. ENGI NEERINIJ
`.PERIODICALS
`
`UNIVERSITY OF WASHiNGTON
`
`twv 1 5 2001
`'"
`-
`
`LIBRARIES
`
`BUSHED QUARTERLY BY THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS · OCTOBER 2001
`
`GE-1025.001
`
`

`
`Journal of
`Engineering for Gas
`Turbines and POUTer
`
`Published Quarterly by The American Society of Mechanical Eng ineers
`
`VOLUME 123 • NUMBER 4 • OCTOBER 2001
`
`TECHNICAL PAPERS
`Gas Turbines: Advanced Energy Systems
`717 The Thermoeconomic and Environomic Modeling and Optimization of the
`Synthesis, Design, and Operation of Combined Cycles With Advanced
`Options
`S. Pelster, D. Favrat, and M. R. von Spakovsky
`727 Development of Methanol Steam Reformer for Chemical Recuperation
`T. Nakagaki, T. Ogawa , K. Murata, and Y. Nakata
`Gas Turbines: Coal, Biomass, and Alternative Fuels
`734 Integration of Advanced Gas Turbines in Pulp and Paper Mills for
`Increased Power Generation (OO-GT-020)
`K. Maunsbach, A. Isaksson, J. Van, G. Svedberg, and L. Eidenster
`Gas Turbines: Combustion and Fuels
`741 Deposit Formation and Mitigation in Aircraft Fuels (99-GT-217)
`L. J. Spadaccini, D. R. Sobel , and H. Huang
`747 In Situ Detailed Chemistry Calculations in Combustor Flow Analyses
`(99-GT-271)
`S. James, M. S. Anand, M. K. Razdan , and S. B. Pope
`757 Reduced NOx Diffusion Flame Combustors for Industrial Gas Turb ines
`(OO-GT-085)
`A. S. Feitelberg, V. E. Tangirala, R. A. Elliott, R. E. Pavri , and
`R. B. Schiefer
`766 Acoustic Resonances of an Industrial Gas Turbine Combustion Syste m
`(OO-GT-094)
`S. Hubbard and A. P. Dowling
`774 Flamelet Model of NO. in a Diffusion Flame Combustor (OO-GT-099)
`D. V. Volkov, A. A. Belokin, D. A. Lyubimov, V. M. Zakharov, and
`G. Opdyke, Jr.
`779 Self-Excited Oscillations in Combustors With Spray Atomizers (OO-GT-10E
`M. Zhu , A. P. Dowling, and K. N. C. Bray
`787 Research on a Methane-Fueled Low NO. Combustor for a Mach 3
`Supersonic Transporter Turbojet Engine (OO-GT-11 3)
`Y. Kinoshita, T. Oda, and J. Kitajima
`796 Application of Macrolamination Technology to Lean, Premixed
`Combustion (OO-GT-11 5)
`A. Mansour, M. Benjamin, D. L. Straub, and G. A. Richards
`803 Integrated Experimental and Numerical Approach for Fuel-Air Mixing
`Prediction in a Heavy-Duty Gas Turbine LP Burner (OO-GT-122)
`G. Mori , S. Razore , M. Ubaldi, and P. Zunino
`810 Experimental and Numerical Investigation of a Planar Combustor Secto r
`at Realistic Operating Conditions (OO-GT-123)
`M. Carl , T. Behrendt, C. Fleing, M. Frodermann, J. Heinze,
`C. Hassa , U. Meier, D. Wolff-Gassmann, S. Hohmann, and
`N. Zarzalis
`
`(Contents continued on inside back cover
`
`Th is journal is printed on ac id -free paper, wh ich exceeds the ANSI Z39.48(cid:173)
`1992 specification lor perm anence 01 paper and library materials. @lTM
`@ 85% recycled content, in clUdi ng 10% post-consumer fibers.
`
`GE-1025.002
`
`

`
`L. J. Spadaccini
`D. R. Sobel
`H. Huang
`
`United Technologies ResearchCenter,
`East Hartford, CT
`
`Deposit Formation and Mitigation
`in Aircraft Fuels
`
`The development of a viable strategy fo r limiting coke deposition involves combining
`synergistic approaches fo r suppressing deposit buildup and reducing its impact on per(cid:173)
`fo rmance. Candidate approaches, including selection of favo rable operating conditions
`(viz., pressure, temperature, heat flux, residence time, and passage size) and coke-tolerant
`heat exchanger designs, were investigated to evaluate their effectiveness and provide a
`basis fo r combining them into a single design philosophy. These approaches were evalu(cid:173)
`ated through testing of current jet f uels in single-tubes and segments of heat exchanger
`configurations at temperatures up to 1000°F, pressures up to 1200 psi and liquid hourly
`space velocities up to 40,000Ih. A key result of this work is the ranking of the importan ce
`of heat exchanger operatin g conditions on carbon deposition, with fu el temperature and
`those param eters that control species diffusion having the most pronounced impact. Resi(cid:173)
`dence time and pressure are of
`lesser importance. Alternative coke-tolerant heat ex(cid:173)
`changer designs f eaturing interchannel communication were evaluated and ranked, with
`several of these concepts demonstrating improvement over continuous passages.
`[DO l: 10.1115/1.1383772]
`
`thermal stability additives. The test result s can be used to define
`guidelines to support fuel heat exchanger/thermal management
`system design and implementation, as well as to broaden the da(cid:173)
`tabase available for validating and upgrading coke depo sition
`models.
`
`Background
`In the temperature regime below approximately 700 °F, oxygen
`dissolved in fuel is the major contributor to therm al decomposi(cid:173)
`tion and coke depo sition . When air-saturated fuel
`is heated to
`temperatures above approximately 350 °F, the dissol ved oxygen
`reacts to form free-radical species (coke precur sors) which initiate
`and propa gate other autoxidation reactions leadin g to depo sit for(cid:173)
`mation . These reaction paths are domin ant at temperatures up to
`approximately 700°F, but become insignificant when the concen(cid:173)
`tration of the dissolved oxygen is reduced from its saturated value
`(- 70 ppm) to very low levels « 5 ppm). It has been show n that
`the dissolved oxygen is consumed as coke depo sit is formed , and
`that increasing fuel
`temperature within the autoxidation regime
`acce lerates oxygen depletion ([ I]).
`
`Intro duction
`High heat sink fuel coolin g technology can be applied to en(cid:173)
`hance engine performance over the entire spectrum of flight re(cid:173)
`gimes. For hypersonic flight, it provides the only means for meet(cid:173)
`ing the coo ling requirements with storable fuels; for advanced
`fighter aircraft,
`it provides a path to enhance performance with
`current materials; and, for lower-speed military and co mmercial
`aircraft, it can increase growth potential and play a key role in
`emissions-reduction strategies. However, utilization of this tech(cid:173)
`nology will
`require
`heating
`the
`fuel
`to
`supercritical
`temperatures-well above the maximum temp erature (- 325°F)
`allowable in conventional fuel systems.
`The principal engine operability issue that will affect hydrocar(cid:173)
`bon fuel coo ling technology is coke formation.
`In gas turbine
`applications, where long duty cycles are required , even low rates
`of deposition will accumulate, degradin g perform ance and, if left
`unchecked, leading to system failure. In hypersonic missile appli(cid:173)
`cations, duty cycles are short, but requirem ents for maxi mizing
`heat sink lead to very high fuel
`temperature operation and the
`potential for accelerated coking. The extent to which the benefits
`of this technology can be realized is directly related to our ability
`to mitigate against coke formation .
`The objective of this study is to investigate coke deposition in
`aircraft fuels and to identify design concepts for minimizing its
`impact. Candidate approaches were investigated to evaluate their
`effectiveness and provide a basis for comb ining them into a single
`desig n philosophy. The individual elements examined include : ( I)
`definition of guidelines for specifying operating co nditions to
`minimize coke deposition in both the subcritical and supercritical
`temperature regimes, and (2) identification and eva luation of heat
`exchanger design co ncepts that have the potenti al for acco mmo(cid:173)
`dating moderate coke buildup without progressive degradation.
`The key variables and component operating conditions exam(cid:173)
`ined correspond to gas turbine and scramjet applications. Tests
`were conducted using let A and lP -8 fuels that were supplied by
`the Air Force Research Laboratory (batches POSF-2926 and
`-3405 , respectively) and with lP- 8+ 100, which include s special
`
`Con tributed by the Internation al Gas Turb ine Institute (IGT I) of T HE AMERICAN
`SOCIETY OF MECHANICAL ENGINEERS for public ation in the AS ME JOURNAL OF
`ENGINEER ING FOR GAS TuRBINES AND POWER. Paper presente d at the Intern a(cid:173)
`tional Gas Turbine and Aeroe ngine Congress and Exhib ition, Indianapol is, IN, Jun e
`7- 10, 1999 ; ASME Paper 99-G T-2 17. Man uscript rece ived by IGT I Del. 1998; final
`revision recei ved by the AS ME Headqu arters Mar. 1999. Associate Ed itor: D. Wisler.
`
`c:
`~
`"iii
`oc.
`
`Q)
`
`'C-o
`
`C)
`o
`..J
`
`Autoxidation
`
`---
`
`Deoxygenated
`
`200
`
`400
`
`800
`600
`Temperature, F
`
`1000
`
`1200
`
`Fig. 1 Deposition rate for jet fuel
`
`Journal of Engineering for Gas Turbines and Power
`Copyright © 2001 by ASME
`
`OCTOBER 2001, Vol. 123 / 741
`
`GE-1025.003
`
`

`
`the dep osition
`temperatu res above approxi mately SOO°F,
`At
`mechanism is characterized by pyrolysis, wherein che mical bond s
`are broken and large alkanes are co nverted into smaller alkanes,
`alke nes , and so me hyd rogen . Th is mec ha nism for coke production
`ca n be initiated by ther mal (homogeneous) and/or ca taly tic (het(cid:173)
`erogeneo us) crac king reacti ons, and follow ed by polymeri zation .
`Th e rate of co ke deposit buildup with increasing tem perature is
`illustrated in Fig . I for a typ ica l je t fuel. As indicated in the figure,
`fuel deoxygen ation redu ces coke de pos ition at low and inte rme(cid:173)
`diate temperatures. At higher temperatures, dep osit buildup fro m
`aerated fuel approach es deposit buildup from deoxygen ate
`fuel because the dissolved oxygen is con sumed and the co ke
`prec urso rs are decomposed ([2] ).
`
`Test Appa ra tus
`Most of the testing was performed with the multiple hea ted(cid:173)
`tub e co king rig show n in Fig . 2. The indivi dual tubes were man u(cid:173)
`factured fro m type 304 stainless stee l and sized to simulate single
`passages in a heat exc ha nger. Th e parallel flow path s enable si(cid:173)
`multaneo us eva luatio n of up to five co nfigurations that may have
`diffe rent geometries or operating co nditions, and are particularly
`efficient for long-duration testin g. A nearl y uniform surface tem (cid:173)
`perature distribution is obtai ned by clasping the tubes between
`iso therma l co pper blocks (S in. x S in. X I in.) that are heated elec (cid:173)
`tricall y by strip heaters attached to the outer surface. In some
`tests, co upons co nsisting of mult iple flow passages were used .
`Prior to a test, 1.5 ga l piston -type acc umulators are filled with
`fue l by a gea r pump. Th e 1.5 ga l capacity allow s a full day of
`testing at moderate flow rates . Longerlhi gh er flow rate tests ca n be
`accommoda ted by refilling fuel on-line, during tests. By applying
`
`sarrple
`
`fuel
`llCQImulata
`
`fl it S'
`
`F
`
`f1ov.metS'
`
`heated
`CDPPEI'
`blocks
`
`burmr
`
`liquid
`cd Iecta
`
`gas
`
`high -pressure nitrogen to the back-side of the pistons in the accu(cid:173)
`mulators, the fuel is then pressuri zed to 150 to 200 psig higher
`than the predetermined test pressures. The five flowpaths are es(cid:173)
`sentially indepen den t; the co nstruc tio n of the rig requi res only that
`the test sect ions opera te at the same back pressure . The flow paths
`may have different fuels, flow rates, additives, coa tings, and/ or
`tube sizes. In eac h of the separa te flowpath s, the fuel is filtered
`and the flow rate is measured by a flow meter and controlled by a
`metering valve . Depend ing on the desired simulation cond itions.
`the fuel ca n be preh eated up to about 350°F, which is below its
`autox ida tive co ke deposition temp eratur e, befo re entering the first
`of two succe ssive test sec tions. Th e test sections can be main(cid:173)
`tained at different
`tem peratures (up to approximately 1200 °F).
`simulating differen t parts of a thermal management sys tem.
`Downstream of the heat excha nge r tubes, the processed fuel can
`be sampled for compositional ana lysis and is quenched in a
`cooler. System pressure
`is maintained by a back-pressure
`regulator.
`Durin g a test, key measurement s includ e the inlet and exi t fue l
`tem peratures and pressures, and copper block and heat exc hange r
`wall temperatures. These measurements pro vide important infor(cid:173)
`mation to correlate performance and coke deposition. In prepara(cid:173)
`tion for determining the time-averaged carbon dep osition rate, the
`heat exc ha nge r tube is removed from the rig at the test comple(cid:173)
`tion, rinsed with hexane and dried in a vacuum oven . The tube s
`are then cut into 1/2- 2 in. length s and the carbon accumulated in
`eac h sec tion is determined using a LECO RC-41 2 Carbon Deter(cid:173)
`minator which qua ntifies carbon deposi tio n (precision of ~ 3 per(cid:173)
`cent) by meas uring carbon dioxide produced in a controlled car(cid:173)
`bo n burnoff. Typica lly,
`the deposit co nsist s of more than 75
`percent ca rbon, with the rema inder being oxygen and hydrogen .
`For the most part. test result s are correlated using the total carbon
`deposit
`in the tube,
`i.e.,
`the sum of the individual sections.
`However,
`the axial profiles are also examined to explain local
`phenom en a.
`
`Operating Conditions
`The range of parameters investigated was selected to reflect that
`suita ble for practical heat exc hanger sys tems and is summarized
`in Table I . Because the flow passages are small, fuel temperatures
`were measured onl y at the inlet. the midpoint (betwee n the first
`and second heated copper blocks), and the exit of the tube s. Fuel
`temperature profiles were ca lculated between the measured values
`based on a uniform surfac e temperature within each block and
`heat tran sfer coefficients determined using the Ditt us-Boelt er co r(cid:173)
`re lation. In the analysis, the complex jet fuel composition is simu(cid:173)
`lated by a mixtu re of 12 hyd rocarbons ([3]) and the physical prop (cid:173)
`erties of the mixture at various temperatures and pressur es are
`estimated using the NIST SUPERTRA PP program ([4]). A repre(cid:173)
`sentative fuel
`tem perat ure profile in successive heat exchanger
`tubes is show n in Fig . 3.
`
`Fuel Temperature. Representative coke deposit
`levels as a
`function of fuel temperature are plotted in Fig . 4. The data are
`correlated in ter ms of parts per million (ppm) of carbon dep osit
`
`Table 1 Test conditio ns
`
`Key Variables
`
`Fuel
`Temperature. F
`Flow rate. Ibmfh
`Tube 10 , in.
`Pressure. psig
`lhsv " , Ilh
`Residence time. s
`Test duration. h
`Tube materia l
`
`Range
`JP-8+ 100
`300- 1000
`1.8- 7.1
`0.03. 0.058. 0.085
`600. 1200
`860-9200
`0.14- 1.40
`7-25
`stainless steel- 304
`
`Fig. 2 Multiple heated-tube coking rig
`
`*Ihsv= volumetric now rare/ru be volume
`
`742 / Vol. 123, OCTOBER 2001
`
`Transactions of the ASME
`
`GE-1025.004
`
`

`
`0.4r--.----,---.---..------.----.----,
`
`·
`
`",.... +.
`
`..
`
`,
`
`JP-8+100
`18-1n.long
`m= 3.5Ibm/hr
`······ 1 P = 600 pslg
`TlUoIout =780-820 F
`L...:.t "'=..=8;..:h.:.-
`-'-i
`
`':;:l
`
`E
`8: 0.3r
`g-
`iiia 0.2
`Gl"C
`
`~~
`
`0.1
`
`o'----'-_----''-_---'-__.J...-_--'-__-'-_-...l
`0.06
`0.04
`0.08
`Tube 10, In.
`
`Fig. 5 Effect of passage size on deposition
`
`above approximately 350 °F, the dissolved oxygen reacts to form
`free-radical species (coke precursors) which participate in other
`autoxidation reactions leading to deposit formation. The surface
`depos ition mechanisms are inextricably tied to the mass transfer
`proce ss. For example, the surface depos ition mechani sms
`
`Fuel +Oz->Deposit
`
`Precursor-s-Deposit
`
`Depositbu1k-> Deposi t
`involve diffusion of disso lved oxygen, coke precursor, and deposit
`formed in the bulk flow to the wall surface. The mass transfer
`length scale, i.e., the mean distance between molecules and the
`wall surface, is linearly proportional to the passage diameter, and
`shorter length scales facilitate diffusion . Turbu lence of the fuel
`flow is also affected by the passage size. Smaller 10 tubes have
`larger Reynolds numbers (i.e., Re=4rh/7TdIJ.) and therefore higher
`mass transfer coefficients, also promoting diffusion and thereby
`deposit ion. As Reynolds number is increased, heat transfer coef(cid:173)
`ficients also increase. Deposition at very high heating rates is
`discussed below.
`
`Increasing the resi(cid:173)
`Resid en ce Time and Spa ce Velocity.
`dence time of fuel in the heat exchanger tubes wou ld be expected
`to increase coke deposition. For this reason, a smaller dia meter
`tube wou ld be expec ted to give less coke deposition than a larger
`diameter tube with the same flow rate. However, as discussed
`above, diffusion is the controlling factor for autoxidative coke
`deposition at the conditions tested. The influence of the fuel flow
`rate on the autox idative coke deposition is show n in Fig. 6. Car(cid:173)
`bon deposition results are normalized by the fuel flow rate and
`indicate that, for a constant fuel outlet temperature, carbon depo(cid:173)
`sition increases with increasing fuel flow rate. The mean Reynolds
`numbers (calculated based on film temperature) and residence
`times in the heated tubes at the three fuel flow rates plotted are
`also show n in the figure. As can be seen , although residence time
`was reduced 50 percent by increas ing the flow rate, the deposition
`doub led. From this and the discussion above , it can be concl uded
`that the enhanced diffusion effect
`is stronger than the impact of
`residence time .
`To isolate the influence of residence time at a constant Rey(cid:173)
`nolds numb er, results from two tests were com pared, namely, a
`high fuel flow rate in a large-diameter tube (3.5 lbm/h, 0.058 in.
`10, Fig. 5) and a low fuel flow rate in a small-dia meter tube (I .8
`lbm/h , 0.030 in. 10, Fig. 6). As the Reynolds numbers are equiva(cid:173)
`lent (i.e., Re=6800) and the carbo n depositions were approxi(cid:173)
`mate ly equal in the two tubes, the com pariso n indicates a balance
`
`JP-8+ 100
`0.03-ln .10
`m .. 3.5Ibmlh
`P .600pslg
`
`800
`
`700
`
`LL
`
`::s
`EII)
`
`EI
`
`D. 600
`
`I)-Gi 500
`
`::s
`LL
`
`400
`
`kf-~- Too· 790F-+- Too· 860 F------1
`
`300 '----'_ --'_ --'-_ --'-_ --'-_ ......._-'-_...I.-_.l....-_.L...----l
`o
`2
`4
`6
`8
`10
`12
`14
`16
`18
`Axial position, in.
`
`Fig . 3 Fuel temperature dis tribution
`
`relative to fuel flow (i.e., masscarbon /mass fuel) and normalized by
`surface area for each of the two inch segments analy zed in the
`LECO carbon determinator. As shown in the figure, the carbon
`deposit ion rate increases with increasing fuel temperature up to a
`point (- 600°F) beyond which any further increase in fuel tem(cid:173)
`perature cause s the deposition rate to decline. As discussed earlier,
`the coke deposit ion over this temperature range comes from the
`reaction of oxygen dissolved in the jet fuel. The oxidation rate of
`the fuel is accelerated by increasing the fuel temperature, thereby
`increasing the coke deposition rate. However, as the reaction pro(cid:173)
`ceeds along the heat exchanger tube, dissolved oxygen is con(cid:173)
`sumed and its concentration decreases. In addition, the decompo(cid:173)
`sition of coke precursors is also accelerated by increasing fuel
`temperature. Thus , the depletion of both dissolved oxygen and
`coke precursors results in the decreasing rate of deposition .
`
`Reynolds Numbe r an d Passage Size. Both heat and mass
`transfer rates are dependent on Reynold s number (Re). The influ(cid:173)
`ence on coke depo sition of a change in Reynolds number and flow
`passage diameter at constant flowrate is shown in Fig. 5 in terms
`of total accumulated carbon . The results indicate that increasing
`the passage diameter (i.e., decreasing Re) reduces carbon deposi(cid:173)
`tion, a trend that has also been reported by Chin and Lefebvre [5]
`and Jones et al. [6]. Moreover,
`if the results were normalized by
`surface area, the effect of passage size on the deposition would be
`even more pronounced. The trend of decreasing deposit rate with
`increasing diameter reflects a balance of mass-diffusion length
`scale,
`turbulence-enhanced mass-diffusion rates and residence
`time influences . When air-saturated fuel is heated to temperatures
`
`/
`
`/ f
`I
`
`0.6
`
`0.5
`
`0.4
`
`0.3
`
`0.2
`
`0.1
`
`o3
`
`00
`
`..c
`
`E1
`CI.
`CI.
`
`Co
`
`E..ogo
`
`';:J
`
`co
`
`.c
`~
`
`A
`
`/ 1
`
`/
`
`JP-8+100
`O.ll3-/n. 10; 18-1n. long
`
`P. 600 polg
`t.6h
`-, T.... 790 -860F
`0:t
`
`-, m .1.6 kgn.
`-,
`
`\
`
`\
`
`...
`
`400
`
`SOO
`
`I
`
`700
`600
`Fuel temperature, F
`
`i
`800
`
`900
`
`Fig. 4 Representative deposition profile
`
`Journal of Engineering for Gas Turbines and Power
`
`OCTOBER 2001, Vol. 123 / 743
`
`GE-1025.005
`
`€
`

`
`decrease in the Reynolds number, as a conseque nce of increases in
`fuel viscos ity and density. Therefore, fuel pressure should have
`only a small effect on mass transfer and residence time at these
`conditions. Th us, as observed in the tests.
`the variation of fuel
`pressure in the liquid range has a minor impact on the autoxida(cid:173)
`tive coke deposition. At higher temp eratures, where the fuel is a
`supercritical vapor,
`increasing pressure will
`increase residence
`time and pyrolytic reactions may be more significantly affected.
`
`Hi gh Heating Rat e. To determine the cok ing characteris tics
`under very high heat flux and low reside nce time conditions, a
`series of tests was performed with resistively heated tubes in the
`single-element Bench-Scale Reactor Test Rig descr ibed in Sobe l
`and Spadaccin i [7]. (These high heat fluxes can not be achieved
`over a substantial length of the tube using the multiple-heated tube
`cok ing rig.) In the single-element bench-scale reactor test rig, the
`heat flux is uniform and the wall tem perature increases from inlet
`to exit (typically from 800°F to 1400°F). The resistively heated
`tubes were 12 in. long and 0.030 in. 10 , and were selected to
`maximize the heat flux, by reducing wetted area at a given power
`input, while avoiding excessive end effects and maintaining
`readily manageable flow rates with subs tantial residence times . In
`these tests, the fuel was not preheated upstream of the tube.
`Tests were run with heat fluxes from 45 to 240 Btu/ft2s at 600
`psi for a duration of seven hours (one run day). For each test. the
`fuel exit temp erature was maintained at 1000°F and both autoxi(cid:173)
`dati ve and pyrolytic co ke was formed. It should be noted that the
`space velocity increases by a factor of approx imately six over the
`range of flow conditio ns exami ned; the residence time is reduced
`by esse ntially the same factor (from 0.4 to 0.07 s).
`The carbon deposition results are shown in Fig. 8. Since the
`tests were run with the same out let temperature but different flow
`rates, this analysis allows a direct com parison of the amo unt of
`carbon deposited for a specific fuel heat sink (heat sink/masslue')'
`These results show that carbon deposition can be reduced by sig(cid:173)
`nificantly increasing heat flux, while increasing flow rate (and
`Reynolds numb er) and decreasing the residence time . This trend
`reflects the changing balance between heating, mixing, and kinet(cid:173)
`ics, and occ urs in spite of the higher wall temperatures associated
`with the higher heat flux. As heat flux is increased substantially
`with the same fuel ex it temp erature, the wall temp erature will also
`increase. At heat fluxes of 45 and 240 Btu/ft2s, the exit (peak ) wall
`tempera tures were 1200 and I450°F, respec tively-well into the
`pyrolytic regime.
`(The carbon distribution indicates approxi(cid:173)
`mately equa l amounts of autox idative and pyrolytic deposition.)
`
`0.6 ~------------------,
`
`JP-8+ 1 00
`0.030-ln. 10 , 12 ·ln. lon g
`P = 600 ps i
`T,u. ' oul = 1000 F
`t = 7h
`
`o
`
`100
`
`0.5
`
`0.4
`
`0 .3
`
`0.2
`
`EQ
`
`.
`Q.
`
`C
`.2
`II
`o
`Q.
`GI
`'tl
`
`~ I
`
`Co-e
`
`III
`Co)
`
`9.2 1bmlh , 1450 F, 85000
`
`200
`2-s
`
`300
`
`Heat flux, Btu/ft
`
`0.5 ,-----,..---r---,........-----.,....-----r----r----,
`
`JP-8+1oo
`P.600 palg
`O.03-In.ID
`T.... __ .20F
`h.h ·
`
`..
`
`I
`I
`----t----~17r========----'----,
`MIIILBI BII'dlnet tiD Tube Itngtb
`•
`'n.
`I
`I
`1.
`:
`I
`- ---j--
`1.
`!
`I
`27
`-,-_--,_...J
`!
`,
`!
`'
`3
`2
`
`- --
`
`.800
`1.8
`3.5
`12000
`'-_-;-__.,....--
`7.1
`17000
`
`0.2'
`0.16
`0.14
`
`4
`Fuel flow rate,
`
`5
`
`Ibm/h
`
`6
`
`7
`
`8
`
`E8
`
`;::
`
`: 0.4
`co
`'iiia~ 0.3
`co.c
`~
`
`0.2
`
`Fig . 6 Effect of flow rate on deposition
`
`between the sma ller diameter (length scale) tendin g to increase
`deposition and the larger diam eter (high residence time) tending to
`decrease deposition. Also, the axia l loca tion of the peak de posit
`was shifted downstream to higher temperature in the smaller(cid:173)
`diam eter
`tube,
`reflecting the effect of
`reaction kinetics on
`deposi tion.
`Liquid hourly space velocity. or lhsv, is defined as the ratio of
`the volum etric flow rate of liquid fuel to the passage volume. Th is
`sca ling parameter incorpora tes the effec ts of the tube 10, length ,
`and residence time . A sma ller tube or higher fuel flow rate gives
`larger space velocities, higher mass trans fer rates and lower resi(cid:173)
`dence times. The influence of the space velocity on the autoxida(cid:173)
`tive co ke deposition is show n in Fig. 7. which illustra tes that. for
`the turb ulent flow co nditions tested (i.e., 4500< Re< 17000),
`the
`autox idative coke depo sit increases with increasing space velocity.
`At higher Reynolds numbers, where turbul ent mixing is enhanced
`and residence time is sign ificantly decreased, it is anticipated that
`deposit forma tio n will no longer be co ntrolled by mass transfer,
`but by the reaction kinetics.
`
`Pressure. The influence of the fuel pressure on autoxidative
`co ke deposition was investigated using a 0.085 in. 10 tube that
`was 18 in. long. At pressures of 600 and 1200 psig, 3.5 lbm/h fuel
`flow and 810°F fue l outlet tempera ture, the total cumulative car(cid:173)
`bon depositions were relative ly low. i.e., 0.062 and 0.050 ppm .
`respecti vely, afte r eight hours duration. Since the fuel is in the
`liquid state over a significant portion of the tube, increas ing the
`fuel pressure from 600 to 1200 psig effec ts only a very small
`
`0.6 , - - - - - - , . . . - - - - - - - - - - - - - - - - - - ,
`
`I
`...... [
`
`!i
`
`1·
`;
`i
`·_-_··-·-·_·i··-·_·-
`
`~ ;
`
`JP-'+100
`P. 600 palg
`T........ 76ll-820 F
`T_.800F
`h.h
`
`················
`
`···
`
`: /"i
`
`_ -..+.."
`
`"
`
`_.-
`
`~. _..__ . ··..··
`
`.
`
`0.5
`
`EQ
`
`.
`': 0.4
`
`coE~
`
`0.3
`
`Q.
`GI
`
`'tl5 0.2
`-eIII
`
`Co) 0.1
`
`o
`
`o
`
`2000
`
`4000
`
`6000
`Ihsv,1/hr
`
`8000
`
`10000
`
`Fig. 7 Effect of space velocity on deposition
`
`Fig . 8 Effect of heat flux on deposition
`
`744 / Vol. 123, OCTOBER 2001
`
`Transactions of the ASME
`
`GE-1025.006
`
`

`
`These results support the hypo thesis that at high Reynolds num(cid:173)
`bers species diffu sion is very rapid and the deposition proce ss is
`limited by the reaction kinetics.
`
`Coke-Tolera nt Designs
`Understanding the impact of operating conditions on deposit
`forma tion enables effective flow passage design . Conventional
`heat exchanger designs, whether co, counter, or cross-flow, are
`charac terized by individual parall el fuel passages. In this type of
`design, carb on depo sits may build up over time , causing a restri c(cid:173)
`tion in a given passage. The restriction , in turn, can reduce the
`coo lant flow in that passage ,
`increasing local surface and fuel
`temp eratures and thereby potentially increasing the coking rate
`itself. Moreover, the increased surface temperature will also result
`in higher temperatures in neighbo ring passages and the fouling
`problem may propagate across the heat exchan ger, resul ting in
`failure. Therefore, candidate configurations for coke-tolerant de(cid:173)
`signs, i.e., designs which have the potenti al for accommodating
`moderate coking rates without progressive degradation, were
`identified, fabricat ed as sectors and tested in the multi-tube coking
`rig. Concepts were evaluated that allow for interch annel commu(cid:173)
`nication, perm itting isolation of coke buildup and thereby coke
`acco mmodation, and enh anced heat
`transfer, due to turbulence
`generation and frequent restarting of the boundary layer.
`A standard test element design, shown in Fig. 9, was developed
`and five different fluid passage geometries were created by ma(cid:173)
`chining or chemical milling type 304 stainless steel. Although
`chemical etching increases surface roughness (from ~ 0 .4 JLm to
`- 4.0 JLm) which can promote deposition (by increa sing surface
`area or turbul ence), the effect is greatly reduc ed or eliminated as
`deposits accumul ate ([8]). As shown in the figure , the overall size
`of all coupons was 8.0 in.XO.55 in. XO.25 in. The intern al patterns
`were as follow s: ( I) Parallel channels-four 0.050 in. XO.050 in.
`straight flow channels separated by 0.025 in. ribs ; (2) Offset par(cid:173)
`allel strip fins-
`same as above, except that the ribs had 0.100 in.
`long stagge red openings cross- connecting the channels each 1.00
`in.
`along the
`length ;
`(3) Circular pin
`fins-O.O lOin. dia .
`XO.030 in. high fins in a 0.045 in.XO.067 in. stagge red grid array;
`(4) Small
`teard rop pin fins-O.OIOin.XO.030 in.
`teardrop fins,
`0.030 in. high, in a 0.033 in.XO.120 in. staggered grid array ; and
`(5) Large teardrop pin fins-O.O10 in. XO.090 in.
`teardrop fins,
`0.030 in. high, in a 0.033 in. XO.352 in. staggered grid arra y. Ad(cid:173)
`ditional key geometric paramet ers of the test elements are sum(cid:173)
`marized in Table 2. The same parameters for three cylindri cal
`tubes are also listed in the table for comparison . The differences in
`these parameters can have significant effects on the heat transfer
`(i.e., the transverse temperature gradient) and coke depo sition.
`The open volumes of all of the test elements listed in the table are
`
`Table 2 Test element geometric patterns
`
`Coupon Type
`
`No. of Pins
`
`Parallel channels
`Offset strip fins
`Circular pin fins
`Small teardrop pin fins
`Large teardrop pin fins
`Tubes:
`0.085-in. ID
`0.058-in. ID
`0.030-in. ID
`
`1586
`1155
`394
`
`Vnof
`In .
`
`0.0706
`0.0715
`0.0745
`0.0725
`0.0727
`
`0.0454
`0.0211
`0.0057
`
`A ~urf~{e
`In .
`
`A . /V!
`ltin.
`
`5.63
`5.63
`6.87
`8.06
`7.64
`
`2.14
`1.46
`0.75
`
`80
`79
`92
`III
`105
`
`47
`69
`133
`
`the mean linear flow velocity of fuel
`therefore,
`quite similar;
`through the coupon and the fuel pre ssure drop should be compa(cid:173)
`rable for the same mass flow rate.
`Tests were conducted using lP-8 fuel instead of lP-8 + 100 to
`accelerate coke deposition and, thereby, reduce the test
`time re(cid:173)
`qu ired to evalu ate the design conc ept s. A 0.058 in. ID tubu lar
`preheater was installed in the first copper block (to raise the fuel
`inlet temperature to 420 °F) and the second copper block, contain(cid:173)
`ing the test elements, was heated to I 150°F. The tests were run
`until the fuel pressure drop across the test element increased sig(cid:173)
`nificantl y, indicating that complete blockage was imminent. Sinc e
`the coke depo sit buildup introduced an additional thermal resis(cid:173)
`tance at the wall , cau sing the fuel temperature to decrease as the
`tests proceed, the fuel exit temp eratures were time averaged.
`Carbon deposit profiles for the large-teardrop pin-fin sector
`align ed in two different direction s of flow are shown in Fig. 10.
`The data corre spond to sixteen 1/2 in. segments of the test ele(cid:173)
`ment. Reversing the or ientation (i.e., flow from the sharp end of
`the teardrop to the blunt end ) facilitates eva luation of the effects
`of reci rculation and turbulence. More carbon deposit was found
`with the fuel flow directi on from sharp end to blunt end, which
`produces a larger and more turbu lent wake flow. To reduce the
`coke depos ition , the fuel flow direction should be arranged from
`blunt end to sharp end for the teardrop pin fins. Consequently,
`only fuel flow from blunt end to sharp end was tested for the small
`teardrop element.
`Carbon depo sit profile s from tests with each of the five different
`fuel passa ges are compared in Fig. II . Despite the differences in
`geometry, carbon dep osition always peaks at an axial position
`corresponding to a fuel temp erature of 550-650°F. This behavior
`is con sistent with cylindrical tube s and is due to the consumption
`of dissolved oxygen and coke precursors. The test data , consisting
`
`0.4,---.------,,---
`
`--.---- - , - ---.--- --,.---r----,
`
`JP·8
`m = 3.51b1h
`P = 600 psig
`T......= 420 F
`TN. out = 985 F
`
`[0.3
`c.
`
`Co
`
`;:l
`
`iii&. 0.2
`
`Gl
`'t:l
`
`co
`
`.c
`~ 0.1
`
`Outlet Tube
`
`Test
`
`Inlet Tube
`
`Fig. 9 Test element design
`
`Fig. 10 Deposit profiles in large-teardrop pin fins
`
`Journal of Engineering for Gas Turbines and Power
`
`OCTOBER 2001, Vol. 123 / 745
`
`°O~--'----=-----'----';-----'--~--'--......:::..J
`8
`6
`4
`2
`Axial positIon, In.
`
`GE-1025.007
`
`

`
`solved oxygen and coke precursors. Pyrolytic deposition begin s at
`approximately 800°F and increases monotonically.
`• The transport/diffusion of reactive species to the wall (gener(cid:173)
`ally the location of maximum temp erature and residence time) is
`the next most important factor affec ting coke deposition for low
`and moderate turbul ent Reynolds numbers. Increasing turbul ence
`enhances heat transfer but also increases autoxidative coke depo(cid:173)
`sition as a result of improved mass transfer. Consequentl y, in(cid:173)
`creasing flow rate or space velocit y usually increases deposit
`buildup in the autoxidative regime , while enlarging the flow
`passage (surface -to-volume ratio ) reduces deposit buildup.
`• Residence time is a less important factor in high-temperature
`heat exc hangers where the coke depo sitio