Documents

10.1016@j.fuproc.2015.09.029.pdf

Description
Fuel Processing Technology 142 (2016) 86–91 Contents lists available at ScienceDirect Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc Research article Effect of dissolved oxygen concentration on coke depositio
Categories
Published
of 6
20
Categories
Published
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
Share
Transcript
  Research article Effect of dissolved oxygen concentration on coke deposition of kerosene Xin-yan Pei, Ling-yun Hou ⁎ School of Aerospace Engineering, Tsinghua University, Beijing 100084, China a b s t r a c ta r t i c l e i n f o  Article history: Received 3 April 2015Received in revised form 28 September 2015Accepted 28 September 2015Available online 8 October 2015 Keywords: Dissolved oxygenCoke depositionSupercritical conditionEntrance effect Coke deposition is an obstacle to the application of fuel cooling. The effect of dissolved oxygen concentration atthesupercriticalpressureonthethermaloxidativecokingdepositionofChineseRP-3kerosenewasinvestigatedusingthermal 󿬂 uidexperiments.Testswereconductedinasmall-diameter,indirectlyelectricallyheated,single-pass straight tube covering the temperature range of the thermal oxidation reaction. The amount of coke depo-sition was measured by weighting.The results of theexperiments for high-dissolved-oxygen, air-saturated, anddeoxygenatedfuelswerecomparedatthesupercriticalpressureof3MPaandafueloutlettemperatureof410°C.In comparison with the air-saturated fuel, the high-dissolved-oxygen fuel experienced a sharp rise in deposits,while the deoxygenated fuel had the lowest deposition quantity. The consumption fraction of the dissolved ox-ygenincreasednotablyabove 200 °C and 󿬁 tted a pseudo- 󿬁 rst-orderreaction. The entrance effect, inducing highdepositions, was interpreted by the temperature gradient and the boundary layer. Finally the relationship be-tween thermal oxidative deposition and heat transfer was examined.© 2015 Elsevier B.V. All rights reserved. 1. Introduction Recent advances in jet aircraft technology have induced an ever-in-creasing heat load on aircraft air cooling systems. Cooled cooling airwithhydrocarbonaviationfuelsisapotentialtechniqueforthethermalmanagement systems of next-generation aircraft [1]. As the jet fuel ex-posed to air, oxygen in the air is dissolved in the fuel. Accordingly, au-toxidation reactions occur in the fuel temperature range of 150 – 400 °C[2].Thejet fuelreactswith dissolved oxygen toproduce oxidizedprod-ucts,anddepositsformontheinner-wallsurface.Thesedepositswouldblock fuel lines, valves, nozzles, and various other aircraft components,potentially resulting in engine function failure [3].Variousexperimentshavebeenperformedonthethermaloxidationstability.Manyphysicalandchemicalfactors[4,5]in 󿬂 uencethethermaloxidationstability,listedastemperature[6],systempressure[7,8], 󿬂 owmassratesorvelocity,testduration[9,10],fuelcompositions[2,11],ad- ditives [12,13], catalysis and surface effect [14,15], especially, the dis- solved oxygen concentration [16,17] which is one of the mostimportant factors for oxidation deposition. It is well cognized that thereactionofdissolvedoxygenwithhydrocarbonsformshydroperoxides,which are intimately involved in deposit formation [18]. The effect of dissolvedoxygenhasbeenstudiedusingmanydifferentdevices,classi- 󿬁 ed as static [19] and dynamic ones [20]. In static experiments, the jet fuel thermal oxidative stability in quartz – crystal microbalances hasbeen examined using crystal electrodes with various metal materials,namely gold, platinum, aluminum, and silver [19,21,22]. In dynamicexperiments, a near-isothermal  󿬂 ow test rig was applied to investigatethe role of dissolved oxygen by reducing the oxygen concentration invarious fuels subjected to a thermal environment, with the carbonburn-off analysis of surface deposits [2,16,23]. It has been indicatedthat control of dissolved oxygen at sub parts-per-million (ppm) levelsgreatly improved the thermal stability and decreased heat exchangerfouling. In addition to the oxygen concentration,  󿬂 uid dynamics andheattransferalsoaffectedfueloxidationdeposition[24,25].Theremov-al of oxygen has been proposed as a promising method for improvingthe thermal stability of fuel [16,17]. Most studies have focused on themechanisms and reactions of thermal oxidation and decompositioncokingforvariousfuels[26 – 28],andfewstudieshaveconsideredtheef-fect of oxygen consumption on deposition and heat transfer.The above researches prove that it is important to understand themainfactors affectingthedeposit formationfor a givenjet fuel. Theob- jectiveofthisworkistoexaminetheeffectofdissolvedoxygenconcen-tration on coke deposition of jet fuel at supercritical pressure andtemperatures.Intheexperiments,theoxygeninthefuelwaspartlycon-sumed,andwasmonitoredattheinletandexitofthereactorduringthetests. The in 󿬂 uence on deposition of the entrance effect, which is char-acterized by the boundary layer and the temperature gradient, wasinterpreted. The relationship between heat transfer and the oxidationstability of RP-3 was investigated based on the Nusselt number (Nu)and deposition rate. 2. Experimental Chinese No.3 (RP-3) kerosene was chosen as the fuel. The physicalproperties of RP-3 aviation kerosene are listed in Table 1 [29]. The Fuel Processing Technology 142 (2016) 86 – 91 ⁎  Corresponding author. E-mail address:  lyhou@tsinghua.edu.cn (L. Hou).http://dx.doi.org/10.1016/j.fuproc.2015.09.0290378-3820/© 2015 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Fuel Processing Technology  journal homepage: www.elsevier.com/locate/fuproc  thermal physical properties of the aviation kerosene are temperature-dependent. The bulk temperature of fuel along the reactor tube couldbe predicted in the simulation results. The thermal physical propertieswere determined by using a three-component surrogate model [30].An indirectly electrically heated tube was employed to simulate aheat-exchanger in which heat transfer and thermal oxidation deposi-tion of fuel occur. As shown in Fig. 1, the experimental apparatusconsisted of four subsystems, i.e., feeding, heating and reactor, data ac-quisition/controlling system, and sampling/online analysis systems. Avacuum pump (6, N86 KT.18, KNF Inc.) and a magnetic stirrer (3)wereappliedtodeoxidizethefuel.Theoxygenconcentrationwasmon-itored via two observation windows  󿬁 xed on the opposite walls of thefuel tank (4) and an oxygen sensor (5, InPro 6850i, Mettler ToledoInc.)dippedinthefuel.Theoxygensensorwasdesignedforcontinuousmeasurement of dissolved oxygen over a 0 – 100% full range, i.e., up to70 ppm. The transmitter, which was coupled to the sensor, was aMettler Toledo dissolved oxygen microprocessor transmitter (Model3000). A mass  󿬂 ow controller (2; CS2000, Seven-star Inc.) was used tocontrol the amount of oxygen bubbled into the deoxidized fuel to thedesired dissolved oxygen concentration. The fuel with a speci 󿬁 c dis-solved oxygen concentration was introduced into the reactor througha  󿬁 lter (7) to remove any impurities. The inlet mass  󿬂 ow rate was setat 1 g/s, and was monitored using a plunger metering pump (8) andcoriolis mass  󿬂 ow meter (11; DMF-1-1-ADMF-DX, Sincerity Inc.). Thefuel pressure stability was maintained using a pressure damper. Nitro-genwasusedtoblowdowntheresidualfuelinallpipesbeforeheating.The reactor straight tube was made of the 316 stainless steel with alength of 450 mm, an inner diameter of 2 mm and a wall thickness of 0.5 mm. The test section was  󿬁 xed horizontally and the heating  󿬂 uxwas simulated by an electrical heating stove. After heating, the fuelwas cooled using a counter- 󿬂 ow water-fuel exchanger (18). The fuelpressure was adjusted to 3 MPa using a back-pressure regulator (19;SS-9833F-1P-B, Xiongchuan Value Inc.) installed after the heat-ex-changer to guarantee the jet fuel in a supercritical phase. The oxygenconcentrations after the thermal experiments were measured continu-ously, but the oxygen sensor could not be placed immediately down-stream of the heated test section exit because of the high exittemperature and pressure. A sampling and online analysis system (21)was therefore installed behind the cooling exchanger for consistentmeasurements. To ensure a constant oxygen concentration along thewholetube,thesealdesignwastestedatpressureof3MPabycheckingtheinletandexitoxygenconcentrationstoverifytheabsenceofoxygenleakagefrom,oradditionto,thesystem.Theresidualswerecollectedina container (20). A pressure difference transducer (22; 300S1AAD1M5,Rosemount Inc.) was used to monitor the pressure drop between theinlet and outlet of the reactor during heating. The outer-wall tempera-ture of the reactor tube was measured using 18 K-type thermocouplesspot-weldedat2.5cmintervalsalongthetube.ThefuelinletandoutlettemperaturesweremeasuredusingK-typethermocouplesinsertedintothe center of the tube. The amount of oxidized deposition was stronglyaffected by the test time, therefore heating was performed for 105 minto guarantee the formation of suf  󿬁 cient deposit for analysis.As the amount of deposit formed by oxidation is smaller than thatformed by thermal cracking, a segmentation method was used toweigh the mass of oxidized deposition [31]. The formation of oxidationdeposits depends on a two-step mechanism:  󿬁 rstly, the chemical for-mationofdepositprecursorsbyfuelreactingwiththedissolvedoxygen;and secondly, the mass transport of deposit precursors to the wall. Asthe fuel heated in the reactor, the coke was gradually deposited on theinner surface of the tube with increase in the temperature and theheating duration. After each experiment, the tube was divided into 11sections with each length of 4 cm. Each segment surface was polishedoutandcleanedwithethanol.Thisoperationwastoeliminateerrorsin-troduced by the insoluble impurities on the surface. And then the seg-ments were dried above 100 °C for at least 1 h to remove residualwater and ethanol. A high-precision balance (ESJ182 – 4, Long-TengInc.)wasusedtoweigheachsegment.Thesegmentswerethenwashedfor at least 2 h in an ultrasonic vibrator  󿬁 lled with potassium  Table 1 Characteristics of RP-3 aviation kerosene.Critical pressure (MPa) Critical temperature (°C) Density (g/cm 3 ) (20 °C) Flash point (°C) Distillation range (°C) Relative molecular weight(g/mol)Averaged molecularformula2.39 372.5 0.7913 50 163 – 212 148.33 C 10.5 H 22 Fig.1. Experimentalsetupforthecokedeposition.1:Oxygencylinder;2:Gasmass 󿬂 owcontroller;3:Magneticstirrer;4:Fueltank;5:Onlineoxygensensor;6:Vacuumpump;7:Filter;8:Plunger metering pump; 9: Pressure damper; 10: Nitrogen cylinder; 11: Mass  󿬂 ow meter; 12: Relief valve; 13: Preheater; 14: Pressure gauge; 15: Thermocouple; 16: Electrical heatingstove; 17: Reactortube; 18: Water-cooled exchanger;19: Backpressureregulator; 20:Collection container;21:Sampling and onlineanalysissystem;22:Pressuredifferencetransducer;23: Data acquisition/controlling system.87  X. Pei, L. Hou / Fuel Processing Technology 142 (2016) 86  – 91  permanganate (125 mg/L) to remove the deposits. The segments werewashed with water and ethanol, dried, and weighed again. The differ-ence between the two weights was the mass of coke deposition.The experimental heat transfer coef  󿬁 cient can be calculated as fol-lows:  h  =  q /( T  w  –  T  b ), where  q  is the heat  󿬂 ux and it depends on theheating power and excludes the heat loss. The system heat loss to theenvironmentcanbeminimizedthroughinsulationandcanbeobtainedthrough calibration without a fuel  󿬂 ow. The inner-wall temperature  T  w was calculated from the measured outer-wall temperature. The bulktemperature  T  b  of the fuel along the reactor tube, was predicted usinga commercial software, i.e., Ansys Fluent [32]. Nu is de 󿬁 ned as( h ∗ d ) /  κ  , where  h  is the convective heat transfer coef  󿬁 cient,  d  is theinnerdiameterofthetube,and κ  isthethermalconductivityofthefuel. 3. Results and discussion Thedissolvedoxygenconcentrationplaysamajorroleincokedepo-sition. The autoxidation of hydrocarbons consists of a complex set of free radical reactions and the dissolved oxygen is involved in fuel initi-atesreactions.Infuel-coolingsystemdesign,itisessentialtoinvestigatethe effect of dissolved oxygen concentration on the fuel thermal stabil-ity. The solubility of oxygen in the fuel increases with increasing fueltemperature. The three dissolved oxygen concentrations at the inletwere 󿬁 xedat32 ppm(high-level), 10 ppm(normal-level, air-saturatedfuel),and0 – 1ppm(deoxidized-level,belowtheprecisionoftheoxygensensor).AslistedinTable2,theheatingpowerandtimewerekeptuni-form for the three cases, ensuring a constant heat  󿬂 ux. The pressure inthe tube and mass  󿬂 ow rates were controlled at 3 MPa and 1 g/s,respectively.As presented in Table 3, the total mass of coke deposition increaseswith increasing concentration of the dissolved oxygen. In comparisonwith the deposition in the normal case, the mass deposition by thedeoxidized fuel is 31% lower. The mass deposition and oxygen con-sumption in the high-level case are nearly three times as much asthoseatthenormallevel.Thus,thetotaloxygenconsumptionisrelatedtomass deposition. Ina 󿬂 owsystem,oxygenconsumptionis controlledbyboththeoxidationkineticsandspeciestransport.Intheautoxidationchain mechanism, the cycle begins with the initiation of thermal de-composition of the hydrocarbon, producing hydrocarbon radicals R·,which react rapidly with dissolved oxygen, forming peroxy radicalsROO·.Theperoxyradicalsthenformhydroperoxides,whichactasiniti-ators, increasing the free-radical pool and the resulting in rapid oxida-tion [33]. Oxygen is therefore important in oxidation depositionreactions.As illustrated in Fig. 2, more than 50% of the oxygen is consumed atthe outlet. The oxygen consumption pro 󿬁 les of these two cases followthe same trend, and consumption increases with increasing outlet fueltemperature.Theoxygenconcentrationwasplottedastheoxygencon-sumptionfraction([O 2c /O 2in ])versus theoutletfueltemperature ( T  out  ),tomaketheresultsindependentoftheinitialoxygenconcentration.Theoxygen consumption fraction ([O 2c /O 2in ]) is de 󿬁 ned as the ratio of theconcentration of the consumed oxygen in the fuel to the concentrationof dissolved oxygen at the inlet of the test section during the experi-ment.Ina 󿬂 owsystem,theoxygenconsumptionreactionfollowspseu-do- 󿬁 rst-order kinetics [34]. Pseudo- 󿬁 rst-order kinetics means thatcomparedwith that of oxygen,the concentrationsof all other reactantsare suf  󿬁 cientfor them to be treated asconstants. Eq. (1) wasused to 󿬁 tthe oxygen consumption data. The same Arrhenius parameters wereused for the normal and high-level cases.ln O 2 c  O 2 in    ¼  aT  out  − 300 ð Þ Z   T  out  300  A exp  − E  = RT  ð Þ dT  − b  ð 1 Þ In Eq. (1),  A ,  E  / R ,  a , and  b  are 10 − 1 , 982 K, 32.98, and 4.943, respec-tively.  T  out   is the thermodynamic outlet fuel temperature, and the sub-scripts  c   and  in  denote consumed oxygen and initial oxygenconcentration, respectively.On the one hand, the oxygen consumption increases greatly above200 °C, which indicates a more violent oxidation deposition reactionsin both cases, on the other hand, the oxygen in the high-level case isconsumed faster than that in the normal case at temperatures below200 °C. When the temperature is above 325 °C, however, the oxygenconsumption fraction of high level is lower than that of the normallevel.Becausetheoxidationdepositionisnotcompletelylinearenlargedwith the increase in the concentration of the dissolved oxygen in thefuel. Under certain conditions of temperature and pressure, theremaybe a peak of surface fouling and the deposition reactions can bemore violent with less dissolved oxygen concentration. The similar re-sults were alsoobservedinthe experiments ofErvinandWilliams [23].As displayed in Fig. 3, the deposition in the high-level case is fargreater than in the other two cases. The deposition rate for the high-level case is the fastest and increases stably along the tube, due to thesuf  󿬁 cient dissolved oxygen in the fuel. Deoxidization is an effectiveway to decrease deposition along the tube, because dissolved oxygenis not present to trigger hydroperoxide formation and stimulate free-radical reactions. When the amount of oxygen is limited, the dominantradical species changes from the peroxy radical to a mixture of peroxy,alkoxy (RO·) and alkyl (vinyl) radicals (R·). Furthermore, the depositsformed in the air-saturated fuel experiment contain mostly sphericalparticles, whereas for the deoxidized fuel, it takes the form of a fused,amorphousvarnish[35].Oxygenisthereforeessential,notonlybecause of its effect on the chemistry of deposit formation, but also in terms of morphology.ItisrevealedinFig.3thattherearehigheramountsofdepositsatthetube inlet in all cases, because of the entrance effect which is deter-mined by the temperature gradient and the boundary layer. A thermalboundary layer is established as a result of heating of the wall whenthe 󿬂 uidentersthetube.Theviscosityofthe 󿬂 uidincontactwiththere-actor surfaceresists themotion of adjacent 󿬂 uid layers and slows themdowngradually.Thethicknessoftheboundarylayergraduallyincreasesinthe 󿬂 owdirectionuntiltheboundaryreachesthepipecenterand 󿬁 llsthe entire pipe. The increasing thickness of the boundary layer acceler-atesthe 󿬂 owofthecentralportion.Thelengthfromthepipeentrancetothepointatwhichthewallshearstressreachesabout2%ofthefullyde-veloped value is called the thermal entrance length,  L *. In a laminar  Table 2 Experimental parameters.Heat  󿬂 ux (kW/m 2 ) Heating time (min) Inlet mass  󿬂 ow rated (g/s) Pressure (MPa) Maximum fuel temperature at outlet (°C)384.6 105 1 3 410  Table 3 Coke deposition and oxygen consumption at different oxygen concentrations.Case Inlet oxygenconcentration(ppm)Total oxygenconsumption(ppm)Total mass of deposition(mg)Deoxidized-level  b 1  b 1 3.53Normal-level 10 6.34 5.13High-level 32 18.2 14.3588  X. Pei, L. Hou / Fuel Processing Technology 142 (2016) 86  – 91  󿬂 ow,thenon-dimensionalhydrodynamicentryisgivenby L ⁎ ≈ 0.05 dRe [36]. The  󿬂 ow at normal temperature (27 °C) of reactor inlet is laminaras Reynolds number (Re) in 751 and the mass  󿬂 ow rate of 1 g/s, there-fore  L * is 75.1 mm.Toinvestigatethein 󿬂 uenceof theinletReontheentranceeffect,anexperiment with a high Re, i.e., 3746, was performed by increasing thefuel temperature at the inlet. As plotted in Fig. 4, the inlet Re waschanged by increasingthe inlet temperature. For Re of 751, the deposi-tionrateinthesectionuptonearly60mmfromtheinletisnearlythreetimesasmuchasthoseintheothersections.Thislengthislessthantheempirical  L * calculated above in that theheatis transferred radiallyup-stream inducing a gradual increase in the inlet temperature to above100 °C. The  󿬂 ow at this temperature is transition  󿬂 ow with Re at1850.TheReiscalculatedbasedontheviscositycoef  󿬁 cientand densitythat are varied with the fuel temperatures [30]. This value is just be-tween laminar  󿬂 ow and turbulent  󿬂 ow. Growth of the boundary layerfor turbulent  󿬂 ow is faster than that for laminar  󿬂 ow, resulting in ashorter  L *, i.e., about 25 – 45 d , mainly because of   󿬂 ow disturbance. Theentrylengthfortransition 󿬂 owinthisexperimentisthereforebetweenthose for the turbulent and laminar  󿬂 ow.In Fig.4, the turbulence inlet condition mitigatestheentranceeffectandleadstoanobservabledecreaseindepositformationattheinlet,be-causetheintensemixingduringrandom 󿬂 uctuationsoutweighstheef-fect of molecular diffusion.Inthethermalentranceregion,thewallshearstressisthehighestatthetubeinlet,wheretheboundarylayerthicknessisthesmallest,anditdecreases along the  󿬂 ow direction; this induces a long residence timefor the 󿬂 ow and enhances deposition. Also, as the initially thin thermalboundary layer develops near the heated wall of the inlet, the radialtemperature gradient is large at the inlet wall but decreases with in-creasingaxialdistance.Notonlythereactionsbutalsothedepositmor-phology are different. In the stable heated sections, the deposit layershavemicro-spheroidalstructures,whereasinthegradienttemperaturesection, a  󿬁 lm with no perceptible small-scale structure, which con-forms to the micro-geometry of the underlying stainless-steel surface,is formed. Laminar  󿬂 ow, high wall shear stress and temperature gradi-ent,andprolongedresidencetimenearthewallallstimulateformationof oxidized deposits near the entrance of the reactor tube.As given in Fig. 5, it is obvious that there are some  󿬂 uctuations inboththewalltemperature andmassof deposits alongthetube.Theex-tentofthewalltemperature 󿬂 uctuationincreaseswithheating.Inaddi-tiontheregionwiththepeakwalltemperatureapproximatelycoincideswith the area of high coke deposition. In this case, deposition on theinnersurfaceincreasesthewalltemperature.InFig.6,thewalltemper-ature, corresponding to a high level deposition at location  “ B ”  in Fig. 5,increases faster than the one at location  “ A ”  during heating process. Fig. 2.  The oxygen consumption fraction with the outlet fuel temperature. Fig. 3.  The deposition rate distribution along the tube. Fig. 4.  The deposition rates for different inlet Re numbers along the tube. Fig. 5.  Wall temperature and deposition pro 󿬁 les at different time and locations.89  X. Pei, L. Hou / Fuel Processing Technology 142 (2016) 86  – 91
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x