HP_2009_08 by androsov.info

AUGUST 2009

HPIMPACT

SPECIALREPORT

TECHNOLOGY

Global energy demand down in 2009

FLUID FLOW AND ROTATING EQUIPMENT

Biorefineries: Fact or fiction?

Economic recovery in sight for chemical industry

Lubrication, piping compressors and heat exchangers

What metals to use in critical-service exchangers

www.HydrocarbonProcessing.com

sky's the limit

OneWireless solutions give you the freedom to extend beyond your limits. From helping you manage your rotating equipment to making your employees mobile and more efficient, Honeywell has helped our customers solve process and business challenges with innovative wireless-enabled solutions. Our OneWireless

universal mesh network supports multiple industrial protocols and applications simultaneously, giving you flexibility without sacrificing reliability or bandwidth. Why stay chained to multiple networks, when there is one that will let you soar. OneWireless.

Select 52 51 at www.HydrocarbonProcessing.com/RS

AUGUST 2009 • VOL. 88 NO. 8 www.HydrocarbonProcessing.com

SPECIAL REPORT: FLUID FLOW AND ROTATING EQUIPMENT

Understanding and maintaining an effective lubrication system Follow these guidelines to improve gearing equipment life J. DeBaecke

Improve gas compression systems with all-welded shell-and-plate heat exchangers They offer benefits compared to shell-and-tube and welded-plate designs S. Gavelin

Consider ceramic bearings for screw compressors

A retrofit of this system at a refining company provided a payout in less than a year

This retrofit resulted in a six-fold increase in service life H. P. Bloch

Improved stepless capacity regulation for reciprocating compressors J. Jin, W. Hong and L. Tang

Cover The cover photo is complements of Onis Inc., with offices worldwide including Houston, Texas. The application is in a Middle East Refinery that was built in 2006 and designed to handle mixed feedstocks. This application is used in the production of jet fuel. The valves with Onis blinds serve as double block-and-bleed with a blind that improves the flow control of the feed stocks. The positive isolation for the by-pass lines ensures exceptional product quality and improves safety.

Predicting liquid hold-up in horizontal stratified two-phase flow A new model has an average error of only 1% H. Firoozfar, N. Kasiri and M. H. Khanof

HPIMPACT 19 Global energy demand down in 2009 21 Economic recovery in sight for chemical industry

HEAT TRANSFER

What metals should be used in critical service heat exchangers? Here are important guidelines when considering titanium for exchanger construction R. Pramanik

SAFETY/LOSS PREVENTION

Are you bypassing process safety programs? Implementing these changes may increase profits and lower incidents M. Sawyer

PROCESS TECHNOLOGIES

Biorefineries: Fact or fiction? Technology advances facilitate building integrated chemical complexes based on renewable feedstocks that can cost-effectively process ethylene derivatives M. Bruscino

PIPING/FLUID FLOW

A support vector classification method for regime identification of slurry transport in pipelines Statistical analysis showed the proposed solution has an average misclassification error of only 1.5% S. K. Lahiri and K. C. Ghanta

DEPARTMENTS 7 HPIN BRIEF • 19 HPIMPACT • 23 HPIN CONSTRUCTION • 25 LETTERS TO THE EDITOR • 86 HPI MARKETPLACE • 89 ADVERTISER INDEX

COLUMNS 9 HPIN RELIABILITY Electric motors and mechanical efficiency 11 HPIN EUROPE Costly flawed paradigm means ‘shaking up’ molecules 13 HPIN CONTROL APC designs for minimum maintenance—Part 3 15 HPINTEGRATION STRATEGIES Intelligent toxic gas detectors: a smart move for the HPI 17 HPIN ASSOCIATIONS ILTA gathers to take annual look in mirror 90 HPIN WATER MANAGEMENT Update on water treatment for ethanol plants

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 Editor Les A. Kane Senior Process Editor Stephany Romanow Process Editor Tricia Crossey Reliability/Equipment Editor Heinz P. Bloch News Editor Billy Thinnes 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) MAGAZINE PRODUCTION Director—Editorial Production Sheryl Stone Manager—Editorial Production Chris Valdez Artist/Illustrator David Weeks Manager—Advertising Production Cheryl Willis ADVERTISING SALES See Sales Offices page 88. CIRCULATION +1 (713) 520-4440 Director—Circulation Suzanne McGehee E-mail: circulation@gulfpub.com SUBSCRIPTIONS

Subscription price (includes both print and digital versions): United States and Canada, one year $140, two years $230, three years $315. Outside USA and Canada, one year $195, two years $340, three years $460, digital format one year $140. Airmail rate outside North America $175 additional a year. Single copies $25, prepaid. 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 & Tech nology 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 us 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 Cheryl Willis at +1 (713) 525-4633 or e-mail EditorialReprints@gulfpub.com

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!

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 © 2009 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

GULF PUBLISHING COMPANY John Royall, President/CEO Ron Higgins, Vice President Pamela Harvey, Business Finance Manager Part of Euromoney Institutional Investor PLC.

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

Select 151 at www.HydrocarbonProcessing.com/RS

Other energy group titles include: World Oil® Petroleum Economist Publication Agreement Number 40034765

Printed in U.S.A

INT ENS E HE AT. A GGR E S S IVE CHE MICALS . E XT R E ME COLD.

WE’RE PUSHING THE LIMITS OF ENDURANCE. NOT YOUR PATIENCE.

MATERIAL TECHNOLOGY ENGINEERED FOR REFINERIES THERMICULITE® 835 Spiral Wound Filler UÊÊ > ` iÃÊÌ iÊÌ Õ} iÃÌÊ>«« V>Ì Ã UÊÊ"ÕÌ«iÀv À ÃÊ}À>« ÌiÊ> `ÊwLiÀ UÊÊ*À Û `iÃÊÌ Ì> ÊvÀii` ÊvÀ Ê Ý `>Ì UÊÊ"vviÀÃÊÌÀÕiÊ ÕÌ>}i Ì ÕÌ>}iÊ>ÃÃÕÀ> Vi UÊÊ,i`ÕViÃÊ Ûi Ì ÀÞÊÀiµÕ Ài i ÌÃ

ALSO AVAILABLE IN: U 815 Tanged Sheet U 815 Cut Gaskets UÊ845 Flexpro™ (kammproﬁle) Facing Select 93 at www.HydrocarbonProcessing.com/RS IÊÓäänÊ À ÃÌÊEÊ-Õ Û> Ê ÀÌ Ê iÀ V> Ê*À `ÕVÌÊ6> ÕiÊ i>`iÀÃ «Ê vÊÌ iÊ9i>ÀÊ Ü>À`Ê,iV « i Ì°

UR GLOBAL YO

K ET

P R O VI

log onto: www.ﬂexitallic.com or call: US +1 281.604.2400 UK +44 (0) 1274 851273

Where do You Want to be on the Performance Curve?

P = People M= Methodologies T = Technologies

Your Company + KBC Produces NextGen Performancen We collaborate with our clients to create unique solutions to their speciďŹ c challenges. Some of these challenges may include: Strategic Challenges

Operating Challenges

U Effective Business Strategy/Decisions U Increased Return on Investments U Enhanced Returns on Acquisitions/Divestitures U Reduced Risk (Strategic, Capital, Other)

U Improved Organisational Effectiveness U Reduced Maintenance Costs U Improved Energy EfďŹ ciency U Behaviour-based Reliability/Performance U Improved Safety Performance U Operational Risk Management

Market Challenges U Enhanced Yields U Effective Responses to Crude/Feedstock and Product Markets U Improved Financial Performance U Market Risk Management

Environmental Challenges U Reduced Emissions U Enhanced Compliance

For 30 years, KBC consultants have provided independent advice and expertise to enable leading companies in the global energy business and other processing industries manage risk and achieve dramatic performance improvements.

For more information on how KBC can help you achieve NextGen Performance, contact us at: AMERICAS +1 281 293 8200 EMEA +44 (0)1932 242424 ASIA +65 6735 5488 answers@kbcat.com U www.kbcat.com Select 82 at www.HydrocarbonProcessing.com/RS

HPIN BRIEF BILLY THINNES, NEWS EDITOR

BT@HydrocarbonProcessing.com

US petroleum deliveries dropped nearly 6% for the first half of 2009, to the lowest level for the January-to-June period in more than a decade. Petroleum deliveries fell to 18.75 MMbpd, down nearly 10% from the peak of 20.75 MMbpd reached in first-half 2005, according to API’s Monthly Statistical Report. First-half 2009 jet fuel deliveries fell nearly 13% from a year ago as a result of reduced air travel, while distillate fuel oil deliveries dropped 8.6% on sluggish demand for freight transportation. Residual fuel oil deliveries fell by 9.1%, reflecting both fuel substitutions and the economic slowdown. First-half 2009 gasoline deliveries slipped less than 1% from already-depressed levels of a year ago, reaching their lowest level for the period since 2003. For the first half of 2009, US crude oil production reached its highest level in four years, at an average of 5.29 MMbpd. June’s refinery inputs rose to nearly 15 MMbpd, the highest level since November.

The European biodiesel industry maintains resilience in the face of unfair international competition and poor market conditions, according to a release from the European Biodiesel Board (EBB). EU biodiesel production in 2008 increased 35.7% to 7.7 million tons, which the EBB found disappointing when compared to increases in 2005 (65%) and 2006 (54%). The EBB lays some of the blame for reduced production increases on what it labels unfair competitive practices from overseas that are affecting the profitability of EU biodiesel producers. “For more than two years, EU biodiesel producers had to compete with heavily subsidized biodiesel from the US known as B99. US B99 has been sold in the EU with a considerable discount, even at lower price than the raw material soybean oil,” the EBB release says. While rankled, the EBB sounded a positive note when revealing that EU legislators have instituted measures to prevent the alleged unfair trade practices that the EBB claims are in violation of WTO principles. The EBB was disappointed by the biodiesel production numbers because it is feeling the pressure from EU mandates requiring 33 million metric tons of biofuel in Europe by 2020, in order to comply with greenhouse gas emission reduction targets. While some of this will have to be imported, the EBB hopes for a greater proportion to be homegrown. While the EBB is proud that, as of July 2009, 276 biodiesel plants exist in the EU, it is disconcerted by the fact that over half of them remain idle. Still, the EBB believes “the EU and its member states have been able to create a genuine framework supporting the deployment of biodiesel and biofuels, notably in the form of mandatory blending targets.”

Enterprise Products Partners plans to merge with TEPPCO to form a publicly traded energy partnership with an enterprise value of more than $26 billion. The combined partnership will own almost 48,000 miles of pipelines comprising 22,000 miles of NGL, refined product and petrochemical pipelines, over 20,000 miles of natural gas pipelines and more than 5,000 miles of crude oil pipelines. Enterprise Products Partners/ TEPPCO’s logistical assets will include approximately 200 million barrels of NGL, refined product and crude oil storage capacity; 27 billion cubic feet of natural gas storage capacity; and a NGL import/export terminal located on the Houston Ship Channel in Texas.

A publicly accessible hydrogen filling station recently opened at Stuttgart Airport in Germany. OMV collaborated with Linde AG and Daimler AG on the project. The hydrogen filling station will serve fuel-cell vehicles of the latest generation. It is being subsidized by the federal state of Baden-Württemberg in Germany, to the tune of €800,000 from a fund known as “Baden-Württemberg’s Campaign for the Future.” This fact explains the logic behind the location of the hydrogen filling station. It is located next to a major transport hub (Stuttgart Airport) and is in close proximity to Daimler AG’s research and development centers. The new hydrogen filling station incorporates ion-compressor technology developed by Linde AG. With this new compression process, cars and electric buses powered by fuel cells can be refueled like conventional vehicles with hydrogen at a pressure of either 350 bar or 700 bar. HP

■ A tsunami of change awaits US refiners A tsunami of change awaits US refiners, according to a recent report from Deloitte (www.deloitte.com). Citing a confluence of factors that will have an adverse effect on the refining industry (increased fuel economy standards for cars, renewable fuel mandates and capand-trade legislation), Deloitte projects that these factors “have turned the future demand curve for gasoline—the most significant product from most US refineries—upside down.” Deloitte leans on the EIA to make its point, citing projections that show the US refining industry’s utilization rate sliding to 78.5% in 2010 as a result of reduced demand and increased capacity. The last time utilization rates were at such a level was 1985. Deloitte is not all doom and gloom, though. The report offers guidance on how to survive the tough new reality. Several key points stand out. Quickly internalizing the new legislative and regulatory rules is important for refiners “to ensure compliance and facilitate strategic decision making.” Reassessing competitive position is encouraged, as the position of many traditional players is likely to shift. The report advises, “A portfolio strategy backed up by a market-by-market analysis of the potential changes in competitor positions is key.” Other recommendations include capturing more of the emerging value chain and noting that disruption creates opportunity. Regarding value chain, refiners could leverage their business knowledge to rapidly embrace new regulations. In the brave new energy world, Deloitte sees the possibilities for “unintended consequences to result in opportunities for those with the knowledge and capital to act. New carbon markets and the evolving market for renewable energies will provide large opportunities for the bold players.” HP HYDROCARBON PROCESSING AUGUST 2009

The Inpro/Seal Company has been in the business of bearing protection for rotating equipment for 32 years and counting. We have been supplying bearing protection for the IEEE-841 motors since they were first introduced to industry. It is only logical that we would expand into the field of motor shaft current mitigation to protect motor bearings. The CDR is:

ROBUST

Machined entirely out of solid corrosion resistant and highly conductive bronze, the CDR/MGS is capable of carrying 12+ continuous amps. They are made exclusively by the Inpro/Seal Company in Rock Island, IL, to ensure consistent quality and same-day shipments when required.

RELIABLE

The CDR and MGS (Motor Grounding Seal) products were developed in our own Research and Experimentation Laboratory and then extensively tested and evaluated by professional motor manufacturing personnel. Our standard guarantee of unconditional customer satisfaction of product performance applies. We stand behind our products.

REALISTIC

When you order a CDR or MGS from Inpro/Seal, you are assured of the complete responsibility for technology and performance from a single source. We want to earn the right to be your first choice for complete bearing protection.

For more information visit www.inpro-seal.com/CDR or contact 800-447-0524 for your Inpro/Seal Representative. Select 78 at www.HydrocarbonProcessing.com/RS

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

Electric motors and mechanical efficiency Reliability-focused readers know quite a bit about oil mist and many use plantwide oil mist systems for their process pumps and electric motor drivers. In fact, some major grassroots olefins plants used oil mist on motors with rolling-element bearings as small as 1 hp (0.75 kW) and as large as 1,250 hp (925 kW) as early as 1975. What’s less known is that oil mist, together with the right viscosity synthetic lubricant, will save considerable energy. In 1980, an important presentation described that a readily available synthetic lubricant, having a viscosity of 32 cSt at a temperature of 40°C (98°F), offered long-term contact surface protection for process pumps and their electric motor drivers.1 Although “only” 32 cSt, the protective effect of this synthetic lubricant was found to be equivalent to that of a base line mineral oil with the higher viscosity of 68 cSt. The same good wear protection could not be achieved with a reduced viscosity mineral oil. The tests also showed the lower viscosity synthetic lubricant providing energy savings of approximately 4% of the normal electric motor power draw. Quantifying the energy savings potential. Using both

oil mist and synthetic lubes makes economic sense. According to the test described in reference 1, the frictional losses in rollingelement bearings can be reduced as much as 37%. On the 65-mm bearings typically used in 15-hp (~11.3-kW) process pumps and electric motors, 0.11 kW could be saved (Fig. 1). While the small absolute value of 0.11 kW per bearing tends to make the savings appear insignificant, petrochemical process pump rotors are typically supported by a double-row radial ball bearing and two angular contact ball thrust bearings. These then represent a total of 4.8 test-equivalent bearings (4 x 0.7) + (2 x 1) = 2.8 + 2 = 4.8. 0.300

Power loss per bearing, KW

MIN 68 SYN 32 0.250

Therefore, the total savings available from an actual motor-driven pump set are 4.8 times the single test bearing energy savings of 0.11 kW; that equals 0.53 kW or 4.7% of 11.3 kW. Assuming that the average pump operates 90% of the time, and rounding off the numbers, this difference amounts to energy savings of 4,180 kWh per year. At $0.10 per kWh, yearly savings of $418 should be expected. By using synthetics on conventionally lubricated equipment, oil replacement schedules are typically extended four-fold; the extended drain intervals more than compensate for the higher cost of synthetic lubricants. With oil mist the bearings run cooler and last longer than those typically lubricated by conventional oil sumps. While open oil mist systems typically consume 12–22 liters (3.1–5.7 gallons) per pump set per year, closed oil mist systems consume no more than 10% of these yearly amounts. Again, at these extremely low make-up or consumption rates, and compared to the cost of mineral oils, the incremental cost of synthetic lubricants is relatively insignificant. Considering annual energy savings per 15-hp pump and driver set to be worth $418, we realize that these savings should be multiplied by the number of pumps actually operating in large refineries—850 to 1,200. Again using $0.10 per kWh, annual savings in the vicinity of $450,000 would not be unusual. A detailed calculation can be found in reference 2; it will prove the point: • Total pump hp installed at the plant = 15 hp x 1,000 = 15,000 hp • Total pump kW installed at the plant = 15,000 x 0.746 = 11,190 kW • Total consumption kWh per year, considering 90% of 8,760 h/yr = 8,760 h x 0.90 = 7,884 h/yr; then 7,884 x 11,190 kW = 88,220,000 kWh/yr • Total US dollar value of yearly energy consumed, assuming $0.10/kWh: = $8,822,000. Total energy savings for 1,000 average-sized pump sets would thus equal 0.047 x 8,822,000 = $414,600. That is an amount that should not be overlooked. HP 1

0.200

LITERATURE CITED Morrison, F.R., Zielinsky, J., and James, R., “Effects of synthetic fluids on ball bearing performance,” ASME Paper 80-Pet-3, February 1980. Bloch, Heinz P., Practical Lubrication for Industrial Facilities, The Fairmont Press, Second Edition, 2009.

0.150

0.100 Oil sump FIG. 1

Oil mist

Power loss plot for the ball bearing tests. The two different oils have different viscosities, but their protective properties are identical. Oil mist reduces power losses, as do carefully selected synthetic lubes.2

The author is HP’s Reliability/Equipment Editor. A practicing consulting engineer with nearly 50 years of applicable experience, he advises process plants worldwide on failure analysis, reliability improvement and maintenance cost-avoidance topics. He has authored or coauthored 17 textbooks on machinery reliability improvement and over 460 papers or articles dealing with related subjects. This excerpt is from his 2nd Edition, Practical Lubrication for Industrial Facilities, which was released in May 2009 by Fairmont Press.

HYDROCARBON PROCESSING AUGUST 2009

In troubled times fierce global competition for premium crudes means that refinery units must have the flexibility to handle heavy, viscous, dirty crudes that increasingly threaten to dominate markets. And flexibility must extend to products as well as crudes, for refinery product demand has become more and more subject to violent economic and political swings. Thus refiners must have the greatest flexibility in determining yields of naphtha, jet fuel, diesel and vacuum gas oil products.

Why Do Many Crude/Vacuum Units Perform Poorly?

Rather than a single point process model, the crude/vacuum unit design must provide continuous flexibility to operate reliably over long periods of time. Simply meeting the process guarantee 90 days after start-up is very different than having a unit still operating well after 5 years. Sadly few refiners actually achieve this—no matter all the slick presentations by engineers in business suits!

modeling. Refinery hands-on experience teaches that fouling, corrosion, asphaltene precipitation, crude variability, and crude thermal instability, and many other non-ideals are the reality. Theoretical outputs of process or equipment models are not. In this era of slick colorful PowerPoint® presentations by well-spoken engineers in Saville Row suits, it’s no wonder that units don’t work. Shouldn’t engineers wearing Nomex® coveralls who have worked with operators and taken field measurements be accorded greater credibility?

In many cases it’s because the original design was based more on virtual than actual reality. There is no question: computer simulations have a key role to play but it’s equally true that process design needs to be based on what works in the field and not on the ideals of the process simulator. Nor should the designer simply base the equipment selection on vendor-stated performance. The design engineer needs to have actual refinery process engineering experience, not just expertise in office-based

Today more than ever before this is important. Gone are the days when a refiner could rely on uninterrupted supplies of light, sweet, easy-to-process crudes.

PROCESS

CONSULTING SERVICES,INC.

3400 Bissonnet Suite 130 Houston, Texas 77005 USA

If you want to explore these issues in technical detail ask for Technical Papers 267 and 268. Ph: [1] (713) 665-7046 Fx: [1] (713) 665-7246 info@revamps.com www.revamps.com

Select 76 at www.HydrocarbonProcessing.com/RS

HPIN EUROPE TIM LLOYD WRIGHT, EUROPEAN EDITOR tim.wright@gulfpub.com

Costly flawed paradigm means ‘shaking up’ molecules Because there’s no such thing as throwing things “away,” it’s time to get out the white coats and do some serious chemistry. We need to use hydrocarbon molecules in a completely different way. Also, we must somehow now shake up two mighty oil industry commandments: “Thou shalt not change fuel specifications,” and “Thou shalt not tinker around with vehicle engines or fuel infrastructure.”

into trucks in a compressed form, or as dimethyl ether? Or, what about taking a good cut of the lighter hydrocarbons used in gasoline and of putting those into diesel engines? (You’d have to alter the holy flash point of the fuel. Yes, that’s right.) • Instead of putting diesel into trucks, how about putting it into ships? • Instead of putting heavy oils into ships, how about turning the heavy ends into coke? Shifting standards. These are undeniably solid pillars of • Instead of separating refinery residues onboard ships and modern hydrocarbon industry thought. The almost unquestionusing incinerators to gasify and vent them to the atmosphere, able logic that reinforces them is that doing either means having a why not take care of them at the refinery, where all produced CO2 LOT of meetings. I sympathize with this; meetings are awful. Furemissions can be separated and injected underground? thermore, refiners are thinking these thoughts as well. One chief • Instead of venting carbon into the air from coal-fired power refiner at a European downstream company recently asked a group stations, why not capture the carbon, not just from coal, but also of engineers. The following question:* from biomass—which is carbon efficient ■ “Time is running out. Strong for generation—and refinery coke? Q: Why not just change the specifications? We have to get very creative about the leadership now is essential to A: Now “throwing things away” is a fashydrogen requirements, but did previous cinating paradigm. There’s some magic prevent this,” — Nobuo Tanaka, generations have a monopoly on creativto it. For example, take a piece of used ity? And, this not only means meetings Executive Director of the printer paper, roll it up into a ball and to rewrite the commandments, but the throw it away over your shoulder. Now International Energy Agency. process will be expensive. turn around and have a look. Is it gone? Has it gone… “away”? As it sits there, does it bear any relation to Carbon and climate. Since a growing number of us accept the dictionary definition “out of existence”? that CO2 does something more odious than simply accelerate A friend of a friend was long-distance sailing in the Pacific. plant growth and that emitting it will cost money soon, perhaps As he and his group crossed the open ocean, they progressively the right thing to do would be to schedule all those meetings. As tied plastic bags of rubbish to the yacht railing. When they made time goes on, I increasingly feel that the extremists in the climate land on a tiny island, they threw it away in a conveniently placed debate are those who want to do nothing at all. Or it is the folcontainer. After a few days in the small harbor, the skipper noticed lowing sentiment, which arrived by e-mail: a truck collecting the waste and decided to follow it. The truck “If we do not develop several large-scale integrated carbon capwent to the edge of a cliff on a peninsula where the current was ture and storage demonstration projects within the next decade, away from the coast… well, you know the rest. we won’t be able to deploy the technology in time to prevent CO2 levels from exceeding allowable limits. If that happens, our only Throwing away. The wisdom of throwing these gases “away,” alternative would be to develop and deploy novel technologies to as if there were such a place, is now generally called into question. remove CO2 from the air, which would be enormously expensive In the meantime, this practice has been extended to the refining, and may not succeed.” HP chemicals, power generation and transport industries. Societies NOTE have allowed several hundred years of free permits to “throw away” * Per Olsson, the Preem Refinery director, suggested redrawing diesel specs gases, but that practice is changing. In Europe, there’s going to be and using surplus light distillates to address the European distillate deficit. He also a price on dumping carbon dioxide (CO2 ) to atmosphere. For this said of gasoline production at his two refineries: “Why bother? We think Swedish reason, I say open the “can of worms” and shake up the molecules. cars will be running solely on batteries and biofuels by 2020.” Here are some fairly wild shots as to how: • Instead of burning natural gas and pumping CO2 and air through turbines, how about just accelerated air, or wind, as we also know it? • Instead of putting light distillates into cars to power them, The author is HP’s European Editor and has been active as a reporter and conference chair in the European downstream industry since 1997, before which he was a how about using electrons from the turbines to charge batteries? feature writer and reporter for the UK broadsheet press and BBC radio. Mr. Wright • Instead of burning gaseous hydrocarbons to power turbines, lives in Sweden and is the founder of a local climate and sustainability initiative. how about picking up T. Boone Pickens’ idea of putting the gas HYDROCARBON PROCESSING AUGUST 2009

I 11

Select 101 at www.HydrocarbonProcessing.com/RS

HPIN CONTROL Y. ZAK FRIEDMAN, CONTRIBUTING EDITOR Zak@petrocontrol.com

APC designs for minimum maintenance—Part 3 This is the third in a series of three editorials about designing advanced process control (APC) applications for an environment of little to no maintenance. Having lamented for years about APC applications failing due to lack of maintenance, we have to accept that a personnel shortage is here to stay. If we want our APC applications to survive in a minimum-maintenance environment we have to make them as robust as possible. My June editorial proposed design rule 1: Avoid cluttering the multivariable predictive controller (MVPC) control matrix. Associate each control variable (CV ) with one, or a maximum of two manipulated variables (MVs). The July editorial continued with rule 2: Avoid nearly redundant CVs, and rule 3: Where near redundancy is necessary, restructure the CVs to improve matrix conditioning. This editorial adds two rules that affect how robust an application is and how much maintenance it would need to remain useful. Design rule 4. Keep the problem as linear as possible. MVPCs rely on linear models of the unit response to MV

changes. However, the real world is nonlinear. Unit response changes with time, throughput and operational mode, and if those changes are not passed on to the MVPC the controller becomes confused and could begin to cycle between two or more unoptimal solutions. Throughput-related nonlinearities are inherent in almost any application because MVPCs mix extensive variables, such as flows, with intensive variables, such as temperatures. If you halve the throughput then the response gain of a temperature CV to a flow MV would double. Commercial MVPCs permit gain scheduling on the fly, which offers a solution to the nonlinearity problem, although I have not personally seen massive use of this gainmultiplying feature. Can we make the matrix linear by designing intensive manipulated variables, using for example yields instead of flows? While product quality CVs are typically intensive, constraint CVs, such as flooding or valve positions, are extensive. Restructuring the MVs would hence eliminate some nonlinearities but introduce others. Should we make automatic use of gain multipliers? Gain multipliers adds complexity, but also robustness. Intuitively I think we should begin to design applications with automated gain scheduling. In a proper maintenance environment the APC engineer could manually scale the gains as needed and make the application run, whereas in the absence of maintenance, when the MVPC control models drift the application would likely be turned off. Are process gains easily predictable? Throughput-related gains are predictable because they are inversely proportional to the throughput, but for other nonlinearities we need more elaborate rules. Still, one cannot design effective APC for a unit without a detailed understanding of the unit, and that includes changes between operational modes.

Design rule 5. Use high-quality inferential models.

I come now to a topic near and dear to my heart. APC makes money by pushing the unit toward constraints, but such a push has value only if accompanied by precise product quality control. In the early years APC relied on simple inferential models (for speed of response) plus onstream analyzers (for accuracy) to achieve quality control. But analyzer reliability, never very strong to begin with, has further deteriorated over the years. The lack of maintenance environment that decimated APC has surely also reduced analyzer reliability. We are now at a point where we must design inferential models to work without being corrected by analyzers, especially for applications that are expected to survive in an environment of minimal maintenance. I have written several articles against the practice of developing inference models via regression analyses.1 A very good process engineer can perhaps specify model inputs correctly, conduct a series of test runs and identify a working model. He/she would need extensive lab support for the initial development as well as for continued testing and redevelopment upon process changes. While I have never been impressed by regression models, with continued lab support and periodic regression analyses maintenance of such inferences might succeed, but in an APC-neglect environment I cannot envision weak inferential models surviving very long. What then is the type of inference model that can survive in a low-maintenance environment? I have made a career out of developing first-principles models,2 and think that the more the model is based on process engineering principles the more reliable the inference is. Even the best inferential models need occasional recalibration, but that recalibration is fairly simple, involving a change of bias or multiplier. Actually, the most common cause of inferential model failure is an erroneous input measurement. An engineer must still be there to first identify the problem, and second to nag the instrument maintenance team, which is also short of people, to repair the offending measurement. HP 1 2

LITERATURE CITED Friedman, Y. Z., “Choosing inferential modeling tools,” HPIn Control, Hydrocarbon Processing, January 2006. Friedman, Y. Z., “Where do you take those inferential models from?,” HPIn Control, Hydrocarbon Processing, November 2008.

The author is a principal consultant in advanced process control and online optimization with Petrocontrol. He specializes in the use of first-principles models for inferential process control and has developed a number of distillation and reactor models. Dr. Friedman’s experience spans over 30 years in the hydrocarbon industry, working with Exxon Research and Engineering, KBC Advanced Technology and since 1992 with Petrocontrol. He holds a BS degree from the Israel Institute of Technology (Technion) and a PhD degree from Purdue University.

HYDROCARBON PROCESSING AUGUST 2009

I 13

We get to the heart of constructability issues. Weâ&#x20AC;&#x2122;ve been repairing and revamping FCCUs (the heart of every refinery) since 1976. This hands-on experience makes us eminently qualified to address complex constructability issues too. We help owners, engineering/construction companies and process design firms bring their proposed solutions to practical conclusion at the site. We perform project studies and analyses, lift engineering and rigging studies, and handle a myriad of other seemingly small but important details. We help you make the transition from engineered drawings to plant fabrication so that all the pieces can be fit perfectly and properly assembled in the field.

1605 South Battleground Road, La Porte, TX 77571 Call: 281-478-6200 or 1-800-478-6206 E-mail: wstrickland@altairstrickland.com www.altairstrickland.com

Call or E-mail us today to learn more.

Select 56 at www.HydrocarbonProcessing.com/RS

Continually Improving, Re-defining And Expanding Our Services For Industry. A RepconStrickland Company

HPINTEGRATION STRATEGIES ALLEN AVERY, CONTRIBUTING EDITOR AAvery@Arcweb.com

Intelligent toxic gas detectors: a smart move for the HPI In the wake of several high-profile industrial accidents, and against the background of an aging installed base and a shrinking workforce, HPI owner/operators should consider upgrading and modernizing the toxic and combustible gas detection equipment installed in their plants. Replacing older conventional devices with intelligent detectors will help to increase visibility of toxic gas data across the organization, improve reliability, and streamline device configuration and calibration. While replacing older, yet still functional, equipment may seem counterintuitiveâ&#x20AC;&#x201D;particularly in the current challenging economic environmentâ&#x20AC;&#x201D;the need for reliable toxic and combustible gas detection equipment remains, and safety and environmental regulations will not be relaxed any time soon. Modern intelligent gas detectors can improve both plant safety and environmental compliance. Users gravitating to intelligent gas detection. Several

trends will drive adoption of intelligent toxic gas detectors in new facilities and for replacing older conventional devices that have been in service for years. Among these are a need for visibility into toxic gas emissions data, stricter government regulations, the ability to remotely validate and diagnose sensors and transmitters to increase reliability and as a response to the transitioning workforce, and the increasing demand for SIL-rated devices in plants. Smart transmitter sales will outpace sales of conventional and low-cost devices, as users seek to utilize recent technological advances to improve visibility into plant safety and device diagnostics. There is also an increasing requirement to archive toxic gas emissions data and alarms, to meet regulatory requirements and for potential litigation by plant employees. This will fuel demand for transmitters that incorporate powerful onboard diagnostics capabilities and use digital communication protocols. Stricter enforcement of safety regulations. To comply

with increasingly tough safety and environmental regulations, owner/operators will need to revisit their safety and gas emissions strategies, assessing the efficacy and reliability of their safety systems and gas detection equipment. This will drive investment in safety systems and toxic gas detectors to mitigate the risk of catastrophic events. Gas detectors are on the front line, warning of hazardous gas emissions before they reach crisis levels. As such, there will be greater emphasis on deploying devices with greater sensitivity and coverage, and installing several types of detectors to ensure redundancy and overlapping coverage, should any one detector fail. Increased implementation of SIL-rated devices. The sharper focus on safety in the wake of a number of high-profile accidents and an increase in regulatory scrutiny have helped to drive the adoption of reliable SIL-rated transmitters for safety instrumented systems to mitigate the risk of catastrophic events. Toxic and combustible gas detectors are an integral part of safety

instrumented systems. Many leading gas detector suppliers offer SIL-rated transmitters, including DrĂ¤ger, General Monitors, Honeywell, Industrial Scientific and Sierra Monitors. Sensor validation and diagnostic monitoring capabilities for generating an audit trail are important to the effectiveness of the entire safety system. Wireless technology helps. The adoption of wireless field

devices by process industry users will continue to affect the dynamics of the market for toxic gas detectors. Although wireless gas detectors still only represent a small percentage of the overall toxic gas detector market, the wireless market is poised for strong growth over the next five years. Wireless technology allows users to install field devices in new measurement points that were previously not feasible due to the high cost of wiring. This is particularly true for hazardous or inaccessible areas. With wireless gas detectors, users can easily and economically expand the coverage of their toxic and combustible gas detection systems. Aging installed base. The installed base of conventional gas

detection equipment is aging. Every year, more and more units reach the end of their life cycle, requiring replacement. Companies are replacing conventional transmitters with the more capable smart transmitters that provide key enabling technology for data visibility, giving users access to diagnostic functionality and reducing requirements for manual calibration. Ironically, with each passing year, there are fewer plant personnel to protect, given the retirement boom, leaving fewer technicians to manage and maintain plant assets, including gas detection equipment. Arguably, relying on older detectors that do not incorporate the latest diagnostic capabilities and digital communication protocols puts the plant and surrounding areas in greater jeopardy. While gas detection equipment can function for decades, the sensors must be periodically calibrated, and it cannot be easily ascertained if a detector or sensor has failed. Many new intelligent gas detectors can calibrate themselves, diagnose potential problems and generate alerts automatically. Recent advances in intelligent gas detectors can give users a wealth of information about plant safety, and provide additional assurance in preventing catastrophic events that jeopardize the plant, its personnel and surrounding communities. Plant owners should take a closer look at their facilities to determine just how effective their existing gas detection equipment is, and avail themselves of the latest safety technologies. HP Allen Avery is an automation analyst for ARC Advisory Group. His focus areas include field systems (flow, level, pressure and temperature) and wireless networks. Prior to joining ARC, Mr. Avery was at Blackstone Research Associates, where he worked on several primary research projects that examined office technology issues. He holds BA and MBA degrees from the University of Rhode Island and a BA degree in studio art.

HYDROCARBON PROCESSING AUGUST 2009

I 15

Kobelco Screw Gas Compressors (API 619)

If you thought reciprocating or centrifugal compressors were the only options for heavy-duty process gas service, we have good news.

A Better Answer for Heavy-Duty Process Gas Service

KOBELCO rotary screw gas compressors are excelling in applications worldwide, including: Oil-Injected Compressor Applications: I Hydrogen for Gasoline & Diesel Desulfrization and Hydrotreating I Fuel Gas Boosting I PP & PE Process Gas I Gas Pipeline Boosting I Coke Oven Gas I Helium Oil-Free Compressor Applications: I Flare Gas Recovery I Offshore Vapor Recovery Unit (VRU) I Refinery Off-Gas, Vent Gas, Coker Gas I Heavy Hydrocarbon Gas I Dirty Gas I Petrochemical Process Gas (Styrene Monomer, Butadiene, LAB, Soda Ash)

â&#x20AC;Ś and more Superior Performance KOBELCO oil-injected screw compressors are robust, with discharge pressures up to 1,500 psig (100 barg) and extremely high compression ratios. Our oil-free screw compressors handle large capacities up to 65,000 CFM (110,000 m3/hr) and difficult gas applications.

Kobelco Screw Compressors â&#x20AC;&#x201D; better technology for heavy-duty process gas compression.

Ask KOBELCO! The Best Solution for Any Gas Compression

Tokyo +81-3-5739-6771 Munich +49-89-242-18424 www.kobelco.co.jp/compressor

Kobelco EDTI Compressors, Inc. Houston +1-713-655-0015 rotating@kobelcoedti.com www.kobelcoedti.com

Select 103 at www.HydrocarbonProcessing.com/RS

HPIN ASSOCIATIONS BILLY THINNES, NEWS EDITOR

bt@HydrocarbonProcessing.com

ILTA gathers to take annual look in mirror The International Liquid Terminals Association (ILTA) held its 29th Annual International Operating Conference and Trade Show on June 8-10, 2009, in Houston, Texas. Over 250 domestic and international companies exhibited in the trade show. More than 3,300 industry personnel attended the three-day event, which included a conference, the trade show and three training workshops. “This year, as the decline in our economy has continued, many companies in our industry have cut back on their travel budgets. However, ILTA’s strong membership base produced an excellent turnout for this year’s conference and trade show,” said ILTA President David Doane. The opening session of the conference featured a dialogue among four business leaders in the petroleum and chemical industries. The participants in the executive roundtable were: Curt Anastasio, president and CEO of NuStar Energy; Hank Heithaus, president of Retail Marketing at Murphy Oil USA; Olav Refvik, founding member of Arrowhawk Capital Partners; and Rod Sands, president and CEO of Explorer Pipeline Co. These executives provided their insights into key issues affecting companies in today’s economic and political climate. For the

2009 ILTA Safety Awards Platinum Award recipients

tional performance in protecting the safety of their employees (see table).

NuStar Energy

Overview of CFATS. The Chemi-

Western Refining Co.

cal Facility Anti-Terrorism Standards (CFATS) are of interest to all in the liquids storage industry. During a late Tuesday morning presentation, Peter Weaver of the ILTA advised attendees that facilities that fall under the regulatory purview of the standards need to submit a security vulnerability assessment (SVA) to the Department of Homeland Security (DHS). This assessment is the basis for the DHS to determine if a particular facility is ready and what tier classification it should fall under. There are four tier risk categories, but all are considered high risk. Dave Chalson of Sunoco Logistics Partners also spoke during this session, reinforcing the tier classification system. He emphasized the complicated nature of the DHS evaluation and the 18 risk-based performance standards that facilities are expected to comply with. The conference concluded with the keynote luncheon on Tuesday. Rick Pitino, head basketball coach for the University of Louisville, gave an inspirational speech about staying positive and productive in today’s world. HP

Excellence Award recipients Asphalt Operating Services Buckeye Terminals CITGO Petroleum Corp. Flint Hills Resources Hess Corp. Houston Fuel Oil Terminal Co. Intercontinental Terminals Co. International Raw Materials JIT Chemical Corp. Marathon Petroleum Co. Motiva Enterprises – New Jersey Complex Petro-Diamond Terminal Co.

first time, this year’s conference featured sessions in Spanish. These technical presentations focused on three key issues for terminal operators—tank emission controls, inventory management systems and vapor recovery technologies. On Tuesday morning, 12 ILTA terminal member companies received the Safety Excellence Award for demonstrated excep-

Representatives from companies that won ILTA safety awards proudly show off the trophy that commemorates their achievement.

Peter Weaver of ILTA and Dave Chalson of Sunoco Logistics Partners offered an overview of CFATS regulations and how they will affect the liquid terminals industry.

HYDROCARBON PROCESSING AUGUST 2009

I 17

C E R T I F I CA EM TI

:2 0 0 0

PRODUCTS:

SERVICES:

CAPABILITIES:

• NEW EQUIPMENT • OIL & GAS • GEOTHERMAL • AIR SEPARATION • REDESIGN & UPGRADE • CONTROL SYSTEMS • MAGNETIC BEARINGS

• ENGINEERING SERVICES • REPAIRS • FIELD SERVICES • SPARE PARTS • ROTOFLOW ® SERVICE & SUPPORT

• 60,000 SQ. FT. FACILITY NORTH OF L.A. • FULL MACHINE SHOP CAPABILITIES • CNC 5-AXIS FOR WHEEL PRODUCTION • WORLD-WIDE REPRESENTATION

FOR A FREE EVALUATION OF YOUR CURRENT TURBOEXPANDER EQUIPMENT CALL

661.294.8290 OR EMAIL US AT SALES@LATURBINE.COM

www.laturbine.com Tel: +1.661.294.8290 | Fax: +1.661.294.8249 29151 Avenue Penn, Valencia, CA USA Select 98 at www.HydrocarbonProcessing.com/RS

IS O

WORLD LEADER In Turboexpander Technology and Equipment

low f o ot port R s in Sup t r d xpe e an E c e Th ervi S

HPIMPACT STEPHANY ROMANOW, SENIOR PROCESS EDITOR

SR@HydrocarbonProcessing.com

Mark Finley, general manager of Global Energy Markets for BP America, shared BP’s view on future global energy at the Baker Institute Energy Forum. According to the BP Statistical Review of World Energy 2009, energy demand for the US will be down in 2009 and part of 2010. In fact, energy demand by non-OECD (Organization for Economic Co-operation and Development) countries exceeded OECD countries in 2008. Over the last five years, non-OECD countries have had the fastest growth rate and have dominated global energy growth since 2000. This is no surprise. There is much hope that China and India will provide corrections for the global energy markets. Yet, this optimism is muted as China becomes more energy efficient and uses less energy (in all forms) in the future. In 2008, the global economic growth rate was 2%—below the 10-year average for the first time since 2003. The US has been officially in recession since December 2007. High volatility in the energy and financial markets have deepened this recession. The extraordinary speed and spread of this recession was transmitted by lack of credit and working capital and then by the collapse of international trade. Recovery remains uncertain. Big economies are replacing consumer spending with government deficit spending, thus slowing down economic growth recovery.

Demand by resource. Crude oil remains the dominant fossil fuel and accounts for 38% of the energy mix. Crude oil prices have been on a seven-year increase cycle. Since 2003, crude oil prices have increased from $30/bbl to the all-time high of $145/bb in July 2008 before abruptly falling to a low of $34/bbl in December 2008. Since then, oil prices have rebounded to over $50/bbl. Globally, oil consumption has fallen; much of this decline is attributed by one OECD nation—the US, the largest gasoline market. In contrast, non-OECD countries showed consumption growth until the economy deteriorated. US crude oil consumption dropped 1.3 million bpd (6.4%) in 2008—the largest volumetric fall since 1980. Gasoline demand declined in response to higher prices, while declining diesel demand reflected the deteriorating state of the economy. The high oil prices of mid-2008 caused unilateral production increases by OPEC. Unfortunately, there is a six-month lag Non-OECD

GDP 2

–1

–2

0 2004 2005 2006 2007 2008

Primary energy

from wellhead to refinery. When crude oil demand began crashing, OPEC producers made a series of production cuts of nearly 4 MMbpd to bring demand and supply closer to parity. Yet, the decline in crude oil supplies is mainly attributed to product declines by non-OPEC producing nations. For example, Mexico, the US, Norway, Russia and Nigeria all experienced production declines. A combination of field maturity, high capital (entry) costs and restrained access to reserves contributed to declining production rates for non-OPEC producers. Yet, production declines by non-OPEC producers were more than offset by production increases by OPEC producers. At the end of 2008, OECD inventories rose by 134 MMbbls—the highest level since 1984. As OPEC nations continue to cut production rates, spare capacity increased in 2009 (Fig. 2). OPEC nations hold the majority of the global crude oil reserves and are in the position to make investments for long-term crude oil production. However, openness to investment has deteriorated, thus constraining crude oil supplies and supporting volatile pricing for energy. Refining. The refining industry suffered

two blows in 2008—new capacity additions and declining product demand from the recession. The middle distillate market had the fastest growth rate in the first half of 2008 as gasoline demand plunged from high prices. During the second half of 2008, spare capacity from new processing facilities came online from China, India and the

Reﬁning margins 10 Reﬁning margins, $/bbl

GDP and primary energy growth, %

OECD

Contraction of the global economy caused strong downward movement of energy prices and consumption. In 2008, prices for all fossil fuels peaked mid-year and then fell. Primary energy growth slowed to 1.4%. Looking back, 2008 was a year with two halves. Prices and energy consumption rose in the first half of the year and then both declined in the remainder of the same year. Yet, non-OECD primary energy consumption exceeded OECD consumption for the first time; these nations account for 51.2% of the global energy consumption (Fig. 1).

BP GIM 10-year average

8 6 4 2 0

2004 2005 2006 2007 2008

Spare capacity growth Spare capacity growth, MMbpd

Global energy demand down in 2009

1.2 0.9

From new investment From lower crude runs

0.6 0.3 0.0

1998 2000 2002 2004 2006 2008

2006

2007

2008

Source: BP Statistical Review of World Energy 2009

FIG. 1

Primary energy growth and GDP for OECD and non-OECD nations.

FIG. 2

Global refining margins and spare capacity—1998 to 2008.

HYDROCARBON PROCESSING AUGUST 2009

I 19

GE Energy

Gasification expertise? It’s what sets us apart. With more than six decades of world-class industry knowledge and expertise, GE Energy offers commercially proven gasification solutions to customers worldwide. GE's gasification technology enables you to convert coal and other low-value hydrocarbons into syngas that can then be used in the manufacture of marketable products—like basic chemicals, steam, hydrogen and power. Operating in 63 sites, our technology delivers a wide range of efficient, reliable gasification customized for the specific needs of your growing business. We’ll work with you to enhance your plant’s competitiveness and profitability. To learn more about our innovative gasification solutions, visit our website at ge-energy.com/gasification.

Select 91 at www.HydrocarbonProcessing.com/RS Long-term license agreements include the gasification process design package; process and instrumentation design review with EPC; start-up support; full technology transfer with various technical support services and operator training.

HPIMPACT Middle East. A surplus of transportation fuels outpaced demand. As margins collapsed, refiners worldwide began cutting utilization rates—the global refinery utilization rates slipped to an average of 85% in 2008. Ethanol continues to displace gasoline volume, thus creating surpluses in some markets, while demand for middle distillates continues to increase globally. Natural gas. Like crude oil, natural gas (NG) prices followed a similar pattern— peaking in July and dropping for the remainder of the year. World gas production grew by 3.8% in 2008 and consumption increased by 2.5% and then fell rapidly in the second half of 2008. Non-OECD countries used more NG than OECD countries for the first time; China’s NG consumption increased by 15%. In addition to diminishing demand due to the global recession, more NG was made available from unconventional gas resources. Nonconventional NG reserves benefited from new investments and technology. In the US market, 50% of new NG reserves were sourced to nonconventional reserves—in particular, shale gas. With new NG production, Henry Hub prices peaked in July 2008 and fell in the second half of the year. NG is trading at a discount to crude oil. Directional drilling has been instrumental in bringing new gas reserves online. Only 19% of global NG is traded by pipeline and 7% as liquefied natural gas (LNG). Lower prices have resulted in a lower drilling rig count. LNG still links producing regions to consumer markets. LNG tanker capacity increased by 19% in 2008. Excess NG capacity in the US market added more flexibility to LNG trade; Atlantic Basin NG was shipped as LNG to Asia Pacific—a 12% increase from 2007. Abundant NG reserves in the US market caused LNG imports to fall by more than 50% in 2008. New LNG projects in Qatar, Russia and Yemen will add more LNG capacity this year and cause more LNG to search for a home. The 2009 NG market is one of too much supply chasing not enough demand. Other fuels. Global coal consumption slowed in 2008, increasing by 3.1%. Coal remains the fastest growing fuel in the world. China remains the largest coal consumer (43% of the world share), and domestic consumption grew by 6.8%—accounting for 85% of global energy market growth. Coal prices are more volatile than crude oil or NG. Excluding China and India, coal

consumption fell in other markets. Carbon dioxide (CO2) taxes favor NG over coal, especially for electrical power generation. Nuclear power. Output dropped 0.7% for a second year. No new units entered service in 2008. Hydroelectric power generation increased 2.8%. China’s hydroelectric power increased by 20.3%; elsewhere, hydroelectric output was weaker. Renewables (solar and wind) account for a small share of total energy consumption and still rely on government subsidies to make inroads in the energy market. Ethanol. Only 0.9% of the global oil

consumption is held by ethanol. Ethanol production has accelerated for a fourth year, rising 31% (700 Mboe/d in 2008). The US market accounts for 62% of the growth in ethanol production; Brazil holds the remainder. US ethanol production increased 600 Mbpd in response to higher blending rates and high gasoline prices. The mid-year slowdown caused the US ethanol industry to be overbuilt—15% of the US ethanol production capacity was idled at year-end. The full report is available online at www.bp.com.

Economic recovery in sight for chemical industry “Recovery is in sight for the US and global chemical industry,” according to the American Chemistry Council’s mid-year 2009 outlook. At mid-year, expectations for 2009 are stabilizing; the freefall is over and the end of the recession is within sight. Market indicators show that the pace of decline in production, employment and business has moderated. The “green shoots” of recovery have emerged for the chemical industry. Monitored economic indicators (including oil and natural gas prices, worldwide production, trade, shipments, inventories, price indices, energy, employment, investment, R&D, EH&S, financial performance, etc.) for the chemicals and petrochemical industries have trended upward for two consecutive months. And labor market indicators have improved slightly. The large imbalance between sales and inventories is slowly closing. Consumers remain wary, however, and the outlook for consumer spending is muted. Rising energy prices will slow economic recovery in the US market. Result: The US economy is expected to fall 2.8% in 2009—the largest annual decline since 1946. In the future, the US economy will gain traction, and expected

growth is 1.6% in 2010 and 3.2% in 2011. Looking abroad, the economic contraction first emerged in the US and quickly spread through the rest of the world. Trade volumes have fallen sharply as financing and demand for goods and services waned. In Q1 of 2009, nearly every producing nation experienced a synchronous contraction. Industrial production in most developed nations posted double-digit declines as global demand faltered and inventories surged. According to the International Monetary Fund’s April outlook, world trade volumes are estimated to be off 11% and the global GDP will contract 1.3% in 2009. Stabilization in emerging Asian economies appears to be underway. In looking ahead, the ACC forecasts that the pace of the downturn will be moderate. Growth in the industry will emerge beginning in Q3 of this year. As product inventories are worked off and demand stabilizes, production will increase. Industrial production growth in the US is expected to be down 11.2% in 2009 before rebounding in 2010. End-user markets depressed. The

recent bankruptcy of two US major automakers has reduced light vehicle production. Light vehicle sales are expected to be $9.9 million in 2009, down from $13.2 million in 2008. Spending on residential construction appears to be stabilizing; yet, it is at a historical low point. Worst decline since 1980s. In May,

US chemistry output was off 9.5% Y/Y and capacity utilization rates dipped to 69.5%— down 7.1% from 76.6% in 2008. Excluding pharmaceuticals, chemical output was off 15.5% Y/Y. Along the supply chain, inventories-to-sales ratios for chemicals have improved over recent months. US output slipped 4.7% in 2008. The ACC expects that the chemistry output to fall 8.1% in 2009 and then to slightly recover in 2010. Excluding pharmaceuticals, 2009 chemical output is expected to fall. A weak global economy will hinder US exports and recovery. The global chemistry industry appears to be improving, led by emerging Asian nations. Global chemical output is expected to slip 6.3% in 2009 before recovering in 2010 and 2011. Growth in developing countries will outpace that in developed countries, which endured steeper output declines. (Report available at www.americanchemistry.com) HP HYDROCARBON PROCESSING AUGUST 2009

I 21

Spread the word!

For more information please contact compressor-mechatronics@hoerbiger.com

Select 61 at www.HydrocarbonProcessing.com/RS

www.hoerbiger.com

The new RecipCOM delivers diagnostics, protection and therapy for your reciprocating compressors.

HPIN CONSTRUCTION BILLY THINNES, NEWS EDITOR BT@HydrocarbonProcessing.com

North America Pearl will be managing the construction of an expansion project for a gas processing plant in North Louisiana. Pearl is currently executing the engineering for the addition of inlet separation, dehydration and utility storage for an existing processing plant in Louisiana. Pearl has also been awarded the management of the construction phase of the project. The inclusion of these services is expected to provide Pearl $1.2 million in revenue.

Europe Hellenic Petroleum has started a new 27,000-bpd fluid catalytic cracking gasoline desulfurization unit in Greece. The unit utilizes CDTECH technologies to produce ultra-low-sulfur gasoline. In addition, this unit has the capability to produce a low-sulfur diesel stream, which can be blended into the diesel pool.

LNG. Air Products’ LNG technology and equipment are to be delivered to the western Mediterranean port site by October 2010. Saipem, in joint venture with Chiyoda, is the contractor on the project for Sonatrach. Air Products will be providing the GNL3Z LNG project its liquefaction process technology and its main cryogenic heat exchanger. SNC-Lavalin has an engineering procurement and construction contract with Sonatrach to design and build natural gas processing facilities in the Sahara desert, approximately 1,200 km southeast of Algiers, Algeria. The €785.5 million contract calls for SNC-Lavalin to build a gas treatment complex, including infrastructure to collect raw gas at four different fields, a natural gas processing facility and a unit to process and reinject carbon dioxide into gas fields. Once in operation, the facilities will produce and process 3.5 billion m3/y of natural gas.

Jacobs Engineering Group Inc. has a contract with Storengy to provide engineering, procurement and construction management services to upgrade its underground gas storage facility in Etrez, France.

Samsung Engineering has a $1.6 billion contract with SATORP for two refinery projects to be constructed in Al-Jubail,

TOTAL Raffinerie Mitteldeutschland GmbH and EDL Anlagenbau Gesellschaft mbH have a new framework contract for engineering services in Germany.

Hydrocarbon Processing maintains an extensive database of historical HPI project information. Current project activity is published three times a year in the HPI Construction Boxscore. When a project is completed, it is removed from current listings and retained in a database. The 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.

Middle East Saudi Aramco Total Refining and Petrochemical Co. finalized the plan to award engineering, procurement and construction (EPC) contracts for the 13 different process packages of its 400,000-bpd full-conversion refinery in Jubail, Saudia Arabia. The refinery will maximize production of diesel and jet fuels and will also produce 700,000 tpy of paraxylene, 140,000 tpy of benzene and 200,000 tpy of polymer-grade propylene. It should be fully operational by the second half of 2013. Air Products has an agreement with Saipem to provide its proprietary process technology and main cryogenic heat exchanger for Sonatrach’s GNL3Z liquefied natural gas (LNG) project in Arzew, Algeria. Sonatrach will produce 4.7 MMtpy of

TREND ANALYSIS FORECASTING

The historical information is available in comma-delimited or Excel® and can be custom sorted to suit your needs. The cost of the sort depends on the size and complexity of the sort you request 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: Lee Nichols P. O. Box 2608 Houston, Texas, 77252-2608 Fax: 713-525-4626 e-mail: Lee.Nichols@gulfpub.com.

Saudi Arabia. The first project is an aromatics plant valued at $700 million. It will produce 700,000 mtpy of paraxylene and 140,000 mtpy of benzene. Samsung Engineering will provide the engineering, procurement and construction of this package on a lumpsum turnkey basis, with completion scheduled for August 2012. The second project is a $900 million delayed coker unit, which was obtained through a strategic collaboration between Samsung Engineering and Chiyoda. By June 2013, the unit is expected to produce 100,000 bpd of light hydrocarbons like LPG and naptha through a thermal cracking process.

Asia-Pacific INPEX Corp. recently started construction of a liquefied natural gas (LNG) receiving terminal at the port of Naoetsu in Joetsu City, Japan. The LNG receiving terminal project is estimated to be completed by 2014. Technip has a lumpsum contract with Ningxia Hanas Natural Gas Co. Ltd. for construction of a mid-scale liquefied natural gas (LNG) plant to be built in Yinchuan, China. The contract covers the engineering, supply of main equipment, procurement and construction management services for facilities for natural gas pre-treatment, liquefaction, LNG storage and loading, utilities, offsites, buildings and other infrastructure. The plant will have two trains with a capacity of 400,000 tpy each, based on an Air Products liquefaction process. The contract is scheduled to be completed in the second half of 2011. Axens has an agreement with Nghi Son Refinery & Petrochemical LLC for basic engineering design for some of the units in the new 200,000-bpd refinery to be constructed in the Thanh Hoa Province of Vietnam. Axens will provide a residue fluidized catalytic cracking (RFCC) unit, a gasoil desulfurization unit and a kerosene desulfurization unit. The facility is scheduled to be operational in 2013. Polymer-grade propylene will be supplied to a downstream polypropylene unit. Axens’ RFCC is part of the FCC Technology Alliance between Axens, Shaw, Total and IFP. HP HYDROCARBON PROCESSING AUGUST 2009

I 23

)DEAS FOR GROWTH )DEAS FOR SUSTAINABILITY

!MMONIA &ERTILIZER s 3YNGAS s (YDROGEN s 2ElNING

/RGANIC #HEMICALS s /LElNS s #OAL 'ASIlCATION s #ARBON #APTURE 3TORAGE

KBR Technology specializes in developing and licensing process technologies worldwide. From reďŹ ning to ammonia, from chemicals to coal gasiďŹ cation, from oleďŹ ns to syngas, KBR Technology helps you accelerate proďŹ tability and sustain growth. For more information, visit technology.kbr.com/HP or email technology@kbr.com ÂŠ 2009 KBR All Rights Reserved K09088 08/09

*OIN US IN $UBAI /CTOBER FOR THE +"2 4ECHNOLOGY #ONFERENCE h4ECHNOLOGY 3OLUTIONS TO -EET 9OUR .EEDS 4ODAY AND 4OMORROWv Visit TECHNOLOGYCONFERENCE KBR COM $UBAI for more information. Select 89 at www.HydrocarbonProcessing.com/RS

LETTERS TO THE EDITOR editorial@HydrocarbonProcessing.com

Words of praise I want to thank you for Hydrocarbon Processing magazine. I consider it one of the most premier, professional and informative trade journals that I receive. Recently, I had a question concerning an article in the January 2009 issue and the editorial staff kindly answered my questions in a very timely and courteous manner. I often refer to and quote your articles on energy and the environment when talking to my state senators, state representatives and congressional representatives staff. Thank you! David Herr

Quibbling over NPSHA I am most surprised that an engineer with a rotating equipment background would write a column regarding pump suction strainers (“HPIn Reliability,” April 2009, p. 9) while never even mentioning the

effect on system net positive suction head available (NPSHA). Having been an engineer dealing with rotating equipment for 40 years, I would first recommend that the pump suction system be as clean as possible. By this, I mean there should be no elements that could present even the slightest pressure drop or start of a flow vortex that could detrimentally affect pump inlet conditions. James E. Bonerigo, PE United Refining Co. Warren, Pennsylvania

Editor’s response We at Hydrocarbon Processing must keep our editorial columns to a single page. That said, we confined our short narrative to “issues,” meaning items that are frequently overlooked. Since, in determining available NPSH, it is extremely rare for competent project engineers to disregard such rel-

Select 152 at www.HydrocarbonProcessing.com/RS

evant parameters as pipe size (affecting flow velocity and friction), block valves, strainers, filters and so on, we elected to not even mention NPSHA. Calculations are available in literally hundreds of texts, including Bloch-Soares’ Process Plant Machinery (ISBN 0-7506-7081-9). Many books also recommend maintaining an appropriate margin (or ratios) between available and required NPSH. Bloch-Budris’ Pump User’s Handbook (ISBN 0-88173-517-5) mentions that these ratios may have to be as high as 20! Please kindly note that we disagree with Mr. Bonerigo’s statement regarding “even the slightest pressure drop.” For one, experiencebased checklists are used by conscientious engineers. Most of them know that progressive corrosion is usually encountered in piping and that “as clean as possible” is a subjective (or even contentious) term. Heinz P. Bloch

HYDROCARBON PROCESSING AUGUST 2009

I 25

Thousands of CCC customers worldwide donâ&#x20AC;&#x2122;t think so. In the industrial environment, no one can afford to sacrifice the high quality of a control system and a responsive customer service, which is often a key to a long-lasting protection of your plant, process, and equipment (your own peace of mind should not be overlooked either!). CCC offers real value in turbomachinery control. We call it the CCC Value Triangle: Profit Enhancement Opportunity. Quality. Service. CCC high-quality, precision-designed control systems and field-proven proprietary control algorithms along with the unsurpassed lifetime engineering support provide real profit enhancement opportunities resulting from:

Preventing unnecessary process trips and downtime Minimizing process disturbances Preventing surge, overspeed and associated damage Automating startup and shutdown Operating close to limits, safely Operating at lowest possible energy levels Minimizing antisurge recycle or blow-off Optimizing load sharing of multiple units We are on your side, at your site and at your service, 24/7, offering lifetime support, guaranteed by our large body of technical resources, experience on thousands of installations worldwide, and institutional memory.

The Right Solution â&#x20AC;&#x201D; The First Time, Every Time P o w e r B e h i n d B e t t e r Tu r b o m a c h i n e r y C o n t r o l â&#x201E;˘ International Headquarters: 4725 121st Street â&#x20AC;˘ Des Moines, Iowa 50323-2316, U.S.A. â&#x20AC;˘ Tel: (1) 515 - 270 - 0857 â&#x20AC;˘ Fax: (1) 515 -270 -1331 E-mail: solutions@cccglobal.com â&#x20AC;˘ Web site: http://www.cccglobal.com

Select at www.HydrocarbonProcessing.com/RS Select 77 Select74 70at atwww.HydrocarbonProcessing.com/RS www.HydrocarbonProcessing.com/RS

FLUID FLOW AND ROTATING EQUIPMENT

SPECIALREPORT

Understanding and maintaining an effective lubrication system Follow these guidelines to improve gearing equipment life J. DeBAECKE, Philadelphia Gear Corp., King of Prussia, Pennsylvania

or gearing equipment owners and operators, the ultimate goal is to achieve a return on their investment; this is done by both maximizing the output, reliability and efficiency of their machinery, and minimizing downtime and operating costs. Continued reliability, successful operation and long life of power transmission equipment largely depend upon the constant supply of lubrication oil of proper quantity, quality and condition. The lifeline of the gearbox is its lubrication system, critical for supporting the drive under all modes of operation. The purpose of a gearbox lubrication system is to provide an oil film at the contacting surfaces of all working components to reduce friction and wear. In addition, the oil serves to remove and dissipate heat from where it is generated, preventing gearing component temperatures from rising to excessive levels. Other lubrication functions include the transfer and/or removal of wear particles, as well as the filtration of rust and corrosion and any other undesirable contaminants. However, failure of the lubrication system to perform any one or more of these functions may result in premature equipment failure. Understanding the role and importance of a lubrication system in the overall life of a gearbox serves as a foundation for understanding the needs for maintaining such an effective system. And that is what this article aims to do—provide maintenance professionals with the tools to properly understand the lubrication needs for extending overall life of their gearbox. This article examines the number of lubricant types available, as well as the systems used to supply such lubricant throughout a gearbox. In addition, proper

maintenance functions are provided for sustaining a functional, effective lubrication system. Understanding lubrication. Lubrication can be defined as the control of friction and wear between adjacent surfaces by the development of a lubricant film between them, called an elastohydrodynamic (EHD) oil film. EHD film thickness between gear tooth surfaces is quite small, usually less than 1.25 micrometers (0.00005 in.). Oil film thickness is significant—if the adjacent surfaces are not fully separated, the EHD film leaves local areas of contact between those surfaces, making them vulnerable to surface fatigue. Viscosity is a characteristic of fluids to resist flowing freely. It is one of the most important characteristics of a lubrication fluid. Lubricating oil viscosity changes appreciably with temperature, and is generally stated at two temperatures: 40°C (100°F) and 100°C (210°F). Viscosity is usually expressed in terms of the time required for a standard quantity of a fluid at a given temperature to flow through a standard opening. Fatigue life of contacting components of a gearbox, such as gear teeth and bearing rollers, is determined by a complex combination of speed, load, lubricant temperature, clearance and alignment. The lubricant’s role in this interaction is determined primarily by speed, viscosity and temperature. The effect of these factors on the fatigue life of elements can be dramatically altered at higher temperatures with lower viscosity, and thinner resultant oil films. Selecting the correct lubricant for any application requires a careful study of expected operational and environmental conditions.

Gear lubricants. Several factors must be considered before choosing a gear lubricant—the unit’s operating speed and load, temperature range and lubricant availability, to name a few. However, the most important parameter in selecting a lubricant is viscosity. High-speed units produce an acceptable oil film at the tooth contact area even with a low-viscosity oil; at lower operating speeds, a thinner oil film is generated, requiring more viscous oils to separate contacting surfaces. Still, often a gearbox will contain both high- and low-speed gear meshes. In these cases, a compromise must be obtained (though in such cases, performance of these gear meshes may be reduced). There are two basic types of lubricants used in gear drive systems: petroleum-based mineral oils and a general category known as synthetic lubricants. Petroleum-based lubricants. Petroleum-based mineral oils are complex mixtures derived from refining crude oil. Petroleum products have been found to excel as lubricants in most applications. Mineral oils are usually compounded with different chemical additives to improve specific properties such as increased lubricant life, resistance to rust and oxidation and even increased load-carrying capacity. High-load oils, called extreme-pressure (EP) gear lubricants, contain selected additives that increase the load-carrying capacity of gearing by forming a film on the metal that provides component separation under higher load conditions. EP lubricants are ideal for use when severe operating conditions are anticipated. Often these lubricants will contain more than one chemical additive for load capacity enhancement over a wide temperature range, most commonly compounds of phosphorous and sulfur. However, EP HYDROCARBON PROCESSING AUGUST 2009

I 27

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

gear lubricants should not be used in gear units containing an internal backstop or an internal friction clutch unless the lubrication types used have been specifically approved by the gearbox manufacturer. Synthetic lubricants. Synthetic lubricants consist of base fluids manufactured by chemical synthesis or molecular restructuring to meet specific physical and chemical qualities desired for certain operating parameters, such as high-temperature thermal and oxidation stability, low-viscosity

variation over a broad temperature range, low-temperature capability and/or long service life. Care must be taken when synthetic lubricants are substituted for previously utilized lubricants. Compatibility with other gearbox components like rubber lip seals, rubber O-ring seals and housing paint must be established. Synthetic lubricants can be up to four times more costly than petroleum-based oils, and are thus generally reserved for problem appli-

Mission: Immersion. Immersion Engineering™™ goes deep to solve your heat transfer problems.

cations such as extremely high or low temperatures, equipment subjected to frequent overloads and equipment with a marginal lubrication system. The largest class of today’s synthetic lubricants is the esters—materials containing the ester chemical linkage. Esters have wide operating temperature ranges and high viscosity indices—thus permitting low-temperature operation, as well as providing good lubrication characteristics at high temperatures. A lubricant’s viscosity index is a measure of how much that oil’s viscosity varies with temperature. Another class of synthetic lubricants is the synthesized hydrocarbons—these lubricants contain many of the advantages of esters (to a lesser extent), but have a similar structure to mineral oils, making them compatible with mineral oils while not being detrimental to seals and paints (esters have low compatibility with some polymeric materials such as those used in seals and paints). Lubricant viscosity selection. In general, the lowest viscosity oil capable of forming an adequate oil film at all operating conditions should be chosen. However, in practice, the lubricant chosen is often a compromise between the requirements of the various lubricated components—such as gears and bearings—and the particular application requirements such as large ambient temperature differentials. Lubrication systems. There are two

Even though you may call us on the phone miles away, we're so deep into your stuff--your fluid, your equipment, your system--we can virtually touch it, see it. Immersion Engineering is a bundle of very specialized services that you can cherry pick. Some are free, some you pay for. We’re the only company offering them all. One thing is for sure; when you need HTF help you need it now. Nobody knows more about the chemistry, Eyeball this selection of services. ■ Fluid Analysis ■ Troubleshooting ■ Fluid Maintenance ■ System Layout ■ Training

performance and applications of heat transfer fluids than we do. So pick a service and call one of our technical specialists. Or, check out our web site for case histories, data sheets, comparisons, user’s guide, tip sheets and technical reports. It’s all there, it’s deep, it’s Immersion Engineering.

HEAT TRANSFER FLUIDS

4 Portland Road West Conshohocken PA 19428 USA

800-222-3611 610-941-4900 • Fax: 610-941-9191 info@paratherm.com

www.paratherm.com

Select 153 at www.HydrocarbonProcessing.com/RS 28

types of gearbox lubrication systems in use: splash lubrication systems and force-feed lubrication systems. The intent of both types of systems are the same, to distribute oil to each component of the gearbox sufficient for lubricating and cooling that component yet minimizing heat generation by oil churning. Splash lubrication. Splash lubrication systems require that the gearbox be filled to a predetermined lubrication oil level. Rotating gear elements within the gearbox must dip into the oil and “sling” it into troughs, pockets or directly to bearings and gear meshes requiring lubrication and cooling oil. Feed troughs are employed to capture oil that is “slung” onto the upper gearbox housing wall by a dipping gear element. This oil drips into the trough which, in turn, distributes that oil to the bearings. Such systems are better suited for gearboxes containing rollingelement bearings than those with journal bearings, which require far more oil. A splash lubrication system requires at a minimum, oil troughs, bearing oil pock-

FLUID FLOW AND ROTATING EQUIPMENT ets, an oil fill location, an oil drain and a breather. In cold ambient temperatures, an immersion heater should be provided in the sump. Cold starting temperatures can cause oil viscosity to be too high to properly distribute oil upon startup. Splash lubrication systems are far simpler and less expensive than force feed, but are applicable only to low-speed gear units. As shaft operational speeds increase, the heat generated in the gearbox becomes excessive, requiring an external, force-feed system to supply larger volumes of lubricant to lubricate and cool gearbox components. In addition, higher-speed units require oil to be precisely introduced at the gear and bearing interfaces; this is accomplished through strategically placed jets to properly lubricate the gear meshes and dedicated bearing oil supply lines. Temperature control/thermal rating. The second primary function of the lubrication system is to provide heat removal. Adequate cooling is necessary to maintain oil viscosity control and oil quality. Conversely, for every gear drive there is a thermal rating; the average power that can be transmitted continuously without overheating the unit and without using any special external cooling method. If the thermal rating is less than the mechanical rating—the load a gearbox can transmit—additional cooling supplied by a force-feed lubrication system is required. Auxiliary cooling can be used in combination with splash lubrication to increase the thermal rating of a gearbox—for instance, air can be forced past the radiating surfaces of the gear casing by strategically placed fans internal to the gearbox and located on a high-speed pinion shaft. In addition, the unit can be cooled by a water jacket; water passages are built into the gear housing, usually at the high-speed end, and heat is carried away by a cooling water flow that is isolated from the lubrication oil sump. To operate a gear unit at maximum efficiency, auxiliary cooling schemes should include thermostatic controls so that the oil is not cooled unnecessarily. Operating with too cool a lubricant increases churning losses. Adding cooling fins to increase the surface area of the gearbox casing can increase the heat transfer from the gear casing to the ambient air. Force-feed lubrication. In a typical force-feed lubrication system, a shaft- or a motor-driven oil pump draws oil from the gearbox sump through a suction pipe. The oil is directed from the pressure side

of the oil pump through a filter to cleanse the oil, and through a cooler employed to cool the oil. A pressure relief valve is typically located before this filter to protect the system from too high an operating pressure. If the filter becomes clogged, the relief valve will permit the unfiltered oil to bypass the filter so the gearbox will continue to receive lubrication and cooling oil albeit unfiltered (unfiltered oil is better than no oil). Another relief valve is often located at the inlet to the gearbox

SPECIALREPORT

to limit the oil feed pressure if the system contains both shaft- and motor-driven pumps, and both are running at the same time. A shaft-driven oil pump is driven by one of the rotating gear shafts of the gearbox. Some lubrication systems will include both motor- and shaft-driven oil pumps. The motor-driven pump can be activated prior to gearbox startup to supply full oil flow requirement to the gearbox prior to shaft rotation until the attached lube oil pump is running at a speed sufficient to

We a prese re nt at Offsh o 08-11 re Europe Sep Aberd tember een

Select 154 at www.HydrocarbonProcessing.com/RS

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

supply full lubrication oil flow to the gearbox, during gearbox coast-down or as a backup in case of failure of the main shaftdriven oil pump. Check valves are located so that the main pump does not pump through the auxiliary system and that the auxiliary pump does not pump into the pressure side of the main oil pump. A bypass is provided at the cooler serving as both a pressure relief valve and/ or a thermostatically controlled valve set so that the pressure drop across the cooler is limited during times when the oil is cold; additionally, temperature and pressure sensors are located at various critical points throughout the system. Relatively little oil is required for lubrication using a force-feed system provided it is properly applied. The bulk of the oil flow is required for cooling the gear tooth flanks and bearings. For demanding high-speed applications, gear tooth meshes are sprayed on either in-mesh or out-mesh sides or, in some instances, both sides. System components. A typical forcefeed lubrication system consists of the following major components:

$100

Pumps. The gearbox oil pump delivers a given quantity of oil over a wide range of oil temperatures and viscosities. In addition, the gearbox pump must be capable of priming itself and overcoming pressure drops in the line between the oil reservoir and the pump suction port. The most common method of lubricant delivery is the positive-displacement lubeoil pump—these pumps deliver a given quantity of fluid with each pump rotor revolution. A positive-displacement pump’s output is directly proportional to its operating speed and offers practically constant oil flow at any particular speed regardless of downstream conditions. Gearbox lubrication pumps can be mounted to the unit and driven by one of the gearbox shafts, or independently mounted with an electric motor or other prime mover driving. When the pump is shaft driven, oil flow will vary directly with shaft speed. In a gear pump, as the gears rotate, fluid is trapped between the gear teeth and the case, and is carried around the pump casing to the pump discharge oil port. Filtration. Gearbox lubrication systems are subject to contamination due to

Million and Counting Aggreko Process Services (APS) has contributed more than $100 million to our customers’ bottom line. Aggreko’s process engineering staff provides innovative temporary solutions designed to address operational, environmental and turnaround issues cost-effectively. We help customers: UÊ «À ÛiÊ ÀÊ > Ì> Ê«À `ÕVÌ ÊÀ>ÌiÃ UÊ iiÌÊ«À `ÕVÌÊÃ«iV wV>Ì Ã UÊ,i`ÕViÊi Û À i Ì> ]Ê «iÀ>Ì > Ê> `ÊÃ>viÌÞÊÀ Ã Ã UÊ,i`ÕViÊ ÕÌ>}iÊ`ÕÀ>Ì Ê> `Ê «>VÌÊon operations UÊ >«ÌÕÀiÊÜ ` ÜÃÊ vÊ «« ÀÌÕ ÌÞ

Contact APS today to learn more or to put our expertise to work for you.

a variety of causes—internal component wear generates particles washed into the oil stream; foreign particles find their way into the system during assembly, maintenance and everyday operation. These contaminants, if uncontrolled, will cause wear and even failure of bearings or gear elements. Lubricant cleanliness is a major concern when looking to maximize geared equipment service life. The lubrication filters play a key role in ensuring that abrasive particles are removed from the system. In addition to the filtration of fluids, the lubrication filters often incorporate a bypass for clogged element conditions, a magnetic drain plug to collect metallic particles and a visual and/or electrical cleanliness indicator. In force-feed lubrication systems, the oil must be supplied through a filter media that is compatible with the lubricant, meets the viscosity requirements without excessive pressure drop and removes particulate matter consistent with the rotating equipment design. The oil filter should be located on the pressure side of the pump so warmer, less viscous oil is being filtered. Filter elements

I AUGUST 2009 HYDROCARBON PROCESSING

Select 155 at www.HydrocarbonProcessing.com/RS

The A-Team Advantage – Comprehensive Services for Enhanced Processes: UÊ" Ã ÌiÊi } iiÀ }ÊÃÌÕ` iÃ UÊ ÃÌ VÌ ÛiÊÀiw iÀÞ }À>`iÊiµÕ « i Ì UÊ*À ViÃÃÊ iÃ } Ê*>V >}iÊ* *® UÊ Û>Ì ÛiÊ> ÌiÀ >Ì ÛiÃÊÌ Ê conventional methods UÊ,i >L ÌÞÊ> `Ê«Ài` VÌ Ê vÊÃÌÀi> Ê factor performance UÊ ÛiÀ>}iÊ vÊÎäÊÞi>ÀÃ½ÊiÝ«iÀ i ViÊv ÀÊ process engineers and project managers UÊ VÕ i Ì>Ì Ê> `Ê>vÌiÀ >VÌ ÊÀi« ÀÌÃ UÊ ÃÌ ÀÞÊ vÊ ÀiÊÌ > ÊÎxäÊ«À iVÌÃÊÌ Ê`>Ìi UÊ ÊV>« Ì> ÊiÝ«i ` ÌÕÀiÃ

www.aggreko.com/northamerica

800.348.8370

FLUID FLOW AND ROTATING EQUIPMENT can be either cleanable and reusable or disposable. Cleanable filter elements are usually made of wire mesh, with cleaning commonly accomplished using an ultrasonic liquid bath. Coolers. In force-feed lubrication systems, the oil inlet temperature to the gearbox is controlled by passing hot sump oil through a cooler. Such a cooler must be capable of achieving the required oil temperature drop when exposed to the maximum ambient air temperature anticipated for the application. However, a generous temperature margin should be applied during design to account for cooler deterioration. The two types of coolers used are liquidto-liquid and liquid-to-air. In oil-to-water (liquid-to-liquid) coolers, hot oil gives up heat to cooler water, resulting in cooler oil and hotter water. Where water is unavailable, radiators are used to blow cooling air over oil tubes. However, air-to-oil (airto-liquid) coolers require larger envelopes than oil-to-water coolers; in addition, hot days will limit the amount of cooling a radiator can achieve. Oil reservoir. The oil reservoir may be integral with the gearbox or separately mounted and connected to the gearbox by piping. The oil level in the reservoir will vary from a maximum when the unit is shutdown and oil has drained from lines and components to the minimum permitted during operation. At shutdown, when lines and components such as coolers and filters drain back into the reservoir, the oil level will be higher than the maximum operating level; thus, the reservoir tank must have sufficient volume to accommodate the drain backflow and still retain some air space at the top. To ensure complete draining for cleaning and oil changes, the unit should be fitted with a drain connection located at or near the bottom of the sump. The oil pump suction line should be located slightly above this reservoir bottom so that any sediment on the bottom is not pulled into the pump suction line. Oil return lines should be piped into the reservoir near the maximum operating level away from the area around the pump suction connection so that the incoming oil must travel the maximum distance to the pump suction. By maximizing this dwell time, the oil has more time to lose any entrained air before it is again circulated through pumps, filters and coolers. To facilitate reservoir inspection and cleaning, sufficiently large access openings must be provided.

Breather. The gearbox breather is used to vent pressure that may be built up in the gearing unitâ&#x20AC;&#x201D;such pressure may result from air entering the lubrication system through seals or the natural heating and cooling of the unit. When a cold gearbox starts up, the heat generated during operation will cause air pressure to build within the gearbox housing. Piping. The lubrication system piping is intended to distribute lubrication oil in accordance with system design requirements and should be as simple and direct as practical. The piping connections for a gearbox can cause problems at assembly and startup since often the responsibility for supplying piping and lubrication system components is split between the gear manufacturer and user. If this is the case, care must be taken to avoid piping complications at installation. It is good practice to have only one external oil feed connection with all other oil passages placed inside the system casingâ&#x20AC;&#x201D;this means that any slight leaks in the piping connections will be internal and harmless. This also means there will be less chance of damage during installation. In all, the piping arrangement must be carefully designed to minimize pressure drops and leak sources. Lubrication monitoring. In provid-

ing reliable service, the lubrication system must incorporate sufficient sensors to allow continuous and complete system condition monitoring. Resistance temperature detectors (RTDs). RTDs allow continuous temperature monitoring at key locations, such as oil supply and drain temperatures, as well as sump temperature. RTDs should be of the duplex type to obtain redundant readings for accuracy or to supply a backup. Temperature switches. Temperature switches are typically used as a trigger to alarm or shut down the gearbox prime mover when excessive temperatures are experienced, and can be permissives for cold temperature startup. These switches are typically located in the main oil reservoir and lube-oil supply lines. Pressure switches. Pressure switches are used as a mechanism to trigger the operation of auxiliary back-up pumps, as well as to initiate a signal for system shutdown when pressure is lost at the main oil inlet to the gearbox. Flow switches. Flow switches are used as a trigger to signal loss of flow below Select 156 at www.HydrocarbonProcessing.com/RS

One Person, Zero Tools, Blinded in Minutes. Invest in Your Plant for the Long-Term Used in Chemical Plants and Refineries around the world for over 30 years, Onis Blinds can improve safety, minimize downtime, reduce emissions, and produce savings each time they are used. Replacing spectacle, slip, and figure-8 blinds,

Line Blind

Onis Blinds can be used in any process including nitrogen, H2S and decoke as well as on feed lines, furnaces and reactor isolation. Onis Blinds are ideal for emergency shut downs or quick isolation for approaching hurricanes. Onis Line Blind Product Range: Size: 1/2” up to 50” Pressure: Vacuum up to 5500Psig Temperature: -152° up to 1400°F

www.onislineblind.com

(713) 840-6377

Select 95 at www.HydrocarbonProcessing.com/RS

the minimum oil demand requirements of the system. Water detectors. Water detectors act as a means to detect the presence of unwanted water in the lubrication oil system. Lubrication system maintenance.

Given the integral role that the lubrication system plays in overall gearbox life and longevity, it must be continually maintained so that the system is functioning at peak performance. It is important to develop a systemic inspection method, condition verification and documentation to avoid any unexpected lubrication system failures, and ultimately equipment damage. The following are areas of concern for maintaining a properly functioning lubrication system. Cleanliness. Dust, dirt, grit and wear particles in the lubricant supply must be kept to a minimum. Filters and strainers should be serviced regularly to avoid circulating contaminants within the oil, as well as to avoid excessive pressure drops that can reduce the quantity of oil supplied to the gear drive. Lubricant condition. The service life of a lubricant is negatively affected by a number of factors, including high temperatures, water and/or emulsions, solid contaminants and operating environment. An oil sample should be drawn from the oil sump at scheduled intervals and analyzed by the lubricant supplier or a reputable maintenance provider. The lubricant supplier should be consulted for typical oil changeout limits for the particular oil used. Sensor/switch settings. An annual check of all switches and sensors should be performed to verify operation as per lubrication system schematic specified settings. System vibration and environmental conditions can alter settings, ultimately affecting critical timing and initiation of sensor functions. Auxiliary pump function. Pumps and other motorized accessories should be checked at scheduled intervals to verify operability, proper oil delivery, pressures and motor power draw. Relief valve settings should be checked to ensure that the required oil delivery is supplied to the gear drive at the proper pressure. Flow and pressure check. Flows and pressure drops at the cooler, filters and inlet to the rotating equipment should be routinely monitored and recorded to identify any adverse trends that might be developing. Cooler condition. An annual check

of cooler condition is important to maintain cooler efficiency. Water-cooled heat exchanger coolant ports should be checked for any fouling or blockage. All sacrificial anodes should be replaced. Air-oil cooler core fins should be checked and cleaned of any dirt buildup that would affect heat transfer efficiency. Breathers. Oil breathers should be checked frequently since they will become dirty. Any blockage in the breather could potentially lead to leakage elsewhere in the drive to relieve pressure. Visual component check. The entire lubrication system should be checked daily for all indicator gauge readings, pipe connections, vibration, bolted connections, oil leaks or seepage, loose accessories and wiring connections. Sound levels. The operating sound level of the pumps should be routinely noted. Any increase in sound level could indicate the presence of air in the lube system, blockage at the pump intake, air leaks in the pump shaft seal, worn or loose parts in the pump, filter blockage or high oil viscosity from the pumped fluid being too cold. Greased points. Some motors and pumps are equipped with greased bearings that must be lubricated at manufacturer recommended intervals. HP

BORSIG

FLUID FLOW AND ROTATING EQUIPMENT

BORSIG GROUP Leading Technology for Innovative Solutions Jules DeBaecke is vice president of engineering for Philadelphia Gear Corp. in King of Prussia. He is responsible for maintaining and enhancing Philadelphia Gearâ&#x20AC;&#x2122;s leadership role within the industry as world-class engineering experts. Mr. DeBaecke joined Philadelphia Gear in 1981 as engineering design manager for the Synchrotorque (hydroviscous clutch) and Marine Divisions, as well as manager of production engineering for all products. Since then, he has also held positions as product manager for the marketing and sales of marine, synchrotorque, test stand, material handling and high-speed product lines; and marketing manager. Prior to 1981, Mr. DeBaecke was employed by the Naval Ship Engineering Center, Philadelphia Division, where he was responsible for research, development, test and evaluation of US Naval main propulsion equipment, as well as fleet machinery maintenance worldwide for this equipment. At the project level, he was recognized as the US Navyâ&#x20AC;&#x2122;s clutch expert for equipment installed on naval vessels including gas turbine-driven cruisers and destroyers, nuclear submarines and aircraft carriers. Mr. DeBaecke holds a degree in mechanical engineering from Drexel University in Philadelphia, Pennsylvania.

Reciprocating and Centrifugal Compressors Pressure Vessels and Heat Exchangers Membrane Technology e.g. Emission Control Systems Boilers and Power Plant Technology Industrial Services For more information, please contact:

BORSIG GROUP Egellsstrasse 21 D-13507 Berlin/Germany Phone: +49 (30) 4301-01 Fax: +49 (30) 4301-2236 E-mail: info@borsig.de www.borsig.de Member of KNM Group Berhad Select 157 at www.HydrocarbonProcessing.com/RS

Select 73 at www.HydrocarbonProcessing.com/RS

FLUID FLOW AND ROTATING EQUIPMENT

SPECIALREPORT

Improve gas compression systems with all-welded shell-and-plate heat exchangers They offer benefits compared to shell-and-tube and welded-plate designs S. GAVELIN, Tranter International AB, Sweden

ultistage gas compression is one of the most common processes in any industrial plant. Applications vary from one industry to another and include natural gas processing, petroleum refining, and manufacturing of chemicals and end-product gases. Pressurizing the gas enables it to be stored and transported more easily. The use of shell-and-plate heat exchangers in place of conventional welded-plate heat exchangers and shelland-tube units provides a lot of benefits. With the shell-and-plate design you avoid the corner welds that are critical for the traditional â&#x20AC;&#x153;four-corner gridâ&#x20AC;? welded-plate design and you also save space and weight compared to the shell-and-tube units. The shell-and-plate design also offers full maintainability with an option for a removable plate-pack core via a bolted cover on one end of the shell that enables the unit to be mechanically cleaned. The potential for improved process reliability and efficiency that this offer is ever more widely appreciated.

Stage 1

Stage 2

Stage 3

Stage 4

Gas

Dryers Stage 5

FIG. 1

To recovery section

Typical gas compression process.

Interstage cooler design. In the design of the interstage cool-

ers in a gas compression process, it is necessary to have a good understanding of the physical laws governing fluid flow and heat transfer in corrugated channels. In the interstate coolers, partial condensation will occur when the gas is cooled between two stages. Since the pressure loss due to fluid friction contributes to decreased compressor efficiency, and thereby increased energy input to the compressor, it is essential that the designer is well aware of the aspects influencing the pressure loss. Several correlations exist on how to calculate twophase pressure drop but the sources in the literature cannot agree upon a common complete adequate theory to predict the two-phase flow behavior. Because of this the engineering contractors need to rely more on the applications expertise of the manufacturers who usually use in-house programs based on empirical or semiempirical correlations to choose a correct size of heat exchanger. Hence, an experienced designer can help the process engineers save energy consumption by proper cooler design. The shell-and-plate design makes it possible to meet the heat transfer requirements with a relatively low pressure drop that contributes to increased process performance. The gas compression process. Fig. 1 presents a typical flow diagram for the gas compression process. The gas is typically compressed in several stages working with equal pressure ratios and mechanical work input. Since the compressor efficiency is a function of the compression ratio single-stage compression is inefficient. The

gas enters the first compressor and is compressed to the first pressure level. When the pressure of the gas is increased the temperature will also increase by the definition of the ideal gas law. The hydrocarbon gas is then cooled in the interstage cooler and condensed liquids are separated from the gas stream in the free water knock-out drum. The separated gas enters a new compression stage and the process starts all over again. It is important to monitor the suction temperature at each stage since it must be kept away from the saturated liquid line to avoid liquid droplets in the compressor. Liquid is constantly removed from the gas prior to each compression stage. Depending on the gas dryness requirements further dehydration might be necessary before final storage or transportation. After the compression all the liquid has usually gone into solution with the vapor phase. If any subsequent corrosion cannot be tolerated the gas must be dehydratedâ&#x20AC;&#x201D;preferably with an absorption unit such as a TEG glycol process. In the glycol contactor the wet gas flows counter current with lean glycol, which absorbs the water vapor from the hydrocarbon gas by chemical affinity. The rich glycol is collected in the bottom of the contactor and flows to the regeneration skid where rich glycol is regenerated to lean glycol by boiling off the water vapor. The dry gas then flows from the contactor to the last compression stage. The gas is compressed to the appropriate discharge pressure and then cooled in a last cooler before being exported to storage or transportation. HYDROCARBON PROCESSING AUGUST 2009

I 35

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

Two-phase pressure drop. The two-phase pressure loss is a

very important factor for overall compressor efficiency (based on energy input vs. enthalpy increase) due to the partial condensation occurring when the gas is cooled. The two-phase pressure drop is a function of vapor quality and, since the quality varies with progression downstream, it is necessary to integrate any correlation that depends on quality over the heat exchanger length to compare with actual measurements conducted in experimental facilities. The most commonly used model for predicting the pressure loss in two-phase flows is the classical Chisholm and Lockhart–Martinelli correlations. In this model the pressure drop of the two-phase flow may be related to the pressure drop of the single-phase flow (as calculated by standard correlations) by a two-phase multiplier defined as:

l 2 =

( p / L)tp ( p / L)l

= 1+

C 1 + X LM X LM2

(1)

where XLM is the Lockhart–Martinelli parameter and C is the Chisholm parameter. The Lockhart–Martinelli parameter is defined as the ratio of liquid to vapor pressure gradients, with the assumption that each phase flows alone in the channel. The parameter takes different forms depending on the assumed or measured flow regime of each phase. The most commonly used assumption is that both phases are turbulent. There are insufficient experimental data for general conclusions to be made for the value of C. However, C seems to vary with hydraulic diameter and several suggestions exist for the C value. It should be noted that the prediction accuracy of any of the existing models is quite low. Since the correlations in the literature are only suitable for certain geometries at different mass fluxes, heat fluxes, and vapor qualities and flow orientations, the uncertainty is within the range of ± 20% and even more in special geometries. Hence, reliable manufacturing data in conjunction with a critical eye of the engineering contractor are essential in predicting the pressure loss. Optimizing the gas compression process. The work

input to the compressor is highly dependent on the pressure ratio between the compressor inlet and outlet and the inlet temperature on the suction side of the compressor. Optimizing a compressedgas system is hence not the easiest task. While each compressor may very well be 70–80% efficient this high efficiency cannot be achieved if one compares the energy input to the increased enthalpy across the total compressor because friction in piping and heat exchangers tend to consume this. The efficiency that can usually be achieved by taking into account all the extra frictional loss may be 65% for a single-stage unit with a 6% drop for every added stage. With five stages one could expect an efficiency to be around 35% (based on energy input vs. enthalpy increase). Since the frictional loss is such an important factor for the overall compression efficiency, the pressure drop across the interstage coolers should be kept as low as possible. Hence, reliable manufacturing data cannot be neglected and as described, it is not always an easy task to predict the two-phase pressure drop. Why shell-and-plates? The benefits of using plate heat

exchangers are well known. They are more efficient, occupy less space, are less heavy, and do not need to be cleaned as often as shell-and-tube heat exchangers. When the material is exotic (high cost) the price will also be less than for traditional shell-and-tube heat exchangers due to less required heat transfer area. Therefore, it will save costs where corrosive fluids are present. However, 36

I AUGUST 2009 HYDROCARBON PROCESSING

traditional welded plate heat exchangers with rectangular plates (four corner grid design) exhibit poor performance because of low resistance to thermal and pressure fatigue and are not ideal for the application. The corner welds very often result in crack propagation that results in unexpected shut down and maintenance. Therefore, this antiquated design should be replaced by more reliable shell-and-plate heat exchangers to improve reliability and availability in the hydrocarbon industry. The shell-and-plate heat exchangers introduced in recent years, have presented solutions to some of the traditional shell-and-tube/ four corners grid limitations. They provide the thermal efficiency and the compactness of a plate-and-frame heat exchanger while handling pressures and temperatures otherwise requiring shelland-tube units. Their excellent resistance to thermal and pressure fatigue makes them superior to other welded technologies. The shell-and-plate design has also proven to withstand challenging process conditions with liquids, gases, steam and two-phase mixtures. The compact design enables very close temperature approaches and the small hold-up volume provides fast startups and close following of process changes. From a thermal point of view the shell-and-plate design is very well suited for vapor/liquid mixtures. The short flow path and the large cross section makes it particularly suited for two-phase flows with a high LMTD between the hot and cold sides of the heat exchanger. Summary of advantages of shell-and-plate compared to other types of heat exchangers: • Less than half the size of a shell-and-tube unit for comparable duties as a result of higher heat transfer coefficient created by the corrugated patterns of their plates • Turbulent flow even at low velocities keeping the plates free from scaling and fouling longer than does the laminar flow seen in shell-and-tube units. • Easy to maintain due to the option for removable plate-pack core via a bolted cover on one end of the shell in combination with a traditional CIP unit. • Superior to any other type of welded heat exchanger due to the lack of corner welds which are critical in other “four-corner grid” designs. Conclusion. Plate technology has been used for more than a

decade in main hydrocarbon processes. Gas compression operations are now beginning to enjoy the advantages of all-welded shell-and-plate heat exchangers. The shell-and-plate concept is superior to a four-corner grid design since the resistance to thermal and pressure fatigue is very much improved, especially for dynamic processes where close following of process changes is needed. Circular shell-and-plate designs eliminate the corner welds that are most susceptible to problems. HP

Stefan Gavelin is the area sales manager (Europe) for Tranter’s oil & gas market segment. He holds an M.Sc. degree in mechanical engineering from the Royal Institute of Technology, Sweden and has specialized in thermodynamics and heat transfer. At Tranter he has the task of researching and finding new application areas, as well as developing sales concepts and initiating product development according to the new applications. Mr. Gavelin also has experience from the refrigeration industry especially in the design of evaporators and condensers. As a member of the HTFS organization, Tranter is constantly aware of the latest developments in two-phase flow theory.

You Get More Than Just a Process Gas Compressor Lubricated up to 1’000 bar, non-lubricated up to 300 bar For longest running time: We recommend our own designed, in-house engineered compressor valves and key components Designed for easy maintenance We are the competent partner with the full range of services – worldwide

Your Beneﬁt: Lowest Life Cycle Costs

More beneﬁts: www.recip.com/api618

Select 55 at www.HydrocarbonProcessing.com/RS

Select 99 at www.HydrocarbonProcessing.com/RS

FLUID FLOW AND ROTATING EQUIPMENT

SPECIALREPORT

Consider ceramic bearings for screw compressors This retrofit resulted in a six-fold increase in service life H. P. BLOCH, HP staff

il-flooded screw compressors with bearings flooded by contaminated oil can be problematic. Retrofitting an oil-flooded screw compressor subjected to severe sour gas (30-40 mol%) and acid gas (35 mol%) conditions with new rolling bearing technology resulted in a six-fold increase in service life. Ceramic bearings were employed in the retrofit; it demonstrates important opportunities to equip oil-flooded screw compressors with advanced ceramic rolling-element bearings. In essence, attractive reliability improvements exist for “wet screw” compressors that process, gather and extract hydrocarbons. The bearing retrofit at the Syncrude oil sands facility can boast the longest-running oil-flooded screw compressor installation with rolling-element bearings under clearly difficult conditions. The new bearings are made of materials specifically selected for severe sour gas use: rings of super-tough stainless steel and rolling elements made of bearing-grade silicon nitride ceramic. Located within the Syncrude upgrading plant in Fort MacMurray, Alberta, Canada, the compressor handles vacuum off-gas from the world’s largest vacuum tower in a process designed to avoid flaring H2S-containing “sour” gas (Fig. 1). The oil-flooded Flare => SO2

machine is critically important to this Syncrude facility. It is nonspared and in continuous operation. The twin-screw compressor incorporates a rather typical bearing arrangement (Fig. 2) with four cylindrical roller bearings absorbing pure radial loads and two four-point angular-contact ball bearings taking the axial loads. The machine is direct-driven by an induction motor at 3,600 rpm and polyglycol oil with a viscosity of 40 cSt at +212°F (+100°C) is used. This synthesized hydrocarbon oil lubricates the bearings and also seals and cools the gas compression space. After being cooled and filtered to nominally 5μ, the oil is divided into separate streams; one is injected into the bearings at +176°F/+80°C, the other enters the compression space. Because it contacts a sour process gas stream and is part of a circulating oil system, the lubricant becomes progressively more contaminated by H2S, brine, light hydrocarbons and other matter. This contamination escalates the risk of overall and stress corrosion, poor lubrication and clogged filters. At times, operating oil viscosities have decayed from initially 37 to only about 3 cSt, again jeopardizing both bearing life and reliability. Ceramic bearing materials. The new and eminently successful rolling elements are made from bearing-grade silicon nitride

15 psia (1.0 bar) Suction scrubber

Gas cooler

Screws

Bearings

Gas out Mixed lube oil and gas 150 psia (10.3 bar)

Oil separator

Oil ﬁlter

Vacuum off-gas H2S 30-40 mol% CO2 35 mol% H2O fully saturated Mole weight = 36

Lube oil supply

Drain

Overhead seal drum

Oil cooler

Sour water stripper FIG. 1

Drain

Schematic of process flow around oil-flooded screw compressors at Syncude (Alberta, Canada).

FIG. 2

Schematic of the oil-flooded twin-screw compressor design and its all-ceramic rolling bearing arrangement. Cylindrical roller bearings absorb radial loads; four-point angular contact ball bearings carry the thrust loads of each rotor. HYDROCARBON PROCESSING AUGUST 2009

I 39

ai_BHS_1.5_en aio

Industry | Marine | Rail | Road

Voith Turbo BHS Getriebe â&#x20AC;&#x201C; Benchmark for Turbo Gears High Speed Epicyclic Gear Units N Experience of > 7,000 delivered units and 75 years N Up to 45 MW or 80,000 rpm N Various casing, toothing, mounting and design options High Speed Parallel Shaft Gear Units N Experience of > 7,000 delivered units and 55 years N Up to 80 MW or 60,000 rpm, gear ratio up to i = 28 N Casing material options N API, AGMA, ISO, DIN compliance High Speed Integral Gear Units N Experience of > 1,000 delivered units and 35 years N Up to 10 impellers N Up to 40 MW or 60,000 rpm, gear ratio up to i = 22

Visit us at Turbo Show Houston, Sept 14-17 2009, Booth # 517/519

High Speed TWINTORS-Diaphragm Couplings N Experience of > 1,000 delivered units and 20 years N Up to 80,000 rpm N Outstanding performance characteristics N API 671, ISO 10441 and 14691, ATEX compliance Rotor Turning Gear Units N Experience of > 1,000 delivered units and 45 years N Design for shaft train integration or available free shaft end N Easy installation and free of maintenance N DIN, ISO compliance Germany Hans-BĂśckler-Strasse 7 87527 Sonthofen, Germany Fon +49 83 21 802 -0 Fax +49 83 21 802 - 685 www.bhs-getriebe.de sales.bhs@voith.com

Select 90 at www.HydrocarbonProcessing.com/RS

USA 4357 Ferguson Drive Cincinnati, Ohio 45245, U.S.A. Fon +1 513 797 - 8100 Fax +1 513 797 - 8103 www.bhsgear-inc.com info@bhsgear-inc.com

FLUID FLOW AND ROTATING EQUIPMENT

commonly called “hybrid” bearings. The silicon nitride ceramic consists of fine elongated grains of beta-silicon nitride in a glassy phase matrix. It provides a combination of favorable properties for rolling elements within bearings such as high hardness, low density, low thermal expansion, high electric resistivity, low dielectric constant and no susceptibility to magnetic fields.

Weibull probability plot included lives: L10 L50 l conf. interval: 90%-2 sided 37-1K-3 compressor bearings long 37-1K-3 compressor bearings short 37-1K-3 compressor bearings

F, %

99 95 90 80 70 60 50 40 30 20 10 5 2 1

2 FIG. 3

3 4 5

103

2 Life, hr

3 4 5

104

Weibull probability plot of past compressor failures, including service period without bearing failure of retrofitted compressor (i.e., equipment change due to slide valve failure). Blue line—summer periods; red line— winter periods; green line—all seasons.

ceramic. Suitable ceramics greatly resist corrosion and impart high wear and surface fatigue resistance to bearings under poor lubrication conditions. Bearings that combine hard stainless bearing-grade steel and bearing-grade silicon nitride rolling elements (Si3N4) are BEGEMANN PUMPSTM

Centrifugal Process Pumps

SPECIALREPORT

Equipment life experience. A statistical life analysis was done based on the actual operating history of this compressor installation. The results are summarized in Fig. 3, plotting the compressor as previously equipped with standard bearings (AISI 52100 steel bearings and an MTBF of 3,800 hours) and the compressor with ceramic retrofit bearings (note that a replacement compressor was installed after 23,300 hours due to a slide valve failure). Interestingly, shorter lives were obtained with conventional AISI 52100 steel bearings during summer operation. In warm weather it is more difficult to keep discharge temperatures sufficiently above the process gas dew point; an increased volume of water vapor condenses as a result. HP

Heinz P. Bloch is HP’s Reliability/Equipment Editor. A practicing engineer and ASME life fellow with close to 50 years of industrial experience, he advises process plants on maintenance cost-reduction and reliability upgrade issues. His 16th and 17th textbooks on reliability improvement subjects were published in 2006 and 2009. An excerpt was taken for this article from Bloch-Geitner, Maximizing Machinery Uptime, pp. 201–228 (Gulf Publishing, ISBN 10:0-7506-7725-2).

Excellent Minerals Solutions

Weir Minerals France

Performance Not promises

Precision Engineered for API 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 Tel: +33 472 817 272 - Fax: +33 472 817 643 www.weirminerals.com

Select 158 at www.HydrocarbonProcessing.com/RS

Select 159 at www.HydrocarbonProcessing.com/RS

PROCESS INSIGHT Selecting the Best Solvent for Gas Treating Selecting the best amine/solvent for gas treating is not a trivial task. There are a number of amines available to remove contaminants such as CO2, H2S and organic sulfur compounds from sour gas streams. The most commonly used amines are methanolamine (MEA), diethanolamine (DEA), and methyldiethanolamine (MDEA). Other amines include diglycolamine® (DGA), diisopropanolamine (DIPA), and triethanolamine (TEA). Mixtures of amines can also be used to customize or optimize the acid gas recovery. Temperature, pressure, sour gas composition, and purity requirements for the treated gas must all be considered when choosing the most appropriate amine for a given application.

Tertiary Amines A tertiary amine such as MDEA is often used to selectively remove H2S, especially for cases with a high CO2 to H2S ratio in the sour gas. One beneﬁt of selective absorption of H2S is a Claus feed rich in H2S. MDEA can remove H2S to 4 ppm while maintaining 2% or less CO2 in the treated gas using relatively less energy for regeneration than that for DEA. Higher weight percent amine and less CO2 absorbed results in lower circulation rates as well. Typical solution strengths are 40-50 weight % with a maximum rich loading of 0.55 mole/mole. Because MDEA is not prone to degradation, corrosion is low and a reclaimer is unnecessary. Operating pressure can range from atmospheric, typical of tail gas treating units, to over 1,000 psia.

Mixed Solvents In certain situations, the solvent can be “customized” to optimize the sweetening process. For example, adding a primary or secondary amine to MDEA can increase the rate of CO2 absorption without compromising the advantages of MDEA. Another less obvious application is adding MDEA to an existing DEA unit to increase the effective weight % amine to absorb more acid gas without increasing circulation rate or reboiler duty. Many plants utilize a mixture of amine with physical solvents. SULFINOL® is a licensed product from Shell Oil Products that combines an amine with a physical solvent. Advantages of this solvent are increased mercaptan pickup, lower regeneration energy, and selectivity to H2S.

Primary Amines The primary amine MEA removes both CO2 and H2S from sour gas and is effective at low pressure. Depending on the conditions, MEA can remove H2S to less than 4 ppmv while removing CO2 to less than 100 ppmv. MEA systems generally require a reclaimer to remove degraded products from circulation. Typical solution strength ranges from 10 to 20 weight % with a maximum rich loading of 0.35 mole acid gas/mole MEA. DGA® is another primary amine that removes CO2, H2S, COS, and mercaptans. Typical solution strengths are 50-60 weight %, which result in lower circulation rates and less energy required for stripping as compared with MEA. DGA also requires reclaiming to remove the degradation products.

Secondary Amines The secondary amine DEA removes both CO2 and H2S but generally requires higher pressure than MEA to meet overhead speciﬁcations. Because DEA is a weaker amine than MEA, it requires less energy for stripping. Typical solution strength ranges from 25 to 35 weight % with a maximum rich loading of 0.35 mole/mole. DIPA is a secondary amine that exhibits some selectivity for H2S although it is not as pronounced as for tertiary amines. DIPA also removes COS. Solutions are low in corrosion and require relatively low energy for regeneration. The most common applications for DIPA are in the ADIP® and SULFINOL® processes.

BR&E

Choosing the Best Alternative Given the wide variety of gas treating options, a process simulator that can accurately predict sweetening results is a necessity when attempting to determine the best option. ProMax® has been proven to accurately predict results for numerous process schemes. Additionally, ProMax can utilize a scenario tool to perform feasibility studies. The scenario tool may be used to systematically vary selected parameters in an effort to determine the optimum operating conditions and the appropriate solvent. These studies can determine rich loading, reboiler duty, acid gas content of the sweet gas, amine losses, required circulation rate, type of amine or physical solvent, weight percent of amine, and other parameters. ProMax can model virtually any flow process or configuration including multiple columns, liquid hydrocarbon treating, and split flow processes. In addition, ProMax can accurately model caustic treating applications as well as physical solvent sweetening with solvents such as Coastal AGR®, methanol, and NMP. For more information about ProMax and its ability to determine the appropriate solvent for a given set of conditions, contact Bryan Research & Engineering.

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

FLUID FLOW AND ROTATING EQUIPMENT

SPECIALREPORT

A retrofit of this system at a refining company provided a payout in less than a year Improved stepless capacity regulation for reciprocating compressors J. JIN, W. HONG,* Institute of Process Equipment, Zhejiang University, Hangzhou, China; and L. TANG, Sinopec Zhenhai Refining & Chemical Company, Ningbo, China

ecause of the conflict between the fixed capacity and the varying operating conditions in process units, it is extremely important to implement stepless capacity regulation of reciprocating compressors to meet the demands of a changing industry. This article presents a new regulation facility to achieve this aim. in the gas and refining industries to compress light gases such as hydrogen, ammonia and methane for their wide adaptability and high efficiency. But regulating capacity efficiently for this type compressor is a long-term problem. Three ways to provide capacity regulation for reciprocating compressors are:1 • Bypass: The energy used to compress the exceeded gas is wasted in the bypass throttling valve, and the consumed energy at the different capacity is the same. • Speed control: Up to 50% capacity could be reduced in this way limited by motor efficiency, and it needs an unacceptable investment for large units. • Clearance pocket: By automatically or manually varying the additional clearance volume, the reciprocating compressor capacity can be steplessly changed from 100% to 40% (or much less). But the clearance pocket will be large in the lower pressure ratio and this method is uneconomical. In many applications, more than one method can be combined to get more regulation quality for the field units. Furthermore, it is difficult for the above principles to provide the sufficient control dynamic and capacity regulation accuracy. Depressing the suction valves in the partial compression stroke to realize capacity regulation has the advantages of being full range, stepless and saving energy. The unnecessary gas flows back to the suction chamber; the quantity of gas flowing back is decided by the depressing time. In this way, only the required gas is compressed with a small amount of energy wasted in flowing back the unnecessary gas through the suction valve. As Fig. 1 shows, the dashed area is the energy savings in one cycle. Up to now, this is the most effective method of capacity regulation for reciprocating compressors. This article presents a new capacity regulation facility that is based on the same method but has a dif* Corresponding author

Energy saved Pressure

Introduction. Reciprocating compressors are still widely used

3 2 Volume

FIG. 1

Typical P–V diagram in stepless capacity regulation.

ferent actuator and control method. This regulation system has the advantage of competitive cost. Hydraulic system. As the key part of the regulation system, the hydraulic distributor produces the “controllable” oil pressure impulse wave.2 That means the periodic time and pressure acting time in one cycle of the hydraulic pressure wave are under control. Fig. 2 shows the assembly sketch of the hydraulic distributor3 which is different from the one in reference 2. There are three individual parts in the whole machine: adjusting motor, synchronous motor and hydraulic distributor. The hydraulic distributor can also be divided into three parts: outer sleeve, middle sleeve and rotor. The outer sleeve is fixed by supports. The middle sleeve can move in the axial direction and does not rotate. The central rotation axis rotates synchronously with the compressor motor to produce the same periodic hydraulic pressure wave. Movement of the middle sleeve is controlled by adjusting the servo motor through the guide screw, and the rotor rotation is driven by the synchronous servo motor. HYDROCARBON PROCESSING AUGUST 2009

I 43

Outer sleeve

Adjusting servo motor FIG. 2

Middle sleeve

FLUID FLOW AND ROTATING EQUIPMENT Hydraulic pressure, bar

SPECIALREPORT

Oil Pressure Release area area inlet

Central rotation axis

Oil outlet

Oil return port

Synchronous servo motor

80 60 40 20 0 -10 0

FIG. 4

100

150 200 250 300 350 Sample points (sample rate: 1K)

400

250

499

Typical diagram of the hydraulic impulse wave.

Assembly sketch of hydraulic distributor. Oil Inlet and outlet Unloader hydraulic piston

Inlet channel

Return channel

Connecting joint

Pressure Release area area

Lifting spring Suction valve

FIG. 3

Torus sketch in pressure area. FIG. 5

Fig. 3 shows the typical torus shape and inlet and outlet oil channels of the rotor. The shape of the pressure area on the rotor periphery is a triangle. Oil pressure enters through the hollow rotor to a pressure area. When this pressure communicates with a port in the middle sleeve, oil is delivered through the outer sleeve to the unloaders. Pressure remains on the unloaders until the rotor rotates to a point where the port in the middle sleeve communicates with a release area on the rotor periphery. Oil then escapes through the outlet oil channel and then back to the hydraulic units. The actual pressure duration on the unloaders depends upon the longitudinal position of the ports in the middle sleeve with respect to the pressure area on the rotor periphery. As the middle sleeve is positioned from right to left, the unloading time is decreased and capacity is increased. There are two sets of pressure and release areas along the torus of the rotor that balance the hydraulic force in the radial direction. Correspondingly, the middle sleeve has two ports. Therefore, compared to compressor motor speed, half rotational speed of the hydraulic distributor is needed to keep synchronous rotation between them. That is to say that one hydraulic distributor can deliver pressure to two sets of unloaders in the double-acting cylinder. The typical pressure impulse wave diagram recorded in the experimental test is illustrated in Fig. 4. 44

Unloader plunger

I AUGUST 2009 HYDROCARBON PROCESSING

Typical sketch of the unloader.

Suction valve reliability and valve plate fatigue. The operating conditions of the reciprocating compressor suction ring valve will be changed greatly during stepless capacity regulation. Inappropriate design will greatly reduce the suction valves’ lift time. According to the research results from the 50–100% range regulation system that used the pneumatic diaphragm actuator to hold the suction valve open, most of the suction valves failed after operating more than one month because of the additional unloader. This conclusion is different from reference 1 that says that adding the pneumatic diaphragm actuator greatly improves the suction valves’ lifetime. As reference 4 (published by author) shows, the hydraulic pressure to unload the suction valve will decide the maximum stress in the suction valve during impact between the unloader and suction valve plate. The inclination of the unloader and valve plate causes the serious stress concentration, and the unloader structure and material also influence the stress distribution of the suction valve plate. So, the assembly error between the unloader plunger and suction valve plate and hydraulic pressure in the unloader should be controlled carefully to guarantee enough suction valve life. Unloader design. To make full use of the original compressor’s part, a new kind of unloader was designed (Fig. 5). The component parts of the unloader include an inlet or outlet oil

Do you see a ladybug? We also see an opportunity for industry to benefit from protecting the environment. Managers within the industrial sector face the never-ending issue of balancing business performance and environmental compliance goals. As the world reference in environmental services, Veolia Environnement helps industry achieve sustainable objectives through our four core service areas: water services and technologies, waste services, energy and facility management, and transportation and logistics.

The environment is our universal challenge.

Select 94 at www.HydrocarbonProcessing.com/RS

veolianorthamerica.com

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

I AUGUST 2009 HYDROCARBON PROCESSING

Pressure, MPa

joint, hydraulic piston, connecting rod, lift0.35 0.45 ing spring, unloader plunger and the rede3/h 0.40 Capacity, 150 m Capacity, 450 m3/h signed suction valve. Oil pressure enters 0.30 0.35 through the inlet oil joint and moves the 0.25 0.30 unloader piston and plunger downward 0.20 0.25 before the suction valve closes. The plunger 0.20 0.15 comes to rest on the valve channels and 0.15 prevents them from closing for a period of 0.10 0.10 time. When hydraulic pressure is released, 0.05 0.05 the pressure drops and the lifting spring 0.00 0.00 drives the plunger and piston upward, -0.05 in the meantime the compression stroke -0.05 -0.10 begins in the cylinder. -0.15 -0.10 To enable retrofit to existing compres0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 -3 3 Volume: x10 m Volume: x10-3 m3 sors, the connecting joint in Fig. 5 should be designed correctly to fit well with the suction FIG. 6 Pâ&#x20AC;&#x201C;V indicator diagram under different reciprocating capacities. valve cover. The hydraulic piston diameter and hydraulic pressure should be chosen synthetically to supply enough force to keep the sealing element open. Normally, a largerTDC sensor diameter piston and lower hydraulic pressure are good choices because high hydraulic Compressor motor pressure will cause pipe pulsation. Pipe system design and installation. Second-stage cylinder First-stage cylinder There are two kinds of pipes: oil supply and oil return. For oil supply pipe, a non-continuous oil supply causes a hydraulic impulse in the pipe. Ways to minimize the pipe vibration should be investigated. One method is to install hydro-pneumatic accumulators that are capable of attenuating hydraulic impulses in the oil inlets of every hydraulic distributor. In addition, high-pressure hydraulic hose in two ends of the pipe could be employed to connect the pipe with the distributor or unloader, and a hard metal Second-stage hydraulic distributor First-stage hydraulic distributor pipe could be located in the middle of the pipe line. The metal pipe is easy to fix with Hydraulic Control a rigid support. unit box Since there will be a period of time when the hydraulic oil flows from the hydraulic FIG. 7 Schematic diagram of the stepless capacity regulation system. distributor to the unloader, the rising pressure in the unloader hydraulic piston lags behind the oil outlet of hydraulic distributor. Hydraulic hose will make this lag longer, so the hard metal the compressor interstage pressure, the regulation system (espepipe needs to reduce this lag as much as possible to get more cially for the first stage) has to always be working. But regulating accurate control in the capacity regulation system. the quantity and response time is different among all stages. We should implement proper control parameters for every stage, such Control strategy and control system. The key problem as the PID parameters. to realize capacity regulation is to keep the hydraulic distributor It is easy to implement a compressor capacity control algoand compressor motor synchronized. To achieve this goal, a PLC rithm in the process distributed control system (DCS). The using a PID feedback strategy is used to control the hydraulic control signal for this regulation system could be the same as the distributor rotation that uses the crankshaft encoder and TDC original control signal in the bypass capacity regulation system. sensor as the original control signal. Lab tests show this method One of the interstage pressures, output pressure or compressor is effective. There is a pressure acting time delay for the unloader flowrate could be chosen as the original control signal. relative to the hydraulic distributorâ&#x20AC;&#x2122;s oil outlet, the rotation phase advance for the hydraulic distributor relative to the compressor Prototype machine and industry realization. A laboramotor is needed to ensure the the suction valves are unloaded tory prototype machine was built, and a reciprocating compressor before TDC. with a 640 m3/h design capacity was used to demonstrate the Generally the suction gas condition changes frequently in the technical feasibility of the capacity regulation system. Laboraprocess. For the sake of keeping the needed capacity and balancing tory test results have shown the effectiveness of the system. The

FLUID FLOW AND ROTATING EQUIPMENT compressor capacity varies from 50 to 640 m3/h. Fig. 6 shows the typical P–V diagram of the compressor with different capacities. After success in the laboratory test, the first industry unit has been installed in the disproportionation alkyl transfer equipment of Sinopec Zhenhai Refining and Chemical Company in China. A two-stage, two-cylinder double-acting compressor pumping hydrogen was fitted with the new capacity regulation system. The H2 is compressed from 12 bars to 37 bars and fed to the hydrogenation plant. The compressor has a power consumption of 600 kW/h and its design gas flow is 14,500 m3/h. The regulation system is able to match the demand of 11,000 m3/h, and 120 kW/h compressor electric power can be saved. According to this situation, its payback period of investment was within one year in this unit, and the compressor operation, which includes startup, shutdown and capacity regulation, is also more stable. The system was put into industrial application in October 2008 and since then has been in full operation. Fig. 7 shows the schematic diagram of the stepless capacity regulation system for the two-stage reciprocating compressor. It replaced the bypass control and it was easy to realize an undisturbed switch to the new system. This capacity regulation system has proven to be in full compliance with the refining company’s expectation in terms of competitive price, energy savings and control dynamics. Studies on system reliability and structure optimization should remain interesting topics for the further research. HP ACKNOWLEDGMENTS The authors acknowledge financial support from China National Key Technology R&D Program (No. 2008BAF34B13).

LITERATURE CITED Steinruck, P. and Ottitsch, F., “Better reciprocating compressor capacity control,” Hydrocarbon Processing, February 1997, pp. 79–84. Wither, Kenneth. H., “Infinitely variable capacity control,” Proceedings of the 1972 Purdue Compressor Technology Conference, 1972, pp. 47–51. Weirong, Hong and Jiangming, Jin, “A time based pressing off suction valves device for reciprocating compressor,” China invention patent, patent number 200610155395.8, 2006. Jiangming, Jin and Weirong, Hong, “Valve dynamic characteristic and stress analysis of reciprocating compressor under stepless capacity regulation,” Proceedings of the Fourth International Symposium on Fluid Machinery and Fluid Engineering, 2008, p415–420.

Jiangming Jin is a PhD student in the Institute of Process Equipment of Zhejiang University. He was a main participant in developing the reciprocating compressor capacity regulation system. Mr. Jin’s interesting research areas are computational fluid dynamics, ring valve dynamic response and condition monitoring for reciprocating compressors. He holds a BSc degree from the China University of Petroleum. Select 160 at www.HydrocarbonProcessing.com/RS 䉴

SPECIALREPORT

Weirong Hong is an associate professor in the Institute of Process Equipment of Zhejiang University. He directs the research group in reciprocating compressor capacity regulation systems. Dr. Hong’s responsibilities cover rotating and reciprocating compressor structure strength, rotor dynamics and fault diagnosis. He holds BSc and PhD degrees from the Zhejiang University, China.

Linbo Tang is an operations manager of Sino-Pec Zhenhai Refining & Chemical Company. He has 13 years of management and operations experience in synthetic ammonia and PX units. Mr. Tang earned an MS degree in 1996 from Liaoning Shihua University with the major of chemical machinery. www.woodgroup-esp.com

Low tolerance for pump problems? No problem.

Pressure to lower ever-increasing maintenance costs and reduce environmental impact has paved the way to better surface pumping solutions. Our multi-stage centrifugal SPS™ Surface Pumping Systems provide versatile, low-maintenance alternatives to many split-case centrifugal, positive-displacement and vertical-turbine pump applications. The SPS is a cost-effective solution for processing, petroleum, mining, water and other industries that require high-pressure movement of fluids. Proven benefits include: UÊ ÜiÀÊ Ì > Ê> `ÊÜ i viÊV ÃÌ UÊ- ÀÌÊV ÃÌÀÕVÌ Ê i>` Ì i UÊ VÀi>Ãi`ÊÀi >L ÌÞÊ> `ÊÀÕ Ì i UÊ ÜÊ ÃiÊ> `ÊÛ LÀ>Ì Ê iÛi Ã UÊ,i ÌiÊ Ì À }Ê> `ÊV ÌÀ UÊ7 À `Ü `iÊÃÕ«« ÀÌ° Call this number: 281.492.5160. Or e-mail sps @woodgroup.com.

Wood Group Surface Pumps Manufactured by Wood Group ESP, Inc.

SPS APPLICATIONS:

Booster Transfer 䊳 Circulation 䊳 Injection 䊳䊳

SP1000: rated 50 to 1000 HP (SP2000: rated 1000 to 2500 HP also available)

“Advertisement”

ReactoGard® V

The Ultimate Protection for Fixed Bed Catalytic Reactors from Eaton.

Using refinery-proven AccuFlux™ media, ReactoGard V automated backwashing systems offer significant advantages in all types of refinery feedstocks. ReactoGard V systems are engineered to operate cleaner, more efficient low flux sizing. This is achieved through improved media materials and AccuFlux element designs that increase filtration area by as much as 300% over previous standards. In addition, a new multi-bank system and controls approach eliminates the headers and reduces the number of valves by up to 85%. The ReactoGard V provides: •

Improved filter productivity

•

Complete media cleaning

•

Superior mechanical integrity for enhanced serviceability and reduced maintenance

•

Reduced product loss through less frequent backwashing

•

Reduced cost of installed filtration capacity

•

Reduced space requirements

ClearAmineTM

A Safe, Green and Economical Solution from Eaton for the Filtration of Amine Gasses.

With the ClearAmine™ automated self-cleaning system there are no bags or cartridges to purchase, no material to inventory and track, no hazardous waste landfill fees and greatly reduced maintenance and handling costs. Over a ten year period, you can expect nearly a 60% savings compared to a disposable media system. Plus, with no cartridges or bags to change out, employee exposure to hydrogen sulfide is minimized or eliminated. Amine filtration is a key factor for trouble-free sulfur removal. Pipe scale, corrosion, iron sulfide, sand and other solids build-up in closed-loop amine systems. They foul flash drums, heat exchangers, strippers, re-boilers and carbon filters and also attract hydrocarbons – which cause system foaming. ClearAmine™ provides an efficient method of removing harmful solids to protect your investment, extend equipment life, and maximize your amine system’s effectiveness. For more information, call Eaton at 888-309-6648. Select 119 at www.HydrocarbonProcessing.com/RS

Your complete source for

Oil & Gas Liquid Filtration • • • •

For more information, call Eaton at 888-309-6648.

ReactoGard® V ClearAmineTM Strainers Bags & Housings

Select 118 at www.HydrocarbonProcessing.com/RS

Model 2596 Automatic SelfCleaning Strainer

This automatic self-cleaning strainer is designed for the continuous removal of entrained solids from liquids in pipeline and process water systems. With an auto control system monitoring the strainer operation, cleaning is accomplished by an integral backwash system. A small portion of the screen element is isolated and cleaned by reverse flow. The remaining screen area continues to strain – providing an uninterrupted process. Also included is the idLTM shaft seal that positively prevents leakage from the backwash shaft at the top of the strainer. Eaton automatic strainers are ideal replacements for both simplex and duplex manual strainers because of their reliability and low maintenance.

Sustainable. Economical. Environmental. Eaton provides complete refinery and petrochemical liquid filtration solutions with installations in thousands of applications worldwide. We stand behind our products with comprehensive parts and service programs, plus innovative purchase/lease options. Count on Eaton to provide complete solutions that are sustainable economically and environmentally while protecting the safety of your workforce. At almost any point in your refinery, effective, efficient filtration can deliver tremendous benefits including:

Longer equipment life Reduced downtime Improved safety

Protection of costly process materials Extended runtime between turnarounds Enhanced environmental stewardship

For more information, visit www.filtration.eaton.com. Or, call 888-309-6648 to speak with an Eaton Oil & Gas Specialist.

For more information, call Eaton at 888-309-6648. Select 117 at www.HydrocarbonProcessing.com/RS

Select 116 at www.HydrocarbonProcessing.com/RS

FLUID FLOW AND ROTATING EQUIPMENT

SPECIALREPORT

Predicting liquid hold-up in horizontal stratified two-phase flow A new model has an average error of only 1% H. FIROOZFAR, N. KASIRI and M. H. KHANOF, Iran University of Science & Technology, Narmak, Tehran, Iran

his article provides a combined theoretical and experimental investigation into predicting hold-up in stratified two-phase flow in a horizontal pipe. The test section, 3m long, with an inside diameter of 30 mm was made of plexiglass to permit visual observation of the flow patterns. The experiments were carried out under various air and water flowrates in the stratified flow pattern. The liquid hold-up was measured by isolating the test section with two quick-closing valves. In the theoretical study, the Hamersma– Hart friction factor correlation1 was reformulated and incorporated to the Wongwises model2 to improve the prediction accuracy of horizontal stratified two-phase flow. This model was compared with experimental data. The comparison shows that the new model is more accurate than theWongwises mechanistic model and the average error is 1%. So, by considering the interfacial friction factor changes the Wongwises model can be improved. Introduction. One of the critical unknown parameters involved in predicting the pressure loss and heat transfer in any gas–liquid system is the void fraction, 1–Hl, or liquid hold-up, Hl, which is the volume of space occupied by the gas or liquid, respectively. Numerous efforts have been made to develop methods to determine liquid hold-up in pipes. Empirical and semi-empirical correlations based on experimental data and diverse theoretical models with different degrees of complexity for predicting liquid hold-up in pipes have been suggested by a number of investigators. Some authors have attempted to find general correlations for liquid hold-up in two-phase flow by curve fitting experimental data (e.g., Lockhart and Martinelli),3 Beggs and Brill4 and Abdul-Majeed.5 These correlations are strongly dependent on the composition of the database used. Other authors have developed models and correlations specific for each flow type: stratified flow (e.g., Agrawal,6 and Chen and Spedding;7 annular flow (e.g., Kadambi8 and Tandon et al.;9 and slug flow (e.g., Bonnecaze et al.,10 Mattar and Gregory,11 Gregory et al.12 and Go´mez et al.13 Most stratified flow models were based on an iterative solution of the two-phase momentum balance, but differed in the model of the interfacial shear stress. Taitel and Dukler14 developed a model for hold-up and assumed that the interface was smooth and the interfacial friction factor was equal to the gas–wall friction factor. In another article, they demonstrated that the hold-up and the dimensionless pressure drop for stratified flow are unique functions of X (Lockhart–Martinelli parameter)3 under the assumption that fg /fi = constant. Kawaji15 predicted hold-up successfully by substituting the ratio of the

gas–wall friction factor and the gas interfacial shear stress into the Taitel and Dukler momentum balance. Wongwises calculated the stratified flow hold-up. His method is based on that of Spedding– Wongwises’ work where the ratio of the interfacial friction factor and gas–wall friction factor is assumed to be a constant. The value of the constant depends on whether the phases are in turbulent or laminar flow. In this work, unlike other studies, variation of interface friction factor with important parameters that affect it were considered. The result of the new model and Wongwises’ model was compared with experimental data. The results show that the new method is more accurate. Experiments. The experimental facility used is shown schematically in Fig. 1. The main system components consisted of the test section, air supply, water supply and instrumentation system. The test section has two parts: fixed and moveable. Each part of 1.5 m length and 30 mm diameter were made of plexiglass to permit visual observation of the flow pattern. The angles of the moveable part can change from 0 to 40°. The water at room temperature and atmospheric pressure was pumped from the storage tank through the rotameter and then to the air–water mixer and the test section. The surplus water was sent back to the storage tank. Air was supplied to the test section by a compressor. Both the air and water streams were brought together in a mixer with a 6-mm inside diameter then passed through the test section. The

FIG. 1

The test facility piping was made of clear plexiglass.

HYDROCARBON PROCESSING AUGUST 2009

I 49

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

the storage tank. Air was supplied to the test section by a compressor. Both the air and water streams were brought together in a mixer with a 6-mm inside diameter then passed through the test section. The air inlet flowrate was measured by a digital flow meter within the range of 0–12,000 lit/hr. Water flowrate was measured by two sets of rotameters for laminar and turbulent regimes within the range of 0–600 lit/hr and 1,000–6,000 lit/ hr. The air pressure was controled by the regulator. During the experiments the superficial gas velocity was maintained at a set value. The superficial liquid velocity was adjusted and, once the system had reached steady state, the hold-up was measured. The liquid hold-up was measured by isolating the test section with two quick-closing valves. After isolating the test section, the first valve opens and the amount of water that was isolated between the two valves was measured, so hold-up was found by dividing this amount of water into the total volume. Four differential pressure transducers were installed in the test section to measure the two-phase pressure drop across the test section. The range of the transducers is 0–100 psig. Experiments were conducted at various air and water flowrates (the moveble part was set at 0° during the experiments). In the experiments, the air flowrate was increased while the water flowrate was kept constant at a preselected value and then the experiments were repeated with other water water flowrates. The system was allowed to approach steady conditions before the air and water flowrates, flow pattern, pressure drop and hold-up were recorded. Mathematical model. This model is based on flow geometry that is presented in Fig. 2. Considering steady-state gas–liquid stratified flow in a horizontal pipe, neglecting droplet entrainment and assuming onedimensional motion for each phase, the momentum balances for each phase yield: dP AL WL S L + i Si = 0 (1) dx

dP AG WG SG + i Si = 0 dx

(2)

Assume that the pressure drop in the liquid and gas phase is equal to:

1 SG S 1 WL L + i Si + = 0 AL AG AG AL

(3)

The shear stresses are evaluated in a conventional manner:

WL = f L

L uL2 2

(4)

WG = fG

G uG2 2

(5)

G (uG uL )2 (6) 2 For correlating the liquid and gas friction factors the form of the Blasius equation is used: i = fi

D u n f L = C L L L L

(7)

D u m FG = CG G G G

(8)

where Dg and Dl are the hydraulic diameter evaluated in the manner as suggested by Agrawal et al.6 DL = DG =

4 AL SL

(9)

4 AG

SG + Si

(10)

Furthermore, the coefficients Cg , CL , m and n used in Eqs. 7 and 8 are calculated by Taitel–Dukler.16 In turbulent flow:

CG = C L = 0.046,m = n = 0.20

(11a)

In laminar flow:

CG = C L = 16,m = n = 1.0

(11b)

Turbulent or laminar flow conditions in each phase are identified by calculating the Reynolds number for each phase. Laminar flow is assumed for superficial Reynolds numbers < 2,000: uSk D K Sk = G, L Re SK =

By substituting ␶ WL , ␶ WG and ␶ i from Eqs. 4–6 into Eq. 3, the following equation was obtained:

hL FIG. 2

Flow configuration used in stratified flow.14

I AUGUST 2009 HYDROCARBON PROCESSING

f u2 S fG G uG2 SG L L L L 2 AG 2 AL f i G uG2 Si 2

1 1 + A =0 L AG

(12)

Delivering the Future... AVEVA software supports the world’s Oil & Gas industries with solutions for the entire engineering lifecycle - from initial design through fabrication and construction, to the lifetime support of facilities. For 40 years, AVEVA has been helping its customers face the challenges of a continually evolving business environment... offering ever more capability, productivity and flexibility, but never at the expense of long-term stability. AVEVA’s unique and proven technology is unrivalled in safeguarding its customers’ data and protecting their significant investment in engineering information technology.

Don’t get left behind... Find out how AVEVA can help deliver a better future for your engineering data at: www.aveva.com or contact us via email on: future@aveva.com

Global head office: AVEVA Group plc Tel +44 (0)1223 556655 Fax +44 (0)1223 556666

Offices, representatives and agents in over 50 countries: www.aveva.com/contact

Select 109 at www.HydrocarbonProcessing.com/RS

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT

The parameters shown in Fig. 2 are calculated as:

h AG = D 2 0.25 cos 1 2 l 1

2 2hl 2 l D 0.25 1

1 1

D 2h AL = D 2 0.25 cos 1 l 1

2h 2

2h l l 1 1 1

+0.25 D D

Start Input Ď L, Ď G, D, ÎźL, ÎźG, uG, uLĎ

(13)

IF Re < 2,000

CG = CL = 0.046, m = n = 0.20

(15)

2h SG = D cos 1 l 1

(16)

2 2h l Si = D 1 1 D

(17)

Substitute Eqs.13, 15, 16, 17 and 20 in Eq.12

FIG. 3

(u ul ) P = fi G G (19) L 2D By the measured friction factor the Hamersmaâ&#x20AC;&#x201C;Hart correlation was reformulated. The reformulated version considered the Reynolds number of mixture flow and both the liquid and gas friction factors were used as parameters of the correlation. The data were fitted with software using the least-squares method for fitting. The best parameter arrangement was selected that has the most R-square [R2 = yy(x)] and least standard deviation. The new correlation for the interfacial friction factor is: Rem0.7130

(20)

Results. Air and water properties (density and viscosity), pipe

diameter, and air and water velocities were used as input param52

I AUGUST 2009 HYDROCARBON PROCESSING

Solution algorithm.

0.8 Predicted Q1 = 150 lit/hr Q1 = 80 lit/hr Q1 = 60 lit/hr

0.6 Liquid holdup

In this new model, we needed the interfacial friction factor so that the effect of both phases was evaluated. So, the new correlation was developed to consider the effects of Reynolds mixture and both liquid and gas friction factors on interfacial friction factor. For this purpose, the pressure drop of two-phase flow was measured by four pressure transducers. By the measured pressure drop and following equation the friction factor was measured:

f g0.003657

HL = fsolve (Eq.12, initial guess) Finish

Wongwises considered that the interfacial friction factor was constant. In the new model, the changes of interfacial friction factor by Reynolds and other parameters that affect friction factor were considered. Hamersma and Hart evaluated the liquid Reynolds number and friction factor as effective parameters.1 Their correlation is: f f i = l Re L0.726 (18) 108

f l 1.0053

CG = CL = 16, m = n = 1.0

(14)

2hl S L = D cos 1 1

f i = 0.01065

ReSK =_____ uSkD YK

0.4

0.2

0.0 0.0

FIG. 4

0.5

1.0 1.5 SuperďŹ cial gas velocity, m/s

2.0

2.5

The airâ&#x20AC;&#x201C;water stratified flow hold-up in horizontal pipe by the new model.

eters and then these TABLE 1. Hold-up prediction parameters and cor- accuracy, of this model compared relations were sub- to the Wongwises model stituted in Eq. 12. E1% E 2% E 3% Ho l d - u p w i l l b e Models New model 0.0212 0.0524 1.582 found when Eq. 12 is solved with the soft- Wongwises model 2.5706 6.8549 18.3937 ware solver. Notice that the air and water velocities were selected at stratified regions. The solution algorithm is shown in Fig. 3. The experiments were done at three different liquid velocities in the stratified region. In

We’ve earned our sea legs.

Think about ITT. Offshore, the challenges are always tougher. And the need for reliability in equipment couldn’t be more critical. ITT is up to the task, every day. Providing pumps, agitators, valves, desalination equipment, electrical feedthru systems, composite piping, switches, vibration isolation systems and more. The truth is, offshore oil and gas simply couldn’t be produced without us. For more information, and to receive our Oil and Gas catalog, visit ResourceITT.com. Doing more with the world’s most precious resources.

Select 67 at www.HydrocarbonProcessing.com/RS

SPECIALREPORT

FLUID FLOW AND ROTATING EQUIPMENT 6

0.8 Predicted Q1 = 150 lit/hr Q1 = 80 lit/hr Q1 = 60 lit/hr

Liquid holdup

0.6

0.4

0.2

0.0 0.0

FIG. 5

0.5

1.0 1.5 Superﬁcial gas velocity, m/s

2.0

2.5

The air–water stratified flow hold-up in horizontal pipe by the Wongwises model.

each test the liquid velocity was fixed and the gas velocity varies from 0.47 m/sec to 2.35 m/sec. In Fig. 4, results from the new mechanistic model were plotted and compared with experimental data. In the next part, liquid hold-up in horizontal stratified twophase flow by the Wongwises mechanistic model was calculated. In this model liquid hold-up was predicted by using the Taitel and Dukler momentum balance16 and the ratio of interfacial friction factor and superficial gas–wall friction factor was assumed to be constant. Results from this model are plotted in Fig. 5. The difference between the new model and the Wongwises model from experimental data was calculated and is shown in Table 1. The average percent error, E1, average absolute percent error, E2, and root mean-square error, E3, are calculated as below. Table 1 shows that the new method is more accurate than the Wongwises mechanistic model. 1 n 1 n 1 E1 = ri E 2 = ri E3 = (ei )2 n n i=1 n 1 i=1 This model has less computation time than other mechanistic models for stratified flow pattern in horizontal piping because it used a simple algorithm. This model, unlike previous models, considered the variation of interface friction factor with Reynolds number and both gas and liquid frictions. So the interface behavior with increasing or decreasing liquid and gas velocity can be computed. The new mechanistic model has the average error 1% that is less than the Wongwises model. So by a new reformulation of the Hamersma–Hart friction factor correlation the Wongwises model can be improved. HP 1

4 5

LITERATURE CITED Hamersma, P. J. and Hart, J., “A pressure drop correlation for gas/liquid pipe flow with a small liquid holdup,” Chemical Engineering Science, 42, 1187–1196, (1987). Wongwises, S., Khankaewr, W., and Vetchsupakhun, W., “Prediction of Liquid Holdup in Horizontal Stratified Two-Phase Flow,” ThammasaIntt. J. Sc. Tech., 3, (1998). Lockhart, R. W. and Martinelli, R. C., “Proposed Correlation of Data for Isothermal Two Phase, Two Component Flow in Pipes,” Chem. Engg. Prog., 45, 39–48, (1949). Beggs, H. D. and Brill, J. P., “A study of two phase flow in inclined pipes,” Journal of Petroleum Technology, 607–617, (1973). Abdulmajeed, G. H.,” Liquid holdup in horizontal two-phase gas–liquid flow,” J. Petrol. Sci. Eng., 15, 271–280, 1996.

I AUGUST 2009 HYDROCARBON PROCESSING

Agrawal, S. S., Gregory G. A., and Govier, G. W, “An Analysis of Horizontal stratified Two Phase Flow in Pipes,” Can. J. Chem. Eng., 51, 280–286, (1973). 7 Chen, J. J. and Spedding, P. L., “An analysis of holdup in horizontal twophase gas liquid flow,” Int. J. Multiphase Flow, 2, 147–159, 1983. 8 Kadambi, V., “Prediction of Void Fraction and Pressure Drop in Two-Phase Annular Flow,” GE Rep, No. 8OCRDI56, (1980). 9 Tandon, T.N., Varma, H. K. and Gupta, C. P., “A void fraction model for annular two-phase flow,” Int. J. Heat Mass Transfer, 28, 191–198, 1985. 10 Bonnecaze, R. H., Erskine, W. and Greskovich, E. J., “ Holdup and pressure drop for two phase slug flow in inclined pipes,” AIChE J., 17, 1109–1113, 1971. 11 Mattar, L. and Gregory, G. A., “Air oil slug flow in an upward-inclined pipe – I: Slug velocity, holdup and pressure gradient,” J. Can. Petroleum Technol., 13, 69–76, 1974. 12 Gregory, G. A., Nicholson, M. K., Aziz, K., “Correlation of the liquid volume fraction in the slug for horizontal gas liquid slug flow,” Int. J. Multiphase Flow, 4, 33–39, 1978. 13 Gomez, L. E., Shoham, O. and Taitel, Y., “Prediction of slug liquid holdup: horizontal to upward vertical flow,” Int. J. Multiphase Flow, 26, 517–521, 2000. 14 Taitel, Y. and Dukler, A. E., “A theoretical approach to the Lockhart–Martinelli correlation for stratified flow,” Int. J. Multiphase Flow, 2, 591–595, (1976). 15 Kawaji, M., Anoda, Y., Nakamura, H. and Tasaka, “Phase and Velocity Distributions and Holdup in High-Pressure Steam Stratified Flow in a Large Diameter Horizontal Pipe,” Int. J. Multiphase Flow, 13, 145–159, (1987). 16 Taitel, Y. and Dukler, A. E,” A model for predicting flow regime transitions in horizontal and near horizontal gas-liquid flows,” AIChE Journal, 22, 47–55, (1976). 17 Eaton, B. A., Andrews, D. E., Knowles, C. R., Silberberg, I. H. and Brown K. E., “The prediction of flow patterns, liquid holdup, and pressure losses occurring during continuous two-phase flow in horizontal pipelines,” J. Pet. Tech., 815–828, (1967). 18 Melkamu, A., Woldesemayat, Afshin, J. and Ghajar, “Comparison of void fraction correlations for different flow patterns in horizontal and upward inclined pipes,” International Journal of Multiphase Flow, 33, 347–370, (2007). 19 Badie, S., Hale, C. P., Lawrence, C. J. and G.F. Hewitt, “Stratified laminar countercurrent flow of two liquid phases in inclined tubes,” International Journal of Multiphase Flow, 29, 1583–1604,(2003).

Hasti Firoozfar began her career in chemical engineering at Arak University, where she graduated with her BSc degree, followed by her post-graduate studies at Iran University of Science and Technology, where she studied for her MSc degrees. Her special fields of interest include fluid mechanics.

Norollah Kasiri began his career in chemical engineering at Glamorgan University, where he graduated with his BSc degree, followed by his post-graduate studies at Swansea University, Wales, UK, where he studied for his MSc and PhD degrees. Dr. Kasiri then joined the Chemical Engineering Department at Iran University of Science and Technology (IUST) as an assistant professor where he established the CAPE center. Over the past 12 years of CAPE activity, Dr. Kasiri has managed to team up professional chemical, process and reservoir engineers, resulting in the presentation and publication of over 180 papers, conclusion of 58 research projects and development of 14 software packages. He is currently with IUST as an associate professor.

Mohammad Hassan Khanof began his career in chemical engineering at Sharif University, Tehran in 1971, graduated in 1975 with a B.sc degree, graduated with an M.Sc degree in chemical engineering in 1977 and an M.Sc degree in mechanical engineering in 1978 at the University of Southern California, US. Since 1987 he has been in the Chemical Engineering Department at Iran university of Science and Technology as an instructor. His special fields of interest include fluid mechanics projects.

HEAT TRANSFER

What metals should be used in critical service heat exchangers? Here are important guidelines when considering titanium for exchanger construction R. PRAMANIK, Consolidated Contractors Co., Abu Dhabi, United Arab Emirates

n heat exchanger applications, titanium and its alloys are increasingly replacing copper-nickel alloys as construction materials for seawater service. Titanium (Ti) offers increased resistance to almost all forms of seawater corrosion, lower density, ability to withstand higher temperatures and advanced manufacturing technology. However, some specific issues related to the design and fabrication of Ti-based shell-and-tube heat exchangers have surfaced (Fig. 1). The presented strategies will focus on developing design criteria to optimize heat exchangers without compromising the quality of fabrication, and thereby, being cost-competitive. This case history will fully explore these issues with special emphasis on the design, material selection and fabrication of various components of shell-and-tube heat exchangers from Ti and Ti alloys. Why Ti? Compared to copper (Cu)-nickel (Ni) and SS 316, Ti is more resistant to most forms of corrosion, e.g., crevice, pitting, sulfide-stress cracking corrosion, fatigue, galvanic, microbial induced corrosion, etc.1 This coupled with advances in manufacturing technologies allows using Ti at much higher temperatures in seawater services, and thus, extending the service life of the exchanger.2 Due to these conditions, Cu alloys are being replaced by Ti as a construction material for heat exchanger in sea water service. The most common Ti grade used for shell-and-

tube heat exchangers is Grade 2; it provides the best combination of strength, ductility and weldability. Grade 2 Ti can withstand process temperatures up to 300°C without any significant loss of strength. With suitable alloying elements, Ti can withstand processing temperatures of 600°C. However, there are some complications when using Ti. Notably, Ti is susceptible to hydriding under various process conditions, including:3 • Metal temperatures exceeding 176°F (80°C) and • Process conditions that generate hydrogen: > Galvanically coupled to certain active metals, e.g., carbon steel (CS) in hydrogen sulfide (H2S)-containing aqueous media > Under alkalinity process conditions, limits of 3 < pH or pH > 12. Hydriding should be considered when determining the metallurgy of various components of shell-and-tube heat exchangers, e.g., baffles, tubesheets, etc. COMPONENT DESIGN AND FABRICATION

We will consider the various components of the shell-and-tube exchange and how construction materials influence capital cost and service life of the component, as well as the exchanger: Tube. Ti tubes are available in three forms—seamless, welded/

cold worked (WC) and welded. All three tube types of tubes can be ordered either as an average or minimum wall thickness; seamless with minimum wall tubes, are highest cost, while welded tube with average wall thicknesses are the lower cost tube. The final choice of tube type depends on user preference and severity of service; seamless tubes are preferred to welded tubes for high-pressure service on the shell side. Generally, seamless tube with an average wall thickness is the most commonly used in seawater services. Tube-wall thickness. In cases where the exchanger is designed as per API-660, the minimum tube-wall thickness should be 1.24 mm (average wall), unless specified otherwise by user specification.4 However, recommendations of API-660 should be used as good engineering practices, even if API-660 is not applicable. FIG. 1

Heat exchangers are constructed to handle sever conditions with different heat-transfer requirements and footprint needs.

Allowable tube-side velocity. Since Ti tubes are more erosion resistant than Cu-alloy tubes, the maximum allowable tubeHYDROCARBON PROCESSING AUGUST 2009

I 55

K T I C O R P : F I R E D H E AT E R S & S C R S Y S T E M S

World Leader in Fired Heaters and SCR Systems ENGINEERING - FABRICATION - CONSTRUCTION Fired Heaters:

SCR Systems:

Refinery Applications Steam Reformers Petrochemical Applications OTSGs Global E-3 Services

Gas Turbines Heaters Boilers FCC Units other fired sources

Please visit www.kticorp.com for a complete list of our products, services & contacts.

KTI Corporation 1990 Post Oak Blvd., Suite 1000, Houston, TX 77056 Tel: (281) 249-2400 Fax: (281) 249-2328 E-mail: sales@kticorp.com KTI - KOREA #612, Kolon Science Valley II, 811, Guro-dong, Guro-gu, Seoul, 152-050, Korea Tel: 82-2-850-7800 Fax: 82-2-850-7828 E-mail: BSKim@kti-korea.com Select 96 at www.HydrocarbonProcessing.com/RS

HEAT TRANSFER side velocity for Ti exchangers is much higher than the Cu-alloy counterparts. Generally, the tube-side velocity inside the Ti tubes should be maintained between 1 m/sec to 4.5 m/sec to optimize maintenance and operating costs. Tube-hole tolerance. Tube-hole tolerance in the tubesheet

should TEMA recommendations, as per section 9.9.3 of API-660.4 U-bend radius. Ti tubing is routinely bent on conventional

tube-bending equipment. Mandrel benders are recommended, particularly for tight bends. Wiper dies and mandrels should be smooth and well lubricated to minimize Ti tubing’s tendency for galling. Tube manufacturer recommendations should also be addressed before finalizing the minimum bend radius. Smaller than the recommended bend radius may require preheating the tube at temperatures of 400°F–600°F before bending. Likewise, using heavier wall tubing for tight bends should be considered to compensate for wall thinning that occurs during bending. Stress relieving of U bends. In all heat treatments of Ti materials, designers should consider that Ti and Ti alloys are highly reactive with hydrogen, oxygen and nitrogen. Accordingly, heat-treatment facilities must be designed to mitigate reactions with tube materials. This particularly applies to hydrogen pickup from the furnace. Heating should be done in a neutral or slightly oxidizing atmosphere, and the heating cycles maintained as short as possible. Due to these conditions, stress relieving of the Ti-based U bends is not always recommended; yet, there are different opinions by heat-exchanger designers on this subject. However, Appendix 6 Section D of ASME Sec VIII Div 1 recommends heat treatment of Ti after the forming operation at 900°F to 1,100°F for 30 minutes for Grades 1, 2, 3 and 7, and 1 hour for Grade 12. Careful review of operating conditions and discussion with the vendor are recommended before specifying the stress relieving of Ti U bends. Tubesheet. The tubesheet should be solid Ti when designed to SB-265 as per ASME recommendations. Although the ultimate tensile strength (UTS) of SB-265 is 50,000 psi, ASME Code Case 2497-1 allows using a UTS of 58,000 psi, resulting in a thinner tubesheet when: • Exchanger diameter is small (when the cost advantage for a cladded tubesheet over a solid tubesheet does not exist). • Conditions favor hydriding and a Ti-cladded CS tubesheet can create a galvanic couple between the Ti tubes and the CS backer plate of the tubesheet. • Seawater is present on the shell side. It is almost impossible to prevent seawater from leaking through the shell-side joint of the tube and tubesheet (which, unlike the tube side of the tube-totubesheet joint cannot be seal welded) undercutting the backside of the Ti cladding and finally leading to failure of the tubesheet. For all other cases, the tubesheet should preferably be cladded, thus, optimizing the exchanger design without compromising quality. The thickness of the cladding should be determined per RB7.8 of TEMA, unless otherwise specified by the user.5 Tube-to-tubesheet joint. The tube-to-tubesheet joint should be expanded and seal welded as recommended in Appendix D of API 660. Any leakage in the absence of seal welding may affect the backer plate behind the cladding and cause corrosion eventually. Other types of tube-to-tubesheet joints include:

• Expanded. When the tubesheet is solid Ti, the joint does not need to be seal welded. However, it is advisable to adopt seal welding for solid Ti tubesheets. Since under high-yield stress, Ti tubes shrink after expansion, thus creating leaks. • Strength welded. If the equipment is in hydrogen service, then the joint should be strength welded as recommended in Appendix D of API-660. • Strength welded and expanded. Done when specified by end users and/or process technology licensors. Baffles, tie rods and spacers. Under particular service conditions, the baffles, tie rods and spacers should be made of Ti when: • Conditions resulting in hydriding are present or there is a chance of contamination of the Ti tube with a CS baffle. • Seawater is on the side shell. For all other process conditions, the CS baffle in combination with Ti tubes can be used by following several measures that mitigate contamination and damage to the tube from vibration. Some options are: • Chamfering the tube holes in the baffle and protecting the tube with fluids such as alcohol so that CS cannot stick on the outside of the tubes. • Properly selecting the baffle spacing and tube pattern and installing partial baffles to prevent excessive lateral displacement of tubes during operations, e.g., when designing exchanger using HTRI the lateral displacement of tubes is always < 10% of the gap between tubes to mitigate vibration. Floating head assembly. The floating head cover and flange of the heat exchanger should be made from a solid Ti and solid formed head, as shown in Fig. 2. Cladded or lined cover, as shown by the second sketch in Fig. 2, is not advisable. Leakage of seawater through the gasketed joint can creep in undetected between the lining and backer plate and eventually damage the floating head assembly since it remains inside the shell cover and is not visible from the outside of the exchanger. Backing ring. When the exchanger uses seawater on the tubeside, the backing ring should be made from CS. Since there is no chance of the backing ring coming in contact with seawater, CS can be used. Conversely, if seawater is used on the shell side, then the backing ring should be Ti.

7 typ. Ti lining (SB265 Gr. 2)

FIG. 2

A solid vs. cladded floating head assembly and lined floating head cover. HYDROCARBON PROCESSING AUGUST 2009

I 57

HEAT TRANSFER

SB 266 CI 2 Loose ﬂange (Tubeside)

TiJA gaskets SA 266 CI 2 Hub ﬂange (Shellside) SA 516 Gr 70 (Shell)

SB 265 Gr-2 (channel) SB 266 Gr-2 lap

FIG. 3

FIG. 4

FIG. 5

Flange stoppers prevent flange rotation for loose-flange girth covers.

FIG. 6

Typical girth flange design with tubesheet.

A larger exchanger with CS-cladded channel cover and smaller exchanger with solid Ti channel cover.

Explosion bonded vs. screwed cover for channel.

Shell and shellside nozzle. With seawater on the tube side,

the shell and shell-side nozzle: • Should be non-titanium when conditions do not result in hydriding. • Should be solid Ti/Ti-cladded when conditions can result in hydriding. With seawater on the shell-side, the shell and nozzle materials should be solid Ti/Ti-cladded. Channel and tube-side nozzle. Since it is always difficult to get Ti-cladded plate in small sizes and it is difficult to overlay Ti on CS, the channel should be constructed of solid Ti materials when seawater is on the tube side. However, the tube-side nozzle can be solid Ti pipe or made from Ti-cladded plate depending on the nozzle size. The channel and tubeside nozzles should be made from non-Ti materials when seawater is on the side shell. Channel cover cladding. Cladding on the channel cover can be either explosion bonded or screwed. Explosion bonded is 58

Ti Gr-2 tubesheet

I AUGUST 2009 HYDROCARBON PROCESSING

better because it ensures no gaps between Ti and CS base metal. Explosion welding is a unique technology that uses energy from a chemical explosive to create fusion and welding conditions. The extreme rapidity of the process does not permit sufficient time to form detrimental inter-metallic compounds that can form during other, slower welding technologies (Fig. 4). In the case of screwed covers, minute gaps can remain between the Ti and the backer plate (as shown in Fig. 4,) which can cause corrosion of the CS backer plate when there is seawater leakage through the gasket between the girth flange and the channel (a phenomenon known as undercutting), finally leading to failure of the channel. Girth flange and nozzle flange. Designers have options

for selecting flange designs: CS loose flange. This design is applied to reduce costs. A typical flange design is shown in Fig. 5. Two important issues in installing a loose-flange design are the flange rotation, which in turn may cause gasket leakage, and longitudinal movement of the loose flange due to the force exerted by jack screws during disengagement of gaskets. One way to solve both issues is to provide a stopper on the loose flange side (Fig. 6). Ti-hub flange. A solid Ti-type flanges can be used when loose flange design cannot be used due to high pressures (< 300 psi) or in critical service. Appendix-2 of ASME Sec VIII Div 1 provides selection criteria for loose flanges, when the equipment is designed as per ASME. However, in seawater service, the flange rating

K T I C O R P : R E VA M P G RO U P

Fired Heater Global E-3 Services EVALUATE - ENGINEER - EXECUTE FIRED HEATER STUDIES ENGINEERED REVAMPS EMERGENCY REBUILDS CONSTRUCTION SERVICES REPLACEMENT PARTS KTI Corporation 1990 Post Oak Blvd., Suite 1000, Houston, TX 77056 Tel: (281) 249-2400 Fax: (281) 249-2328 E-mail: sales@kticorp.com

KTI - KOREA #612, Kolon Science Valley II, 811, Guro-dong, Guro-gu, Seoul, 152-050, Korea Tel: 82-2-850-7800 Fax: 82-2-850-7828 E-mail: BSKim@kti-korea.com

Please visit www.kticorp.com for a complete list of our products, services, and contacts. Select 97 at www.HydrocarbonProcessing.com/RS

HEAT TRANSFER Gaskets. When contact with seawa-

Flange arrangements in descending order of cost

ter is possible, the gaskets should be a Ti double-jacket graphite face (DJGF) type. Sometimes a Monel DJGF is specified to 1 3 reduce costs. However, there is a chance for galvanic corrosion to occur between Stainless steel lap flange, bolts, Ti bolts, nuts the Monel and Ti, resulting in pitting at nuts and washers. and washers. the gasket contact face. To mitigate pitting Ti welded stub end, (M.S.S-SP43-Type A) problems, a sacrificial (zinc) anode can be with minimum weld to design code and extra installed on the seawater side. This may weld runs for large machined radius at rear. Ti slip-on flange. cause blocking of Ti tubes by zinc erod(Note: Type B stub end rear 2 with reduced radius at rear ing from the sacrificial anode. So, the Ti is cheaper alternative). DJGF without a sacrificial anode is a bet4 Ti bolts, nuts ter option over the Monel DJGF with a and washers. sacrificial anode. Stainless steel lap flange, bolts, nuts The designer should evaluate gasket and washers. materials for the tubesheet when fluids on 5 6 Same as 3 but with mild steel epoxy Same as 4 but with mild steel epoxy either side of the tubesheet are different coated flange, bolts, nuts and washers. coated flange, bolts, nuts and washers. (Fig. 5). Since the flange type on either Source: Titanium for offshore and marine applications–a designers and users handbook, side of the tubesheet is different (the one Titanium Information Group on the seawater side being the loose type, while the flange on the process fluid side is FIG. 7 Possible flange design using Ti and Ti alloys in descending order of installed costs.6 the integral type,) the possibility of flange rotation on the seawater side will be aggrararely exceeds 150 psi, unless 11/13 rule of the API requires a vated if the gaskets of either side are different (e.g., spiral-wound higher design pressure for seawater to address tube rupture. The gasket on the side shell and a Ti DJGF on the tubeside). It is CS loose tube flange is the most commonly used flange type in Ti suggested to install the same gasket type on either side of the shell-and-tube heat exchangers (Fig. 7). tubesheet (as shown in Fig. 5). Ti stub end.

Ti weld neck ﬂange.

Options in construction materials. Continuous advances in manufacturing technology of Ti and its alloys should be kept in mind while specifying the various details. Vendor and user recommendations should be considered before finalizing the design of Ti shell-and-tube heat exchangers. HP 1

3 4 5 6

HIGH ACCURACY FLOW METERS FOR HIGH TEMPERATURES AND HIGH PRESSURES – – – – – – – – –

non-intrusive ultrasonic clamp-on technology for temperatures up to 750 °F independent of process pressure multi-beam for high accuracy wide turn down installation without process shut down no maintenance no pressure loss standard volume calculation

Cu Ni Ti CS www.flexim.com usinfo@flexim.com FLEXIM Instruments LLC CA: (510) 420-6995 NY: (631) 492-2300 TX: (281) 635-2423

NOMENCLATURE Copper Nickel Titanium Carbon steel

Ritabrata Pramanik is the lead mechanical engineer of one of the oil and gas development projects in the Middle East for Consolidated Contractors Co. in Abu Dhabi. He holds a degree in mechanical engineering from Jadavpur University. He holds a post graduation degree from IIT, Madras in heat transfer and thermal power. Since then, Mr. Pramanik has held numerous positions of responsibility in esign of heat exchangers with Engineers India Ltd., two leading manufactures of air-cooled heat exchangers, and Bechtel India. In 2005, he joined Consolidated Contractors Co., as a senior heat transfer engineer and was involved in the design and engineering of all heat transfer equipment for a refinery expansion project in the Middle East. Since 2006, he has been working in his present position.

Select 161 at www.HydrocarbonProcessing.com/RS

5687_Flexim_Anzeige-HydrocProc 1

LITERATURE CITED Mountford Jr., J. A., “Titanium—properties, advantages and applications solving the corrosion problems in marine service,” Paper 02170, Corrosion 2002. Pugh, S., G. F. Hewitt and H. Muller-Steinhagen, “Fouling During the Use of Seawater as Coolant—The Development of a ‘User Guide’,” Paper 3, 2003 ECI Conference on Heat Exchanger Fouling and Cleaning: Fundamentals and Applications. Table A.41, NACE MR0175/ ISO 15156-3:2003. “Shell and Tube Exchangers for General Refinery Service,” API-660/ISO16812. Standards of the Tubular Exchanger Manufacturer Association, 8th Ed. Titanium for offshore and marine applications—A designer’s and user’s handbook, Titanium Information Group.

11.01.2008 11:20:16 Uhr

SAFETY/LOSS PREVENTION

Are you bypassing process safety programs? Implementing these changes may increase profits and lower incidents M. SAWYER, Apex Safety Consultants, Houston, Texas

here is notable similarity between financial blunders making headline news recently and the attitudes of some corporate officials toward process safety. Consider the expenses a company is willing to pay out once mistakes are exposed. For example, AIG continues to pay millions of dollars to public relations firms, crisis management companies and lobbyists in efforts to improve their reputation and corporate image after their financial inadequacies became public. If a fraction of this money had been proactively invested in ethics and sound investment practices, AIG would not likely be on the verge of financial ruin. Expenses. Now, compare the expenses of financial blunders that

corporations face with the expenses associated with catastrophic incidents that have impacted companies within the process industries. One recent and costly incident, both financially and in loss of life was the BP Texas City explosion. BP has reportedly paid out in excess of $2 billion in conjunction with claims and business interruptions arising from its Texas City incident.1 The primary difference in AIG and BP is that the Texas City incident likely impacted fewer lives financially than AIG’s downfall. It is noteworthy that no fatalities were attributed to AIG’s financial ruin. A substantial cost portion for the Texas City incident was due to the large number of fatalities and arising litigation. This comparison excludes the emotional and humanitarian sides of process incidents and relates solely to financial aspects. Process incidents are expensive and some companies never recover from a major incident. BP is still financially viable; however, Union Carbide was unable to fully recover from the Bhopal disaster and was eventually purchased by Dow in 1999. While only a few incidents are as devastating as Bhopal or as catastrophic as Texas City, the financial cost of a process incident may easily reach tens of millions of US dollars. Recent litigation in Houston, Texas, that involved a contractor fatality at a small landfill gas processing plant resulted in a $9.76 million verdict.2 This amount reflects only the jury award; it does not include other indirect costs associated with the incident investigation, loss in production, etc. This was one of many events litigated in US courts with the potential of more process incidents to be settled outside the courts. In basic engineering economy courses, undergraduate students learn fundamental principles designed to rationally analyze how much present capital should be invested to prevent spending “x” amount in the future. Given the costs associated with a process incident vs. the cost of preventing or mitigating the incident, rational companies would likely elect to expend the capital needed

to prevent the incident. Corporations should not imagine they are invincible to process incidents; a simple Web search will quickly prove otherwise. So, why are many corporations reluctant to apply proactive safety programs costing only fractions of what reactive measures to incidents may cost later? This question is perplexing and it has been asked by safety professionals countless times. Yet, no rational answer has been proposed by either corporate management or safety professionals. Even when corporations know about costly incidents that have occurred at other similar plant sites, they typically fail to assess whether the incident could happen at their plant. There are numerous cases where a similar incident happened within the same corporation, at the same plant site or at plant sites that were in close proximity. In most cases, these incidents were known to management, yet were allowed to reoccur at great financial expense (e.g., Valero nitrogen asphyxiation incidents at Delaware City, Delaware; Paulsboro, New Jersey and Texas City, Texas). Minimizing capital expenditures needed for safety critical maintenance, testing, training, etc., while maximizing plant production is a simple recipe for short-term increase in profits, however, this could lead to disastrous consequences. Risk taking. Plant management has grown to understand the risk

of bypassing process safety all too well. However, many corporations have placed such a high reward for risk taking that it is simply too alluring. This practice will continue if management is allowed to reward themselves through short-term increases and as long as the reward is sufficiently high and the penalty remains relatively low in comparison. Corporate management does not intend on causing an incident, and it understands the financial burden associated with an incident, but, appealing high profits increase risk taking and subsequently may increase the likelihood of a costly incident. Safety standards. No legislated process safety standard or good will by professionals will deter this type of risk taking until corporations take away the incentive for management to bypass process safety. Historically, corporations have been reluctant to penalize management for bypassing process safety. In 2005, BP experienced its worst process incident to date, yet paid hefty bonuses to the Texas City plant management that year.3 This reflects a total disconnect from process safety philosophy and can easily be construed as rewarding dangerous risk taking. The process safety professional’s dilemma is–how to make corporate board members aware of process safety’s importance in long-term, sustainable profitability. HYDROCARBON PROCESSING AUGUST 2009

I 61

SAFETY/LOSS PREVENTION It was recently reported that the implementation of the Occupational Safety and Health Administration’s Process Safety Management program resulted in incident reductions within the process industries.4,5 This seems feasible since program implementation designed to enhance process safety should be effective in reducing incidents, thus protecting profits. Likewise, data acknowledging the reduction of process incidents should be useful in justifying process safety programs to management and corporate board members, although it would seem that safety program justification could be based solely on the costs associated with historical incidents.

■ The familiar “we take responsibility” is

the spin that corporations release to the public after an incident and this adage is old, tired and meaningless. It is more economically feasible to take proactive incident prevention steps than costly reactive measures after an incident. While process incidents may be reported as a downward trend, an accurate account of process-related fatalities and injuries may likely be hidden within incident statistics. The incident numbers may be down with the severity increasing, although there are no firm statistics to support this hypothesis. The severity may likely be camouflaged because contractors that are injured or killed at process sites are not necessarily included in the process plant statistics. These statistics are included with the contractor’s industry code, North American Industry Classification. Reporting the severity of process incidents is critical because there is a direct correlation between the incident’s severity and financial cost— excluding moral and humanitarian issues. Program implementation. If process safety programs

are successful in controlling incidents and therefore protecting profits, then why aren’t corporations rushing to implement such programs? In fact, a recent US Environmental Protection Agency (US EPA) report estimates that a significant number of process sites having large chemical inventories have not filed a risk management plan (RMP).6 This report also acknowledged that 944 corporations submitted RMPs, reporting an incident as of November 2007 (RMP incident definition, 40 CFR 68.42). This represents a relatively low percentage and provides no detail regarding incident severity; thus, no financial correlations can be drawn from the data. It would appear that, in this current economic environment corporations would closely analyze issues affecting profitability and search for areas that could maximize the effectiveness of expenditures (i.e., providing the maximum risk reduction worth). The correlation is quite clear—process incidents can represent a significant financial burden that can be effectively deterred by implementing a process safety program. Process safety advocate, Trevor Kletz, has spent many years writing on this subject—urging corporations to proactively take responsibility for process safety as a means of protecting profits 62

I AUGUST 2009 HYDROCARBON PROCESSING

and lives. The familiar “we take responsibility” is the spin that corporations release to the public after an incident and this adage is old, tired and meaningless. It is more economically feasible to take proactive incident prevention steps than costly reactive measures after an incident. Acknowledging that the corporation takes responsibility does nothing to counter its financial obligations after an incident. Proactive corporations must be open to implementing a number of changes to control the financial impact that may result from process incidents. To accomplish this goal, process companies should consider the following implementations: • Institute quality control functions within the process safety program; process safety management compliance audits alone are not adequate control measures and cannot be relied upon for quality control. • Abolish management incentives and bonuses based solely upon profitability, which rewards risk taking where process safety may be adversely impacted. • Revise attitudes toward safety responsibility; it is no longer acceptable to depend upon the old slogan, workers are responsible for their own safety, management is responsible for maintaining a safe workplace and protecting the company’s profitability. • Process safety professionals at plant sites should report to corporate officers and not to plant site management. It is ineffective to have personnel reporting to the site management that administers the programs being evaluated by safety personnel. • Budgets for maintenance, inspection, testing, training and others that impact process safety should be processed through “management of change” and justified at a level above the plant site management. • Prioritize safe plant maintenance criteria in accordance with recognized and generally accepted good engineering practices for mechanical integrity. • Benchmark process safety practices along with other corporate functions. • Implement process safety economic guidelines and requirements at all corporate levels. HP 1 2 3

5 6

LITERATURE CITED “First Explosion Trial Ends in Settlement,” CBS News Release, September 18, 2007, Galveston, Texas. Vaughn, G. and P. C. Barnes, “$9.76 Million Awarded to Family,” Press Release, February 5, 2009, Houston, Texas. Arenaza, M., E. Ramon, D. G. Crow and J. G. Crow, et al. vs. BP Products North America Inc., Case No. 05CV0337-A, July 10, 2006, Parus Deposition, Galveston County, Texas. “Accident Epidemiology and the RMP Rule—Learning from a Decade of Accident History Data for the US Chemical Industry,” Risk Management and Decision Processes Center, The Wharton School of the University of Pennsylvania, and the Office of Emergency Management, EPA, December 18, 2007. “Fatality/Catastrophic PSM Incidents,” Presentation of OSHA’s Refinery Net Effective Personnel Perspective, September 4, 2008. “EPA Can Improve Implementation of the Risk Management Program for Airborne Chemical Releases,” US EPA Office of the Inspector General, Report No. 09-P-0092, 2009.

Mike Sawyer is a consulting engineer at Apex Safety Consulants, Inc. in Houston, Texas. He has over 25 years of experience in general industrial safety applications, detailed analysis of processes and incident investigations. Some notable investigations include ESSO’s Longford gas plant incident in Australia and BP’s 2005 Texas City Refinery explosion.

The only sulphuric acid pump with a maximum flow of 14,000 gpm (3,200 cubic meters/hour), 800 hp (600kw), 15.3 feet high (4.7 meters) and weighing in at 14,000 lbs (6,400kg). Specifically designed and built to accommodate the largest sulphuric acid plants in the world. And, like King Kong, able to resist and protect your plant from incidents that could overwhelm smaller pumps.

Standard replacement parts always available. Emergency service or parts to any major airport worldwide within 72 hours. Lewis Pumps – serving molten sulphur and sulphuric acid plants everywhere since 1891. Customers in over 100 countries can’t be wrong.

And if “King Kong” Lewis is a little too much for you, we have the pump to fit your specific needs. Guaranteed! Select 87 at www.HydrocarbonProcessing.com/RS

CHAS. S. LEWIS & CO., INC.ÊUÊnÈÓxÊ À> ÌÊ, >`ÊUÊ-Ì°Ê Õ Ã]Ê ÃÃ ÕÀ ÊÈÎ£ÓÎÊ1°-° ° /i i« i\ÊÎ£{®Ên{Î {{ÎÇÊUÊ/i iv>Ý\ÊÎ£{®Ên{Î Ç È{ÊUÊ > \ÊÃ> iÃJ iÜ Ã«Õ «Ã°V Ê

ANSWERS FOR ALL YOUR PROCESSING NEEDS

Focus on Green & Sustainable Technology

In the Exhibit Hall: s /VER %XHIBITORS ,EADING MANUFACTURERS OF PROCESS EQUIPMENT SYSTEMS AND TECHNOLOGY FOR mUIDS SOLIDS AND PLANT SERVICES s .EW 4ECHNOLOGY 7ORKSHOPS !N INTERACTIVE OPPORTUNITY TO NEW DISCUSS SPECIlC SOLUTIONS WITH PARTICIPATING EXHIBITORS

In the Classroom: s .EW !)#H% $AY n /NE DAY #ONFERENCE DEVELOPED BY !)#H% NEW ADDRESSES PROVEN SOLUTIONS TO ADAPT TO YOUR PROJECTS s #HEMICAL %NGINEERING -AGAZINE %DUCATIONAL #ONFERENCE â&#x20AC;&#x201C; 4WO DAY PROGRAM PRESENTING PRACTICAL SEMINARS FOR MORE EFlCIENT PROCESS OPERATIONS s .ANO 'REEN 4ECHNOLOGY 3YMPOSIUM n 4WO DAY 3YMPOSIUM AND 7ORKSHOP ON THE OPPORTUNITIES THAT CURRENTLY EXIST TO INCORPORATE NANOTECHNOLOGY INTO YOUR OPERATIONS DEVELOPED BY I2!0

OnLine: NEW s .EW -ATCHING 3ERVICE CONNECTS YOU WITH THE EXHIBITORS THAT HAVE THE RIGHT PRODUCTS AND SOLUTIONS YOU NEED

More to see, do and learn at the process industries #1 event. &REE 3HOW 2EGISTRATION AND )NFORMATION

www.chemshow.com

November 17-19, 2009 | *AVITS #ONVENTION #ENTER s .EW 9ORK #ITY %NDORSING !SSOCIATIONS

Produced and managed by:

Telephone: 203-221-9232 s % MAIL info@chemshow.com

AlChE

Select 84 at www.HydrocarbonProcessing.com/RS

PROCESS TECHNOLOGIES

Biorefineries: Fact or fiction? Technology advances facilitate building integrated chemical complexes based on renewable feedstocks that can cost-effectively process ethylene derivatives M. BRUSCINO, Scientific Design Co., Little Ferry, New Jersey

ntegrated complexes for petroleum-derived processes are common. However, the same types of complexes are not typical for processes based on bio-renewable feedstocks. New technology advancements overcome processing hurdles that allow cost-effective production of ethanol-derived ethylene derivatives that can compete with traditional processing methodologies. Background. Over the last 90 years, feedstocks for most down-

stream petrochemicals are petroleum based. Due to rising prices of crude oil and concerns over the environment, renewed interest for applying alternative feedstocks—in particular, bio-renewable feedstocks—has emerged. Among the many bio-derived feedstocks, ethanol has received the most interest. Ethanol production is based on fermenting readily available agricultural products such as sugar cane and corn. Fermentation technology is simple and commercially developed and can be easily transferred to less developed nations. During the 1950s and 1960s, ethanol was largely used to produce commodity chemicals in Brazil, India and Australia due to: • Locally available, fermentable agricultural materials • Lack of indigenous hydrocarbon feedstock resources.

cheaper and larger volume based on cellulose biomass such as sugarcane bagasse • Environmental concerns • Energy supply security issues. Global supplies. Fig. 1 shows worldwide ethanol production for 2006. Due to the renewed interest in ethanol, worldwide ethanol production is projected to more than double over the next 15 years to a worldwide capacity of more than 120,000 million liters by 2020, as shown in Fig. 2. The main reason for ethanol production expansion, especially in the US, is its use as an octane additive for gasoline. As oil prices rise, emphasis will shift toward

13.5 billion gallons (51 billion liters) ethanol India 4%

China 8%

EU 2%

Brazil 33%

Other 14%

New ethanol technologies. Following a decline in inter-

US 39%

FIG. 1

Global ethanol prod., MM/liters

est, the ethanol-based industry staged a comeback during the energy crisis of the early 1970s as a possible option to reduce dependence on foreign oil imports. During this time, a fullyintegrated ethanol-to-ethylene glycols process was developed as an alternative to petroleum-based feedstock for glycols production. In 1989, the first commercial plant using this integrated technology was commissioned in India in 1989. This facility was expanded several times over the last 19 years; it is currently producing ethylene oxide (EO)/ethylene glycol (EG) from ethanol derived from molasses. In Brazil, more than 30 products were derived from ethanol, several with installed capacities above 100,000 tpy during this period. However, as naphtha prices declined throughout the 1980s and 1990s, the production for many ethanol-based chemicals became uneconomical. As the price of oil increased recently, interest in ethanol-derived chemical processes renewed. Other interesting reasons for ethanol-based chemicals include: • Significant productivity improvements and cost reductions in ethanol production from technology evolution • Increasing availability of fuel ethanol, with ambitious expansion plans in the US and Brazil • Promise of an ethanol-production process that is both

120,000 100,000 80,000

Worldwide ethanol production in 2006.1

Other EU US Brazil

60,000 40,000 20,000 0 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020

FIG. 2

Expansion of global ethanol production capacity—1975 to 2020.2 HYDROCARBON PROCESSING AUGUST 2009

I 65

PROCESS TECHNOLOGIES Ethanol comparative production cost

Chemicals from ethanol

Ethanol production costs (mid 2006), cents/l

FIG. 3

45 40 35 30 25 20 15 10 5 0

Ethylamines Ethyl ether Ethyl acrylate

Acetic acid Acetic anhydride Ethyl acetate Vinyl acetate Cellulose Acetate n-Butanol 2-Ethyl-hexanol Butadiene

(1.06) 0.28

(1.48) 0.39

(1.85) 0.49

(1.63) 0.43

0.2 0.1

Chemicals that can be produced from ethanol.1

FIG. 5

India (Molasses)

Thailand (Molasses)

Colombia (Molasses)

0.0

Ethanol production cost by various nations and feedstocks.3

Ethanol

Net raw materials Other*

Ethylene

Brazil sugarcane mill Cane = $23/ton

EDC

Others

US corn dry mill Corn = $2.5/bushel

*Includes: utilities, waste treatment, labor, maintenance, plant overhead, depreciation

FIG. 4

0.3

(1.25) 0.33

Thailand (Cassava)

Polyethylene Ethylene Dichloride Ethylene oxide (EO) Ethylene glycol (EG) Ethoxylates Ethanolamines Glycol ether Styrene

0.4

(1.51) 0.40

US (Corn)

Acetaldehyde

0.5

Brazil (Sugarcane)

Ethylene

US$/liter/(US$ gallon)

Ethanol

EU (Beets)

0.6

The ethanol production cost differential of using sugar cane (Brazil) over corn (US).1

using ethanol for chemical/petrochemical production. Fig. 3 lists some of the well-known chemicals/petrochemicals that can be produced from ethanol.

TEG t Gas dehydration t 4PMWFOU t 1MBTUJDJ[FST t )VNFDUBOU t 1PMZFTUFS SFTJOT

Brazilâ&#x20AC;&#x201D;an ethanol nation. Brazil has a large economic advan-

tage over other areas including the US and Europe. This nation is the lowest-cost producer for ethanol production from sugar cane, which is the main ethanol source in Brazil. Corn is the main ethanol feedstock in the US and Europe. Fig. 4 illustrates the production cost differential when using sugar cane over corn for ethanol production. Competitive cost structures. From Fig. 4, the cost differential in ethanol production between the US and Brazil is about $0.10/l. Due to the well-developed ethanol production and distribution capabilities of Brazil, this cost differential holds up when compared to other ethanol producers worldwide, as shown in Fig. 5. Worldwide, based on the well-developed ethanol production infrastructure in Brazil, the Brazilian ethanol-to-ethylene production is the most competitive with petroleum-derived ethylene. Other areas, such as India and China, can also be competitive based upon local situations, such as sugar production capabilities and local government incentives.

DEG

FIG. 6

t 1PMZFTUFS resins t "OUJGSFF[F t 1PMZPMT t 5&( t 5FYUJMF BHFOUT

Others

Ethoxylates

t 4UFSJMBOU t 'VNJHBOU t 4PMWFOU t 1&(4

t /PO JPOJD 4VSGBDUBOUT

MEG

EOA

t 1PMZFTUFS t "OUJGSFF[F

t (BT USFBUJOH t 4PMWFOUT

Glycol ethers t )ZESBVMJD nVJET t 4PMWFOUT

Chemical/petrochemicals based on ethanol and ethylene.

bed reactors, the reaction is either isothermal or adiabatic. This discussion will focus on the adiabatic, fixed-bed reactor ethanolto-ethylene process. Fig. 7 is a schematic of an adiabatic fixed-bed ethanol-to-ethylene process. In this process, ethanol is vaporized and then superheated in a fired heater before being fed to the first in a series of dehydration reactors. The endothermic dehydration reaction requires reheating between reactors to drive the reaction to 99% ethanol conversion or with high selectivity to ethylene. The dehydrator effluent stream is cooled and compressed. After compression, the ethylene stream is washed with caustic, dried and purified in an ethylene column followed by a stripper to reduce carbon monoxide (CO) levels in the final ethylene product.

Ethanol-to-ethylene processes. Ethylene production is

the primary chemical application from ethanol; the ethylene is then used to produce other major downstream ethylene derivatives as listed in Fig. 6. The ethanol-to-ethylene reaction takes place in the vapor phase, using either fixed-bed or fluidized-bed reactors. For fixed66

I AUGUST 2009 HYDROCARBON PROCESSING

Fully integrated ethanol-to-ethylene complex. One

way to reduce the current pricing disadvantage of ethanol-derived ethylene, as compared to petroleum-derived ethylene, would be to take advantage of the benefits from building a fully-integrated chemical processing unit. The complex would use sugar cane

PROCESS TECHNOLOGIES BFW from/to other area Ethanol

Dehydration reactors MP steam

Ethanol puriﬁcation (optional)

Steam

Water purge

Waste heat boiler

Ethanol dehydration reactors feed heater

BFW

Quench system

Vent

Refrig.

Polymer grade ethylene product

Process water Caustic

Steam

Water purge Ethylene Caustic wash system compressor FIG. 7

Dryer

Ethylene column

Steam Energy recovery

Heavies purge Stripper

Flow diagram of polyethylene based ethanol as the feedstock.

Performance and Reliability. Our DATUM®

Ethylene from ethanol for EO production Reactor feed/efﬂuent exchanger Ethanol feed preheater

Dehydration condenser Ethanol vaporizer

Ethylene compressor

Stm. Ethylene wash tower Steam generator

Ethanol

FIG. 8

Ethanol puriﬁcation (optional)

Water

BFW from/to EO reactor area

Dehydration reactor

Ethylene To EO reaction area

Ethanol dehydration reactor feed heater

Cond. Aqueous purge MP steam

Condensate Aldehyde stripper

Process condensate purge

compressors are inspiring a lot of loyalty. And no wonder—they consistently demonstrate superior performance. They are a leader in the industry in efﬁciency and pressure rise per case. And the modular design of the DATUM compressor allows easy maintenance and increased availability—making it the most advanced turbocompressor for the oil, gas, and process industries. Along with great performance, you’ll get the D-R heritage of continuous innovation and worldwide, total life-cycle support. Not to mention the fastest cycle times in the industry. For more information contact D-R today.

Flow diagram of EO/EG plant using ethanol as the feedstock.

as feed to produce downstream ethylene derivatives such as vinyl chloride monomer (VCM), ethanolamines and highdensity polyethylene (HDPE). Such an integrated process was developed in the 1980s for EO/EG production from ethanol as shown in Fig. 8. When the product ethylene is used to feed an integrated EO or EO/EG plant, the ethanol is vaporized, diluted with steam and fed to a single-bed dehydration reactor. The steam carries sufficient heat to provide endothermic heat of reac-

tion; this yields almost full conversion with a high selectivity to ethylene. The ethylene from the dehydrator is cooled, compressed and washed in preparation for mixing with the EO reactor cycle gas. The advantages of integrating the ethanol dehydration and EO/EG production process allow using excess steam generated in the process to reduce capital costs associated with ethylene product purification. Such advantages were demonstrated in a commercial plant operating in India since the late 1980s. Table 1 compares the

The Americas: (Int’l +1) 713-354-6100 ESA: (Int’l +33) 2-35-25-5225 Asia-Paciﬁc: (Int’l +60) 3-2093-6633 info@dresser-rand.com

www.dresser-rand.com

Select 162 at www.HydrocarbonProcessing.com/RS

re fo to be 09 00 ok 20 Bo July =C 4 31 VE SA

ERTC

ERTC 14th Annual Meeting

9 –11 November 2009 . Maritim Hotel, Berlin www.gtforum.com/ertc-annual-meeting

Join the biggest oil refining meeting in Europe

Sponsor companies

In what ways can you lessen the impact of the economic crisis on petroleum products’ demand?

ERTC

What are the latest developments in heavy oil upgrading? Sustainable solutions for sulfur from DuPont and Veolia

How should we manage and operate refineries in a CO2 emission-constrained world?

Media partners HYDROCARBON PROCESSING®

Get the answers at the ERTC 14th Annual Meeting

3 easy ways to book now and save = C 400 Quote HCP07 to get your discount Online: www.gtforum.com/ ertc-annual-meeting Call: +44 (0)870 240 8859 Email: saffa.khan@gtforum.com

ERTC Annual Meeting delegates learn about the latest technologies and processes driving change in our industry. Topics for discussion include: • Biofuels regulation and their impact on your refinery processes • Maximisation of distillates production and minimisation of gasoline production • Latest market evolution in heavy fuel oils • Energy efficiency and novel approaches to recover low temperature heat

New for 2009 Celebrate the industry’s achievements at the Gala Dinner To mark the dynamism and continuing success of our industry – despite some challenging times – we will host the first ERTC Annual Meeting Gala Dinner. Discuss the issues raised during the day and network over fine wine and good food.

Select 58 at www.HydrocarbonProcessing.com/RS

PROCESS TECHNOLOGIES integrated and nonintegrated process. As shown in Table 1, there are both capital and operating cost advantages to integration. Biorefinery complex. As demonstrated by the production of polyhydroxybutyric acid (PHB), a sugar cane mill producing sugar and ethanol can be integrated with an ethanol-to-EO plant.5 The dehydration plant will be designed to produce excess ethylene that will be used to produce downstream ethylene derivatives such as VCM or HDPE. The EO produced from the ethanol will be used to produce ethanolamines. The concept of fully integrating a chemical plant with a sugar mill incorporates key aspects of using readily available renewable sources such as bagasse to produce steam and electricity. Power/utilities benefits. Producing EO and ethanolamines consumes substantial steam and electricity. However, the sugar mill can be used as the main source to provide energy requirements by using the residue material (mainly bagasse) to generate needed energy and at the same time serving as a means of fulfilling waste-disposal and environmental requirements.5 It is possible that the wastewater generated from the chemical processes can be sprayed on the sugar cane fields and used as fertilizer.5 Another advantage of this complex is that the net carbon balance is close to zero since all of the carbon used in the process, both as raw material feedstock and as fuel for energy generation, comes from the sugar cane, and carbon dioxide emitted from the complex is reabsorbed by the sugar cane during its growth. Case history. The design basis for the complex is to produce 50,000-metric tpy (mtpy), ethanolamines and 60,000 mtpy excess ethylene, that can be used for VCM and/or HDPE. To produce the desired quantity of final products, 90,000 mtpy of ethylene must be processed from the dehydration of ethanol. The 90,000-mtpy

ethylene corresponds to 156,500-mtpy ethanol (100%). In a typical sugar mill that is producing both refined sugar and ethanol, about 2 wt% of the raw sugar cane is converted to ethanol. Therefore, to yield 156,500 mtpy of ethanol, over 8 million mtpy of raw sugar cane must be processed. As stated earlier, the complex can be energy self-sufficient by using residue from the sugar cane (bagasse) to generate the needed energy for the mill and the downstream process units. Typically, about 29% of the sugar cane is left over as bagasse. This material can then be used to generate the electricity and steam necessary for the complex. Burning bagasse can generate about 2 tons of steam for every ton of bagasse consumed. In the past, the boilers and steam generators were typically run inefficiently to dispose of as much bagasse as possible. The steam requirements for a mill are usually about 550 kg of steam per ton of cane processed.5 Steam requirements were reduced to about 350 kg of steam per ton of cane by optimizing the mills for heat generation, increasing the use of secondary steam in the juice heaters and vacuum pan, and increasing the number of stages in multiple effect juice evaporators.5 Optimizations in the ethanol distillation towers also reduced mill steam demand. By improving the overall mill steam requirements, it is now possible to use excess energy generated by the burning of the bagasse to integrate downstream chemical processes such as ethanol-to-EO process. Fig. 9 is a schematic of the integrated complex, showing how steam generated from the bagasse operates the sugar mill and is integrated into both the ethanol-to-EO and EO-to-ethanolamines processing units.

HAVER & BOECKER

The Solution Provider

The cost-saving packing system

HAVER ® FFS DELTA NT

Future options. As shown in Fig. 9,

efficient integration of a sugar cane mill with typical chemical process units can be achieved. Optimizing the efficiency of the mill steam cogeneration unit in which the cane residue (bagasse) is burned can produce sufficient thermal and electrical energy to operate downstream chemical process TABLE 1. Comparison of integrated plants. Along with and non-integrated ethylene production the energy efficiency benefits provided Integrated Non-integrated through integration Reactor Single bed Multi-bed with designs, this concept inter-stage heating provides the other Ethylene purification Single caustic Multiple columns and benefits including: wash column cryogenic distillation • Shared environCatalyst performance 99.8% conversion 99% conversion mental facilities, such 99.4% selectivity 96% selectivity as bio-ponds and Capital cost factor 1 1.5 waste-heat facilities Select 163 at www.HydrocarbonProcessing.com/RS 䉴

Filling of free-flowing granular, micro-granular or powdery bulk materials into plastic tubular film bags made of PE or PP

HAVER & BOECKER, Germany Phone: +49 2522 30-271 E-mail: chemie@haverboecker.com

www.haverboecker.com

The designation ® indicates a registered trademark of HAVER & BOECKER OHG in Germany. Several indicated designations are registered trademarks also in other countries worldwide. M 914-E4 Select 163 at www.HydrocarbonProcessing.com/RS

PROCESS TECHNOLOGIES Sugar cane 1,074 tph

Milling

• Reduced fuel oil requirement by burning sugar cane waste products—every ton of bagasse used saves about 1.1 bbls of fuel oil • Lower carbon footprint since carbon source is bio-renewable • Applies proven technology and does not present any unique processing risks. HP

Bagasse 279 tph 240 tph Steam and electrical power generation 480 tph steam

Juice treatment

61.25 tph 376 tph steam

Ethanol production

Sugar production Ethanol 19.3 tph

Sugar 89 tph 24 tph

EO 4.8 tph

EOA 6.25 tph

18.75 tph steam

LITERATURE CITED Bohlmann, G. M. “Process economics of ethanol production in Brazil: An indepth, independent technical and economic evaluation by the PEP program,” June 27, 2007. Retrieved Dec. 3, 2008 from www.bbibiofuels.com/few/2007/ FEW07/FEW07-WS04--Bohlmann.pdf. César, M. N., “Chemicals from ethanol, “Presented at SRIC Brunch, San Antonio, March 2008. Diniz, P. , “Renewable energy for a better world,” Nov. 14, 2007. Retrieved Dec. 3, 2008 from www.gsb.stanford.edu/gmp/events/documents/COSAN_ diniz_pp.pdf Macedo, I. C. and L. A. H. Nogueira, “Balanço de energia na produção de açúcar e álcool nas usinas cooperadas,” Copersucar Technical Bulletin, 31:2227, 1985. Nonato, R. V., P. E. Mantellato, and C. E. V. Rossell, “Integrated production of biodegradable plastic, sugar and ethanol,” Applied Microbial Biotechnology, No. 57, pp. 1–5, 2001.

Michael Bruscino is the licensing manager for Scientific FIG. 9

An integrated biorefinery uses sugar cane to process EO and other downstream derivatives.

• Reduced total plot areas due to savings in utility, environmental and other shared offsites • Lower transportation costs • Less storage requirements for raw materials and products

Design (SD) Co. He is responsible for the sales and marketing of SD’s technologies and products used within the petrochemical industry. Previously, Mr. Bruscino was the technical director for SD’s EO catalyst and worked with the R&D, product development, technology groups in the development of new products, and provided sales and marketing services for the commercialization of these products along with technical service to clients. He holds a BS degree in biology from Seton Hall University, along with a BS degree in chemistry from Montclair State University and an MS degree in chemical engineering from Manhattan College.

In this economy, y, plant p optimization p is critical...

Gain the insight you need to

MAKE AN IMPACT

The Yaws Handbook of Thermodynamic Properties www.GulfPub.com/YawsThermodynamic

Valve Handbook, 2nd Edition www.GulfPub.com/ValveHandbook

Valve Selection Handbook www.GulfPub.com/ValveSelection

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

I AUGUST 2009 HYDROCARBON PROCESSING

Advanced Piping Design www.GulfPub.com/AdvPipeDesign

WinHeat 4 NEW FEATURES! www.GulfPub.com/winheat

Phone: +1-713-520-4428 l +1-800-231-6275 Email: svb@GulfPub.com Online Store: www.GulfPub.com

PIPING/FLUID FLOW

our distinct regimes were found to exist (namely sliding bed, saltation, heterogeneous suspension and homogeneous suspension) in slurry flow in pipelines depending upon the average flow velocity. In the literature, few correlations have been proposed for identifying these regimes. Regime identification is important for slurry pipeline design since it is a prerequisite for applying different pressure drop correlations in different regimes. However, available correlations fail to predict the regime over a wide range of conditions. Based on a databank of around 800 measurements collected from the open literature, a method has been proposed to identify the regime using support vector machine (SVM) modeling. Statistical analysis showed that the proposed method has an average misclassification error (AME ) of 1.5%. A comparison with selected correlations in the literature showed that the developed SVM method noticeably improved predicting regime over a wide range of operating conditions, physical properties and pipe diameters. Introduction. Transporting slurries through pipelines is com-

mon in the solids handling, mineral and petrochemical industries and its huge power consumption has been drawing attention in recent years. The need and benefits of accurately predicting slurry pipeline pressure drop during the design phase is enormous since it provides better selection of slurry pumps, better minimization of power consumption and helps to maximize economic benefit. Power consumption costs are a substantial portion of operational costs for the overall pipeline transport. For that reason great attention has been paid to reducing the hydraulic losses. The pressure drop correlations available in the literature are applicable to the particular regime for which they were developed. The correlations show bad prediction of pressure drop when they are applied for other regimes. Thus, regime identification becomes important for slurry pipeline design since it is a prerequisite for applying different pressure drop correlations in different regimes. Accurate prediction of slurry pressure drop and understanding different regime formation makes it possible to minimize energy and water requirements. Experimental observations have shown that different correlations should be used in each of the identifiable flow regimes. * Corresponding author

Although this is a logical approach, it is not straightforward to apply. The main difficulty arises because it is not easy to define the boundaries between the flow regimes. These boundaries are poorly defined because they are based on visual observations of particle motion in small laboratory pipelines. Many researchers have attempted to establish correlations among the relevant experimental variables that can be used to define the flow regime boundaries. These attempts have met with only limited success and an approach developed by Turian and Yuan is the most popular and promising.1 Turian’s approach claims to provide a completely self-consistent definition of the flow regime boundaries that results directly from the head loss correlations and no additional correlations are required to define the boundaries. The method has the additional advantage that it is based on a large database of reliable experimental data, and consequently, the method can be used with confidence for practical engineering work. A databank of around 800 measurements collected from the open literature was used in Turian and Yuan’s calculations and the results of flow regime prediction were found to be very poor (Table 1). These poor results motivated this work and an attempt has been made to develop a method to identify the flow regime. To facilitate design and scale-up of pipelines and slurry pumps, there is need for a correlation that can predict flow regime over TABLE 1. Systems and parameters studied12 Slurry system: coal/water, coal/brine, ash/water, copper ore/water, sand/water, gypsum/water, glass/water, gravel/water, iron/water, iron/kerosene, high density material/water, iron tailings/water, limestone/water, limonite/water, plastic/water, potash/brine, sand/ethylene glycol, nickel shot/water, iron powder/water, ore/ water Pipe diameter, cm

1.27–80

Particle diameter, cm

0.0017–0.868

Liquid density, gm/cm3

0.77–1.35

Solids density, gm/cm3

1.15–8.90

Liquid viscosity, gm/cm-sec

0.008–1.9

Solids concentration, volume fraction

0.005–0.561

Velocity, cm/s

18.22–456.28 HYDROCARBON PROCESSING AUGUST 2009

I 71

Stationary deposits with ripples

Blocked pipe

Volumetric solids concentration FIG. 1

Heterogeneous flow regimes in terms of speed versus volumetric concentration.

a wide range of operating conditions, physical properties and particle size distributions. Industry needs quick and easily implementable solutions. The model derived from first principles is no doubt the best solution. But in the scenario where the basic principles of regime identification modeling accounting for all the interactions for slurry flow are absent, the numerical model may be promising to give some quick, easy solutions for slurry flow regime prediction. The field of machine learning is expanding, and many new technologies are growing using these principles. Among the various existing algorithms, one of the most recognized is the so-called SVM

I AUGUST 2009 HYDROCARBON PROCESSING

FIG. 2

Water

Symmetric ﬂow

Lenticular deposits

Asymmetric ﬂow

Suspended with saltation

Moving bed

Velocity

Suspended with moving bed

Slurry 1

Stationary bed

Fully suspended

Pressure drop per unit of length

PIPING/FLUID FLOW

Flow speed

Transitional mixture velocity and pressure drop.

for classification which permits creating systems that, after training from a series of examples, can successfully predict the output at an unseen location performing an operation known as induction. Since the present form of SVM was proposed, SVMs have been used in various application areas, and their classification and regression power have been investigated in depth from both experimental and theoretical points of view.2–8 The SVM is a modern mechanism for two-class classification, regression and

Select 165 at www.HydrocarbonProcessing.com/RS

In addition to tube ﬁttings, we also make valves, regulators, ﬁlters, and happier customers.

Contrary to what you may think, we’re much more than a tube ﬁtting company. And we have our obsession with Customer Focus to thank for that. Yes, we’re known throughout the world for our tube ﬁttings. And yes, we’ve been at it for over 60 years. But when companies are looking harder than ever for greater value, it’s our broad range of products, including orbital welders, modular systems, and a complete line of hose, that helps us offer more than you expect. See for yourself at swagelok.com/moreproducts.

Select 63 at www.HydrocarbonProcessing.com/RS

Safety

S C HAN

Energy & Environment

Automation and Control Technology and Technique Discover the latest developments, advancements, and practices in key instrumentation, automation, and control technologies at ISA EXPO 2009, the most significant conference, exhibition, and educational event for manufacturing professionals. Learn from world-renowned experts who are defining the next generation of technology, speak directly with the leaders who are changing the way businesses operate their plants and produce their products, and benefit from the lessons learned by others. Hear all sides of the automation story at North America's only unbiased, comprehensive automation and control exchange. Register now for ISA EXPO 2009 in Houston, TX, 6-8 October at www.isa.org/isaexpo. Enter promotion code W12319 for FREE Exhibition Registration.

ISA EXPO 2009 Endorsing Organizations: 速

Select 81 at www.HydrocarbonProcessing.com/RS

SIX

Wireless & Networking

EV NE EN

Security

Process Automation & Control

Enterprise Integration

PIPING/FLUID FLOW clustering problems. SVM, developed by Vapnik,9,10 is gaining popularity in the classification and regression fields due to many attractive features and promising empirical performance. Compared with traditional neural networks, SVM possesses prominent advantages: A strong theoretical background provides SVM with high generalization capability and can avoid local minima. SVM always has a solution that can be quickly obtained by a standard algorithm (quadratic programming). SVM need not determine network topology in advance, which can be automatically obtained when the training process ends. SVM can be regarded as a representation of information reducing (dimension reducing). It can solve high-dimension problems and, therefore, avoid the “curse of dimensionality.” Building on these studies, the focus of this work is to develop a unified correlation for predicting flow regimes in pipelines that can be useful for design engineers. This correlation has been derived from a broad experimental data bank collected from the open literature (800 measurements covering a wide range of pipe dimensions, operating conditions and physical properties). Based on the potential of SVM to classify complex functions, an attempt has been made in the present study to explore the computational capability of SVM to predict slurry flow regime. Four flow regimes for settling slurries. To develop the

regime identification procedure, it is necessary to understand the four flow regimes for settling slurries. These are (Fig. 1): • Flow with a sliding bed • Flow with a saltation • Heterogeneous mixture with all solids in suspension • Pseudo homogeneous or homogeneous mixtures with all solids in suspension. The tendency that the solid particles have to settle under the influence of gravity has a significant effect on the slurry behavior that is transported in a horizontal pipeline. The settling tendency leads to a significant gradation in the slurry solids concentration. The solids concentration is higher in the lower sections of the horizontal pipe. The extent of the solids accumulation in the lower section depends strongly on the slurry velocity in the pipeline. The higher the velocity, the higher the turbulence level and the greater the ability of the carrier fluid to keep the particles in suspension. The upward motion of eddy currents transverse to the main flow direction is responsible for maintaining the particles in suspension. At very high turbulence levels, the suspension is almost homogeneous with very good solids dispersion, while at low turbulence levels the particles settle toward the channel floor and can, in fact, remain in contact with the flow and are transported as a sliding bed under the influence of the pressure gradient in the fluid. Between these two behavior extremes, two other more or less clearly defined flow regimes can be identified. When the turbulence level is not high enough to prevent any particle deposition on the channel floor, the flow regime is described as being a heterogeneous suspension. As the slurry velocity is further reduced, a distinct transport mode known as saltation develops. In the saltation regimes, there is a visible layer of particles on the channel floor and these are continually picked up by turbulent eddies and dropped to the floor again further down the pipeline. The solids, therefore, spend some of their time on the floor and the rest in suspension in the flowing fluid. Under saltation conditions the solids concentration is strongly nonuniform. The flow regime depends strongly on the particle size and density that

make up the slurry. For example, a higher turbulence concentration is required to keep larger and heavier particles in suspension than is required for smaller and less dense particles. The four flow regimes can be represented by a plot of the pressure gradient versus the average mixture speed (Fig. 2). The transitional velocities are defined as: • V1: velocity at or above which the bed in the lower half of the pipe is stationary. In the upper half of the pipe, some solids may move by saltation or suspension. • V2: velocity at or above which the mixture flows as an asymmetric mixture with the coarser particles forming a moving bed. • V3 or Vc: velocity at or above which all particles move as an asymmetric suspension and below which the solids start to settle and form a moving bed. • V4: velocity at or above which all solids move as a symmetric suspension. Head loss correlations for separate flow regimes.

The saltation and heterogeneous suspension regimes have been studied most widely and the best known correlation for the excess pressure gradient due to the presence of solid particles in the slurry is due to Durand, Condolios and Worster which is given by:11

Pf ,sl Pfw Pfw

= C ( C D Fr ) 1.5

where Ω is a constant, C is the volumetric fraction of solids in the suspension, CD is the drag coefficient at terminal settling

www.GulfPub.com/EnhancedOilRecovery Email: svb@GulfPub.com Tel: +1 713.520.4426 | +1 800.231.6275 Select 166 at www.HydrocarbonProcessing.com/RS

PIPING/FLUID FLOW velocity and Fr is the Froude number. The value to be used for the constant Ω is uncertain and values between 65 and 150 are reported in the literature. Because this correlation does not apply to all flow regimes, the experimental data cannot be used to fix the value more precisely. Errors of 100% and more in the calculated value of ␾ can result. While the Durand- Condolios- Worster correlation is useful in the heterogeneous suspension flow regime, it deviates more and more from actual conditions in the other flow regimes. Experimental observations have shown that different correlations should be used in each of the identifiable flow regimes. Using the experimental data, Turian and Yuan established that the excess pressure gradient in each flow regime can be correlated using an equation of the form:

fsl fw = KC fw C D* Fr

(1)

The coefficients K, ␣, ␤, ␥ and ␦ have values that are specific to each flow regime. Using experimental data gathered from experiments in each flow regime, the best available values of these parameters in each flow regime are given by: Sliding bed (regime 0):

fsl fw = 12.13C 0.7389 fw 0.7717C D* 0.4054Fr 1.096

(2)

Heterogeneous suspension (regime 2):

fsl fw = 30.11C 0.868 fw 1.200C D* 0.1677 Fr 0.6938

(4)

Homogeneous suspension (regime 3): fsl fw = 8.538C 0.5024 fw 1.428C D* 0.1516 Fr 0.3531

(5)

Fairly consistent trends in the variation of the correlating parameters can be seen in the four correlations. Here C *D is the drag coefficient at the slurry terminal settling velocity. Regime boundaries (Turian and Yuans approach).1

The flow regime boundaries are defined in a self-consistent manner by noting that any two regimes are contiguous at their common boundary and, therefore, each of the two correlation equations must be satisfied simultaneously. For example, the boundary between the sliding bed regime (regime 0) and the saltation regime (regime 1) must lie along the solution locus of the equation: 12.13C 0.7389 fw 0.7717C D* 0.4054 Fr 1.096 = 107.1C 1.018 fw 1.046C D* 0.4213Fr 1.354

which is simplified to: 2

Saltation (regime 1):

Fr =

fsl fw = 107.1C 1.018 fw1.046C D* 0.4213Fr 1.354

15 October 2009

(3)

V = 4679C 1.083 fw1.064C D* 0.0616 Dg[s 1]

The regime transition number for transitions between regimes

•

The Houstonian Hotel

Awards DESIGNING A

BRIGHTER TOMORROW RESERVE YOUR TABLE TODAY www.WorldOil.com/Awards Sponsors |

(6)

Media Sponsor |

Select 167 at www.HydrocarbonProcessing.com/RS

Supporting Organization|

PIPING/FLUID FLOW TABLE 2. Regime identification by Turian and Yuan approach

VT*

Rep*

R 01

R 02

R 03

R 12

R 13

R 23

Regime identified by Turian

0.0043

3.48

0.31

91.18

1.15

11.89

0.03

0.06

0.34

0.24

0.12

0.0198

12.40

0.22

39.35

1.79

0.66

1.75

0.55

1.19

0.58

0.14

571

0.0162

1.48

0.08

0.10

256.98

0.13

3.16

2.25

0.91

1.08

1.51

1.40

1,808

0.0121

6.25

0.08

0.10

256.98

0.75

9.82

7.38

3.60

4.09

5.25

168.02

540,919

0.0029

6.01

21.27

364,749.95

0.46

4.30

0.04

0.02

0.25

0.08

0.01

35.42

21451

0.0065

1.33

7.17

25,924.56

0.46

1.30

0.01

0.09

0.04

0.01

9.54

133,201

0.0041

0.67

1.85

1,801.85

0.41

0.32

0.01

0.00

0.03

0.01

0.00

3.99

444,889

0.0031

1.29

0.52

209.55

0.78

8.05

0.01

0.11

0.07

0.03

2.71

712,183

0.0027

2.80

0.27

73.84

1.27

3.10

0.03

0.02

0.18

0.08

0.02

1.84

183,357

0.0038

0.47

0.13

24.73

2.36

4.07

0.01

0.02

0.07

0.06

40.57

219,153

0.0037

2.36

8.01

33,165.86

0.46

1.13

0.02

0.01

0.10

0.03

0.00

121.96

482,163

0.0030

1.90

17.44

217,110.66

0.47

2.27

0.01

0.09

0.04

0.01

8.32

73,124

0.0048

3.84

1.55

1,296.18

0.43

0.51

0.06

0.01

0.14

0.04

0.00

8.45

120,243

0.0042

1.32

1.58

1,271.32

0.43

0.31

0.02

0.01

0.05

0.01

0.00

8.45

120,243

0.0042

1.32

1.58

1,271.32

0.43

0.25

0.02

0.00

0.05

0.01

0.00

4.86

9,702

0.0080

0.69

0.71

83.18

1.20

0.08

0.03

0.01

0.04

0.02

0.00

4.86

9,915

0.0079

0.72

0.71

83.18

1.20

0.07

0.03

0.01

0.04

0.02

0.00

4.50

7,283

0.0086

0.39

0.63

68.21

1.33

0.35

0.01

0.04

0.03

0.01

4.50

8,295

0.0083

0.51

0.63

68.21

1.33

0.18

0.02

0.01

0.04

0.02

0.01

4.50

8,238

0.0083

0.50

0.63

68.21

1.33

0.11

0.02

0.01

0.04

0.02

0.00

Sly No.

dp*

Rew

2.93

114,476

2.41

251

1.40

Regime identified by experiments

4.50

9,153

0.0081

0.62

0.63

68.21

1.33

0.06

0.03

0.01

0.04

0.01

0.00

0.50

1,170

0.0135

0.67

0.01

0.02

1,555.21

0.14

2.89

3.96

0.89

1.67

5.74

0.50

1,035

0.0139

0.52

0.01

0.02

1,555.21

0.08

2.56

2.90

0.66

1.14

3.37

4.29

12,147

0.0075

1.74

0.58

47.56

1.61

0.69

0.06

0.04

0.16

0.08

0.02

4.29

12,467

0.0075

1.84

0.58

47.56

1.61

0.47

0.07

0.04

0.15

0.07

0.02

4.29

12,531

0.0075

1.86

0.58

47.56

1.61

0.35

0.08

0.03

0.14

0.06

0.01

7.06

36,156

0.0057

0.57

1.23

889.04

0.47

8.74

0.00

0.01

0.06

7.06

43,904

0.0055

0.84

1.23

889.04

0.47

13.57

0.00

0.02

0.09

57.61

320,958

0.0033

1.81

10.48

61,613.94

0.47

8.72

0.01

0.12

0.07

0.03

24.00

80,400

0.0047

2.93

5.09

12,463.99

0.43

0.94

0.03

0.01

0.13

0.04

0.00

24.00

63,624

0.0050

1.83

5.09

12,463.99

0.43

0.55

0.02

0.01

0.08

0.03

0.00

6.67

65,755

0.0049

0.63

1.14

759.31

0.49

6.83

0.00

0.01

0.06

0.05

0.04

6.68

88,261

0.0046

1.13

1.14

762.30

0.49

3.17

0.01

0.08

0.05

0.02

6.64

98,422

0.0045

1.42

1.13

748.45

0.49

1.64

0.01

0.08

0.04

0.01

7.06

107,380

0.0044

0.63

1.23

889.04

0.47

0.15

0.01

0.00

0.03

0.01

0.00

7.06

114,246

0.0043

0.71

1.23

889.04

0.47

0.17

0.01

0.00

0.03

0.01

0.00

11.30

383,507

0.0032

7.99

2.28

2,629.26

0.39

32.74

0.03

0.05

0.47

0.25

0.08

11.30

218,234

0.0037

2.59

2.28

2,629.26

0.39

4.16

0.01

0.13

0.06

0.01

11.30

230,107

0.0036

2.88

2.28

2,629.26

0.39

1.99

0.02

0.01

0.12

0.04

0.01

6.68

88,261

0.0046

1.13

1.14

762.30

0.49

3.17

0.01

0.08

0.05

0.02

6.64

98,422

0.0045

1.42

1.13

748.45

0.49

1.64

0.01

0.08

0.04

0.01

0 and 1 is defined by: R01 =

Fr 4679C 1.083 fw 1.064C D* 0.0616

(7)

and this number must be unity on the boundary between these two regimes. The transition numbers for the other possible transitions are found in the same way and are given by:

R02 =

R12 =

Fr 0.1044C

0.3225

6.8359C

0.2263

fw 1.065C D* 0.5906

Fr fw 0.2334C D* 0.3840

HYDROCARBON PROCESSING AUGUST 2009

(8)

(9)

I 77

PIPING/FLUID FLOW >1 Start

R01 <1 R02

R12

<1 >1

R13

<1 R03 <1

Not 1, 2 or 3 regime is 0 FIG. 3

R13 = R23 = R03 =

R23

Not 0, 1 or 2 regime is 3

R23

Not 0, 1 or 2 regime is 3

Not 0, 1 or 3 < 1 regime is 2

Not 0, 1 or 2 regime is 3

Not 0, 2 or 3 regime is 1

>1 <1

H1 H2 Support vectors

Not 0, 1 or 2 regime is 3

Not 0, 1 or 3 regime is 2

Margin

Decision tree for establishing flow regime.

Fr 12.522C

0.5153

fw 0.3820C D* 0.5724

Fr 40.38C 1.075 fw 0.6700C D* 0.9375 Fr 1.6038C

0.3183

fw 0.8837C D* 0.7496

(10) (11) (12)

Origin

FIG. 4

-b w y=0

Separation of two classes by SVM.

These numbers define the boundaries between any two flow regimes a and b by the condition Rab =1. It is possible to identify the regime that applies to a particular set of physical conditions quite simply from a knowledge of the transition numbers Rab . If a < b the value of Rab increases monotonically as the velocity increases. At low velocities, Rab < 1 and with increasing velocity, the value of Rab will eventually pass through 1.0. This must signal a transition out of regime a. The following simple rules will fix the flow regime: If Rab < 1 the regime is not b. If Rab > 1 the regime is not a. These inequalities must be tested for the combinations of a and b as shown in Fig. 3. No more than three of the transition numbers need to be calculated to fix the flow regime uniquely. Notice that these rules test the flow regimes negatively and a single test will never suffice to define the flow regime. It is always necessary to test at least three different combinations of a and b to get a definitive identification of the flow regime (Fig. 3). The applicable flow regime can be identified quickly and easily using the figure and the appropriate equation can be selected from the equations to calculate the slurry friction factor. A computer program of the method was developed and the results are shown in Table 2. The following calculation algorithm is used: 1. The drag coefficient at terminal settling velocity, CD*, is evaluated by balancing drag and buoyancy force: * dp3 ( s f )g C D = f VT2 dp 2 4 6 2

(13)

4( s f )gdp (14) C D* = 3 f VT 2 The particle Reynolds number at terminal settling velocity is given by: 78

Select 168 at www.HydrocarbonProcessing.com/RS

PIPING/FLUID FLOW TABLE 3. Some of the input and output data for SVM training Input to SVM Solid density, Fluid density, gm/cc gm/cc

SL No

Particle dia, cm

Solid conc., vol. fraction

0.0130

0.0280

1.8340

0.0100

0.3000

0.2200

0.3040

0.2200

0.8680

6 7

Output Fluid viscosity, gm/cm-sec

Pipe dia, cm

Fluid velocity, cm/sec

Regime

0.9980

0.0098

7.6200

147.22

2.8200

0.9820

0.0130

5.0000

332.50

–1

1.6700

1.3250

1.5200

7.6200

85.98

–1

0.3040

1.6700

1.3250

1.5200

10.1600

204.14

–1

0.1650

1.5300

0.9980

0.0098

20.8000

254.85

0.1830

0.0560

1.5300

0.9980

0.0098

4.0000

52.55

–1

0.0300

0.1700

3.3600

0.9980

0.0098

10.3000

126.73

–1

0.0140

0.0220

2.6900

0.9980

0.0098

20.7000

210.62

–1

0.0090

0.1250

3.0000

0.9980

0.0098

20.7000

337.16

0.0060

0.0140

3.1000

0.9980

0.0098

14.9000

120.59

0.1113

0.1920

4.5500

0.9980

0.0098

8.2500

260.32

0.3346

0.1000

4.5500

0.9980

0.0098

15.0000

315.00

0.0390

0.4750

1.9840

1.1460

0.0114

5.2200

138.99

–1

0.0410

0.3220

1.9840

1.1420

0.0120

10.7600

117.23

–1

0.0410

0.3830

1.9840

1.1420

0.0120

10.7600

117.23

–1

0.0540

0.3600

2.6500

1.3500

0.0560

5.2450

76.73

–1

0.0540

0.4200

2.6500

1.3500

0.0560

5.2450

78.41

0.0500

2.6500

1.3500

0.0560

5.2450

57.60

0.0500

0.1200

2.6500

1.3500

0.0560

5.2450

65.60

0.0500

0.1800

2.6500

1.3500

0.0560

5.2450

65.15

–1

0.0500

0.4200

2.6500

1.3500

0.0560

5.2450

72.39

–1

0.0200

0.1800

2.6500

1.1320

0.3820

5.2450

75.29

–1

0.0200

0.2400

2.6500

1.1320

0.3820

5.2450

66.60

–1

0.0490

0.1200

2.6500

1.0950

0.0575

5.2450

121.61

0.0490

0.1800

2.6500

1.0950

0.0575

5.2450

124.81

0.0490

0.2400

2.6500

1.0950

0.0575

5.2450

125.45

0.0250

0.0050

2.6500

0.9980

0.0098

5.1500

68.80

–1

0.0250

0.0050

2.6500

0.9980

0.0098

5.1500

83.54

–1

0.2040

0.0250

2.6500

0.9980

0.0098

15.0000

209.68

–1

0.0850

0.2150

2.6500

0.9980

0.0098

5.0800

155.10

0.0850

0.2150

2.6500

0.9980

0.0098

5.0800

122.73

0.0242

0.0080

2.6000

0.9990

0.0100

7.6000

86.52

0.0242

0.0300

2.6000

1.0000

0.0100

7.6000

115.90

–1

0.0242

0.0700

2.6000

1.0000

0.0101

7.6000

130.15

–1

0.0250

0.3000

2.6500

0.9980

0.0098

10.3000

102.16

–1

0.0250

0.3180

2.6500

0.9980

0.0098

10.3000

108.70

0.0400

0.0300

2.6500

0.9980

0.0098

10.3000

364.88

0.0400

0.0620

2.6500

0.9980

0.0098

10.3000

207.63

–1

0.0400

0.1370

2.6500

0.9980

0.0098

10.3000

218.93

–1

0.0242

0.0300

2.6000

1.0000

0.0100

7.6000

115.90

–1

0.0242

0.0700

2.6000

1.0000

0.0101

7.6000

130.15

–1

dpVT f (15) μf It is not possible to solve CD* from Eq. 13 directly because it is a function of terminal settling velocity, VT , and particle diameter, dp, through the relationship of Abraham or Turton-Levenspiel. This problem can be solved by introducing a dimensionless number, ␾ 1*, as follows: Re *p =

4( s f ) fg 3 dp 1* = C D* Re *p = 3μ f 2

(16)

As per Karamanev (1996) CD* can be calculated as: 432 0.517 C D* = * 1+ 0.0470 1*0.66 + 1+154 1* 0.33 1

(17)

(

)

HYDROCARBON PROCESSING AUGUST 2009

I 79

PIPING/FLUID FLOW Performance check of Turian Yuan’s approach. As

mentioned earlier, over the years researchers have amply quantified the flow regime of slurry flow in pipelines. In this work, about 800 experimental points have been collected from 20 sources from the open literature spanning 1950–2002. The data were screened for incompleteness, redundancies and evident inaccuracies. This wide-range database includes experimental information from mainly two regimes namely saltation (regime 1) and heterogeneous flow (regime 2). These two regimes were selected since they have great practical significance and all practical hydrotransport is carried out in the heterogeneous suspension region because it has the lowest pressure drop and subsequently less power requirement. Table 1 indicates the wide range of the collected databank for regime identification. Table 2 shows some of these data. Regime identification. Now Table 2 data were exposed to Turian and Yuan’s calculation and the regimes were identified as shown in Table 3. It is evident from the last two columns of Table 3 that the Turian Yuan approach did not produce promising results and failed to correctly identify the regimes. These poor results motivated this work and an attempt has been made to explore the new support vector classification methodology to identify the flow regime correctly. SVM modeling. Basically, identifying regimes is a classification problem and recently SVM has emerged as a great tool for solving classification problems. This technique can be subdivided into two distinct parts: • Learning: consists of training the SVM with examples at its disposition • Prediction: where new samples are inserted in which the result is not known. Each example can be written as a pair (input, output) where input is the data set and output makes up how it should be cataloged. Mathematically, examples can be regarded as a pair (x, y) where x is a vector of real numbers and the output can be a Boolean value (like as yes/no, 1/–1, true/false) or a real number.

In the first case, we speak of a classification problem, in the latter of a regression problem. For conventional purposes, the group of examples that make up the SVM is called the training set, whereas the group that contains the examples used in the prediction is called the test set. This name derives from the fact that generally, error is calculated based on the total results in the test and this in turn measures the system quality. In simple terms, the SVM can be thought of as creating a line, or hyperplane between two data sets. If we imagine a twodimensional case, the action of the SVM can be shown easily. In Fig. 4, a series of data points for two different classes of data are shown, black circles (class A) and white squares (class B). The SVM attempts to place a linear boundary between the two different classes and orientate it in such a way that the margin (represented by the dotted lines) is maximized. In other words, the SVM tries to orientate the boundary in such a way as to ensure that the distance between the boundary and the nearest data point in each class is maximal. The boundary is then placed in the middle of this margin between the two points. The nearest data points are used to define the margin, and are known as support vectors (represented by the gray circles and square). Once the support vectors have been selected, the rest of the feature set is not required since the support vectors contain all the information need to define the classifier. Mathematics behind SVM algorithm. Mathematically, an SVM can be defined comparatively easily. The explanation that follows is an overview of the functioning of an SVM; for a more Divide the whole databank as 80% for training and 20% for testing (chosen randomly) Start SVM using a linear kernel Start with C = 10,000 Start with kernel parameter = 1

TABLE 4. Different kernel types Case

Name of kernel

Case 1

Linear

Case 2

Polynomial

Case 3

Gaussian radial basis function

Case 4

Exponential radial basis function

Case 5

Splines

Case 6

B splines

After completing training, evaluate the AME for test set Yes Change the kernel parameter and fresh start the new run

All the kernel parameters tested? Yes All the C values tested?

Change the C value and proceed for new run

Yes All kernel types tested?

Change the kernel type

Yes Find out which combination of kernel type, kernel parameter and C gives the lowest AME

Imput space FIG. 5

Nonlinear transformation from input to a higherdimensional feature space.

I AUGUST 2009 HYDROCARBON PROCESSING

Stop

Feature space FIG. 6

Flowchart for SVM algorithm implementation.

PIPING/FLUID FLOW

MisclassiďŹ cation, %

36.5

AME

30 17

20 10

3.5

1.5

erbf

Polynomial Linear Kernel type

rbf

SVM (AME) performance for different kernel types.

FIG. 7

detailed explanation, it is suggested that one of the several excellent tutorials is consulted. For any point that lies on our boundary line, we can write (w.x) + b = 1

where w is a vector that defines the boundary, x is an input, or data vector, and b is a scalar threshold value. At the margins, H1 and H2, where the support vectors are situated, the equations for class A and B, respectively, are: (w.x ) + b = 1 and (w.x ) + b = 1 w and b may have to be scaled for this to be the case. Therefore, as the support vectors correspond to the extremities of the data for a given class, anything that belongs to class A will conform to the equation (w.x ) + b and (w.x) + b 1 respectively, for class B. Combining these two functions, a decision function can be created to determine whether a given data point belongs to class A or B. This is defined as: f (x ) = sign[(w.x) + b] (18)

space). In SVMs, this is achieved through use of a transformation, called kernel, that transforms the data from an N-dimensional input space into a Q-dimensional feature space: s = (x) (20) Fig. 5 shows the possible effect that this transformation has on some fictitious feature set, and how the separability of the data changes after the transformation. Substituting the transformation into Eq. 19, this gives: f (x ) = sign( vi( (x) (xi)) + b)

(21)

Making transformations into higher-dimensional space is relatively computationally intensive to perform a dot product on the results. A kernel can be used to perform this transformation and the dot product in one step if the transformations that are allowed are restricted to those that can be replaced by an equivalent kernel function. In this way, it is possible to reduce the computational load but retain the effect of the higher-dimensional transformation. The kernel function is defined as:

K (x, y) = (x ) ( y) substituting into Eq. 21, we get the final basic form of the SVM: f (x ) = sign( viK (x, xi) + b) i

vi is used as a weighting factor to determine which of the vec-

In trying to find the best boundary between our data sets, we are in effect trying to find a solution of w that allows this. It can be shown that the solution is of the general form:9,10 n

w = vixi i=1

where xi are the support vectors that have been kept from training. Substituting this into Eq. 18, we get: f (x ) = sign( vi(x.xi) + b)

(19)

A case has been constructed for a linear boundary in two dimensions; however, there will be cases where the linear boundary in input space will be unable to separate two classes properly. At first this seems a large problem, however, by transforming the data into a higher-dimensional space, it is possible to create a hyper plane that allows linear separation in the higher dimension (which corresponds to a curved surface in the lower dimensional input Select 169 at www.HydrocarbonProcessing.com/RS

Easy Open Scalable

CADWorx

Plant Design Suite

PIPING/FLUID FLOW tors in the input are actually support vectors, i.e., 0 < vi < . Input vectors with a corresponding vi ¼ 0 or 1 are not support vectors, and may be for all intents and purposes, discarded. The kernel used for the experiments in this article is the RBF kernel, defined by the equation:

Why is CADWorx one of the fastest growing and most productive plant design suites on the market?

Ř 3LSLQJ Ř 6WHHO Ř &DEOH WUD\s/ducting Ř &ROOLVLRQ GHWHFWLRQ Ř (TXLSPHQW Ř %LOOV RI PDWHULDO Ř ,VRPHWULFV Ř )ORZ VFKHPDWLFV Ř ,QVWUXPHQW ORRS GLDJUDPV Ř %L GLUHFWLRQDO OLQNV WR DQDO\VLV SURJUDPV Ř :DONWKURXJK DQG YLVXDOL]DWLRQ

CADWorx delivers! &RQWDFW XV WR ILQG RXW KRZ \RX FDQ LPSURYH \RXU GHVLJQ HIILFLHQF\

+1 281-890-4566 sales@coade.com DOWNLOAD FREE DEMO

www.coade.com

Because it has all the tools to produce intelligent plant designs, including:

Select 170 at www.HydrocarbonProcessing.com/RS

K (x, y) = e

2 x y 2 2

The width of the RBF kernel parameter, s, can be determined by an iterative process that selects the optimum width based on the full feature sets. The RBF kernel was chosen after a number of empirical experiments, which indicated that the RBF kernel seemed to offer the best performance of a number of different kernels on this type of data. The previous introduction covers the case of separable data, for nonseparable data the case is slightly different. The values of vi, the support vectors, can be thought of as indicating the degree of ‘‘importance’’ that each support vector has to the overall boundary definition. By limiting the values that vi can take, this, therefore, limits the effect that any one support vector has on the overall solution, and can be used to control the generalization properties of the SVM by reducing the effect that outliers have on the boundary defined by the support vectors. For the nonseparable case, therefore, the constraint is 0 < vi <C . For a separable case, C can be thought of as equal to infinity, while for a nonseparable case, the value of C may be varied depending upon the number of allowable errors in the trained solution; few errors are permissible for high C values, while low C values will allow a higher proportion of errors in the solution, although with a comparatively smooth boundary between the classes. Thus, to control the generalization capability of the SVM there are a few free parameters: the limiting term C (also known as the misclassification penalization term) and the kernel parameters can be controlled by the user; the number of different kernel parameters depends on the type of kernel in use, however, for the RBF kernel, only one parameter is in use. The free parameters are set at the beginning of the training run, and can be used to alter the SVM performance as desired. The three main characteristics of SVM algorithms are: first, that they implement the approach called Structural Risk Minimization to reduce the generalization error; second, that they work on highdimensional feature spaces by means of

a dual formulation in terms of kernels; and third, that the prediction is based on hyper planes in these feature spaces, that may correspond to quite involved surfaces on the input space, and that can handle outliers in the training set by means of soft margins or soft tubes. Training and testing. Training an

SVM consists of an iterative process in which the SVM is given the desired inputs along with the correct outputs for those inputs. It then seeks to alter its margin, w, and bias, b, to try and produce the correct output (within a reasonable error margin). If it succeeds, it has learned the training set and is ready to perform upon previously unseen data. If it fails to produce the correct output it rereads the input and again tries to produce the correct output. The margins and bias are slightly adjusted during each iteration through the training set (known as a training cycle) until the appropriate margins and bias have been established. Depending upon the complexity of the task to be learned, many thousands of training cycles may be needed for the SVM to correctly identify the training set. Once the output is correct, w, and, b, can be used with the same SVM on unseen data to examine how well it performs. SVM learning is considered successful only if the system can perform well on test data on which the system has not been trained. This capability of a SVM is called generalizability. SVM-based classification for regime identification. Developing

the SVM-based regime identification had been started with collecting a large databank. The next step was to perform a support vector classification, and to validate it statistically. Data collection. As mentioned earlier,

about 800 experimental points have been collected from 20 sources from the open literature spanning 1950–2000. Table 3 shows some of these data used for support vector classification. Identifying input parameters. After

an extensive literature survey, all physical parameters that influence regime identification are put in a so-called “wish-list.” Based on the extensive literature survey, the input variables such as pipe and particle diameters, solids concentration, solids and liquid density, and viscosity and velocity of the flowing medium have been finalized to

PIPING/FLUID FLOW ■ The present study

focuses on classifying and identifying two major regimes: heterogeneous suspension and saltation. These two regimes are chosen purposely since pressure drop is minimum in the heterogeneous regime and thus, most important from a power consumption point of view. predict different regimes in a slurry pipeline. Some portion of the input and output data used for support vector classification are shown in Table 3. Support vector classification. The present study focuses on classifying and identifying two major regimes: heterogeneous suspension and saltation. These two regimes are chosen purposely since pressure drop is minimum in the heterogeneous regime and thus, most important from a power consumption point of view. The method developed here can be extended for classifying other regimes also. Initially, all the data related to these two regimes were collected from the open literature. Six parameters were identified as input (Table 3) to the SVM and the +1 or –1 is put as the target. The output was designated as +1 for heterogeneous suspension flow regimes data and –1 for saltation regimes data. Since the magnitude of inputs greatly differs from each other, they are normalized in –1 to +1 scales. Eighty percent of the total dataset was chosen randomly for training and the remaining 20% was selected for validation and testing. All the kernel types listed in Table 4 have been trained, validated and tested for best prediction using this databank. Three different parameters to be evaluated to design a successful classification model are: kernel type, C, kernel parameter, i.e., polynomial degree, etc. Since the prior knowledge is not there regarding the suitability of a particular value of any of the three parameters, the strategy adopted here is holistic and summarized in Fig. 6. The SVM performance was evaluated exhaustively for all parameter combinations.

All the kernel types listed in Table 4 are tested with all combinations of capacity control, C, and degree of kernel. The degree of kernel was varied from 1 to 6, capacity control varied from 10,000 to 0.1 (typically six values: 10,000, 1,000, 100, 10, 1 and 0.1). Each run was exposed with the same training and testing data and the percentage of misclassification (error %) was calculated for each run. The statistical analysis of SVM prediction is based on the following performance criteria: The average misclassification error (AME ) should be minimum. AME = (number of misclassification in test data/total number of test data)100 Optimizing parameters in regression. After collecting the large databank,

an exhaustive search was made to optimize the three SVM parameters so that prediction error, i.e., misclassification error (AME ) became minimum. Step 1: Start with a kernel type (say erbf kernel), assume C is 10,000 and the kernel parameter (␴ in this case) is 1. Run the SVM algorithm with 640 training data and find out the average misclassification error percentage on 160 test data (which were not used for training). Repeat the procedure by varying the kernel parameter from 1 to 6 and find the AME for each run. Step 2: Step 1 is repeated by varying C (typically six values: 10,000, 1,000, 100, 10, 1 and 0.1; total 6X6 runs) and AME for each run was calculated. Find the values of C and, ␴, that give the lowest AME. This may be considered as the best possible solution for that particular kernel type (erbf in this case) and is listed in Fig. 7. Step 3: All the kernel types given in table are exposed to the same input and output data and steps 1 and 2 are repeated for all the kernel types and summarized in Fig. 7. The best kernel performance was judged by AME. The other performance parameters, namely execution time and storage requirement, are neglected since they are not important for this type of study. Fig. 7 summarizes the performance of the different kernels. It is evident from Fig. 7 that the erbf kernel is the most promising since it gives the lowest AME (1.5%). It is evident from the table that the erbf Kernel with capacity control, C, of 10,000 and ␴ equal to 1 produces the best results among all other algorithms and predicts the velocity regimes with an average error of 1.5%. This can be considered as very good model agreement for the wide range

Do you have 20+ years of experience in hydrocarbon production or processing operations? Would you like to take your place on a team of dedicated Professionals?

Do you have a passion for teaching? International training company with over 40 years in the industry is seeking professional, self-motivated instructors for our Operator training program. You will have the opportunity to travel internationally and explore new cultures while teaching operator courses. The ideal candidate will have a Bachelor’s degree in engineering or science in addition to 20+ years of experience in hydrocarbon production/processing environments. He/she will understand process control & instrumentation, be able to read process/engineering drawings and use technical terminology correctly. Excellent communication and public speaking skills are required. Prior training experience a plus. Training is provided. Submit your resume to: trainingcorrespondance @gmail.com Select 171 at www.HydrocarbonProcessing.com/RS

PIPING/FLUID FLOW 6

of experimental data compared to the predictions by the standard available models, especially for slurry flow. Recall that all the 800 experimental data collected from the open literature were exposed to the Turian Yuan formulas for regime identification and the AME calculated was 25%. The present work has reduced the misclassification error from 25% to 1.5%. The advantage of the present method is that users don’t have to calculate the CD, settling velocity, Froude number, R01, R02, R03, R12, R13, R23, etc., as in case of Turian Yuan’s approach to evaluate the regime. In the present approach, regime will be evaluated more accurately from the basic slurry flow data (pipe dia., solids conc., fluid velocity, etc.) effortlessly and the user will be relieved to calculate the other parameters stated above. Once the regime has been evaluated correctly, appropriate pressure drop correlations (Eqs. 2–5) can be used for the each regime. This will help to choose the appropriate correlations for pressure drop and accurately predict the pressure drop in the design phase. HP

Scholkopf, B., Burges, J., Smola, A., “Advances in Kernel Methods: Support Vector Machine,” MIT Press, 1999. 7 Schölkopf, B., Platt, J. C., Shawe-Taylor, J., Smola A. J. and Williamson, R. C., “Estimating support of a high-dimensional distribution,” Neural Comput., 13, 1443–1471, 2001. 8 Smola, A., Murata, N., Schölkopf, B. and Muller, K., “Asymptotically optimal choice of epsilon–loss for support vector machines,” Proc. ICANN, 1998. 9 Vapnik, V., The Nature of Statistical Learning Theory, Springer Verlag, New York, 1995. 10 Vapnik, V., Statistical Learning Theory, John Wiley, New York, 1998. 11 Durand, R. and Condolios, E., compterendu des Deuxiemes journees de L Hydraulique, Paris, societe Hydrotechnique de France, 29–55 (1952). 12 Hsu, F. L, Turian, R. M, Ma, T. W, “Flow of noncolloidal slurries in pipelines,” AIChE Journal, 35, 429–442, 1989. 13 Karamanev, D. G., “Equations for the calculation of the terminal veolicty and drag coefficient of the solid spheres and gasa bubbles,” Chemical Engineering Communications, 174, pp. 75–85, 1996.

LITERATURE CITED Turian, R. M., and Yuan, T. F, “Flow of slurries in pipelines,” AIChE J., 23, 232–243, 1977. Agarwal, M., Jade, A. M., Jayaraman, V. K. and Kulkarni, B. D., “Support vector machines: A useful tool for process engineering applications,” Chem. Engg. Prog, 57–62 (2003). Burges, C., “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery, 2(2), 1–47 (1998). Cherkassy, V., Shao, X., Mulier, F. and Vapnik, V., “Model Complexity Control for Regression Using VC Generalization Bounds,” IEEE Transaction on Neural Networks, Vol. 10, No. 5, pp. 1075–1089, 1999. Hastie, T., Tibshirani, R. and Friedman, J., The Elements of Statistical Learning Data Mining Inference and Prediction, Springer, 2001.

Sandip Kumar Lahiri has 15 years’ technical services and

1 2

3 4

operation experience in petrochemical industries across the globe. He has an M.Tech qualification in chemical engineering and is doing research on multiphase flow and CFD. Mr. Lahiri’s interests include simulation, optimization, APC and soft sensor development.

Kartik Chandra Ghanta is a professor in the department of chemical engineering of the National Institute of Technology, Durgapur, India. He has 16 years of teaching and research experience. Dr. Ghanta’s fields of interest are slurry flow and modeling.

The economy is down. Projects are harder won. NOW

is the time to be better. In three full conference days you will have the chance to attend over 60 classes.

CADWorx

CAESAR II

PV Elite

™

TANK

™

Sessions are tailored from beginner to advanced levels. So, the piping designer that wants to know more about stress analysis. They’re covered. The seasoned stress engineer looking to improve the work flow between engineering and design. We have them covered too!

If you want to be a better designer, engineer or manager than you already are, this is the time.

September 28–30, 2009 Houston, TX ★ USA

The Woodlands Waterway Marriott Hotel & Convention Center

Join us! ★ Questions? Call the CUC2009 conference hotline at (+1) 781-803-3735.

www.coadeuserconference.com 84

I AUGUST 2009 HYDROCARBON PROCESSING

Select 172 at www.HydrocarbonProcessing.com/RS

HydrocarbonProcessing.com

WEBCAST Live Event September 10 , 2009 Heinz Bloch— Maintenance and Reliability Trends in the Refining, Petrochemical, Gas Processing and LNG industries Hydrocarbon Processing’s Reliability/Equipment Editor Heinz Bloch will be interviewed in his first webcast on maintenance and reliability trends in the refining, petrochemical, gas processing and LNG industries. With forecast maintenance spending of $59.1 billion in 2009 (source: HPI Market Data 2009) and concerns about profitability, efficiencies and the need to maintain operability, this webcast is a must watch for maintenance professionals worldwide. Heinz, as an editor for Hydrocarbon Processing for 10 years, has built a dedicated following worldwide in his area of responsibility. He holds six US patents and has authored over 460 technical papers and 17 books on machinery. He was an Exxon Chemical Co. machinery specialist and held positions worldwide before retiring after 24 years with Exxon. His deep personal and technical understanding in the area of maintenance and reliability and current trends will be presented via an interview with Les Kane, Hydrocarbon Processing’s Editor. The format will be three 15-minute segments followed by a live question-and-answer segment. To be a part of this exciting, one of a kind event for the HPI, visit www.HydrocarbonProcessing. com and click on the HP Webcast link to register for the one-hour live event on September 10, 2009. Attendees of the live event will have the opportunity to submit questions to the speaker. For those unable to make it, the webcast will be available on-demand for 12 months. Please contact Tiffany.Tosh@GulfPub.com if you have any questions or need further details.

20 09

HPWEBCASTS