Evolution of Surface Deposits on a High-Pressure Turbine ...tom/gas_turbines/BYU_Deposition...With...

I

at�rt�aepibdsata

tacucsol

tmcS9

J

Downlo

Jeffrey P. Bons

James E. Wammack

Jared Crosby

Daniel Fletcher

Department of Mechanical Engineering,Brigham Young University,

Provo, UT 84602

Thomas H. FletcherDepartment of Chemical Engineering,

Brigham Young University,Provo, UT 84602

Evolution of Surface Deposits ona High-Pressure TurbineBlade—Part II: Convective HeatTransferA thermal barrier coating (TBC)-coated turbine blade coupon was exposed to successivedeposition in an accelerated deposition facility simulating flow conditions at the inlet toa first stage high pressure turbine (T=1150°C, M =0.31). The combustor exit flow wasseeded with dust particulate that would typically be ingested by a large utility powerplant. The turbine coupon was subjected to four successive 2 h deposition tests. Theparticulate loading was scaled to simulate 0.02 parts per million weight (ppmw) ofparticulate over 3 months of continuous gas turbine operation for each 2 h laboratorysimulation (for a cumulative 1 year of operation). Three-dimensional maps of thedeposit-roughened surfaces were created between each test, representing a total of fourmeasurements evenly spaced through the lifecycle of a turbine blade surface. From thesemeasurements, scaled models were produced for testing in a low-speed wind tunnel witha turbulent, zero pressure gradient boundary layer at Re=750,000. The average surfaceheat transfer coefficient was measured using a transient surface temperature measure-ment technique. Stanton number increases initially with deposition but then levels off asthe surface becomes less peaked. Subsequent deposition exposure then produces a secondincrease in St. Surface maps of St highlight the local influence of deposit peaks withregard to heat transfer. �DOI: 10.1115/1.2752183�

Keywords: deposition, roughness, turbines

ntroductionLand-based and aircraft gas turbines ingest large quantities of

ir during operation. Contaminants found in the atmosphere or inhe fuel are known to generate deposits on the blade surfaces1,2�. Once formed, these deposits roughen the blade surfacesesulting in an increase in skin friction and in the convective heatransfer rate between the exhaust gases and the turbine blades3,4�. Deposits can also clog vital coolant passages, thus furtherggravating the thermal load of the turbine components �5�. Theffects of elevated levels of surface roughness on turbomachineryerformance have been studied for over 25 years. These studiesnclude fundamental flat-plate wind tunnel research �6�, multi-lade cascade facilities �7�, and full system level tests �8� in ad-ition to numerous computational studies �9�. These studies allupport the expected result that roughness increases surface dragnd heat transfer �though to varying degrees�. For turbomachinery,his translates to higher heat loads, accelerated part degradation,nd lower stage efficiencies.

Unfortunately, because this deposition process takes place overhousands of hours �for a land-based turbine�, little is knownbout the evolution of deposition on turbine blade surfaces. Aompanion paper �10� reports on a combustor simulator that wassed to evolve deposits on a turbine coupon with thermal barrieroating �TBC�. This was done in the Turbine Accelerated Depo-ition Facility �TADF� which is a specialized combustor capablef creating deposits at a vastly accelerated rate and under control-able conditions. The facility is described in detail in Ref. 11, but

Contributed by the International Gas Turbine Institute of ASME for publication inhe JOURNAL OF TURBOMACHINERY. Manuscript received September 8, 2006; final

anuscript received November 15, 2006; published online March 25, 2008. Reviewonducted by David Wisler. Paper presented at the ASME Turbo Expo 2006: Land,ea and Air �GT2006�, Barcelona, Spain, May 8–11, 2006, Paper No. GT2006-
1257.
ournal of Turbomachinery Copyright © 20

aded 23 Apr 2008 to 128.187.0.164. Redistribution subject to ASME

the essential principle is that of matching the product of the par-ticle concentration in the air flow �measured in parts per millionweight, ppmw� and the number of hours of operation. Depositswere generated at constant operating conditions over four succes-sive 2 h tests, simulating 3 months of gas turbine operation foreach test. The companion paper Part I �10� reviewed the evolutionof surface roughness character with successive deposition for thethree most common material systems found on modern gas tur-bine airfoils: uncoated superalloy, thermal barrier coating �TBC�,and oxidation resistant coating. A significant observation from thisstudy was that deposit roughness size �Rz� and shape �� f� expe-rienced a temporary lull in growth during deposit evolution for theconditions studied. This occurred as deposits filled the isolatedpeaks created during initial operation. Because of the well-documented link between roughness size and surface heat transfer,the objective of this study is to explore the implications of thissurface evolution on the heat load to the turbine.

Roughness ModelsOf the three surface treatments included in the Part I study, the

TBC1 coupon was selected for this convective heat transfer studydue to its relevance to modern gas turbines. After each 2 hourdeposition test �denoted as “burns” in what follows�, the surfacetopology of the coupon deposit was measured using a HommelInc. T8000 profilometer equipped with a TKU600 stylus. The sty-lus tip was conical in shape and had a 5 �m tip radius. Three-dimensional surface representations were constructed from a se-ries of two-dimensional traces separated by 10 �m. Data pointswithin each trace were also separated by 10 �m, yielding a squaredata grid. Three-dimensional surface representations and rough-ness statistics were produced with the Hommelwerke HommelMap software. The statistics of most interest for this study were
the centerline averaged roughness, Ra, the average roughness
APRIL 2008, Vol. 130 / 021021-108 by ASME

license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

hc

tt1tctaz

w

Ts

tttttt3edtr�RTbBtdaBFt

spoRWasoim

2smoTptc1

0

Downlo

eight, Rz, and the average forward-facing surface angle, � f. Theenterline averaged roughness was calculated using the following

Ra =1

IJ�i=0

I−1

�j=0

J−1

�zi,j� �1�

Rz was calculated as the mean of the vertical distance betweenhe highest peak and deepest valley every 10 mm of a 2D surfacerace. Based on the scaled roughness model images shown in Fig., this appeared to be a representative roughness dimension forhe deposits in this study. The average forward-facing angle wasalculated using the methodology proposed by Bons �12� in whichhe surface is traversed in the flow direction and forward-facingngles are averaged with all leeward-facing angles set equal toero �Eq. �2��

�̄ f =1

I �i=0

I−1

�i �2�

here, if �zi+1−zi��0

�i = tan−1� zi+1 − zi

yi+1 − yi� else, �i = 0

ypically, Rz and � f are calculated as the average of multiple 2Durface traces.

The TBC1 coupon experienced continuous deposit growth overhe entire four-burn sequence in the TADF. In addition, during thehird and fourth burns, significant TBC spallation occurred nearhe coupon edges. Since spallation roughness was not the focus ofhis study, scaled roughness models were made from a region athe coupon center, where spallation was minimal. Figure 1 con-ains a series of 3D topologies showing the deposit evolution on a.3 mm�4.2 mm region away from the spallation at the coupondges. Burns 2–4 appear to fill the valleys between the initialeposit peaks created during the first period of exposure. Thisrend is even more evident when reviewing the evolution ofoughness statistics for the scaled models of the TBC1 couponFig. 2�. As outlined previously, the roughness height �i.e., Ra andz� on the TBC1 coupon increased markedly during the first burn.his was then followed by a gradual increase over the next twourns. Finally, the roughness height again increased rapidly withurn 4. The average forward facing angle followed a similar

rend. The initial rise in � f with Burn 1 was followed by a slightecline with Burn 2 and a leveling off with Burn 3. Finally, theverage forward facing angle climbed back above 7 deg withurn 4. It is noteworthy that the levels of Ra, Rz, and � f shown inig. 2 are comparable to those reported for measurements on ac-

ual turbine hardware �3,4�, as discussed in Part I.The topological measurements from the Hommel profilometer

hown in Fig. 1 were used to create the CNC-compatible code androduce the scaled models. Model scaling was guided by the ratiof roughness height to boundary layer momentum thickness,z /�. This ratio was maintained near a value of 0.5 �4,13,14�.ith this as guidance, a scaling factor of 20 was determined to be

ppropriate. Because of this scaling factor, and due to the limitedize of the largest usable rectangular region of roughness, theriginal data formed 25% of the total model surface. The remain-ng 75% of the 22 cm�38 cm wind tunnel space provided for the

odel was filled by mirroring this section of data.With the CNC code, the roughness models were cut out of

.54-cm-thick Plexiglas acrylic sheets using a modified counter-ink. The countersink was chosen because it most closely approxi-ated the tip of the profilometer stylus that scanned the surface

riginally. Both are conical in shape with a 90 deg included angle.his was to ensure that the model contours would be as close asossible to the original surface measurements. A comparison ofhe scaled model surfaces to the original coupon roughness indi-ated that while the roughness height was well matched �within
0%�, the angle was not. The countersink tool that was used, with
21021-2 / Vol. 130, APRIL 2008


a 0.5 mm tip radius, flattened out the transitions in the surfaceproducing a 25% drop in mean forward facing angle. These ad-justments are reflected in the roughness statistics for the scaledmodel shown in Fig. 2.

Experimental FacilityA schematic of the research facility used for the experiments is

Fig. 1 Surface topologies of 3.3 mmÃ4.2 mm from surfacedeposit on TBC1 coupon after each burn. Vertical scale showspeak roughness height.

shown in Fig. 3. The open loop wind tunnel uses a main flow

Transactions of the ASME


bsbmrlcstc�T

bbfFffslbiTretddbCama

J

Downlo

lower to provide a nominal mass flow of 1.8 kg /s to the testection. A heat exchanger at the main flow blower discharge cane used to vary the flow temperature from 20°C to 50°C. Theain flow enters a 0.6-m-diameter conditioning plenum before

eaching the square test section. This conditioning plenum has oneayer of perforated aluminum plate followed by 7.6 cm of honey-omb straightener, and five layers of fine screen. A circular-to-quare foam nozzle conducts the flow from the plenum cross sec-ion to the 0.38 m by 0.38 m acrylic test section. With thisonditioning, 2D flow uniformity of �0.4% in velocity and1°C is obtained over the center 0.18 m of the test section span.he freestream turbulence level in the wind tunnel is 0.5%.At 1.52 m from the plenum exit a knife-edge boundary layer

leed with suction removes the bottom 2.7 cm of the growingoundary layer. The top wall of this final section pivots about itsorward end in order to adjust the pressure gradient in the tunnel.or the tests presented here, the wall was adjusted to produce zeroreestream acceleration over the roughness test panels. At 2.5 cmrom the boundary layer suction point, a 2-mm-diameter cylinderpans the test section to trip the boundary layer to turbulent. Theeading edge of the roughness panel is located 1.04 m from theoundary layer suction point. The roughness panels are installedn a 0.22 m streamwise gap in the lower wall as shown in Fig. 1.he tunnel then continues 0.3 m beyond the trailing edge of the

oughness panels. In this configuration, the flow in the tunnelxperiences a transition from a smooth to rough wall condition athe leading edge of the roughness panels. This experimental setupeparts from traditional roughness experiments in which the entireevelopment length of the boundary layer is roughened. Studiesy Antonia and Luxton �15� and more recently by Taylor andhakroun �16� show that this smooth to rough transition results inn initial overshoot in cf and St followed by a fairly rapid adjust-ent to the appropriate rough-wall values. Both researchers report

n adjustment length equivalent to 3–4 boundary layer thicknesses

Fig. 2 Roughness statistics for the scaled model surface

Fig. 3 Wind Tunnel Facility

ournal of Turbomachinery


with up to 20% initial overshoot. To mitigate the effect of thistransition region, the heat transfer data were taken on the down-stream half of the roughness section �beyond the expected adjust-ment length of approximately 9 cm�.

As additional verification of this testing procedure, heat transferdata were also acquired with multiple upstream roughness panelsextending to 1 m upstream of the measurement point, and the datashowed no perceptible effect beyond the uncertainty of the mea-surement. Thus, all data were acquired with the smooth to roughtransition at 1.04 m from the tunnel boundary layer bleed. Themean elevation of the roughness panel was aligned vertically withthe upstream smooth panel to prevent acceleration or decelerationof the bulk flow �U��.

Flow velocity was measured using a pitot-static probe with aco-located flow thermocouple with 0.13 mm bead diameter forflow temperature measurement. The two instruments are posi-tioned at midspan just outside the boundary layer and downstreamof the roughness panel during the transient testing. A second ther-mocouple was located upstream of the test panel and further awayfrom the wall in the freestream, as shown in Fig. 3. Uncertainty inthe velocity measurement was within �1.5% at flow rates of in-terest.

A boundary layer pitot probe with a co-located 0.13 mm beaddiameter flow thermocouple was used to determine the boundarylayer velocity and thermal profiles at the St measurement location.With the smooth test panel installed at the test section, the velocityboundary layer thickness was approximately 22 mm, with mo-mentum and displacement thicknesses of 2.7 mm and 3.6 mm,respectively �shape factor=1.36�. The thermal boundary layershape varies during the St measurement due to the transient test-ing procedure and nonconstant wall temperature. However, thethermal layer thickness remains roughly equivalent to the velocityboundary layer thickness �approximately 2 cm�.

St Measurement. For the heat transfer measurements, a FLIRThermacam SC 3000 infrared �IR� camera system is mountedwith the lens fit into a hole in the acrylic ceiling of the tunnel. Thecamera has a sensitivity of 0.03°C �at 30°C� and allows framingrates of approximately 1 Hz. The IR camera was focused on a67 mm �crossstream��83 mm �streamwise� field of view. Thislimited field of view was centered at a distance of L=1.21 m fromthe leading edge of the tunnel floor. This puts the streamwiseposition of the heat transfer measurement roughly 6 cm down-stream of the center of the roughness panel. The 240�320 pixels of the FLIR camera allowed excellent resolution ofthe roughness features during the transient experiments.

The Stanton number was determined from this surface tempera-ture history using a three-dimensional unsteady conduction finite-volume algorithm with time-varying boundary conditions. Thetime-varying surface heat flux is calculated from the surface nor-mal temperature gradient in the solid. This technique assumes thepanels are a semi-infinite solid at constant temperature at time t=0. To accomplish this, the entire test section was soaked at roomtemperature for several hours before testing. Using the main flowheat exchanger, a hot-gas air flow �50°C� was then initiated in-stantaneously while monitoring the freestream velocity and tem-perature, as well as the average surface temperature �with the IRcamera�.

Radiative heat flux from the test plate to the surrounding tunnelwalls was always less than 1% of the calculated convective heatflux. The thermophysical properties, thermal conductivity �� andthermal diffusivity ��=� /cp�, for the plastic panels were deter-mined experimentally by Thermal Properties Research Labora-tory. The measurements yielded the following values: �=0.196 W /m K�6% and cp=1330 J /kg K�3%. The plasticdensity is 1188 kg /m3�2%.

During testing the 25-mm-thick acrylic roughness panel wasmounted on a second 25-mm-thick smooth acrylic panel with the
same material properties. A thermocouple sandwiched between
APRIL 2008, Vol. 130 / 021021-3


t5onlsifowTtftwtcvbtrni6fwt

R

ispfptsnr�hchsHwnwr

Ff

0

Downlo

he two panels indicated no change in temperature during the firstmin for the typical test case. Due to the low thermal diffusivity

f the acrylic plates, test times shorter than 5 min yield a Fourrierumber �Fo=�t /b2� less than 1 /16, which is the semi-infiniteimit. This confirmed the use of the semi-infinite conduction as-umption in the data processing. Also, it was discovered that thenfrared measurement is sensitive to the temperature of the sur-aces surrounding the roughness panels. This occurs because somef the radiation that is incident on the camera originates from theind tunnel enclosure and is reflected off the roughness panel.he magnitude of this component of radiation changes as a func-

ion of the tunnel wall temperature. The FLIR software accountsor this by allowing the user to specify the ambient enclosureemperature. Since the heat transfer test was transient, this inputas adjusted in post-processing to track the tunnel wall tempera-

ure as a function of time. Six 50 �m bead diameter thermo-ouples were flush mounted to each of the acrylic test panels toerify the IR surface temperature measurement. This allowed cali-ration of the camera to within �0.3°C of the initial surfaceemperature and corroborated the St calculation procedure for theough panels relative to the smooth panel. The area-average Stumbers presented in this paper were calculated by averaging thenitial 250 s of the plate’s transient response over the full7 mm�83 mm camera viewfield. The smooth plate St value wasound to be within 1% of a standard correlation. Repeatability wasithin �3% and bias uncertainty was estimated at �0.00015 for

he smooth plate measurement of St0=0.00228 at ReL=750,000.

esults and DiscussionHeat transfer measurements were made at constant flow veloc-

ty �ReL=750,000� for each of the four deposit models and amooth baseline panel. Figure 4 shows the Stanton number com-uted using the area-averaged surface temperature as obtainedrom the IR camera measurement described above. Each dataoint in Fig. 4 represents the average of at least three separateransient tests, usually conducted on different days. Error barshow the range of measurements for each data point. The Stantonumber values are presented as a percent difference between theough surface value �StR� and the smooth surface reference valueSt0�. The smooth reference panel used in the wind tunnel studyad a centerline average roughness of Ra=0.5 �m. This is signifi-antly lower than the roughness level of Ra=12 �m that wouldave been obtained if the preburn polished TBC surface had beencaled by the same factor of 20 used for the roughness panels.owever, the roughness Reynolds number �Reks� for both cases isell below the aerodynamically smooth limit for these wind tun-el conditions. Thus, the percent change in St would be similarith either reference. Also shown in the figure are the model

ig. 4 Stanton number augmentation and roughness statisticsor the scaled roughness models

oughness statistics from Fig. 2 for comparison. These statistics

21021-4 / Vol. 130, APRIL 2008


are computed for the portion of the roughness model surface inthe direct field of view of the IR camera and may differ somewhatfrom those presented in Part I.

The trend in Stanton number follows closely the trend in rough-ness statistics. The St augmentation levels off between Burns 2and 3 followed by a marked rise with Burn 4 as the roughnessheight and mean angle experience a resurgence. Data accumulatedby Bons �12� using roughness characterizations obtained from ser-viced turbine components indicate that the effect of roughness oncf and St is reduced as the average forward facing angle decreasesfor the same mean roughness height �Rz�. Bons proposed a corre-lation for the dependency of ks /k on � f.

ks

k= 0.0191�̄ f

2 + 0.0736�̄ f �3�

Though the skin friction coefficient was not measured for theroughness models, an empirical correlation proposed by Schlich-ting was used to estimate its value �17�

cf = �2.87 + 1.58 log�L/ks��−2.5 �4�This was combined with a correlation for Reynolds analogy �RA�factor �RA=2St /cf� proposed by Bons for a turbulent boundarylayer over rough surfaces �12� to obtain an empirical estimate forSt

RA

RA0= 0.5�1 + exp�− 0.9

ks

�� 5�

The value for RA0, the smooth wall value of Reynolds analogy,for a turbulent boundary layer was 1.2.

The results of this empirically based St correlation �Eqs.�3�–�5�� are shown with the experimental data in Fig. 5. The pre-dictions are higher than the experiment in all cases. A differentcorrelation based on the roughness Reynolds number �Reks� byDipprey and Sabersky �18� is also shown for comparison

St =0.5cf

1 + 0.5cf�5.19 Reks0.2 Pr0.44 − 8.5�

�6�

Both models capture the general trend of increasing roughnesswith repeated exposure, although Eq. �6� shows better overallagreement.

The spatial resolution of the infrared camera combined with the3D finite-volume analysis technique allows a detailed evaluationof the heat transfer around specific roughness features in additionto the area-averaged St measurements presented above. The un-steady conduction algorithm properly accounts for lateral conduc-tion within the three-dimensional roughness features on the scaledmodel. Table 1 contains the minimum, average, maximum, and

Fig. 5 Comparison of experimental St augmentation with em-pirical prediction

standard deviation of St for each of the roughness surfaces. The



v8F

oerwtabtrhmtttpiwpobradtstaptmhs

C

assl3rtorf

B

1234

J

Downlo

alues in the table represent only a 20 s time average from0 to 100 s after flow initiation, whereas the area-averaged data inigs. 4 and 5 are averaged over the full 250 s test time.In all four cases, there is a substantial variation in heat transfer

ver the rough surfaces. In general, roughness peaks can experi-nce 2–3 times higher heat transfer than the area average. Theseesults are consistent with measurements from Henry et al. �19�ho reported up to a factor of 2.5 heat transfer augmentation at

he leading edge of spherical protuberances in a turbulent bound-ry layer. This effect is partially due to the energetic interactionetween the flow and the roughness peak as the flow accelerateso navigate around the obstruction. In addition, roughness peakseach further from the wall into the thermal boundary layer whereigher temperature fluid �on average� is present; thus, the elevatedeasurements in these regions. This is actually welcome news for

urbine thermal load management. Since a substantial portion ofhe heat transfer augmentation occurs at these elevated peaks ofhe deposit, the path to the metal substrate �through the depositeak� is longer, resulting in a greater thermal resistance. Thus, thencrease in local metal temperature beneath the deposit �and TBC�ill be less severe. Unfortunately, flow separation behind the de-osit roughness peaks generates highly turbulent flow in the wakef the protuberances. This is shown in Fig. 6 which contains side-y-side surface height and St contour plots for the region nearoughness peaks on the Burn 1 and 2 models. Though not as highs the heat transfer at the roughness peak, the wake regions �in-icated by circles in Figs. 6�a� and 6�d�� still experience heatransfer augmentation that is 75% higher than the surroundingurface when compared to the smooth surface heat transfer. And ifhe deposit is thinner in this wake-affected region, the lack ofdded insulation from the deposit could result in a local hot spotenetrating deep into the TBC layer. Finally, the standard devia-ion data in Table 1 show a trend similar to that exhibited by the

ean forward-facing angle in Fig. 4. As might be expected, theeat transfer variations become more pronounced as the depositurface becomes more peaked.

onclusionsSuccessive deposits were generated on a TBC coupon in an

ccelerated test facility at a gas temperature and velocity repre-entative of first stage high-pressure turbines. The coupon wasubjected to four successive 2 h deposition tests. The particulateoading was scaled to simulate 0.02 ppmw of particulate over

months of continuous gas turbine operation for each 2 h labo-atory simulation �for a cumulative 1 year of operation�. Convec-ive heat transfer measurements were made using scaled modelsf the deposit roughness after each exposure period. Based on theesults presented in this study, the following conclusions are of-ered:

�1� For the conditions studied, deposit roughness size �Rz� andshape �� f� experienced a temporary lull in growth duringthe deposit evolution. This produced a comparable plateauin measured heat transfer coefficient �St� as deposits filledthe valleys between isolated peaks created during initialoperation. The implication is that heat load managementmay be most severe during the initial phase of deposition,

Table 1 Spatial statistics for heat transfer coefficient

urn Min St Avg St Max StStd dev St

�%�

0.00078 0.00273 0.00794 20.10.00138 0.00294 0.00637 19.40.00105 0.00292 0.00675 23.10.00001 0.00309 0.00797 27.7

when the deposit peaks are distributed over the relatively



smooth surface. The subsequent valley filling reduces theroughness signature as well as provides additional insula-tion to the blade. Existing Stanton number correlationsbased on various roughness statistics captured the trend inthe experimental data; and

�2� Surface roughness features can create local St variations ofup to a factor of three from the area average. Isolated peaksresult in substantially elevated levels of convective heattransfer in the wake directly downstream of theprotuberance.

AcknowledgmentVarious individuals provided invaluable support to this research

effort. The authors would particularly like to thank Dr. Tom Taylorof Praxair Surface Technologies and Gerry McQuiggan from Si-emens who generously donated coupon specimens for this study.Thanks also to Ken Forster, Kevin Cole, and Spencer Grange forperforming a myriad of helpful tasks. This work was sponsored bya generous grant from the BYU Mechanical Engineering Depart-ment and by the US Department of Energy—National EnergyTechnology Laboratory through a cooperative agreement with theSouth Carolina Institute for Energy Studies at Clemson Univer-sity. The views expressed in this paper are those of the authors anddo not reflect the official policy or position of the Department ofEnergy or U.S. Government.

NomenclatureI total number of y direction surface

measurementsJ total number of x direction surface

measurementsL wind tunnel length from bleed to measurement

location �1.21 m�M Mach numberPr Prandtl number

RA Reynolds analogy factor �2St /cf�Ra centerline averaged roughness �mm� �Eq. �1��

ReL flow Reynolds number U�L /�Reks roughness Reynolds number u�ks /�

Rt maximum peak-to-valley roughness �mm�Rz k=mean peak-to-valley roughness �mm�St Stanton number �h /cpUe�T flow temperature �°C�

U� wind tunnel velocity �13 m /s�b plate thickness �25 mm�cf skin friction coefficientcp specific heat

i y direction indexing variablej x direction indexing variablek average roughness height �Rz�

ks equivalent sandgrain roughnesst timex surface dimension perpendicular to the gas

streamy surface dimension parallel to the gas streamz height of an individual roughness element� thermal diffusivity

� f average forward-facing angle �Eq. �2�� thermal conductivity� kinematic viscosity� boundary layer momentum thickness density

Subscripts0 flat plate baselineR roughness
s surface
APRIL 2008, Vol. 130 / 021021-5


R

0

Downlo

� freestream

eferences�1� Borom, M. P., Johnson, C. A., and Peluso, L. A., 1996, “Role of Environmen-

tal Deposits and Operating Surface Temperature in Spallation of Air PlasmaSprayed Thermal Barrier Coatings,” Surf. Coat. Technol., 86–87, pp. 116–126.

�2� Wenglarz, R. A., and Fox, R. G., Jr., 1990, “Physical Aspects of DepositionFrom Coal-Water Fuels Under Gas Turbine Conditions,” J. Eng. Gas TurbinesPower, 120, pp. 9–14.

�3� Tarada, F., and Suzuki, M., 1993, “External Heat Transfer Enhancement toTurbine Blading due to Surface Roughness,” ASME Paper No. 93-GT-74.

�4� Bons, J. P., 2002, “St and Cf Augmentation for Real Turbine Roughness withElevated Freestream Turbulence,” ASME J. Turbomach., 124, pp. 632–644.

�5� Kim, J., Dunn, M. G., and Baran, A. J., et al., 1993, “Deposition of VolcanicMaterials in the Hot Sections of Two Gas Turbine Engines,” J. Eng. GasTurbines Power, 115, pp. 641–651.

Fig. 6 Heat transfer coefficient and surface heightroughness models. „Flow direction is from top to botare circled: „a… Burn 1 surface height contour plot „mmcontour plot „mm…; and „d… Burn 2 St contour map.

�6� Acharya, M., Bornstein, J., and Escudier, M., 1986, “Turbulent Boundary Lay-

21021-6 / Vol. 130, APRIL 2008


ers on Rough Surfaces,” Exp. Fluids, 4, pp. 33–47.�7� Hoffs, A., Drost, U., and Bolcs, A., 1996, “Heat Transfer Measurements on a

Turbine Airfoil at Various Reynolds Numbers and Turbulence Intensities In-cluding Effects of Surface Roughness,” ASME Paper No. 96-GT-169.

�8� Blair, M. F., 1994, “An Experimental Study of Heat Transfer in a Large-ScaleTurbine Rotor Passage,” J. Turbomach., 116�1�, pp. 1–13.

�9� Guo, S. M., Jones, T. V., Lock, G. D., and Dancer, S. N., 1998, “Computa-tional Prediction of Heat Transfer to Gas Turbine Nozzle Guide Vanes WithRoughened Surfaces,” J. Turbomach., 120�2�, pp. 343–350.

�10� Wammack, J. E., Crosby, J., Fletcher, D., Bons, J. P., and Fletcher, T. H., 2008,“Evolution of Surface Deposits on a High-Pressure Turbine Blade—Part I:Physical Characteristics,” ASME J. Turbomach., 130�2�, p. 021020.

�11� Jensen, J. W., Squire, S. W., and Bons, J. P., 2005, “Simulated Land-BasedTurbine Deposits Generated in an Accelerated Deposition Facility,” ASME J.Turbomach., 127, pp. 462–470.

�12� Bons, J. P., 2005, “A Critical Assessment of Reynolds Analogy for TurbineFlows,” ASME J. Heat Transfer, 127, pp. 472–485.

�13� Bogard, D. G., Schmidt, D. L., and Tabbita, M., 1998, “Characterization and

ts for a 67 mmÃ83 mm region on the Burn 1 and 2as indicated.… Wake regions with elevated St values„b… Burn 1 St contour map; „c… Burn 2 surface height

plotom…;

Laboratory Simulation of Turbine Airfoil Surface Roughness and Associated



J

Downlo

Heat Transfer,” J. Turbomach., 120�2�, pp. 337–342.�14� Barlow, D. N., and Kim, Y. W., 1995, “Effect of Surface Roughness on Local

Heat Transfer and Film Cooling Effectiveness,” ASME Paper No. 95-GT-14.�15� Antonia, R. A., and Luxton, R. E., 1971, “The Response of a Turbulent Bound-

ary Layer to a Step Change in Surface Roughness. Part 1: Smooth to Rough,”J. Fluid Mech., 48, pp. 721–726.

�16� Taylor, R. P., and Chakroun, W. M., 1992, “Heat Transfer in the TurbulentBoundary Layer with a Short Strip of Surface Roughness,” AIAA Paper No.92-0249.



�17� Schlichting, H., 1979, Boundary Layer Theory, 7th ed., McGraw-Hill, NewYork.

�18� Dipprey, D. F., and Sabersky, R. H., 1962, “Heat and Momentum Transfer inSmooth and Rough Tubes at Various Prandtl Numbers,” Int. J. Heat MassTransfer, 6, pp. 329–353.

�19� Henry, R. C., Hansman, R. J., and Breuer, K. S., 1995, “Heat Transfer Varia-tion on Protuberances and Surface Roughness Elements,” J. Thermophys. HeatTransfer, 9�1�, pp. 175–180.

APRIL 2008, Vol. 130 / 021021-7


Evolution of Surface Deposits on a High-Pressure Turbine ...tom/gas_turbines/BYU_Deposition...With...

Documents

Transcript of Evolution of Surface Deposits on a High-Pressure Turbine ...tom/gas_turbines/BYU_Deposition...With...