NASA TECHNICALM E MORANDUM

NASA TM X-71607

X

5<

<

/(NASA-TM-X-71607) SEALING TECHNOLOGY FORAIRCRAFT GAS TURBINE ENGINES (NASA)15 p HC «••» CSCL 21E

N7U-33229

OnclasG3/28 48758 J

SEALING TECHNOLOGY FOR AIRCRAFT

GAS TURBINE ENGINES

by L. P. Ludwig and R. L. JohnsonLewis Research CenterCleveland, Ohio

TECHNICAL PAPER proposed for presentation atTenth Propulsion Conference sponsored byAmerican Institute of Aeronautics and Astronauticsand Society of Automotive EngineersSan Diego, California, October 21-24, 1974

https://ntrs.nasa.gov/search.jsp?R=19740025116 2018-04-18T10:19:02+00:00Z

SEALING TECHNOLOGY FOR AIRCRAFT GAS TURBINE ENGINES

* **L. P. Ludwig and R. L. Johnson

National Aeronautics and Space AdministrationLewis Research Center

Cleveland, Ohio

REPRODUCIBILITY OF THEPAGE IS POOR

Abstract

Experimental evaluation under simulated engineconditions revealed that conventional mainshaftseals have disadvantages of high gas leakage ratesand wear. An advanced seal concept, the self-acting face seal, has a much lower gas leakage rateand greater pressure and speed capability. In en-durance tests (150 hr) to 43 200 rpn the self-acting seal wear was not measurable, indicatingnoncontact sealing operation was maintained even atthis high rotative speed.

A review of published data revealed that theleakage through gas path seals has a significanteffect on TSFC, stall margin and engine mainte-nance. Reducing leakages by reducing seal clear-ances results in rubbing contact, and then the sealthermal response and wear determines the final sealclearances. The control of clearances requires amaterial with the proper combination of rub toler-ance (abradability) and erosion resistance. In-creased rub tolerance is usually gained at the ex-pense of reduced erosion resistance and vice versa.

I. Introduction

A multiplicity of seals are used in gas tur-bines to restrict gas leakage, provide thrust bal-ancing, maintain thermal gradients, meter coolinggas flow, and protect bearing sumps. Since a gasturbine has many sealing locations (blade tips,interstage, mainshaft, etc.), the accumulative ef-fect of deficient sealing practice can be appreci-able. In fact, operating experience reveals thatsealing practice has a significant impact in threeareas: engine performance, stall margin and gen-eral engine maintainability.

The effect of sealing on performance has re-ceived some documentation. Reference 1 calculatesa 21 percent power loss attributable to gas sealleakages (mainshaft, gas path, flanges, etc.) in asmall gas turbine engine. Reference 2 gives a de-tailed review of the effect of seal clearance onperformance of transport and fighter engines.This study indicated that a 2i percent change inTSFC can be expected for a nominal change in sealclearances.

There is considerable evidence that stallmargin can be affected by incorporating variousgeometries (e.g., grooves) into the blade tipshrouds;(3) also, the tip clearance magnitude af-fects stall margin.

A study on reliability and maintainability ofseven small military engines now in use, showsthat one of the main causes of unscheduled engineremoval is oil leakage through carbon mainshaftseals.W In fact, it ranks as the third most

common cause (foreign object damage and impropermaintenance being first and second) out of a list of28 failure modes; and in reference 4 it is statedthat "many in the engine design, management, andprocurement communities tend to minimize the sealproblem and correspondingly downplay the value ofimprovement in this area."

Dirt ingestion through high leakage seals atthe bearing sumps degrades bearing life. Otherproblems which can occur as a result of high airflowinto the bearing lubrication system are increasedgearbox pressure (tends to lower seal life) and in-creased 'bearing cavity pressure which may limit oilflow into the bearing compartment and thereby affectbearing life.

This paper will consider the state-of-the-artof two general types of seals in a gas turbine: themainshaft seals and the gas path seals (e.g., com-pressor blade tip). A comparison of the wear andgas leakage rates obtained for three conventionalmainshaft seals (ring, circumferential and faceseals for small engines) and for an advanced seal,the self-acting seal, is made. The effect of leak-age through gas path seals is reviewed and effect ofrub interactions discussed. Experimental data (wearand erosion) are presented for abradable turbineshroud materials.

II. Apparatus and Test Seal Description

Mainshaft Seal Rig

The rig bearing compartment is typical of smallhigh-speed gas turbine sumps. Sealing positionswere located forward and aft of the bearing, whichenabled simultaneous testing of two seal assemblies.

The bearing compartment drains by gravity intoa static air-oil separator. Desired air pressureis introduced into the cavities adjacent to the testseals, and the air that leaks past the two testseals is passed through a flowmeter downstream ofthe air-oil separator to obtain a measure of sealleakage.

Conventional Mainshaft Seals

Head, Seals Section.**Chief, Lubrication Branch.

Mainshaft seals are used in gas turbine enginesto restrict gas leakage into the bearing compart-ments (sumps). The air leaking through the sealprevents oil leakage out. In some engines the gaspressure and temperature are relatively low, so thata single labyrinth seal to restrict gas leakage intothe sump is adequate (Fig. 1). At low pressures,the efficiency loss due to seal leakage is not sig-nificant. However, a major disadvantage of thelabyrinth seal, as compared to the rubbing contactseals (circumferential and face), is that thehigher leakage requires corresponding larger duct-ing requirements which results in easier passage ofairborne water and dirt into the sump. It shouldbe emphasized that dirt is very detrimental tobearing life. In addition, for labyrinth seals,

reverse pressure drops must be avoided to precludehigh oil loss. Further, high temperature com-pressor air leaking into oil sumps can lead tosump fires.

As engine pressures increase, the high leakagerates associated with labyrinth seals start to be-come significant from a performance penalty andsump design standpoint. In small engines, theavailable space for the bearing sumps tends to bevery limited, hence the amount of gas leakage thatcan be handled in the sump is limited.

In large engines with high gas pressure dif-ferential (i.e., 241 N/cm2 (350 psi)), multiplelabyrinth seals (see schematic in Fig. 2) are usedto protect the bearing from the high temperature,high pressure, turbine cooling gas. These multiplelabyrinth seal systems take up considerable spaceand are generally too costly and difficult toaccommodate in small engines. Thus, multiple lab-yrinth seal systems have significant disadvantageswhen applied to small high pressure engines.

The floating ring seal can be considered to bea type of labyrinth seal, which, because the ringis not restrained from moving in a radial direc-tion, can be set up to operate with less leakagegap clearance than the conventional labyrinthseals.(5) The floating ring assembly (Fig. 3(a)),which is free to rotate in the seal case, is com-posed of a carbon ring shrunk into a steel retain-ing band. The retaining band is used to controlthe expansion rate of the composite ring and to re-inforce it against compressive and rotationalstress. The ring assembly is designed to have acoefficient of thermal expansion similar to that ofthe seal runner. Thus, the clearance between therunner outside diameter and the carbon inside diam-eter will be constant (independent of temperature).The ring assembly is not restrained from rotatingand in operation may rotate (at a speed less thanshaft speed) because of local and/or intermittentcontact with the shaft.(5)

Circumferential seals operate with very lowgas leakages and are therefore attractive, but thepressure differential capability for these pressureunbalanced seals is low (less than 41 N/cm2 (60psi)) because of rubbing contact. The circumfer-ential segmented seal (Fig. 3(b)) is a carbon ringconsisting of three 120° segments held together bya garter spring on the outside diameter. When thering is installed on the runner, the gaps betweenthe adjacent ends of the segments is a source ofair leakage into the bearing cavity. During oper-ation, when the carbon wears, the garter springforces the segments radially inward. When about0.127 mm (0.005 in.) of radial wear has occurred,the adjacent segment ends butt up and the sealthen operates as a close-clearance labyrinth.(5)Other circumferential seal designs (not evaluatedin this program) incorporate multiple rings oroverlapping joints to eliminate the leakage at thegaps between the segments. In addition pressurebalancing of the segments can result in a moderateincrease in the pressure and speed capability.(6)

The conventional face seal (Fig. 3(c)) con-sists of a rotating seat that is attached to theshaft and a nonrotating carbon sealing nose thatis free to move in an axial direction and thusaccommodate engine thermal expansion. The second-ary seal (piston ring) is subjected only to the

axial motion (no rotation) of the nose; a wavespring provides axial mechanical force to maintaincontact. The sealing nose is pressure balanced withan area ratio of 0.645 (for pressure balance defi-nition, see Ref. 7). Pressure balancing is alsoapplied to the secondary carbon seal ring, bothaxially and radially.

Self-Acting Seal

A self-acting face seal (Fig. 4) is similar toa conventional face seal except for the added fea-ture of a self-acting geometry, similar to that em-ployed in a gas lubricated thrust bearing. Whenthe shaft rotates, the sealing faces are separateda slight amount (in the range of 0.00025 to0.00127 cm (0.0001 to 0.0005 in.)) by the force gen-erated by this self-acting lift geometry. The pos-itive separation results from the balance of sealforces and the gas film stiffness of the self-acting geometry. Analyses of the self-acting sealconcept and experimental feasibility studies forlarge aircraft gas turbine engines have been de-tailed in several programs.C'~15)

The self-acting geometry can be any of thevarious types used in gas thrust bearings. In theseal shown in Fig. 4, the self-acting geometry con-sists of a series of shallow recesses (shroudedRayleigh steps) arranged circumferentlally insidethe sealing dam (see Fig. 4, section A-A). An im-portant point to note is that the lift pads arebounded at their inside diameter and outside diam-eter by the sealed pressure, Pj. (This is accom-plished by feed slots that communicate with theannular groove directly inside the sealing dam.)Therefore, a pressure gradient due to gas leakageoccurs only across the seal dam. Thus the effectsof force changes due to seal face deformation areminimized.(1QJ

Gas Path Seal Rigs

The gas path seal materials were evaluated inwear (abradability) and erosion rigs. The wear rigconsisted of a single rotating labyrinth knife edgeinto which a shroud specimen (60° arc) was moved ata predetermined rate in order to effect a rub pene-tration. The shroud specimen was heated to simulateengine operation.

In the erosion rig, a hot gas stream (Mach0.35) containing A1203 particles was directed atshroud specimens. Typical erosion tests were20 minutes in duration.

Materials

Four conventional type abradable materials andtwo experimental materials were evaluated in thewear and erosion rigs. The conventional type mate-rials were:

(1) 1/16 inch size honeycomb of Hastelloy Xmaterial (material A)

(2) Porous cermet with metal honeycomb support(material B)

(3) Porous metal with diatomaceous earth andhoneycomb support (material C)

(4) Sintered fiber metal of NiCrAlY (material D)

The experimental materials evaluated were:

(1) Sintered fiber metal of NiCrAlY with aplasma spray of nichrome/glass coating (ma- .terial E),

(2) Sintered fiber metal of NiCrAlY with aplasma spray of nichrome/glass/CaF2 coating(material F) .

III. Results and Discussion

Comparison of Mainshaft Seal Performance

Three conventional mainshaft seals (floatingring, circumferential segmented ring, and face) anda self-acting face seal were evaluated in simulatedgas turbine operation.

During high-speed testing (above 122 m/sec(400 ft/sec)) of the floating ring seal, substan-tial carbon wear occurred. (Wear was to be ex-pected at the 213 m/sec (700 ft/sec) point sincethe calculated operating gap closes to 0.0025 mm(0.0001 in.).) The results of the testing revealedthat the leakages were relatively high, and that itdoes no good to attempt to obtain lower leakages byreducing the initial clearances because the sealwears to a clearance value that is dictated byshaft runout and dynamic effects. In fact, sealsstarting with low initial radial clearances can endup with running clearances greater than the sealassembled with intermediate clearances; this isthought to be due to a thermal rub interaction(see section on gas path seals). The changes indiametral gap are shown in table I for two ringswith different initial diametral gaps. The tableshows the radial gaps after operating at slidingspeeds of 91, 122, 152, 183, and 213 m/sec (300,400, 500, 600, and 700 ft/sec). The ring with thesmallest initial gap (forward seal) 0.0610 mm(0.0024 in.) wore, in the first test, to a diam-etral gap of 0.135 mm (0.00530 in.). The aft sealdid not wear because the initial gap was largeenough to accommodate dynamic shaft motion.

It was found during the test program that thecircumferential segmented seals wore excessively athigh speeds and pressures, and eventually operatedas labyrinths when enough wear allowed the segmentsto butt together. At this point the amount of ad-ditional wear that takes place depends on runningspeed; the higher speeds cause more shaft radialmotion and therefore more wear.

Conventional face seal evaluation covered arange of speeds and air presures at ambient temper-atures. Typical conditions and resulting airflows, bearing cavity pressures, and seal tempera-tures are listed in table II. The seal operatedsuccessfully over a wide range of conditions, in-cluding 213 m/sec (700 ft/sec), with a pressuredifference of 117 N/cm2 (170 psi) . The total faceseal carbon nose wear for 20 hours was 0.0051 mm(0.0002 in.) on the forward seal and 0.0102 mm(0.0004 in.) on the aft seal. The fact that thetemperature did not exceed 394 K (250° F) at slid-ing speeds of 152 m/sec (500 ft/sec) suggests theseals were operating on an air film during much ofthe running time.

Table III contains the gas leakage data fortwo self-acting seals operating over a pressuredifferential range from 23 to 111 N/cm2 (34 to

REPRODUCIBILITY OF THORIGINAL PAGE IS POOR

161 psi) and a sliding speed range from 91 to183 m/sec (300 to 600 ft/sec). This is a rotativespeed range of 27 300 to 54 600 rpm. Neither theforward nor the aft carbon nose or seal seat showedany wear during this evaluation. Thus, the sealingsurfaces were separated by a gas film over the en-tire matrix of operating variables. This suggeststhat the gas bearing film stiffness was sufficientto prevent rubbing contact under the high inertiaforces that are associated with high rotativespeeds. (*- >

Note in table III that the seal leakage in-creases as the sliding speed increases (for anygiven pressure differential). This leakage increaseis due to a slight increase of the sealing gap be-cause the increased lift force produced by the liftpads (dynamic effect). As would be expected, theleakage increases as the pressure increases.

To further explore the operating limits of theself-acting seals, 150 hours of endurance operationat ambient temperature was conducted as follows:

Speed

a/sec ft/sec

Air pressuredifferential

N/cm2 psi

Time,hr

102122137145145

334400450475475

103103103103124

149.7149.7149.7149.7179.7

2822652015

Air temperature varied throughout the test butwas generally from 372 to 408 K (200° to 275° F).Inspection of the seals after the 150 hours revealedthat the wear to each seal was insignificant (lessthan 0.0012 mm (0.00005 in.)j.

A comparison of the gas leakage rates of thevarious seal configurations is shown in Fig. 5. Ingeneral, the plot shows that self-acting face sealhas the potential of significantly reducing leakageas compared to the conventional seals.

Of the conventional configurations, face sealsallowed the least air flow at high pressure differ-entials. Circumferential segmented seals are astight as face seals at moderate operating condi-tions; however, experience and the subject test pro-gram results have shown that at pressure differen-tials above 41.4 N/cm2 (60 psi) and speeds above107 m/sec (350 ft/sec), these (unbalanced) circum-ferential segmented seals rapidly wear out andfinally operate as labyrinths. In that case thereis little to choose between circumferential, rotat-ing ring, and labyrinth seals in terms of air flows.

To gain some perspective of the magnitude ofair flow under discussion, engine experience hasshown that excessive air flow into a bearing packageincorporating seals of the size used in the testprogram would be in the order of 0.012 kg/sec(0.029 Ib/sec). Taking midpoint values of the rangeof pressure differentials in Fig. 5, the face sealcould not meet this criterion at pressure differ-entials above approximately 85 N/cm2 (123 psi) ; andthe limiting pressure differential for circumferen-tial segmented seals (which wear rapidly), rotatingring seals, and simple labyrinths would be approxi-mately 40 N/cm2 (58 psi). The self-acting seal,however, did not reach the limiting leakage rate

and had a leakage of 0.0046 kg/sec (0.0102 Ib/sec)at a pressure differential of 107.6 N/cm^(156.0 psi). In general the self-acting seal hadabout one third the leakage of the conventionalface seal.

Gas Path Seals

Gas path sealing either in the primary or sec-ondary gas flow paths, is a complex engineeringproblem since these.seals operate in extreme andvarying temperature environments under large cen-trifugal stress conditions, high peripheral speeds,and must withstand occasional rubbing contact.These rotating and stationary seal elements can .move radially and axially with respect to one .another due to engine thermals and transient condi-.tions. However, in proposing improvements, itmust be kept in mind that the foremost requirementis reliability; that is, the seal system must notenter a self-destruct rub mode when hard rubs occurand adverse blade damage must not be produced.

Considering all the seals, the primary gaspath seals (Fig. 6) have the greatest impact onTSFC. Generally, the gas path sealing clearanceschange with each engine condition (idle, takeoff,climb, etc.); the support structure dimensionalchanges being large relative to the seal clear-ances. One portion of the design problem is thatof compensating for these relatively large dis-placements through design features and judicioususe of materials. The trend toward higher enginepressures and temperatures will tend to increasethese seal displacements.

Typically, a secondary gas path system con-tains flow restrictions in parallel and series.The functions of the secondary seals include cool-ing, maintaining proper thermal balance, pressuriz-ing, venting and thrust balancing. Because of thecavity interconnections across the seals, a prob-lem at one seal location can affect other parts ofthe system (such as changing the thrust bearingload). Changes in the turbine cooling flow are ofparticular concern since the trend is to use mate-rials at temperatures which result in the minimumacceptable low cycle thermal fatigue life. Thetotal seal clearance, of course, is a function ofdifferential thermal growth, wear, and erosion.The crux of the gas path seal (primary or secondaryflow) problem is to maintain control of clearancesthat need to be made as small as practical.

Effect of Clearance

Reference 1 points out that gas leakage ef-fects become more acute as engine size decreasesbecause the ratio of engine circumference to engineflow area increases. For example, the leakageanalysis reported in Ref. 1 indicates that theengine flange and labyrinth seal leakage is 2 per-cent of the mass flow for a 5 pound per secondengine, and only 0.3 percent for a 120 pound persecond engine. Reference 1 presents a detailedanalysis of the clearance effects in a hypotheticalengine with a 5 pound per second air flow. Theengine was designed for zero leakage and perform-ance calculated; performance was then determinedwith typical leakage values (based on seal tests)assigned to each seal position. The results areshown in Fig. 7. The most significant losses arein the labyrinth seals, and in particular the lab-yrinth seals at the discharge of the centrifugal

compressor and at the turbine bearing locations.The calculated overall effect (which includes lossesthrough flanges and vane pivots) is a 21 percentloss in power and a 10 percent increase in SFC.

In large axial- compressors, overall efficiencylosses of 2 to 6 percent have been indicated^) .andtypical data are shown in Fig. 8 where compressorefficiency is plotted as a function of clearance toblade span ratio.

In turbines the blade tip clearances tend to belarge because of significant thermal growths andthermal transients. Tip clearance to blade height'ratios of 1 to 4 percent are common. The efficiencyloss, however, depends on the degree of turbine re-action because the pressure difference across thestage has a direct effect on.the amount of leakagethrough the tip clearance. Experimental efficiencyeffects for reaction and impulse turbines have beenpublished;(16,17) these data are shown in Fig. 9.In general high reaction turbines show about a3 percent loss in efficiency for an, increase in tipclearance of 1 percent of blade height; for impulseturbines the loss is about 1.6 percent for an in-crease in clearance of 1 percent of blade height.

Effect of Shroud Geometry

Stall margin is affected by both clearance andseal shroud geometry (casing treatment). The term"casing treatment" has been used when holes, slots,grooves, etc., are added to the shroud surface inthe region over the blade tips. Many geometrieshave been investigated.(3,18) jn general, it hasbeen found that stall margin can be improved byadding casing treatment if stall inception is atthe outer wall. Reference 19 attributes the im-provement to reduced downwash from the tip leakageflow. . .

Effect of Rub Interaction

As previously stated, clearance control is thecrux of the seal problem and generally, the sealingclearances change with each engine" condition. Rel-ative displacements of the seal components can begrouped into four categories:'^)

(1) Symmetric differential growth - This effectis principally due to thermal growth differences,but pressure and body forces are usually also sig-nificant. It is vital to consider not only thetotal growth but also the thermal response rate.Thus, the shroud cooling and mounting of the sealto the mating parts represents an integrated designproblem.

(2) Ovalization - An example is displacementdue to localized introduction of hot gas thatcauses asymmetric displacements. Also mountingpractice may allow case sag.

(3) Beam bending displacements - These occurfor various reasons; some are: rotor thermal bow,mounting, aeroelastic vibration of seal parts, andmaneuver loads. (Maneuver displacements can belarge.) . •

(4) Machining tolerance buildup

In attempting to maintain close clearances,accumulation of adverse displacements causes rubs(usually localized). If no wear occurs to the

blade (or knife edge) and all wear occurs in theshroud, then only the local clearance increase isgenerated. If wear occurs to the blades (knife),then the clearance has increased around its fullannulus. Thus, an ideal situation would be zerowear to blades with all wear occurring in theshroud; a practical goal is 10 percent of the wearto the blade (knife) and 90 percent in the shroud;this can be termed a 10:1 rub tolerance.

Thermal Response

Intimately associated with rub tolerance isthe thermal response of the seal system when a ruboccurs. This thermal response can be divided intotwo parts - an overall response and a local re-sponse .

The overall response is dependent on the rubseverity and the overall thermal expansion of thestatic and rotating parts. The differential expan-sions can lead to a self-destruct wear mode. Thatis, the rub itself generates enough heat to causethermal growth to increase the severity of the rub.This type of rub interaction has been responsiblefor a series of engine failures. Thus, the firstseal design criteria is avoidance of self-destructrub interaction.

It has been found that the wear, in stable rubinteractions, is dependent on the initial sealclearance. In some engines if the labyrinth sealis assembled with a small initial clearance, a rubinteraction is severe and produces significant wearthat results in large final clearance. But smallerfinal clearances can be obtained if the seals areassembled with intermediate size initial clearances,because then, the thermal response associated withthe rub interaction is less severe and less wear

The change in stator and rotor diameters de-pends on the temperature change (due to the heatinput) and coefficient of thermal expansion. Theheat transfer problem is complex because of thecomplex geometry and temperature gradients. But,in general, it is desirable to maximize the heatinput into the shroud and minimize the heat inputinto the rotor; this tends to minimize the differ-ential thermal growth and the corresponding normalforce between the rotor and shroud.

In addition to the overall thermal response,the local thermal response has a significant in-fluence on the rub interaction. For example, ex-perimental data show that when a labyrinth knifeedge rubs against a shroud segment, the rubbing cantake place over just a small segment (~5° arc) ofthe 360° of knife edge (see Fig. 10). Thus, theheat input is highly localized and a local thermalbump is generated that expands, rubs harder, andfinally wears away. This is then followed by rub-bing over a second small segment which grows andthen wears, ecc. Evidence of localized rubbing isindicated by the heat discoloration of the knifeedge (Fig. 11) that was rubbed against a shroudspecimen. This type of local rub interaction hasbeen investigated from a fundamental analyticalstandpoint, but application of the tip seal problemhas not been attempted.(20-22)

Rub Tolerance

In addition to the thermal considerations, the

REPRODUCIBILITY OF THPORIGINAL PAGE IS POOR

rub tolerance of tip seals is of prime importance,and research effort is currently being directedtoward development of rub tolerant tip seals thatcan operate in turbine environments for long periodsof time without degradation.(23) As an example ofthis work, Fig. 12 shows the torque produced byrubbing of a simulated labyrinth knife edge againsta turbine shroud specimen. The data are for a rub-bing interaction at 183 m/sec (600 ft/sec) with thelabyrinth knife edge penetrating the shroud at arate of 0.254 mm (0.010 in.) per second. The fourconventional type materials showed a wide variationin rubbing torque with the magnitude•of the rubbingtorque being an indication of the wear that occurredto labyrinth knife edge. The experimental materialsevaluated have nickel-chromium-glass flame sprayedcoatings on a fiber metal base. One experimentalmaterial contained calcium fluoride, as a high tem-perature lubricant, in the.coating. The effect ofthis high temperature lubricant can be seen(Fig. 12) to have reduced the reaction torque by afactor of two. Figure 13 shows the clean cut(groove) produced by the knife edge in this experi-mental material.

In addition to reaction torque, erosion rateis a measure of the material potential in engineapplications. Figure 14 shows erosion rates for thefour conventional type materials and the two experi-mental materials. Note that conventional materialtype D, a sintered fiber metal that had low reactiontorque and hence caused low knife edge wear, had ahigh erosion rate. On the other hand, materials Band C had low erosion rates but high torque whichsuggests low abradability (high blade wear). Theexperimental materials (E and F) with the flamesprayed coatings showed the lowest erosion rates.Generally materials with good abradability had poorerosion characteristics and vice versa. The exper-imental materials showed a potential for achievingboth good erosion and good abradability. The ex-perimental materials were the first attempts to usethe surface film concepts. Optimization of bothfilm materials and supporting deformable structuremay give further improvements in tribiological anderosion properties.

Summary of Results

Four types of shaft seals were evaluated undersimulated gas turbine operating conditions whichincluded pressures to 148 N/cm' (215 psi). The re-sults of this experimental evaluation revealed thefollowing:

1. The self-acting face seal operated withoutrubbing contact. This was evidenced by lack ofwear. Of particular interest was the successfuloperation at 54 600 rpm (183 m/sec (600 ft/sec));this was taken as evidence that the gas film stiff-ness was high enough to prevent rubbing contactunder high inertia force conditions.

2. Self-acting face seal leakage was signifi-cantly lower than that of the three conventionalseal types.

3. Conventional contact seals may not be satis-factory in future advanced engines because of wearlife and excessive air leakage flow.

a. Of the conventional seals tested, theface seal configuration was the most successfulat limiting air leakage flow; however, at air-

to-oil pressure differentials above approxi-mately 85 N/cm2 (123 psi) , air flow was con-sidered excessive.

b. The circumferential segmented seal(unbalanced type) configuration operated wellat moderate conditions, but at air-to-oilpressure differentials of 41.4 N/cm2 (60 psi)and speeds above approximately 107 m/sec(350 ft/sec), it wore very rapidly and even-tually operated as a labyrinth.

c. Ring seal clearances, which are deter-mined by shaft dynamics and thermal rub re-sponse, results in air leakage rates that arecomparable to labyrinth seals.

Published data on gas path seals was reviewed;wear and erosion evaluations were made on commer-cial turbine shroud materials and on some experi-mental materials using simulated labyrinth knifeedge rubbing into a shroud specimen. This studyrevealed:

1, Gas path seals have significant effects onengine efficiency and compressor stall margin.

2, The thermal response and rub tolerance ofthe gas path seal are significant factors in thedetermination of the final operating clearances.

3, In four conventional materials there was awide variation in their wear, torque reaction, anderosion resistance. Generally, materials with goodabradability had poor erosion and vice versa.Shroud materials with flame spray coatings on a de-formable substrate show promise of providing moreoptimal tribiological as well as erosion proper-ties.

References

1. Paladini, W., "Static and Rotating Air/Gas SealEvaluation," CW-WR-70-024F, ASTIA AD-730361,June 1971, Curtis-Wright Corp., Wood-Ridge,N.J.

2. Mahler, F. H., "Advanced Seal Technology," PWA-4372, ASTIA AD-739922, Feb. 1972, Pratt &Whitney Aircraft, East Hartford, Conn.

3. Prince, D. C., Wisler, D. C., and Hilvers,D. E., "Study of Casing Treatment Stall Mar-gin Improvement Phenomena," R73AEG326, Mar.1974, General Electric Co., Cincinnati, Ohio;also available as CR-134552, 1974, NASA.

4. Rummel, K. G. and Smith, H. J. M., "Investiga-tion and Analysis of Reliability and Main-tainability Problems Associated with ArmyAircraft Engines," D210-10571-1, ASTIA AD-772950, Aug. 1973, Boeing Vertol Co.,Philadelphia, Pa.

5. Lynwander, P., "Development of Helicopter En-gine Seals," LYC-73-48, Nov. 1973, AvcoLycoming Div., Stratford, Conn.; also avail-able as CR-134647, 1973, NASA.

6. Machine Design, Sept. 13, 1973.

7. Parks, A. J., McKibbin, A. H., Ng, C. C. W^,and Slayton, R. M., "Development of MainshaftSeals for Advanced Air Breathing PropulsionSystems," PWA-3161, Aug. 1967, Pratt S, WhitneyAircraft, East Hartford, Conn.; also availableas CR-72338, Aug. 1967, NASA.

8. Povinelli, V. P. and McKibbin, A. H., "Develop-ment of Mainshaft Seals for Advanced AirBreathing Propulsion Systems, Phase 2," PWA-3933, June 1970, Pratt & Whitney Aircraft,East Hartford, Conn.; also available as CR-72737, June 1973, NASA.

9. Ludwig, L. P. and Johnson, R. L., "Design Studyof Shaft Face Seal with Self-Acting Lift Aug-mentation. Ill - Mechanical Components,"IN D-6164, Apr. 1971, NASA.

10. Ludwig, L. P., Zuk, J., and Johnson, R. L., ""Design Study of Shaft Face Seal with Self-Acting Lift Augmentation. IV - Force Bal-ance," TN D-6568, May 1972, NASA.

11. Zuk, J., Ludwig, L. P., and Johnson, R. L.,"Quasi-One-Dimensional Compressible FlowAcross Face Seals and Narrow Slots. I - Anal-ysis," TN D-6668, June 1972, NASA,

12. Zuk, J. and Ludwig, L. P., "Investigation ofIsothermal, Compressible Flow Across a Rotat-ing Sealing Dam. I - Analysis," TN D-5344,Sept. 1969, NASA.

13. Zuk, J. and Smith, P. J., "Computer Program forViscous, Isothermal Compressible Flow Acrossa Sealing Dam with Small Tilt Angle," TND-5373, Aug. 1969, NASA.

14. Zuk, J.; Ludwig, L. P., and Johnson, R. L., "De-sign Study of.Shaft Face Seal with Self-ActingLift Augmentation. I - Self-Acting Pad Geom-etry," TN D-5744, Apr. 1970, NASA.

15. Povinelli, V. P. and McKibbin, A. H.., "Develop-ment of Mainshaft Seals for Advanced AirBreathing Propulsion Systems, Phase 3," PWA-4263, July 1971, Pratt & Whitney Aircraft,East Hartford, Conn.; also available as CR-72987, 1971, NASA.

16. Szanca, E. M., Behning, F. P., and Schum, H. J.,"Research Turbine for High-Temperature CoreEngine Application. II - Effect of Rotor TipClearance on Overall Performance," TN D-7639,Apr. 1974, NASA.

17. Kofskey, M. G. and Nusbaum, W. J., "PerformanceEvaluation of a Two-Stage Axial-Flow Turbinefor Two Values of Tip Clearance," TN D-4388,Feb. 1968, NASA.

18. Boyce, M. P., Schiller, R. N., and Desai, A. R.,"Study of Casing Treatment Effects in AxialFlow Compressors," ASME Paper 74-GT-89,Zurich, Switzerland, 1974.

19. Boyce, M. P. and Desai, A. R., "Clearance Lossin a Centrifugal Impeller," in IntersocietyEnergy Conversion Engineering Conf., 8th,Philadelphia, Pa., 1973, pp. 638-642.

REPRODUCffilLITY OF THEORIGINAL PAGE IS POOR

20. Barber, J. R.: Thermoelastic Instabilities inthe Sliding of Conforming Solids," Proceed-ings of the Roy. Society, Vol. 312, :to. 1510,Sept. 1369, pp. 381-394.

21. Burton, R. A., Heilikar, I). , and Kilapaiti,S. R., "Thermoelastic Instabilities in aSeal-Like Configuration," Wear, Vol. 24,1973, pp. 177-188._

22. Dow, T. A. and Burton, R. A., "The Role of Wearin the Initiation of Thermoelastic Instabili-ties of Rubbing Contact," ASM£ Paper 72-Lub-45, iiew York, H . Y . , 1972.

23. Shiembob, L. T., "Abradable Gas Path Seals,"July 1974, Pratt & Whitney Ai rc ra f t , Eastiiartford, Conn. ; also available as CR-134689,1974, NASA.

Table I Floating ring seal post test diametral measurements

Test speed, m/sec (ft/sec)

New 91 (300) 122 (400) 152 (500) 183 (600) 213 (700)

Forward Diametral gap, mm 0.06096 0.12446 0.12700 0.12700 0.12700 0.21590seal (in.) (0.00240) (0.00530) (0.00500) (0.00500) (0.00500) (0.00850)

Aft Diametral gap, mm 0.13462 0.13462 0.13716 0.13462 0.14478 0.16510seal (in.) (0.00520) (0.00530) (0.00540) (0.00530) (0.00570) (0.00650)

Table II Typical face seal test data

Rpm

27364536455463

300400500400500600700

Speed

m/sec

91122152122152183213

ft/sec

300400500400500600700

Air pressuredifferential

N/cn

61.61.62.111.113.113.117.

>2 psi

4017182

89.88.90.162.164.165.170.

0500000

Air(two

kg/sec

0.007.007.006.025.023.021.016

flowseals)

Ib/sec

0.016.015.013.055.050.045.035

Sealtemperature

K

350363394

OF

170195250

Table III Self-acting face seal evaluation

Rpm Speed

m/sec ft/sec

Air pressuredifferential

N/cm psi

2736455427364554

300400500500300400500600

9112215218391122152183

300400500600300400500600

23.23.23.22.111.110.109.107.

41114766

34.33.33.32.161.160.159.156.

05505500

Air(two

kg/sec

<0.0006<.0006<.0006.0011.0023.0032.0036.0046

flowseals)

Ib/sec

<0.0013<.0013<.0013.0024.0050.0070.0079.0102

Sealtemperature

K

333352371392364373386402

°F

140174210246196212236263

oooi

W

' COMPRESSORPRESSURE

Fig. 1. - Single stage labyrinth seal.

SUMP FAN VENTPRESSURE PRESSURE PO'

COMPRESSORDISCHARGEPRESSURE

CS-56863

Fig. 2. - Multiple stage labyrinth seal system.

HOUSING^

AIR LEAKAGE s

AIR SIDE

FLOATING RING

OIL SIDE

(a) FLOATING RING SEAL

WAVE SPRING

AIR LEAKAGE

ANTI-ROTATION PIN

1-— SEGMENTED

(b) UNBALANCED CIRCUMFERENTIAL SEGMENTED-RING SEAL

WAVE SPRING

SECONDARY SEAL -

- HOUSING

ROTATING SEAT

(c) FACE SEAL

Figure 3. - Conventional shaft seals (ref. 5).

REPRODUCIBILITY OP THEORIGINAL PAGE IS POOR

v-12 SELF-ACTING\ LIFT PADS EQUALLY SPACED

vDO^oooI

w

60.960 .mm dia.(2.40 in.)

51.816 mm dia.(2.04 in.)

0.0229/0.0152 mm(0.0009/0.0006 in.

-0.889 mm(0.035 in.)

1. SPRING PLATE2. COMPRESSION SPRING3. SPRING PIN4. HOUSING5. CARRIER

66.4144 mm dia.\ (2.536in.)

^THERMOCOUPLE

6. PISTON RING (SECONDARY SEAL)7. BELLOWS SPACER8. OIL DAM AND HEAT SHIELD9. ROTATING SEAT

10. NOSE PIECE

Figure 4. - Self-acting (ace seal design (ref. 5i.

oLU

53— i

i/i

LU

O

fc

cc.

.10

.08

.06

.04

.02

n

r-

8- 3

_ io_!i

-

.05

.04

.03

.02

,01

r LABYRINTH SEALS, ROTATINGRING & WORN OUTCIRCUMFERENTIAL SEALS-

CONVENTIONALFACE SEAL

/- SELF-ACTING/ FACE SEAL

PRESSURE DIFFERENCE, N/CM

1 1 150 100 150

PRESSURE DIFFERENCE, PSI

Fig. 5. - Comparison of seal configurations.

200

oI—Q_

oo

O

Q-co

LU >- —u-i —i o: oo•^ CD |_ i—

£i|iH> <C uj Q_

uj >±1 o;

oo

0

OCO

Q-

_ 5oo o

CT>c0}

_i O<±. Q-

e ts

>- LUCD 00

iSOl JO !N30d3d

a00<c

cmCO<c

s.in3.

_§IX

o>L3

10r-

Q-

>-

ott

O HIGH BYPASS RATIO FANQ RESEARCH COMPRESSOR AD RESEARCH COMPRESSOR CO MULTISTAGE COMPRESSOR

0.01CLEARANCE/SPAN'

0.03

vDOOooI

W

Fig. 8. - Typical compressor efficiency penalty as a function of bladeclearance-to-span ratio (Ref. 2).

LUOo;

o

LU

O

LUO

0

-4

-8

-12

-16

-20

o -•;

-IMPULSEJURBINEx

1

REACTIONTURBINE- '

I I I 1 I) 2 4 6 8 10 12

ROTOR TIP CLEARANCE, PERCENT OFANNULAR PASSAGE HEIGHT

Fig. 9. - Effect of rotor tip clearance onperformance for various turbines(Ref. 16).

JL_ KNIFE

[I EDGE THERMAL

A BUMP

O.D. OFLABYRINTHKNIFE EDGE

SHROUDSEGMENT

INFRAREDPYROMETER

Figure 10. - Local thermal response of labyrinth knife edge rubbing againsta shroud segment in bench tests.

Figure 11. - Labyrinth disk and knife edge showing heat discoloration due tothermal bumps generated in rubbing contact against a shroud specimen.Rubbing speed, 183 m/s (600 ft/sec).

OooI

W

ro

01 x 'H3JGW-N01M3N

I A I I I I ILT\ \r Lf\ < i uS e^ LT\IT\

81- 'Nl ' NOIlDVrd3iNI

no

100

90

80

"s 70

o- 60

COo2 50

40

30

20

10

=J}

MATERIAL TYPE

A, HONEYCOMBB, POROUS CERMETC, POROUS METALD, FIBER METALE^ EXPERIMENTAL

MATERIALTYPE

1033 1144 1265TEMPERATURE, °C

1366

1400 1600 1800TEMPERATURE, °F

2000

Fig. 14. - Erosion rates of shroud material specimens;gas velocity, 0.35Mach number; A^O^, particlerate, 2.72kg/hr (6lb/hr).

NASA-Lewis