Carbon Vol. 27. No. 6. pp. H03-814. 1989 COOS-6223189 $3.(K) + .

OO
Printedm GreatBritain. 6 1989Pergamon
Pressplc

EVALUATIONS OF BLISTER-LIKE FRACTURES AND

CRITICAL PV-VALUES OF CARBON MECHANICAL
SEALS BY THERMAL SHOCK TEST

S. SATO and K. KAWAMATA

Faculty of Engineering, University of Ibaraki. Hitachi 316, Japan

and

J. AIZAWA
Sakuragawa Works, Hitachi Chemical Industry Ltd., Hitachi, 316. Japan

(Received 20 September 1988; accepted in revised form 30 January 1989)

Abstract-The formation of a blister-like fracture, one of the troublesome phenomena observed on

the sliding surface of a carbon mechanical seal, is simulated by a model depicting the distribution of
thermal stresses in a locally heated disk. The critical pressure-velocity values (PV-values) of a carbon
mechanical seal are expressed in two ways. each containing a parameter, which, for the edge crack, is
given by the thermal shock resistance, A, and, for the center of blister on the sliding face, by the
thermal shock fracture toughness, V. These parameters were experimentally measured for 11 kinds of
carbons by using a series of disk testing methods. Practical performances of mechanical seal rings made
of these carbons are discussed in relation to the determined parameters.
Kev Words-Carbon mechanical seal. blister-like fracture, PV-value, thermal shock resistance, thermal
shock fracture toughness.

1. INTRODUCTION ductivity, E; Young’s modulus and a; thermal ex-

pansivity) for the central region of a blister, and by
Mechanical seals are widely used as sealing elements
the thermal shock resistance[4] A( =cr,klEa, a,; ten-
for rotating axes of fluid-handling machines. The
sile strength) for the circumferential region of carbon
demanded conditions have become more and more
ring. In this study, these two parameters, V and A,
stringent as the operating velocity and performance
were experimentally measured for 11 kinds of me-
of the machines using the mechanical seal increase.
chanical seal carbons by applying the measuring
Various kinds of carbons and graphite are usually
technique, which involves local arc discharge heating
used as mating ring materials for mechanical seals
of a disk specimen. The measured results showed
under the requirements that the sealing material pos-
good correspondence with those obtained from the
sesses the desired sealing capability and lubricating
practical performance tests of these carbon materials
property needed to protect the sliding surface. Un-
using the mechanical seal testing apparatus.
der the conditions of high load and high velocity, if
the frictional heat generated on the sliding faces does
not diffuse sufficiently, an overall thermal defor-
2. CARBON MATERIALS FOR MECHANICAL SEAL
mation of the sealing material or localized frictional
heating can occur, thus causing the formation of a As is well known, the fundamental structure of a
slightly bulged surface, the so-called blisters, the pit- mechanical seal is such that it performs the function
ting or cracking of which often causes the mechanical of sealing a rotating axis by mutually sliding the two
seal to lose its sealing capability[ 1,2]. individual end-surfaces of a rotating ring and a fixed
In this study, the breaking out of blisters on the ring, which are lapped together precisely. In general,
sliding surface of a mechanical seal and the adjacent high-hardness, high-strength superhard alloy mate-
cracking along the ring circumference are assumed rial is used as the rotating ring in combination with
to be caused by the superposition of the thermal carbon-graphite material, which possesses self-lu-
stress due to local friction heat generation and the bricating property, as the fixt ring. Proper selection
sliding contact stresses produced, by the slight up- and combination of these ring materials must take
heaval of part of the blisters. It is shown that the into consideration not only chemical properties such
critical pressure-velocity (PV)-values, which specify as the nature and properties of the sealing fluid, but
the design conditions of a mechanical seal system, also mechanical conditions such as pressure, circum-
can be expressed as a function of a couple of material ferential velocity, temperature, and oscillations based
parameters, which are given respectively by the ther- on the operating conditions of the machine. There-
mal shock fracture toughness[3] V( = KJckl Ea, K,c; fore, as is the case with the sliding bearings, the PV
mode I fracture toughness volume, k; thermal con- value, which is the product of contact surface pres-

803
804 s. SAT0 et al.
sure and circumferential velocity corresponding to a mechanical seal by using a resistance factor con-
the maximum seal pressure difference of the system, taining a thermal shock parameter. Golubiev[6] ap-
becomes an important design parameter for the slid- proximated the ring-shaped sliding surface of a me-
ing motion of a mechanical seal. chanical seal by infinite plates and carried out elastic
The carbon-graphite materials are indispensable analysis of unsteady, thermal stress by assuming that
for use as the fixed ring of a mechanical seal because the upper surface is subjected to a uniform heat flux,
they possess good self-lubricating property and ma- q, by dry friction, while the lower surface is being
chinability and are highly resistance to chemicals and cooled. In this case, the maximum thermal stress,
thermal stresses. Moreover, the moderate roughness uxr can be expressed in terms of compressive stress
of their surfaces is known to contribute to the for- on the high-temperature upper surface and the ten-
mation of a fluid film on the sliding surface. The sile stress on the low-temperature lower surface, both
carbon-graphite material, to be used as a mechanical at time infinity, as follows:
seal, is prepared by mixing filler cokes or powdered
graphite with binder pitch, followed by molding and a,= TEaqhl2k, (1)
baking. Sometimes, it is further thermally treated at
high temperatures to undergo graphitizing. Since the where h is the thickness of the plate, the fracture,
volatile components of the binder vaporize during in this case probably caused by the tensile force of
these processes, the material is prone to becoming the lower surface. These theories, however, fail to
porous. Since the material thus prepared, which used explain the featured condition that the actual dam-
directly as a mechanical seal, can easily cause leak- age of a mechanical seal is a loss of sealing capability
age of the sealing fluid, it is further treated by blend- due to the formation of blisters on the sliding surface
ing with proper amounts of resins or pitches, fol- or the occurrence of the edge cracks.
lowed by heating, to increase it impermeability as
well as wear-proof property. Recently, impermeable 3.2 Approximation by thermal stress due to local
graphites of mesophase pitch carbon, named H-10 heat generation on a disk
and -11, were developed without using any impreg- We explain in the following that the damages to
nation. These materials, which do not contain for- a mechanical seal are caused by the formation of
eign impregnation, have good mechanical and ther- both the local blisters on the sliding surface and the
mal properties. H-11 is especially suitable for use edge trackings, attributable to thermal stresses due
under the conditions of high velocity, high load, and to local heat generation on the sliding surface. It will
high temperature. be shown that these thermal stresses can be ex-
In this study, experimental measurements of the pressed as a function of the PV-values. Recently,
thermal shock resistance and the fracture toughness Takeuti et a/.[71 performed a transient thermal stress
were performed for 11 kinds of carbons. Of these, analysis for the case where an eccentric circular heat
8 kinds of carbons were further tested for practical source exists on the end surface of a thick disk. Fig-
use as mechanical seals. The results of these two
kinds of experiments are compared and discussed.

3. THERMAL STRESS FAILURE IN

MECHANICAL SEALS

3.1 Damages on the seal face

By observing the fixed ring of a carbon mechanical
seal, which has lost its sealing capability, one can
often detect localized dull spots on the sliding surface
or the occurrence of cracks along its peripheral edge.
Under the microscope, the dull spots can be seen to
be comprised of a large number of small pitting de-
fects linked with numerous fine cracks. A scan using
a roughness tester can further reveal the existence
of small bulges, called blisters, on these spots. This
type of fracture, produced by local changes in the
sliding conditions, such as mechanical or thermal L/R= 1
load, dry-friction, cooling malfunction, or fluctua-
tions in pressure and rotating speed, can cause local
deformation and growth of defects and increases the
frictional heating and leakage, leading eventually to
loss of sealing capability.
Mayer]51 stated in his book that the cracking due
to thermal stresses of sliding surfaces is the biggest
problem for a mechanical seal and can induce dam- Fig. 1. Simulation of blister is the contact face of mechan-
age and leakage. He appraised the performance of ical seal by thermal stress in locally heated disk model.
 Carbon mechanical seals 805
ure 1 explains the situation posed by this thermal stress on the radius opposite to the radius containing
stress problem by considering the disk, shown with the heat source becomes negligibly small. Now, sup-
a dotted line, as part of the ring. We assume that pose that there is a time period during which the
the thermal stress condition on the ring sliding sur- axial compressive load on the sliding surface can be
face can be simulated approximately by that of the neglected and that the maximum principal stress is
area adjacent to the circular heat source. Under such zero; then the maximum shearing stress. ;i,,, =
an assumption, the compressive thermal stress at the -a ee,,,/2, is produced at the center of the circular
center of the heat source may be somewhat smaller heat source. This shearing stress becomes the cause
than that of the circular disk, with a restricted bound- of the primary fracture since it is more than 1.2 times
ary due to a larger constraint. For a similar reason, larger than the maximum tensile stress at the outer
the tensile thermal stress at the outer edge adjacent edge of the disk. The secondary fracture will prob-
to the heat source is probably somewhat smaller. ably cause trackings due to the tensile stress, CVRRm,,Yr
However, it is possible to make a good estimate of at the outer edge, adjacent to the heat source. These
the related material properties, though not neces- stresses, of course, vary according to the ratio of the
sarily the absolute value of the thermal stress. Tak- heated radius a/R and the ratio of eccentric posision
euti et al. calculated the temperature and stress dis- c/R, and that iTeen,axtends to become larger the larger
tributions in detail for the case with a thick disk of the value of c/R and the closer it is to the outer
radius R, thickness ratio L/R = 1, eccentric position circumference. These stresses also become the cause
c/R = 0.5. heat generation per unit area Q,, within of repeated fatigue failures due to on-off operations
a circular heat source of u/R = 0.2, and the non- of the machine.
dimensional heat transfer coefficient of 0.1, by tak- Suppose the PV-value of a mechanical seal pro-
ing the elapsing time as a function of Fourier number duces heat generation rate. Q, per unit area per unit
T. Figure 2 shows the distribution of nondimensional time on the sliding face; then Q can be expressed as
thermal stress in the radial direction, a,, and that in follows:
the circumferential direction. iYAR,on a radius con-
taining the circular heat source at T = 1. Here, the Q = fPVlj, (3)
nondimensional thermal stress, 0, can be expressed
by the following equation: where f is the friction coefficient, j is the mechanical
work equivalent of energy (0.102 kg . m/W . s). For
i? = uk(1 - v)/EaQ,,R, (2) the heat generation rate, Q, given by eqn (3). the
local heat concentration factor S,, can be derived by
where v is the Poisson’s ratio of the material. The assuming that the sliding motion is due to local con-
stress distribution shown in Fig. 2 remains virtually tact, i.e.
unchanged for r > 0.5, and has a tendency to con-
verge to the maximum thermal stress. According to Q,, = S,Q. (4)
the figure, both i?,, and a,, attain the maximum com-
pressive stress, a,,,,, and (TRBmar, respectively, at the The value of S, can be assumed by taking into con-
center of the heated circle. Although i?,, is zero along sideration the allotment of sliding surface pressure
the outer circumferential surface of the disk, a,, in- and the fact that the area of the disk and the total
creases to the maximum tensile stress, Ceemax,at the area of the ring are respectively about 25 times and
outer circumferential section near the heated circle. 950 times larger than that of the circular heat source.
As can be presumed from Fig. 2, the compressive In the later calculations, S, is taken to be 100 by
assuming that several of these circular heat sources
exist on the sliding ring surface.
0.04
From eqns (3) and (4), the maximum dimension-
alR=0.2 less thermal stress, Oeemar
, produced at the outer edge
CIRd3.5 of the ring can be expressed as follows:
0.02
L/R= 1
a “B”I,iX
= o,,(l - v)kj/Ea(b/2)S,fPV. (5)
-1
% I
Assuming that the fracture occurs when iT,, attains
the value of the tensile strength u, of the material,
the critical pressure-velocity factor, PV, can then be
expressed as:

[PV], = a,k(l - ~)ii{a,,,,,E~s,f(bi2)}. (6)

The central section of the circular heat source is

situated in the compressive stress field, and it is pos-
sible that the shearing stress due to the principal
Fig. 2. Thermal stress distributions in the locally heated stress difference may cause propagation of the crack-
disk. ing. The propagation of cracking starts when the
806 S. SAT0 et al.
mode II stress intensity factor K,, due to shearing According to the above equations, PV values are
stress attains the mode II fracture toughness valve only a function of A or V so far as material properties
K,,,[8,9]. The nondimensional shearing stress, are concerned. It is, therefore, possible to estimate
?,,, = -??,,,,I2 + -5,,,,/2 is expressed as fol- the critical PV values of the carbon materials by
lows: determining A or V.

Fm, = Z&(1 - v)kjlS,~)E~(b/2)S~fPV,

4. EXPERIMENTAL METHODS
(7)
4.1 Thermal shock testings of carbon
where c denotes the crack length and S, is the shape
mechanical seal
factor. For a coin-shaped crack of diameter 2C, S,
4.1.1 Disk specimens. In this study, we have pre-
is given by
pared 11 kinds of carbon materials for mechanical
seal as shown in Tables 1 & 2. Disk specimens of 30
s, = 4 cos oln(2 - Y), (8)
mm in diameter and 3 mm in thickness were used
where o is the angle formed by T,,, and the direction to find the thermal shock resistance A, the thermal
of the crack[lO]. shock fracture toughness V, and the related mode I
Experimental measurement of Kllc caused by and mode II fracture toughness values.
thermal stress is not simple; however, the ratio, The disk specimens to be used in the thermal shock
K,,clKIc, where K,, denotes the mode I fracture
fracture toughness measurement were machined to
toughness value of brittle materials, such as carbons, have an edge slit of length C = 4.5 mm. The ma-
has been known to have the fixed relationship[ll] chining was carried out by first making a slit using
a miling cutter of thickness 0.2 mm. The tip radius
S, = K,,,IK,, % 1.2.
of the slit was then sharpened by using a thin razor
(9)
blade while performing measurement using an op-
Therefore, the critical pressure-velocity factor due tical comparator. The disk specimens to be used in
to shearing thermal stress can be expressed as: the fracture toughness measurement were machined
to have a center slit of 2C = 12 mm. The two tips
[PV], = &(l - u)K,,kj of the slit were also finished sharp.

+ {F,,, Ea(b/2)fS,S,<(w)}. (10) 4.2 Experimental metho& for practical use of

mechanical seals
In eqns (6) and (lo), Y is taken to be a constant 4.2.1 Testing apparatus for mechanical seal. Fig-
irrespective of the material, and material properties ure 3 shows a general view of the laboratory-made
have been lumped together as follows: mechanical seal testing apparatus, which is capable
of testing two sets of mechanical seals at a time.
c+,klEa = A (11) 4.2.2 Testing methods. The mechanical seal test-
KlcklEa = V. ing apparatus was operated under the conditions
(12)
shown in Table 3, which closely simulate those en-
We have defined A and V to be the thermal shock countered in practice. The sealant fluid (C-heavy
resistance[3] and the thermal shock fracture tough- fuel oil), contained in a pressure vessel, was pres-
ness[4], respectively. Using these symbols, eqns (6) surized using nitrogen gas and sent to a seal box
and (10) can be expressed as: under pressure. In this way, each carbon specimen
was repeatedly test-operated under a constant PV-
[PV], = A(l-v) jlS,f(b/2)~‘ee,,, value of 70 kg/cm2 . m/s(=6.86 MPa . m/s). The
(13)
resulting thermal stress fracture, leakage, blisters,
[PV], = V(l-u) jS,/f(b/2)S,S,v’(pc)‘r,,,. (14) and other injurious damages on the sliding surface

Table 1. Conditions of carbon-graphite specimens for mechanical seal

Specimen Impregnation Filler Heat treat

H-l - Pitch coke-graphite 950°C

H-2 Furan Pitch coke-graphite 160°C 2 h
H-3 Pitch epoxy Pitch coke-graphite 9oo”c
210°C
H-4 - Soot-graphite 950°C
H-5 Furan Soot-graphite 160°C 2 h
H-6 Pitch, phenol, Soot-graphite 760°C 2 h
polyamid imide 250°C 2 h
H-7 - Pitch coke 950°C
H-8 Furan Pitch coke lWC2h
H-9 Phenol Pitch coke 250°C 4 h
H-10 - (Pitch carbon, mesophase) 3000°C
H-11 - (Pitch carbon, mesophase) 3000°C
 Carbon mechanical seals x07

Table 2. Physical properties of carbon-graphite specimens for mechanical seal

Specimen designation H-1 H-2 H-3 H-4 H-5 H-6 H-7 H-8 H-9 H-10 H-11

Apparent density p (g/cm3) 1.70 1.83 1.86 1.73 1.85 1.81 1.60 1.73 1.75 1.65 2.00
Shore hardness H, 59 77 75 65 90 93 80 100 100 115 85
Bending strength o,, (MPa) 47 63 58 41 59 62 49 62 71 72 88
Compressive strength o, (MPa) 98 176 176 137 216 235 176 196 196 245 196
Tensile strength o, (MPa) 25 27 29 19 26 25 25 27 30 39 35
Young’s modulus E (GPa) 12 14 14 13 16 16 12 14 15 18 15
Thermal conductivity k (Wim”C) 35 35 41 38 38 19 12 12 12 17 136
Thermal expansion 01(x 10mh/C) 4 5 5 3 7 10 4 7 8 3

of the carbon mechanical seal were observed using be described later. Figure 5 shows the comparison
a microscope and a roughness meter. of thermal shock resistances, A, which were calcu-
lated from the upper and lower limits of the thresh-
old electric powers. The upper and lower bounds of
5. EXPERIMENTAL RESULTS AND DISCUSSIONS
the cross-hatched area show respectively the limiting
5.1 Thermal shock testiqs of carbons value of A, for which either all the specimens are
5.1.1 Experiments on thermal shock resistance. The damaged or no damage of the specimen occurs. Ac-
thermal shock resistances of 11 kinds of mechanical cording to these data carbons H-11, -10. and -4 show
seal carbons were determined by applying arc dis- comparatively large thermal shock resistances, fol-
charge heating to the disk specimens using various lowed by carbons H-l and -3. For the specimens
electric powers. For a heating time, t*, the threshold made of the same material, those that have under-
electric power, which causes cracking to occur, was gone resin impregnation tend to show weakened
measured. The heat time t,, which was about 3 to thermal shock resistance. This trend was clearly
6 seconds, corresponds to a nondimensional thermal found for H-4, -5. and -6, but not for H-l and H-3.
diffusion time, T = 0.25, of the disc specimens. If By comparing carbons H-l, -4, and -7, which are
no crack occurs during the heating time, a larger porous basic graphite materials, it can be seen that
electric power was applied, and the critical threshold the thermal shock resistance becomes larger the larger
electric power was determined. Thermal stress frac- the thermal conductivity, k. and the smaller the lin-
tures can be clearly observed during the heating pro- ear thermal expansion coefficient, (Y. However, no
cess; however, they tend to close up as the specimen definitive trend can be found for mechanical strength,
is cooled down to room temperature. Figure 4 shows such as tensile strength. For reference, values of the
the typical appearance of the cracks and the mag- thermal shock resistance parameter, a,k/Ea, cal-
nified crack paths in the disk specimens. These cracks culated using room-temperature properties (tensile
resemble those that occur on the blisters of the seal strength CT,,thermal conductivity, k, elastic coeffi-
surface during the practical mechanical seal tests to cient E, linear thermal expansion coefficient, a) are

Fig. 3. Mechanical seal testing apparatus.

808 S. SAT0 et al.
Table 3. Testing condition by the mechanical seal H-11
testing apparatus
H-10

Shaft diameter 40 mm H-9

Rev. speed 3000 rpm H-8
Oil pressure 0.68 MPa
H-l
Spring pressure 0.10 MPa
Surface pressure 0.94 MPa H-6
[PV] value 7.06 MPa . m/s H-5
Sealing liquid C-heavy oilt H-4
Counter face ring Tungsten carbide
H-3
Testing time 30 min, on; 60 min, off
Cycle 1000 cycles H-2
H-l
t 12 cst (lOO°C).

A W/mm
shown in Table 4 for the 11 kinds of carbon graphite
materials. As compared with the thermal shock re- Fig. 5. Thermal shock resistance of carbon-graphite for
mechanical seal.
sistances calculated in this study using the threshold
electric power, they show good relative correspon-
dence, although some discrepancies exist due to the of porous basic materials, by using the aforemen-
difference in the measured temperature. However, tioned apparatus under the conditions specified in
there are certain specimens that show poor relative Table 3. The operating time was 100 cycles, totaling
correspondence due to the impact of thermal stress 100 hours. Temperature recordings during the tests
on the resin impregnation. indicated that the maximum temperature and the
5.1.2 Experiments on thermal shock fracture minimum temperature at 1 h/l cycle were 100-147°C
toughness. Figure 6(a) shows the propagation of and 51-74”C, respectively. As stated previously, the
cracking in the carbon mechanical seal during ther- practical test apparatus used in this study is capable
mal shock fracture toughness tests. Figure 6(b) is the of testing two sets of mechanical seals at a time. For
magnified picture showing the growth and propa- convenience of further discussions, we distinguish
gation of cracks from the tip of edge slit. Figure 7 two sets of mechanical seals as 0 the cylinder head
compares the thermal shock fracture toughness val- side (left) and 0 the motor side (right). Table 5
ues, V, calculated from the upper and lower limits shows the mean values of wear, leakage, and the
of the threshold electric power determined for the frequency of occurrence of blister-like cracking after
nine carbon materials. Carbons H-l, -3, and -11 are the operations using the five-step evaluation scheme
seen to have large values. It can be observed that for a total of six sets of mechanical seals. The leakage
the values of the thermal shock fracture toughness was measured using a graduated cylinder for side 0
do not always show a proportional correspondence and side 0 of the mechanical seal, respectively.
with the magnitude of ihe thermal shock resistance. However, it was not possible to determine the exact
In Table 4, the measured thermal shock fracture location of the leak. For 0 and 0 mechanical seals
toughness are compared with the calculated values there is a tendency that the wear of 0 is larger.
of the parameter, &klEa, with the relevant prop- Blister-like damages were not recognized on rings
erties evaluated at room temperature. H-10 and -11, but a few fine cracks were recognized
5.1.3 Fracture toughness experiments. Cracking of on the edge of the rings. The results of practical tests
a disk having a slit at its center, by a diametral shown in Table 5 have good correlations with the
compression, usually starts from the tip of the slit.
It propagates immediately, leading to final fracture. H-11
The values of fracture toughness were calculated H-10
based on this fracture load. The propagation of cracks H-9
normally proceeds along the voids or defects of the H-8
carbon material. Figure 8 compares the mode I and H-7
mode II fracture toughness values for the 11 kinds H-6
of carbon specimens. Resin impregnated materials H-5
are seen to have larger fracture toughness values H-4
than unimpregnated materials. The mode II fracture H-3
toughness values were always larger than the mode
H-2
I values. The ratio, S3 = KllclKlc, was found to be
H-l
about 1.1-1.4, irrespective of the material[9,11]. In
the calculations to be discussed later, S, was taken 0 5 10 15
to be.1.2. V W/mm+
5.2 Practical tests of mechanical seal
The practical tests of mechanical seal were carried Fig. 7. Thermal shock fracture toughness of carbon-graph-
out for seven kinds of carbons, with the exception ite for mechanical seal.
 Carbon mechanical seals X09

Table 4. Comparisons between measured values of the thermal shock resistances and the
fracture toughness and their calculated values from relating properties of specimens

Specimen A (W/mm) o,k/Ea (W/mm) V (W/mm’ :) K,, klEa (W/mm’ ‘)

H-l 5.86 18.2 13.3 23.3

H-2 2.93 14.0 9.0 17.4
H-3 5.75 17.4 IO.5 21.2
H-4 x.57 IX.7 10.0 22.1
H-5 2.19 0.3 6.X IO.1
H-6 2.43 2.1 7.1 3.2
H-7 3.82 6.1 7.9 7.6
H-X 3.07 3.3 7.1 4.3
H-Y 3.14 3.0 7.9 3.3
H-IO X.25 12.4 9.0 7.0
H-II 23.5 77.0 14.1 53.6

thermal shock resistance and the fracture toughness. 5.3 Mechanisms of blister formation
although the number of tests is small and the scat- 5.3.1 Observations of the blister. Figure 9 shows
tering of data is considerable, especially that for the typical appearances of blister-like failures found on
leakage. the sliding faces of carbon rings. A blister, at first

H-2

I I

+-+ (a) 0.1 mm

10 mm

Fig. 4. Typical fracture appearances in the testings of the thermal shock resistance of carbon disk
810 S. SAT0 et al.
glance, looks like a cloudy spot locally losing its gloss the increase of friction coefficient.
on the lapping surface. Under the microscope, some 5.3.2 Mechanism breaking out of blisters. Abar[l2]
pitting defects and numerous linear cracks connect- gave an explanation that thermal stress cracking on
ing the defects can be observed, as shown in Fig. 9. the sliding surface is produced at the moment the
Scanning the blister by using a surface roughness sealing is rapidly cooled by the sealant fluid leaked
meter, a slight bulge, about l-2 km in height as a to the frictional surface. This is based on the simple
whole, can be found, as shown in Fig. 10. On the concept that the cracking of the frictional surface is
bulge sharp crevices are caused by the cracking of due to tensile stress. Mayer et al.[S], on the other
pittings. Such blisters were not found for H-10 and hand, suggested that thermal stress cracking is caused
-11 carbons, probably because they do not include by nonuniform frictional heating of the sliding face.
plugging materials by resin impregnations. In this According to these authors, thermal stress trackings
study, C-heavy oil was used as the sealant. As can can cause the formation of small pitting defects on
be seen from Fig. 9, however, the heat generated the surface of the sealing, although no specific ex-
by the sliding motion causes scorched C-heavy oil to planation was given on the process of such forma-
adhere to part of the inner edge of the rings, thus tion .
hindering heat transfer from the sliding surface. This, The uresent authors consider the mechanism of
in turn, causes local thermal expansion, promoting

,---.-, (a) H-11 (b) , ,

lOll!Ill 0 .1 mm
Fig. 6. Typical fracture appearances in the testings of thermal shock fracture toughness of carbon disks.
 Carbon mechanical seals XII

H-10. -

H-9, - 0

H-8 0

H-7 0

H-6 -

H-5 a 0

H-4 I 0

H-3 m 3

H-2 a 0

H-l 0
I L I
0 0.5 1.0 1.5
kc , Knc MNrn-f
Fig. X. Mode I and 11fracture toughnesscsofcarbon-graphite for mechanical seal.

breaking out of a blister, characterized by bulging, pits and develop into blisters. During such a process,
pitting. and cracking on the sliding surface of a car- the sealant pressure on the sliding face or cavitation
bon mechanical seal, to be as follows: although the probably will help promote the propagation of crack-
sliding surface is lapped precisely, it has fine convex ing or pitting. Among these stresses, the maximum
spots that are subjected to local pressure exceeding compressive stress occurs at the center of the
the mean contact pressure. Since these convex spots locally heated section. When the compressive stress
bear large PV-values locally, concentrated genera- is S,( K,,, / K,, k 1.2) times larger than the tensile
tion of frictional heat ensues. If the carbon material stress component, the compressive stress becomes
has undergone resin impermeability treatment, its the predominant cause for shearing fracture. In other
generally larger thermal expansion coefficient causes words, when the shearing stress calculated from the
softening expansion to occur locally, thus increasing three principal stresses exceeds a critical stress in-
friction as well as local PV-values and heat gener- tensity factor for a defect size contained in the ma-
ation. Such a spot then becomes the field of thermal terial, mode II fracture occurs according to the frac-
stress due to heat generation and sliding contact ture mechanics principle. At the peripheral edge near
stress. Due to these shearing stresses, the small the heating part on the sliding ring, maximum tensile
defects preexisting in the material grow into mode stress occurs as shown in Fig. 2. If the tensile stress
II-type trackings, which, by sliding motion, produce component exceeds the critical tensile strength of
the material, tensile stress fracture occurs according
to the maximum principal stress principle corre-
Table 5. Experimental results of carbographite rings of
mechanical seal sponding to the mode I fracture. With regard to the
mechanism of breaking out of blisters, we have stud-
Wear Leakage ied quantitatively 0, the maximum shearing thermal
Specimen OLm) (ml) Crackingst stress by eqn (7); 0, the maximum tensile thermal
stress by eqn (5); CC,the contact stress accompanied
H-l
H-2 Y 24 3 with friction by the equation of Smith and Liu(l3];
H-3 5 13 2 and 0, the growth of a coin-shaped crack subjected
H-4 - to internal pressure(l41. Table 6 shows the estimated
H-5 15 4s 5
results for carbon H-2. It was revealed that the stresses,
H-6 13 25 4
H-7 - calculated using the assumed constants, S,, SZ, and
H-X 7.5 16 3 Si, for the case in which the aforementioned items,
H-Y I3 28 5 0, 0, and 0 are included. exceed the tensile and
H-IO 4 3 1 shearing strengths of carbon materials. This corre-
H-l I 2.5 1 I
spondence. however, is no more than an indication
i-Blistering. pitting, and hair cracking. A smaller that the assumed thermal stress conditions indeed
number is better in the five step evaluation. qualitatively approximate the contact heat genera-
812 s. SAT0 et al.
 Carbon mechanical seals

H-2

H-3

Fig. 10. Roughness on blister of carbon ring recorded by a roughness meter

tion on the surface of the mechanical seal. The quan- it was shown that PV-values, the fundamental sys-
titative correspondence is probably accidental. Fur- tem design parameters, can be characteristically ex-
ther, even if the shearing stress at the center of heating pressed as a function of thermal shock resistance A
area reaches the shearing strength in the compressive and thermal shock fracture toughness V of the sliding
stress field, it only causes the occurrence of very fine carbon materials. By evaluating both A and V, the
cracks at a localized spot, such as that for a blister, performance of a sliding material can be easily pre-
and does not lead to total fracture, by virtue of the dicted. Experimental measurements of A, V, and
constraint effects. The foregoing calculations were fracture mechanics properties were carried out for
carried out by assuming quasisteady state conditions. 11 kinds of typical carbon sliding materials. Al-
For the mechanical seal, however, repeated stresses though the measured values of A and V differ some-
need to be considered due to the cyclic operation of what from the calculated values using room tem-
actual machines. Under such conditions. it is pos- perature physical properties, due to temperature
sible that trackings may propagate even with a difference and the nonlinearity of fracture, a good
smaller stress. agreement was observed for estimating the relative
magnitude of the strength.
From the practical tests of mechanical seals, which
6. CONCLUSIONS were performed for several kinds of carbon mate-
Based on a simple idea that the problem of heat rials, the amount of leakages and wears were mea-
generation on the sliding surface of a mechanical sured and the conditions of damage on the sliding
seal can be simulated by the distribution of thermal surfaces were observed. The results showed that the
stresses in a thick disk having a local heat source. magnitudes of wears and damages have good cor-
respondences with the magnitudes of A and V.
Table 6. Estimations of stresses, blisters, and cracking Subsequently, the mechanism of breaking out of
occurring on the sliding face of mechanical seal (H-2) blister-like fracture was analyzed quantitatively
for the case in which a circular hole with cracks is
Center (T,,,,) End (u.,,,,)
subjected to thermal stress, sliding contact stress,
Thermal stress (MPa) 0 39.6 0 23.1 and internal pressure. Though depending on the
Contact stress (MPa) 0 12.7 0 19.9 magnitudes of the constant terms assumed, these
Total (MPa) 52.3 43.0 calculated stresses were found to be close to the
Material strength (MPa) 33.0 27.5 tensile and shearing strengths, and showed fairly

CAR
27:6-C
814 S. SATO et al.
good correspondences with the actual occur- 5. E. Mayer, In Axiale Gleitringdichtungen. VDI-Verlag,
rences of blister-like damages. These good cor- Dusseldorf (1970).
respondences tend to suggest that the simulation ‘. A. I. Golubiev, Proc. Fifth Inter. Conf. Fluid Sealing,
A2. Warwick (1971).
used in this study for analyzing the damages on , Y. Takeuti, R. Ishida, and M. Tsuji, Trans. JSME A-
the sliding surface of a mechanical seal is ade- ’ 48, 747 (1982).
quate. 8. H. Awaji and S. Sato, J. Sot. Mater. Sci. Japan 27,
349 (1978).
9. S. Sato, H. Awaji, K. Kawamata, A. Kurumada, and
REFERENCES T. Oku, J. Atomic Energy Sot. Japan 28, 1172 (1986);
Nucl. Engng. Design 103, 291 (1987).
A. Washida, In Point MechanicalSeal, p, 120. Nikkan- 10. H. Tada, In The Stress Analysis of Cracks Handbook,
Kogyo-Shimbunsha, Tokyo (in Japanese) (1972). p. 247. Del Research Corp., Hellertown, PA (1973).
R. P. Paxton, In Manufactured Carbon; A Self-Lubri- 11. H. Awaji and S. Sato, J. Engng. Mater. Tech. ASME-
eating Material for Mechanical Devices, p. 76 CRC H 100, 175 (1978).
Press, Inc., Boca Raton, FL (1979). 12. J. W. Abar, J. Am. Sot. Lubr. Eng. 20,381 (1964).
S. Sato, K. Sato, Y. Imamura, and J. Kon, Carbon 12, 13. J. 0. Smith and C. K. Liu, J. Appl.
. . Mech. ASME-E
309 (1975). 20, 157 (1953).
S. Sato, H. Awaji, and H. Akuzawa, Carbon 16, 103 14. J. C. Newman, op. cit. [lo] p. 193; NASA Tech. Note
(1978). D-6376 (1971).