,x ‰êk ~‰ ,xì ‘ i) An Overview of ‘Swaram’ A Language for Programming in Tamil
Overview Pacific Northwest
Pacific Northwest National LaboratoryOperated by Battelle for the U.S. Department of EnergyDeconMSn – A Software Tool for Determination of Accurate Monoisotopic Masses of Parent Ions of Tandem Mass SpectraAnoop M. Mayampurath1, Navdeep Jaitly1, Samuel O. Purvine1, Matthew E. Monroe2, Kenneth J. Auberry1, Joshua N. Adkins2, and Richard D. Smith2 1Environmental & Molecular Sciences Laboratory, 2Biological Sciences Division, Pacific Northwest National Laboratory, Richland, WA 99352OverviewWe present a new software tool for tandem MS analyses that: • accurately calculates the monoisotopic mass and charge of high–resolution parent ions • accurately operates regardless of the mass selected for fragmentation • performs independent of instrument settings • enables optimal selection of search mass tolerance for high mass accuracy experiments • is open source and thus can be tailored to individual needs • incorporates a SVM-based charge detection algorithm for analyzing low resolution tandem MS spectra • creates multiple output data formats (.dta, .MGF) • handles .RAW files and .mzXML formats • compatible with SEQUEST, MASCOT, X!TandemMethodsHigh Resolution PrecursorStartParent_MZNET 0.6Bscore2/3 0.01/0.03Xscore2/3/ratio 19794/0/01 0.8 0.6 0.4 0.2Xscore2/3_Loss (H20, NH3, CO) ….Peak Distribution [0:mz:2mz:3mz:4mz] 0.4:0.5:0.02:00.25 0.2 0.15 0.1 0.05Fscore2/3 9.1/5.2ConclusionsDeconMSn: Accurately determines the monoisotopic mass and charge of high resolution parent isotopic distributions Improves narrow mass tolerance-based SEQUEST searchesLow Resolution PrecursorProbability Distribution Value Start0.25 0.2 0.15 0.1 0.05 0 0818.92+3 +2Get ParentPeak from headerParentPeak Fragmentation SpectrumGet ParentMz from spectrumNext SpectraCS = 1? NY0.20.40.60.810 00.511.522.5 x 105.dta CS = 1NETXscore20 -50Fscore251015Figure 5 : Example probability distributions for selected feature vectors for each low-res MS/MS scan.Implements a SVM-based charge state detection algorithm to handle low resolution tandem mass spectra Incorporated as part of Decon2LS, which is available for download at /THRASHIntroductionPeptide identification through tandem MS is a commonly used technique in proteomic-based research. Analyzing MS/MS fragmentation results from Thermo Fisher Scientific mass spectrometers currently involves using the program extract_msn (Figure 1) to create .dta file representations of spectra. Each .dta file contains a list of observed peaks in the MS/MS spectra and the corresponding parent ion charge and protonated mass, which extract_msn determines based on spectra characteristics.Is THRASH a success? N .dta CS +2/3 Y .dta MonoMass/CS from CalCS .dta MonoMass/CS from THRASHGuess charge based on peak spacing (CalCS)Calculate SVM FeaturesResultsHigh Resolution Precursor SpectrumMass Differential (DelM) vs. SEQUEST cross correlation (Xcorr) (filtered for fully tryptic peptides, DeltaCn >0.1, forward sequences only) Histogram of Mass Differential (delM) populationAcknowledgementsThis research was supported by the National Institute of Allergy and Infectious Disease (NIAID) and the National Center for Research Resources (NCRR). Experimental portions of this research were performed at the Environmental and Molecular Sciences Laboratory (EMSL), a U.S. Department of Energy (DOE) national scientific user facility located at the Pacific Northwest National Laboratory (PNNL) in Richland, Washington. PNNL is a multi-program national laboratory operated by Battelle for the DOE under Contract No. DE-AC05-76RLO 1830.NIs CalCS a success?NLast Spectra?YNormalize Determine Charge from SVMYCreate .dtas for each spectraFigure 4 : Flowchart describing the DeconMSn algorithm for a low-resolution tandem MS dataset. DeconMSn overcomes the lack of peak profile information by incorporating a supportvector-machine-based charge detection algorithm that identifies the most likely charge of a parent ion. • A trained support vector machine (SVM) [6] is used to assign a charge (+1, +2, +3, or +4) to a scan based on feature vector valuesDeconMsnReferences166 1660 76Figure 1: Thermo’s software tool “extract_msn” Problem: The instrument acquisition software can record the wrong parent monoisotopic mass (e.g., Figure 2). Thus, a .dta file is created with the monoisotopic mass +/-n amu from the correct mass (n being an integer).Both .dtasNIs THRASH_CS = CalCS_CS?• Figure 5 shows the feature vector of 19 features based on [7] that is calculated for an example MS/MS scan • Ambiguous spectra are assigned charge states +2 and +3Y Choose the lower mono_massFigure 2: Example illustrating -1 amu difference between mono mass in dta and actual mono massFigure 3: Flowchart describing the DeconMSn algorithm for high-resolution precursor dataset for each MSn spectrum. DeconMSn identifies the parent spectrum and m/z peak distribution from the high-resolution raw spectra, and then extracts the monoisotopic neutral mass through deisotoping, regardless of which mass was chosen for fragmentation • Same core as Decon2LS [2] (written in C++ on .NET platform) • Autocorrelation based peak fitting routines [3] are used for charge state detection. • Averagine [4] and Mercury[5] used to create theoretical profiles which are matched to observed profiles • CalCS determines charge by stepping from the parent peak by a distance calculated as 1.003/(cs) to determine the existence of a peak.231788721. Horn, D.M., et al. Automated Reduction and Interpretation of High Resolution Electrospray Mass Spectra of Large Molecules. J. Am. Soc. Mass Spectrom. 2000, 11, 320-332. 2. Jaitly, N., et al. Open Source Tools for the Accurate Mass and Time (AMT) Tag Proteomics Pipeline, Proc. of ASMS 2006, Seattle, WA. 3. Senko, M. W., et al. Automated assignment of charge states from resolved isotopic peaks for multiply charged ions. J. Am. Soc. Mass Spectrom. 1995, 6, 52–56. 4. Senko, M. W., et al. Determination of monoisotopic masses and ion populations for large biomolecules from resolved isotopic distributions. J. Am. Soc. Mass Spectrom. 1995, 6, 229–233. 5. Rockwood, A. L., et al. Rapid Calculation of Isotope Distributions. Anal. Chem. 1995, 67, 2699–2704. 6. Kanu S., et. al. SVM And Kernel Methods Matlab Toolbox. Perception Systemes et Information, INSA de Rouen, 2005 France 7. Klammer, A. A., et al. Peptide charge state determination for lowresolution tandem mass spectra. Proc. of 2005 IEEE Comp. Systems Bioinformatics. Conf. 2005 Stanford, CT..dtaLow Resolution Precursor Spectrum47 filtered hits unique to extract_msnContact InformationAnoop M. MayampurathEnvironmental Molecular Sciences Laboratory Pacific Northwest National Laboratory P.O. Box 999, Richland, WA 99352 e-mail: anoop.mayampurath@ Phone: (509) 376-52672437 filtered hits commonOur solution: Use a combination of THRASH [1] and chargebased peak finding routines to deisotope the parent isotopic distribution.98% of all filtered peptides found49 filtered hits unique to DeconMSn。
shiva英文介绍
shiva英文介绍Deep within the rich tapestry of Hindu mythology, Shiva stands as a towering figure, embodying the primal forces of creation, destruction, and preservation. His name resonates with awe and reverence, symbolizing the ultimate power and transcendent wisdom. In this essay, we delve into the enchanting world of Shiva, exploring his various avatars, attributes, and the profound impact he has had on the Hindu religion and culture.Shiva, often referred to as the "Destroyer," represents the force of transformation and rebirth. He is one of the Trimurti, the three primary deities in Hinduism, alongside Brahma (the Creator) and Vishnu (the Preserver). Each deity embodies a different aspect of the divine, creating a harmonious balance in the universe.Shiva's iconography is rich and diverse, often depicting him with a snake coiled around his neck, symbolizing the infinite cycle of life and death. His third eye, located between his eyebrows, represents his ability to perceive the truth beyond the illusions of the world. He is often shown wearing a crown of snakes and holding atrident, a weapon that represents the three qualities of nature: sattva (goodness), rajas (passion), and tamas (darkness).One of Shiva's most famous avatars is Ardhanarishvara,a form that embodies both the masculine and feminine principles. This duality represents the union of Shiva and his consort, Parvati, symbolizing the harmony of opposing forces. Ardhanarishvara is often depicted with one half of the body male and the other half female, further emphasizing this union.Another significant avatar of Shiva is Nataraja, the "Dancer of the Cosmos." This form represents Shiva's roleas the conductor of the universe, his dance symbolizing the cycles of creation, destruction, and rebirth. Nataraja is depicted with four arms, each performing a different action: creating, destroying, preserving, and Granting liberation.Shiva's influence extends far beyond the realm of mythology. He is worshiped by millions of Hindus worldwide, who believe that by invoking his name and meditating on his form, they can achieve spiritual enlightenment andliberation from the cycle of birth and death. Shiva templesare a common sight in India and Nepal, often serving as centers of worship and pilgrimage.The festivals dedicated to Shiva are also numerous and vibrant. Mahashivaratri, the night of Shiva's great festival, is celebrated with fervor and devotion. On this day, Hindus fast, meditate, and perform rituals to honor the deity, seeking his blessings and guidance.In conclusion, Shiva stands as a towering figure in Hindu mythology and religion. His enchanting avatars, profound attributes, and the impact he has had on millions of Hindus worldwide make him a deity worthy of reverence and worship. As we delve deeper into the rich tapestry of Hindu mythology, we discover that Shiva is not just a deity but a symbol of the primal forces of the universe, a reminder of the harmony and balance that exists within all of creation.**湿婆:迷人的印度教神话与神祇**在印度教神话的丰富织锦中,湿婆(Shiva)矗立为一位雄伟的神祇,体现了创造、毁灭和保存的原始力量。
高考英语听力理解主旨大意归纳单选题30题
高考英语听力理解主旨大意归纳单选题30题1.What are the speakers mainly talking about?A.Going on a trip.B.Buying a gift.C.Visiting a friend.D.Planning a party.答案:B。
本题主要考查对对话主旨大意的理解。
通过听力内容可知,两人在讨论买什么礼物合适,所以正确答案是B。
A 选项“去旅行”在对话中未提及;C 选项“拜访朋友”不是对话的主要内容;D 选项“计划一个派对”在对话中也没有体现。
2.What is the conversation mainly about?A.Ordering food.B.Cooking a meal.C.Cleaning the kitchen.D.Decorating the house.答案:A。
对话围绕着两人讨论点什么食物展开,所以正确答案是A。
B 选项“做饭”、C 选项“打扫厨房”和D 选项“装饰房子”在对话中均未涉及。
3.What are the speakers discussing?A.Watching a movie.B.Reading a book.C.Playing a game.D.Doing homework.答案:A。
听力内容中两人在谈论看哪部电影,故正确答案是A。
B 选项“读书”、C 选项“玩游戏”和D 选项“做作业”与对话主旨不符。
4.What is the main topic of the conversation?A.Taking a bus.B.Riding a bike.C.Driving a car.D.Walking to school.答案:B。
对话主要围绕着骑自行车出行展开,所以答案是B。
A 选项“坐公交车”、C 选项“开车”和D 选项“步行去学校”在对话中未被提及。
5.What are the speakers talking about?A.Planting flowers.B.Watering the garden.C.Painting the house.D.Fixing the fence.答案:C。
超实用备战高考英语考试易错题——阅读理解:主旨大意题(大陷阱) (解析版)
易错点17 阅读理解主旨大意题目录01 易错陷阱(3大陷阱)02 举一反三【易错点提醒一】标题类易混易错点【易错点提醒二】段落大意类易混易错点【易错点提醒三】文章大意类易混易错点03 易错题通关养成良好的答题习惯,是决定高考英语成败的决定性因素之一。
做题前,要认真阅读题目要求、题干和选项,并对答案内容作出合理预测;答题时,切忌跟着感觉走,最好按照题目序号来做,不会的或存在疑问的,要做好标记,要善于发现,找到题目的题眼所在,规范答题,书写工整;答题完毕时,要认真检查,查漏补缺,纠正错误。
易错陷阱1:标题类易混易错点。
【分析】标题类是对中心思想的加工和提炼,可以是单词、短语、也可以是句子。
她的特点是短小精悍,多为短语;涵盖性、精确性强;不能随意改变语言表达的程度和色彩。
如果是短语类选项,考生容易混淆重点,此时应当先划出选项的关键词。
此类题和文章的中心主题句有很大关系。
中心主题句一般出现在第一段,有时第一段也可能引出话题,此时应当重点关注第二段和最后一段,看看是否会出现首尾呼应。
易错陷阱2:段落大意类易混易错点。
【分析】每个段落都有一个中心思想,通常会在段落的第一句或最后一句体现,这就是段落主题句。
如果没有明显的主题句时,应当根据段落内容概括处段落大意。
有时考生还会找错文章对应位置,盲目选词文中相同的词句,而出现文不对题的现象。
易错陷阱3:文章大意类易混易错点。
【分析】确定文章主旨的方法是:先看首尾段或各段开头再看全文找主题句,若无明显主题句,就通过关键词句来概括。
如,议论文中寻找表达作者观点态度的词语,记叙文中寻找概括情节和中心的动词或反映人物特点的形容词。
文中出现不同观点时,要牢记作者的观点彩色体现全文中心的。
此时,要注意转折词,如:but, however, yet, in spite of, on the contrary等。
【易错点提醒一】标题类易混易错点【例1】(浙江省义乌五校2023-2024学年高三联考试题)The scientist’s job is to figure out how the world works, to “torture (拷问)” Nature to reveal her secrets, as the 17th century philosopher Francis Bacon described it. But who are these people in the lab coats (or sports jackets, or T-shirts and jeans) and how do they work? It turns out that there is a good deal of mystery surrounding the mystery-solvers.“One of the greatest mysteries is the question of what it is about human beings — brains, education, culture etc. that makes them capable of doing science at all,” said Colin Allen, a cognitive scientist at Indiana University.Two vital ingredients seem to be necessary to make a scientist: the curiosity to seek out mysteries and the creativity to solve them. “Scientists exhibit a heightened level of curiosity,” reads a 2007 report on scientific creativity. “They go further and deeper into basic questions showing a passion for knowledge for its own sake.” Max Planck, one of the fathers of quantum physics, once said, the scientist “must have a vivid and intuitive imagination, for new ideas are not generated by deduction (推论), but by an artistically creative imagination.”......ong as our best technology for seeing inside the brain requires subjects to lie nearly motionless while surrounded by a giant magnet, we’re only going to make limited pro gress on these questions,” Allen said.What is a suitable title for the text?A.Who Are The Mystery-solversB.Scientists Are Not Born But MadeC.Great Mystery: What Makes A ScientistD.Solving Mysteries: Inside A Scientist's Mind【答案】C【解析】文章标题。
Two-Phase Flow Patterns and Flow-Pattern Maps-Fundamentals and Applications
Lixin Cheng Laboratory of Heat and Mass Transfer(LTCM),Faculty of Engineering(STI),École Polytechnique Fédérale de Lausanne(EPFL),Station9,Lausanne CH-1015,Switzerlande-mail:lixincheng@Gherhardt Ribatski Department of Mechanical Engineering, Escola de Engenharia de São Carlos(EESC),University of São Paulo(USP), São Carlos,São Paulo13566-590,Brazile-mail:ribatski@p.brJohn R.Thome Laboratory of Heat and Mass Transfer(LTCM),Faculty of Engineering(STI),École Polytechnique Fédérale de Lausanne(EPFL),Station9,Lausanne CH-1015,Switzerlande-mail:john.thome@epfl.ch Two-Phase Flow Patterns and Flow-Pattern Maps: Fundamentals and Applications A comprehensive review of the studies of gas-liquid two-phaseflow patterns andflow-pattern maps at adiabatic and diabatic conditions is presented in this paper.Especially, besides other situations,this review addresses the studies on microscale channels,which are of great interest in recent years.First,a fundamental knowledge of two-phaseflow patterns and their application background is briefly introduced.The features of two-phaseflow patterns andflow-pattern maps at adiabatic and diabatic conditions are reviewed,including recent studies for ammonia,new refrigerants,and CO2.Then,fun-damental studies of gas-liquidflow patterns andflow-pattern maps are presented.In the experimental context,studies offlow patterns andflow-pattern maps in macro-and mi-croscale channels,across tube bundles,at diabatic and adiabatic conditions,under mi-crogravity and in complex channels are summarized.In addition,studies on highly vis-cous Newtonianfluids(non-Newtonianfluids are beyond the scope of this review)are also mentioned.In the theoretical context,modeling offlow-regime transitions,specific flow patterns,stability,and interfacial shear is reviewed.Next,flow-pattern-based heat transfer and pressure drop models and heat transfer models for specificflow patterns such as slugflow and annularflow are reviewed.Based on this review,recommendations for future research directions have been given.͓DOI:10.1115/1.2955990͔1IntroductionGas-liquid two-phaseflows at both adiabatic and diabatic con-ditions are very complex physical processes since they combinethe characteristics of deformable interface,channel shape,flowdirection,and,in some cases,the compressibility of one of thephases.In addition to inertia,viscous and pressure forces presentin a single-phaseflow and two-phaseflows are also affected bythe interfacial tension forces,the wetting characteristics of theliquid on the tube wall͑contact angle͒,and the exchange of mass,momentum,and energy between the liquid and vapor phases.De-pending on the operating conditions,such as pressure,tempera-ture,mass velocity,adiabatic or diabaticflow,channel orientation ͑the effect of gravity,which in nonvertical channels tends to pull the liquid to the bottom of the channel͒,andfluid properties ͑widely different combinations of different classes offluids such as air-water,steam-water and liquid and vapor phases of refriger-ants͒,various gas-liquid interfacial geometric configurations occurin two-phaseflow systems.These are commonly referred to asflow patterns orflow regimes.Many differentflow patterns havebeen defined by various researchers͓1–4͔,and the nature of theflow patterns varies with channel geometry and size͑macro-andmicroscale͒,fluid physical properties,flow orientation,flow pa-rameters,adiabatic or diabatic condition,etc.Furthermore,tran-sient two-phaseflows andflow oscillations are also important top-ics but are beyond the scope of the present review.speedflow͑as in criticalflow͒partial vaporization of the liquidmay occur even though there is no heat addition.Gas dissolutionor desorption in the liquid phase may also contribute in some instances to mass exchange in gas-liquidflows,especially forflu-oroinerts.One example of adiabatic two-phaseflows is the trans-portation of gas-oil mixtures in pipelines.Diabatic two-phase flows with heat transfer occur duringflow boiling,flow conden-sation,or gas-liquid two-phaseflows with heat addition or re-moval.These are confronted in steam generators,boiling water reactors,boiling and condensation of refrigerants used in air con-ditioning,refrigeration and heat pump systems,petrochemical processes,and so on.Without knowing the localflow patterns, one cannot correctly calculate the thermal/hydraulic design pa-rameters.In fact,the physical mechanisms controlling two-phase pressure drops and heat transfer coefficients are intrinsically re-lated to the localflow patterns͓1–4͔,͓10–23͔,and thusflow-pattern prediction is an important aspect of two-phase heat trans-fer and pressure drops.To predict localflow patterns,two-phaseflow-pattern maps are used.These are generally two-dimensional graphs with transition criteria to separate the areas corresponding to the variousflow regimes.Over the past decades,numerous studies offlow patterns have been conducted for various tube configurations such as in-side vertical,horizontal,and inclined channels͑macro-and micro-channels͒and other complex geometries such as inside enhanced tubes,in compact heat exchangers,across tube bundles,and under microgravity conditions,for which numerousflow-pattern maps have been proposed.Mostflow-pattern maps have been developed for adiabatic conditions,e.g.,the Hewitt and Roberts͓5͔flow-pattern map for vertical upflow and the Baker͓6͔,Taitel and Dukler͓7͔,Hashitume͓8͔,and Steiner͓9͔flow maps for horizon-talflow,just to name a few.Regarding diabaticflow-pattern maps, they should include the effect of heatflux and dryout on theflow-pattern transition boundaries and revert to an adiabatic map when the heatflux tends to zero.In principle,adiabatic two-phaseflow maps are not applicable to diabatic conditions,although this is often done.Such extrapolation of adiabaticflow maps to diabatic conditions is,in general,not reliable and also lacks the influence of heat transfer on the localflow patterns and their transitions. With respect to diabatic two-phaseflows,one of the earliest di-abaticflow-pattern maps is that of Kattan–Thome–Favrat͓10–12͔, which was developed according to their experimental observa-tions and heat transfer data forfive refrigerants͑R-134a,R123, R402a,R404a,and R502͒under evaporation conditions.Their flow map was then the basis of theirflow boiling heat transfer model for evaporation in horizontal tubes in the fully stratified,Published online July30,2008.Transmitted by Assoc.Editor J.N.Reddy.stratified-wavy,intermittent,and annularflow regimes and for an-nularflow with partial dryout at the top of the tube.Physically,itis connected to the local heat transfer characteristics and mecha-nisms by use of simplified two-phaseflow structures to accountfor any dry perimeter predicted to occur and may be applied toboth adiabatic and diabatic conditions.Since then,a number ofmodifiedflow maps have been developed for differentfluids such as R134a,R407c,R22,R410A,ammonia͑R717͒,and CO2͑R747͒under evaporation and/or condensation conditions͓13–23͔on the basis of the Kattan–Thome–Favratflow map.These will bediscussed in Secs.2–4.The vast majority of technical calculations on two-phaseflows are made without any reference whatsoever toflow patterns. Nearly all two-phase pressure drop correlations in the literature and reference books are purely empirical without reference to the flow patterns that they cover.These include such leading methods as those of Martinelli and Nelson͓24͔,Lockhart and Martinelli ͓25͔,Chisholm͓26͔,Grönnerud͓27͔,Müller-Steinhagen and Heck͓28͔,and Friedel͓29͔.Furthermore,most of the leading flow boiling heat transfer correlations do not contain any informa-tion on theflow patterns,such as those of Chen͓30͔,Shah͓31͔, Gungor and Winterton͓32͔,and Kandlikar͓33͔.Such methods are typically most accurate for annularflow,but in fact they cannot themselves identify when this regime occurs,nor do they use explicitly an annularflow structure in the prediction method. This poses the following question:Areflow patterns andflow-pattern maps helpful in practical design?Certainly,this seems to be the case.The relationships for two-phase pressure drops are likely to be significantly different for aflow consisting of a dis-persion of bubbles͑bubblyflow͒than for aflow consisting of a liquidfilm on the channel wall with a central gas core͑annular flow͒.Recent work has demonstrated that two-phase pressure drops can be more accurately predicted by giving attention to specificflow patterns in a generalflow-pattern-based model ͓22,34–36͔.Furthermore,models that have a sound theoretical basis are more likely to be reliable and generally applicable than those that are purely empirical.Calculation methods based on flow patterns andflow-pattern maps accounting for two-phase flow structure effects will ultimately supersede those ignoring the influence of theflow regimes.For heat transfer,the aforemen-tionedflow-pattern-based heat transfer models͓10–23͔provide more accurate heat transfer predictions and attempt to intrinsically relate the heat transfer mechanisms to the localflow patterns. Therefore,flow patterns andflow-pattern maps play an important role in improving the prediction models for two-phase pressure drops and heat transfer coefficients.The earliest papers onflow patterns andflow-pattern maps date back to the early1950s,and after that an avalanche of papers have been published on this subject.Numerous research reviews on flow patterns have been presented by different authors.Rouhani and Sohal͓37͔presented an overall literature review covering commonly observedflow regimes in horizontal and vertical pipes, different types offlow pattern maps,experimental techniques for direct and indirect determination offlow regimes,flow-regime transition criteria based on correlations and theoretical deriva-tions,and also the effects of wall roughness,heatflux,andflow accelerations onflow-regime transitions.Collier and Thome͓1͔, Carey͓2͔,Hewitt͓3͔,and Thome͓4͔also provided summaries or reviews on two-phaseflow patterns andflow maps.Generally, they provided an introduction to the fundamental knowledge for conventional size channels with various orientations,such as hori-zontal,vertical,and inclined channels.However,they did not pro-vide information on two-phaseflow patterns andflow maps in microscale channels.Therefore,this review also addresses the studies on microscale channels.In recent years,emphasis has been put on the characteristics of two-phaseflow and heat transfer in small and microscaleflow passages due to the rapid development of microscale devices ͓38–48͔.Due to the significant differences of transport phenom-ena in microscale channels as compared to conventional size channels or macroscale channels,one very important issue should be clarified about the distinction between microscale and macros-cale channels.However,a universal agreement is not clearly es-tablished in the literature.Instead,there are various definitions on this issue.Shah͓45͔defined a compact heat exchanger as an exchanger with a surface area density ratioϾ700m2/m3.This limit trans-lates into a hydraulic diameter ofϽ6mm.According to this defi-nition,the distinction between macro-and microscale channels is 6mm.Mehendale et al.͓46͔defined various small and mini heat ex-changers in terms of hydraulic diameter D h:•micro heat exchanger:D h=1–100m•meso heat exchanger:D h=100m–1mm•compact heat exchanger:D h=1–6mm •conventional heat exchanger:D hϾ6mmAccording to this definition,the distinction between macro-and microscale channels is somewhere between1mm and6mm. Based on engineering practice and application areas such as refrigeration industry in the small tonnage units,compact evapo-rators employed in automotive,aerospace,air separation,and cryogenic industries,cooling elements in thefield of microelec-tronics,and microelectromechanical systems͑MEMS͒,Kandlikar ͓38͔defined the following ranges of hydraulic diameters D h, which are attributed to different channels:•conventional channels:D hϾ3mm•minichannels:D h=200m–3mm •microchannels:D h=10–200mAccording to this definition,the distinction between small and conventional size channels is3mm.There are several important dimensionless numbers,which are used to represent the feature offluidflow in microscale channels. According to these dimensionless numbers,the distinction be-tween macro-and microscale channels may be classified as well. Triplett et al.͓47͔definedflow channels with hydraulic diameters D h of the order of,or smaller than,the Laplace constant L,L=ͱg͑L−G͒͑1͒as microscale channels,whereis the surface tension,g is the gravitational acceleration,andL andG are,respectively,liquid and gas/vapor densities.Kew and Cornwell͓43͔earlier proposed the confinement num-ber Co for the distinction of macro-and microscale channels,Co=1D hͱ4g͑L−G͒͑2͒which is actually based on the definition of the Laplace constant. Based on a linear stability analysis of stratifiedflow and the argument that neutral stability should consider a disturbance wavelength of the order of channel diameter,Brauner and Moalem-Maron͓48͔derived the Eotvös number Eöcriterion for the dominance of surface tension for microscale channels,Eo¨=͑2͒2͑L−G͒D h2gϾ1͑3͒The definition of a microscale channel is quite confusing be-cause there are different criteria available as described above. Cheng and Mewes͓42͔made a comparison of these different criteria for microscale channels.Figure1shows their comparable results for water and CO2,which shows the big difference among these criteria.So far,the distinction of microscale channels is still in dispute.In this review,the distinction between macro-and mi-croscale channels by the threshold diameter of3mm is adopted due to the lack of a well-established theory but is in line with those recommended by Kandlikar͓38͔.Using this threshold diam-eter enables more relevant studies to be included.So far,there is a little information on two-phaseflow patterns at microgravity conditions and in complex configurations such as tube bundles,enhanced channels,U-bends,heat exchangers,and so on.In addition,only a limited number of studies have been done on two-phaseflow patterns during condensation.Further-more,a number of theoretical studies have been performed on specificflow patterns andflow-pattern instabilities.Therefore,the present paper aims to address many of these related topics and to provide a comprehensive review of what has been learned from this research.It should be mentioned that only Newtonian and highly viscous Newtonianfluids are addressed here͑non-Newtonianfluids are beyond the scope of this review͒.In what follows,different aspects offlow patterns in a general scope arefirst presented,and published literatures are mentioned in accordance to their relevance to the specific topics.Then,sev-eral leadingflow-pattern maps will be presented together with a discussion of their limitations and future requirements.Following this,a detailed review of the studies classified according to spe-cific topics will be presented.Finally,attention will be turned to flow-pattern-based heat transfer and pressure drop models.2Schematics of Two-Phase Flow Patterns and Flow-Pattern MapsFirst,because of the variety of names and definitions offlow patterns,a discussion offlow patterns in vertical and horizontal tubes for adiabatic and diabatic conditions is presented.Then, several leadingflow-pattern maps are presented and their limita-tions are discussed.2.1Flow Patterns.In this section,flow patterns for adiabatic and diabatic conditions are described and discussed by category. Vertical Adiabatic Two-Phase Flows.Figure2͑a͒shows the most commonly observed two-phaseflow patterns in a vertical tube.Bubblyflow occurs when a relatively small quantity of gas or vapor is mixed with a moderateflow rate of liquid.Increasingbubbles separated by liquid slugs.Each of these bubbles occupiesnearly the entire channel cross section except for a thin liquidlayer on the wall,and their length is typically one to two times thechannel diameter.An increase in both gas and liquidflow rateswill lead to an unstableflow pattern,which is called churnflow.Arelatively higher gasflow rate generates a wispy-annularflow pat-tern,which is not observed or recognized as such in many studies.Very high gasflow rates may cause some of the liquidflow to beentrained as droplets carried along with the continuous gas phasein annularflows.At even higher gasflow rates,all the liquid issheared from the wall to form the mistflow regime.Horizontal Adiabatic Two-Phase Flows.Figure2͑b͒shows themost commonly observedflow patterns for cocurrentflow of gasand liquid in a horizontal tube.Two-phaseflow patterns in a hori-zontal tube are similar to those in a vertical tube,but distributionof the liquid is influenced by gravity.In general,mostflow pat-terns in horizontal tubes show a nonsymmetrical structure,whichis due to the effect of gravity on the different densities of thephases.This generates a tendency toward stratification in the ver-tical direction,with the liquid having a tendency to occupy thelower part of the channel and the gas,the upper part.Stratifiedflow is usually observed at relatively lowflow rates of gas and Fig.1Comparison of various definitions of threshold diam-eters for microscale channels:…a…water and…b…CO2by Chengand Mewes†42‡Fig.2…a…Schematic offlow patterns in vertical upward gas-liquid cocurrentflow;…b…schematic offlow patterns in horizon-tal gas-liquid cocurrentflow†1‡liquid.As the gas and liquid flow rates are increased,the smooth interface of the liquid becomes rippled and wavy.This pattern is called a stratified-wavy flow.If the liquid flow rate is further in-creased while the vapor flow is maintained low,an intermittent flow pattern will develop in which gas pockets or plugs are en-trapped in the main liquid flow and then a plug flow will develop.If flow rates of gas and liquid increase together,a so called slug flow regime will develop.The main distinction between slug and plug flow is in the more pronounced nature of intermittent liquid mass separated by a larger gas bubble.With further increase in the gas flow alone,annular flow will develop.The gas flow in the core of an annular flow may entrain a portion of the liquid phase in the form of droplets,and in some cases the liquid film may also entrain some small bubbles.At relatively large liquid flow rates,with little gas flow,one would observe the so called dispersed bubble flow in which the liquid phase is in the dispersed form of the scattered bubbles.At very high gas flow rates,the mist flow is reached,which can begin at the top perimeter where the annular film is the thinnest and then progress downstream to the bottom perimeter.Flow Patterns in Inclined Channels .There are relatively few experimental observations on flow patterns in inclined tubes,not-withstanding the technical importance of such flows.Hewitt ͓3͔summarized this topic briefly.In short,flow patterns in inclined channels seem to have the same basic structures as in vertical and horizontal flows except for the limitation or total suppression of the churn regime.Flow Patterns in Other Applications .A limited amount of in-formation is available in the literature in a variety of other specific applications,which are necessarily mentioned here,such as verti-cal downward flow and tube bundles.Collier and Thome ͓1͔pre-sented a brief introduction for rectangular channels,internal grooves,helical inserts,obstructions,expansions,contractions,bends,coils,and annuli.Rounhani and Sohal ͓37͔presented a brief summary of downward cocurrent flow.Relatively few stud-ies on flow patterns in cocurrent downward flow are reported in the literature.However,all of the flow patterns of cocurrent up-ward flow may also appear in the downward flow situations.Hewitt ͓3͔presented a brief summary of flow patterns in complex geometries such as in rod bundles and inside shell and tube heat exchangers.Thome ͓4͔provided some detailed information on flow patterns and flow maps for two-phase flows over horizontal tube bundles.However,there is a scarcity of information on these topics.Flow Patterns in Countercurrent Flows .There is very little published information regarding flow patterns in countercurrent flow situations.Rounhani and Sohal ͓37͔also presented a brief summary of countercurrent flows.In general,flow regimes in countercurrent flows have a very limited range of existence due to the fact that a continuous increase in the flow rate of either phase would lead to so called flooding,which means that the passage of the other phase would be blocked and cocurrent flow would be established.In horizontal channels,countercurrent flow may exist only as a stratified-smooth or stratified-wavy flow.In vertical channels,it exists only for downward liquid flow against rising vapor.The observed patterns are limited to annular,churn,and plug flows.So far,there is a scarcity of information on this topic.Flow Patterns at Diabatic Conditions .At diabatic conditions,two-phase heat transfer coefficients and pressure drops are closely related to the local flow patterns and vice versa.Therefore,flow patterns are very important in the heat transfer and pressure drop predictions.For flow boiling ͑evaporation ͒,consider a vertical tube heated uniformly over its length with a low heat flux and fed with sub-cooled liquid at its base at such a rate that the liquid is totally evaporated over the length of the tube.Figure 3shows the flow patterns in diagrammatic form,the various flow patterns that may be encountered over the length of a vertical tube heated by a uniform heat flux,together with the corresponding heat transfer regimes.Figure 4shows a schematic representation of a horizon-tal tubular channel heated by a uniform heat flux and fed with subcooled liquid.Flow patterns formed during evaporation in a horizontal tube may be influenced by departures from thermody-namic and hydrodynamic equilibrium.Asymmetric phase distribu-tions and stratification introduce additional complications.Impor-tant points to note from a heat transfer viewpoint are the possibility of intermittent drying and rewetting of the upper sur-faces of the tube in slug and wavy flows and the progressive dryout over long tube lengths of the upper circumference of the tube wall in annular flow.At higher inlet liquid velocities,the influence of gravity is less obvious,the phase distribution be-comes more symmetrical,and the flow patterns become closer to those as in vertical flow.For condensation,Fig.5illustrates the flow patterns typically observed during condensation inside a horizontal tube ͓49͔.At the inlet,film condensation around the circumference of the tube pro-Fig.3Schematic of flow patterns and the corresponding heat transfer mechanisms for upward flow boiling in a vertical tube †1,2‡Fig.4Schematic of flow patterns and the corresponding heat transfer mechanisms and qualitative variation of the heat trans-fer coefficients for flow boiling in a horizontal tube †1,2‡duces an annularflow,with some droplets entrained in the central high velocity vapor core.As condensation continues,the vapor velocity falls and reduces the influence of vapor shear on the condensate,and the influence of gravity forces increases.At high flow rates,slug and bubbleflows are eventually reached,while at lowflow rates large magnitude waves and then stratifiedflow are formed.2.2Flow-Pattern Maps.Flow-pattern maps may be classi-fied primarily into two types:empiricalflow-pattern maps,whichare generallyfitted to the observedflow-pattern database and the-oretical or semitheoreticalflow-pattern maps whose transitions are predicted from physical models of theflow phenomena.Theoret-ical or semitheoreticalflow-pattern maps are developed according to theflow structure and are sometimes related to heat transfer mechanisms and diabatic ually,only twoflow parameters are used to define a coordinate system on which the boundaries between the differentflow patterns are charted,such as the superficial gas and liquid velocities.Transition boundaries are then proposed to distinguish the location of the variousflow re-gimes as in a classical map.Mostflow maps are only valid for a specific set of conditions and/orfluids,although efforts are made to propose generalizedflow maps.In this section,only several leadingflow-pattern maps are presented.Generally,flow-pattern maps for other applications such as microscale channels,en-hanced heat transfer tubes,compact heat exchangers,and tube bundles and at microgravity conditions have been proposed by modification of these leadingflow maps.Empirical Flow-Pattern Maps.One of the best known empiri-calflow-pattern maps for horizontalflow is that of Baker͓6͔shown in Fig.6.It was based on observations of cocurrentflow of gaseous and condensate petroleum products in horizontal pipes and was constructed with two parameter groupsG G/and G L͑where G G and G L are gas and liquid mass velocities,respec-tively͒,taking into account their physical properties by introduc-ing the following parameters:=ͩGALWͪ1/2͑4͒=ͩWͪͫͩLWͪͩWLͪ2ͬ1/3͑5͒whereA,W,W,andW are the density of air,the density ofwater,the surface tension of water,and the dynamic viscosity ofwater,respectively,at1atm pressure and room temperature whileL,G,,andL are the liquid density,gas density,surface ten-sion,and liquid viscosity,respectively.The correction factorsandhad previously been used for correlating data onfloodingpoints in distillation columns.Although Baker’sflow map coordi-nates include these apparently relevant variables for scaling a va-riety of different conditions,later investigations have shown thatthis map does not adequately predict horizontalflow regimes innumerous situations,as pointed out by Rounhani and Sohal͓37͔.One of the leading empiricalflow-pattern maps for vertical up-flows is that of Hewitt and Roberts͓5͔shown in Fig.7.On thismap,the coordinates are the superficial momentumfluxes of therespective phases.Both air-water and steam-water data could berepresented in terms of this plot,which thus covers a reasonablywide range offluid physical properties.All the transitions areassumed to depend on the phase momentumfluxes.Wispy-annularflow is a subcategory of annularflow,which occurs athigh massflux when the entrained drops are said to appear aswisps or elongated droplets.Generally,the accuracy in determining transition lines on aflowmap is in part dependent on the number of experiments carried outand on the adopted coordinate systems as well.There are manyother coordinate systems forflow-pattern maps used by differentinvestigators,for example,superficial gas and liquid velocitiesu GS and u LS͑m/s͒,and mass velocities G͑kg/m2s͒versus vaporqualities x.According to Troniewski and Ulbrich͓50͔,the coor-dinates used inflow maps may be divided into three groups:͑1͒Phase velocities orfluxes:gas and liquid superficial veloci-ties u GS and u LS͑m/s͒or gas and liquid superficial massfluxes G GS and G LS͑kg/m2s͒and gas and liquid massflowrates M G and M L͑kg/s͒.Use of these parameters,while Fig.5Schematic offlow patterns for horizontal gas-liquidcocurrentflow in condensation†49‡Fig.6The Baker†6‡flow-pattern map for horizontal gas-liquidcocurrentflowFig.7The Hewitt and Roberts†5‡flow-pattern map for verticalupward gas-liquid cocurrentflowundoubtedly being the most convenient,does not ensure creation of a universal flow-pattern map for different two-phase mixtures.͑2͒Quantities referring to the two-phase flow homogeneousmodel are the transformations of the parameters from group ͑1͒such as total velocity u T ,total mass flux G T ,Froude number based on total velocity Fr T ,void fraction ,and quality x ,and they are only useful for the description of some flow-pattern maps.͑3͒Parameters including the physical properties of phases suchas liquid and gas Reynolds numbers Re L and Re G ,Baker correction factors and ,gas and liquid kinetic energies E G and E L ,and others;this formulation gives the best pos-sibility for attaining a universal flow-pattern map.A summary of the coordinates used in flow maps can be found in Refs.͓50,51͔.Theoretical or Semitheoretical Flow-Pattern Maps .There have been various attempts at a theoretical or semitheoretical descrip-tion of flow-pattern transitions.For such a description to be suc-cessful,it should be suitable for extrapolation to a wide range of conditions.Perhaps the most comprehensive treatment of flow-pattern transitions in horizontal flow on a semitheoretical basis is that of Taitel and Dukler ͓7͔.It has been proven successful in predicting a fairly wide range of system conditions.Figure 8shows the Taitel and Dukler flow-pattern map.The parameter groups,which are based on semitheoretical derivations for differ-ent flow-pattern transitions in horizontal or slightly inclined chan-nels ͑is the angle of inclination ͒,are as follows:X =ͫ͑dp /dz ͒L ͑dp /dz ͒Gͬ1/2͑6͒Fr =G G͓G ͑L −G ͒Dg cos ͔1/2͑7͒T =͉ͫ͑dp /dz ͒L ͉g ͑L −G ͒cos ͬ1/2͑8͒K =FrͫG L D Lͬ1/2͑9͒where X is the Martinelli parameter,͑dp /dz ͒L is the frictional pressure gradient as if the liquid in the two-phase flow were flow-ing alone in the tube,͑dp /dz ͒G is the frictional pressure gradient as if the gas in the two-phase flow were flowing alone in the tube,Fr is theFroude number,D is the tube diameter,g is the accelera-tion due to gravity,L is the liquid density,G is the gas density,and L is the liquid viscosity.They suggested the K versus X coordinate with a theoretically derived boundary curve,C ,for transition from stratified-smooth to stratified-wavy flow.The Fr versus X relationship was proposed for the transitions between stratified-wavy,annular-dispersed ͑droplets ͒,dispersed bubble,and intermittent ͑plug or slug ͒flows.The theoretically determined transition curves A and B ͑at X =1.6͒between the said regimes were also given in those coordinates.Finally,T versus X was proposed for defining the transition between dispersed bubble and intermittent ͑plug or slug ͒flow regimes with the transition line D .The transition curves shown in Fig.8are for the case of zero inclination angle ͑horizontal ͒.All the transition criteria used by Taitel and Dukler have some theoretical bases,although they are sometimes rather tenuous.As pointed out by Hewitt ͓3͔,it should be remembered that there is an essential arbitrariness in the inter-pretation of flow-pattern data,and thus it is unlikely that perfect prediction methods will ever emerge.Diabatic Flow-Pattern Maps .In the case of diabatic two-phase flows such as flow boiling ͑evaporation ͒and condensation,very few maps have been proposed.Important factors influencing these flows and their transitions are nucleate boiling,evaporation or condensation of liquid films on what could otherwise be dry parts of the perimeter,and acceleration or deacceleration of the flows.For example,nucleate boiling in an annular film tends to increase the film’s thickness and change the void profile near the wall,or vigorous nucleate boiling in an otherwise stratified flow can com-pletely wet the upper perimeter,thus increasing liquid entrainment in the vapor core.It is desirable that diabatic flow-pattern maps include the influences of heat flux,dryout,etc.,on the flow-pattern transition boundaries.One such map is that of Kattan–Thome–Favrat ͓10–12͔for evaporation inside horizontal channels.This was developed based on five refrigerants ͑pure fluids R134a and R123,the azeotropic refrigerant mixture R502,and two near azeo-tropic mixtures R402A and R404A ͒under flow boiling conditions by modification of the Steiner map ͓9͔,which in turn is a modified Taitel–Dukler flow map ͓7͔.Figure 9shows the Kattan–Thome–Favrat flow-pattern map ͑solid lines ͓͒10–12͔compared to the Steiner map ͓9͔͑dashed lines ͒evaluated for R410A at T sat =5°C in a 13.84mm internal diameter tube at different heat fluxes ͓36͔.In the Kattan–Thome–Favrat flow map,stratified,stratified-wavy,intermittent,annular,bubbly,and mist flows are encountered.The map includes a diabatic method for predicting the anticipation of the onset of dryout at the top of the tube in evaporating annularFig.8The Taitel and Dukler †7‡flow-pattern map for horizontal gas-liquid cocurrent flow:coordinates of curves A and B are Fr versus X ,coordinates of curve C are K versus X ,and coordinates of curve D are T versus X。
Cai_2001_Comparison-of--solid-pmicroextraction-for-the-determination-of-volatile-flavor-components
930 (2001) 1–7Journal of Chromatography A,/locate/chromaComparison of simultaneous distillation extraction and solid-phase microextraction for the determination of volatile flavor componentsa,b b a,b ,*Jibao Cai ,Baizhan Liu ,Qingde SuaDepartment of Chemistry ,University of Science and Technology of China ,Hefei ,230026,PR ChinabResearch Center of Tobacco Science ,University of Science and Technology of China ,Hefei ,230052,PR ChinaReceived 6April 2001;received in revised form 28June 2001;accepted 28June 2001AbstractTraditional simultaneous distillation extraction (SDE)and solid-phase microextraction (SPME)techniques were compared for their effectiveness in the extraction of volatile flavor compounds from various mustard paste samples.Each method was used to evaluate the responses of some analytes from real samples and calibration standards in order to provide sensitivity comparisons between the two techniques.Experimental results showed traditional SDE lacked the sensitivity needed to evaluate certain flavor volatiles,such as 1,2-propanediol.Dramatic improvements in the extraction ability of the SPME fibers over the traditional SDE method were noted.Different SPME fibers were investigated to determine the selectivity of the various fibers to the different flavor compounds present in the mustard paste samples.Parameters that might affect the SPME,such as the duration of absorption and desorption,temperature of extraction,and the polarity and structure of the fiber were investigated.Of the various fibers investigated,the PDMS–DVB fiber proved to be the most desirable for these analytes.©2001Elsevier Science B.V .All rights reserved.Keywords :Solid-phase microextraction;Simultaneous distillation extraction;Mustard paste;Volatile organic compounds;Flavor compounds1.Introductionformed.Several extraction and concentration meth-ods have been used;among them are liquid–liquid The determination of volatile components in a extraction [2],liquid–liquid extraction with ultra-mixture is a process widely used in many disciplines,sound [3],simultaneous stream distillation extraction such as environmental,food,forensic,fragrance,oil,[4],solid-phase extraction [5],and other techniques pharmaceutical and polymer analysis.The method of [6,7].The main reason for extraction is to obtain a choice for many of these analyses is simultaneous more concentrated samples,to eliminate interfering distillation extraction (SDE)[1]followed by GC or substances and to improve detection limits for spe-GC–MS analysis.Extraction and concentration are cific compounds.usually necessary before analysis by GC is per-Solid-phase microextraction (SPME)[13]is a relatively new technique that is able to address the need for concentrating the analytes in the headspace *Corresponding author.Tel.:186-551-360-6642;fax:186-[8].SPME uses a small (1-cm long)piece of fused 551-360-3388.E -mail address :qdsu@ (Q.Su).silica,on which a liquid phase,similar to a GC0021-9673/01/$–see front matter ©2001Elsevier Science B.V .All rights reserved.PII:S0021-9673(01)01187-6930 (2001) 1–72J.Cai et al./J.Chromatogr.Astationary phase,has been coated to absorb the waterbath maintained at30,50and708C,respective-desired analytes and concentrate them on thefiber.ly,to optimize temperature of extraction.The vial The selectivity of the extraction of target analytes in was submerged only as far as necessary to submerge the gaseous phase can be significantly altered the solid phase of the sample,to help keep the SPME through the use of different liquid phase on thefiberfiber cool,which is a desired condition for SPME.[9].This is because as the temperature of thefiber Mustard paste,which is usually served as a spice increases,the partition coefficient decreases[10]. for foodflavoring,has become increasingly popular.The SPME holder was secured and thefiber extend-The main ingredients of mustard paste are Brassica ed into the headspace,and thefiber was equilibrated nigra and Brassica alba seeds.The most predomi-for20,40and60min,respectively,to optimize the nant constituent of Brassica nigra is allyl isothio-time of extraction.Thefiber was then retracted, cyanate,which accounts for more than90%of the removed from the vial,and placed immediately into total volatile compounds.The most predominant the injector of the GC at2508C.Injection was constituent of Brassica alba is sinalbin disulfide.accomplished by extending thefiber in the heated Consequently,the analysis of the volatileflavor inlet for3,5and7min,to optimize the time of compounds in mustard paste can identify the mustard desorption,while the injector operated in the splitless varieties.mode for2min.The additional time of exposuretime in the injector port allowed thefiber to becleaned of any compounds that may not be desorbed 2.Materials and methods in the initial minute.Preliminary studies indicatedthat the above procedure allowed for reproducible, 2.1.Materials quantitative transfer of target analytes into the injec-tor port of the GC–MS.Mustard pastes(made in Japan)were purchasedfrom a local supermarket.The mustard paste is2.3.Sampling conditions of SDEcomposed of mustard,sorbitol,corn oil,salt,water,artificialflavor,xanthan gum,turmeric,and artificialSimultaneous distillation–solvent extraction was color(FD&C Yellow no.5,FD&C Blue no.1).Thecarried out in a microversion apparatus,as described components of the standard solutions were all pur-elsewhere[11].Dichloromethane(chromatography-chased from Sigma(St.Louis,MO,USA).Standardgrade reagent,Merck)and n-tetradecane were used solutions were used to optimize GC–MS and SPMEas solvent and internal standard,respectively.For conditions.All solutions were stored at48C.each extraction,10g of mustard paste and250mldistilled water were placed in a500-mlflask,30ml 2.2.Sampling conditions of SPMEdichloromethane was placed in a50-mlflask,streamdistillation was stopped after2h,while the solvent For the SPME determinations,a SPME holder andextraction was continued for a further15min.The threefibers(100-m m PDMS,65-m m PDMS–DVBextract was concentrated to 1.0ml at558C by and65-m m CW–DVB)were used(Supelco,Belle-Kuderna-Danish apparatus(NE-1,Japan).The in-fonte,PA,USA).Thefibers were conditioned underjection volume was2.0m l with a split ratio of20:1. helium at2908C for4–5h prior to use.BetweenA series of three consecutive extractions was per-uses,fibers were kept sealed from ambient air byformed on different aliquots of mustard pastes in inserting the tip of the SPME needle into a smallorder to evaluate the repeatability of SDE method. piece of septum to prevent accidental contamination.The sampling procedure involved placing2–3g ofsample into a20-ml vial and sealing with a screwtop 2.4.Condition of GC–MSseptum-containing cap.The SPME needle was theninserted through the septum and suspended in the Autosystem TurboMass GC–MS(Perkin-Elmer, headspace of the vial.The vial was placed in a USA)was used.A30m30.25mm Supelco-wax930 (2001) 1–73J.Cai et al./J.Chromatogr.Aquartz capillary column(Supelco,Bellefonte,PA, 3.Results and discussionUSA)with0.25-m mfilm thickness was used toresolve the volatiles with the following temperature parison of SDE and headspace SPME programming:initial oven temperature was608C,techniquekept for2min;then was raised to2408C at48C/min,and kept at2408C for15min.Helium was used as As shown in Table1,traditional SDE technique carrier gas with column head pressure at10kPa.could extract all the volatileflavor compounds of Programming split/splitless(PSS)injector tempera-mustard paste,except for1,pared ture was at2508C.In SPME analysis,the I.D.of the with SPME,SDE could also extract high-molecular-injection liner was1.5mm;the desorption time was mass and low volatility compounds such as oleic 5min in splitless mode;and the time of the splitless acid,9-hexadecenoic acid and palmitic acid in was2min.In the analysis of SDE extract,the I.D.of volatileflavor compounds of mustard paste.So the the injection liner was4.0mm;the split ratio was traditional SDE technique seemed more comprehen-20:1,and the amount of injection was2m l.The sive but less sensitive to trace rge temperature of the GC–MS transfer line was2508C.amounts of furfural and furfural alcohol were found The MS was operated at1708C in the electron in the SDE extracts,perhaps arising from pyrolysis impact mode(70eV),scanning from m/z33to350or hydrolysis during the SDE process.In fact,if in0.3s with an0.2-s interval time of the scan;the these compounds were mustard paste volatileflavor voltage of the photoelectric multiplier tube(PMT)components,they should easily be extracted by was230V.The mass spectral identifications of the SPME.However,they were found only in SDE volatiles were carried out by comparing to the extracts.Representative TIC chromatograms of vola-NIST98(National Institute of Standards and Tech-tileflavor compounds from mustard pastes are shown nology,Gaithersburg,MD,USA)mass spectral in Fig. pared with SDE,SPME showed library as well as to the Wiley6.0(Wiley,New York,enormous advantages:simplicity,rapid solvent-free NY,USA)mass spectral library.extraction,low cost,little interference,no apparentTable1GC–MS identification of mustard paste volatiles and peak area percentagesPeak t Compound name Peak area(%)Rno.(min)100-m m65-m m65-m m SDEPDMS CW–DVB PDMS–DVB1 5.081-Propene,3,3-thiobis0.0220.0130.0170.009 28.10Thiocyanic acid,methyl ester0.0160.0270.0210.011 313.56Allyl isothiocyanate98.5863.6293.2498.84 413.83Furfural ND ND ND0.081 514.20Diallyl disulfide0.0220.0080.0130.012 617.50Methyl allyl trisulfide0.0140.0120.0130.006 717.881,2-Propanediol 1.12835.77 3.125ND 819.87Furfural alcohol ND ND ND0.125 923.37Diallyl trisulfide0.0150.0140.0140.005 1026.93Butylated hydroxyl toluene0.0700.0130.0410.011 1128.475-Methyl-tetrahydrothiophen-2-one0.0510.3300.1950.014 1231.25Ethanol,1-methoxy-,benzoate0.0090.0410.0270.011 1334.722-Phenylethyl isothiocyanate0.0140.0210.0160.012 1449.67Palmitic acid ND ND ND0.508 1550.619-Hexadecenoic acid ND ND ND0.799 1650.90Oleic acid ND ND ND 1.522 ND,not determined.930 (2001) 1–7 4930 (2001) 1–75J.Cai et al./J.Chromatogr.ATable2Repeatability of SPME(n55)and SDE(n53)Peak Compound name RSD(%)no.100m m65m m65m m SDECW–DVB PDMS PDMS–DVB11-Propene,3,3-thiobis 6.23 3.71 6.74 2.032Thiocyanic acid,methyl ester 5.768.73 4.73 3.743Allyl isothiocyanate8.64 3.52 2..22 1.614Furfural––– 3.495Diallyl disulfide 5.66 5.758.08 2.536Methyl allyl trisulfide–8.779.24 3.6871,2-Propanediol 4.60 4.207.99–8Furfural alcohol––– 1.369Allyl trisulfide7.919.72 3.29 2.2710Butylated hydroxyl toluene 5.817.55 5.82 3.52115-Methyl-tetrahydrothiophen-2-one9.557.44 4.29 2.4512Ethanol,1-methoxy-,benzoate7.91 4.75 6.48 1.45132-Phenylethyl isothiocyanate 6.689.148.89 2.3714Palmitic acid––– 3.09159-Hexadecenoic acid––– 1.3416Oleic acid––– 2.15–,not determined.21less than0.3m g l for most of analytes.The aqueous layer at thefiber-gas interface increased relative standard deviation(RSD)is better than9%.with increasing temperature,so that more analytes This method was applied to a food sample(mustard were absorbed at higher temperature if equilibrium paste)using an external calibration.had not been reached.The decreasing absorptionwith increasing temperature at708C was presumably 3.2.1.Extraction temperature due to the distribution constant decreasing with The extraction temperature profile obtained using increasing temperature.The absorption process was a PDMS–DVBfiber is shown in Fig.2.Optimum exothermic,thus lowing the temperature increased extraction efficiency was achieved at508C.The the distribution constant at equilibrium.In practical lower absorption of most analytes at308C was due to applications when the extraction was stopped before the decreased rate of diffusion of the analytes.The reaching the equilibrium,not only thermodynamic rate of diffusion of the analytes through the static but also kinetic aspects became important.An ex-traction temperature of508C was selected for thisstudy using the threefibers,because this temperaturewas relatively easily maintained,and the improve-ment in sensitivity at higher temperature was notnecessary.3.2.2.Extraction timeThe extraction time profile obtained using PDMS–DVBfiber is shown in Fig.3.For the PDMS–DVBfiber,the equilibrium condition for the absorption ofthe most analytes was almost reached after40min.Factors that influenced the equilibration period wereinvestigated by Pawliszyn and co-workers[13–15].The equilibration rate was limited by(1)the masstransfer rate of the analytes through a thin static Fig.2.Extraction temperature profile for65-m m PDMS–DVBfiber.Extraction time,40min;desorption time,5min.aqueous layer at thefiber-gas interface,(2)the930 (2001) 1–76J.Cai et al./J.Chromatogr.A3.2.parison of differentfibersThree differentfibers were evaluated to determinewhichfiber most effectively extractedflavor volatilesfrom mustard paste samples.Thefibers that wereused to extract analytes from the headspace ofaliquots of the same sample for comparison of therelative extraction effectiveness were100-m mPDMS,65-m m CW–DVB and65-m m PDMS–DVB,respectively.The results of the experiments on thesethreefibers are summarized in Fig.5.These resultsshow that,of thefibers evaluated,the PDMS–DVBfiber proved to be the most effective in extractingflavor volatiles overall,followed by the PDMSfiber,then the CW–DVBfiber.Therefore,the65-m m Fig. 3.Extraction time profile for65-m m PDMS–DVBfiber.Extraction temperature,508C;desorption time,5min.PDMS–DVBfiber was used for all subsequentcomparison experiments.The PDMS–DVBfiber was distribution constant of thefiber coating and(3)the chosen as a representative to investigate the duration thickness and kinds of thefiber coating.Extraction of absorption and desorption,temperature of absorp-periods of40min were chosen for the threefibers tion,detection of limits,and the precision of SPME since it was approximately equivalent to the time in this investigation.required to run GC in this experiment.The PDMS–DVBfiber performed the most effec-tive extractions,for this analysis,because thefiber 3.2.3.Desorption time coating was composed of a mixed coating containing The desorption time profile obtained using the PDMS,a liquid phase that favored the absorption of PDMS–DVBfiber is shown in Fig.4.A desorption nonpolar analytes,as well as DVB,a porous solid period of5min was enough to desorb the analytes that favored the adsorption of the more polar ana-from the PDMS–DVBfiber(temperature of the GC lytes.There was little difference between the PDMS injection port,2508C).So a desorption period of5fiber and the PDMS–DVBfiber in extracting the min was used for the threefibers(temperature of the nonpolar analytes(butylated hydroxytoluene),but GC injection port,2508C).No carryover of any the more polar disulfide and trisulfide were extracted, volatileflavor component was observed.on average,three times better by the PDMS–DVBFig. 4.Desorption time profile for65-m m PDMS–DVBfiparison of100-m m PDMS,65-m m PDMS–DVB and Extraction temperature,508C;extraction time,40min.65-m m CW–DVBfibers.930 (2001) 1–77J.Cai et al./J.Chromatogr.Afiber as measured by peak area.The CW–DVBfiber,sitivity allowed fast,accurate determinations of which was more selective towards polar analytes,didflavor compounds and easy performance of analyses. show enhanced extraction effectiveness of the polar Consequently,SPME was suitable for simple,rapid, analytes,but was less effective with the nonpolar routine screening,while SDE was used for proper analytes.quantitative analysis.Profiling of different mustard paste samples were3.2.5.Repeatability performed.Differentfibers were investigated withA series offive consecutive extractions were the bestfiber found to be65-m m PDMS–DVB.The performed on different aliquots of mustard pastes in optimal parameters for SPME sampling were also order to evaluate the repeatability of the headspace investigated.SPME(HS-SPME)method.The precision of theHS-SPME method was good and the RSD valueswere between2.22and9.72%for all the11volatile Referencesflavor compounds in mustard paste(Table2).[1]A.Orav,T.Kailas,M.Liiv,Chromatography43(1996)215.3.3.Mustard pastes determined by SPME–GC–MS[2]D.Martinez,F.Borrul,M.Calull,J.Chromatogr.A827(1998)105.[3]T.Hankemeier,S.J.Kok,R.J.J.Vreuls,U.A.Th.Brinkman,J. In total,11volatile compounds in mustard pasteChromatogr.A841(1999)75.were identified,which accounts for99%of TIC peak[4]A.J.Nunez,J.M.H.Bemelman,J.Chromatogr.294(1984) area as shown in Table1.The four methods made up361.[5]T.Hankemeier,E.Hooijschuur,R.J.J.Vreuls,U.A.Th.Brink-one another and validated mutually.Since allylman,T.Visser,J.High Resolut.Chromatogr.21(1998)341. isothiocyanate was the main volatile constituents of[6]M.D.Burford,S.B.Hawthorne,ler,J.Chromatogr. the mustard pastes(Table1),the main ingredient ofA685(1994)79.the mustard pastes was Brassica nigra seeds.[7]B.Gawdzik,T.Matynia,Chromatography38(1994)643.[8]Z.Zhang,M.Yang,J.Pawliszyn,Anal.Chem.66(1994)844A.[9]X.Yang,T.Pepard,LC–GC13(1995)882.4.Conclusions[10]Z.Zhang,J.Pawliszyn,Anal.Chem.67(1995)34.[11]M.Godefroot,P.Sandra,M.Verzele,J.Chromatogr.203 Traditional SDE analysis of volatile compounds is(1981)325.a widely used technique.However,for many analy-[12]C.L.Arthur,L.Killam,K.Buchhliz,J.Pawliszyn,J.Berg,Anal.Chem.64(1992)1960.ses,the SDE method lacked the sensitivity and[13]C.L.Arthur,J.Pawliszyn,Anal.Chem.62(1990)2145. convenience needed to perform adequately.SPME[14]D.Louch,S.Motlagh,J.Pawliszyn,Anal.Chem.64(1992) had the ability to perform these analyses where SDE1187.fell parison of SDE and SPME showed[15]D.W.Potter,J.Pawliszyn,J.Chromatogr.625(1992)247. that SPME determinations offlavor compounds[16]D.C.Garcia,S.Magnaghi,M.Reinchenbacher,K.Danzer,J.High Resolut.Chromatogr.19(1996)257.were,on average,more sensitive under the con-ditions employed in this study.The increased sen-。
designing website attributes to induce experiential encounters
Designing website attributes to induceexperiential encountersMing-Hui Huang *Institute of Electronic Commerce,National Chung Hsing University,250Kuo Kuang Road,Taichung 402,TaiwanAbstractCan websites be designed to be both utilitarian and hedonic?This article approaches this question by identifying Web attributes,their direct impacts on experiential flow,and their direct and indirect impacts on the utilitarian and hedonic aspects of Web performance.The results presented here support the proposal that,as an information-laden medium,a success-ful website must be able to use its attributes to satisfy both the information and entertainment needs of users.#2003Elsevier Science Ltd.All rights reserved.Keywords:Experiential flow;Utilitarian;Hedonic;Web performance;Web attributes1.IntroductionThe Web is evolving into an environment that caters for a range of activities,including entertainment,exploration,communication,and ers not only search the Web for specific information,they also surf it for entertainment,stimu-lation,and to socialize.The search for specific information is the result of goal-directed need,whereas surfing the Web for other purposes is the result of the experiential need of users (Eliashberg &Sawhney,1994;Hoffman &Novak,1996;Novak,Hoffman,&Yung,2000;Wolfinbarger &Gilly,2001).Users use and look for different Web attributes depending on their needs (Singh &Dalal,1999).The ability to create and manage websites that fulfill one or more of the needs of users will become a distinguishing feature of successful e-business.The utilitarian function of websites in meeting goal-directed need is well docu-mented.This article considers the other side of the coin:can websites be designed to *Tel./fax:+886-4-22854910.E-mail address:mhhuang@.tw (M.-H.Huang).426M.-H.Huang/Computers in Human Behavior19(2003)425–442generate an experiential experience and to enhance hedonic performance?To address this issue,we examined numerous Web attributes and divided these attri-butes into relevant components such as complexity,novelty,and interactivity.We also investigated the individual and collective impacts of these properties on experi-entialflow,an intrinsically enjoyable experience that some believe is an essential feature of commercially compelling websites(Novak et al.,2000).The Web attri-butes andflow experience were further linked to both the utilitarian and hedonic aspects of Web performance.Because the Web mixes goal-directed and experiential behavior,our results can be used to develop and evaluate websites in terms of the extent to which they satisfy these two needs.2.The concepts2.1.Web attributes:complexity,novelty,and interactivityAttributes are features or aspects of a ers see each website as a bundle of attributes with varying capacities to satisfy their needs.Attributes can be tech-nology or user-oriented.Technology-oriented attributes are the structural properties of a site such as hyperlink multimedia modalities,whereas user-oriented attributes are the qualitative experiences of users in relation to the structural properties of a site,for example navigability and demonstrability.We investigate the user-oriented attributes of websites,because they have the greatest influence on user decision-making.Although we do not explicitly consider the structural properties of websites,the user-oriented approach includes within it user preferences in regard to structural properties such as multidimensionality,par-ticipation,richness,and demonstrability.Drawing upon information theory,the concept of information rate is used to gauge the complex spatial and temporal arrangement of stimuli in a website setting(Mehrabian&Russell,1974).The spatial and temporal arrangement of attributes varies from site to site.These attributes can be described as multidimensional,rich,selective,multiple,large-scale,broad,active, responsive,interactive,participatory,dynamic,demonstrable,imaginary,surprising, innovative,new,flexible,or multimedia.Complexity and novelty have been identified as the two major categories for attributes in physical settings(Berlyne,Craw,&Salapatek,1963;Donovan&Rossiter, 1982;Hwang&Lin,1999;Iselin,1988).Nevertheless,the attributes of a website cover a larger variety of categories than those of traditional media.Interactivity has emerged as a critical attribute in computer-mediated communications,because it is seen as the key advantage of the medium(Ha&James,1998).Thus,we identified complexity,novelty, and interactivity as the three broad categories for website attributes:plexityComplexity refers to the amount of information that a site is perceived to offer, including aspects such as the number of alternatives,the number of attributes,and the variation in the information provided by the attributes(Berlyne et al.,1963;M.-H.Huang/Computers in Human Behavior19(2003)425–442427Campbell,1988;Donovan&Rossiter,1982;Hwang&Lin,1999;Iselin,1988).Both the topical range and structural properties of a website contribute to how it is per-ceived.For example,sites that convey information through a broad range of text and graphics in a multidimensional format are likely to be perceived as complex.2.1.2.NoveltyNovelty can take the form of novel experiences or information,or both in com-bination(Bianchi,1998).It refers to the aspects of website attributes that usersfind unexpected,surprising,new,and unfamiliar.Novelty can be created by freshness of content and innovation in information technology.Novelty is often conceptualized as the opposite of familiarity,and is associated with a lack of experience with the website or its offerings(Berlyne et al.,1963).Many Web technologies provide attri-butes that are novel to users,such as multimedia modalities and the simulation of humanlike characteristics.2.1.3.InteractivityInteractivity describes the extent of information exchange between a website and its users.Interactivity is the attribute that most distinguishes websites from other media.Although the meaning of interactivity may seem self-evident,the term has been applied to widely divergent phenomena and has been defined in a number of ways.Synthesizing the results fromthe literature(see Fig.1),the following core conceptual domains were identified for interactivity(Burgoon,Bonito,Bengtsson,& Ramirez,1999/2000;Ha&James,1998).Also given are examples of technology attributes corresponding to these conceptual domains.Responsiveness:the degree to which a site is perceived to respond to users’needs.An e-mail management system is one such technology attribute showing responsiveness.Individualization:the extent to which a site is perceived to provide informa-tion personalized to the unique needs of each user.For example,the use of information-filtering agents to provide users with personalized information achieves this purpose.We see this in popular sites such as My Yahoo!and My Dell.Navigability:the extent to which a site is perceived to have unrestrained connectedness,including links to other parts of the site and to other sites to allow easy information retrieval.For site visitors,hypertext in websites can create a feeling of being connected to the world by allowing visitors to jump, with minimal effort,from one point in cyberspace to another.Reciprocity:the extent to which a site is perceived to provide two-way infor-mation exchange between the site and users.A reciprocal loop can be formed by a company providing information to users at a website along with a way for users to respond to this information.Posting answers to FAQs is one example of reciprocity.Synchronicity:the extent to which users feel that they have real-time bi-directional feedback.With a higher degree of synchronicity,the informationprovided is more immediately available to users.Online customer services and online chat make synchronicity possible.Participation :the extent to which a site and its users actively interact.Forexample,a site may allow visitors to modify or add information to the site.When users are given the opportunity to create content,information exchange is facilitated.Demonstrability :the extent to which a site is perceived to simulate or incor-porate humanlike characteristics.Demonstrability reflects the media richness of a site.It is influenced by whether a site format utilizes one or more mod-alities such as text,audio,visual,or touch,and the extent to which a site supports a variety of symbols to present diverse information.2.2.Experiential flowAccording to the theory of flow,experiential flow is the optimal and enjoyable experience in which we feel ‘‘in control of our actions,masters of our own fate ...weer and technology attributes of interactivity illustrated.428M.-H.Huang /Computers in Human Behavior 19(2003)425–442M.-H.Huang/Computers in Human Behavior19(2003)425–442429 feel a sense of exhilaration,a deep sense of enjoyment’’(Csikszentmihalyi,1990,p.3).Applying the concept of optimalflow to the Web environment,flow is a sub-jective human and computer-mediated interaction experience,representing the user’s perception of the interaction with a site as playful and exploratory(Hoffman& Novak,1996;Novak et al.,2000).The consideration offlow is important for understanding user behavior on the Web.The virtual hypermedia environment of the Web incorporates various types of interactivity(human–human,human–com-puter,and computer–computer)and possesses unique characteristics which distin-guish it fromthe physical world(Hoffman&Novak,1996).This interactivity and its distinction from everyday activities may provide Web users with an environment in which to experienceflow(Chen,Wigand,&Nilan,1999).The overallflow experience can be divided into four components:(1)the percep-tion of a sense of control over the computer interaction,(2)the degree to which the user’s attention is focused on the interaction,(3)the curiosity aroused during the interaction,and(4)the extent to which the userfinds the interaction intrinsically interesting(Trevino&Webster,1992;Webster,Trevino,&Ryan,1993).The four components reveal thatflow is both cognitive and affective.It represents the optimal experience that stems from people’s perceptions of challenges and skills in given situations(Csikszentmihalyi,1975);the experience offlow requires a Web user to meet challenges imposed by Web environments(Hoffman&Novak,1996).In theflow state,the computer user concentrates on the activity to such an extent that they have little attention left to consider anything else(Ghani&Deshpande,1994;Hoffman&Novak, 1996;Novak et al.,2000).Flow is fun(Privette,1983).The state of mind that results from achievingflow is extremely gratifying(Novak et al.,2000)and self-motivating(Trevino &Webster,1992).When in aflow state,an individualfinds the activity intrinsically interesting(Csikszentmihalyi,1975).Flow is most often experienced in activities that reward the participants(Csikszentmihalyi,1975),for example the immediate feedback provided by computers imparts enjoyment to users(Webster et al.,1993).2.3.Utilitarian and hedonic aspects of Web performanceUsers visit websites not only for information,but also for entertainment.Through the inclusion of attributes that cause users to perceive a site as abundant,interactive, or novel,a website can be created that provides users with opportunities to experi-enceflow and that meets both their utilitarian and hedonic needs.We identified two aspects of Web performance:utilitarian and hedonic.The utili-tarian aspect of Web performance is the evaluation of a website based on the assessment by users regarding the instrumental benefits they derive from its non-sensory attributes.It is related to the performance perception of usefulness,value, and wisdom(Batra&Ahtola,1990).Utilitarian performance results from user vis-iting a site out of necessity rather than for recreation;therefore,this aspect of per-formance is judged according to whether the particular purpose is accomplished (Davis,Bagozzi,&Warshaw,1992;Venkatesh,2000).The hedonic aspect of Web performance is the evaluation of a website based on the assessment by users regarding the amount of fun,playfulness,and pleasure they430M.-H.Huang/Computers in Human Behavior19(2003)425–442experience or anticipate fromthe site.It reflects a website’s entertainm ent value derived from its sensory attributes,from which users obtain consummatory affective gratification(Batra&Ahtola,1990,Crowley,Spangenberg,&Hughes,1992).A website performs well in the hedonic aspect when users perceive the site to be enjoyable in its own right,apart from any performance consequences that may be anticipated(Davis et al.,1992;Igbaria,Schiffman,&Wieckowski,1994;Venkatesh, 2000).3.Methodology3.1.ProcedureThe data for the study were collected in May and June2000through a ques-tionnaire survey of Web users with varied demographic profiles and Web usage levels.The initial questionnaire was generated fromexisting academ ic and practi-tioner-oriented literature in electronic commerce and marketing.It was tested on Web users who either worked for companies in information management related positions or who studied in management relatedfields.The research team contacted more than400Web users through their academic and continuing education pro-grams and obtained243valid questionnaires.Each Web user wasfirst asked to rate Web descriptors according to how appro-priately they described the characteristics of the websites that they most frequently visited.This approach covers all types of websites,an important requirement for any attempt to construct a general description of Web attributes.The fact that these websites are the most frequently visited by the user suggests that they successfully meet the user’s needs.After rating the descriptors,each user was then asked to indicate his or herflow state in the site and evaluate its performance.3.2.MeasuresThefinal questionnaire consisted of37items for15Web attributes,12experiential flow measures,and10Web performance measures(see Appendix).First,the Web characteristics were described by15descriptors capturing the complexity,novelty, and interactivity attributes.The15descriptors were purified from58Web attributes on the basis of the results of exploratory and confirmatory factor analyses.The complexity attribute contained the items multidimensionality,richness,multiplicity, large size of scale,and breadth.The novelty attribute included the items imagina-tion,surprise,innovation,and novelty.Finally,the interactivity attribute was com-posed of items such as activity,responsiveness,interactivity,participation, dynamics,and demonstrability.These attributes represent the attributes of websites as perceived by their users,and are potentially useful cues for users and website designers.Second,a12-itemLikertflow scale was used to m easure the experientialflow in the site(e.g.‘‘This website allowed me to control the computer interaction’’orM.-H.Huang/Computers in Human Behavior19(2003)425–442431‘‘When navigating through this website,I was aware of distractions’’)(Webster et al.,1993).This scale contains four components with three questions per component. Responses were scored on seven-point scales ranging fromstrongly disagree(1)to strongly agree(7).Finally,the evaluation of the site performance was measured using a10-item two-factor7-point semantic differential Web performance scale.The10items were selected froma pool of42candidates,which was collected through a review of the relevant literature,and on the basis of the results of confirmatory factor analyses. This scale is useful for assessing the competitive advantages of websites and the benefits derived from them.Evaluative criteria such as order,wisdom,reliability, effectiveness,and correctness were useful for determining the utilitarian aspects of Web performance,whereas evaluative criteria such as pleasant,nice,entertaining, agreeable,and soothing provided an assessment of the hedonic Web performance.3.3.AnalysisWe used structural equation modeling,which is more powerful than a simple corre-lation,to analyze the link between Web attributes,experientialflow,and Web perfor-mance(see Fig.2).In the model,Web attributes were assumed to have direct impacts on experientialflow,and have direct and indirect impacts on the utilitarian and hedonic aspects of Web performance.Specifically,users are hypothesized to judge Web perfor-mance based on the Web attributes and on whether they experienceflow in the site.In addition to examining the overall measure offlow,an analysis was also con-ducted using the four components offlow(see Fig.3).In this model,the direct impacts of Web attributes on Web performance were omitted from the analysis for simplicity.Significant impacts(P<0.1)arereported. Array Fig.2.The direct and indirect impacts of Web attributes on the utilitarian and hedonic aspects of Web performance,treatingflow as an overall experience.!Positive impact;ÁÁÁÁc Negative impact;$Corre-lation.Path coefficients and t-values are reported.4.Results4.1.Descriptive statisticsThe majority of the Web users were male (59.3%)with ages ranging from 18to 51.The user profile was in agreement with the regular Georgia Tech Graphics,Visuali-zation &Usability (GVU)survey distribution of males and females (Bellman,Lohse,&Johnson,1999).The users surveyed tended to spend more than 3h per week on the Web (74.6%),with 24%spending more than 10h per ers most commonly use the Web for a combination of work and entertainment (75.3%).The websites that they most frequently visited fell primarily into the categories Enter-tainment (19.2%),News and Media (18.5%),Business and Economy (16.6%),Reference (11.3%),and Computers and Internet (9.9%).4.2.Major findingsThe results of the survey are fourfold.First,we used accepted psychometric tech-niques to identify viable Web attributes and aspects of Web performance.FromtheFig.3.The direct impacts of Web attributes on the experiential flow,and the indirect impacts on the utilitarian and hedonic aspects of Web performance.!Positive impact;ÁÁÁÁc Negative impact;$Corre-lation.Thick arrows denote the paths being significant at 0.05(|t -value|>1.96),while thin arrows denote the paths being significant at 0.1(|t -value|>1.64).Path coefficients and t -values are reported.432M.-H.Huang /Computers in Human Behavior 19(2003)425–442M.-H.Huang/Computers in Human Behavior19(2003)425–442433 confirmatory factor results we captured the existence of three broad categories of Web attributes,designated as complexity,novelty,and interactivity.In total these three broad sets consisted of15specific attributes associated with the most fre-quently visited websites(see Appendix).The reliability and variance extracted(in brackets)for complexity,novelty and interactivity were0.80(42.9%),0.67(34.1%), and0.70(28.4%)respectively.The dimensionality of the site performance scale was similarly confirmed.The reliability and variance extracted(in brackets)for utilitar-ian and hedonic performance were0.86(54.6%)and0.89(62.6%)respectively. Second,results of confirmatory factor analyses for theflow scale suggested the omission of one unreliable statement,leaving11questions representing Control, Attention,Curiosity,and Interest.The reliability and variance extracted(in brack-ets)for control,attention,curiosity,and interest were0.68(52.5%),0.82(60.6%), 0.71(45.2%),and0.72(48.3%),respectively.Third,treatingflow as an overall experience results of the structural equation modeling analysis showed that Web attributes are linked to experientialflow differ-entially,and can meet both goal-oriented and experiential needs directly(see Fig.2). Complex websites are deemed by users to be useful(coefficient=0.19,t=2.79). Interactivity is the key to creating experientialflow(coefficient=0.27,t=2.73)and is the most important attribute for determining hedonic Web performance(coeffi-cient=0.39,t=2.97).Novelty is contingent in that it facilitatesflow(coeffi-cient=0.13,t=2.17),but undermines hedonic Web performance (coefficient=À0.24,t=À2.62).Analysis of the components offlow(see Fig.3) showed that complexity distracted attention(coefficient=À0.27,t=À1.93),novelty excited curiosity(coefficient=0.27,t=3.18),and interactivity increased control (coefficient=0.67,t=3.30),curiosity(coefficient=0.57,t=3.80),and interest(coef-ficient=0.38,t=3.30).Fourth,experientialflow was found to enhance both utilitarian(coefficient=0.57, t=2.56)and hedonic(coefficient=1.17,t=3.44)Web performance(see Fig.2). Thus,our study establishes the following relationship between online user experi-ence and online marketing outcome variables:the more profoundly users are in a flow state during their site visits,the more likely they are to consider the site as useful and pleasant.Consideration of the components offlow(see Fig.3)showed that attention is utilitarian(coefficient=0.07,t=1.75),control(coefficient=0.20, t=2.22)and interest(coefficient=0.80,t=3.80)are more hedonic than utilitarian (shown in thick arrows pointing to the hedonic performance),and curiosity is both utilitarian(coefficient=0.23,t=1.97)and hedonic(coefficient=0.33,t=2.52).The different components of experientialflow can be used strategically for specific man-agement purposes.5.DiscussionWhat are the implications for management and research of thesefindings?By identifying the major Web attributes,assessing their connection to experientialflow, and linking both to utilitarian and hedonic Web performance,this article presents434M.-H.Huang/Computers in Human Behavior19(2003)425–442the key features influencing information seekers and entertainment surfers and thus facilitates the process of designing and evaluating websites.ing Web attributes to induce experientialflowThe present results show that complexity distracts attention,novelty excites curi-osity,and interactivity increases control,curiosity,and interest.Thesefindings sug-gest the following strategic use of Web attributes to induce experientialflow.plexity distracts attentionWe found that complexity tends to distract users from relevant information. When faced with an abundance of information,users feel unable to absorb it,are more easily distracted,and think about other things during site ers may be overwhelmed by the perceived complexity of a website,and consequently find it difficult to concentrate.Our results are consistent with the information over-load perspective,which suggests a negative relation between complexity andflow (Hoffman&Novak,1996;Hwang&Lin,1999).Users often feel stressed when required to deal with excessive information.When users are confronted by a site offering a wide variety of options,they may be overwhelmed and believe that their skills are inadequate for meeting the challenges posed by the plexity represents a challenge to users’cognitive capacity;thus,users often perceive a web-site with too much information to be unpleasant.For this reason,‘‘keeping it sim-ple’’is an acknowledged value-adding practice that is used by leading e-businesses such as FedEx,Alamo,Direct Line and Dell Excel to make it easy for customers to do business with them(Willcocks&Plant,2001).5.1.2.Novelty excites curiosityWe found that novelty can act as a curiosity generating mechanism that arouses the imaginations of users and captures their interest in a site.Similarfindings were obtained elsewhere.In a study of catalog shopping,Stell and Paden(1999)found that novel stimuli arouse the curiosity of readers and motivate them to examine the catalog information further.Novelty is thought to be an innate human preference, and varied,novel,and surprising stimuli can elicit sensory curiosity(Bianchi,1998). Users gain excitement and pleasure from seeking out new things.Therefore,incor-porating novel elements into a website can attract curious users and bring out the enjoyable experience offlow.5.1.3.Interactivity increases control,curiosity,and interestWe found that interactivity has a consistent positive impact on the control,curi-osity,and interest components offlow,but has little effect on the attention compo-nent.Similarfindings have been reported by Novak et al.(2000),who found that higher levels of interactivity are not associated with more focused attention.In summary,interactivity enhances the subjective feeling of‘‘having control’’over the interaction with the site,excites the curiosity of users,and makes navigation intrin-sically interesting.Interactivity is considered an important determinant of experientialflow(Novak et al.,2000).Interactive Web attributes increase‘‘the degree to which participants in a communication process can exchange roles and have control over their mutual discourse’’(Rogers,1995).Because the Web is a computer-driven environment,it can performa variety of interactive input–output functions.Creating sites that can receive and respond to input fromusers is the essence of Web interactivity,because the interactivity offered by such sites not only enhances users’sense of control in a manner that is not possible with more static technologies(Ha&James,1998),but also enhances and alters the entertainment experience of users(Bryant&Love, 1996).Functionalities such as e-mails,questions and answers on a website can pro-vide users with such an interactive environment.For example,the interactive Web activities of users of Internet discussion groups,including information retrieval on the Web,reading and posting on newsgroups,and reading and replying to e-mail, have been found to be associated withflow during Web use(Chen et al.,1999).5.2.The consequences offlow:utilitarian or hedonic?The results of our study indicate thatflow elicits favorable Web evaluations for both the utilitarian and hedonic aspects.Flow was found to have a stronger impact on the hedonic aspect,as reflected in the greater number of components offlow that are linked to hedonic performance(see the thick arrows in Fig.3).The components offlow naturally distinguish themselves into the four categories cited above,with attention facilitating utilitarian performance,curiosity facilitating both the utilitarian and hedonic aspects of Web performance,and control and interest exerting greater influence over hedonic performance.This suggests thatflow can be both utilitarian and hedonic,depending on the sources of theflow experience.5.2.1.Attention is utilitarianWe found that attention is a significant facilitator of utilitarian Web performance. Attention is a series of activities in which users selectively allocate cognitive resour-ces(Kahneman,1973).Because human cognitive resources are limited,it is impos-sible for users to process all of the information available at any given moment. Consequently,the cognitive systemis constantly selecting inform ation for further ers use the attention mechanism to control the choice of stimuli that will be allowed,in turn,to control his or her behavior.Retaining the attention of users is therefore ers who are completely absorbed in navigating the Web are more likely to retain what they perceive than users who are not.Attention facilitates the learning needed to progress to instrumental usage involving browsing and searching(Hoffman&Novak,1996;Novak et al.,2000;Webster et al.,1993).5.2.2.Curiosity is both utilitarian and hedonicThe curiosity component offlow is a facilitator of both utilitarian and hedonic Web performance.Curiosity is seen as a motivational variable that represents desire for information(Loewenstein,1994).It can be an extrinsically motivated desire for information to resolve uncertainty(Loewenstein,1994).This type of informationseeking is the goal-directed Web use that treats the acquisition of information as a means to some further end(Steenkamp&Baumgartner,1992).Curiosity can also be an intrinsically motivated desire for information that is not driven by a utilitarian goal(Posnock,1991).This type of information seeking is the experiential Web use that treats the acquisition of information as an end in itself(Steenkamp&Baum-gartner,1992).The two sources of curiosity and their differential linkages to the two aspects of Web performance suggest that the outcome(the utilitarian aspect)and the process(the hedonic aspect)of information acquisition are both relevant to Web performance.Web users may examine and explore websites in which they have little interest simply because their curiosity is aroused by the characteristics of the sites.5.2.3.Control and interest are more hedonicThe control and interest components are dual-facilitators of Web performance, although they have a greater impact on hedonic performance than on utilitarian performance.Control implies the freedom to act on the ers who feel that they have control over the human and computer interaction,and who feel in control of actions andfinal choices,feel more confident than users who do not feel in control.In a study of Internet discussion groups,Web users reported that they felt a sense of control when engaged in Web activities such as navigation and searching(Chen et al.,1999).These Web activities not only enhance the sense of control,but also increase the enjoyment of Web users.In experiential Web use,users are intrinsically interested in the process of naviga-tion,not the outcome of ers seekflow primarily because it brings themenjoym ent.For exam ple,Hoffman and Novak(1996)suggested that Web surfers exploring the Web in their daily quest for the latest interesting sites do so in an intrinsically motivated experientialflow state.It represents the intrinsic motiva-tion of users toward‘‘playfulness’’and‘‘having fun’’during computer interactions, which is related to feelings of pleasure and satisfaction derived fromwebsites(Ven-katesh,2000).6.Implications6.1.Three keys to inducing experiential encountersThis article lends credence to the use of Web attributes to induce experiential encounters.The empirical results verify that websites can be designed with attributes that shape experientialflow,and both the Web attributes and this experientialflow determine the utilitarian and hedonic Web performance.6.1.1.Interactivity is the keyInteractivity is a critical concept in computer-mediated communications,because it is the key advantage of the medium(Burgoon et al.,1999/2000;Ha&James, 1998).Despite its importance in Web management,the effective use of interactivity。
Pool boiling heat transfer on copper foam covers with water as working fluid
Pool boiling heat transfer on copper foam covers with water as working fluidYongping Yang,Xianbing Ji,Jinliang Xu *Beijing Key Laboratory of New and Renewable Energy,North China Electric Power University,Beijing 102206,PR Chinaa r t i c l e i n f oArticle history:Received 26July 2009Received in revised form 23December 2009Accepted 12January 2010Available online 12February 2010Keywords:Copper foam Pool boiling Boiling curvea b s t r a c tPool boiling heat transfer with porous media as the enhanced structure is attractive due to its simple geometry and easy operation.However,the available studies focus on low porous porosities.Metallic foams provide large porous porosities that have been less studied in the literature.In this paper a set of copper foam pieces were welded on the plain copper surface to form the copper foam covers for the pool boiling heat transfer enhancement.Water was used as the working fluid.Enhancement of pool boiling heat transfer compared with plain surface depends on the increased bubble nucleation sites,extended heat transfer area,and resistance for vapor release to the pool liquid.Effects of pores per inch (ppi)of foam covers,foam cover thickness,and pool liquid temperatures are examined.It is found that temperatures at the Onset of Nucleate Boiling (ONB)are signi ficantly decreased on copper foam covers compared with on plain surfaces.Heat transfer coef ficients of foam covers are two to three times of the plain surface.A large ppi value provides large bubble nucleation sites and heat transfer area to enhance heat transfer,but generates large vapor release resistance to deteriorate heat transfer.Therefore an optimal ppi value exists,which is 60ppi in this paper.Generally small ppi value needs large foam cover thickness,and large ppi value needs small foam cover thickness,to maximally enhance heat transfer.Effect of pool liquid temperature on the heat transfer enhancement depends on the ppi value.For small ppi value such as 30ppi,lower pool liquid temperature can dissipate higher heat flux at the same wall superheat.However,the heat transfer performance is insensitive to the pool liquid temperatures when large ppi values such as 90ppi are used.Ó2010Elsevier Masson SAS.All rights reserved.1.IntroductionDue to the limitation of the conventional forced convective fan cooling method and increased heat generation of electronic components,reliable cooling methods for electronic devices shall be developed.The required heat removal rate for a CPU is 6.25W/cm 2or more and a printed circuit board produces about 10W/cm 2.In the near future the heat flux of the electronic devices will be more than 100W/cm 2.The widely used fan cooling method using forced convective air as the working fluid,however,is not suitable for high heat fluxes such as larger than 10W/cm 2.Besides,it has problems such as noise,electrical failure,and high power consumption.Therefore,a direct immersion cooling has been suggested as an alternative to the forced convective air cooling methods.Liquids have advantages of higher thermal conductivity,density,and speci fic heat over air,ensuring high heat flux that is to be dissipated.Many studies have tried to improve the pool boiling heat transfer by coating a heater surface with a thin,porous layer of particles and reduce the cost and size of equipments.The available studies focus on the boiling heat transfer enhancement using porous media with low porosities.Bergles and Chyu [1]studied pool boiling heat transfer using a porous structure with a porous porosity of 50e 65%,a porous-layer thickness of 0.38mm,the range of particle sizes of 75%between 74m m and 44m m.The result shows that the porous coating can improve boiling heat transfer signi ficantly and decrease the surface superheat.Rainey and You [2]performed an experimental study of “double enhancement ”behavior in pool boiling from heater surfaces simu-lating microelectronic devices immersed in saturated FC-72at atmospheric pressure.The term “double enhancement ”refers to the combination of two different enhancement structure enhancement techniques:a large-scale area enhancement (square pin fin array)and a small-scale surface enhancement (microporous coating).Results showed signi ficant increases in nucleate boiling heat transfer coef ficients with the microporous coating to the heater surface.Liter and Kaviany [3]demonstrated modulated (periodically non-uniform thickness)porous-layer coatings,as an example of*Corresponding author.Tel./fax:þ861051976819.E-mail address:xjl@ (J.Xu).Contents lists available at ScienceDirectInternational Journal of Thermal Sciencesjou rn al homepage:/locate/ijts1290-0729/$e see front matter Ó2010Elsevier Masson SAS.All rights reserved.doi:10.1016/j.ijthermalsci.2010.01.013International Journal of Thermal Sciences 49(2010)1227e 1237capillary artery-evaporator systems,to enhance the pool boilingcritical heat flux nearly three times over that of a plain surface.The modulation separates the liquid and vapor phases,thus reducing the liquid-vapor counter flow resistance adjacent to the surface.Theories are suggested for two independent mechanisms capable of causing the liquid choking that leads to the critical heat flux.Kim et al.[4]studied the nucleate pool boiling heat transfer enhancement mechanism of microporous surfaces immersed in saturated FC-72.They measured bubble size,frequency,and vapor flow rate from a plain and microporous platinum wire using the consecutive photo technique.It is found that the microporous coatings enhance nucleate boiling performance through increased latent heat transfer in the low heat flux region and through increased convection heat transfer in the high heat flux region.Ghiu and Joshi [5]conducted visualization study at atmospheric pressure from top covered enhanced structure for a dielectric fluorocarbon liquid (PF 5060).The single layer enhanced structures were fabricated in copper and quartz,having an overall size of 10mm by 10mm and 1mm thick.The heat transfer performance of the enhanced structures was found to depend weakly on the channel width.The internal evaporation has a signi ficant contri-bution to the total heat dissipation.Parker and El-Genk [6]studied enhancements in nucleate boiling of FC-72liquid on porous graphite and compared results with those on a smooth copper surface of the same dimensions (10mm by 10mm).Also investigated is the surface temperature excursion at boiling incipience and the obtained values of CHF are compared with those of other investigators.Results showed no temperature excursion at boiling incipience on porous graphite but as much as 14K on plain copper surface.The heat transfer coef fi-cients are signi ficantly higher than those on copper surface and the values of CHF are 63e 94%higher than on copper surface.Hwang and Kaviany [7]found that the porous surface can enhance the critical heat flux (q CHF )and reduce the superheat across the wick in pool boiling.Recently Min et al.[8]developed a new fabrication method (hot-powder compaction)to make 2-D and 3-D modulated coat-ings for enhanced pool boiling performance.The results show that the maximum measured critical heat flux (q CHF )of 2-D and 3-D modulated coatings are 3.3and 2.0times that of the surface without coatings (plain).The critical heat fluxes strongly depend on the modulation wavelength,while particle diameter and porosity have little effects.The porous porosity of their study is 43.8%.The above review of the pool boiling heat transfer on porous media surfaces such as porous coatings (particles),porous graphite refers to low porous porosity such as less than 65%.Metallic foam is a new kind of porous media for the heat transfer applications,having large surface to volume ratio and low density.Metallic foam is a structure characterized by thin fibers,or ligaments,of metal joining several others in a random manner throughout the volume.The porous porosity can be larger than 90%.Fewer studies have been reported on the pool boiling heat transfer enhancement with metallic foam structures in the open literature.Several reports on this topic can be found in some conference papers which are shortly described as follows.Arbelaez et al.[9]reported pool boiling heat transfer of FC-72in highly porous metal foam heat sinks.The porous porosities were in the range of 90e 98%and pore sizes had the range of 5e 40ppi.It is shown that the temperature excursion usually observed for fluo-rinert fluids at the onset of nucleate boiling is not present.The low porosity samples exhibit a signi ficantly enhanced heat transfer in the low heat flux regions for the same pore size.Enhanced heat transfer is observed with an increase in the foam ppi for similar porosity.Athreya et al.[10]studied effects of orientation and geometry on the pool boiling heat transfer of FC-72in high porosity aluminum metal foam heat sinks.It is found that high ppi samples deteriorate heat transfer in the vertical orientation.The low ppi sample first decreases and then increases the heat transfer coef ficients with reduction in the foam height.Moghaddam and Ohadi [11]investigated pool boiling heat transfer of water and FC-72on thin blocks bonded with copper foams of 80ppi,90%porosity,30ppi,95%porosity,and graphite foam of 75%porosity.On the 30ppi copper foams,signi ficant enhancement was observed in boiling of water.But no enhance-ment was observed on the 80ppi copper and graphite foams.A substantial enhancement was achieved on all the foams with FC-72as the working fluid.Y.Yang et al./International Journal of Thermal Sciences 49(2010)1227e 12371228In this paper we study the pool boiling heat transfer on copper foam covers with water as the workingfluid.Experiments were performed at atmospheric pressure.Significant heat transfer enhancement is observed.By using copper foam covers tempera-tures at the ONB can be decreased by13K,maximally.Heat transfer coefficients with copper foam cells can be two to three times of those with plain surface,maximally.Bubble nucleation sites, extended heat transfer area and resistance for produced vapor release are the key factors to influence the enhanced heat transfer on copper foam covers.There are optimal ppi values and foam cover thickness to enhance heat transfer,maximally.Coupling of pool liquid temperature and foam cover thickness is observed to influence heat transfer.2.Copper foam parametersCopper foams for the ppi of30,60and90at the porosity of0.88 are shown in Fig.1(a e c).Photos of foam covers with the porosity of 0.95are not given here because they have small difference with those for the porosity of0.88.Copper foams have open-celled structures composed of dodecahedron-like cells,possessing12to14pentagonal or hexagonal faces.Porosity and ppi are the two parameters to influenceflow and heat transfer.The ligament cross section depends on porosity,and changes from a circle at3¼0.85to an inner concave at3¼0.97,where3is the porosity(Calmidi[12]).A unit cell of the foam is shown in Fig.1d,with the assumed tetrakaidecahedron shape.The ligament length is l with its diameter of d f.Thus the volume of a unit cell is V¼8ffiffiffi2pl3.The circumcircle diameter of the foam cell isffiffiffiffiffiffi10pl,which can be regarded as the pore diameter of d p, i.e.d p¼ffiffiffiffiffiffi10pl,not considering the ligament thickness.The foam cell parameters were measured by a Leica M-type microscope(Germany)and are given in Table1.The larger the ppi, the smaller the pore diameter of d p and ligament diameter of d f are. At the given ppi,larger porosity such as0.95leads to slight larger d p and smaller d f,compared with the lower porosity of0.88.3.The test section and experimental setupThe geometry and dimensions of the copper block test section are given in Fig.2a.There arefive6.0mm diameter holes in which five cartridge heaters were inserted,providing heating power to the copper block,in the bottom part of the copper block.A maximum power of100W at the applied AC voltage of220V can be provided by each heater.The copper block has a middle part,having three 1.0mm diameter holes,inside which three K-type thermocouples are inserted.A rectangular plate with a thickness of3.0mm is located at the top of the copper block.A plain smooth,sand pol-ished copper surface is regarded as the reference surface for the boiling heat transfer experiment,with the size of12.0mm by 12.0mm.For the pool boiling heat transfer enhancement,the top copper surface was welded with copper foam covers,withfive different thicknesses of1.0,2.0,3.0,4.0and5.0mm,respectively. The copper block was cleaned by methanol and baked in an oven. Then it was taken out of the oven and heated by thecartridge Fig.1.The copper foam photos for3¼0.88(a,b and c)and a unit foam cell(d).Y.Yang et al./International Journal of Thermal Sciences49(2010)1227e12371229heaters until its temperature reached the melting temperature of the tin at the top copper surface,leaving a thin tin film.The esti-mated tin thickness at the copper surface is 0.1mm,which was about 2e 5%of the total foam cover thickness.A clean copper foam cover was being put on the copper surface.The copper foam was welded with the copper block tightly by turning off the cartridge heaters.The thermal resistance was small between the copper surface and foam cells by the welding technique.Then the whole copper block assembly was ready for the experiment.Fig.2b shows the experimental setup and measurement systems.A transparent glass chamber containing the experimental equipments and liquid has the size of 125Â127Â145mm.A stainless steel plate (3)is located at the bottom of the glass chamber.A rectangular hole was drilled at the center of the stain-less steel plate to fit the copper block and the stainless steel plate by filling Te flon and epoxy glue between them for seal.The hardware arrangement ensures the copper foam exposed in the pool liquid.The part of the copper block under the plate (3)was surrounded by a glass sheath (1).As the thermal insulation material,glass fiber was filled in the gap between the copper block (2)and glass sheath (1),as shown in Fig.2b.A stainless steel plate (6)with a 2.0mm thickness forms the top cover of the glass chamber.An inclined 6.0mm diameter coiled copper tube (11)was arranged along the internal wall surface of the glass chamber.There are two holes on the plate (6)to fit the two ports of the coiled tube.The tap water is flowing in the copper tube (11)to yield a desired pool liquid temperature by adjusting the flow rate of tap water using the valve (10).In a corner of the glass chamber there is an auxiliary heater (12).The auxiliary heater (12)was turned on automatically meanwhile the valve (10)was turned off if the pool liquid temperature was below the desired value.This situation only took place at small heating power applied on the test section.For most cases it is necessary to maintain a suitable flow rate of the tap water in tube (11),with the auxiliary heater (12)turned off.The boiling induced vapor entered a re flux condenser (7).The condensed liquid returned to the glass chamber by gravity.The forced convective air dissipates heat to the environment through a fin heat sink.The condenser was vented to atmosphere by the valve (8)through a side branch tube.Thus,atmospheric pressure was always kept in the glass chamber.A K-type thermo-couple (9)measures the pool liquid temperature.The right part of Fig.2b shows the power supply and measurement systems.The power supply system consists of a 220V voltage stabilizer,a voltage transformer and a power meter,giving the power reading.The pool liquid temperature and three thermocouple signals were recorded by a Hewlett e Packard data acquisition system (see Fig.2b).Before the formal experiment,we charge liquid (water)in the glass chamber and remove the non-condensable gas in the liquid.The copper foam cover was horizontally positioned.The top liquid level was higher than the top foam cover by 100mm.The cartridge heaters were turned on to vigorously boil the liquid for one hour to remove the non-condensable gas in foam cells and liquid.After the pool liquid reaches the environment temperature the whole system is ready for experiment.Water has good thermal performance and it is non-flammable,non-poisonous.Thus it is compatible to many kinds of material and is widely used as the working fluid.Pool boiling heat transfer experiments using water as the working fluid can be found in refs.[13e 17].An alternative liquid widely used in pool boiling heat transfer experiments is FC-72,such as reported in refs.[2,4,6,9].Water has larger surface tension force and latent heat of evaporation than other fluids such as FC-72.The physical prop-erties of water and vapor at atmospheric pressure are listed in Table 2.During each experiment,we started from a small heat flux 1e 2W/cm 2on the copper foam and speci fied the pool liquid temperature.The heat transfer was considered to reach a steady state if the variation of the copper block temperature was smaller than 1 C in ten minutes.We recorded the pool liquid temperature,the three temperatures on the copper block and the power meter reading.Then the heat flux is increased by a small step of 2-5W/cm 2,and the above procedure is repeated.In the present study,the copper block below the surface immersed in the pool liquid of water was well thermallyinsulated.abFig.2.Copper block test section (a)and experimental setup (b),all dimensions are in mm.Table 1Y.Yang et al./International Journal of Thermal Sciences 49(2010)1227e 12371230Thus one-dimensional thermal conduction heat transfer within the middle part containing T 1,T 2and T 3(see Fig.2a)can be assumed.Such assumption can also be found in refs.[18e 20].Based on the one-dimensional heat conduction equation,de finition of the heatflux is written as q ¼Àk s dTdz j base surface,where k s is the copper thermal conductivity,dT j base surface is the temperature gradient atthe base surface,z is the coordinate perpendicular to the base surface.A least square correlation of temperatures versus z was written as T ¼a 0þa 1z,where a 0,and a 1are constants correlated based on T 1,T 2,and T 3(see Fig.2a).The heat flux uncertainty was estimated to be smaller than 6.0%.The surface superheat D T sat is de fined as the surface temperature of T w subtracting T sat ,where T w is the temperature at the base surface,T sat is the saturation temperature of water at atmospheric pressure.Heat transfer coef-ficient is calculated ash ¼q =ðT w ÀT bulk Þ(1)where T bulk is the pool liquid temperature.The surface tempera-ture,surface superheat,and pool liquid temperature have themaximum uncertainties of 0.3 C.Performing the standard uncer-tainty analysis,we obtain the maximum relative uncertainty of h of 8.52%.In order to evaluate the heat transfer performance enhanced by copper foams,a heat transfer enhancement ratio is de fined as the heat transfer coef ficient on copper foam covers divided by that on plain smooth surface,i.e.,En ¼h /h o .This study covers the following data ranges:ppi of 30,60,90;porosity of 0.88and 0.95;foam cover thickness of 1.0,2.0,3.0,4.0and 5.0mm;surface superheat from À10to 23K;surface heat flux up to 171W/cm 2.It is noted that the heat flux is based on the top copper surface area of 12.0mm by 12.0mm.The foam cell area is not involved in the computation of heat flux.4.Results and discussion4.1.Effect of foam ppi on the heat transfer performancePores per inch (ppi)strongly in fluences pore diameter of d p ,affecting liquid suction towards the foam covers and vapor release from the foam cells.Thus,the value of ppi has signi ficant effect on the pool boiling heat transfer.Fig.3shows effect of ppi on boiling curves for various conditions.It is seen that foam covers signi fi-cantly enhance heat transfer.For comparison,the wall superheats are in the range of 11e 15K at the boiling incipience on the plain smooth surface.Boiling incipience takes place at low wall super-heats such as 1e 4K when foam covers are used (see Fig.3).For the foam porosity of 0.88shown in Fig.3(a e b),small differences of boiling curves are identi fied at low wall superheats for D T sat <10K,especially for the pool liquid temperature of 60 C and foam cover thickness of 3.0mm (see Fig.3a).Boiling curves are identi fied to be different among various ppi if wall superheats are larger than 10K,for which 60ppi foam covers have better thermal performance than 30and 90ppi foam covers.For porosity of 0.95,boiling curves are different for the three foam covers of 30,60and 90ppi over the whole range of wall superheats.Foam covers of 60ppi have better thermal performance than the others.An effective validation of the experimental results is to compare the obtained critical heat fluxes with predictions by the well known Zuber correlation [21],which is written asq CHF ¼ph fg r 0:5fÀs g Àr f Àr g ÁÁ0:25(2)The computed critical heat flux by Eq.(2)is 110W/cm 2for thesaturation pool boiling heat transfer on the large plain surface with water as the working fluid.Saylor et al.[22]noted that q CHF was relatively constant for large heater surfaces and increased for decreasing heater size past a certain transition point.Bar-Cohen and McNeil [23]suggested the dimensionless transition heater size asL tran ¼L Àg Àr f Àr g Á=s Á0:5¼20(3)where L is the transition heater size.The present heater size of 1.2cm is suf ficiently smaller than the transition heater size of 5cm by Eq.(3).Thus the predicted critical heat flux should consider the heater size effect and be larger than the value of 110W/cm 2by Eq.(2).Rainey and You [24]recommended the experimental deter-mined curve of q CHF,small /q CHF,large ,which is 1.22for the present case,where q CHF,small and q CHF,large are the critical heat fluxes at small heater size and large heater size,respectively.Therefore the pre-dicted critical heat flux considering the small heater size effect is 135W/cm 2.Our measured critical heat flux is 165W/cm 2for the saturated boiling heat transfer on the smooth plain surface,which is 22%higher than the predicted value of 135W/cm 2,showing the reasonable results that we obtained.It is noted that critical heat fluxes are higher for the subcooled boiling heat transfer than those for the saturated boiling heat transfer.Critical heat flux data on the subcooled boiling heat transfer are not obtained in the present paper because temperatures at the bottom part of the copper block test section are very high.Data of critical heat flux on the subcooled boiling heat transfer are not marked in the figures in the present paper.Enhancement of boiling heat transfer on porous surface depends on the balance between the liquid suction capability towards the porous structure,and the vapor release resistance to the pool liquid environment.Heat transfer coef ficients are given in Fig.4for both the plain smooth surface and copper foam surfaces.Slopes of heat transfer coef ficients versus heat fluxes are decreased when the heat flux q is larger than 60W/cm 2for the two porosities of 0.88and 0.95(see Fig.4a and c).At high heat fluxes,boiling inside foam structures are violent.The increased vapor release resistance to the pool liquid decreases slopes of heat transfer coef ficients against heat fluxes.As shown in Fig.4a,heat transfer coef ficients are suddenly decreased at the heat flux of about 60W/cm 2for the 90ppi foam covers.The right column of Fig.4gives the heat transfer enhancement ratios versus heat fluxes (see Fig.4b and d).For both the two sub figures,the heat transfer enhancement ratios are initially increased to a maximum value at low heat fluxes.Beyond the maximum point the heat transfer enhancement ratios (h /h o )are decreased with increases in heat fluxes.The two sub-figures of Fig.4b and d further identify that nucleate boiling is dominated by the increased nucleation sites in foam structures at low heat fluxes.But the increased vapor release resistance decreased the heat transfer enhancement ratios at largerheatTable 2Y.Yang et al./International Journal of Thermal Sciences 49(2010)1227e 12371231fluxes.The heat transfer coef ficients are more than three times of those on the plain smooth surface maximally for the 60ppi foam covers.For all the cases demonstrated in Fig.4,the heat transfer enhancement ratios are larger than unity.Fig.5shows the heat transfer enhancement ratios versus foam ppi.Under the same conditions,heat transfer enhancement ratios display parabola distributions and attain maximum values at 60ppi,showing great effect of foam ppi on the heat transfer performance.This phenomenon is explained as follows.In a general sense,enhancement of pool boiling heat transfer on foam covers is attributed to the combined effect of an extended surface area,an increased nucleation site density,the resistance for vapor release from the foam cells,and a capillary-assist liquid flow towards the foam cells.The liquid supply and vapor release occur as a liquid e vapor counter flow resisting each others'motion.Melén-dez and Reyes [25]gave a correlation to compute the vapor flow rate escaping from the porous coverings:m ¼p128 r g sm g!3d 3p d!(4)where r g and m g are the vapor density and viscosity,respectively.Eq.(4)indicates the in fluence of the thermal physical properties (r g ,s ,m g )and the porous parameters (3,d p ,d ).A larger vapor mass flow rate represents a smaller resistance for vapor release.At the same porosity 3,low ppi foam covers have large pore size of d p ,leading to a large value of m ,which is helpful for the heat transfer argument.On the other hand,low ppi foam covers have larger pore diameter of d p ,leading to the decreased capillary pumping of liquidflow towards foam cells characterized by 2s /d p .Due to the above two opposite effects of the ppi values on the pool boiling heat transfer,there is an optimal ppi value for the pool boiling heat transfer enhancement,which is 60ppi in this paper.4.2.Effect of foam cover thicknessFoam cover thickness also signi ficantly affects the heat transfer performance.Fig.6shows the boiling curves with different foam cover thicknesses.It is noted that different set of thicknesses of foam covers were used for different ppi in Fig.6.This is because the optimal foam cover thickness is changed for different ppi.The optimal foam cover thickness is 4.0mm for 60ppi (see Fig.6b)among the four thicknesses of 2.0,3.0,4.0and 5.0mm.For 30ppi foam covers,the thickness of 2.0mm was not tested because it will be broken when it is sliced due to the large pore diameter of 2.76mm (see Table 1).Thus only three thicknesses of 3.0,4.0and 5.0mm were tested (see Fig.6a).The foam cover thickness of 3.0mm provides better thermal performance among the three thicknesses of 3.0,4.0and 5.0mm.For 90ppi foam covers,the minimum thickness that we can fabricate is 1.0mm.As shown in Fig.6c,the thickness of 3.0mm begins to deteriorate the heat transfer,it is not necessary to test the thickness larger than 3.0mm.Thus the foam cover thicknesses of 1.0,2.0and 3.0mm were tested.The optimal thickness is 2.0mm for 90ppi at relative large wall superheats such as D T sat >8K .The general trend is that the optimal foam cover thickness is decreased when the ppi values areincreased.Fig.3.Effect of ppi on boiling curves under the subcooled and saturation pool liquid conditions.Y.Yang et al./International Journal of Thermal Sciences 49(2010)1227e 12371232Similar to the effect of foam ppi value on the nucleate boiling heat transfer,the foam cover thickness also has two opposite effects on the boiling heat rger foam cover thickness provides more nucleation sites and extended heat transfer area,enhancing heating transfer.On the other hand,larger foam cover thickness generates larger vapor release resistance to the pool liquid,dete-riorating heat transfer.Therefore,there is an optimal foam cover thickness to enhance heat transfer.Different foam ppi values have different optima foam cover thicknesses,inferring the combined effect of foam ppi value and foam cover thickness.Fig.7illustrates the boiling curves,heat transfer coef ficients and heat transfer enhancement ratios.Again,a suitable foam cover thickness of 4.0mm yields the optimal heat transfer performance among the four foam cover thicknesses of 2.0,3.0,4.0and 5.0mm for 60ppi.Fig.8shows the heat transfer enhancement ratios versus the foam cover thickness.Maximum heat transfer enhancement ratios are reached at the foam cover thickness of 4.0mm.The available studies of heat transfer on porous media covers are mainly focused on the low porosities such as less than 0.3.Metal foams provide signi ficantly large porosities.For the two porosities of 0.88and 0.95used in the present study,the heat transfer performance shows mini difference between the two porosities for most cases (see Fig.9).However,for the saturation pool boiling heat transfer experiments,it is found that the porosity of 0.88has slightly better thermal performance at larger wall superheats.4.3.Effect of pool liquid temperaturesBoiling curves are provided in Fig.10at pool liquid temperatures of 60,80and 100 C on both plain surface and foam covers.For boiling heat transfer on plain surface,higher pool liquid tempera-tures result in larger heat flux dissipated at the same wall super-heats,due to the easy generation of bubbles and agitated flow field by the bubbles at higher pool liquid temperatures.However,this trend is totally changed when foam covers are used.For all the runs tested in this paper,lower pool liquid temperatures lead to higher heat flux dissipated when other parameters are the same.Heat transfer enhancement in foam cells is caused by the combined effects of increased bubble nucleation sites,increased heat transfer area,and thermal conduction along the foam cell networks.FortheFig.4.Effect of ppi on heat transfer coef ficients under the saturation pool liquid conditions.ppi2030405060708090100h / h o.81.01.21.41.61.82.02.2Fig.5.Effect of ppi on the heat transfer enhancement ratio under the saturation pool liquid conditions.Y.Yang et al./International Journal of Thermal Sciences 49(2010)1227e 12371233。
culture 英语作文
Culture is the heart and soul of a nation,encompassing a wide array of elements such as language,art,music,dance,literature,and traditions.It is the collective expression of a societys history,values,and beliefs,shaping the identity of its people.Language:The most direct way to understand a culture is through its language.It is not just a means of communication but also a repository of a cultures wisdom and knowledge. Each language has unique expressions,idioms,and phrases that reflect the cultural nuances.Art:Art is a visual representation of a cultures aesthetic values and historical events. Paintings,sculptures,and installations tell stories of the past and present,often serving as a mirror to societal changes and developments.Music and Dance:Music and dance are universal languages that transcend borders.They are integral parts of cultural celebrations,rituals,and ceremonies.Traditional music and dance forms are preserved and passed down through generations,maintaining cultural continuity.Literature:Literature is a reflection of a societys intellectual and emotional landscape.It provides insights into the thoughts,dreams,and aspirations of a culture.Classic works of literature often become the cornerstone of a cultures educational curriculum,shaping the minds of its youth.Traditions and Festivals:Traditions and festivals are the heartbeat of a culture.They are occasions that bring communities together,celebrating shared values and beliefs.These events often involve rituals,food,and customs that are unique to a particular culture. Cuisine:Food is a fundamental aspect of culture.It is more than just sustenance it is a form of expression and a way of life.Each culture has its own unique cuisine,which is a result of its history,geography,and climate.Architecture:The buildings and structures of a culture are a testament to its architectural prowess and aesthetic sensibilities.They tell the story of a societys technological advancements and artistic achievements.Fashion:Fashion is a dynamic aspect of culture that reflects the changing trends and tastes of a society.It is a form of selfexpression and a way to assert ones identity within a cultural context.Religion and Philosophy:Religion and philosophy are the spiritual and intellectualbackbone of a culture.They provide a moral compass and a framework for understanding the world and ones place in it.Cultural Exchange:In todays globalized world,cultural exchange is more important than ever.It fosters understanding,tolerance,and respect among different cultures,promoting peace and harmony.Preservation and Evolution:While it is essential to preserve the richness of a culture,it is equally important to allow for evolution and adaptation.Cultures that remain stagnant risk losing their relevance and vitality.Challenges and Opportunities:Cultures face numerous challenges,such as globalization, which can lead to cultural homogenization.However,these challenges also present opportunities for innovation and the creation of a new cultural synthesis. Understanding and appreciating the diversity of cultures is crucial in fostering a more inclusive and harmonious world.By embracing cultural differences,we can learn from one another and enrich our own cultural experiences.。
An implicit upwind algorithm for computing turbulent flows on unstructured grids
Computers Fluids Vol.23,No.1,pp.1–21,1994An Implicit Upwind Algorithm for ComputingTurbulent Flows on Unstructured Grids.W.Kyle Anderson and Daryl L.BonhausNASA Langley Research CenterHampton,Virginia23665–5225An implicit,Navier-Stokes solution algorithm is presented for the computation of turbulentflow on unstructured grids.The inviscidfluxes are computed using an upwind algorithm and the solution is advancedin time using a backward-Euler time-stepping scheme.At each time step,the linear system of equations isapproximately solved with a point-implicit relaxation scheme.This methodology provides a viable and robustalgorithm for computing turbulentflows on unstructured meshes.Results are shown for subsonicflow over a NACA0012airfoil and for transonicflow over a RAE2822 airfoil exhibiting a strong upper-surface shock.In addition,results are shown for3–element and4–elementairfoil configurations.For the calculations,two one–equation turbulence models are utilized.For the NACA0012airfoil,a pressure distribution and force data are compared with other computational results as well as withparisons of computed pressure distributions and velocity profiles with experimental data areshown for the RAE airfoil and for the3–element configuration.For the4–element case,comparisons of surfacepressure distributions with experiment are made.In general,the agreement between the computations and theexperiment is good.1.IntroductionFor computingflows on complicated geometries such as multielement airfoils,the use of unstructured grids offers a good alternative to more traditional methods of analysis.This is primarily due to the promise of dramatically decreased time required to generate grids over complicated geometries.Also,unstructured grids offer the capability to locally adapt the grid to improve the accuracy of the computation without incurring the penalties associated with global refinement.While work remains to be done to fully realize their potential,much progress has been made in computing viscousflows on unstructured grids.While several advances have been made for computing turbulentflow on unstructured grids(e.g.[1] [2]),probably the most mature and widely used code for computing two-dimensional turbulent viscousflow on unstructured grids is that of Mavriplis[3].In this reference,the solution algorithm is a Galerkinfinite-element discretization and a Runge-Kutta time-stepping algorithm is used in conjunction with multigrid to obtain very efficient solutions.The turbulence model predominantly utilized in this code is that of Baldwin and Lomax[4] although extensions have been made to include a two-equation turbulence model[5].Other modifications to this code are presented in reference[6]in which backward-Euler time-differencing is used in conjunction with GMRES [7]to produce results which are competitive with multigrid for the cases considered.The use of upwind differencing offers several advantages over a central-differencing formulation for computing viscousflows.For example,in references[8]and[9],it is clearly shown that with theflux-differencing scheme of Roe[10]the resolution of boundary layer details typically requires only half as many points as with a central-differencing code.As discussed in reference[11],the poor performance of the central-difference formulation is attributed to the scalar artificial dissipation formulas commonly used to damp odd-even oscillations and to provide non-linear stability.For upwind calculations on unstructured grids,Barth[12]has described methodology for utilizing Roe’s approximate Riemann solver[10]for the inviscidflux computations and a Galerkin formulation for the viscous terms.In this work,a sparse matrix solver is used in conjunction with a Runge-Kutta time-stepping algorithm forupdating the solution at each time step.The turbulence model included is that of Baldwin-Barth[13]and sample computations are shown over a single-element airfoil as well as a two-element airfoil in a wind tunnel.For the current study,theflux-difference-splitting of Roe is used for computing the inviscid contribution to the flux and an implicit solver based on backward-Euler time differencing is utilized for updating the solution.At each time step,the linear system of equations is approximately solved with a red-black type relaxation procedure.This method circumvents the need to assemble large matrices and,therefore,significantly reduces the required memory. As in reference[12],the Baldwin-Barth turbulence model is used for computingflows at high Reynolds number.In addition,a recently developed turbulence model due to Spalart and Allmaras[14]is also utilized and comparisons between solutions obtained with each model are shown.Results are shown for subsonicflow over a NACA0012 airfoil and for transonicflow over a RAE2822airfoil as well as for theflow over two multielement airfoils.Detailed comparisons are made with available experimental data.These comparisons include velocity profiles at particular locations along the surface as well as pressure and skin friction distributions.2.SymbolsA matrixA area of control volume.Also used in definition of numericalfluxa speed of soundb column vector for least squaresCFL Courant-Friedichs-Lewy numberC3constant used for Sutherlands lawc chord lengthc b1;c b2;c w1constants used in Spalart-Allmaras turbulence modelc1;c2constants used in Baldwin-Barth turbulence modelD diagonal components of AD ij elements of Dd distance to nearest surfaceE total energy per unit volume~Ffluxes of mass,momentum,and energy~Fiinviscid contribution to thefluxes~Fvviscous contribution to thefluxesf;g components of inviscidfluxesf v;g v components of viscousfluxesf2function used in Baldwin-Barth turbulence modelf w;f t1;f t2functions used in Spalart-Allmaras turbulence modele k1thermal conductivity in freestreamKarman constantL reference lengthl distance between centroids adjacent to an edgel d length of edge in dual meshM1freestream Mach numberN number of edges meeting at a nodeN d number of edges making up the boundary of the control volumeb n unit normal vector~n directed area vectorb n d unit normal to a boundary for dual mesh~n L;~n R directed areas formed by connecting the midpoint of an edge to the centroids of thetriangles to the left and right,respectivelyb n x;b n y x and y components of a unit normalO all off-diagonal components of AP production of turbulent kinetic energyP r Prandtl numberp pressureQ conserved state vector,Q=[ u v E]T.Also used to denote an orthogonal matrixq primitive state vector,q=[ u v p]Tq+;q0primitive state vector on a cell boundary obtained from extrapolationq L;q R primitive state vector at the nodes on either side of an edgeq x0;q yx and y components of gradient at nodeq x;q y components of heatfluxR residual for a control volume.Also used to denote an upper triangular matrix Re Reynolds Numbere Rtmodified turbulent Reynolds NumberRHS Right hand sideR6Riemann invariantse S production term for Spalart-Allmaras model~r vector from vertex to an edge of the dual control volumer11;r12;r22components of upper triangular matrix Rref denotes reference conditionS entropyT temperaturet timeU velocity normal to boundary of control volumeu;v Cartesian velocities in x and y directionsW x i;W y i least square weights for computing gradientsx;y Cartesian coordinatesx i;y i coordinates of mesh verticesx0;y0coordinates of a central vertexangle of attackratio of specific heats,taken as1.4laminar viscosityt turbulent viscosity=t t=e dependent variable for Spalart-Allmaras turbulence modeldensityconstant for Spallart-Allmaras turbulence modelconstant for Baldwin-Barth turbulence modelxx; xy; yy shear stress terms8numericalfluxflux limiterboundary of cellerning EquationsThe governing equations are the time-dependent Reynolds-averaged Navier-Stokes equations.The equations are expressed as a system of conservation laws relating the rate of change of mass,momentum,and energy in a control volume of area A to thefluxes of these quantities through the volume.The equations(nondimensionalized by free stream density,e 1,speed of sound,e a1,temperature,e T1,viscosity,e 1,thermal conductivity,e k1,and a reference length,L)are given asA @Q@t+I@~F1^n dl0I@~Fv1^n dl=0(1)where b n is the outward-pointing unit normal to the control volume.The vector of dependent variables Q,and the flux vectors~F i and~F v are given asQ=264uvE375(2)~Fi=f b i+g b j=264uu2+puv(E+p)u375b i+264vvuv2+p(E+p)v375b j(3)~Fv=f v b i+gvb j=264xxxyu xx+v xy0q x375b i+264xyyyu yx+v yy0q y375b j(4)Here,~F i and~F v are the inviscid and viscousflux vectors respectively;the shear stress and heat conduction terms are given asxx=( + t)M1Re23[2u x0v y](5)yy=( + t)M1Re23[2v y0u x](6)xy=( + t)M1Re[u y+v x](7)q x=0M1Re( 01)P r+tP r t@a2@x(8)q y=0M1Re( 01)P r+tP r t@a2@y(9)The equations are closed with the equation of state for a perfect gasp=( 01)2E0u2+v21=23(10)and the laminar viscosity is determined through Sutherland’s law1=(1+C3)1(T=T1)3=2(11)where C3=198:6460:0is Sutherland’s constant divided by a free stream reference temperature which is assumed tobe460 Rankine.4.Solution AlgorithmTheflow solver is an implicit,upwind-differencing algorithm in which the inviscidfluxes are obtained on the faces of each control volume using theflux-difference-splitting technique of Roe[10].For the current scheme,a node-based algorithm is used in which the variables are stored at the vertices of the mesh and the equations are solved on non-overlapping control volumes surrounding each node.The viscous terms are evaluated with afinite-volume formulation which is equivalent to a Galerkin-type approximation and results in a central-difference-type formulation for these terms.The solution at each time step is updated using the linearized backward-Euler,time-differencing scheme.At each time step,the linear system of equations is approximately solved with a subiterative procedure in which the unknowns are divided into two groups(colors)according to whether the node number is even or odd.For each subiteration,the solution is obtained by solving for all the unknowns in one color before proceeding to the next one.This corresponds to a red-black type of iterative algorithm for solving the linear system.4.1.Finite Volume SchemeThe solution is obtained by dividing the domain into a finite number of triangles from which control volumes are formed that surround each vertex in the mesh.Equation 1is then numerically integrated over the closed boundaries of the control volumes surrounding each node.These control volumes are determined by connecting the centroid of each triangle to the midpoint of the edges as shown in figure 1.These non-overlapping control volumes combine to completely cover the domain and are considered to form a mesh which is dual to the mesh composed of triangles formed from the vertices.6Figure 1.Control V olume Surrounding Node The numerical evaluation of the surface integrals in equation 1is conducted separately for the inviscid and viscous contributions.For a finite volume formulation,the inviscid contribution can be approximated using midpoint integration of the fluxes over each edge of the dual mesh that defines the boundary of the control volume.i.e.I@ ~F 1^n d dl d =I @ ~F 1d~n d N d X i=180q +;q 0;b n d i 12l d i (12)Here N d is the number of edges from the dual mesh that make up the surface of the control volume and l d i is the length of the edge.Also,8(q +;q 0;b n d i )is a numerical flux formed from data on the left (q +)and right sides (q 0)of the face which are determined by extrapolation from the surrounding data.Details of this procedure are presented in a later section.Note that the distinction between the left and right hand side of a face is determined by the direction of the normal which has been specified a priori and is considered to point from the left side of the face to the right.The flux calculations for a node are made by distributing the contributions from each of the edges.Since the associated normal vector is directed outward from the control volume on the left,the contribution of each edge to the integral in equation 12is added to the control volume to the left and subtracted from that on the right.A simpler and computationally more efficient process than the one described above can be achieved by replacing the two directed areas from the dual mesh that join at the midpoint of an edge in the triangular mesh with a single directed area.For example,in figure 2,the average directed area is defined as~n =~n L +~n R (13)This is identical to the directed area vector normal to the line formed by connecting the cell centroids of the adjoining cells.Observe that for this choice of directed area,if the numerical flux on the face is formed from the arithmetic average of the fluxes at the two nodes,80q +;q 0;b n 1=12 ~F (q L )+~F (q R ) 1b n (14)the resulting scheme is equivalent to a Galerkinfinite-element method[12].Note that in equation14,q+is simply taken to be the data at the left node(q L)and q0represents the data to the right(q R).The inviscid boundaryputing an Average Normalintegral contribution can now be written asI @ ~F1^n dl=I@~F1d~nNXi=18q+;q0;b n12l i(15)where N is the number of edges in the triangular mesh incident to the node under consideration and8(q+;q0;b n) represents the numericalflux on the newly defined face.For obtaining an upwind scheme,the numericalfluxes on the edges of the control volume are obtained using Roe’s approximate Riemann solver[10].Thesefluxes are formed from data on either side of the face as8=1 ~Fq+;b n1+~Fq0;b n11j A(b q;b n)jq+0q01(16)Here,~F(q+;b n)and~F(q0;b n)are the inviscidflux vectors given by equation3formed from the data on the left side (q+)and right side(q0)of the face,respectively.The matrix j A(b q;b n)j is formed from the variables on the cell face which are determined using an averaging procedure described in reference[10].Note that an equally valid alternative to this formulation is to form the numericalflux on the face as8=1 ~F(q L)+~F(q R)1b n01j A(b q;b n)jq+0q01(17)where~F(q L)and~F(q R)are theflux vectors evaluated from the values at the nodes on either side of the edge instead of from data extrapolated to the edge.As discussed above and in[12],with the matrix term neglected the formation of theflux in this manner yields a Galerkin discretization.Although results are only shown below in which inviscidfluxes are obtained using equation16,solutions obtained using equation17have also been obtained and no observable difference in pressures,skin frictions,or velocity vectors have been seen.Forfirst-order-accurate differencing,the data on the left and right sides of the cell face(q+and q0)are set equal to the data at the nodes lying on either side of the cell face.For higher-order differencing,the primitive variables are extrapolated to the boundaries of the control volumes using a Taylor series expansion about the central vertex so that the data on the face is given byq face=q node+5q1~r(18)where5q represents the gradient of the variables at the node and~r is the vector extending from the vertex to the midpoint of each edge.In the above equation,is a variable that ranges from zero to one and is used to control oscillations that may occur at steep gradients.For the current study,is determined using theflux-limiting procedure described in reference[15].Note that although the data on either side of the cell face(q+and q0)will be discontinuous,a single-valuedflux is obtained through equation16or17.For evaluating the gradient,5q,a least squares procedure is used in which the data surrounding each node is assumed to behave linearly.Referring tofigure3as an example,the data at each node surrounding the center node may be expressed asq i=q0+q x0(x i0x0)+q y(y i0y0)(19)By expressing the data in a like manner at each of the N surrounding nodes,an N22system of equationsis1Figure3.Nodes For Least Square Reconstruction of Data formed which can be solved to obtain the gradients at the nodes2 6641x11x2...1x N1y11y2...1y N3775q xq y=8>><>>:q10q0q20q0...q N0q09>>=>>;(20)This represents an over-determined system of linear equations,Ax=b which may be solved in a least squares sense using the normal equation approach in which both sides are multiplied by the transpose of the coefficient matrix,A,so that a222system of equations is obtained.A T Ax=A T b(21) Unfortunately,the sensitivity of the solution obtained using this technique is dependent on the square of the condition number of A[16].For problems on grids which are highly stretched,the accuracy of the process can be severely compromised.Therefore,in the current study,a Gram-Schmidt process is used in which the system of equations is solved by decomposing the A matrix into a product of an orthogonal matrix Q,and an upper triangular matrix R,i.e.A=QR(22) so that the solution is obtained byx=R01Q T b(23) A similar procedure has been used for computing inviscidflows on unstructured meshes in references[17]and[18]. Further details of least squares procedures can be found in reference[16].With this procedure,the numerical difficulties associated with solving linear systems with near rank deficiency is significantly reduced over the use of normal equations.Further improvement may be achieved through the use of Householder transformations[16]or singular value decompositions[16].The Gram-Schmidt process,however, allows for the easy precomputation and storage of weights so that the gradients at each node can be calculated by“looping”over the edges in the mesh and distributing the contribution of each edge to each of the nodes.The resulting formulas for calculating the gradients at the center node infigure3are given asq x0=NXi=1W x i(q i0q0)q y=NXi=1W y i(q i0q0)(24)where the summation is over all the edges that connect to the node(N=5infigure3)and the weights are given byW x i=x i0x0r211r12r11r222(y i0y0)0(x i0x0)r12r11(25) W y i=1r222(y i0y0)0(x i0x0)r12r11(26)wherer11="NXi=1(x i0x0)2#1=2(27)r12=N Pi=1(x i0x0)(y i0y0)r11(28)r22="NXi=1(y i0y0)0(x i0x0)r12112#1=2(29)Note that equation20yields an unweighted least squares procedure in which all the data surrounding the central node are given equal consideration.Numerical experiments have been conducted which indicate that for reconstructing nonlinear data on highly stretched meshes using equation18,the unweighted formulation is far superior to either inverse distance weighting or the use of gradients calculated with Green’s theorem.It was found, however,that for computing the actual values of gradients,inverse distance weighting and Greens theorem give very similar results,both of which are more accurate than unweighted least squares.Their failure to accurately reproduce surrounding data via equation18is attributable to theflawed assumption of linearly varying data.Therefore,for reconstructing data on boundaries of control volumes,the unweighted least squares procedure is used.When actual gradients are required,as in the production terms for the turbulence models,Green’s theorem is utilized.For computing the viscous contribution to the residual,R@ b Fv1b n dl,afinite volume approach is followed.Forexample,the surface integrals of the viscous contributions involve terms such as the two terms belowZ@ [( + t)u x]1b n x dlZ@[( + t)v y]1b n x dl(30)Recall that the boundary of the control volume is formed by connecting the centroid of each triangle to the midpoint of each edge as shown infigure1.By assuming a linear distribution in each triangle,gradients along these boundaries may be considered as constant.Therefore,quantities such as those in brackets in30above arefirst evaluated in each triangle of the mesh where + t are computed from an average of the surrounding nodes and the gradients are computed using Green’s theorem.Although this approach is from afinite-volume point of view,the computation of the viscous terms in this manner leads to identical formulas as a Galerkin approximation as given in reference[12].Note that for calculating viscousflows,the aspect ratio of the triangles in the boundary layer can be very large.Babuska[19]has shown in two dimensions that high aspect ratio triangles are not always detrimental as long as no angle is too close to180degrees.In the present study,all the grids have been generated using the method described in reference[20]which triangulates a set of points derived from structured grids.The resulting meshes tend to have very few cells with angles larger than120degrees and therefore the accuracy of the computations is not severely degraded.4.2.Time Advancement SchemeThe time advancement algorithm is based on the linearized backward-Euler time-differencing scheme which yields a system of linear equations for the solution at each step given by[A ]n f 1Q g n =f R g n (31)where [A ]n=A 1t I +@R n @Q (32)The solution of this system of equations is obtained by a relaxation scheme in which f 1Q g n is obtained through a sequence of iterates,f 1Q g i which converge to f 1Q g n .Note that several variations of classic relaxation procedures have been used in the past for solving the Euler equations on unstructured grids (see for example [21],[18],[1]and [22]).To clarify the scheme,[A ]n is first written as a linear combination of two matrices representing the diagonal and off-diagonal terms [A ]n =[D ]n +[O ]n (33)The simplest iterative scheme for obtaining a solution to the linear system of equations is a Jacobi type method in which all the off-diagonal terms (i.e.[O ]n f 1Q g ),are taken to the right-hand side of equation 31and are evaluated using the values of f 1Q g i from the previous subiteration level i.This scheme can be represented as [D ]n f 1Q g i+1=2f R g n 0[O ]n f 1Q g i3(34)The convergence rate of this process can be slow but it may be accelerated somewhat by using the latest values of f 1Q g as soon as they are available.This can be achieved by adopting a red-black type of strategy in which all the even-numbered nodes are updated first,followed by the solution of the odd-numbered ones.This procedure can be represented as [D ]f 1Q g i+1=h f R g n 0[O ]n f 1Q g i+1ii (35)where f Q gi +1i is the most recent value of Q and will be at subiteration level i+1for the even numbered nodes which have been previously updated and will be at level i for the odd numbered nodes.For Laplace’s equation,the use of a red-black strategy offers a factor of two improvement in the asymptotic convergence rate over point Jacobi [23].Numerical experiments for the Euler and Navier-Stokes equations indicate that a factor of approximately 1.6is attained in practice.While the use of the red-black algorithm offers improvement over the Jacobi iteration strategy,the convergence of the linear system can still be slow,particularly on grids with highly stretched cells necessary for turbulent Navier-Stokes calculations.Fortunately,it is not necessary to fully converge the linear system to provide a robust algorithm that remains stable at time steps much larger than an explicit scheme can provide.Although the use of high CFL numbers is desirable for convergence of the nonlinear system,the greatest benefits of large time steps are only obtained when the linear system is solved well at each time step.Unfortunately,large time steps are generally detrimental to the convergence of the ly,the convergence of the subiterative procedure is greatly enhanced by smaller time steps resulting from larger diagonal contributions.Therefore,a reasonable compromise must be made to allow CFL numbers small enough for good convergence of the linear system but large enough to also provide good convergence of the nonlinear system.In the current study,it has been found that although computations with CFL numbers of 500or more remain stable,the best convergence for the Navier-Stokes equations on large grids is achieved for moderate CFL numbers between 100and 200.To enhance the convergence to steady state,local time-stepping is used.The time-step calculation is based strictly on inviscid stability considerations for each node as given by1t =CFL A R@ (36)where U is the velocity normal the boundary of the control volume.4.3.Boundary ConditionsThe boundary conditions on the body correspond to no-slip and a prescribed wall temperature.They are enforced by modifying the matrix terms in equation35to appropriately reflect the desired boundary conditions. To more clearly demonstrate the procedure,a slightly expanded representation of one of the rows in equation35 is given by2 6 4D11D12D13D14D21D22D23D24D31D32D33D34D41D42D43D443758><>:11 u1 v1E9>=>;=8><>:RHS1RHS2RHS3RHS49>=>;(37)where RHS represents both the residual and off-diagonal terms on the right hand side of equation35and D ij represents the individual components of one of the diagonal blocks in[D].The density can be determined from the continuity equation during the solution process from thefirst row of equation37.However,the contribution to the continuity residual along the boundary involves an integration around the dual mesh surrounding the node and a segment of the body surface as depicted infigure4.The contribution from the surface,assuming zero velocity at the wall is identically zero.The second and third rows are modified so the solution of equation37maintains a zero velocity at the nodes on the solid boundaries.Further,the fourth row is altered to preserve a constant temperature which is set to the adiabatic wall temperature[24]T wall T1=1+pP r012M21(38)The constant wall temperature assumption is used to relate the change in energy at the wall to the change in density1E wall=T wall( 01)1 (39)The resulting matrix now reflects the enforcement of appropriate boundary conditions at the wall and is given by2 664D11D12D13D14 0100 00100T wall ( 01)00137758><>:11 u1 v1E9>=>;=8><>:RHS19>=>;(40)In the far-field,the data at the nodes are not explicitly set but are obtained through the solution process in much the same manner as the points interior to theflowfield.The only distinction between a far-field boundary node and an interior node comes from the fact that the enforcement of the boundary condition is reflected in the residual calculation.Referring tofigure4,theflux along the boundary between i and i+12is calculated from a weighted average of a characteristic reconstruction of the data at i and i+1.Note that the characteristic reconstructions at these nodes are only utilized to form theflux on the boundary;the actual values at the nodes are updated through equation35.i-1/2Figure4.Geometry for Calculating Flux on Solid and Far-Field Boundaries。
2021《文献检索》单选题3(附答案)
2021最新更新,手动整理,谢谢支持!下载之后搜题更方便(题库持续更新)国家开放大学《文献检索》单选题3(附答案)1.[单选题]下列哪项不属于EIsevier电子期刊数据库提供的Email提示服务?()(4分)A.全文提示B.检索提示C.期刊提示D.引文提示2.[单选题]我国电子图书产品最早出现于哪个年代?()(4分)A.20世纪90年代B.2004年C.20世纪70年代D.20世纪60年代E.20世纪80年代3.[单选题]通过Elsevier电子期刊数据库检索后,点击“Email Articles”按钮可以Email发送检索结果,对于此功能描述不正确的是()。
(4分)A.可以发送记录的摘要信息B.可以发送记录的引文信息C.可以发送记录的链接D.可以发送记录的全文(PDF/HTML格式)4.[单选题]对药物检索提供了专指的药物副主题词和投药途径的数据库是以下哪个?()(4分)A.SinoMedB.PubMedC.EMBASE.CINAHL5.[单选题]下列哪项不属于文献检索系统质量的评价要素?()(4分)A.索引体系的完善程度B.被用户使用的数量C.信息标引深度D.查全率、查准率6.[单选题]以下哪项是世界上资源最丰富的生物医学文献信息中心?()(4分)A.美国国家卫生研究院B.世界卫生组织C.美国疾病控制与预防中心D.美国国家医学图书馆7.[单选题]通过EIsevier电子期刊数据库检索韩启德发表的期刊论文,在检索框内输入作者姓名,下列选项中正确的是()。
C(4分)A.qide hanB.han qideC.qd hanD.han qd8.[单选题]关于特尔斐法描述哪项是正确的?()(4分)A.该方法由特尔斐创建的B.特尔斐是一位情报学家C.该方法也称专家法D.特尔斐是英国学者9.[单选题]数据库中检索韩冬季(Han Dongi)的文章,以下哪项检索式输入错误?()(4分)A.Han DJB.Han D JC.HanD.HarE.J10.[单选题]早期的0PAC系统是在何时由美国-些大学图书馆和公共图书馆共同开发的?() (4分)A.20世纪70年代末B.20世纪70年代中C.20世纪90年代末D.20世纪90年代中11.[单选题]中国科学引文数据库来源检索的检索结果中,相关文献包括哪些?()(4分)A.作者相关、关键词相关、参考文献相关B.题名相关、关键词相关、机构相关C.作者相关、主题词相关、文摘相关D.题名相关、主题词相关、来源出版物相关12.[单选题]以下哪项是目前常用的国内参考文献管理软件?()(4分)A.EndNoteB.BiblioscapeC.NED.Reference Manager13.[单选题]通过Web of Science数据库检索浙江大学附属医院科研人员发表的文献,最好选择下列哪一个检索式进行精确地址检索?() (4分)A.zhejiang Univ NEAR hospB.zhejiang Univ AND hospC.zhejiang Univ SAME hospD.zhejiang Univ WITH hosp14.[单选题]当需要评价某人的科研学术成就,评价某种期刊的质量,评价某一组织机构的科研水平时,应尽可能采用()进行检索。
introduction of xi'an
Xian, the eternal city, records the great changes of the Chinese nation just like a living book. Called Chang’an in ancient times, Xian is one of the birthplaces of the ancient civilization in the Yellow River area of the country. During Xian's 3,100 years development, 13 dynasties such as Western Zhou (11th century BC - 771 BC), Qin (221 BC - 206 BC), Western Han (206 BC - 24 AD) and Tang (618 - 907) placed their capitals here. Xian is the capi tal of Shaanxi province, located in the southern part of the Guanzhong Plain. With the Qinling Mountains to the south and the Weihe River to the north, it is in a favorable geographical location surrounded by water and hills. It has a clear distinction between the four seasons. Except the colder winter, any season is relatively suitable for traveling. The cultural and historical significance of the area, as well as the abundant relics and sites, help Xian enjoy the laudatory title of 'Natural History Museum'.
科学展览概况英语作文
科学展览概况英语作文An Overview of the Scientific Exhibition.In the ever-evolving landscape of knowledge and discovery, scientific exhibitions have become a vibrant platform for showcasing the latest advancements in various fields of science. These exhibitions not only provide a window into the wonders of science but also serve as a catalyst for inspiration and curiosity, encouraging individuals to delve deeper into the mysteries of our universe.The scientific exhibition I recently attended was a veritable feast for the senses, offering a comprehensive overview of the latest scientific developments. As I entered the vast exhibition halls, I was immediately struck by the diverse array of exhibits, ranging from cutting-edge technology to fascinating natural phenomena.At the forefront of the exhibition was a sectiondedicated to advancements in artificial intelligence (AI). Here, visitors were treated to a display of robots that could perform complex tasks with uncanny precision. One such robot, designed for medical applications, was able to perform surgical procedures with a degree of accuracy that surpassed even the most skilled human surgeons. Another remarkable exhibit was a virtual reality (VR) headset that allowed users to immerse themselves in a simulated environment, providing a unique perspective on scientific experiments and discoveries.Moving on, I encountered a section focused on space exploration.。
PowerUp
Supports multiple configuration options, such as output voltage, current limit, etc., which can be adjusted according to actual needs.
Key players in the market include major energy companies, technology providers, equipment manufacturers, and service providers
Competitive strategies adopted by these players include mergers and acquisitions, collaborations and partnerships, new product launches, and expansions into new markets
The market size is primarily driven by the increasing demand for energy efficient and sustainable power solutions, as well as the growing option of advanced technologies and innovations in the energy sector
future planning
01
Introduction
Purpose and background
To provide a comprehensive overview of the PowerUp project, its goals, and the current state of development
Pressure drop, gas hold-up and heat transfer during single
Pressure drop,gas hold-up and heat transfer during singleand two-phase flow through porous mediaM.Jamialahmadi a ,H.Mu ¨ller-Steinhagenb,c,*,M.R.IzadpanahdaUniversity of Petroleum Industry,Ahwaz,IranbInstitute for Thermodynamics and Thermal Engineering,University of Stuttgart,Pfaffenwaldring 38–40,Stuttgart 70569,GermanycInstitute of Technical Thermodynamics,German Aerospace Center (DLR),GermanydThe School of Engineering,University of Kerman,Kerman,IranReceived 23October 2003;accepted 22July 2004Available online 23September 2004AbstractPressure drop,bubble size,gas hold-up and convective heat transfer have been studied both experimentally and theoretically at constant wall heat flux for single and two-phase flow through unconsolidated porous media.Single-phase pressure drop and heat transfer coefficients have been measured over a wide range of particle size,heat flux and liquid flow rate.The conservation equations and the Kozeny-Carman equation are used to describe single-phase flow pressure drop and convective heat transfer through the porous media.The measured pressure drops have been used to evaluate the validity of the predictive expressions available in the literature.Mathematical models are developed for the prediction of temperature profiles and single-phase heat transfer coefficients,which predict the experimental data with good accuracy.A large number of new experimental data are presented on two-phase pres-sure drop,bubble size,gas hold-up and heat transfer coefficients for co-current upward gas/liquid flow through beds of different particle sizes under constant wall heat flux.The experimental data suggest the existence of two distinct regimes,i.e.homogeneous and heterogeneous flow.The experimental data on two-phase pressure drop and gas hold-up have also been compared with the pre-diction of published correlations.Finally,mathematical models are presented for the prediction of pressure drop,bubble size,gas hold-up and heat transfer which predict the experimental data with good accuracy.Ó2004Elsevier Inc.All rights reserved.1.IntroductionSingle and two-phase gas/liquid flow through porous media composed of stationary granular particles is fre-quently encountered in many diverse fields of science and engineering,ranging from agricultural,biomedical,mechanical,chemical and petroleum engineering to food and soil sciences.Classical research areas of chemical engineering dealing with porous media include filtration,drying,and multi-phase flow in packed columns and catalytic reactors.In all these instances it is necessary to predict design parameters such as friction factor,pressure drop,bubble size,gas hold-up,heat and mass transfer coefficients in order to determine the desired operating conditions and the size of the equipment re-quired for the specific purposes.Therefore,expressions are needed to predict these parameters accurately in por-ous media in which fluids are flowing either alone or as gas/liquid mixtures.In the last decade there has been a steady effort,both experimentally and theoretically,to improve the knowledge of single-phase flow and heat transfer in porous media.These studies have been re-viewed by several investigators (barnous and Bories,1975;Cheng,1978;Stankiewicz,1989;Tien and Vafai,1990).In most cases,Darcian flow was0142-727X/$-see front matter Ó2004Elsevier Inc.All rights reserved.doi:10.1016/j.ijheatfluidflow.2004.07.004*Corresponding author.Address:Institute for Thermodynamics &Thermal Engineering,University of Stuttgart,Pfaffenwaldring 38–40,Stuttgart 70569,Germany.Tel.:+497116862358;fax:+497116862712.E-mail address:hans.mueller-steinhagen@dlr.de (H.Mu ¨ller-Steinhagen)./locate/ijhffInternational Journal of Heat and Fluid Flow 26(2005)156–172assumed;however,some researchers have extended their work into non-Darcianflow as well(e.g.Vafai and Tien, 1981;Hunt and Tien,1988).Several studies concentrate on property variations of the bed,such as porosity and their effect on the heat transfer process(Vafai et al., 1985).A comprehensive review of these investigations can also be found in the books of Bear and Bachmat (1990),Kawiany(2002)and Nield and Bejan(1999). Several correlations have been recommended for the prediction of pressure drop and heat transfer in porous media saturated withfluids.Although these models can describe the hydrodynamics of single-phaseflow quite well,there are nevertheless some disagreements in the predictions.The studies on single-phase heat transfer have mostly dealt with a gaseous medium as the saturat-ingfluid and constant wall temperature boundary condi-tion.The available information on heat transfer under constant wall heatflux and in the non-Darcian regime is inadequate.Much less information is available on hydrodynamics and heat transfer of two-phase gas/liquidflow through porous media.Two-phaseflow through porousmedia M.Jamialahmadi et al./Int.J.Heat and Fluid Flow26(2005)156–172157is an important area,which covers a broad spectrum of engineering disciplines including geothermal systems (Sondergeld and Turcotte,1977;Cheng,1978),oil reser-voir engineering(Scheidegger,1974;Bear,1972),post-accident analysis of nuclear reactors(Lipinski,1981; Tung and Dhir,1988),multi-phase packed bed reactors (Shah,1979),condensation enhancement and thermal energy storage(Plumb et al.,1990;Vafai and So¨zen, 1990).Many attempts have been made to identify the flow regimes encountered when gas and liquidflow con-currently through porous media(Larkins and White, 1961;Eisenklam and Ford,1962;Turpin and Hunting-ton,1967).These investigators observed differentflow regimes,namely the homogeneous,transition and heter-ogeneous regimes,by establishing a constant liquidflow rate through the bed and then increasing the gasflow rate.They correlated the frictional pressure drop according to a Lockhart–Martinelli approach based on the knowledge of single-phase frictional losses for the gas and the liquid phaseflowing alone in the -parison of the correlation by Turpin and Huntington (1967)with Larkins and White(1961)data shows differ-ences of30–50%.Saada(1972)used data of Eisenklam and Ford(1962)to construct aflow diagram of two-phaseflow through beds offinely divided particles. Khan and Varma(1997)also presented their experimen-tal data in the form of aflow map and then compared their results with those reported by Saada(1972).They concluded that the map suggested by Saada is inade-quate for identifying the variousflow regimes in packed beds.Khan and Varma(1997)also presented different correlations for frictional pressure drop for eachflow re-gime and argued that separate correlations for the differ-ent regimes were advantageous when compared to the correlation of Turpin and Huntington(1967).Goto and Gaspillo(1992)studied gas/liquidflow through small packing material and presented correlations to predict frictional pressure drop in a manner similar to that of Lockhart and Martinelli.They used the Ergun (1952)correlation to estimate the single-phase pressure drop in packed beds.A general alternative description of two-phaseflow has been proposed by Hassanizadeh and Gray(1988,1993).They emphasized that the aver-aging of microscopic drag forces leads to a macroscopic non-linear theory forflow,but that the average of micro-scopic inertial terms is negligible in typical practical circumstances.Avraam and Payatakes(1995)studied two-phaseflow in porous media using a micro-model pore network of the chamber-and-throat type,etched into glass.During each experiment,the pore-scaleflow mechanisms were observed,the mean water saturation was determined with image analysis and the correspond-ing relative permeabilities and fractionalflow have been calculated.The subject of two-phase heat transfer in porous media has gained considerable attention during the last few years.The studies mostly concentrate on drying of different porous media,condensation in porous media, heat pipe application and geothermal applications. Dybbs and Schweitzer(1973)formulated the problem of non-isothermalflow in porous media for low Reyn-olds number where the inertia term is negligible.Dinul-escu and Eckert(1980)studied the problem of mixture migration due to a temperature gradient in porous media.They performed a one-dimensional analysis of this problem and produced an analytical solution.Lycz-kowski and Chao(1984)reported studies on two-phase drying.In their work,the effect of non-condensable components is not taken into consideration.Several authors have used a one-dimensional model to analyze condensation in porous media(i.e.Nilson and Romero, 1980;Ogniewicz and Tien,1981;Vafai and Sarkar, 1987).The problem studied consists of a porous slab subjected to different environments on two sides.Vafai and Whitaker(1986)also studied a different problem using a two-dimensional transient minarfilm condensation in a porous medium was analyzed by White and Tien(1987a,b).The non-slip boundary con-dition was employed for the velocity at the wall and an exponential function for the porosity.So¨zen and Vafai(1990)analyzed the transient forced convective condensingflow of a gas through a packed bed,with quadratic drag effects incorporated.The effects of ther-mal non-equilibrium in connection with the condensing flow of a vapor have been included in numerical simula-tions by So¨zen and Vafai(1993),Amiri and Vafai(1994) and Amiri et al.(1995).They found that the local ther-mal equilibrium condition was dependent on particle Reynolds number and independent of thermophysical properties.A literature review on heat transfer during two-phase flow in porous media and packed beds reveals that most research which had been carried out in this area was di-rected towards phase change in porous media,where the enteringflow is usually a liquid or a vapor.Published correlations are based on the energy equation which is solved either analytically or,in most cases,numerically under constant wall temperature.However,there is little information on heat transfer in porous media with two-phaseflow as the feed.Zhukova et al.(1990)undertook a thorough review of the literature on two-phase gas/liq-uidflow in stationary beds and concluded that research on heat transfer with ascending gas/liquidflow through porous media under constant wall heatflux was practi-cally non-existent.In order to obtain prediction models for pressure drop,bubble size,gas hold-up and heat transfer for co-current two-phase gas/liquid upward flow through packed beds offine particles under con-stant wall heatflux,there is need for better understand-ing of the role of the various operating parameters and of the mechanisms of heat transfer.The principal aim of the present investigation was to measure pressure158M.Jamialahmadi et al./Int.J.Heat and Fluid Flow26(2005)156–172drop,bubble size,gas hold-up and heat transfer coeffi-cients in packed beds over a wide range of superficial gas and liquid velocities,heat fluxes and solid particle physical properties.The predictions of various correla-tions from the literature are compared with these exper-imental data.Theoretical models are presented for the prediction of pressure drop,bubble size,gas hold-up and heat transfer coefficients to correlate the presented experimental data.2.Experimental equipment and procedure 2.1.Test rigFig.1illustrates the schematic diagram of the exper-imental set-up.The basic components of the test rig are the test section,the peristaltic pump,tank,flow meter and gasometer.The stainless steel test section is 32mm in diameter and 580mm long.The tube was packed with glass beads of different diameters hence forming porous media with different system pore size distribution.Six pressure transducers are installed at equal distances along the bed,which can measure pressures up to 4bar with very high accuracy.At the inlet and outlet of the test section two knit-meshes are kept in place by flanges,which contain the particles and prevent them from leaving the bed.All tubing and fittings are made of stainless steel.For those experiments,where flow pat-terns and gas hold-up have been investigated,the stain-less steel test section was replaced by a perspex test section with a diameter of 37mm and a length of 600mm.The liquid is pumped from the reservoir tank to the test section using a peristaltic pump.Adjusting the revolutions of the pump controls the flow rate of the liquid phase.The gas phase flows from a compressor via a filter and pressure regulator to the test section.The air flow rate is adjusted by a flow meter at the inlet of the bed and measured again precisely at the outlet using an online gasometer.For the two-phase flow measure-ments,the two fluids are mixed shortly before the test section.Heating is achieved by a Thermocoax resistance heat-ing wire which is placed in a spiral groove around the pipe and embedded by high temperature soldering tin to ensure good contact with the test section wall.The bed extends above and below the heated section to en-sure uniform distribution of the flow.Longitudinal grooves accommodate thermocouples measuring the wall temperatures.The local temperature of the wall is measured using thermocouples,which are located close below the heat transfer surface.The ratio between the distance of the thermocouples from the heat transfer surface and the thermal conductivity of the wall material (s /k )w was determined by calibration measurements using the Wilson plot technique.The heat transfer sur-face temperature can be calculated using this ratio,the heat flux and the thermocoupletemperature.Fig.1.Porous medium test apparatus.M.Jamialahmadi et al./Int.J.Heat and Fluid Flow 26(2005)156–172159T S ¼T TC À_q s =k ðÞwð1ÞBulk temperature is measured using six thermocouples which are inserted along the length of the bed.Another eight thermocouples are used to measure the wall tem-perature at two different positions along the bed.All sig-nals are fed into a data acquisition system which is connected to a desk top computer.The gas and liquid phases used in this investigation were air and distilled water,respectively.The physical properties of the fluids and the particles are given in Table 1.2.2.Experimental procedure and data reduction To start the experiments,clean and dried pre-weighted particles of a given size are poured into the test section and placed into a shaker for 1h.Then the bed is leveled,its height recorded and the test section installed back to its position in the test loop.Initially only liquid is pumped through the bed,and then the gas flow is gradually raised while the liquid flow rate is kept con-stant.The power supply to the test heater is then switched on and maintained at a predetermined value.Before any readings are taken,the system is left to itself for about 1h to establish homogeneous and steady-state conditions throughout.Finally,the data acquisition sys-tem is started to record pressures,heat flux,tempera-tures,gas and liquid flow rates.Gas hold-up is measured using the bed expansion method.This tech-nique relies on the instantaneous isolation of the bed from both liquid and gaseous feed.This is achieved by the use of quick action isolation valves on both gas and liquid inlets.It is assumed that the difference in the liquid level is produced by the gas hold-up in the sys-tem;according to the definition that gas hold-up is the fraction of the volume occupied by the gas bubbles.All experimental runs were performed with gradually increasing air flow rate while keeping the water flow rate constant.The local heat transfer coefficients are defined as:h ¼_q T S ÀT Cð2ÞThe power supplied to the test heater is calculated from the measured current and voltage drop.The average of five thermocouple readings is used to determine the dif-ference between wall and bulk temperature for each location.The experimental runs were performed in an arbi-trary sequence and some experiments were repeated to check the reproducibility of the experiments,which proved to be good.The range of experimental parame-ters covered in this investigation is summarized in Table 2.3.Experimental results 3.1.Single-phase flowIn single-phase flow,the pressure gradient across the bed is a function of system geometry,bed porosity,bed permeability and physical properties of the liquid phase.Considerable progress has been made in establishing the velocity–pressure drop relationship and several correla-tions have been developed to describe the hydrodynam-ics of these with acceptable accuracy.Some of these models are summarized in Table 3.The Kozeny-Carman model (1927)is still the most widely used correlation for the prediction of pressure drop and recommended in al-most all chemical engineering books for Darcian and non-Darcian flow regimes.Darcy Õs model is universally used for fluid flow through porous media in oil reservoir engineering and in almost all reservoir simulators.Pressure drop was measured at six positions along the porous medium in the direction of flow,for a range of liquid velocities.Typical measurements for single-phase,Table 1Physical properties of packing material and fluids Materials d s (mm)d K (m À2)q s (kg/m 3)l (kg/ms)k s (W/mK)Glass beads30.38 1.7·10À92500– 1.051.50.367 1.0·10À92500– 1.0510.3650.79·10À92500– 1.050.4–0.60.362 1.71·10À102500– 1.050.18–0.250.3618.56·10À112500– 1.05Mineral sand 0.25–0.4250.375 4.62·10À112300–5.34Distilled water –––9960.798·10À30.613Air–––1.17740.00001840.026Table 2Range of operating parameters Liquid velocity 0.004–0.7cm/s Gas velocity 0–30cm/sHeat flux1000–5000W/m 2Bulk temperature 20–90°C Inlet temperature 20–30°C System pressure1–2bar160M.Jamialahmadi et al./Int.J.Heat and Fluid Flow 26(2005)156–172liquid flow are depicted in Fig.2.A linear relationshipexists between pressure drop and distance in the direc-tion of flow in the bed.The slope of this proportionality increases sharply as the liquid velocity increases.Fig.3shows a typical comparison between measured and pre-dicted pressure drops for a bed packed with 0.5mm par-ticles as a function of modified liquid phase Reynolds number Re m,l .While all correlations predict an increasein pressure drop with increasing liquid velocity,the var-iation between the actual values from the different corre-lations is quite considerable.The best agreement between measured and calculated values is obtained with the correlations suggested by Kozeny-Carman (1927)and by Ergun (1952)for both,the Darcian and non-Darcian flow regime.As expected Darcy Õs model can predict the experimental data only at low liquidTable 3Correlations suggested for the prediction of single-and two-phase flow pressure drop Reference CorrelationDarcy (1856)D P L ¼l k Áu f Single-phase,Darcian flow regimeBlake (1922)D P L ¼k l a g ÁG 2g c q fÁa d 3Single-phase,Darcian and non-Darcian flow regimesKozeny-Carman (1927)D P L ¼150ð1Àd Þ2d 2s d 3l f Áu f þ1:75ð1Àd Þd 3d s Áq u 2f Single-phase Darcian and non-Darcian flow regimes Leva (1947)D P L ¼k ð1Àd Þg cd 3d s Gl 1:9Ál 2k 1:1q d 3sSingle-phase,Darcian and non-Darcian flow regimes Ergun (1952)D P ¼150ð1Àd Þ2d s d 3l f Áu f þ1:75ð1Àd Þd 3d sÁq u 2f Single-phase,Darcian and non-Darcian flow regimesand for Re m ;g1Àd ¼1À2000Larkins and White (1961)logD P t =L D P l =L þD P g =L¼0:416ðlog v Þ2þ0:666where v ¼D P lL=D P g L0:5Two-phase gas/liquid flow forhomogeneous and heterogeneous flow regimesFord (1960)D P t L ¼0:0407g q l l l l gÁRe 0:29l ÁRe 0:57g Two-phase air/water flow,for d s =1mm and d c =4.52cm Turpin and Huntington (1967)D P t L¼2q g u 2sgD e g cÁf t where ln f t ¼8À1:12ln Z À0:0769ðln Z Þ2þ0:0152ðln Z Þ3;Z ¼Re 1:167g 0:767l Two-phase flow and for L G ÀÁ0:24¼1–5and Re 1:167g0:767l ¼0:1–1000Saada (1972)D P t L ¼0:027g q l Re 0:35l Re 0:51g d cd s1:15Two-phase flow and for Re l =2.1–153.2and Re g =15–600Goto and Gaspillo (1992)D P t L ¼y ÁD P l L þD P g Lwhere ln y ¼0:55ln ðv =1:2Þ2þ0:666Two-phase flow,homogeneous and heterogeneous flow regimesKhan and Varma (1997)D P t L¼u 2sl q l 2d sÁf where ;For bubbly flow :f ¼3Â107Re 0:18g Re À1:7l d s dc 1:5For pulse flow :f ¼2:36Â107Re 0:26g Re À1:7l d s d c1:5For spray flow :f ¼3:91Â105Re 1:12g Re À1:82ld s d c1:5Two-phase flow,bubbly,pulse and spray flow regimesM.Jamialahmadi et al./Int.J.Heat and Fluid Flow 26(2005)156–172161Reynolds numbers.Despite the substantial similarity in the end results,the Kozeny-Carman correlation is a semi-empirical model based on a cylindrical pore model, whereas the Ergun equation is purely empirical.The Kozeny-Carman correlation predicts the present experi-mental data with an absolute mean average error of about3.5%.This semi-empirical correlation is expressed as follows:D P f L ¼1501ÀdðÞ2d3d2sl fÁu fþ1:751ÀdðÞd3d sq f u2fð3ÞEq.(3)is a form of the Forchheimer(1901)equation for homogeneous packed beds offine particles and unidirec-tionalflow(Nield and Bejan,1999).The two terms on the right hand side of Eq.(3)can be recognized as vis-cous and inertial contributions.At low Reynolds num-bers the second term is negligible and Eq.(3)reduces to DarcyÕs equation.At high Reynolds numbers,when thefluid inertia is important,the second term becomes dominant.3.2.Two-phaseflow3.2.1.Flow patternContrary to single-phaseflow,the problem of pres-sure drop during two-phase,gas/liquidflow in porous media is still largely unresolved.Fig.4is a schematic representation of visual and photographical observa-tions.At a constant liquid velocity,two different regimes may be distinguished,as the superficial gas velocity is in-creased from0to20cm/s.At low gas velocities,a homo-geneousflow regime prevails and the tiny bubbles pass through the bed without significant collisions or coales-cence.Since theyflow at almost the same velocity as the liquid phase,the additional turbulence due to the second phase is only small.In this regime,the bubble diameter is generally determined by pore size,surface tension and buoyancy(Lydersen,1989):d b¼md pÁrgðq lÀq gÞ!1=3ð4ÞFor bubble columns,d p is the orifice diameter of the gas distributor and m=1.For homogeneousflow in porous media,a value of m=0.09has been obtained from the analysis of the experimental data.Replacing pore diam-eter d p in terms of particle size d s(see Eq.(9))gives:d b¼0:09rÁd sÁdgÁqlÁð1ÀdÞ1=3ð5ÞFig. 4.Sketches offlow in homogeneous and heterogeneousflowregimes.162M.Jamialahmadi et al./Int.J.Heat and Fluid Flow26(2005)156–172In the heterogeneous flow regime,bubbles coalescence within a few centimeters after entering the porous med-ium at the bottom,to form large longitudinal bubbles.In the present investigation,it was found that L bd b%6–10ð6ÞEq.(6)is in agreement with the findings of Tung and Dhir (1988).The large bubbles occupy almost the whole cross-section of the pores and create significant turbu-lence and mixing effects,since they flow considerably faster than the liquid phase.3.2.2.Pressure dropOnly limited published information is available on pressure drop for two-phase flow in porous media,and no generally accepted prediction model has yet been produced.Fig.5shows the effect of the modified gas phase Reynolds number Re m,g on the measured two-phase pressure drop for a bed of 1mm particles,with the modified liquid phase Reynolds number Re m,l as parameter.At each liquid flow rate two regimes can be observed,according to the influence of Reynolds num-ber on the pressure drop.At low gas velocities (i.e.homogeneous flow regime)the pressure drop depends only moderately on the gas phase Reynolds number.In this regime the interaction between bubbles and liq-uid is negligible and hence the turbulence and mixing are low.At higher gas velocities (i.e.heterogeneous re-gime)the interaction between bubbles is high,leading to the formation of large bubbles and the associated tur-bulence/mixing effects.The measurements also show that the effect of liquid phase flow rate is more pro-nounced for the homogeneous regime than for the heter-ogeneous regime and that it loses its importance as thegas flow rate increases.The pore-scale flow mechanisms and their relationship to the mean liquid phase satura-tion can be found elsewhere (Avraam and Payatakes,1995).Some of the more widely used correlations available in the literature for prediction of pressure drop during two-phase flow in porous media are compiled in Table 3.Their predictions are compared with the experimental data for a bed packed with 0.5mm diameter particles in Fig.6.In most cases,there are considerable discrepan-cies between the predicted and measured values.As far as the supporting data for these models are concerned,they are confined to a very narrow range of system geometry as well as operating parameters (Izadpanah,1999),and this may be one of the main reasons why the predicted results are so scattered.The best agree-ment between measured and predicted values is obtained with the correlation suggested by Larkins and White (1961).This correlation can predict the present experi-mental data with an absolute mean average error of 28%.3.2.3.Gas hold-upGas hold-up plays two major roles in the evaluation of the transport phenomena in porous media:(i)it pro-vides the volume fraction of the phases present in the system and hence their residence time;(ii)the gas hold-up in conjunction with knowledge of the mean bubble diameter allows the determination of the gas/liq-uid interfacial area.Typical gas hold-up measurements for a bed of 1mm particles are shown in Fig.7as a func-tion of superficial gas velocity and liquid phase velocity.Again,homogeneous and heterogeneous flow may be distinguished.In the homogeneous regime,gashold-upFig.5.Experiment and predicated of two-phase pressure drop as a function of gas phase Reynolds number.M.Jamialahmadi et al./Int.J.Heat and Fluid Flow 26(2005)156–172163increases rapidly as gas velocity increases,whereas in the heterogeneous regime this effect is considerably less pro-nounced.It is important to note that gas hold-up is sig-nificantly decreased as the liquid velocity is increased.The liquid phase,which is the wetting phase,takes up the free space in the porous medium;hence less space would be available for the gas phase as the liquid flow rate increases.Furthermore,at higher liquid flow rates,the gas bubbles will wash out of the pores faster,which results in lower gas hold-up.Gas and liquid phase veloc-ities have also a strong effect on the flow regime.It is essential to know the range of parameters over which a particular flow regime prevails and the conditions under which the transition occurs.Transition from homogeneous to heterogeneous flow occurs when the density of the bubbles in the pore space increases to such an extent that the interaction and coalescence between bubbles becomes important.Several correlations have been suggested for the pre-diction of gas hold-up in porous media.These correla-tions are summarized in Table 4.A typical comparison between the measured and predicted gas hold-up values is shown in Fig.8.While the best trend is obtained from the correlation suggested by Larkins and White (1961)the deviation between the predictions of the various cor-relations and the measured data is quiteconsiderable.Fig.6.Prediction of two phase pressure drop from variouscorrelation.Fig.7.Experimental and predicated gas holdup as a function of superficial gas velocity.164M.Jamialahmadi et al./Int.J.Heat and Fluid Flow 26(2005)156–172。
- 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。
G.R. Prakash and K.K. Ravi Kumar
(Software Engineers, Verizon, Chennai, India)
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB Introduction: “Swaram” is a simple, general-purpose and procedural language designed for programming in Tamil. The authors of this paper are the designers and developers of a working version of the compiler and a virtual machine (interpreter) as a ‘proof of concept’. The sample implementation was done as an academic project, when the authors were pursuing their MCA degree in PSG College of Technology, Coimbatore, India. This paper provides a brief overview on motivation for the project, design considerations, technical overview of the implementation of Swaram. Few sample programs are also provided for illustrating the programming aspects of Swaram. Motivation: Knowledge of English language is a pre-requisite for computer programming (as of now, at least in India) and computing should not be the luxury of a privileged few, who know English. To enable the common man to work in his mother tongue, programming languages should be available in regional languages as well. This is a well-recognized need. Providing a preprocessor (the simplest alternative), a fullfledged native-language programming environment (for example, Visual Basic language and environ-ment in French), or even a full-fledged programming language (for instance, the Ruby language for Japanese) are few of the well-known means to achieve the same end. Two important questions naturally arise here. Which is the best alternative? If a preprocessor will do, isn’t it an overkill to provide a programming language? There is no single and simple answer for these questions. There are many perspectives from which we have to approach this question. Let us answer these questions indirectly by taking-up analogies for providing design and usage perspectives. Today, printing high-resolution documents is taken for granted in the printing industry. However it was not the case before the arrival of PostScript. Desktop publishing was a real nightmare with presence of innumerable document formats, font types, character encoding and image
____________________________________________________________________________________ 96
Tamil Internet 2003, Chennai, Tamilnadu, India
representation schemes. Today, PostScript is the industry standard for desktop publishing; it is flexible, extensible and can provide high-quality images – more than that, it is device independent! PostScript is not a file format – it is an object-oriented, interpreted language! Even in small hand-held printing devices, it is enough that the device has a PostScript interpreter - it can print the same documents used for professional publishing. That is the difference a language can make, even for printing documents. Yes, a language provides power and flexibility, but most importantly, it provides a means of abstraction – that is why it has a unique advantage over other schemes to get the work done. Let us see another analogy from the field of education. English teachers find one major problem in teaching the language to non-native speakers in countries like India. Students tend to think in their native language and do word-by-word translation into English! Such mental ‘transliteration’ doesn’t work for even gaining natural fluency if not gaining mastery over it. In just the opposite way, we write software that is essentially English, and put a layer over that to ‘transliterate’ it - it is just unnatural! When a programming language directly supports program-ming in a regional language, the barrier itself is removed; and at the same time it becomes a natural and effective tool for programmers to code in their native language. This should explain why native Japanese users find Ruby language to be very convenient, natural and useful though there is wide availability of programming languages based on English with a Japanese. From the brave vision that programming should be available universally, our language derives its name from the musical term ‘Swaram’ - since music is an universal language! Swa-ram, in fact derives its main idea and inspiration from the Ruby language. Ruby is an interpreted, scripting and object-oriented language with extensive support for Japanese script. The primary motivation for the designer of Ruby, Yukihiro Matsumoto, was to support Japanese for scripting, since he found that no good scripting language could help him for that purpose. Design Considerations The objective of the authors was to design and implement a programming language and related tools especially for Tamil. The design considerations for such a language is as follows. Such a programming language: 1) Should be for general purpose and easy to use and learn 2) Should be platform independent and portable 3) It should be capable of having specialized support and functionality for various specific aspects of the Tamil language 4) Should use a well-established program translation technology and programming paradigm 5) Should be capable of supporting the Tamil language fonts directly and without much effort from the programmer 6) Should be easy to extend support to other Indian languages 7) Should be similar to widely used programming languages like C/C++/Java so that the learning curve will be minimal for those who already know programming and make the best use of it. ‘Swaram’ is thus designed as a general-purpose, easy-to-learn, platform-independent and portable, both compiled and interpreted and extensible language.