Heinrich J.G., Aldinger F. (Eds.) Ceramic Materials and Components for Engines
Подождите немного. Документ загружается.
Table
1.
CSGT Engine Test Summary (Aug.
1995
-
June 2000)
Total Test Time
Time at Full Load
Max. Time on
Components
I
Blades
I
Nozzles
I
Combustor Liners*
Ceramics
1
NT164,GN-IO,AS-800 Si3N4
I
SN-88 Si3N4
I
DLC/H-ACI SiC/SiC (CVI and MI)
BFG SiC/SiC (CVI and MI)
1512
hrs
91
hrs
>
2
1,000
hrs
1483 hrs
74 hrs
>
16,000
hrs
1052
hrs
(AS-800) 68
hrs
>
10,000
hrs (H-ACI SiC/SiC CVI Outer
I
Single Assembly
I
I
LinerhlI Inner LinerEBC)
BLADE DEVELOPMENT
Figure
3
shows the evolution of the blade
design. The Centaur
50s
has 62 cooled MAR-M247
metal blades with an oxidation resistant Pt-aluminide
coating. The CSGT blade design has an airfoil shape
that is almost identical to that of the metal blade,
except for the absence of cooling passages. Two
attachment designs were developed. The first design is
that of a “dovetail” which requires the use of a
compliant layer to act
as
a buffer for the
ceramic/metallic interface. Maximum steady state
stresses
in
a 55” dovetail design were predicted to be
214 MPa at the blade root neck under the platform at a
temperature of 682°C. An alternative design concept
was a blade with a double-lug support attachment to
the disk with a ceramic pin. A compliant layer is not
necessary for this “pinned root” design due to the
“round-round” interface geometry. This design was
abandoned because:
(I)
the maximum steady state
stress was higher than for the dovetail blade (283 MPa
vs. 214 MPa), (2) manufacturing was very challenging
because of tight tolerances and intensive machining,
and
(3)
some of the fabricated parts failed
in
cold spin
proof testing.
Figure 3. Blade Design Evolution in the CSGT
Program
Various compliant layer systems were
evaluated
in
cyclic attachment testing. A nickel-base
0.13 mm thick compliant layer system withstood
5000
hrs of testing
(>
200, 24 hr cycles, from room
temperature to 682°C) indicating that it was likely to
perform satisfactorily in an engine operating in
continuous duty mode. Cold spin testing at 125% of
design centrifugal force (CF) load was conducted
as
a
standard proof test for all blades prior to engine testing.
While rig and short term engine tests of
dovetail blades fabricated from GN-
I0
(AlliedSignal
Ceramic Components
-
CC) and NT-164 (Norton
Advanced Ceramics) silicon nitrides were successful,
these first generation design materials could not
provide the required service life based on the
CARES/Life life prediction analysis [6]. For the
30,000 hrs of service life required for commercial
operation, a set of 62 GN-I0 or NT-I64 blades has a
probability of survival (POS) of about
0.10
(based on
fast fracture and slow crack growth failure modes),
insufficient for acceptable component and engine
durability. AS-800 (CC) and SN-281 (Kyocera
Industrial Ceramics Corporation) silicon nitrides,
which became available later
in
the program, were
selected for second generation component fabrication.
Both of these materials have POS values in excess of
0.995
for a set of 62 blades, sufficient for meeting the
program life and reliability goals. Details of the blade
design and development have been reported [7,8].
Figure 4. Centaur
50s
first Stage Disk with Failed
Ceramic Blade Residues
A
milestone
100
hr cyclic endurance engine
test with AS-800 first stage blades with two compliant
layer systems (Ni-alloy
+
BN, and Ni-Alloy
+
Pt)
terminated after 58 hrs due
to
failure of the first stage
blades. There was no evidence of foreign object
damage (FOD)
as
a source of component failure
initiation. While most of the blades had ruptured
above the platforms, only segments of the blade roots
remained
in
alternating slots in one sector
of
the disk
spanning twelve blades (see Figure 4). The more
heavily damaged blades were associated with the Pt-
coated compliant layer system, which caused the
dovetail to pinch in the disk following cyclic engine
testing. The attachment system was subsequently
modified and the failed compliant layer system was
eliminated
[8].
These modifications to the blade
attachment system were evaluated in a successfil
262
second
100
hr test, which included 12 cold starts and
15
hot restarts. Final inspection revealed the ceramic
and metal components to be in excellent condition.
The first CSGT field test was launched in
May 1997 at the ARCO Western Energy (currently
Texaco) Bakersfield enhanced oil recovery site when
the Centaur
50s
SoLoNOx engine retrofitted with AS-
800 (CC) first stage blades and enhanced SiC/SiC
(DuPont Lanxide Composites) CFCC liners started
operation. Figure
5
shows the CSGT engine being
installed in the package at the ARCO site. The CSGT
engine accumulated 948 hrs over the duration of the
first field test in addition to 104 hrs qualification/
endurance testing at Solar, giving a total test time of
1052 hrs at
full
load.
At 948 hrs of operation the
engine shut down as a result of a blade failure, caused
in
all likelihood because of impact damage from a
dislodged metallic locating pin from the inner
combustor liner. Failure analysis indicated that all
blade failures had occurred above the platform, which
excluded failure of the dovetail blade attachment/
compliant layerldisk interface. A second engine field
test with the same ceramic component configuration
was started
in
March 1998. After only 352 hrs of field
testing at
full
load, the field test was again terminated
due to FOD of the ceramic blade airfoils. Various
methods of improving the impact resistance of the
silicon nitride airfoils are being examined, improving
the materials fracture toughness, redesigning the
airfoils to have more robust leading edges, and energy
absorbing coatings.
Figure
5.
CSGT CentaurSOS Engine Being Installed at
ARCO Bakersfield Site
NOZZLE DEVELOPMENT
Figure
6
schematically represents the nozzle
design evolution in the CSGT program. The Centaur
50s
metal engine has
15
two-vane cooled first stage
FS-4 14 metal nozzles, with a WRh-aluminide coating
for oxidation protection. The ceramic nozzle
development started with a single vane segment,
representing one half of the current metal nozzle with
tip shoe removed, for ease of ceramic fabrication. The
initial nozzle design strategy was to maintain the airfoil
profiles
of
the metal nozzle, thereby maintaining the
basic aerodynamic performance of the turbine.
However replacing. the metal airfoils with a ceramic
vane of the same geometric configuration would result
in
an unacceptably high steady state stress level (480
MPa) at the hottest temperature location (1288OCL
FS-414 Metal
CSGT Straight CSGT Bowed
Nozzle
Vane Nozzle Nozzle
Figure
6.
Program
Nozzle Design Evolution in the CSGT
Alternative design concepts considered
included airfoils detached from either the inner or the
outer shroud, segmentedholted airfoils (radial and
axial), segmented shrouds, various pinned attachments
and various clamped attachments, as well as a cooled
nozzle. The reasonable alternatives required complete
redesign of the nozzle airfoil. The airfoil chord was
reduced by
50%,
and the airfoil was bowed, axially and
tangentially, in order to reduce the steady state stresses
to the desired level of about 200 MPa.
A
cantilevered
"hook" support system accommodates the ceramic-to-
metal interface at the first stage diaphragm. There are
42 single vane ceramic nozzle segments compared to
15
two-vane metal nozzles to avoid rotor vibratory
issues.
A
peak transient stress of 242 MPa
(25%
span,
10%
chord location) was predicted 3
s
after a fast stop
shutdown is initiated. CARES/Life analysis and
thermal gradient proof testing indicated that these
stress levels were acceptable for long term operation.
SN-88
OHexoby
SA
0
SN 253
,
*I
168
I
!
0
2
4
6
8
10
TIME,
Hr
(OOO)
062UbOZJM
Figure 7. Long Term Creep Test Data for Candidate
Ceramic Nozzle Materials
The POS was 0.9934 for a set of 42 SN-88 nozzles.
Figure
7
summarizes creep data for candidate nozzle
materials, which formed the basis for selection of SN-
88
(NGK
Insulators, Ltd.) for first generation nozzle
fabrication.
Figure 8. SN-88 Nozzle
-
Chipping at the Inner
Attachment Shroud Occurred During Engine Test
263
All nozzles were proof tested in a thermal
gradient rig and in a mechanical attachment rig prior to
engine testing.
A full set of
42
SN-88 nozzles was
tested for one hour at full load in the CSGT Centaur
50s
developmental gas turbine. Some minor chipping,
found at the inner and outer shroud areas contacting the
metal, suggested a localized excessive contact stress
condition (see Figure
8).
The ceramic nozzle/metallic
attachment was redesigned (see Figure 9) by faceting
the diaphragm and removing material at the diaphragm
contacting edge locations of each ceramic nozzle.
Edge loading on the nozzle inner shroud was reduced
and no chipping has been observed in subsequent
nozzle engine tests. Details of the ceramic nozzle
design have been documented in the literature [7,9].
Figure 9. Redesigned Diaphragm with Faceted Nozzle
A planned
100
hour nozzle engine test was
initiated which included cold and hot engine restarts
and shutdown cycles that progressively increased the
severity of stresses and thermal gradients on the
nozzle. Borescoping after 68 hours of cyclic engine
testing revealed cracking in several airfoils, and the test
was terminated for detailed failure investigations. The
cracking likely originated from environmentally
induced degradation (temperature, water vapor in the
hot section) involving removal of the protective silica
layer from the nozzle airfoil followed by stress-
corrosion type cracking of the silicon nitride.
Application of an environmental barrier coating (EBC)
will be required to retain nozzle materials integrity in
service. An alternative nozzle material, SN-282 silicon
nitride (Kyocera Corporation), was selected for the
fabrication of the new nozzles, because of its reported
improved oxidation and creep resistance compared to
SN-88.
A
100
hr
engine test with the new nozzles
is
scheduled for the fall of
2000.
COMBUSTOR LINER DEVELOPMENT
The ceramic combustor was designed to be
interchangeable with the production Centaur
50s
dry
low-NOx lean-premix (SOLONOX) combustor, which
incorporates louver-cooled Hastelloy@
X
combustor
liners.
[lo].
To achieve the desired emissions
reductions, only the cylindrical sections in the primary
combustor zone were required to be replaced with
ceramic parts. The ceramic combustor liner evolution
is shown in Figure
10.
Initially, both monolithic
(silicon carbide) and CFCCs (SiC/SiC, AI2O3/Al2O3)
were considered as combustor liner materials. Because
of their relatively low fracture toughness, monolithic
liners had to be segmented, and designs based on tiles
and rings were conceived. Because of the much
higher fracture toughness values CFCC liners can be
fabricated
as
integral parts. The materials were tested
first in a subscale configuration using 20 cm diameter
by 20 cm long cylindrical cans, sequentially for 1,
10,
and
100
hrs. SiC/SiC CFCC materials performed best
in the subscale combustor testing, completing
successfully the 100 hr test which included
200
cycles
(15
minutes between -48OOC and -1177°C).
Following the test visual inspection, and NDE at
Argonne National Laboratory showed no evidence of
spallation or delamination. The AI2O3/Al2O3 CFCC
(McDermott, Inc., formerly Babcock and Wilcox
-
B&W) suffered longitudinal cracks because of matrix
shrinkage, and Hexoloy@ SA (Carborundum) Sic tiles
cracked because of contact stresses between the tiles
and the adjacent metallic support structure. Also, these
tiles were too thick to accommodate the thermal shocks
during the test. NT-230 Sic rings (Norton Advanced
Ceramics) were never tested because of fabrication
defects.
SiC/SiC CFCC materials were selected for the
primary zone cylinders. A layer of compliant insulation
between the CFCC parts and the metal housing
minimizes radial contact stresses. Small axial gaps
between the ends of the CFCC parts and the metal
housing prevent end loads.
All
pressure loads are
carried by the metal housing, giving the CFCC
cylinders an easy mechanical ride at steady state
conditions.
A
variety of SiC/SiC CFCC’s have been
tested including 2-D and 3-D CFCCs fabricated by
BFGoodrich Aerospace and 2-D CFCC’s fabricated by
DuPont Lanxide Composites (DLC, currently
Honeywell Advanced Composites, Inc., H-ACI).
A
set
of enhanced SiC/SiC CVI CFCC liners (DLC) is
shown in Figure
10.
The inner and outer liners are 33
cm and
75
cm in diameter, respectively. They are
20
cm long, and are 0.2
-
0.3 cm thick. The surface of the
liners is coated with a CVI Sic seal coat
(-
0.25
mm
thick).
Figure
10.
Combustor Liner Evolution in the CSGT
Program
Having successfully completed the proof
testing
in
atmospheric and high pressure rigs, full scale
SiC/SiC CFCC liners were subsequently tested, either
by themselves, or with silicon nitrides blades and
264
nozzles in the Solar Centaur
50s
CSGT test engine at
the
10
10°C TRIT. Test conditions were representative
for a wide range of operating modes in commercial
operation including steady state,
as
well
as
the transient
conditions associated with start-ups, shut-downs and
trips. Inspections during and following the test showed
the liners generally to be in good condition with no
evidence of degradation. Favorable NOx
(<
15 ppmv)
and CO
(<
10
ppmv) emissions were observed.
The first 948 hr CSGT ARCO Western
Energy field test, mentioned in the blade development
section, also included a set of enhanced SiC/SiC (DLC)
CFCC liners. Post-test inspection revealed that both
liners, but in particular the inner liner, had oxidized.
The mechanism of oxidation involves the formation of
volatile SiO and Si(0H)x species under the conditions
of gas velocity, temperature, total pressure, and water
vapor partial pressure in the combustor. The
phenomena of accelerated surface recession of silicon
based ceramics in the gas turbine combustion
environment
has
been reported
[
1 11.
Several material changes were incorporated in
later generations of CFCC liners to further mitigate the
effects of environmental degradation, including
replacing the ceramic grade Nicalon fibers with
Hi-
Nicalon, increasing the density of the CFCC liners, and
increasing the thickness of the protective seal coat used
for the CFCC liners. Also, in later testing the inner
liner has been fabricated using the melt infiltration
(MI) process, which gives an inherently denser and
higher conductivity CFCC than the CVI process.
Figure
11
shows a composite photograph of
an
enhanced SiC/SiC outer liner after 5028
hrs
of
cumulative exposure
in
three subsequent field tests and
pre-field test qualification tests. The surface oxidation
which is clearly visible is highly correlated with the
temperature of the liners monitored with
thermocouples, which in turn is correlated with the fuel
injector impingement locations. The CFCC material
degradation from the field test closely matched the
degradation encountered in simulated gas turbine
environmental exposure testing at Oak Ridge National
Laboratory (ORNL)
[
121, which enables the lower cost
rig test at ORNL to be used to effectively screen
various CFCC liner systems. Additional details of the
rig and field testing of the CFCC liners have been
presented elsewhere
[
131.
11111
OeSFFloW
11
11
1
Figure
1
1. Composite Photograph of Enhanced
SiC/SiC (DLC) CFCC Outer Liner after 5,028 hr
Engine Testing
In
the most recent field test which started
in
April 1999, environmental barrier coatings (EBCs) to
further increase the life of the SiC/SiC CFCC liners,
have been incorporated in the liner system. Figure 10
shows a set of SiC/SiC CFCC liners with EBCs
developed by Pratt
&
Whitney and United
Technologies Research Center. As of the middle of
June 2000, EBC coated liners at the Texaco (ARCO)
site had accumulated over 10,000
hrs
of service life.
Apart from spallation correlated with pre-existing
discontinuities in the inner liner and minor surface
imperfections in the outer liner, the EBC appeared to
have afforded adequate protection.
Solar is planning continued field testing at the
Texaco site in the 2000-2001 time frame. CFCC liners
with EBCs have also been incorporated into a Centaur
50s
engine at the Malden Mills Industries textile
factory
in
Lawrence, Massachusetts. The test was
initiated in August 1999 and is scheduled to continue
for at least
8,000
hours. As of the end of May 2000,
the CFCCs at the Malden Mills site had accumulated in
excess of
5,000
hrs of service life. A second Centaur
50s
engine at the Malden Mills site is scheduled for a
CFCC retrofit in August 2000, bringing the number of
Centaur
50s
engines with CFCC retrofits operating
in
continuous duty service to three. While the Malden
Mills engines operate with many more starts and stops
(1
09
as
of end of May 2000) than the Texaco engine
(38 as of end of May 2000), the EBC apparently is
equally protective under both sets of conditions.
SECONDARY COMPONENT DEVELOPMENT
Secondary component redesign was made
necessary by the following changes relative to the
production Centaur
50s
engine: (1) increase of TRIT
from 1010°C to
1
121"C, (2) first stage nozzle design
changed from integral to separate tip shoe to reduce
fabrication complexity, and (3) compliant clamping
and sealing for the uncooled ceramic nozzle. Materials
for the stage 1 disk, the nozzle support housing and
diaphragm were changed to accommodate the higher
metal temperatures. A more compliant face seal was
selected to minimize all mechanical loads imparted to
the ceramic nozzle. The design changes have been
documented in the literature [7,9].
SUMMARY
Functional ceramic designs were developed for the
Solar Centaur
50s
first stage blade, nozzle, and
combustor liners. The ceramic components were
fabricated and evaluated
in
rigs and engines.
AS-800 silicon nitride blades were evaluated
in
a
field test at the Texaco (formerly ARCO Western
Energy) industrial cogeneration site in Bakersfield,
California.
A
maximum of 948
hrs
of full load
field testing on one set of AS-800 blades was
accumulated before the test terminated because
of
impact failure of the blades. Improving the impact
resistance of monolithics will be required for
rotating components of gas turbines.
One set of SN-88 Silicon nitride nozzles was
tested in-house for 68 hrs. Material degradation
265
associated with the removal of the protective oxide
layer was observed. The incorporation of
an
environmental barrier coating (EBC) will be
required for extended engine life.
Testing of SiC/SiC CFCC combustor liners has
been most successful to-date. Over
2
1,000
hrs
of
engine testing had been accumulated by midJune,
at the Texaco (Bakersfield, California) and Malden
Mills (Lawrence, Massachusetts) test sites. The
engines with the SiC/SiC CFCC liners have shown
significant emissions reduction compared to the
all-metal engine. NOx and CO levels below
15
ppmv and
10
ppmv, respectively, have been
observed. The longest engine test with one set of
SiC/SiC CFCC liners with a protective EBC
system had accumulated 10,000
hrs
as
of mid-June
2000.
ACKNOWLEDGEMENTS
The authors wish to acknowledge the support of
personnel of the U.S. Department of Energy
-
William
P. Parks, Jr., Stephen
Waslo,
Patricia Hoffman, and
Debbie Haught
-
for
the work performed under the
Ceramic Stationary
Gas
Turbine Development program
(DOE Contract No. DE-AC02-CE40960).
REFERENCES
1.
2.
3.
4.
5.
6.
M. van Roode, W.D. Brentnall, P.F. Norton, and
G.P. Pytanowski, “Ceramic Stationary Gas
Turbine Development”, ASME paper 93-GT-309,
ASME TURBO EXPO, Ohio, USA., May 24-27,
1993.
V.
Parthasarathy, M. van Roode, J.R. Price,
S.
Gates,
S.
Waslo, and P.
Hoffman, “Review of
Solar’s
Ceramic Stationary
Gas
Turbine
Development Program, 6* International
Symposium on Ceramic Materials and
Components for Engines, Arita, Japan, October
M.
van Roode, J.R. Price,
S.
Gates,
S.
Waslo,
and
P. Hoffman, “Ceramic Stationary Gas Turbine
Development
-
Opportunities and Challenges
-
“Proceeding of the 7* International
Gas
Turbine
Congress, Kobe, Japan, November 14-1 9, 1999.
J.R.
Price,
0.
Jimenez,
V.
Parthasarathy,
N.
Miriyala, and D. Leroux, “Ceramic Stationary
Gas
Turbine Development Program
-
Seventh Annual
Summary”, ASME paper 2000-GT-0075, ASME
TURBO EXPO, Munich, Germany, May 8-1
I,
2000.
S.
Gates, Jr.,
J.
Price, V. Parthasarathy,
N.
Miriyala, and
0.
Jimenez, “Recommended
Direction of the Solar/DOE Ceramic Stationary
Gas
Turbine Program”, ASME paper 2000-GT-
0076, ASME TURBO EXPO, Munich, Germany,
May 8-1
I,
2000.
N.N.’Nemeth, L.P. Powers, L.A. Janosik, and J.P.
Geykenyesi, “Durability Evaluation of Ceramic
19-23, 1997. pp. 259-264, 1997.
V0l.1,
pp.65-72
7.
8.
9.
Components Using CARES/LIFE, ASME paper
94-GT-352, ASME TURBO EXPO, The Hague,
The Netherlands, June 13-1
5,
1994.
Transactions
of ASME, Journal of Engineering for Gas Turbine
andPower,
118
(I),
150-158, 1996.
P.F. Norton, G.A. Frey, H. Bagheri, A. Fierstein,
C. Twardochleb,
0.
Jimenez, and A. Saith,
“Ceramic Stationary
Gas
Turbine Development
Program
-
Design and Life Assessment
of
Ceramic
Components”, ASME paper 95-GT-383, ASME
TURBO EXPO, Houston, Texas, USA, June 5-8,
1995.
0.
Jimenez, J. McClain, B. Edwards,
V.
Parthasarathy,
H.
Bagheri, and
G.
Bolander,
“Ceramic Stationary
Gas
Turbine Program
-
Design and Test
of
a Ceramic Turbine Blade”,
ASME paper 98-GT-529, ASME TURBO EXPO,
Stockholm, Sweden June 2-5, 1998.
L. Faulder,
J.
McClain, B. Edwards, and
V.
Parthasarthy, 1998, “Ceramic Stationary Gas
Turbine Development Program
-
Design and Test
of a First Stage Ceramic Nozzle”, ASME paper
98-GT-528, ASME TURBO EXPO, Stockholm,
Sweden. June 2-5.1998.
10.
K.O.
Smith, and’ A. Fahme, “Testing of a
Full
Scale Low Emissions, Ceramic Gas Turbine
Combustor”, ASME paper 97-GT- 156, ASME
TURBO EXPO, Orlando, Florida, USA, June 2-
5,1997.
11.
J.L. Smialek, R.C. Robinson, E.J. Opila, D.S. Fox,
and N.S. Jacobson, “Sic and Si3N4 recession due
to volatility under combustor conditions”, Adv.
Composite Mater, 8[
I],
33-45, 1999.
12.
K.L.
More, P.F. Tortorelli, M.K. Ferber, L.R.
Walker, J.R. Keiser, W.D. Brentnall,
N.
Miriyala,
J.R. Price, 1999, “Exposure of Ceramics and
Ceramic Matrix Composites in Simulated and
Actual Combustor Environments”, ASME paper
99-GT-292, ASME TURBO EXPO, Indianapolis,
Indiana, USA, June 7-10, 1999.
13. N. Miriyala, and J.R. Price, “The Evaluation of
CFCC Liners after Field-Engine Testing in a Gas
Turbine
-
II”,
ASME paper 2000-GT-0648, ASME
TURBO EXPO, Munich, Germany, May 8-1
I,
2000.
14. H.E. Eaton, G.D. Linsey, K.L. More, J.B. Kimmel,
J.R.
Price, and
N.
Miriyala, “EBC Protection of
SiC/SiC Composite in Gas Turbine Combustion
Environment”, ASME paper 2000-GT-063
1,
ASME TURBO EXPO, Munich, Germany, May
8-
1
1,2000.
266
AN INVESTIGATION ON PASTE
FLOW
IN A PRESS-MOULDED
CERAMIC DOME
X.
Huang*, A.
S.
Burbidge*’
**
and
S.
Blackburn*9
**
(*).
IRC
in Materials for High Performance Applications,
(**).
School
of
Chemical Engineering,
The University
of
Birmingham, Birmingham, B15 2TT,
U.K.
been produced by machining. This method wastes
energy and material and is consequently expensive.
Precision press moulding offers a near net solution to
the manufacture of this component.
ABSTRACT
Paste flow
in
a press moulded ceramic dome behaves as
a non-rheometric flow, for which a fully theoretical
analysis is still difficult since ceramic pastes are highly
viscous non-Newtonian fluids with a yield stress.
Experiments and computer simulations, on the other
hand, possess many advantages for the analysis of the
behaviour
Of
such
paste
flows
in
the
dome’
Flow patterns, which are a reflection of the rheological
properties of ceramic paste, are now recognised as
Ceramic domes
are
used
in
hemispheres
and
spheres
for
omnidirectional hydrophones and may potentially be
applied
in
Hi-fi tweeters. In addition by the application
of electromagnetic fields during the moulding process
it
is possible to influence the functionality of the material,
a
feature
not
possible
by
the
machining
route.
playing
a
key
in
determining
the
performance
Of
the
Ceramic paste flow
in
a press mould
is
complicated. The
deformation mode in the paste is triaxial, where the
paste
suffers
compressive
strain
in
the
normal
direction
and elongational strains
in
both circumferential and
radial directions. The stress acting on the paste in such a
deformation consists of six independent components and
is a second order tensor quantity. When the principal
axes of deformation of the paste are aligned with the
final product. This paper presents some preliminary
results of an investigation into the flow behaviour of a
ceramic paste
in
a press-moulded ceramic dome using
both experiment and computer simulation. In order to
clearly display the flow patterns, a paste sample
consisting of three different coloured layers was pressed
in a dome mould and the pressed sample was analysed.
-
chosen coordinate frame, the analysis becomes possible
using Benbow and Bridgwater equations
[I]
as
illustrated
in
the
work
by
Huang
and
Oliver
[2,
3,
41.
The experimental visualisations show three regimes of
paste flow
in
the pressed-mould dome, which should be
characterised by different models. The flow pattern in
each regime is highly dependent on the location of the
paste front as it advances. The flow patterns obtained
using Flow-3D computer simulations are consistent with
the experimental results at large strains.
INTRODUCTION
Modern manufacturing is being forced to reduce costs
and material wastage through energy saving and better
processing. Net shape manufacturing addresses these
requirements by developing process routes, which give
the desired shape with minimum waste and reworking.
Ceramic domes are frequently made
of
piezoelectric
materials, which generate a charge when strained and
change shape when an electrical field is applied. Thus
they can be used as a transducer material. Lead
zirconate titanate (PZT) is one of such materials. Items
made from PZT ceramics can be formed in a wide
variety of shapes by pressing and moulding processes.
Typical shapes include discs, plates, curved plates,
cylinders and hemispheres, depending on the desired
applications. Perfect hemispherical dome structures are
difficult to produce in ceramics and have conventionally
In the authors’ previous study
[5],
a simple construction
of ceramic dome, which was called a “top hat”, was built
to approximately model the paste flow
in
the real dome.
In this work, the paste movement in the “top hat” was
divided into three steps as shown in figure
2.
Each of
these steps could then be considered to be a near
rheometric flow, with which some approximate solutions
could be obtained and could then be superposed to
estimate the overall Solution. In the first step, the paste
flowed as a squeezed film. Then, there existed both a
squeeze film flow and an annular flow. When the paste
front reached the rim of the “top hat”, three flows,
squeeze
film
flows both
in
the top part and the rim, and
annular
flow
in
the space between two cylinders occur
simultaneously. For each step, a semi-empirical equation
was given by the analysis of the paste flow, which gave
good agreement with experimental results
For paste flow such as those being investigated here, it
is more convenient to use visualisation and computer
modelling for characterisation. This paper presents some
preliminary results obtained using both experimental
visualisation and computer simulation. Since the moving
front of the paste
in
the mould has a sharp free surface,
267
the code Flow-3D, which incorporates a free surface
tracking algorithm, was employed in this investigation.
Upper Part
Paste
V
......
stopping the pressing process at various times and
sectioning the paste through the central point it was
possible to visualise the flow behaviour during the
triaxial deformation.
Paste
‘
Lower
female
part
Figure
2.
A real press mould for making ceramic dome.
(b) step
2
04
......
0
1 2
3
4
shear rate(l/s)
Figure
3.
Rheological property of the PZT paste.
(c) step 3
Figure
1.
A simplified press mould apparatus showing
the paste flow development.
EXPERIMENTAL
A commercial lead zirconate titanate (PZT) powder
(PZT-SA, Morgan,
UK)
was used to make a paste by
mixing with an aqueous binder system consisting of
polyvinyl alcohol (PVA, Nipon Gohsei)/water. Pre-
mixed PZT powder and polymer bindedsolvent was
milled under high shear stress on a twin-roll mill for
10-
15
min to form a tape with the thickness of
4
mm.
Press
forming was carried out by placing the press mould,
as
shown
in
figure
2,
in a Denison mechanical testing
machine. The pressing process was performed at
cross
head speed of
5
mm/min.
The rheological properties of the pastes were
investigated using a Rosand capillary rheometer. Three
dies (3
mm
in diameter) with the die length to die
diameter ratio
(UD)
of
16,
10,
and
0.08
were used. The
range of the extrusion velocities was
0.25
to
2
mm/min.
The experimental data are shown in figure
3.
In the flow visualisation work, food colourings were
used as dyes and no change in the paste flow properties
was caused by the colour change. It was possible to
make a “sandwich” of three coloured discs of paste. By
SIMULATION AND NUMERICAL
MODEL
The Flow-3D simulation software package has been
used for this study since previous studies have
demonstrated its capabilities in producing reasonable
predictions of macroscopic flow behaviour during
mould filling in the casting process
[6].
Compared to the
conventional mould filling process, paste flow during
dome press moulding has similar features, characterised
by a sharp free surface. Coupled with the fluid flow and
free surface tracking algorithms, the code provides
insight into the details of flow patterns in the pressed
mould. In this study, the effect of non-Newtonian
behaviour of the ceramic paste on the deformation has
been taken into account. Assuming the paste to be
incompressible, the governing equations and boundary
conditions can be written.
Paste free surface evolution:
where
F
is the fractional volume-of-fluid contained in a
given cell;
ui
is the velocity.
Ai
is the open area fraction
associated with the flow in the ith direction and
V
stands
for the open volume fraction of the cell to flow. The
right hand side of the equation has been taken to be zero
268
since no phase change has been assumed to take place
during the pressing process.
Momentum conservation:
(2)
where
P
is the averaged pressure, and
g,
and
fi
are
gravitational acceleration and external body forces,
respectively. In equation
(2),
p
denotes apparent shear
viscosity of the paste, which can be described by the
Carreau equation:
(3)
where and are shear viscosity at the highest shear
rate and zero shear rate, respectively.
4
and
ill
stand
for parameters relative to time constants for the given
material. denotes the shear rate.
n
represents the flow
behaviour index. The above parameters for the ceramic
paste used in this work have been experimentally
determined, as
lu,
=
0.05
MPa.s,
r(10
=
73.5 MPa.s,
4
=
0.405,
ill
=
700
s
and
n
=
0.2.
The Carreau equation is
fitted to the flow curve for the PZT paste
as
shown in
figure 3.
The boundary conditions specified in the simulation
are
that the upper male part is being pressed down with a
velocity of
5
dmin and the lower female part is
stationary. The paste flows with no shear stress on the
contact surface. The simulation starts when the upper
mould just touches the paste, which has been assumed to
be placed in the lower mould. A mesh with the minimum
grid size of
0.2
mm
was employed in this work as shown
in
figure
4.
Figure
4.
The mesh used in the computer,simulation of
the press mould.
RESULTS
FLOW VISUALISATIONS
The materials used were pre-pressed into a sheet and cut
into a circular cake of constant thickness. Three discs of
the paste were placed together with the central layer
being a different colour. Flow visualization results are
shown in figure 5. It can be seen from figure 5(a) that
when the top mould touches the paste, the deformation
rate of the upper part of the paste is obviously larger
than the lower. This deformation process continues as
the paste spreads to fill the space in the mould as the
upper punch moves down (figures 5(b), (c) and (d)).
Ultimately this deformation process results in the upper
layer being thinned to such an extent that in places the
central layer can be seen at the surface as shown
in
figure 5 (d). The upper layer is largely displaced to the
rim by the operating mechanism. It should be noticed
here that there are some cracks in the samples as seen in
figure
5
(a), (b) and
(c),
which are caused by removing
the sample from the mould. Also, cutting the samples
after the pressing process can sometimes make two
different colours
of
the paste mix on the surface of the
cross section as shown in figure 5 (a).
(a)
t
=
30
s
@)
t
=
60
s
(c)
t
=
102
s
(d)t=
138s
Figure 5. Paste flow visualisation during a pressing
process with the downward speed
5
dmin
FLOW3D SIMULATIONS
Figure
6
shows numerical simulation results. It can be
seen that the simulation results are consistent with the
experimental visualisations at high strains. The
deformation patterns observed in experiment are
reasonably modelled except at early time intervals as in
figures 6(b) and (c), in which the numerical results
deviate significantly when compared with the flow
visualisation. The reason for this case is unclear, but this
result is highly suspect due to the high stress at the
contact with the stationary female mould block. After
the whole paste is deformed, the simulation result can
truly reflect the real paste flow during pressing. The
maximum deformation is taking place around the
moving punch. Comparing the photograph shown in
figure 5(d) with the simulation result in figure 6(g), both
show that there exists a recirculation zone in the rim of
the mould. However, the existence of such recirculation
269
zone does not affect the final product (a sphere
combining two domes) since the rim part has to be
removed once the pressing process is complete.
(a) t
=
0
s
(b)
t=30s
(c) t
=
60
s
(d)
t
=
90
s
(e) t
=
108
s
(f)t=126s
(g) t
=
144
s
(h) t
=
156
s
Figure
6.
Computer modelling of the
PZT
paste flow
patterns during a pressing process with the downward
speed
5
mndmin.
DISCUSSION
The paste deformation first occurs near its top surface,
which contacts with the moving upper male part, and the
bottom edges, which are in contact with the concave
surface of lower female part as shown by the
visualisation method. The visualisation and computer
simulation results show that the paste gradually deforms
to match the surface shape of the upper mould and
reduce the gap between the paste and the lower part of
the mould. However in the early stages the simulation is
unable to reflect this as all the deformation takes place
at the top punch and only later is the bottom die filled.
This cannot be correct as the point stress where the paste
meets the bowl must be exceptionally high.
When the bottom surface of the paste has not fully
touched the whole concave surface of the lower mould,
the paste flow is mainly taking place in the upper part of
the paste where the paste is squeezed due to the
downward movement of the upper mould. Once the
bottom of the paste is in full contact with the concave
surface of the lower female part, i.e. the gap between the
paste and lower mould has been filled by the paste, paste
starts
to
flow along the decreasing gap between the
upper and lower mould. This flow pattern can be
regarded
as
an annular flow where the radial dimension
and thickness of such a flow are dependent on time.
Based on this assumption, the paste movement in the
mould can be divided into two flow regimes with regard
to the lowest point of the convex surface of the upper
mould
as
shown in figure
7.
The flow pattern of the
paste section located under this point may be regarded
as the squeeze film flow while the motion of the other
part above this point may be approximately treated as an
annular flow. The latter changes rapidly with time. Also,
the deformation of the paste is noticeably three
dimensional. It is therefore manifest that it is unlikely to
find a general analytical formulation to describe the flow
of the paste in such situation because the paste is a non-
Newtonian fluid with a yield stress. However, it is
interesting to note that the movement of the paste in this
area is highly organised
as
can be seen from the
visualisation and simulation results. The upper layers of
paste thin and move towards the rim while the lower
layers tend to remain in the dome.
Figure
7.
The flow regions of the paste in the dome
mould.
When the paste front enters the rim zone, the flow
approximates a squeeze film flow between two parallel
plates. Therefore the paste flow in the mould during
pressing can be approximately described by these three
modes, squeeze film flows
in
the rim and in the region
below the lowest point, and an annular flow in the
curved gap above the lowest point.
In summary, it is evident that the deformation
of
the
paste in the dome mould during pressing is a process in
which the paste is extended in both radial and
circumferential directions whereas it is deformed
compressibly in the vertical direction. These
deformations are dependent on time and the relative
movement of the moulds or gap between the upper and
lower moulds. Compared with the previous work
[5],
simplifying the real press dome into a “top hat” shape is
reasonable to allow approximate theoretical analysis.
270
Therefore, semi-empirical formulations developed there
may be applied to the true dome with some necessary
modifications; this will be addressed
as
part of further
study in this project.
CONCLUSIONS
Paste flow in a press moulded ceramic dome has been
investigated using both visualisation and computer
modelling. It is found that the paste flow in such mould
behaves like a non-rheometric flow, for which a fully
analytical analysis is difficult since ceramic pastes are
highly viscous non-Newtonian fluids with a yield stress.
Preliminary experiment and computer simulation results
have been presented and conclusions have been drawn
as follows:
(1)
Experimental visualisations indicate the existence of
three regimes of paste flow patterns in the pressed-
mould dome, which should be characterised using
different models.
(2)
Flow patterns, which reflect the rheological
properties of ceramic paste, are found to play a key role
in
determining the performance of final product.
(3)
The flow pattern in each regime is highly dependent
on the location of the paste front and time.
(4)
The flow patterns obtained using Flow-3D computer
simulations are consistent with the observed results at
high strains. This demonstrates that it is possible to use
computer modelling for the design
of
the paste forming
processes but care in their application is needed at low
stains. The rheological behaviour of such ceramic paste
in
such flows can be analysed using numerical
simulation although a full analysis remains to be
developed.
ACKNOWLEDGEMENT
This work is financially supported by European
Regional Development Funding and IRC
in
Materials
for High Performance Applications at The University
of
Birmingham. The authors would like to thank Drs B.
Su
and D. Pearce for the help in providing the PVA paste
and the pressed mould.
REFERENCES
(1)
J.
Benbow and
J.
Bridgwater, Paste Flow and
Extrusion, Clarendon Press, Oxford, (1993)
3
1-36.
(2)
X.
Huang
&
D. R. Oliver, Processing properties of
ceramic paste in radial flow.
British Ceramic
Proceedings (1997) 58, 103-112.
(3) D. R. Oliver and
X.
Huang, Squeeze film testing of
ceramic pastes.
British Ceramic Transactions
(accepted
for publication in
2000).
(4)
X.
Huang, Rheology of ceramic pastes. PhD
Thesis,
the University of Birmingham. (1998).
(5)
X.
Huang, A.
S.
Burbidge,
D.
R.
Oliver and
S.
Blackburn, Approximate flow analysis of the paste
forming process for a simplified ceramic dome.
British
Ceramic Transactions
(to be submitted).
(6)
X.
Yang, T. Din and
J.
Campbell, Liquid metal flow
in moulds with sprue off-set relative to runner.
International Journal
of
Cast Metal Research,
(1998)
11, 1-12.
27
1