situation, the designer may use the DTS strips strategically in the vulnerable regions of the bundle and stay with the original optimum design. Table 1 summarizes the results for application of DTS technology to enable the optimum design of an overhead condenser. During the design phase, an optimum design was identified with a pair of 36-in. shells. The pressure drop limit was satisfied, but a vibration problem was identified. To avoid the vibration problem, the design would have been modified and would require a pair of 44-in. shells with no-tubes-in-window baffles. Instead, designers opted to install DTS technology to alleviate the vibration problems; thus a smaller exchanger could be used. Summary. Commercially proven AVT solutions are becoming accepted in industry. These solutions have been included as an option in Heat Transfer Research, Inc.’s (HTRI’s) vibration analysis software and are considered among standard FIV mitigation options when evaluating vibration potential in shell-and-tube heat exchangers (Fig. 2). HP

Amar S. Wanni received his Ph.D. in mechanical engineering from the Pennsylvania State University, in 1981. He has carried out extensive research studies in the field of heat transfer and has published over 30 articles. Dr. Wanni has also worked for two companies (HTRI and Simulation Sciences) that market heat exchanger design and process simulation software packages. Since 1997, he has been a senior heat transfer specialist at ExxonMobil Research and Engineering Co. Dr. Wanni has over 30 years of experience in designing, monitoring and troubleshooting of heat exchangers in a wide variety of services. He also has over 15 US and foreign patents related mainly to heat exchangers, including vibration mitigation

Zdenka F. Ruzek is currently the licensing coordinator at ExxonMobil Research

FIG. 2

HTRI—screen shot showing FIV mitigation applications.

I DECEMBER 2010 HYDROCARBON PROCESSING

and Engineering Company (EMRE), Technology Sales and Licensing, located in Fairfax, Virginia. As a licensing coordinator, she provides a variety of sales and project coordination services to EMRE licensing customers. Ms. Ruzek has over 30 years of experience in petrochemical industry, primarily in capital project execution. She holds MS degree in engineering (industrial and operations research) from Columbia University in New York, and BS degree in engineering (Cartography) from CVUT (Czech Technical University) in Prague.

HEAT TRANSFER

BONUSREPORT

Optimize heat exchanger installations This case study proves that a systematic approach for thermal design can identify substantial savings in projects M. MANDAL, Fluor Daniel India Pvt. Ltd., New Delhi, India

he optimal thermal design and selection of heat exchangers can be very challenging; it involves many variables. The thermal design of an exchanger with very viscous and high-fouling fluids in a refinery service can be quite complicated as compared to other condensing or reboiling services due to high costs and maintenance problems. In a refinery, high viscosity and high-fouling materials come from the crude column or vacuum-column bottoms (high-temperature zone). These materials have higher temperatures and, therefore, they offer a high energy potential. To use this high energy level, these streams are invariably used as heating media in preheat trains of crude and vacuum units. The challenges that a thermal designer faces when handling such fluids include: • Selecting the optimum number of shells • Utilizing pressure drops across the exchanger • Cleaning and maintenance • Maximizing and sustaining thermal performance over the years • Optimizing the cost of procurement. This article discusses a case study where substantial savings were realized via optimizing heat exchangers handling viscous and fouling fluids.

higher viscosity and lower flowrate. This design resulted in several anomalies. Tables 1 and 2 list the salient design features for this exchanger network. Thermal design revised. To make the basics correct, the thermal reviewer changed the fluid sides in 45° (rotated square) layouts using software and checked the performance. The HVGO side heattransfer coefficient increased from 600 to 850, thus increasing U, the overall heat transfer coefficients, from180 to 230. Therefore, the surface-area requirement dropped from 5,050 m 2 to 3,900 m 2 , and the number of tube passes decreased from eight to four. Shell-side and tubeside velocities were quite adequate 0.5 m/ sec and 1.7 m/sec, respectively, and the pressure drop on the shell side and tube side were less than allowable. Due to the lower surface area, the number of tubes per shell dropped from 1,380 to 1,060 and the shell size from 1,500 mm to 1,300 mm ID. Table 3 summarizes these results. Since the client preferred to use 19.05-mm tubes, this design was further improved with the smaller shell size. However, since tube-side pressure was exceeding the maximum allowable pressure drop with four passes, the number of passs was reduced to two and tube side velocity was

Case study. During the design-review

TABLE 2. Thermal design data

stage, a thermal reviewer from the project management consultancy team/client noticed that the surface area requirement was exceptionally high. The total design heat-transfer surface area was more than 5,000 m2 as the exchanger was designed with eight shells (four in series, two in parallel combinations). When reviewed, it was noted that, instead of routing heavy vacuum gasoil (HVGO) with lower viscosity and higher flowrate on tube side, the atmospheric residue (AR) was routed through the tube side, which has a comparatively

1 m/sec. Therefore, there was no further gain in surface area. Cost saving. The overall comparison of the original design and the new design are listed in Table 4. Analysis. Viscosity played a major role in revising the existing heat exchanger design. Generally, the higher the viscosity of the operating fluid, the greater the difficulty in the heat exchanger optimization. Invariably, the highly viscous fluids are also dirtier fluids in terms of fouling resistance and, thus, it acts like a double-edged sword on the thermal heat exchanger designer. Viscosity and fouling potential almost always TABLE 1. Salient process data Particulars

Shell side

Tube side

Process fluid

HVGO, liquid

AR, liquid

40.87

Heat duty, Gcal/hr Flowrate, kg/hr

967,000

595,000

Temperature, °C

308 to 244

167 to 281

Viscosity, cp

0.29 to 0.63

14 to 2.2

Operating pr., kg/cm2 abs

10.2

Allowable pressure drop, kg/cm2

2.75

5.5

Fouling resistance, hr m2 °C/kcal

0.0001

0.0012

Particulars

Shell side

Tube side

Process fluid

HVGO (liq.)

AR (liq.)

Calculated pressure drop,

kg/cm2

1.3

4.2

Velocity, m/sec

0.76

1.4

MTD

45.3

U Effective surface area,

179 m2

5050

Shell details

1,500 mm ID, 6,096 mm long

Tube details

1,380 nos., 25.4 mm OD, 16 BWG, 6,096 mm long, 31.75 mm pitch, square layout

Material of construction

Shell: carbon steel (CS) clad 317L stainless steel (SS), tube sheet: 317L SS, tubes: 317L SS channel/channel cover: A/B/E/F- CS, C/D/G/H- CS clad 317L SS HYDROCARBON PROCESSING DECEMBER 2010

I 53

BONUSREPORT

HEAT TRANSFER

go hand in hand except for lube oils. The challenges faced by a thermal heat exchanger designer when designing a heat exchanger handling both high viscosity and fouling fluids are: • Minimizing heat transfer area • Optimizing number of shells in series and parallel arrangement • Choosing the correct sides for fluids • Maximizing the velocity in the shell and tube sides • Maximizing the calculated pressure drop utilization within allowable pressure drop

• Exchanger geometry adopted to increase NRe. Maximize heat transfer area. This is most important because the greater surface area means more materials required as the tube surface is linked to the cost of tubes, cost of shell, higher tube-sheet thickness, larger channel ID and thickness, etc. Higher tube-surface area leads to higher shell diameter as tube length, in most cases, will be limited to clients’ specification and market availability. Larger shell diameter will mean higher shell thickness. When tube material is stainless steel

TABLE 3. Revised thermal design data Particulars

Shell side

Tube side

Process Fluid

AR, liquid,

HVGO, liquid

1.6

1.7

Velocity, m/sec

0.5

1.7

MTD

45.2

Calculated pressure drop, kg/cm2

230

Effective surface area, m2

3900

Shell details

1,300 mm ID, 6,096 mm long

Tube details

1,060 nos., 25.4 mm OD, 16 BWG, 6,096 mm long, 31.75 mm pitch, Rotated square layout

Material of construction (proposed)

Shell: A/B/E/F-CS, C/D/G/H-CS clad 317L SS, tube sheet: A/B/E/F- CS clad 317L, C/D/G/H- 317L SS, tubes: 317L SS, channel/channel cover: CS clad 317L SS

(SS) instead of carbon steel (CS), the savings on surface area has more value as SS costs are much greater than CS. In this case study, both the shell and channel diameter were reduced, and the tube-surface area decreased from 5,050 m 2 to 3,950 m 2. Therefore, the overall bundle weight was reduced. The SS weight being proportional to weight, the cost of the equipment was much lower. Optimizing shell in series and parallel arrangement. When the total heattransfer area required for a heat exchanger is greater than the maximum heat-transfer area that can be incorporated in one shell, multiple shells are applied. When removing the bundle, the maximum heat-transfer area is determined by the maximum permissible tube bundle weight, which, in turn, depends on the crane’s carrying capacity. In larger plants such as 6 million tpy to 9 million tpy refineries, the tubebundle weight can be 15 tons to 20 tons, and the maximum heat-transfer area will increase accordingly. For fixed-tube-sheet exchangers, bundle weight is not a constraint so that a very large heat transfer (1,000 m2 or greater) can be incorporated in a single shell. In this case study, since the heat duty was quite large, the surface-area requirement was also high.

TABLE 4. Overall comparison between original design and revised design with cost saving Original design Particulars Fluid type

Shell side

Revised design Tube side

Shell side

Tube side

HVGO

Flowrate, kg/hr

967,000

595,000

967,000

Temperature, °C

308 to 244

167 to 281

No. of shells required Calculated surface area, m2 Each shell size, mm x mm

8 (4 in series 2 in parallel)

308 to 244

8 (4 in series 2 in parallel)

5,050

3,950

1,500 x 6,096

1,270 x 6,096

Tube size, mm

25.4

19.05

Tube pattern

90° (square)

45° (rotated square)

179

222

Shell

CS + SS 317L Clad

Coldest 4 shells CS, Hottest 4 shells CS + SS 317L Clad

Tube

SS 317L

Tube sheet

SS 317L solid

Coldest 4 shells CS + SS 317L Clad, Hottest 4 shells SS 317L Clad both side or Solid SS317L

No. of pass

U value Material of construction

HVGO velocity, m/sec

0.75

1.0

Atm. residue velocity, m/sec

1.47

0.53

HVGO pr. drop, Kg/cm2 (allow. 1.75)

1.75

0.55

Atm. residue pr. drop, Kg/cm2 (allow. 4.5)

4.4

2.1

Total cost, million, Euro

5.9

Total savings, million, Euro

—

2.1

I DECEMBER 2010 HYDROCARBON PROCESSING

HEAT TRANSFER Within client specified limits of 6,096 mm tube length, the number of shells would, therefore, be greater than one. There was an additional requirement due to temperature cross (hot fluid-outlet temperature being lower than cold fluidoutlet temperature). When there is a temperature cross, two or more shells in series are adopted. The greater the cross, the greater the number of shells in series. To improve the shell-side heat-transfer coefficient through utilizing allowable shellside pressure drop and velocity, multiple shells in series are adopted. But if the pressure drop exceeds more than is allowable, then the shell-side fluid flow is divided by considering shells in parallel. This is very crucial to selecting the optimum number of shells in parallel and series-type arrangements such that pressure drop is maximized within the allowable limit. The shell- and tube-side velocities are in an acceptable range, and the U value is maximum leading to the least surface area. In this presented case study, two shells in series were required theoretically. However, those two shells were very large in ID, and the surface area and bundle weight Hot were many times over the bundle HVGO weight limit set by the client in the project specification. Therefore, smaller shells were tried in series and parallel arrangements to arrive at the optimized surface area, the highest velocities and the optimized pressure drop.

there are other parameters that can influence section of fluid sides as they cause the tube-surface area to increase or decrease. The thermal engineer aims to produce a thermal design, which is lesser in cost, but high in ease of operation and maintenance. Sometimes, there will be confusion in selecting a particular side and the design parameters for yield a decision contradiction. The safe bet will be making two designs having opposite fluid sides and then deciding which is best. The key factors that a thermal engineer must analyze when allocating fluid sides are: • Fluid viscosity, fouling and cleaning requirements • Fluid pressure and temperature • Flowrate.

transfer coefficients. In 80% of the cases, the higher viscosity fluid will have a higher fouling resistance; keeping it on the shell side will mean difficulty in maintenance as the tube bundle may have to be pulled out more frequently for cleaning. Cleaning the inside of tubes is preferable; as in this case, the bundle does not need to be pulled out. So, there is a dilemma over what will be the more preferable option. In such situations, it is prudent to take clients in confidence and seek their views. Fouling resistance of AR was marginally higher than HVGO, but the viscosity of AR was much higher than HVGO. Therefore, it was prudent to take advantage of shell-side turbulence on the higher viscous atmospheric crude and get higher heat-transfer coefficients.

Fluid viscosity, fouling and cleaning requirement. Higher viscosity

Fluid pressure and temperature.

fluid between hot and cold fluids must be placed on the shell side to maintain more turbulence and corresponding higher heat HDT ATM residue

Choose correct fluid sides.

Cold ATM residue

Correct allocation of fluid sides, i.e., which fluid needs to be placed on the shell side or on the tube side, is very crucial for maintenance and operability of the exchanger. Along with this,

Cold HVGO FIG. 1

Series and parallel arrangement of shells.

Tube OD

FIG. 2

Flow

90°

Pitch

Square pitch, 90° layout

Tube OD

tc h

Pitch

60°

h tc Pi

Flow

90°

When the operating pressure fluid is allocated on the tube side, only channel/channel cover, the floating head cover (if required) and tubes will be affected. However, if the same is allocated on the shell side, along with tubes and floating head cover, the shell will also be affected and require higher pressure applicability. The latHot ter being a higher cost option HVGO since shell length is more costly than channel length; shell length directly impacts the total cost for equipment. High-temperature fluids require higher alloy steel or SS. Therefore, it is prudent to keep high-temperature fluids on the tube side as only the channel/ channel cover, the floating head cover (if required) and tubes will be affected. However, if the same is allocated on the shell side, along with the tubes and floating head cover, the shell material will Tube OD

Tube OD

Flow

BONUSREPORT

60°

Inline triangular pitch, 60° layout

Diamond square pitch, 45° layout

Pitch Triangular pitch, 30° layout

Suitable tube layout arrangement for this case history.

HYDROCARBON PROCESSING DECEMBER 2010

I 55

BONUSREPORT

HEAT TRANSFER

also require higher cost materials of construction. Due to the higher temperature of the HVGO, all shells required SS 317L materials in the original design. However, with the revised design, due to the lower temperature of the AR, colder shells A/B and shell E/F were chosen to be of CS only. Due to the same reason (lower temperature of shell-side fluid), the tube sheet of the colder shells A/B and shell E/F were chosen to be of CS + SS 317L clad (tube side) only. This was quite a cost savings! Flowrate. Fluids with lower flowrates if routed through the shell side will result in higher shell-side coefficients, leading to higher overall coefficients. For same amount of fluid, the shell side provides more turbulence than the tube side due to the baffles that force fluid flow in diverted directions many times. In this case, the atmospheric crude flowrate was lower than the HVGO flowrate, and, therefore, the shell side was the better option. Maximize shell and tube side velocity. Velocity is an important cri-

terion for improving heat transfer coefficients in any heat-exchanger design. Espe-

cially in high viscosity service, maintaining a decent velocity on the shell side or tube side is very important to obtain a good heat transfer coefficient that is linked to an overall heat transfer coefficient and optimized surface area. Atmospheric velocity in the shell side and the HVGO velocity in the tube side were adequate considering fouling tendency for both fluids. Utilization of pressure drop. In any thermal design, one of the main aims of a thermal engineer is to use the pressure drop both on the shell and tube sides to the fullest within the allowable pressure drop to maximize the calculated heat transfer coefficients. As the pressure drop is proportional to the square of velocity, an increase in the pressure drop means higher velocity is also linked to the heat transfer coefficient. Tube layout arrangements. When handling dirty fluids on the shell side, especially in refineries, floating head exchangers are used extensively where tubes are laid out on a square (90°) or rotated square (45°) pitch. With the shell-side fluid being dirty, outside surfaces of the tubes require periodic cleaning and, therefore, in such

MICROTHERM SlimFlex

cases, TEMA specifies a minimum cleaning lane of 6 mm or ¼ in. The basic guideline for selecting either a 45° or 90° layout is shell side Reynolds number. Square pitch (90°) will be more suitable when the Reynolds number is greater than 7,000 as the shell-side heat transfer coefficients will be higher than rotated square (45°) layouts in such turbulent flow condition. However, when the Reynolds number for the shell-side fluid is less than 4,000, then flow conditions are laminar, and the shell-side heat transfer coefficients will be higher than the 90° layouts. When the Reynolds number is between 7,000 and 4,000, it is ideal to check both 90° and 45° before finalization of design. In this case, after reversing fluid sides (atmospheric crude in shell side), it was noticed that the NRe varied in the range of 1,700 to 5,700 with existing square-tube pitch configuration. Shellside heat transfer was 650 kcal/m2-hr-°C. Therefore, the rotated square configuration was also checked. With the rotated square layout, the NRe came out in the range of 1,400 to 4,800. The shell-side heat transfer was 940 kcal/m2-hr-°C. Overview. The overall saving (1/4th of the total equipment cost) in this case study shows that a proper and thoughtful design can save substantial costs. In a refinery, there is a substantial number of equipment items, and each item may have a story to tell. The role of a design reviewer on a project management consultancy team can be quite demanding as the consultant has to check the design given by the contractor/licensor is as per client’s specification and is optimum in cost and utility. When the client trusts the review team fully and gives proper support, the output from the team can result in huge cost savings as depicted here. HP

“Microtherm on a roll what could be simpler?” • 36” (914mm) wide rolls in .2” (5mm) and .4” (10mm) thicknesses • Multiple times more efficient than conventional insulations • Very low thermal conductivity over full temperature range • Capable of sustained exposure to 1832 °F (1000 °C) • Fully hydrophobic throughout the material to repel water • Fast and simple to cut and shape directly from the roll Microtherm - Truly the Best Performance at High Temperatures

MICROTHERM

LITERATURE CITED Heat Transfer Research Institute (HTRI) 2 Tubular Exchanger Manufacturers Association (TEMA)

Aerogel

Manas K. Mandal is a senior design engineer for

Calcium Silicate

Ceramic Fiber Mineral Wool

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

Thermal Conductivity (W/m-K) at 600 °C Mean

C1676 ASTM Standard for Microporous

www.microthermgroup.com Microtherm Inc. +1 865 681 0155 Microtherm NV +32 3 760 19 80 Nippon Microtherm +81 3 3377 2821

Select 163 at www.HydrocarbonProcessing.com/RS 56

0.160 Data Per ASTM Testing Standards

Fluor Daniel India Pvt. Ltd. He has more than 25 years of experience in the field of process heat transfer, cost optimization studies, process design & operations, process revamp, project control and energy management. Prior to Fluor, he has worked for Hindustan Petroleum’s Mumbai refinery and Engineers India Ltd. Delhi. Mr. Mandal has presented many papers in various seminars on heat transfer, energy management and process improvement. Mr. Mandal holds a BTech. degree in chemical engineering from the Indian Institute of Technology, Delhi and Masters degree in financial management from Jamnalal Bajaj, Mumbai.

HEAT TRANSFER

BONUSREPORT

Convert waste heat into eco-friendly energy New developments, such as the organic Rankine cycle, help operations go ‘green’ A. BOURJI, J. BARNHART, J. WINNINGHAM and A. WINSTEAD, WorleyParsons, Houston, Texas

ending new rules on air emissions will require hydrocarbon processing facilities and other manufacturers, to monitor and to reduce greenhouse gas (GHG) emissions from combustion sources. Waste streams, such as flue gas, contain GHGs as well as heat energy. Innovative thinking investigates using the organic Rankine cycle as one way to recover energy from flue-gas streams and yield “green” energy for the facility. Background. An intergovernmental panel on climate change

recently declared that “warming of the climate system is unequivocal, as is now evident from observations of increases in the global average air and ocean temperatures, widespread melting of snow and ice and rising global average sea level.”1 With the present Obama administration committed to emissions trading, an influential coalition of major US businesses and environmental groups, the US Climate Action Partnership (USCAP), unveiled its blueprint and policy recommendations early this year.2 The 26 corporate members of USCAP include Alcoa, British Petroleum, ConocoPhillips, Ford, Rio Tinto, Shell and Xerox. The recommendations are generally in line with scientific conclusions on what is necessary to stabilize global warming and they suggest these emission reduction targets: • 97%–102% of 2005 levels by 2012 • 80%–86% of 2005 levels by 2020 • 58% of 2005 levels by 2030 • 20% of 2005 levels by 2050. Fred Krupp, president of the Environmental Defence Fund (EDF), a USCAP member, described the plan as an “Obamaera” blueprint. USCAP proposes a “cap-and-trade” program covering as much of the economy’s GHG emissions as politically and administratively possible. This includes large stationary sources and hydrocarbon-based carbon dioxide (CO2) emitted by fuels used by remaining sources. The regulation point for large stationary sources should be the point of emission. The point for transportation fuels should be at the refinery gate or with importers.2 A tremendous burden in complying with new environmental regulations that are adopted will fall on the power industry. More important, the hydrocarbon processing industry (HPI) will also be required to make substantial changes and investments. The statement “putting transportation-fuel regulation at the refinery gate,” implies that the GHG impact from cars,

trucks and planes will have to be met by the fuel producer. Refiners will not only have to reduce their production emissions to the levels listed previously, but they will also have to absorb reductions inside the fence to, in effect, achieve the same reduction levels for their fuels. With these environmental issues, new challenges arise along with some opportunities to meet regulations and to be profitable. One way to indirectly mitigate emissions is to reduce any unnecessary energy losses. Many industrial processes discharge large quantities of energy as waste heat contained in flue gases from furnaces, heaters, kilns or boilers. Flue gas carries significant amounts of excess heat. A cost-effective method to recover this waste heat is an organic Rankine cycle (ORC) system. This system extracts energy from flue gas and converts it to electrical power. An ORC installed in a flue-gas stream can often generate enough electricity to offset the system’s operating costs and achieve a favorable return on capital investment. The produced surplus electricity from the site can either be used internally or sold to the power grid. Furthermore, most flue-gas treatment methods, from fluegas scrubbing to carbon sequestration, require that the flue gas is cooled prior to treatment. ORC technology can be of great benefit when combined with flue-gas treating technologies. The addition of an ORC system can improve the economics of fluegas treatment made necessary by environmental or permitting requirements by generating power with flue-gas heat that would otherwise be lost. An ORC resembles a typical Rankine cycle, but instead of circulating water as the working fluid, a refrigerant is used. An ORC also operates at lower temperatures than a steam system. Thus, an ORC is more efficient at recovering waste heat from low-temperature flue gas than a traditional Rankine cycle using water, thus reducing flue-gas outlet temperatures below what is possible in a steam system. Key evaluation parameters. When considering an ORC

application, the key parameters for feasibility and economics are: • Flue-gas flowrate • Flue-gas temperature • Flue-gas composition • Ambient air temperature (site climatological data) • Cooling water conditions (if the site has cooling water). HYDROCARBON PROCESSING DECEMBER 2010

I 57

BONUSREPORT

HEAT TRANSFER

The flue-gas flowrate and temperature determine the quantity of heat available for recovery. Also, the flowrate governs the size and operating temperatures and the pressures of the ORC equipment. For accurate process simulation and equipment sizing, the flue gas composition must also be known. The condenser, which is usually an air cooler, is the “heat sink” for the Rankine cycle. Its operating temperature is important to the system’s thermodynamics and is essential for sizing the condenser. If the site has a cooling water system, it may prove to be more economical to use cooling water for condensing purposes, instead of air cooling. Cooling water supply conditions are necessary to evaluate this alternative. Basics of an ORC for flue-gas heat recovery. The

basic ORC is illustrated in Fig. 1. Liquid refrigerant flows from the surge drum to the pump, where the fluid is pressurized and sent to the evaporator. The evaporator vaporizes the refrigerant by cross-exchanging it with the flue gas. If water condenses from the flue gas in the evaporator, it will be removed through a drain line. The vaporized refrigerant enters the turbo-expander/ generator where the fluid produces usable work in the form of electrical energy. After expansion, the refrigerant enters the condenser where it undergoes a phase change returning to the surge drum as a liquid. Understanding the temperature-entropy diagram.

The temperature-entropy diagram for a typical refrigerant is Condenser Surge drum M

Turbo-expander/ generator

Flue-gas inlet

Flue-gas outlet

Pump

Evaporator

Process flow diagram of an ORC.

Critical point Saturation line

Temperature

Power recovery efﬁciency, kW/MMscfd

FIG. 1

Isobars

Tf Tv Liquid

Vapor

Tc Ts 2-phase

shown in Fig. 2. The dashed line superimposed on the diagram illustrates the temperature, pressure and phase changes that the refrigerant undergoes as it circulates through a simple-cycle ORC system. The dome-shaped curve represents the saturation line. The area inside the curve is the two-phase region; single-phase liquid conditions exist to the left of the curve, with vapor conditions to the right. Tv is the vaporization temperature of the refrigerant, and Tc is the condensing temperature. The lower left-hand corner of the “box” formed by the dashed line represents the conditions, which exist at the surge drum and pump suction. From there, the fluid follows the saturation line up to the vaporization temperature, Tv . This part of the curve represents the action of the pump (adding pressure and moving the curve across the isobars), and the initial heat input of the evaporator (adding sensible heat to raise the temperature of the liquid to its boiling point). The horizontal line across the top of the cycle represents the vaporization of the refrigerant in the evaporator. Next, shown in the top right-hand corner, there’s a short jog up along the isobar, which represents superheating in the evaporator. The descending part of the cycle, along the right side of the “box” in the diagram, represents the refrigerant going through the expander. In an ideal, theoretical cycle, this line would be vertical; but in a real expander, inefficiency adds entropy, producing the rightward slant to the line. The last part of the cycle represents the action of the condenser, which follows the isobar down to the condensing temperature, Tc , then moves across the envelope, indicating a phase change and finally arriving back at the surge drum as a liquid. In an ideal condenser, the last line would be horizontal; the slight slant to the line represents a pressure drop in the condenser. The temperatures, Tf and Ts , in Fig. 2 are the flue-gas temperature and the heat-sink temperature, respectively. These values indicate the range of temperatures over which the refrigerant can operate. The differences between Tf and Tv , and between Ts and Tc , are determined by cost-effective exchanger design. An exchanger requires at least a few degrees differential temperature between hot and cold sides. As these differential temperatures decrease, the required heat-transfer area and, therefore, the capital cost of the exchanger increases. The condensing and vaporization temperatures each correspond to a refrigerant vapor pressure. The flue-gas and heat-sink temperatures determine the operating pressures of the ORC equipment.

10 9 8 7 6 5 4 3 2 1 0 200

Entropy FIG. 2

Temperature-entropy diagram for a typical refrigerant.

I DECEMBER 2010 HYDROCARBON PROCESSING

FIG. 3

250

300 350 400 Flue-gas temperature, ºF

450

500

Recovery efficiencies vs. flue-gas temperature (steel mill).

HEAT TRANSFER

10 9 8 7 6 5 4 3 2 1 0 200

FIG. 4

250

300 350 400 Flue-gas temperature, ºF

450

Recovery efficiencies vs. flue-gas temperature (cement plant).

10 9 8 7 6 5 4 3 2 1 0 200

FIG. 5

Power recovery efﬁciency, kW/MMscfd

Refrigerant selection. Selecting an optimal refrigerant is essential to the successful implementation of a flue-gas wasteheat recovery system. It is tempting to choose a refrigerant based solely on maximizing the power recovery efficiency, but there are other factors to consider as well. Depending on the specific application and location of the installation, it may be necessary to select a working fluid that is environmentally friendly, safe and economical (inexpensive and readily available). Refrigerant blends should also be considered. The choice of refrigerant or refrigerant blends must be evaluated on a case-bycase basis. Designers and marketers of ORC systems and their component equipment have process simulations and empirical data required to accurately model and to optimize ORC systems specific to any flue-gas stream. A refrigerant to be used in an ORC waste-heat recovery system should have a relatively high decomposition temperature, enabling it to tolerate the high temperatures of the exchanger

tubes contacting the flue gas in the evaporator. An ORC refrigerant should also have high critical and condensing temperatures, which are necessary to operate efficiently in the range between typical flue-gas temperatures and ambient temperature of the heat sink (see Fig. 2). The critical point, a physical property of any fluid, is located at the top of the saturation curve and indicates the temperature and pressure above which the fluid enters the supercritical region. In the supercritical region, the isobars are nearly vertical and a single-stage expander will not experience the full pressure drop required for efficient energy recovery without entering the twophase region. In general, the envelope within which an ORC system can operate expands with a higher refrigerant critical point, allowing for higher operating temperatures and pressures and greater heat-recovery efficiencies. Power recovery efﬁciency, kW/MMscfd

Air cooling vs. water cooling. The condenser in an ORC will usually be an air cooler, since most locations such as cement plants or steel mills, do not have a cooling water network. A typical refinery or petrochemical plant has a cooling water system, and installing a cooling water exchanger may be more economical. Also, a cooling water exchanger occupies less space than an air cooler, saving plot space, which may be important if the ORC is being added to an existing facility. Identifying the best method to cool and condense the refrigerant requires that consideration be given to factors including onsite availability of cooling water, utility costs of using cooling water vs. the electricity consumed by the air-cooler fans, plot space availability in the proposed ORC location, and capital costs of each exchanger design for the given duty requirement.

BONUSREPORT

500

250

300 350 400 Flue-gas temperature, ºF

450

500

Recovery efficiencies vs. flue-gas temperature (refinery).

10 9 8 7 6 5 4 3 2 1 0 200

FIG. 6

250

300 350 400 Flue-gas temperature, ºF

450

500

Recovery efficiencies vs. flue-gas temperature (petrochemical plant).

TABLE 1. Flue gas properties of various sources Composition, mol% Source

Flowrate, MMscfd

Temperature, °F

H 2O

CO2

SO2

66 to 1,728

225 to 1,000

3.00

5.00

17.00

75.00

0.00

Cement plant

15 to 544

300 to 700

5.18

33.55

1.33

59.74

0.20

Refinery

4 to 1,339

350 to 500

13.86

6.93

2.98

76.23

0.00

671+

260 to 1,500

26.33

3.82

1.63

68.22

0.00

Steel mill

Petrochemical plant

26 to

HYDROCARBON PROCESSING DECEMBER 2010

I 59

BONUSREPORT

HEAT TRANSFER

Some refrigerants have favorable environmental traits such as low ozone depletion potential (ODP) or low global warming potential (GWP). Further, many are not considered a volatile organic compound (VOC) in the US. Using an environmentally friendly refrigerant adds to the overall environmental benefit and “green” potential of an ORC system installed for waste-heat recovery. Refrigerant R-245fa (1,1,1,3,3—Pentafluoropropane) is an example of a refrigerant with an overall favorable combination of thermodynamic and environmental qualities. The characteristics of R-245fa make it particularly suitable for flue-gas heat recovery applications using ORC technology. It has a relatively high critical point, a zero ODP, and a low GWP and it is not a listed VOC.3 Determining potential net power recovery. The net

power generation of an ORC system equals the power output of the expander/generator, less the power consumed by other equipment such as the air-cooled condenser and liquid refrigerant pump. In many potential applications, an ORC system can recover enough waste-heat energy to not only provide enough electrical power to offset the entire operating cost of the system, but also to produce a net surplus that can supplement the site’s electrical system or be sold to the power grid. Examples to estimate the net electrical power that an ORC can generate at any given flue gas temperature are shown in Figs. 3–6. These figures are developed using typical flue-gas compositions from various sources including steel mills, cement plants, refineries and petrochemical plants. Table 1 summarizes typical flue-gas compositions, temperatures and flowrates of various sources. Case Study: ORC economic evaluation. A case study was developed to examine the estimated costs, power recovery and

Total installed cost $0.06 per kWh $0.08 per kWh $0.10 per kWh

25 ROI, %

50 40

0 Base 1A A

FIG. 7

3A Base 1B 2B 3B Base 1C B C

Initial cost, $ millions

return on investment (ROI) for an ORC system to recover wasteheat energy from the flue gas from a steel mill. Three scenarios (A, B and C) are identified, each with different flue-gas flowrates and inlet temperatures, as shown in Table 2. The flue-gas composition is consistent for each of the three scenarios and corresponds to that of a steel mill, presented in Table 1. Each of the three scenarios is subdivided into four cases: Base, 1, 2 and 3, and only variables within the ORC system itself are varied. • Base Case has the pump discharge pressure set so that the refrigerant leaves the evaporator as a superheated vapor stream. • Case 1 differs from the Base Case, as the refrigerant has a higher pump discharge pressure. The pressure has been increased to the extent that the refrigerant leaves the evaporator as a saturated vapor rather than as a superheated vapor. • Case 2 is analogous to Case 1, but the ambient temperature has been lowered from 90°F to 62°F, the actual average ambient temperature for the location of the installation. This reduces the load on the condenser and increases the allowable pressure drop across the turbine, thereby increasing total power recovery. The higher temperature is a more conservative approach and is necessary for accurate equipment design. However, the lower temperature provides a more realistic prediction of the power recovery. • Case 3 is identical to Case 2, except that the flue-gas outlet temperature is lowered from 185°F to 100°F, allowing increased heat transfer in the evaporator. Lowering the outlet temperature to this degree forces some water vapor in the flue-gas stream to condense, providing additional heat transfer due to the phase change. However, since acid-gas condensation is occurring, the material specification of the evaporator must be altered to accommodate the corrosive service, adding to the capital cost. Table 3 lists the results of the case study. The total equipment cost represents the estimated purchase cost of the major equipment items. Cost data are provided by equipment suppliers and in-house historical estimating databases. The total installed cost (TIC) is approximated using a factor of 2 (the TIC equals the purchase cost of major equipment multiplied by 2). This repreTABLE 2. Flue-gas flowrates and temperatures Scenario

Flowrate, MMscfd

Temperature, °F

66.2

1,000

1,135

350

1,728

225

Case study comparison.

TABLE 3. Case study economics Scenario A Base

Scenario B

Base

Scenario C 3

Base

Total equipment cost, $ million $3

$3.3

$3.7

$6.7

$7.1

$11.4

$6.5

$6.3

$29.3

Total installed cost, $ million

$6.6

$7.4

$13.5

$14.2

$22.7

$13.1

$13.9

$12.6

$58.6

707

899

1,364

1,549

2,484

3,102

4,455

7,308

916

892

1,364

5,367

ROI @ $0.06 per kWh

5.6%

7.2%

10.0%

10.1%

8.8%

11.1%

15.1%

15.5%

3.4%

3.1%

5.2%

4.4%

ROI @ $0.08 per kWh

7.5%

9.6%

13.3%

13.5%

11.8%

14.8%

20.1%

20.6%

4.5%

4.1%

6.9%

5.9%

ROI @ $0.10 per kWh

9.3%

12.1%

16.6%

16.9%

14.7%

18.4%

25.2%

25.8%

5.6%

5.1%

8.6%

7.3%

Total Power Recovery, kW

I DECEMBER 2010 HYDROCARBON PROCESSING

HEAT TRANSFER sents the initial capital investment required for the ORC system. The total power recovery indicates the quantity of net power produced by the system that can be utilized by the facility or sold to the grid. The power requirements of the individual equipment items contained within the ORC system have been accounted for in this value. The ROI values included are calculated on an earnings before interest and taxes (EBIT) basis and are simply the percentage of initial capital investment displaced by annual net income. Fig. 7 illustrates the effects that the various cases and the variation in energy prices have on the ROI. Options. In the near term, the markets will witness increased investment and investigation by many industries, including the HPI, to curb emissions and improve energy efficiency. These efforts arise in an attempt to meet looming regulations. A new industry has evolved out of compliance with ever-increasing regulations. This is an industry tasked with developing processes and facilities to capture emissions, to improve overall efficiencies and to reduce carbon footprint. A waste-heat-recovery system, to some effect, relates to each of these tasks. Any source of flue gas that is of significant quantity and reasonable temperature is a potential candidate for waste-heat recovery using an ORC system. These systems, which can be retro-fitted into existing plants, have an appealing ROI, providing additional income without increasing emissions. The installation of an ORC system will improve the overall efficiency of the facility and offer easier implementation of other eco-friendly technologies such as carbon capture-sequestration in the future. HP

1 2 3

BONUSREPORT

LITERATURE CITED Intergovernmental Panel on Climate Change, â&#x20AC;&#x153;Climate Change 2007: Synthesis Report.â&#x20AC;? USCAP Blueprint for Legislation Action, Jan. 15, 2009. www.us-cap.org Genetron 245fa Applications Development Guide, Honeywell, 2000. (Genetron is a registered trademark of Honeywell.) Ali Bourji is a senior technical director at WorleyParsons in Houston. Dr. Bourji received his BS and MS degrees in chemical engineering from the University of Houston and a PhD from Lamar University. He is a professional engineer and a member of AIChE and NPRA.

John Barnhart is a WorleyParsons vice president. A graduate of the University of Houston and a professional engineer, his background includes plant operations as well as project and company management. Mr. Barnhart is a member of AIChE.

Jimmy Winningham is a process engineer at WorleyParsons in Houston. He received a BS degree in chemical engineering from Texas A&M University.

Alan Winstead is a supervising process engineer at WorleyParsons in Houston. He received a BS degree in chemical engineering from Rice University.

Select 164 at www.HydrocarbonProcessing.com/RS HYDROCARBON PROCESSING DECEMBER 2010

I 61

Select 59 at www.HydrocarbonProcessing.com/RS

PROCESS CONTROL AND INFORMATION SYSTEMS

Implementing advanced process control on ammonia plants Benefits exceeding the guarantee were realized through APC by simultaneously stabilizing and optimizing the plant in the presence of significant feed density variations that impact the entire plant operation P. BANERJEE and K. V. SIVA RAMA BRAHMAM, ATME Group, Kuwait; and S. AL-AZMI, L. NAYFEH and K. AL-AZMI, Petrochemical Industries Company, Kuwait

dvanced process controllers (APCs) were implemented on the ammonia and steam-generation facilities comprising two ammonia production lines of different capacities interacting with a common downstream purge-gas recovery unit (PGRU) and ammonia absorption units. The steam facility has several boilers and super heaters contributing to the steam grid. Plant operation was significantly affected by large variations in the natural gas (NG) density that sometimes led to trips if prompt actions were not taken. Density variation affects the plant operation since NG serves as the feed and also fuel for the reformer and boilers. The operational stability problem was compounded due to manual compressor operation in one of the ammonia production lines and recycle fuel gas from the PGRU. After implementing the APCs, the plant stabilized and maximized the benefits from the plant assets. Key strategies included in the controller that lead to the operational benefits are discussed in this article.

Introduction. The ammonia manufac-

turing facility at Petrochemical Industries Company (PIC), Kuwait, has two different capacity ammonia production lines referred to as the “low” and “high-capacity” ammonia lines. The manufacturing processes of both lines are similar, but the operational practices are slightly different due to manual compressor operation for the “low-capacity” ammonia line and other equipment/valve constraints. Downstream of both the ammonia lines is a common ammonia recovery unit and a PGRU recycling fuel gas and H2 back to the ammonia lines. Fig. 1 outlines an overview of the ammonia facility. NG is first washed in the MEA section to remove the bulk of the sulfur and CO2. The washed NG is compressed and desulfurized to remove the residual organic sulfur to protect the catalysts in the reformer and low-temperature shift converter from

poisoning. Desulfurized NG is mixed with steam in a fixed steamto-carbon ratio, and steam is reformed in the primary reformer to obtain H2, CO and CO2. Further reforming takes place in the secondary reformer where partial hydrocarbon combustion takes place in the presence of process air. Also, air added to the secondary reformer maintains the H2-to-N2 ratio in the right proportion required for ammonia synthesis. CO is then converted to CO2 in high- and low-temperature shift converters. The resultant CO2 is then removed in a CO2 absorption column using a Benfield HPC solution and is sent to the urea plant. CO2 and CO in the reformed gas from the secondary reformer are poisons for the ammonia synthesis catalyst. Hence, the remaining amount of CO and CO2 in the reformed gas is removed in the methanation section by converting them to CH4, which is an inert for the ammonia synthesis. The reformed BFW Boilers/super-heater system

FG Steam export

Steam APC

Steam import from WHB

Air

Steam header system CO2 to urea plant

Higher-capacity ammonia line Recycle FG Off gas

Ammonia abs.

NG fuel gas (FG)

NG feed

Ammonia products

Recycle FG Recycle LP/HP H2

Purge gas

PGRU

Air Lower-capacity ammonia line Ammonia APC

FIG. 1

Overview of ammonia facility and APC boundaries.

HYDROCARBON PROCESSING DECEMBER 2010

I 63

PROCESS CONTROL AND INFORMATION SYSTEMS Process air

NG feed

MEA absorption

Feed comp.

Air comp.

Process steam

Desulfurization Comb. air

Steam

Primary and secondary reformer

BFW Recycle FG from PGRU and ammonia abs.

resulting snowball effect from the recycle fuel gas recovered from the PGRU. Having appropriate model gain ratios and fast control action, and ensuring complete combustion in the primary reformer, were key in addressing the issue.

SpeciďŹ c gravity

APC project methodology. PIC decided to implement an APC in 2006 on Steam a guaranteed benefit basis on the ammonia CO2 to and steam-generation facility immediately urea plant CO2 HP/LP shift Syn. gas Methanation after commissioning the distributed control comp. converters removal Recycle LP/HP system (DCS). A benefit assessment study H2 from PGRU Steam BFW was conducted in 2004 prior to the DCS commissioning and PIC took up the APC implementation project after the DCS was Offgas to ammonia abs. commissioned. Ammonia Purge gas to PGRU synthesis loop Standard APC project implementation Ammonia product and refrigeration methodology, along with rapid deployment strategies,1 were employed to enhance the Amm. project implementation schedule and obtain comp. superior controller models. The APC project is typically implemented FIG. 2 Ammonia process of a manufacturing line. in a sequence of stepsâ&#x20AC;&#x201D;namely the pretest, step test, model development and commissioning, followed by the post audit. During 7/13/2006 7/19/2006 7/25/2006 7/31/2006 the pretest, regulatory and field instrumentation performance was 0.9991 031B 102 separator OH reviewed and tuned. During the step test, the automated step0.9560 testing application was used to automatically perturb the plant 0.9130 to reduce the implementation time and to enhance the model 0.8699 quality.2 Separate controllers were initially implemented for the high- and low-capacity ammonia lines. Later, controllers for the 0.8268 ammonia lines, the PGRU and ammonia absorption sections were 0.7838 merged into a single controller to account for the interactions. 0.7407 Inferential models were deployed to back up the online analyz0.6977 ers that are considered as controlled variables in the controller. The 0.6546 inferential property prediction application consistently maintains 00:00:00 00:00:00 00:00:00 00:00:00 the analyzer predictions whenever the analyzers are not available or go bad. The inferential models are bias corrected with both FIG. 3 Typical variation in NG specific gravity. analyzer values and lab results. A separate controller was deployed for the steam system. The steam controller distributes load between the boilers based on gas from the methanation section is called the ammonia synthesis their efficiencies. The total load to the boilers is supplied by a gas. The amount of process air to the secondary reformer is conmaster steam header pressure controller configured in the DCS. trolled to maintain the H2-to-N2 ratio in the synthesis gas. Each ammonia line is managed by a different set of operatSynthesis gas is converted to ammonia in the converter. The ing personnel, and the responsibility for operating the PGRU, partially converted synthesis gas is chilled to separate out the liquid ammonia absorption and each of the steam boilers is also divided ammonia product. The unconverted reactants and uncondensed between the two ammonia lines. The APC subcontroller modammonia are recycled back to the converter after mixing with the ules were accordingly distributed between the DCS operating makeup synthesis gas. Part of the ammonia is used for chilling the panels relevant to their sphere of operation. The post-benefit converted synthesis gas using the ammonia refrigeration cycle. study revealed that the benefits accrued due to APC exceeded the The steam-generation facility comprises four boilers and three guaranteed benefits. super heaters contributing to a common steam header. The major operating challenge was to handle the significant variations in the NG density that upset the unit operation leading APC structure. Two APCs were implementedâ&#x20AC;&#x201D;one for the to plant trips if prompt actions were not taken. The operational ammonia lines, ammonia absorption and PGRU, and the second challenges were compounded due to the interactions between controller for the steam system. The controllers are configured to the ammonia lines, the PGRU and ammonia absorption units, run at a 30-sec. frequency to handle the density variations. analyzer malfunctions and manual operation of some of the comThe ammonia controller includes the MEA absorption, pressors that were effectively addressed by the APC. reforming, shift conversion, CO2 removal, methanation, syntheThe key issue for the APCs was to overcome the impact of sis and refrigeration sections of both ammonia lines along with NG density variation, since NG is both feed and fuel, and the the PGRU and ammonia absorption units. The APC is divided 64

I DECEMBER 2010 HYDROCARBON PROCESSING

PROCESS CONTROL AND INFORMATION SYSTEMS

Controller objectives. The APCs were successfully imple-

mented to meet these objectives: • Effectively controlling of the plant in the presence of disturbances in feed NG density • Maximize plant capacity by minimizing the variations arising due to various disturbances • Minimize steam-to-carbon ratio at the primary reformer • Minimize excess O2 in primary reformer flue gas • Minimize CH4 slip at the secondary reformer exit • Minimize synthesis gas venting from the ammonia trains • Maintain H/N ratio in the synthesis loop • Minimize the purge gas from the synthesis loop by optimizing inerts in the synthesis loop • Effectively controlling the off-gas disturbances to the primary reformer fuel coming from the PGRU • Control CO2 absorber’ CO2 slip outlet within limits • Control primary and secondary reformers outlet temperatures and other constraints within limits. Effective control against natural gas feed density variations. NG density variations are very high at the plant site.

that, as the feed NG specific gravity increases, APC decreases the NG feed while maintaining the % H2 in the synthesis loop and CH4 slip at the secondary reformer outlet. The NG feed specific gravity analyzer is a key feedforward variable for adjusting the NG feed, primary reformer fuel, process steam, etc., to achieve operational stability in the presence of density variation. The strategies that enable the APC to minimize the effect of NG density variation are: • Resolving near colinearity and the condition number of the model steady-state gain matrix resulted in a robust controller design that enabled the controller to take the “just right amount of move” to the disturbances arising out of the density variation. Unresolved colinearity otherwise leads to excessive controller action, leading to more process disturbance. • The controller was dynamically tuned to take fast action, and the control move plans of some of the MVs were skewed to take even faster action. TABLE 1. Ammonia and steam controller size S no

Controller variables

Ammonia controller Steam controller

# Manipulated variables (MV)

# Feed-forward variables (FF)

# Controlled variables (CV)

128

# Sub controllers

Requirement of feed NG for 100% throughput vs. feed NG speciﬁc gravity Feed NG ﬂow

into subcontrollers according to the sections in the plant for operational ease. The steam controller includes all the boilers and super heaters at the ammonia facility contributing to the steam grid. Each boiler and super heater is a subcontroller. The APC benefit was realized due to increased throughput, reduction in steam-to-carbon ratio, minimizing the methane slip, maintaining the H/N ratio in the synthesis loop to the target value and by reducing the purge gas flow.

29,000 27,000 26,000 25,000 23,000 21,000 17,000 15,000

Effectively controlling the plant against these NG density variations and optimizing the plant performance at the same time is a difficult task. This was achieved through proper design, accurate model building, gain-matrix analysis and robust MPC tuning. Carbon quantity in the feed and the heating value changes 0.65 0.75 0.55 0.85 0.95 as the NG density changes, which affects operation of the entire Feed NG speciﬁc gravity plant. Density variation affects all the key operating paramFIG. 4 Feed NG specific gravity vs. feed NG flow. eters, i.e., H2 production, reformer firing, NG feed compressor operation, etc. The following actions are required at a rapid pace to handle the feed density variation: • NG feed to the primary reformer needs to be adjusted to control the synthesis loop H/N ratio. Fig. 4 shows the operator guideline for adjusting the NG feed with specific gravity changes at design capacity. • Reformer fuel needs to be adjusted to address the duty requirements, reformer outlet temperature, CH4 slip exiting the secondary reformer, and excess O2 in the primary reformer radiant zone. • NG compressor suction pressure needs to be adjusted to keep the compressor speed within the operating region. • Actions are required in the reformer downstream units to minimize propagation of the disturbances. APC was configured to take the required FIG. 5 Control against the NG density fluctuations. actions on time to keep the plant steady against the density variations. Fig. 5 shows

1.00

HYDROCARBON PROCESSING DECEMBER 2010

I 65

PROCESS CONTROL AND INFORMATION SYSTEMS Purge gas Fp l

Converter

Syn. loop Fs l Separator

Boundary for inerts balance

Liq. NH3 product

Fr l Recycle syn. gas

FIG. 6

Stabilizing primary reformer outlet temperature.

Makeup syn. gas F il i

FIG. 9

Inerts balance for estimating H/N in the synthesis loop.

By effectively controlling the reformer section, the variations in the primary reformer outlet temperature have been significantly reduced in the presence of density variations. Fig. 6 shows that the primary reformer outlet temperature is controlled within +/- 2 degrees representing about a 70% reduction in the variability. Minimizing CH4 slip exiting secondary reformer.

FIG. 7

Minimizing % CH4 at the secondary reformer outlet against NG specific gravity variations.

CH4 slip at the secondary reformer outlet is another important performance parameter that has a direct impact on the efficiency of the operation. Slippage of CH4 from the reformer contributes to inerts in the synthesis loop, requiring more synthesis gas purging to sustain the throughput and NH3 conversion per pass. Fig. 6 shows that the APC has reduced the CH4 slippage variation by about 50% even in the presence of feed NG density variations. The controller moves feed gas and fuel gas to the primary reformer quickly and in right proportions against the NG specific gravity variations. Minimizing excess O2 in primary reformer flue gas.

FIG. 8

Stabilizing % excess O2 in the primary reformer radiant zone.

Stabilizing primary reformer outlet temperature.

Primary reformer outlet temperature control is a multi-variable problem since it is affected by several variables, i.e., primary reformer fuel pressure, NG density, feed NG flow, process steam flow, offgas recycle to reformer fuel, etc. The primary reformer furnace side is constrained by parameters such as the suction damper opening of the induced-draft fan, combustion air blower speed, etc. This has an impact on throughput maximization and in the H/N ratio control of the synthesis loop. 66

I DECEMBER 2010 HYDROCARBON PROCESSING

The percentage of excess O2 in the primary reformer radiant zone is mainly controlled by adjusting combustion air and fuel gas to the primary reformer. As NG density varies, so varies the recycle offgas from the PGRU and ammonia absorption, making the percent of excess O2 control difficult. To maintain the percent of excess O2 above a lower limit, the fuel gas is decreased and that often limits the plant throughput. A calculation is configured in the APC to adjust the minimum lower operating the percent of excess O2 limit to ensure complete combustion in case of a changing proportion of offgas from the PGRU to the primary reformer fuel gas. The amount of offgas from the PGRU changes with NG density variations. Fig. 8 shows that the APC reduced the percent of excess O2 variability by about 75%, thereby succeeding in maintaining an optimal value to sustain maximum throughput. H/N ratio calculation and control. Maintaining the H/N ratio at the desired stoichiometric value of ~ 2.99 in the synthesis loop is another important objective to maximize ammonia conversion. As the NG feed density varies, so varies the percent of H2 in the synthesis loop, thereby varying the H/N ratio. H/N ratio

PROCESS CONTROL AND INFORMATION SYSTEMS TABLE 2. Summary of benefits realized through APC

5.0

Low-capacity line High-capacity line

Increase in throughput, %

0.9

2.5

Reduction of S/C ratio

0.2

0.12

Reduction of CH4 slip exit secondary reformer

0.1

0.07

control is not straightforward since its dynamics do not bear a linear relationship with the independents primarily because of inerts accumulation in the synthesis loop. Moreover, no direct H/N ratio measurement was available; hence, an inferential calculation was used for estimation. In the makeup synthesis gas the analyzers are available for measuring the percent of CH4 and the percent of H2. The remaining components, the percent of Ar and the percent of N2, are estimated. The proportion of Ar in the makeup synthesis gas is estimated from the process air supplied to the secondary reformer and N2 is calculated by balance. The analyzers are available for measuring the percent of CH4, H2 and NH3 in the synthesis gas loop at the converter inlet that comprises the makeup and recycle synthesis gas. To calculate the synthesis gas H/N ratio at the converter inlet, the composition of the remaining components, the percent of Ar and the percent of N2, are estimated from the inerts balance shown in Fig. 9 and as described below. CH4 and Ar are inerts in the synthesis loop that will keep on accumulating unless a portion of the synthesis gas is purged. In Fig. 9 the total inerts (%CH4 + %Ar) are represented by Ii in the makeup gas and by I in the synthesis loop.

PV UL LL SS ET OL CL

4.5 Ratio

Benefit

4.0 3.5 3.0 2.5 06:00

FIG. 10

09:00

12:00

External target control of ammonia H/N ratio.

Since the components Ar and CH4 are inerts in the synthesis loop, the proportions of each component in total inerts remain approximately constant at steadystate in the synthesis loop. As shown below, let Xi be the percentage of CH4 in total inerts (Ii) in the make-up synthesis gas and X at the converter inlet: Xi =

%CH 4 %CH 4 + % Ar

%CH 4 X = %CH 4 + % Ar

Makeup

SynLoop

Then at steady state, Xi ≈ X. The dynamic model that represents the change in these proportions of inerts from make-up synthesis gas to converter inlet

ARE YOU TAKING FULL ADVANTAGE OF HYDROCARBON PROCESSING?

Discover all the beneﬁts of being a premium subscriber and gain full access to HydrocarbonProcessing.com

Subscriber Only Benefits 12 monthly issues in print or digital format, and premium access to HydrocarbonProcessing.com, including: • All the latest issues and Process Handbooks • HP’s extensive archive containing eight years of back issues • A subject/author index of print articles with links to articles currently available online. • Monthly e-newsletters providing an early preview of upcoming special editorial features and exclusive content.

SUBSCRIBE TODAY! Log on to www.HydrocarbonProcessing.com or call +1 (713) 520-4440 As a Hydrocarbon Processing premium subscriber, you will receive full access to WorldOil.com as well as World Oil magazine in your choice of print or digital format. Start your subscription today. Subscribe online at www.HydrocarbonProcessing.com or call +1 (713) 520-4440

Select 165 at www.HydrocarbonProcessing.com/RS HYDROCARBON PROCESSING DECEMBER 2010

I 67

PROCESS CONTROL AND INFORMATION SYSTEMS (synthesis loop) is approximated by a unity-gain first-order transfer function with a dead-time. The time constant and the dead-time are estimated using APC modeling software that uses the relationship between the conditions of make-up synthesis gas and converter inlet in the synthesis loop. After calculating X, the percent of Ar at the converter inlet is estimated using: (100 X ) %CH 4 SynLoop % Ar SynLoop = X The remaining percent of N2 can then be derived by balance to calculate the H/N ratio of the synthesis loop. 16%

76%

34%

26% FIG. 11

20%

High-capacity ammonia plant beneﬁts Throughput maximization S/C minimization CH4 slip minimization

Low-capacity ammonia plant beneﬁts NG compressor antisurge bypass minimization S/C minimization Throughput maximization CH4 slip minimization

20%

Since it is desired to operate the H/N ratio in a very narrow range, an external target is configured to maintain this ratio while allowing it to deviate softly in case of disturbances. Fig. 10 shows that the APC closely maintains H/N ratio near the user set external target. Summary of the benefits realized. The return on investment for this project is less than six months. The contributors to the benefits are shown in Table 2. In addition, the APC scheme effectively controls the plant against the feed NG density variations, maintains H/N ratio in the synthesis loop, controls CO2 slip from the HPC absorber, maintains various reactor temperatures, minimizes excess O2 in the primary reformer, handles disturbances in the reformer fuel gas coming from recycle purge gas/offgases, etc. The contributors to the overall economic benefit are summarized in Fig. 11. HP ACKNOWLEDGEMENT The authors thank their respective Managements for their support. The ATME authors thank Mr. Anand Shah, Advisor to ATME for building the APC– DCS interface.

LITERATURE CITED Kalafatis, A., K. Patel, M. Harmse, Q Zheng and M. Craik, “Multivariable step testing for MPC projects reduces crude unit testing time,” Hydrocarbon Processing, February 2006. McIntosh, A., J. Cooke and M. Harmse, “Techniques for Implementing Large-Scale DMCplus Controllers,” 7th annual IEEE Advanced Process Control Applications for Industry Workshop, Vancouver, Canada, April 28–30, 2003.

Summary of ammonia line economic benefits. Pranob Banerjee is a services manager with ATME Group, Kuwait and heads the APC group. He is a chemical engineer with 25 years of industrial experience and holds a PhD degree in APC from the University of Alberta, Canada. Dr. Banerjee has APC implementation experience in refinery, LNG/NGL, fertilizer, ethylene and other petrochemical processes. Previously he worked with Engineers India Ltd. and Reliance Industries Ltd. in India.

K. V. Siva Rama Brahmam is a senior staff engineer with ATME Group, Kuwait. He is a chemical engineering graduate from the Andhra University College of Engineering, Visakhapatnam, India. Mr. Brahmam has 11 years of industrial experience in production, technical services and APC implementation with technologies from AspenTech (DMCplus) and Shell/Yokogawa (SMOC). Previously he worked at Indian Oil Corporation and Nagarjuna Fertilizers & Chemicals in India.

Salem Al-Azmi is an operations manager at PIC, Kuwait. He is a chemical engineering graduate from the Kuwait University. Mr. Al-Azmi has 20 years of industrial experience in operations, technical services and process control.

Lutfy Nayfeh is a senior specialist in process engineering at PIC, Kuwait. He is a chemical engineer with 30 years of experience in the operation of ammonia and urea plants and he holds a masters degree from the Kuwait University.

Khalid Al-Azmi is a process control engineer at PIC, Kuwait. He graduated in chemical engineering (BS) from Toledo University, Ohio, USA. Mr. Al-Azmi has 12 years of experience in projects and process control. He was involved in a urea revamp and new granulation projects at PIC. He has expertise in FOXBORO DCS, Aspentech Infoplus 21, DMCplus, Aspen IQ and Aspen Watch software. Select 166 at www.HydrocarbonProcessing.com/RS 68

PROCESS CONTROL AND INFORMATION SYSTEMS

Critical concepts in fieldbus system design differ from a conventional DCS Here are some key topics calling for a conceptual change from the traditional DCS D. MAJUMDER, Invensys India Pvt., Ltd., Chennai, India

ith the advent of Foundation Fieldbus (FF) and its wide deployment in the industry, there is a paradigm shift in the concept of the commonly used terms/specifications referred to in a conventional DCS and which needs to be understood in context of the FF. Numerous papers, documents and publications pertaining to FF are available, specifically of the concept, the design aspects, etc., and the Fieldbus Foundation website: http:// www.Fieldbus.org. This article will cover, in short, some key topics that call for a conceptual change from the traditional DCS and the FF-based DCS. Key areas include: • Special power supply requirement and speed/bandwidth limitation for FFbased systems • The power trunk concept and intrinsically safe power supply to be used in hazardous areas. • Control on the wire and control on the DCS, scan time, end-to-end loop response time and macro cycles. • The pros and cons of engineering control on the wire vs. control in the DCSbased FF system. • Brief description of the commonly used terms such as function blocks, VCR, etc. • Post-commissioned plant add-on and loop modification changes, specifically from an Indian context/scenarios. FF is a system and not a bus, and the system is an integrated control system involving a DCS, FF control module, power supply, cables, terminators and FF field devices with each manufactures’ standard. (Inline with Fieldbus Foundation’s guidelines and host interoperability testing and certifications.)

Special power supply required.

For FF-base systems, special power supply requirments are necessary: • The FF system communicates at 31.25 Kbps and at the same time powers the devices. (DC voltage, like any conventional DCS-based field instruments, ranges from 9–32 VDC.) So the power supply for FF devices does the dual-purpose supply of regulated DC power to the device and also carries the FF communication at a high frequency. • Based on the philosophy followed, the power supply can be for intrinsic safety or general purpose with its own guidelines. • Any DC power supply has its own low-impedance regulator circuit and also carries the high-frequency digital signal (like AC signals representing in “0” and “1”). The two are carried by a single cable pair. There should be proper isolation and protection to use both in the FF power supply unit. Example. For digital signal communication, we need the loop to be closed and, if this is done, the power supply DC regulator is shorted and there will not be any power communication. Similarly, if there is no loop closure, there won’t be any digital communication. Hence, there is need for a proper special power supply unit with proper filters for the FF system with required design end terminators for each H1 segment. Note: • FF power supplies are designed considering the Fieldbus Foundation’s guidelines like a balanced power supply unit, etc. Segment terminators are a combination of a resistor and a capacitor, with the resis-

tor closing the loop for the digital signal and the capacitor, preventing power supply short protection. • The signal measurement at 31.25 KHz used in FF measurement is done by measurement using the Manchester encoding technique, where the current from the source is modulated with time and the voltage measurement at the receiving-side resistor due to the variation of each device using filters to eliminate noise and signal distortion. • The conventional DCS working on 4-20 ma DC current works at around 10 Hz (through DC) and FF at 31.25 KHz, and due to the communication high frequency and integrity there is a distance limitation depending on the cable specifications and specifically the AWG. This is also a main reason for the communication speed/bandwidth being limited to 31.25 KHz and not higher since it would call for drastic and stringent specifications for the communication cables and other practical aspects. However, this speed is sufficient for process control. Power trunk and intrinsically safe power supply in hazardous areas.

This is applicable only in cases of hazardous-area application, where power to the field needs to be limited to avoid explosions in the presence of hazardous gas in the field. This is achieved usually by using the intrinsic safety philosophy. In a conventional DCS, for each loop there is an IS barrier in the control room that limits the power to the field and that is coupled with intrinsically certified instruments in the field. However, in FF, since the power supply for each segment is common, no single HYDROCARBON PROCESSING DECEMBER 2010

I 69

PROCESS CONTROL AND INFORMATION SYSTEMS TABLE 1. Foundation Fieldbus considerations Factor

Description

Remarks

Control execution philosophy on the wire or DCS

Control in DCS: The control function is executed using the DCS control block algorithm (e.g., PID, ratio control, etc.) with FF inputs and outputs connected to field devices and communication schedules in each segment done by the FF link/segment schedule master and usually the control system FF interface module with control system MMI for operations as setpoints, etc. Control on wire: In this case, all remains the same except that the control function execution for each loop is performed in the control blocks available in the FF field devices and usually the output device. (See also point 6 that covers an important aspect.)

Number of devices per H1 segment

The number of addressable devices per H1 segment can be a maximum of 32 (with repeaters a 240 maximum) but are often limited by practical application of cable length, loop scan and response time requirement, the power supply driving capability, the number of software and connection blocks available, the hazardous-area classification requirement, the philosophy adopted from safety and operational aspects. Usual practice is to have around 8–10 FF devices per segment depending largely on the project-specific philosophy.

Type and function blocks available in each FF field device and their execution time

The FF device function blocks are modeled/designed in line with a conventional DCS and are based on the standard Fieldbus Foundation definition and some additional as per the manufacturers. They are inputs, outputs, algorithm and also mode. The field-device function blocks definition and other device details come in soft form and can be used/imported in the DCS. The field device and functionalities are tested and certified for usage with the host DCS.

Process scan and end-to-end response times for the control.

This is a major shift from conventional practices since all loops in an H1 segment work in isochronous mode (i.e., all events occurring at the same time interval) and for all devices in the segment. All the segment devices shall have the same response time. The first definition point is the “macro-cycle,” the overall execution cycle time in an FF segment and the process/project specific requirement of macro-cycle. The typical fastest macro-cycle is 200 msec. The fastest response time or even scan controller time possible for an open or closed loop in the FF in that case as per the FF standard, irrespective of any system, shall be 200 msec. (This is even if there is only one open loop in the H1 segment and with control on the wire or control DCS. It will not be correct to calculate the response time of each loop by using the macro-cycle segment scan time and dividing by number of devices in the segment since the FF, H1 segment communicates in isochronous mode.) Also see point 6.

The FF, H1 segment module that controls the H1 link segment scheduling activity scan time defined in terms of “macro cycle” and the selectable options.

Every HI segment or link needs a master for communication and control activity and is called the link-activity scheduler. This is called the link-activity scheduler master and it schedules all the activity in the link like input and output processing, control algorithm execution, data transmission, initiating new data exchange device with host controller/ configuration communication, etc. Some of the activities are scheduled specifically for the known device activities as defined and are termed as scheduled activities. Besides, there are other activities such as alarm message and asset details, and diagnostics, Time synchronization, stamping, FW upgrading, adding new devices in the segment, etc., are unscheduled activity. Macro-cycle is the cyclic time in which the LAS executes and repeats itself in doing all the activities for that H1 segment and is actually the sum of the “scheduled + unscheduled activity”. The macro-cycle time is selectable and the typical options are 200 msec, 250 msec, 500 msec, 1 sec, 10 sec and 60 sec. The scheduled activity time can be planned and calculated depending on the number of devices in the segment/link, the activity and functions defined. The parameter details for each device in execution time (MIN_CYCLE_T) are required to calculate the scheduled activity time. The unscheduled activity time is usually calculated by considering not less than 50% time of the scheduled activities. The LAS can have a backup to take care of the failure, and the LAS device can be an LAS master-capable device. In the DCS system FF, the FF interface module is the LAS which is usually redundant. LAS activity schedule and timing in the control system, shown in the control system, only depicts the response and time schedule of the entities in it and not the actual control loop span/response time.

barrier can be used and hence, the advent of the IS power supply for the segment that takes care of the IS requirement. Intrinsically safe FF power supply.

This IS power supply has a power limitation to meet the hazardous-area requirement and is limited to around 180 ma and 11.5 volts. This limits the number of devices that can be mounted on the segment as each device. Besides, for all the devices in the segment, the FF power supply is nonredundant since it is not possible to design a redundant power supply and at the same time meet the IS criterion. (There is another prevalent nonindencive practice concept and here the FF power supply unit can deliver more power but still be nonredundant, and the area classification limits the usage of this concept.) 70

I DECEMBER 2010 HYDROCARBON PROCESSING

Power trunk FF power supply unit.

Another prevalent concept/philosophy used is the power trunk in the FF segment, where the power from the control room to the field is taken in non-IS restriction until the junction box (JB) is located in the field. This is the trunk in the FF segment and, in the explosion-proof JB and a set of IS barriers is used for individual instruments. This power supply can be redundant, and it has the capability to supply more power around 350 ma and 25 volts until the junction box and then IS barriers limit the power to individual devices. This concept is basically non-IS power until JB and the drop in the trunk is negated and then the IS barrier takes care of each device. A hybrid IS concept, and relatively more devices can be accommodated per the FF H1 card segment with no single failure point of all devices

due to the power supply. Note that, in this case, the JB design and barrier mounting arrangement are critical since IS barriers in case of live replacement by opening the explosion-proof JB, have minimum impact on plant safety in the hazardous area. It’s a compromise to an extent in a hazardous area but it is well accepted and taken care of in most JB designs. In hazardous areas with IS, the number of FF devices per segment depends on the powering philosophy adopted in addition to the other considerations in FF design and detailing. Scan time, end-to-end loop response time and macro-cycles.

This is a paradigm shift in concept from that of the conventional 4-20 ma DCS and often coupled with trying to verify/mea-

PROCESS CONTROL AND INFORMATION SYSTEMS TABLE 1. Foundation Fieldbus considerations (cont.) 6.

The H1 segment module (FF module) communication with the DCS controller and communication modes.

For the DCS, the FF module with LAS activity scheduling for each link is like any module/card and the DCS works in asynchronous mode or at a predetermined/fixed data exchange rate between themselves. The FF link module along with the H1 segment works, in isochronous mode. This point is important in estimating overall loop response or scan time. This is one of the key points on having control in the DCS/HOST or control on the wire. Irrespective of time synchronization between the LAS and the HOST controller, each has its own execution schedule and the two cannot match. In case of control on the host/DCS, actual data control execution processing of the data example in a PID block in the controller or open loop data acquisition point where data are sent to MMI: May not be the latest available from the link and HOST controller PID processing at a very fast rate may not be as per the overall FF isochronous macro-cycle response, as seen from the LAS activity scheduled display. Now this also happens in a conventional DCS with each IOP/ channel and PID algorithm in the controller working in asynchronous mode as per the schedule or prefixed scan rate. The point here in this case is to stress in deciding the overall effect on the FF system scan/response time and deciding on the macro cycle rate for the H1 segment and its associated design/detailing.

The DCS controller scan and response times for close control loops. (refer to 6 )

Scan time: The DCS controller scan time (in control in the DCS) is determined at the rate at which say here for a PID block scan is set. Now this processing is done with the FF segment data that, in turn, depends on the link/segment macro-cycle and also on each FF block execution time in the link and scheduling. The scan rate of any DCS control block as a rule should not be faster than the corresponding relevant function execution in the segment. Response Time (end-to-end response): The end-to-end response for control in the DCS depends on the following factors: 1. The macro-cycle time and the link /segment schedule 2. The data exchange scan time/rate between the controller and the link

FF field device refresh

Another important aspect of FF-based systems moving away from the conventional DCS is the transducer block of each device. This block, which refreshes data, works independently and is referred to as refresh time (the time required from the process control for data collection from the sensor and also exchanges data with device function blocks). This refresh time works on its own and has no relationship with the function block LAS. The data collected from the sensor are stored in the resource block buffer and is used by the function block. Thus, the data available to the device function blocks are not current but are at least one refresh cycle old. Typically, the refresh cycle time is < 1 sec.

sure for end-to-end response in the field using a conventional testing/measuring apparatus as is done in the conventional DCS system. The FF-based control system, with all its devices, is an integrated complete system. Each component in the FF application has unique characteristics and affects the overall system scan and end-to-end response times. In a typical FF system, the key factors are: • Control execution philosophy on wire or on DCS • Number of devices per H1 segment • Type and function blocks available in each FF field device and their execution time • Loop scan time and end-to-end response time for the control • The FF H1 segment module that controls the H1 link segment scheduling activity scan time defined in terms of “macro cycle” and the selectable options • The H1 segment module communication with the DCS controller and also with DCS MMI network. The communication data exchange rate and time synchronization • The DCS controller scan and response times for close control loops • FF field device refresh function block. Its importance in the above from the process control perspective. Table 1 outlines these factors: • The concept of scan time and endto-end loop response time in an FF-based

AI device E T

AI # 1 AI # 2 AI # 3 AI # 4 AI # 5 AI # 6 AI # 7 AI # 8

100

FIG. 1

200

300

400

E = Execution time =< 50 msec T = Transmission time =< 50 msec

500

600

700

800

900 1,000 Msec

An H1 segment with eight devices.

system is a major change from the conventional DCS, but is, at the same time, fully deterministic. • The scan rate and loop response time to be decided considering the overall process tolerance (which should be acceptable to most industry processes if the FF system is used). • Demanding too fast a scan rate and response time of any individual functionality in the FF system for the control loops may not drastically impact overall loop performance despite loading individual controller resources. • FF control on the wire, wherever possible in control schemes, apparently looks

better in terms of overall look response. However, the characteristics/features and each function block execution time in the devices and overall H1 segment performance and data exchange with controllers are factors that need to be considered. Typical example: In context of the

above discussions, following is an example of an H1 segment with eight FF AI field devices and the macro-cycle for the segment set at 1,000 msec. Brief explanation with eight AI function blocks: The FF communication is independent of the refresh cycle (see note 1 below). HYDROCARBON PROCESSING DECEMBER 2010

I 71

PROCESS CONTROL AND INFORMATION SYSTEMS The LAS initiates a new macro-cycle (here at 0 msec). Based on the LAS configuration, the AI #1 function block starts execution and in this case, after 50 msec, the data value can be obtained and, after a 50-msec delay the value can be transmitted for AI #1. Meanwhile, since the AI #1 value transmission starts in parallel, function block AI #2 execution begins and in the same way for other devices in the segment. For your application, under worst-case conditions and the macro-cycle is defined to 1,000 ms: After starting a macro-cycle, a minimum 50 ms delay time is required, and after 450 ms (8 x 50 ms-data transmission + 50-ms delay time) all the data are in the host. Required data transmission time is 400 msec, which leaves 600 msec for all other services (unscheduled activity, which is acyclic). Now, considering the device refresh time, (i.e., the time taken by the device in its own refresh cycle to get its value from the sensor independently of the other cycle times) the value in the host is older (1,000 msec macro-cycle + refresh cycle time). Refresh time: An example of refresh time for an FF transmitter measuring process temperature using a thermocouple as a primary sensing element is: The temperature sensor value will be refreshed cyclically (here called refresh cycle). The refresh cycle is independent of all other (important) cycle times. This refresh cycle is < 1 sec. This process value will be stored in an internal buffer of the FF field device. In context of the tabular points described (macro-cycle set as 1,000 msec; consider the device refresh time as 50 msec). Consider

data exchange from the FF interface module using the DCS controller function blocks: • The LAS master/gateway receives data from AI #1 block on the first macro-cycle (start time at 0 msec) after 100 msec (50 for function block execution + 50 for transmission). On the next cycle after completing the first macro-cycle (i.e., data from the second macro-cycle) at 1,050 msec (start time 0 msec). (These data are 1,000 msec old due to the refresh cycle, so the data available at the LAS host in the first cycle is 1,050 msec old and, in the second cycle a maximum of 1,050 msec older (after the first cycle the AI #1 data are refreshed 72

I DECEMBER 2010 HYDROCARBON PROCESSING

independent of the LAS schedule and also 1,050 msec older). • Now the LAS master/Gateway publishes or has the data ready for the DCS controller dependent on the data exchange rate. So add the delay of exchange rate. Note that this is an example of data acquisition and, in case of closed loop control relate with the controller PID block, processing time after the above two steps and the write data exchange. So time to exchange data between the controller and the H1 isochronous FF link working with available data (some older too) involved marginally effects of overall FF-based loop latency including that of open loops. Control on the wire vs. control on DCS-based FF system. For the

FF system implemented using the DCS FF function blocks, the biggest advantage is they all look a like; displays blocks and characteristics (also depends on the FF device capabilities). While in FF, engineering for control on the wire incorporates function-block displays as supplied by each FF device supplier and incorporating the same in the DCS for operation. Hence, for engineering and operations using DCS FF function blocks, the benefits are ease in operating FF and non-FF devices in the controller and reduced engineering efforts. Commonly used terms. Every FF

device consists of three basic blocks: 1. Resource block—describes characteristics of device 2. Transducer block(s)—represent local connections for physical I/O 3. Function block(s)—provide the device control and I/O behavior 4. All blocks have parameters • Standard (FF defined) • Manufacturer defined; all blocks have MODE. 5. Only FBs have inputs, outputs and algorithms and represent the interface between the FF device and the FF communication system. VCRs: • VCRs are handled by the network management part of the host and devices. • They are dedicated communication channels (like phone numbers). • Host VCR limits are specified per link (not per FF module). • VCR information is found on the NM (basic) tab of all devices and links; including maximum number and number used.

Each FF device and the LAS mater/gateway module has a defined number of function blocks and VCR handling capability. Rules: Each connected device uses twoVCRs of its own and consumes two link (host) VCRs (one is for device status and one is for alerts/alarms). Each FB INPUT connection consumes one VCR from the device and one from the device or link it connects to. Each FB OUTPUT connection consumes two VCRs from the device and two from the device or link it connects to—the second VCR is the BAK CAL. It is important to consider the application and the number of function blocks and VCRs available in the LAS master/gateway and also the availability of the same in the FF devices. Post-commissioned plant add-on and loop modification changes.

The post-commissioning scenario and maintenance aspect in FF-based installation calls for more meticulous planning and considering the existing system load. Replacement factors that need to be considered are function block execution time, algorithm availability used in the earlier configuration, and reconfiguration and engineering to be done online. The unscheduled loading for downloading after configuration also plays a major role in replacing existing field devices online. The LAS activity schedule is affected when adding a new FF device in the H1 link. Besides the physical parameter considerations, such as the power requirement, spur and overall segment length are also to be considered in modifying an existing FF installation. HP

Debasish Majumder is the director of engineering excellence center of Invensys India Private Ltd., based in Chennai, India. He is responsible for engineering deliverables for control and automation for Invensys projects in India. His 22 years of experience encapsulates all aspects from basic to detailed engineering, project engineering, site installation and commissioning, handling projects, maintenance and business development. He is intensively involved with globalization program for engineering and development and well versed in the the rapid development from pneumatic controls to electronic control systems and intensively engaged with the rapid development from single-loop electronic controller, relay based interlocking system, distributed control system and PLCs, etc. He is involved in recent Foundation Fieldbus, wireless technology, advanced process controls and hi-fidelity operator training simulators. Mr. Majumder holds a Bachelor of Technology degree in instrument technology from Madras Institute of Technology, Chennai, Anna University.

ROTATING EQUIPMENT/STANDARDS

What is new in API 610 11th Ed. (ISO 13709 2nd Ed.)? Updates to global specifications address pump reliability and much more F. KORKOWSKI, Flowserve, Vernon, California; R. L. JONES, Consultant, Houston, Texas; and J. D. SANDERS, Fluor, Sugar Land, Texas

henever a new edition of a global specification such as the International Organization for Standardization (ISO) or the American Petroleum Institute (API) is released, there is often confusion about the explicit details and rationale of the changes made. This article will address changes that have been incorporated into the new ISO 13709 2nd Edition - ANSI/API Standard 610 11th Edition, Centrifugal pumps for petroleum, petrochemical and natural gas industries. It will specifically discuss significant changes that will impact pump reliability as well as other key changes. Background. Before reviewing these changes, it is important

to understand the background surrounding this new API 610 edition. Developed cooperatively with the ISO 13709 2nd Edition, which was released on December 15, 2009, the API 610 11th Edition was completed in July 2010 and was completed in July 2010 and published in September. The ISO 13709 and API 610 documents are identical in content with the exception of a few minor editorial corrections in the API 610 version. API documents are routinely updated every five years. The API 610 9th Edition was released in January 2003 and reissued as the 10th edition in October 2004. The API taskforce/ISO Working Group began updating efforts on the 11th edition in 2006. The team addressed the latest developments concerning rotating equipment, including reliability issues, industry issues and proposed changes based upon proven engineering and operating practices. Collaboration with other industry groups such as the Hydraulic Institute, International Electrotechnical Commission (IEC), National Electrical Manufacturers Association (NEMA) and ASTM ensured that the 11th edition reflected those organizations’ latest updates. Historically, new editions have transitioned into worldwide usage over a period of approximately two years. During this timeframe, engineers typically choose to make purchases that embrace a new edition’s changes, especially those affecting equipment reliability. Changes addressing reliability. For the new API 610 11th Edition, there are three significant changes with potential impact on pump reliability. These include: • The addition of Annex K Section K.1 Shaft Stiffness Guidelines for Overhung Pumps, which pertains to OH2 horizontal centerline mounted overhung pumps and OH3 vertical inline pumps

• The addition of Annex K Section K.2 Bearing “system” life considerations for OH2, OH3, BB1,BB2, and BB3 pumps • The expansion of Torsional analysis, rewritten to explain when each type of analysis is required. Annex K Section K.1 Shaft Stiffness Guidelines for Overhung Pumps. The API taskforce/ISO Working Group

reviewed input from a number of users/contractors who reported evaluating shaft stiffness and discovering wide shaft flexibility among manufacturers. The shaft flexibility index was developed as a straightforward tool to evaluate a true API pump design vs. one that is purportedly labeled API but does not meet the standard’s design requirements. Shaft stiffness became the differentiator. Fig. 1 shows a simple overhung rotor with D equaling the shaft diameter under the mechanical seal sleeve and L equaling the distance from the impeller centerline to the radial bearing. Shaft flexibility index or ISF in its shortened expression is ISF = L 3/D 4. Fig. 2 shows historical data from various overhung pumps. In this figure, Kt, the pump “sizing” factor, is equal to a pump’s BEP flow x TDH/rotating speed. This illustration charts smaller pumps as having higher L 3/D 4 than larger pumps. The guideline for L 3/D 4 is that as long as a pump is below the line of shaft stiffness it is following industry practice. If L 3/D 4 exceeds the line by 20%, then the customer should seek justification from a pump manufacturer for its design. Fig. 2 represents modern design overhung pumps. Some of the overhung designs reviewed in constructing Fig. 2 actually exceeded the L

FIG. 1

Overhung pump.

HYDROCARBON PROCESSING DECEMBER 2010

I 73

ROTATING EQUIPMENT/STANDARDS TABLE 1. Old vs. new OH2 designs Refinery A

QTY conversions

QTY repairs before upgrade

Run time, months before upgrade

QTY repairs after upgrade

Run time, months after upgrade

MTBR, before upgrade

MTBR, after upgrade

129

2,114

511

1,458

265

149

3,103

471

Total

351

6,674

1,248

104 A B C D E F G

103

ISF = L3/D4

102

10-1

10-2 10 FIG. 2

102

103

104 K1 = QH/N

105

106

107

Overhung pump shaft flexibility vs. size (USC unit s.) (Source: ISO 13709 2nd Ed./API 610 11th Ed. Appendix K Figure K.3).

L 3/D 4 shaft stiffness line by a factor of 10. It is important to realize that providing the customer with the L 3/D 4 ratio is an “if specified” bulleted item in API. Table 1 shows the differences between previous OH2 designs with characteristic long slender shafts and extremely high L 3/D 4 values and today’s robust designs with shorter shaft spans, larger diameters and very low L 3/D 4 values. In this sampling, the mean time between repair (MTBR) improved from < two years to > four years. (Values derived by dividing run time by number of repairs.) This table illustrates the importance of shaft stiffness evaluations and the positive impact that properly designed pumps can have on MTBR performance. Annex K Section K.2 Bearing ‘system.’ Life considerations for OH2, OH3, BB1, BB2 and BB3 pumps are discussed here. For decades API 610 has required each individual bearing to be designed for a life of 25,000 hours (i.e., three years) continuous operation (at rated flow) and 16,000 hours at maximum radial and axial loads, typically minimum continuous stable flow. Recent work on other API standards raised the issue of this being inconsistent with the API requirement for pump design to be suitable for a three-year uninterrupted run. The identified problem is that “system” life is shorter than the shortest life of the individual bearings in the system. For years, all manufacturers have treated this requirement as applicable to “each” bearing instead of the bearing “system.” The 11th edition 74

I DECEMBER 2010 HYDROCARBON PROCESSING

has added a new formula to this section that calculates system life, noting that the combination of both radial and thrust bearings (system) should comply with the ISO/API bearing life requirements. Results from this formula show that system life is shorter than the shortest life of the individual bearings in a system. For example, if each bearing by itself had a life of 37,500 hours, the bearing system life, when calculated by the new formula, would be only 25,000 hours. Bearing loads, and ultimately bearing life, are functions of pump type, type of impeller (single suction vs. double suction), impeller configuration (balance holes, no balance holes), wear ring diameters, suction pressure and bearing types. Traditionally, API 610 has had various requirements that build a “safety factor” into selection of the bearings. It is also a fact that these requirements be derived from the largest set of hydraulics for each bearing housing size. This means that, for all or most other hydraulics, API 610 pump designs should automatically exceed the bearing life requirements. This positively impacts API 610 pump manufacturers who should not have to change their designs. However, in cases where the bearing system life number does not comply, discussion is needed. Pump manufacturers have a number of tools that they can use in these more extreme applications. Changing bearing type or size, or perhaps using unbalanced construction in a pump, are some of the typical ways to increase bearing life. Applying bearings larger than those in current service may come at the price of increased bearing temperature or may even produce bearing skidding. So, in these cases, discussion of the available options is imperative. The requirement to provide bearing life figures is also an “if specified” bulleted item in API. Torsional analysis. The subcommittee rewrote this section in response to persistent questions from customers regarding the types of torsional analysis and when each type is required. A flow chart was composed to provide a simple “yes/no” decision-tree method for practical guidance. The three types of analysis are: a) Undamped natural frequency b) Steadystate damped response analysis c) Transient torsional. Other key changes. Key changes to address include:

Basic nomenclature. Net positive suction head required (NPSHR) has been replaced with NPSH3—net positive suction head required, in meters (feet). This more accurately reflects that NPSH testing is based upon a 3% head drop. Therefore, the “R” was deleted and the “3” was added to the NPSH term (NPSH3). ‘Flammable’ and ‘hazardous’ terms. These terms have been removed from the entire API 610 11th Edition document. There has been much controversy surrounding the meaning of “flammable” and “hazardous.” National Fire Protection Association

ROTATING EQUIPMENT/STANDARDS (NFPA) and other agencies define these terms differently, and, in reality, the purchaser decides what is flammable and/or hazardous. The purchaser can decide to use an API pump in a selected service. Thus, there is no impact if the terms are dropped from the document. NACE MR 103 and NACE MR 175. Distinction has been made to help understand when each National Association of Corrosion Engineers (NACE) document applies. NACE MR103 becomes the key document applicable to oil refineries, liquefied natural gas (LNG) plants and chemical plants. The traditionally used NACE MR175 is now specifically noted as applying to sulfide- and chloride-stress-corrosion cracking services in oil and gas production facilities and natural-gas sweetening plants. For years, NACE MR175 was the only NACE document that was applied for materials subjected to stress-corrosion cracking covered in API 610. Upon further investigation, it was learned that NACE MR103 was, in fact, more applicable to the majority of equipment purchased to API 610, so it has been added. Pump performance testing. The document now emphasizes that performance testing should be conducted based upon the “uncertainty” requirements of ISO 9906, which primarily addresses instrumentation controls. Test tolerances have changed to +/- 3% across the board for differential heads of 0 m to > 300 m (1,000 ft), representing a slight shift in the mid- to high-head regions. There has been a tightening in the low-head range, which is now defined as 0 m to 75 m (250 ft) vs. the previous 0 to 150 m (500 ft). Defined recorded test points have also slightly changed, as reflected in Table 2. Non-destructive testing (or NDE). Guidelines for NDE were first introduced in the 10th edition. The subcommittee members agreed that more definitive guidelines for pressure casings, nozzles and connection welds were needed. The 11th edition now expands NDE to include when certain types of NDE are required. New Table 14 defines “three inspection classes”—I, II and III. Class I applies when minimal visual inspection is needed—basically for all services. Class II applies to situations when the casing is >80 percent MAWP and > 200°C (392°F) and requires magnetic particle (MT) and dye penetrant (PT) inspections. Class III is for extremely hazardous services. These include services for fluids with: low specific gravity (< 0.50 sg) with temperatures to 200°C (393°F), low specific gravity (< 0.70 sg) with temperatures > 200°C (392°F) and temperatures > 260°C (500°F). All require additional radiographic (RT) and ultrasonic (UT) inspections. Annex N: Pump data sheets and electronic data exchange. Data sheets have been extensively improved. Previously designed to be completed with a pencil, the new datasheets are now electronic. Rather than words and a circle appearing for each datasheet option, there is simply a blank with a “drop down” list. In addition to their ease of use on a computer, the new datasheets eliminate all the circles and choices, making them shorter. Pipe gusseting. This new item details pipe gusseting (when required). With more customers seeking gusseting of pipe connections to the pump casing, the subcommittee decided to include these details in the ISO/API document. Note: However, that gusseting remains an “if specified” by customer item. Conclusion. Even though these new changes and additions to

API 610 11th Edition (ISO 13709 2nd Edition) are designed to increase MTBR, it is also essential to: • Choose correctly designed pumps with the right materials for your specific applications

TABLE 2. Defined recorded test points, comparison of 10th and 11th editions 11th Edition

10th Edition

1. Shutoff

2. Minimum continuous stable flow

3. Between 95% and 99% of rated flow

3. Midway between minimum and rated flow

4. Between rated and 105% rated flow 4. Rated flow 5. Approximate BEP flow (if rated is not within 5% of BEP flow)

5. Maximum allowable flow (120% BEP as a minimum)

6. End of “allowable” operating range

• Follow best practices for pumping • Assess complete operational systems to identify and eliminate problematic equipment. The result will be greater reliability and efficiency and lower operating costs. HP

Frank Korkowski is product manager for Flowserve educational services and marketing manager for the API 1 & 2 stage process pumps. He has spent 36 years in various pump roles with Ingersoll Rand, Ingersoll-Dresser Pumps and currently Flowserve. Positions have included project manager for nuclear pumps, supervisor application engineering, business unit alliance manager, team captain and product manager for overhung process pumps. Mr. Korkowski received his BS degree in industrial engineering from the New Jersey Institute of Technology, with post graduate studies in engineering and business administration at Lafayette College and Fairleigh Dickenson University. He was one of the Flowserve representatives on the API 610 Subcommittee taskforce for producing the ISO 13709/API 610 11th Edition document.

Roger L. Jones is a rotating equipment consultant currently providing support to the Ras Tanura Integrated Project. He spent 32 years in various positions at different Shell companies. In his career, he has held numerous technical and managerial positions. These positions have been in chemical plants and refineries, major capital projects and in engineering consulting roles. Most recently, he spent three years in China on the Changbei Gas Field Development project. Mr. Jones has been in the rotating equipment field for more than 30 years. He received his BS and MS degrees in mechanical engineering from Kansas State University and is a registered professional engineer in the state of Texas. He represented Shell on the API Subcommittee on mechanical equipment and is a former chairman of the subcommittee. Mr. Jones is the taskforce chairman of API 610 and the International Standards Organization (ISO) convenor for the 13709 Working Group. He is the previous chairman of the International Standards Coordinating Committee of the API and head of the US delegation to the various ISO technical committees governing standards for refining and offshore equipment. Mr. Jones is a former member of the International Pump Users Symposium Advisory Committee.

Jack D. Sanders is a Senior Fellow in mechanical engineering with Fluor in Sugar Land, Texas. His responsibilities include: preparation and review of specifications, equipment selection and evaluation; coordination with equipment suppliers and other engineering disciplines; testing; and installation of rotating equipment. He has worked in the application of rotating equipment in the petrochemical, refining and power generation industries. Mr. Sanders has specific experience in crude, vacuum, hydrotreating, FCCU, delayed coking, polyethylene, ethylene glycol, olefins and offsites. He has more than 40 years of experience in rotating equipment. Prior to joining Fluor, he worked for two API pump manufacturers and was a plant engineer at two refineries. Mr. Sanders received his BS degree in mechanical engineering from The University of New Mexico. He is a registered professional engineer in the state of Texas. He represents his company on the API Subcommittee on mechanical equipment, is a member of API 610 and API 671 taskforces, and is serving as the taskforce chair for API 611. HYDROCARBON PROCESSING DECEMBER 2010

I 75

20 11

GULF PUBLISHING COMPANY EVENTS

SAVE THE DATE PROCESS CONTROLS AND INSTRUMENTATION CONFERENCE 9-11 March 2011 • Moody Gardens, Galveston, Texas Hosted by both World Oil and Hydrocarbon Processing, the Process Controls & Instrumentation Conference will be devoted to advancing process control and instrumentation in the oil and gas industry. www.GulfPub.com/PCI

MARKETING IN THE OILFIELD CONFERENCE August 2011 • Houston, Texas The Marketing in the Oilfield Conference provides an environment to learn new ideas and strategies in addition to numerous opportunities to network with fellow upstream and downstream marketing peers. This conference focuses on industry hot topics related to marketing, social media, communication issues and includes featured keynote experts and presentations relevant to the topic in focus. www.GulfPub.com/MITO

WOMEN’S GLOBAL LEADERSHIP CONFERENCE IN ENERGY & TECHNOLOGY October 2011 • Houston, Texas Hosted by both World Oil and Hydrocarbon Processing, the Women’s Global Leadership Conference in Energy & Technology is the largest women’s event in the industry, and the only one that focuses on discussing the industry’s key environmental, economic, professional development and human capital issues in one setting. Attendees leave the conference with an increased understanding of the full range of pertinent issues and an increased ability to be change agents in our industry. This conference continues to encourage the growth and leadership of women in the industry and the respect and knowledge of energy and technology. www.WGLNetwork.com For more information about Gulf Publishing Company events or to work with us to create a new event, visit www.GulfPub.com/Events, e-mail Events@GulfPub.com, or call +1 (713) 520-4475. (Event topics and dates are subject to change.)

GULF P U B L I S H I N G C O M PA N Y

INSTRUMENTS AND NETWORKS

Development of support vector regression-based soft sensor Application was used in a commercial ethylene glycol plant S. K. LAHIRI, S. SAWKE, N. KHALFE and J. AL GHAMDI, National Institute of Technology, Durgapur, India

ow carbon dioxide (CO2) in a cycle gas loop of an ethylene glycol (EG) plant improves catalyst selectivity and overall plant economics. CO2 produced as a byproduct in an ethylene oxide (EO) reactor is removed by a recycle gas purification process. In this process, the carbonate and bicarbonate ratio in lean carbonate solution is considered as an important quality control (QC) variable since it largely depends on CO2 removal efficiency. If there is a process malfunction or if operating in less than optimal conditions, the CO2 content in a cycle gas loop will continue to rise until corrective action is taken. Lab results must be obtained for the carbonate and bicarbonate ratio. This laborintensive process can be overcome by implementing a technological solution in an accurate form and a robust mathematical model capable of real-time QC variable prediction. For well-understood processes, the correlation structure for QC variables as well as input choices may be known in advance. However, the recycle gas purification process is too complex and the appropriate correlation form and input variable choices are not obvious. This article describes a systematic approach to the inferential measurement developments of the carbonate and bicarbonate ratio, using support vector regression (SVR) analysis. Given historic process data, a simple SVR-based soft sensor model is found capable of identifying and capturing the cause-and-effect relationship between operating variables (model input) and QC variables (model output). Special care was taken to choose input variables, so that the final correlation and regression co-efficient was understood by process engineers. The developed soft sensor was applied in a commercial EG plant with a commercially available online data historian interface that satisfactorily predicted the carbonate and bicarbonate ratio in real time. Soft sensors have been reported to supplement online instrument measurements for process monitoring and control. Both model-based and data-driven soft sensors have been developed. If a first principle model (FPM) describes the process accurately, a model-based soft sensor can be derived.1 However, a soft sensor based on detailed FPM is computationally intensive for real-time applications. Difficulties associated with the construction and solution of phenomenological models for soft sensors necessitates exploring alternative modeling formalisms. Modern measurement techniques enable large operating data amounts to be collected, stored and analyzed, thereby rendering data-driven soft sensor development a viable alternative. In conventional empirical soft sensor modeling, appropriate linear or nonlinear models are constructed exclusively from the

process input-output data without invoking the process phenomenology. A fundamental deficiency of the conventional empirical modeling approach is that the data-fitting model structure (functional form) must be specified a priori. Satisfying this requirement, especially for a nonlinearly behaving process, is cumbersome and involves heuristically selecting an appropriate nonlinear model structure from numerous alternatives. In the last decade, artificial neural networks (ANNs) and more recently SVR analysis has emerged as two attractive tools for nonlinear modeling, especially in situations where the development of phenomenological or conventional regression models become impractical or cumbersome. Recently, SVR2,3 is gaining popularity over an ANN due to its many attractive features and promising empirical performance. The salient SVR features include: • SVRs are exclusively data-based nonlinear modeling paradigm and similar to ANNs. • Models are based on the structural risk minimization principle that equips them with greater potential to generalize. • Parameters are obtained by solving a quadratic optimization problem. • The objective function, being a quadratic form, possesses a single minimum, thus avoiding the heuristic procedure involved in locating the global or the deepest local minimum on the error surface. • Inputs are first nonlinearly mapped into a high dimensional feature space wherein they are correlated linearly with the output. Although the SVR paradigm foundation was laid out in the mid1990s, its chemical engineering applications such as fault detection4,5 have emerged only recently. Building on these studies, the focus was to develop a soft sensor for estimation of carbonate to bicarbonate ratio in a CO2 removal unit of an ethylene glycol plant. It was reported2,4,7 that SVR-based soft sensors can handle noise in process parameters [this type of noise is common in process indication of distributed control systems (DCSs)] and gives better performance than normal nonlinear regression-based soft sensors. Based on SVR analysis potential to regress a complex function, an attempt was made to explore the SVR computational capability in the field of soft sensor development in the petrochemical industry. What follows presents a systematic approach using robust SVR techniques to build a soft sensor from available process measurements. SVR-based soft sensor modeling. SVR is an adaptation of a recent statistical learning theory-based classification paradigm, namely support vector machines.2 The SVR formulation follows HYDROCARBON PROCESSING DECEMBER 2010

I 77

INSTRUMENTS AND NETWORKS

f(x) + d f(x)

f(x) – d y J Data points Points outside tube Support vectors Fitted by SVR x FIG. 1

Support vector schematic regression using ␧ – sensitive loss function.

a structural risk minimization (SRM) principle, as opposed to the empirical risk minimization (ERM) approach that is commonly employed within statistical machine learning methods and also in training ANNs. In SRM, an upper bound on the generalization error is minimized as opposed to ERM which minimizes prediction error on the training data. This equips the SVR with greater potential to generalize the input-output relationship learned during its training phase for making good predictions for new input data. The SVR is a linear method in a high dimensional feature space that is nonlinearly related to the input space. Though the linear algorithm works in the high dimensional feature space, in practice it does not involve any computations. This is due to the kernel usage since all necessary computations are performed directly in the input space. Basic SVR concepts are introduced while the literature cited gives more detailed descriptions. Consider a training data set g = {(x1, y1), (x2, y2),…, (xP, yP )}, such that xi N is a vector of input variables and yi is the corresponding scalar output (target) value. Here, the modeling objective is to find a regression function, y = f (x), such that it accurately predicts the outputs {y} corresponding to a new set of input-output examples, {(x, y)}, that are drawn from the same underlying joint probability distribution, P(x, y), as the training set. To fulfill the stated goal, SVR considers the following linear estimation function:

f (x ) =< w, (x) > +b

(1)

where w denotes the weight vector; b refers to a constant known as “bias”; f (x) denotes a function termed feature, and < w, (x) > represents the dot product in the feature space, , such that : x ,w . In SVR, the input data vector, x, is mapped into a high dimensional feature space, , via a nonlinear mapping function, , and a linear regression is performed in this space for predicting y. Thus, the problem of nonlinear regression in lower dimensional input space N is transformed into a linear regression in the high dimensional feature space, . Accordingly, the original optimization problem involving nonlinear regression is transformed into finding the flattest function in the feature space, , and not in the input space, x. The unknown parameters w and b are estimated using the training set, g. To avoid over-fitting and thereby improving the generalization capability, following regular functions involving an empirical risk summation, the complexity term w 2, is minimized:7

Rreg[ f ] = Remp[ f ]+ w P

= C( f (xi) yi) + w

=0 otherwise (3) where ␧ is a precision parameter representing the tube radius located around the regression function, see Fig. 1. The region enclosed by the tube is known as “e-intensive zone”. The SVR algorithm attempts to position the tube around the data as shown in Fig. 1. The optimization criterion in Eq. 3 penalizes those data points whose y values lie more than ␧ distance from the fitted function, f (x). In Fig. 1, the size of the stated excess positive and negative deviations are depicted by ␨ and ␨*, which are termed “slack” variables. Outside the [–␧, ␧] region, slack variables assume nonzero values. The SVR fits f (x) to the data in a manner such that: • The training error is minimized by minimizing ␨ and ␨*. • w 2 is minimized to increase the flatness of f (x) or to penalize over-fitting function complexity. Research shows that Eq. 43 possesses a finite number of parameters that can minimize the regularized function in Eq. 2. P

f (x, , *) = ( i i*)K (x, xi) + b

I DECEMBER 2010 HYDROCARBON PROCESSING

(2)

(4)

i=1

where ␣i and ␣i* (>=0) are the coefficients (Lagrange multipliers) satisfying ␣i ␣i* = 0, i = 1, 2, …, P, and K(x, xi) denotes the so called “kernel” function describing the dot product in the feature space. The kernel function is defined in dot product terms of the mapping function as given by:

K (xi, xj) =< (xi), (xj ) >

(5)

The advantage of Eqs. 4 and 5 is that for many set choices { i(x )} , including infinite dimensional sets, K form is analytically known and very simple.8 Accordingly, the dot product in the feature space i, can be computed without actually mapping the vectors xi and xj into that space (i.e., computation of [x(i) and (x( j )] . There exists several choices for the kernel function K; the two commonly used kernel functions, namely, radial basis function (RBF) and nth degree polynomial, are defined in Eqs. 6 and 7, respectively. xi xj K (xi, xj) = exp 2 2

K (xi, xj) = 1+ (xi, xj )n

i=1

where Rreg and Remp denote the regression and empirical risks, respectively; w 2 is the Euclidean norm; C(.) is a cost function measuring the empirical risk, and > 0 is a regularization constant. For a given function, f, the regression risk (test set error), Rreg ( f ), is the possible error committed by the function f in predicting the output corresponding to a new (test) example. The input vector is drawn randomly from the same sample probability distribution, P(x, y), as the training set. The empirical risk Remp ( f ), represents the error (termed “training set error”) committed in predicting the outputs of the training set inputs. Minimization task described in Eq. 2 involves: • Minimizing the empirical loss function, Remp ( f ) • Obtain the smallest w as possible, using the training set g. The commonly used loss function is the “e-insensitive loss function” given as:2 C ( f (x ) y) = f (x ) y For f (x ) y

(6)

(7)

In Eq. 4, the coefficients ␣i and ␣i* are obtained by solving the following quadratic programming problem.

INSTRUMENTS AND NETWORKS Maximize:

K2CO3 ⫹ CO2 ⫹ H2O r 2KHCO3 2KHCO3 r K2CO3 ⫹ CO2 ⫹ H2O

R( *, ) = 0.5 ( i * i)( j * j)K (xi, xj) i, j =1

i=1

( i * + i) + yi( i * i) subject to constraints: P

0 i, i* C , i and ( i * i) = 0

(8)

i 1

Having estimated ␣, ␣* and b, using a suitable quadratic programming algorithm, the SVR-based regression function takes the form: P

f (x,w) = f (x, , *) = ( i * i)K (xi, x ) + b

(9)

i=1

where vector w is described in Lagrange multiplier terms ␣ and ␣*. Owing to the specific character of the previous described quadratic programming problem, only some coefficients, ( i * i)K are non zero and the corresponding input vectors, xi, are support vectors (SVs). SVs can be informative data points that compress the training set information content. The coefficients ␣ and ␣* have an intuitive interpretation as forces pushing and pulling the regression estimate f (xi) toward the measurements, yi. For Eq. 9, the bias parameter, b, can be computed as follows: yi f (xi)b = 0 For i (0,C ) b = yi f (xi)b = 0 + For i* (0,C ) where xi and yi denote the ith support vector and the corresponding target output, respectively. In SVR formulation, C and ␧ are two user-specified free parameters; while C represents the tradeoff between the model complexity and the approximation error, ␧ signifies the ␧-insensitive zone width used to fit training data. The stated free parameters together with the specific kernel function form, control accuracy and generalization performance of the regression estimate. The judicious procedure selection of C and ␧ is further discussed in reference material.9

(12) (13)

The CO2 absorption by carbonate solution and CO2 stripping in the regenerator column is a continuous process and has a long-term effect on the reactor performance and the overall glycol plant economics. The carbonate solution’s CO2 removal capacity depends mainly on carbonate flow, total carbonate strength in solution, the carbonate to bicarbonate ratio and the inlet cycle gas CO2 concentration. The CO2 removal section’s removing capacity has to be gradually increased throughout the catalyst life, about three years. The catalyst selectivity drops from start of run (SOR) to end of run (EOR) and consequently the CO2 generation in the EtO reactor gradually increases. Therefore, carbonate flow gradually increases from SOR to EOR over the catalyst’s life. The carbonate to bicarbonate ratio represents the regenerator column performance and has a profound effect on CO2 absorption in the contactor. If the carbonate to bicarbonate ratio in the lean carbonate solution is low, it will reduce the CO2 absorption capacity in the contactor while the CO2 outlet gas will increase. The reason for the low carbonate to bicarbonate ratio is that less heat is applied in a regenerator column, thus bicarbonate to carbonate conversion is low. Unlike other parameters that affect unit performance, the carbonate to bicarbonate ratio is not measured on a continuous basis and only offline laboratory sampling is available. There is a need for continuous online ratio analysis as it will help to monitor and adjust the regenerator performance. Due to nonlinearity of the dynamics and complex electrolyte chemistry involvement, it is very difficult to develop a first principle-based model for the regenerator and contactor. Also, a few of the phenomenological-based models that were available in literature were too complex and had a long execution time, thus making them unsuitable for use in online soft sensors. With this information, the SVR study was implemented for making an online soft sensor. How to make an online soft sensor for carbonate to bicarbonate ratio. The following gives detailed information

on proper soft sensor construction. Process description. In an EG plant, ethylene oxide (EtO) is

produced by a gas phase catalytic ethylene and oxygen reaction in a shell-and-tube type reactor at 20 barg pressure and high temperature, refer to Eq. 10. The side reaction, Eq. 11, also occurred at that temperature and pressure and undesired CO2 was produced. Primary reaction: C2H4 ⫹ ½ O2 r C2H4O Secondary reaction: C2H4 ⫹ 3O2 r 2CO2 + 2H2O

(10) (11)

The CO2 was removed from the cycle gas system since it would decrease the catalyst selectivity and increase the system pressure. The CO2 from a cycle gas system is removed by a recycle gas purification process. Eq. 12 illustrates that CO2 is absorbed in a contactor by contacting cycle gas with hot potassium carbonate (K2CO3) solution. The outlet carbonate solution from the contactor [containing unreacted K2CO3, potassium bicarbonate (KHCO3) and water] is flashed in two low-pressure flash vessels to remove dissolved ethylene and methane. The liquid outlet from the flash vessel is fed to the regenerator column, where K2CO3 is again converted back to carbonate using heat, Eq. 13. The regenerated carbonate solution is pumped and recycled back to the contactor column for reuse.

Input selection. Based on extensive literature surveys and plant

operating experience, the input variables illustrated in Fig. 2 were selected that have an effect on the carbonate to bicarbonate ratio. Data collection. Average hourly plant operation data was col-

lected from a plant information management system (PIMS) for an entire year. The data were collected at different catalyst ages and different carbonate flow to cover a wide range of operating data. The carbonate/bicarbonate ratio data were collected from lab sample three times per week for one year. Data regression. Initially, all input and output data was put

in a spreadsheet and data alignment was done. Based on process experience and looking into the dynamics of the process, plant operating data were averaged on a 3-hr basis and aligned with the less frequent lab analysis. Nonlinear data regression was tried to get a correlation that would represent the ratio as function of all input variables listed above. All type of equation was tried (polynomial with varying degrees, exponential radial basis functions, spline etc.). Input parameters were arranged in those equation that gave engineering sense. The statistical analysis of prediction is based on the following performance criteria: HYDROCARBON PROCESSING DECEMBER 2010

I 79

INSTRUMENTS AND NETWORKS

Tags DI-1 F1 F2 T1 F3

Description Online density meter of lean carbonate Reboiler steam flow Direct steam flow to regenerator Regenerator bottom temperature Lean carbonate flow

1st ﬂash drum

P-4

P-5

Cycle gas out

2nd ﬂash drum

P-8

P-7 F3

Carbonate to bicarbonate ratio

Input parameter list

1.9 1.7 1.5 1.3 1.1 0.9 0.7 0.5

Actual lab analysis Prediction

Lean P-11 carbonate

FIG. 4

Heat exchanger

Rich carbonate P-1 Contactor

P-14

P-19

Regenerator P-10

DI-1

SVR prediction

P-2 Live LP steam

Cycle gas in

P-12

Reboiler T1

P-9

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0

9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 Sample number

Actual vs prediction of carbonate to bicarbonate ratio using RBF kernel in SVR.

0.2

0.4

0.6

Carbonate pump

Carbonate to bicarbonate ratio

FIG. 2

1.9 1.7 1.5 1.3 1.1 0.9 0.7 0.5

Block diagram of a CO2 removal unit.

Actual lab analysis Prediction

FIG. 3

Actual vs prediction for simple nonlinear regression model.

1. The average absolute relative error (AARE) should be minimum 1 N ypredicted yactual AARE = N 1 yactual 2. The correlation co-efficient (CC) should be maximum (as near to unity) ( yactual yactual )( ypredicted ypredicted ) CC = ( yactual yactual )2

( ypredicted ypredicted )2

After several trial and errors it was found that simple nonlinear regression performances were very poor and no equations gave acceptable AARE and CC. The prediction was unable to catch the increase and reduction of ratio that were considered basic requirements for a soft sensor (refer to Fig. 3). After an extensive literature survey, the SVR was found promising to correlate such types of difficult parameters. A computer program was made for an SVR, based on the information gathered. Results and discussions. After collecting around 160 data

sets, 80% was segregated as the training set and 20% (chosen

I DECEMBER 2010 HYDROCARBON PROCESSING

1.4

1.6

1.8

Parity plot for SVR prediction and actual lab analysis of carbonate to bicarbonate ratio.

TABLE 1. SVR algorithm performance for different kernel types Sr #

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 Sample number

FIG. 5

0.8 1.0 1.2 Actual lab analysis

⑀

Kernel parameter

10,000

0.01

10,000

0.01

Kernel type

AARE

Polynomial

0.067

0.637

RBF

0.012

0.937

Spline

0.068

0.611

randomly) as the test set. SVR was performed on the training set data but performance was judged by the test set data. To obtain an optimal SVR model, it was necessary to examine the effects of kernel function and other algorithm specific4 parameters; the three kernel functions tested were polynomial, RBF and spline. Among these, RBF resulted in the least AARE values for test sets (refer to Table 1). The number of SVs used by the SVR algorithm for fitting the carbonate to bicarbonate ratio model was 144, which corresponded to 90% of the data. The optimal values of SVR-specific parameters were width of RBF kernel(s), cost coefficient (C ) and loss-function parameter (␧), along with minimized training and test error are listed in Table 1. These values were attained after running numerous SVRs for all different combinations of the three parameters. Also listed are correlation coefficient values for the test set predictions along with the corresponding AARE values for different models. An SVR model comparison was predicted and the corresponding target values of the carbonate and bicarbonate ratio is depicted in Fig. 4. As Fig. 4 shows, there is very good agreement in the SVR prediction and the actual lab analysis of the carbonate and bicarbonate ratio. Fig. 5 represents the prediction success of SVR algorithm against actual lab analysis. One or two outlier data in the parity plot may be due to the misrepresentative lab analysis that also shows the SVR tolerance algorithm against outliers. Implementation and benefit. After a satisfactory SVR prediction performance, the formula was applied to another EG plant.

INSTRUMENTS AND NETWORKS Data and performance were found quite satisfactory. This illustrates the generalization capability and robustness of a developed soft sensor. The soft sensor formula was put in an onlilne data historian interface to operate on a real-time basis. Performance was monitored in a commercial EG plant for one month against lab analysis. Agreement had occurred between the lab analyses, and it captured the downward and upward trend of the ratio properly. After receiving an online real-time indication for the carbonate and bicarbonate ratio, it was easy for a panel operator to vary steam in a regenerator to maintain the ratio within an acceptable limit. By controlling the ratio, the overall performance of a CO2 removal unit is improved, which in turn has a long-term benefit in catalyst selectivity and overall EG plant economics. The prediction error (within 1.2%) and high correlation co-efficient (0.937) renders SVR algorithm promising for future soft sensor development. HP 1 2 3 4

tor machines: A useful tool for process engineering applications,” Chemical Engineering Progress, pp. 57–62, January 2003. Jack, L. B. and A. K. Nandi, “Fault detection using support vector machines and artificial neural networks augmented by genetic algorithms,” Mechanical System Signal Processing, Vol. 16, pp. 372–390, 2002. Cherkassky, V., and Y. Ma, “Practical selection of SVM parameters and noise estimation for SVM regression,” Neurocomputing, 2002.

Dr. Sandip Kumar Lahiri is a professional chemical engineer with more than 17 years experience in plant production, process engineering and APC development. He’s published over 25 papers in international journals related to artificial intelligence, flow modeling and soft sensor development.

Sunil Sawke is a professional chemical engineer with more than 20 years experience in process engineering, plant operations, technical training and commissioning of petrochemical plants. He has numerous papers published related to developments of soft sensors, process simulations, debottlenecking and training strategies.

LITERATURE CITED Vapnik, V., The Nature of Statistical Learning Theory, Springer Verlag, New York 1995. Vapnik, V., Statistical Learning Theory, New York, 1998. C. Burges, “A tutorial on support vector machines for pattern recognition,” DataMining and Knowledge Discovery, Vol. 2, No. 2, pp. 1–47, 1998. Smola, A. J., B. Schölkopf and K. R. Müller, “The connection between regularization operators and support vector kernels,” Neural Networks, Vol. 11, pp. 637–649, 1998. Schölkopf, B., J. C. Platt, J. Shawe-Taylor, A. J. Smola and R. C. Williamson, “Estimating support of a high-dimensional distribution,” Neural Computation, Vol. 13, pp. 1443–1471, 2001. Prasad, M., M. Schley, L. P. Russo and B. Wayne Bequette, “Product property and production rate control of styrene polymerization,” Journal of Process Control, Vol. 12, No. 3, pp. 353–372, 2002. Agarwal, M., A. M. Jade, V. K. Jayaraman and B. D. Kulkarni, “Support vec-

Nadeem Khalfe is a lead engineer with seven years experience working in ethylene oxide ethylene glycol plants. Mr. Khalfe has a BTech qualification in chemical engineering, along with experience in troubleshooting and debottlenecking petrochemical plants.

Jamal Al Ghamdi is a shift coordinator with 18 years of experience in petrochemical plants. Mr. Al Ghamdi’s experience includes ethylene dichloride, vinyl chloride monomer and ethylene oxide ethylene glycol plant operation.

Get the Information You Need to Compete in 2011… Order the Full Version of tthe HPI Market Data 2011

The HPI’s most comprehens comprehensive and trusted forecast of capital, maintenance and operating expenditures for the local and global HPI, the HPI Market Data 2011 includes over 75 pages of expanded coverage and detailed analysis for: • • • • • •

HPI Economics Equ Maintenance and Equipment Natural Gas/LNG Petrochemicals Refining En Health, Safety and Environment

f Includes a BONUS CD featuring: • More than 10 years of trends on global construction activity and spending from previous forecasts. Box • HPI Construction Boxscore Updates 2009 and 2010. Hydro • 20 articles from Hydrocarbon Processing picked by the ke trends and issues. editors to highlight key

ORDER YOUR COPY TODAY and make strategic decisions in 2011 and beyond! Online: www.GulfPub.com/2011hpi Call: +1 (713) 520-4426

Select 167 at www.HydrocarbonProcessing.com/RS

HYDROCARBON PROCESSING DECEMBER 2010

I 81

HPI MARKETPLACE 7!"!3( 3%,,3 2%.43

Water Scale Solvent

"/),%23 $)%3%, '%.%2!4/23 -/Ă&#x160; , 9Ă&#x160;- ,6

nĂ¤Ă¤Â&#x2021;Ă&#x2021;Ă¤{Â&#x2021;Ă&#x201C;Ă¤Ă¤Ă&#x201C; Ă&#x153;Ă&#x153;Ă&#x153;Â°Ă&#x153;>L>Ă&#x192;Â&#x2026;ÂŤÂ&#x153;Ă&#x153;iĂ&#x20AC;Â°VÂ&#x153;Â&#x201C; 8\Ă&#x160; n{Ă&#x2021;Â&#x2021;x{ÂŁÂ&#x2021;ÂŁĂ&#x201C;Ă&#x2021;Â&#x2122; Ă&#x160; n{Ă&#x2021;Â&#x2021;x{ÂŁÂ&#x2021;xĂ&#x2C6;Ă¤Ă¤

â&#x20AC;˘ No Premixing â&#x20AC;˘ Pleasant Odor â&#x20AC;˘ Non-Corrosive â&#x20AC;˘ Low-Foaming â&#x20AC;˘ Non- Hazardous â&#x20AC;˘ Biodegradable â&#x20AC;˘ Non-Flammable â&#x20AC;˘ Changes color when spent â&#x20AC;˘ NSF Registered (A3)

Removing water scale is difficult but not with Summit Sublime Descaler. It is safe for your employees, equipment and the environment. When water scale comes in contact with Sublime it disappears as a gas. There is no chipping, chiseling, brushing or high pressure blasting.

Summit Industrial Products 1.800.749.5823 Distributors Wanted

Select 205 at www.HydrocarbonProcessing.com/RS

Select 201 at www.HydrocarbonProcessing.com/RS

SURPLUS GAS PROCESSING/REFINING EQUIPMENT NGL/LPG PLANTS: 10 â&#x20AC;&#x201C; 600 MMCFD AMINE PLANTS: 60 â&#x20AC;&#x201C; 5,000 GPM SULFUR PLANTS: 10 â&#x20AC;&#x201C; 1,200 TPD FRACTIONATION: 1,000 â&#x20AC;&#x201C; 15,000 BPD HELIUM RECOVERY: 75 & 80 MMCFD NITROGEN REJECTION: 25 â&#x20AC;&#x201C; 80 MMCFD ALSO OTHER REFINING UNITS We offer engineered surplus equipment solutions.

Bexar Energy Holdings, Inc. Phone 210 342-7106 Fax 210 223-0018 www.bexarenergy.com Email: info@bexarenergy.com Select 202 at www.HydrocarbonProcessing.com/RS

$6450. 3&13*/54 5BLF BEWBOUBHF PG ZPVS FEJUPSJBM FYQPTVSF (JWF ZPVSTFMG B DPNQFUJUJWF BEWBOUBHF XJUI SFQSJOUT

$BMM PS TBMFT!GPTUFSQSJOUJOH DPN 'PS BEEJUJPOBM JOGPSNBUJPO QMFBTF DPOUBDU 'PTUFS 1SJOUJOH 4FSWJDF UIF PGGJDJBM SFQSJOU QSPWJEFS GPS )ZESPDBSCPO 1SPDFTTJOH Select 203 at www.HydrocarbonProcessing.com/RS

I DECEMBER 2010 HYDROCARBON PROCESSING

Select 204 at www.HydrocarbonProcessing.com/RS

HPI MARKETPLACE CA Co PE-O mp PE lian N t! HTRI Xchanger SuiteÂŽ â&#x20AC;&#x201C; an integrated, easy-to-use suite of tools that delivers accurate design calculations for â&#x20AC;˘ shell-and-tube heat exchangers â&#x20AC;˘ jacketed-pipe heat exchangers â&#x20AC;˘ hairpin heat exchangers â&#x20AC;˘ plate-and-frame heat exchangers â&#x20AC;˘ spiral plate heat exchangers

â&#x20AC;˘ fired heaters â&#x20AC;˘ air coolers â&#x20AC;˘ economizers â&#x20AC;˘ tube layouts â&#x20AC;˘ vibration analysis

Interfaces with many process simulator and physical property packages either directly or via CAPE-OPEN. Select 206 at www.HydrocarbonProcessing.com/RS

Heat Transfer Research, Inc. 150 Venture Drive College Station, Texas 77845, USA

HTRI@HTRI.net www.HTRI.net

Select 210 at www.HydrocarbonProcessing.com/RS

Select 207 at www.HydrocarbonProcessing.com/RS

NOISE

CONTROL ENGINEERING

HFP Acoustical Consultants Houston TX

Calgary AB

(888) 789-9400

(888) 259-3600

(713) 789-9400

(403) 259-6600

E-mail: info@hfpacoustical.com Internet: www.hfpacoustical.com Select 208 at www.HydrocarbonProcessing.com/RS

0IPE 3TRESS 0ROCESS 3IMULATION 0ELLETIZING $IE $ESIGN (EAT 4RANSFER !NALYSIS &INITE %LEMENT !NALYSIS #OMPUTATIONAL &LUID $YNAMICS 6ESSEL %XCHANGER -ACHINE $ESIGN 2OTOR $YNAMICS 3TRUCTURAL $YNAMICS 3PECIALISTS IN DESIGN FAILURE ANALYSIS AND TROUBLESHOOTING OF STATIC AND ROTATING EQUIPMENT WWW KNIGHTHAWK COM

(OUSTON 4EXAS 4EL s s &AX s s

Select 209 at www.HydrocarbonProcessing.com/RS

Select 211 at www.HydrocarbonProcessing.com/RS HYDROCARBON PROCESSING DECEMBER 2010

I 83

Bill Wageneck, Publisher

SALES OFFICES—EUROPE

SALES OFFICES—OTHER AREAS

FRANCE, GREECE, NORTH AFRICA, MIDDLE EAST, SPAIN, PORTUGAL, SOUTHERN BELGIUM, LUXEMBOURG, SWITZERLAND, GERMANY, AUSTRIA, TURKEY

AUSTRALIA—Perth Brian Arnold

2 Greenway Plaza, Suite 1020 Houston, Texas, 77046 USA P.O. Box 2608 Houston, Texas 77252-2608 USA Phone: +1 (713) 529-4301, Fax: +1 (713) 520-4433 E-mail: Bill.Wageneck@GulfPub.com www.HydrocarbonProcessing.com

Catherine Watkins 30 rue Paul Vaillant Couturier 78114 Magny-les-Hameaux, France Tél.: +33 (0)1 30 47 92 51, Fax: +33 (0)1 30 47 92 40 E-mail: Watkins@GulfPub.com ITALY, EASTERN EUROPE Mediapoint and Communications SRL Corte Lambruschini - Corso Buenos Aires, 8 5° Piano - Interno 7 16129 Genova - Italy Phone: +39 (010) 570-4948, Fax: +39 (010) 553-0088 E-mail: Fabio.Potesta@GulfPub.com

IL, LA, MO, OK, TX Josh Mayer 5930 Royal Lane, Suite 201, Dallas, TX 75230 Phone: +1 (972) 816-6745, Fax: +1 (972) 767-4442 E-mail: Josh.Mayer@GulfPub.com

AK, AL, AR, AZ, CA, CO, FL, GA, HI, IA, ID, IN, KS, KY, MI, MN, MS, MT, ND, NE, NM, NV, OR, SD, TN, TX, UT, WA, WI, WY, WESTERN CANADA Laura Kane

CT, DC, DE, MA, MD, ME, NC, NH, NJ, NY, OH, PA, RI, SC, VA, VT, WV, EASTERN CANADA Merrie Lynch

Lilia Fedotova

Alfred Bilyk Brazmedia Rua General Jardim, 633 Cj 61 01223 011 São Paulo SP, Brazil Phone: +55 (11) 3237-3269 Fax: +55 (11) 3237-3269 E-mail: Brazil@GulfPub.com

Pacific Business Inc. Phone: +81 (3) 3661-6138, Fax: +81 (3) 3661-6139 E-mail: Japan@GulfPub.com

Anik International & Co. Ltd. 10/2 Build. 1,B. Kharitonyevskii Lane 103062 Moscow, Russia Phone: +7 (495) 628-10-333 E-mail: Lilia.Fedotova@GulfPub.com

INDONESIA, MALAYSIA, SINGAPORE, THAILAND Peggy Thay

UNITED KINGDOM/SCANDINAVIA, NORTHERN BELGIUM, THE NETHERLANDS

20 Park Plaza, Suite 517, Boston, MA 02116 Phone: +1 (617) 357-8190, Fax: +1 (617) 357-8194 Mobile: +1 (617) 594-4943 E-mail: Merrie.Lynch@GulfPub.com

Phone: +86 13802701367, (China) Phone: +852 69185500, (Hong Kong) E-mail: Iris.Yuen@GulfPub.com

JAPAN—Tokyo Yoshinori Ikeda

RUSSIA/FSU

2 Greenway Plaza, Suite 1020, Houston, Texas, 77046 Phone: +1 (713) 520-4449, Fax: +1 (713) 520-4459 E-mail: Laura.Kane@GulfPub.com

CHINA—Hong Kong Iris Yuen

BRAZIL—São Paulo

Fabio Potestá

SALES OFFICES—NORTH AMERICA

Phone: +61 (8) 9332-9839, Fax: +61 (8) 9313-6442 E-mail: Australia@GulfPub.com

Michael Brown 1 Ladythorn Crescent Bramhall Stockport Cheshire SK7 2HB, UK Phone: +44 161 440 0854, Mobile: +44 79866 34646 E-mail: Michael.Brown@GulfPub.com

Publicitas Major Media (S) Pte Ltd Phone: +65 6836-2272, Fax: +65 6297-7302 E-mail: Singapore@GulfPub.com PAKISTAN—Karachi S. E. Ahmed Intermedia Communications Karachi-74700, Pakistan Phone: +92 (21) 663-4795, Fax: +92 (21) 663-4795

DATA PRODUCTS AND CLASSIFIED SALES REPRINTS

Lee Nichols Phone: +1 (713) 525-4626, Fax: +1 (713) 525-4631 E-mail: Lee.Nichols@GulfPub.com

Phone: +1 (866) 879-9144 ext. 194 E-mail: rhondab@FosterPrinting.com

The HEAT is On! The Fundamentals of Piping Design By Peter Smith 262 pages • Hardcover • Pub date: April 2007 ISBN: 978-1-933762-043 • Price: $175 Written for the piping engineer and designer in the field, this first part of the two-part series helps to fill a void in piping literature, since the Rip Weaver books of the ‘90s were taken out of print.

Go to www.GulfPub.com and get the latest upgrade to our best-selling software! UPGRADED FOR 2009!

WinHeat 4 Put control back into the hands of the process engineer www.GulfPub.com/Winheat

An intermediate-level handbook covering guidelines and procedures on process plants and interconnecting piping systems.

The Planning Guide to Piping Design By Richard Beale, Paul Bowers and Peter Smith 300 pages • Hardcover • Pub date: September 2010 ISBN: 978-1-933762-37-1 • Price: $175 The Planning Guide to Piping Design covers the entire process of planning a plant model project from conceptual to mechanical completion, and explains where the piping lead falls in the process along with his roles and responsibilities.

To place an order, visit www.gulfpub.com or call +1 (713) 520-4426. 84

I DECEMBER 2010 HYDROCARBON PROCESSING

RECOMMENDED GUIDE:

Heat Exchangers Selection, Rating and Thermal Design www.GulfPub.com/HeatExchangers

HEAT E EXCHANG XCHANG XC X GERS SELECTION GERS ON AND THERMA THERM RM RMA MAL M MA A AL DES DESIGN ESIGN

Advanced Piping Design By Rutger Botermans and Peter Smith 250 pages • Hardcover • Pub date: May 2008 ISBN: 978-1-933762-18-0 • Price: $175

GULF P U B L I S H I N G C O M PA N Y

+1-713-520-4426 l +1-800-231-6275 Email: svb@GulfPub.com I www.GulfPub.com

FREE Product and Service Information—DECEMBER 2010 HOW TO USE THE INDEX: The FIRST NUMBER after the company name is the page on which an advertisement appears. The SECOND NUMBER, appearing in parentheses, after the company name, is the READER SERVICE NUMBER. There are several ways readers can obtain information: 1. The quickest way to request information from an advertiser or about an editorial item is to go to www. HydrocarbonProcessing.com/RS. If you follow the instructions on the screen your request will be forwarded for immediate action. 2. Go online to the advertiser's Website listed below. 3. Circle the Reader Service Number below and fax this page to +1 (416) 620-9790. Include your name, company, complete address, phone number, fax number and e-mail address, and check the box on the right for your division of industry and job title. Name ________________________________________________________

Company ________________________________________________________

Address ______________________________________________________

City/State/Zip ____________________________________________________

Country ______________________________________________________

Phone No. _______________________________________________________

FAX No. ______________________________________________________

e-mail ___________________________________________________________

This Advertisers’ Index and procedure for securing additional information is provided as a service to Hydrocarbon Processing advertisers and a convenience to our readers. Gulf Publishing Co. is not responsible for omissions or errors.

This information must be provided to process your request: PRIMARY DIVISION OF INDUSTRY (check one only): A B C F G H J P

䊐-Refining Company 䊐-Petrochemical Co. 䊐-Gas Processing Co. 䊐-Equipment Manufacturer 䊐-Supply Company 䊐-Service Company 䊐-Chemical Co. 䊐-Engrg./Construction Co.

JOB FUNCTION (check one only): B E F G I J

䊐-Company Official, Manager 䊐-Engineer or Consultant 䊐-Supt. or Asst. 䊐-Foreman or Asst. 䊐-Chemist 䊐-Purchasing Agt.

ADVERTISERS in this issue of HYDROCARBON PROCESSING Company Website

Page

RS#

ACS Industries Inc. . . . . . . . . . . . 32 (158)

Company Website

Page

RS#

Company Website

GE Oil & Gas . . . . . . . . . . . . . . . . . 8

(64)

Merichem Company . . . . . . . . . . 38 (160)

www.info.hotims.com/29426-158

Alfa Laval Packinox . . . . . . . . . . . 26 (155) Altair Strickland. . . . . . . . . . . . . . 12

(51)

www.info.hotims.com/29426-152

www.info.hotims.com/29426-153

www.info.hotims.com/29426-53

Bryan Research & Engineering . . . 30 (113) www.info.hotims.com/29426-113

Siemens AG. . . . . . . . . . . . . . . . . 25 Circulation . . . . . . . . . . . . . . . . 67 (165)

Spraying Systems Co . . . . . . . . . . 87

Market Data Book . . . . . . . . . . . 81 (167)

www.info.hotims.com/29426-156

Süd-Chemie . . . . . . . . . . . . . . . . . 6

(90)

www.info.hotims.com/29426-90

Software . . . . . . . . . . . . . . . . . . 84 HP Marketplace . . . . . . . . . . . 82–83 Kobe Steel Ltd . . . . . . . . . . . . . . . 14

Construction Boxscore . . . . . . . . . 29 (157) www.info.hotims.com/29426-157

(80)

Uhde GmbH . . . . . . . . . . . . . . . . 16

(70)

www.info.hotims.com/29426-70

www.info.hotims.com/29426-80

Linde AG . . . . . . . . . . . . . . . . . . . 22 (57)

www.info.hotims.com/29426-166

(89)

Unifrax . . . . . . . . . . . . . . . . . . . . 50

(68)

www.info.hotims.com/29426-68

www.info.hotims.com/29426-89

www.info.hotims.com/29426-57

Veolia Environment . . . . . . . . . . . 21

Finder Pompe SpA . . . . . . . . . . . . 61 (164) www.info.hotims.com/29426-164

www.info.hotims.com/29426-93

(62)

www.info.hotims.com/29426-62

Trachte USA . . . . . . . . . . . . . . . . 68 (166)

Carver Pump Company . . . . . . . . 28 (156)

Flexitallic LP . . . . . . . . . . . . . . . . . 5

SNC-Lavalin Engineers & Construction Inc. . . . . . . . . . . 40 (161) www.info.hotims.com/29426-161

Events— Save the Date. . . . . . . 76

www.info.hotims.com/29426-154

Costacurta SpA Vico . . . . . . . . . . 21

(63)

www.info.hotims.com/29426-63

www.info.hotims.com/29426-167

C & I Engineering Inc . . . . . . . . . . 24 (154)

(96)

www.info.hotims.com/29426-96

www.info.hotims.com/29426-165

(53)

www.info.hotims.com/29426-159

Selas Fluid Processing Corp . . . . . 19

Books . . . . . . . . . . . . . . . . . . . . 84 (54)

www.info.hotims.com/29426-163

Prosim . . . . . . . . . . . . . . . . . . . . 36 (159)

Gulf Publishing Company

www.info.hotims.com/29426-54

Axens . . . . . . . . . . . . . . . . . . . . . 88

www.info.hotims.com/29426-72

(59)

www.info.hotims.com/29426-59

Aveva AB . . . . . . . . . . . . . . . . . . 42

(72)

Greene, Tweed & Co. . . . . . . . . . 20 (153)

Ametek Process Instruments . . . . 18 (152) ARC Collaborative MFG . . . . . . . . 62

Microtherm . . . . . . . . . . . . . . . . . 56 (163) GE Power & Water . . . . . . . . . . . 10

www.info.hotims.com/29426-51

RS#

www.info.hotims.com/29426-160

www.info.hotims.com/29426-64 www.info.hotims.com/29426-155

Page

(93)

Linde Process Plants . . . . . . . . . . 22

(81)

www.info.hotims.com/29426-81

Maxon Corporation . . . . . . . . . . . 41 (162) www.info.hotims.com/29426-162

Weir Minerals France . . . . . . . . . . . 4 (151) www.info.hotims.com/29426-151

Yokogawa . . . . . . . . . . . . . . . . . . . 2

(74)

www.info.hotims.com/29426-74

For information about subscribing to HYDROCARBON PROCESSING, please visit www.HydrocarbonProcessing.com HYDROCARBON PROCESSING DECEMBER 2010

I 85

HPIN CONTROL PIERRE R. LATOUR, GUEST COLUMNIST clifftent@hotmail.com

Process control practice renewal—consequences In my April 2010 editorial, I initiated and justified a call for renewal of process control and IT practice in the hydrocarbon processing industry (HPI) with a review of the basics. In August 2010, I reviewed the purpose of process control and IT. In the October 2010 editorial on performance, I described a standardized method for determining financial performance of instrumentation, control systems, IT and CIM.1 Now, I turn to the consequences for exceeding limits and violating specs of controlled variables (CVs) and key performance indicators (KPIs). Unforeseen occurrences. In 1995, a major refiner indicated

experiencing unforeseen occurrences in four refineries costing $60 million annually. The problem is “unforeseen.” Limits and specs. Engineers and operating managers specify the location values for equipment limits, quality specifications, alarms, tolerances and safety factors. Obviously, limit and spec settings are critical to long-term HPI performance. One can operate far from them, uneconomically, or too close to them, uneconomically. These limit values are input to LP planning and scheduling systems as problem-boundary limits. The optimum solution invariably lies at an intersection of a combination of constraints, but the location of the intersection is set by these input values. They are input to online process NLP optimizers (and MVC as “equal concern errors”) as problem boundary limits, as well. The optimizer determines the best constraint combination (rarely an interior hilltop), but the location of the intersection is set by these input values. (This is why process optimizers don’t actually determine the best process control setpoint values; those are inputs set by people.) Inside limits. Engineers use process models to determine the physical consequences of moving setpoints toward limits. They associate an economic factor to determine the financial merit from the move. This may be the slope of an LP profit function. Beyond limits. Engineers often say exceeding limits is forbidden because disaster follows, and the consequences are too hard to contemplate—let alone model. Never do it. That is too easy. Sometimes, this is because standard tools like process models, LP and online NLP optimizers do not handle these external models well. Yet, operators and control systems must do something when an alarm sounds. Here is the opportunity. Slopes, drop-offs and cliffs. Determine what happens as a limit is violated or a spec is exceeded.1 Model it physically and economically. Some quality violations have a smooth price penalty depending on degree; some have a discrete penalty independent of degree. Some increase equipment wear and tear gradually; others destroy equipment suddenly. Some emissions are harmless; others are deadly. Knowledge of these consequences is essential to optimally avoiding and mitigating them. The penalty slope and magnitude of any exterior cliff should be modeled with as much fidelity as the interior process models. 86

I DECEMBER 2010 HYDROCARBON PROCESSING

Consequences of consequences. If the violation cliff

is severe, the HPI normally operates safely away. If the violation penalty is small, then the HPI normally operates closer to it. I have seen a gasoil pour-point spec with a mild penalty for exceeding and a strong credit for approaching; therefore, exceeding it was common. I have also seen study of limit violations leading to contingency plans that manage and mitigate them when they inevitably occur—shrinking the cliff. The financial consequences for inadequate attention to modeling limit violation consequences are consequential. The problem is that the approach tolerances are vague, ad hoc and empirical. Risky tradeoffs everywhere. Working on computer control of many refineries and petrochemical plants around the world since 1966, I can confirm that every HPI operator thinks this way but must rely on experience and judgment to set setpoints unless they adopt a risky tradeoff optimizer using the best information available to quantify the thinking process and stay on top of their plant and business daily. Process control renewal. The objective is not simply to renew the practice of process control. The goal is to run better. The heart of the matter is setting setpoints right by optimizing risky tradeoffs rigorously. Historic practice relies too much on fallible human experience, judgment and empiricism without a standard procedure. The remedy is for the HPI to adopt rigorous methods for setting setpoints and process control, and IT adopting this decision process as their guide for control-performance determination and information requirements value. Conclusions. Associating a profit tradeoff with each CV, one

can connect alarms and process control; HPI plant operation and its surroundings; operations to customers; process control and IT to profit centers. Formerly unforeseen occurrences are foreseen. The best way to improve HPI operation is to do it the way everyone has been doing it all along, but with more rigor, science, mathematics, economics and knowhow. This was offered to the HPI in 1996, 1 worth > $1/bbl depending on how well the refiner understood the process, business and CV clifftents. HP 1

LITERATURE CITED Latour, P. R., “Process control: CLIFFTENT shows it’s more profitable than expected,” Hydrocarbon Processing, December 1996, pp. 75–80. Republished, “Advanced Process Control and Information Systems for the Process Industries,” Gulf Publishing Co., 1999, pp. 31–37.

The author, president of CLIFFTENT Inc., is an independent consulting chemical engineer specializing in identifying, capturing and sustaining measurable financial value from HPI dynamic process control, IT and CIM solutions (CLIFFTENT) using performance-based shared risk–shared reward (SR2) technology licensing.

Spray Nozzles

Spray Analysis

Spray Control

Spray Fabrication

Why Leading Refineries and Engineering Firms Rely on Us for Injectors and Quills

Retractable Injector, Slurry Backflush Quill, Water Wash Quill (bottom to top)

Water-Jacketed Injector for High-Temperature Applications Computational Fluid Dynamics (CFD)

Manufacturing quality and flexibility. Need a simple quill or multi-nozzle injector? Insertion length of a few inches or several feet? 25# or 2500# class flange? High-pressure, high-temperature and/or corrosion-resistant construction? Special design features like a water-jacket, air purge or easy retraction for maintenance? Tell us what you need and we’ll design and manufacture to your specifications and meet B31.1, B33.3 and CRN (Canadian Registration Number) requirements.

CFD shows the change in drop size based on nozzle placement in the duct.

D32 (μm) 220

165

Z = 0.6 m

Nozzle spraying in-line with duct

Proven track record. We’ve manufactured hundreds of injectors for water wash, slurry backflush, feed and additive injection, SNCR and SCR NOx control, desuperheating and more. Customers include Jacobs Engineering, Foster Wheeler Corp, Shaw Group, Conoco Phillips Co, Shell, Valero and dozens more. Learn More at spray.com/injectors Visit our web site for helpful literature on key considerations in injector and quill design and guidelines for optimizing performance.

110

55 0

Design validation with process modeling. Let us simulate the injection environment to identify potential problems. We can model gas flow, droplet trajectory and velocity, atomization, heat transfer, thermal stresses, vibration and more to ensure optimal performance.

Nozzle spraying at 45° in duct

1-800-95-SPRAY | spray.com |

Select 62 at www.HydrocarbonProcessing.com/RS

Specify and order standard nozzles spray.com/ispray

Your objectives in focus Make the most of today’s and tomorrow’s challenges with leading-edge solutions from Axens - Clean and alternative fuel technologies - Petrochemicals - Energy efficiency - High performance catalysts & adsorbents - Revamps

Single source technology and service provider ISO 9001 – ISO 14001 – OHSAS 18001 www.axens.net Select 53 at www.HydrocarbonProcessing.com/RS