Evaluation of gas-liquid separation performance of natural gas filters Evaluation of gas-liquid separation performance of natural gas filters

Evaluation of gas-liquid separation performance of natural gas filters

  • 期刊名字:石油科学(英文版)
  • 文件大小:764kb
  • 论文作者:Li Baisong,Ji Zhongli,Yang Xue
  • 作者单位:School of Mechanical and Electronic Engineering
  • 更新时间:2020-09-13
  • 下载次数:
论文简介

438Sc(2009)6:438-444DoI01007/sl21820090067-zEvaluation of gaS-liquid separation performanceof natural gas filtersLi Baisong, Ji Zhongli and Yang XueSchool of Mechanical and Electronic Engineering, China University of Petroleum, Beijing 102249, ChinaAbstract: Fibrous filters are often used to remove contaminants including both dusts and liquid dropletsfrom natural gas. This paper aims to evaluate the gas-liquid separation performance of three types ofcartridge filters used in the West-East natural gas transmission project. The comparison of the originalpressure drop of clean filters and the evolution of pressure drop as liquid droplets deposited in the filtermedia are described. The original pressure drops of these filters were similar but the pressure dropsat a steady state were different. Fractional efficiency was used to study the separation performarcartridge filters. Droplets at the outlet of the filters had small diameters, no more than 3 um, but werenumerous. The effect of filtration velocity on gas-liquid separation performance was analyzed. Higherfiltration velocity indicated better gas-liquid separation performance. Finally the quality factor related topressure drop and filtration efficiency was applied to evaluate the gas-liquid separation performanceKey words: Filtration, natural gas, gas-liquid separation, fractional efficiency, pressure drop1 IntroductionThey suggested that using higher surface energy fibers innist filters might allow lower levels of liquid retention thatNatural gas from different gas fields includes both resulted in wet filters with a lower pressure drop. hajra etdusts and liquid droplets. These small droplets and oil al( 2003)studied the effects of temperature, humidity, andmist existing in natural gas seriously affect the operating filter media with the addition of polymer nanofibers on gas/cycle of gas transmission pipelines and large centrifugal liquid filtration performance. Charvet et al (2008)lookedcompressors. However, it is very difficult to separate these into such operation parameters as filtration velocity and theliquid aerosols from natural gas. It was reported that some concentration of liquid droplets influencing pressure drop andfilters used in natural gas transmission pipelines were low in filtration efficiency of gas-liquid filtration with a horizontalgas-liquid separation performance and short in service life filter. Mullins et al(2004; 2005; 2006)carried out extensive(Li, 2004). Consequently, there is a need to evaluate the gas- experiments to investigate the wetting processes of a fiberliquid separation performance of filters used in natural gas liquid system microscopically during air filtration.Vasudevantransmission pipelines and to give some suggestions about and Chase(2004)used a novel combination of two types offilter materials and filter structureglass media to get better performance in terms of capturea large amount of research has been concerned with the efficiency and quality factorgas filtration of solid aerosols( Destephen and Choi, 1996; There are some discrepancies of evolution of filtrationWalsh, 1996; Thomas et al, 2001; Benesse et al, 2006; Huang efficiency aslets deposit on filters. Some studieet al, 2006; Song et al, 2006). In contrast, little work so far showed that the liquid collected by the filter generated ahas been dedicated to the filtration of liquid droplets from reduction in efficiency( Contal et al, 2004; Rayor and Leithgas. Contal et al(2004)gave a description of clogging of a 2000; Rising et al, 2005). Rayor and Leith(2000)thoughtfiber filter by submicron droplets and divided the filtration the presence of liquid on fiber surfaces could lower overallrocess into four stages. In the first stage the pressure drop efficiency because some fiber sections were unavailableand penetration increased slowly as a small surface mass for droplet collection. In contrast, Conder and Liew(1989)deposited In the second stage the penetration increased observed filtration efficiency increased with the loading ofexponentially until reaching a maximum. The third stage was liquid aerosolscharacterized by a sharp rise in efficiency coinciding with HeHowever, little literature on gas-liquid filtrationan exponential increase in pressure drop. Finally, a pseudo- perforcartridge filters is available(li et al, 2007)stationary state was established. Letts et al (2003)conducted The al中国煤化工 valuate the gas-liquidtests to observe glass, polyester, and polyaramid fibers separmicroscopically as these fibers collected liquid droplets. SeverCN MHGgas carriWest-East natural gastransmission project (a famous natural gas pipeline which*corRespondingauthoremailjizhongli63@vip.sina.comansports natural gas from West China to East China) haveReceived October 22. 2008been evaluated for pressure drop and filtration efficiencyPet. Sci(2009)6:4384442 ExperimentalThe mists were generated by an air atomizing nozzleSUll(Spraying Systems Co, Shanghai, China). The2.1 Experimental apparatusgenerated liquid was di-2-ethylhexyl sebacate(DEHS)with adensity of 913 kg/mand very low saturated vapor pressure.The experimental set-up illustrated in Fig. I was The liquid and compressed air were mixed internallycomposed of a droplet generator unit, a series of filter holders, produce an atomized sprayand a sampling unit. Several cartridge filters mostly used in A Welas 2000 light-scattering spectrometer(Palasnatural gas transmission pipelines were installed horizontally Karlsruhe Germany) which determines particle concentrationthe holders. Through acrylic glass holders, the filtration and particle size was used to measure the liquid aerosols atprocess was observed conveniently. a liquid collector was the outlet of the test cartridge filter. It has a measurementput at the inlet of the test cartridge filters to collect the liquid range of 0. 3-40 um and a small measurement volume fordeposited on the filter holder; and another liquid collector concentrations up to 10 particles/cmwas put at the outlet to collect the drainage from the cartridgefilters.The pressure drop across the filter was monitored 2.2 Characteristics of the cartridge filterscontinuously with a U-tube pressure gaugeThree types of cartridge filters used in the West-natural gas transmission project were tested (Table 1)2. ThePitot tubepacking density of the filter media was calculated as followswhere mr is the mass of the fibers in the filter, kg: V is thevolume of the filter, m; and Pr is the density of the fibersLiquid collectorsFig. 1 ExTable 1 Characteristics of the cartridge filters testedOutside diameter Filter thicknessMaterialCellulose0.2781 PleatedCellulose and acrylicester-styrene copolymer92.l0.44980 PleatedCGlass113980 Unpleated2.3 Experimental methodsFig 2 presents the comparison of the outlet massconcentration of a test filter measured with two methods theTo shorten the clogging process of filtration, a high mass test was carried out with cartridge filter A at a filtration velocityconcentration of liquid droplets was used. This could not be of 0.04 m/s and an inlet mass concentration of 15 g/m. Themeasured by the Welas 2000 light-scatlering spectrometer as it test was conducted three times under the same conditionswas out of the range of measurement. The mass concentration The membrane filter was weighed after 80 minutes. Thewas measured by comparison of the weight changes of liquid agreement of experimental results from two methods showsin the liquid tank supplying the spraying nozzle. The mists the level of repeatability and reliability of the experimentalmeasured by a Malvern 2600 laser diffraction particle sizer resultsMalvern Instruments Ltd, worcestershire UK)had a Sautermean diameter(SMD)of approximately 20 um and an index 3 Results and discussionranging from 2. 1 to 2.6Both the light-scattering spectrometer( Welas 2000)andthe membrane filter sampling system were used to measure3.1 Evolution of pressure drothe mass concentration of outlet liquid droplets. A glass-fiber Fig中国煤化工 een pressure drop andmembrane filter was used in the membrane filter sampling filtratio-ers. Pressure drop wassystem, which has a very high separation efficiency(more a nearCNMH Glocity. The three typesthan 99.97% at 0.3 um). Comparing the results from these of clean filters had similar pressure drops at a given filtrationtwo measuring systems, the repeatability and reliability of velocity of 0.02-0.10 m/experimental results could be clarified.Fig. 4 shows the evolution of pressure drop when liquidPet. Sci.(20096438-444838Time. mTime. minTime. minFig. 2 Comparison of two measurement methodso Welas measurement; Weighing measurementdroplets deposited in the filter. The experiments were The change of pressure drop in filter A and filter B couldperformed at a filtration velocity of 0.06 m/s and an inlet be divided into three stages. In the beginning, the pressuremass concentration of 15 g/m. The pressure drop increased drop increased slowly as a small number of liquid dropletsas liquid droplets deposited in the filter. Finally, the pressure deposited. Then, the pressure drop increased sharply after thedrops leveled off and the filtration reached a steady state. number of droplets deposited in the filter reached a criticalvalue. Finally, the pressure drop reached a plateau. In contrast,the filter C having a low packing density did not have ald of sharp pressure drop. Filter A andC Filter Astructures. However, filter B was an mixture of acrylicestera2000styrene copolymer and cellulose, and filter A just consistedF山cof cellulose. The added acrylicester-styrene copolymercould lower the surface tension of the filter medium, whichresulted in less liquid retained in the filter media and then adecrease in the pressure drop. Fig. 5 shows the SEM(scanningelectronic microscopy)images of the used and unused filtersThe nature of filter had a great effect on the liquid retained inthe filter. The more the liquid retained in the filter the higher002Filtration velocity, m/s3.2 Droplet mass concentration at the outletFig. 6 shows the droplet mass concentrations at the outletFig. 3 Characteristics of original pressure drop of clean filtersof the filters when the pressure drop reached a steady stateThe experiments were carried out at a filtration velocity of0.06 m/s and an inlet droplet mass concentration of 15 g/m7500The mass concentrations at the filter outlet were measuredaFilter Awith the Welas 2000 spectrometer every 5 minutes. Thedroplet mass concentration at the outlet did not changewhen the pressure drop remained constant. The droplet mass4500concentration at the outlet was determined by droplet size anddroplet number concentration. Fig. 7 presents the droplet sizedistributions at the outlets of the three filters, Small dropletswere densely distributed at the outlet of the filters in the150aprocess of gas-liquid filtration. The mean diameter of outletdrople中国煤化工 number concentrations12×101.8x1024x104Generated liquid per unit area, g/m2CNMHGvely. The droplet sizedistributions at the outlet of filters b and c were similarFIg. 4 Evolution of pressure drop at a filtration velocity of 0.06 m/sbut the droplet number concentration at the outlet of filter CPet. Sci(2009)6:438-444was greater than that at the outlet of filter B. As a result, the equivalent fiber diameter of the filter may be different from itsdroplet mass concentration at the outlet of filter C was larger initial state. The filter of higher packing density and smallerthan that at the outlet of filter B, as shown in Fig. 6. When equivalent fiber diameter at the steady state had a better gas-the droplets deposited on the fibers, the packing density and liquid separation performance.(a)Unused filter A(b)Used filter A(c)Unused filter B(d) Used filter B500um中国煤化工(e) Unused filler CCNMHGFig. 5 SEM images of the used and unused filtersPet. Sci.(2009)6:438-444▲▲山^▲▲▲么Filter As80乏···●●·●●●Time, minDroplet diameter, um(a)Filter AFig. 6 Droplet mass concentrations at the outlets ofthe filters at a steady sta3.3 Fractional efficiency for different-sized dropletsFor a specific droplet diameter, fractional efficiency canbe calculated as follows7p=(1-Naw/N-d)×100%where Nourlet and Ninet are the number of droplets at the outletand the inlet, respectivelyThe outlet droplets were measured by the light-scatteringspectrometer(Welas 2000). And the inlet droplets weremeasured by the laser diffraction particle sizer(Malvern2600), which has a different resolution from the light-Droplet diameter,scattering spectrometer. In order to calculate fractional(b)Filter Befficiency, the results measured by the laser diffractionparticle sizer should be changed. The droplet size distributionat the inlet of the filter was obtained according to the Rosin-Rammler distribution function(Zheng et al, 2006). Fig. 8shows the comparison of fractional efficiency of differentcartridge filters for different-sized droplets. The experimentswere carried out at a filtration velocity of 0.06 m/s and adroplet mass concentration of 15 g/m'at the inletThe filters were very effective for liquid droplethan 2 um. Filter B had higher fractional efficiency than the巴E5>other two filters for droplets smaller than 2 um. Therefore,filter b should have lower mass concentration at the outletthan the others. as to the comparison of filter A and filter C, itseemed a bit more complex. Filter a was more effective than▲A▲▲▲▲▲filter C for droplets larger than 0.8 um but less effective fordroplets smaller than 0.8 um(c)Filter C3. 4 Influence of filtration velocityFig. 9 represents the changes in the pressure drop andFig. 7 Outlet droplet size distributions at a filtration velocity of 0.06 m/sthe droplet mass concentration at the outlet at differenthe same inlet d8 nas concentration of 15 g/mi中国煤化工Higher filtration velocity led to higher pressure drop and PrCN MH Gnc are two importantlower droplet mass concentration at the outlet of the filter. factors to evaluate the pertormance of filter media.BetterSimilar results have also been obtained by other researchers filter performance means higher filtration efficiency and( Contal et al, 2004; Rising et al, 2005; Charvet et al, 2008)lower pressure drop. Generally, pressure drop and filtrationPet.Sci.(20096:43844where Coutke and Cinket are mass concentrations of droplets atthe outlet and inlet, respectively; and Ap is the pressure drophrough the filter.Table 2 presents the comprehensive evaluation ofthree filters under the conditions of the inlet droplet massconcentration of 15 g/mand the filtration velocities of0.06 and 0.04 m/s. It can be seen that filter C had thehighest quality factor, which meant it had the best filtrationperformance.Table 2 Comprehensive evaluation of three cartridge filtersDroplet diameter, umFiltration velocityig. 8 Fractional efficiency for different-sized droplets1000408840050000631101.226Cl1511902618248C4.8700.034 ConclusionsFiltration velocity, m/s1)A Welas 2000 light-scattering spectrometer and a(a)Filter Ahigh precision membrane filter sampling system wereused to measure the mass concentration of liquid dropletsThe agreement of the results from these two measurementmethods showed the repeatability and reliability of40 experimental results30 cartridge filters used in the West-East natural gas transmissionproject was investigated. These filters were very effective forliquid droplets larger than 2 um.3)The characteristics of the filter influenced the pressuredrop and the filtration efficiency considerably. Structureoptimization and surface treatment of the filter can improvegas-liquid separation performance4)The filtration velocity influenced significantly the gas-Filtration velocity, m/sliquid separation performance of the filter in the range from(b)Filter C0.03 to 0.06 m/s. Higher filtration velocity led to higherFig. 9 Influence of filtration velocity on pressure drop and dropletpressure drop and lower droplet mass concentration at themass concentration at the outlet of the filteroutlet of the filter.Referencesfficiency have the similar trend. Higher filtration efficiencysometimes leads to higher pressure drop The quality factor Benesse M, Lecoq L and Solliec C Collection efficiency of a wovenfilter made of multifiber yam: experimental characterization duringbeing independent of filter thickness, provides a convenientan filter modeling hased on a two-tier single fiberquality for comparing filter performance( Hajra et al, 2003)中国煤化工6.37:94989The higher quality factor indicates better performance of the Charvetfilter media. The quality factor is given by:CNMH Gring Research and Design2008.86:5695763) Conder JR and Liew T P. Fine mist filtration by wet filters-ll: efficiencyof fibrous filters. Journal of Aerosol Science. 1989. 20: 45-57Pet. Sci.(20096:438444Contal P, Simao J, Thomas D, et al. Clogging of fiber filters by modelling of clamshell droplets on vertical fibers subjected tosubmicron droplets: phenomena and influence of operatingcavitational and drag forces. Journal of Colloid and Interfaceconditions. Journal of Aerosol Science. 2004. 35: 263-278Science.2005.284:245-254Destephen J A and Choi K J. Modeling of filtration processes of fibrous Mullins B J, Braddock R D, Agranovski I E, et al. Observationfilter media. Separation Technology. 1996. 6: 55-67Frising T, Thomas D, Bemer D, et al. Clogging of fibrous filters by liquidgravitational and drag for pets on vertical fibers subjected toaerosol particles: experimental and phenomenological modelingScience.2006.300:70472study. Chemical Engineering Science. 2005. 60: 2751-2762Rayor P C and Leith D. The influence of accumulated liquid on fibrousHajra M G, Mehta K and Chase GG. Effects of humidity, temperature, filter performance. Joumal of Aerosol Science. 2000. 31: 19-34and nanofibers on drop coalescence in glass fiber media Separation Song C B, Park H S and Lee K w. Experimental study of filter cloggingand Purification Technology. 2003. 30: 79-88with monodisperse PSL particles. Powder Technology. 2006. 163Huang B, Yao Q, LiSQ, et al. Experimental investigation on the particle 152-159capture by a single fiber using microscopic image technique. Powder Thomas D, Penicot P, Contal P, et al. Clogging of fibrous filters byTechnology2006.163:125-133solid aerosol particles: experimental and modeling study. ChemicalLettsG M, Rayor PC and Schumann R L Selecting fiber materials to Engineering Science. 2001. 56: 3549-3561improve mist filters. Joumal of Aerosol Science. 2003. 34: 1481- Vasudevan G and Chase GG. Performance of B-E-glass fiber media in1492coalescence filtration Journal of Aerosol Science. 2004. 35: 83-91LiB S, Ji Z L and Chen H Y. Study of the gas-liquid separation Walsh D C. Recent advances in the understanding of fibrous filterperformance of natural gas filters. Natural Gas Industry. 2007. 10behaviour under solid particle load. Filtration Separation. 1996123-125(in Chinese)33:501-506Li K. Actuality and improvement of large natural gas filter. Oil Field Zheng G B, Kang T H, Chai Z Y, et al. Applied the Rosin-RammlerEquipment. 2004. 33: 124-126(in Chinese)distribution function to study on the law of coal dust particle-sizeMullins B J, Agranovski I E, Braddock R D, et al. Effect of fiber distribution. Journal of Taiyuan University of Technology. 2006. 3:orientation on fiber wetting processes. Journal of Colloid and64-66 (in Chinese)nterface Science. 2004. 269: 449-458Mullins B J, Braddock R D, Agranovski I E, et al. Observation and(Edited by Sun Yanhua)中国煤化工CNMHG

论文截图
版权:如无特殊注明,文章转载自网络,侵权请联系cnmhg168#163.com删除!文件均为网友上传,仅供研究和学习使用,务必24小时内删除。