You are on page 1of 9

Quality Requirements for General Electric Ductile Iron

Castings Exceeding 10Tons Used for Wind and Gas


Turbine Components - Revisited

by G. K. Bouse and J. Parolini1, N.Rojek2, and C.Zhou3

1
(General Electric ENERGY, Materials & Processes Engineering, Greenville, SC,
2 3
USA; Warsaw, Poland, and Shanghai, China.)

Abstract: Nodular Ductile Iron (NDI) components >10Tons for gas


or wind turbines frequently utilizes either ASTM A395 or EN1563
(EN-GJS-400-18U-LT) material. This paper reviews the actual cast
components beyond external damage and microshrinkage, and
focuses on darkened metal (chunky graphite) exposed by rough or
finish machining (where the machined plane has intersected a
solidification hot spot). Within the hot spot it will be shown that a
chunky graphite morphological change has occurred. This will cause
reductions in tensile and fatigue properties. Steps for identifying,
investigating, and eliminating such indications are also discussed.

Key words: nodular ductile iron, chunky graphite, dark spots, wind
turbines, gas turbines, low cycle fatigue, UTS, YS, % EL, nodularity.

Introduction. The use of ASTM A395 component could be scrapped at


and EN1563 (EN-GJS-400-18U-LT, machining or at the point of assembly.
JS1049) are common for the casting of
gas and wind turbine parts, GE takes a very active role in the
respectively. The ASTM specification production of these components,
is common for gas turbine inlets and focusing on risk reduction techniques.
compressor cases, and the EN spec is These techniques use tools from the
well known for wind components such Six Sigma® toolkit, and focus on
as hubs and bedplates [1], as shown in FMEA or failure mode and effects
Figure 1. The use of these materials analysis, RCA or root cause analysis,
for modern components routinely and CA or corrective action. The focus
exceeds 10 metric tons (T), and 200 is very intense during production of the
mm in wall thickness. These materials first few components. By deciding
continue to work well for turbines as what steps of the manufacturing
megawatts increase, but require processes are likely to fail, FMEA’s are
continued monitoring as parts size conducted both within GE, and the
increases or as new suppliers are supply chain to reduce production
added [2]. Additionally, it is important risks. It is the goal of the supply chain
to insist on good quality now to provide to identify deficiencies early, determine
for the ever-increasing parts of the their cause, and then fix the
future. As such, foundries are being cause/control so it doesn’t happen
challenged to reduce variability, which again. This will provide components at
will reduce the chances that a the point of assembly which meet all
required properties, preventing the

1
Figure 1. Cutaways of the GE 1.5MW wind turbine (left) and gas turbine (right)
showing the locations of the bedplate and hub, inlet and compressor case,
which are made of nodular ductile iron.

Supplier Customer/GE
seldom has one been deficient.
GE
Finds defect in GE
Therefore, with the appearance of
Finds defect in Finds defect at
Their facility Mfg facility project site. chunky graphite, the Materials
Engineer may be required to define
additional quality considerations to
Rule of 10’s include tests from within the casting,
$100,000 conveniently removed by a trepan drill,
$10,000
in the location of a planned hole.
$1,000 However, language within specifica-
tions may use “circular” reasoning to
GE COSTS accept the component (and avoid
Figure 2. Typical “Rule of 10’s” for GE corrective action), unless of course
costs when a supplier’s compo- additional agreements are made. The
nent has a deficiency. arguments go something like this:

escalation of remediation costs, which • The component shall not contain an ab-
typically grow by factors of 10 from normal form of graphite, and
• Use of metallography is permitted to de-
manufacture, point of assembly, to
termine the graphite form.
customer site, as shown in Figure 2. • Abnormal becomes normal over time, so
Participation in this process has the metallography is avoided.
added beneficial effect to make
suppliers World Class. But machine shops, where the chunky
graphite is most noticeable, seldom
Besides a trepan specimen through have the tools to examine the
the component, a cast-on or attached microstructure. Even if metallography
test coupon is the next best way to is used, moving the microscope
gauge the capability of the casting. objective as little as 1 mm can present
The reader must be clear to an entirely new view or form of
understand the minimum UTS, %EL, graphite - from being primarily nodular
etc for the ASTM and EN to being primarily chunky. Even a
specifications are for the test coupons, specification that utilizes darkening as
usually not the casting itself. The criteria is subjective unless visual
coupon is cooled on 5 sides, and very standards are created. It is the

2
authors opinion the criteria to eliminate spot, that is the first argument
the darkening is much easier to define, advanced. Our GE counterparts who
than criteria involving normality and must assemble these components
metallography. concentrate first on fit (dimensions)
and so may not notice the dark spots,
Designers expect uniform properties and are surprised when someone from
everywhere within the casting, and so Materials Engineering points out they
the supply chain must work to remove are not like all the others before them.
these dark spots, as there are Wind parts have only a small window
numerous papers in the literature that for review, as they are usually coated
associates dark spots or degeneration for corrosion protection soon after
to some property deficiency. For GE, machining, and therefore the dark
this realization came in the early spots cannot be reviewed at the point
1980’s [3], which draws on a reserve of assembly. When a component with
of publications as far back as 1967, so discoloration is found, the foundry is
the problem has been around a long requested to review their records, but
time. Then, two articles by Gagne and in many cases, a specific difference in
Argo in 1986 [4] characterized the the process for that one casting is not
chunky graphite form and properties.
However, at that time, modeling of the
mechanical stress state was in an
early stage and designers didn’t know
what to do with the information, unless
there was some service related issue.
As modeling tools have improved, the
ability to analyze a small volume of the
darkened metal (based upon a change
in microstructure) for full life have
become available, and may allow the
use of a casting that contains some
darkening, depending upon its Figure 3. Darkened area on a large
location. Thus, the casting might be ductile iron component, 200 mm
used instead of being rejected, which wide. The darkening was below a
helps improve casting yield. riser.
Eventually though, it would be desired
to eliminate the darkening through
continuous improvement.

Quality Considerations. An “ab-


normal” region on a machined ductile
iron casting is shown in Figure 3. An
extreme abnormal structure is shown
in Figure 4. Truly, these are both
features that do not require much
training to identify. Figure 4. An extreme example of a
darkened surface on a ductile
It may be intuitive that graphite iron component, showing a
degeneration and large castings just cellular structure. This groove
go together. Indeed, when GE and the goes into the core section of the
foundry supplier sit down to discuss casting wall.
acceptance or rejection criteria for a
large casting that might contain a dark

3
MELTING AND INOCULATION TREATMENT CHEMISTRY
POURING
POURING TIME INOCULANT TYPE Mg SOURCE CARBON EQUIVALENT

POUR TEMP Mg AMOUNT RE ELEMENTS


FADE OF INOCULANT
MELTING METHOD METHOD
OTHER RESIDUAL ELEMENTS
% PIG, % SCRAP LATE INOCULATION FADING TIME
%SCRAP
. Chunky
Graphite
FACING SAND SIZE OF DOW NSPRUE CHILL LOCATION
MOLDING SAND SIZE AND POSITION
OF RUNNERS NUMBER OF CHILLS
CORE SAND
RISER MODULUS CHILL SIZE
. .
TYPE OF W ASH.
SLEEVES CHILL MATERIAL
.

MOLDING GATING AND RISERING CHILLS

Figure 5. A cause-and-effect or “fish bone” diagram commonly used to identify


processes with an influence on the formation of chunky graphite.

found. It is important to GE to deter- It’s a fact that raw materials, foundry


mine the cause of the problem, and returns (recycle), sand, and chills
usually requires the supplier to perform change with time, accompanied by
an RCA. The RCA might involve the personnel changes that put variation
creation of a cause-and-effect or into the manufacture of the mold,
“fishbone” diagram, Figure 5, so as to molten metal, or casting. With smaller
narrow the study for the root cause. components, maybe only 12 elements
Then experiments might be performed measured for heat chemistries were
using a simplified shape to recreate adequate for good control. But now,
the hot spot containing the chemistry as component size has increased (and
and cooling rates of interest. This has hot spots become more troublesome),
been done with success at some of and price pressure forces the use of
GE’s foundry suppliers, as long as the less pig iron and more returns, the
experimental shape is at least ~100 x foundry may have to measure more
150 mm in cross-section and solidifies than 12 elements. Rare Earth (RE)
in a manner similar to the actual elements such as Cerium (Ce), in low
component. Nowadays, most risk can concentrations, are not easy to
be removed from an experiment of this measure, and their effect on the
type using solidification software, microstructure is definitely interrelated
which is common throughout the sand with the metal purity, time, and
casting industry. temperature during the iron-making
process.
Once a root cause of the degeneration
is narrowed, a CA is put into place, to Any discoloration found during mach-
reduce the risk for a reoccurrence. As ining needs to be fed back quickly to
a result, the foundries will usually the foundry, so adjustments in the
track/control additional characteristics casting process can be made. This is
of their process, in an attempt to possible if the foundry is in charge of
reduce total variability. machining – they know where the hot
spots are likely to occur, and will watch
The foundry may track more variables, for them as an internal control. But
but often lack of data or the statistical casting and machining may be
savvy or manpower to identify trends. contracted separately, even in different

4
countries. The end result is once the
casting leaves the foundry, the chance
of recording this deficiency decreases
considerably. It then takes a larger
effort by the OEM Quality Engineers to
fix a foundry problem, especially if the
foundry no longer has access to the
casting.

To combat any lingering quality issues,


additional requirements are placed
upon drawings or purchase orders by Figure 7. Example of a microstructure
GE, for each supplier to document any where there is a high percentage
graphite degeneration, including size of chunky graphite amongst &
and locations. This will eventually help between the nodules. This
the foundry move the risers to less material was stripped from a
critical areas (usually off the joining heavy casting during a bolt
flanges), so that if the graphite tightening operation.
degeneration does occur, it is in less
critical areas, and therefore the grey surfaces associated with a flake
component has a greater probability of graphite casting. Note in this case, the
reaching full life. weak material was within threads of a
bolt that stripped it from a large casting
Microstructural Characterizations of during assembly. Consider this reason
Discolored Machined Surfaces. A #1 to avoid graphite degeneration
good ductile iron microstructure in a around a threaded hole.
heavy section is shown in Figure 6,
having >90% nodularity and a nodule Degradation of Mechanical Prop-
density of ~50/mm2. By contrast, a erties. There are numerous examples
microstructure not meeting these in the literature where the mechanical
criteria is shown in Figure 7. properties of the dark spots are
measured vs good material. Figure 8
The microstructure in Figure 7 is is a typical plot for real nodular ductile
exactly the type of structure that iron castings. This data came from a
causes darkening on the machined series of parts and several suppliers,
surface of a ductile iron casting. The Strength vs %Nodularity (Trepan) - Nodular Ductile Iron
darkening is similar to the traditional

UTS
MIN. UTS
UTS or 0.2%YS (M Pa)

0.2% YS
MIN. 0.2%YS
YS

MINIMUM
NODULARITY

PERCENT NODULARITY

Figure 8. A plot representing UTS and


Figure 6, a good example of the micro- YS, showing a drop in the prop-
structure of a nodular iron cast- erty as the nodularity is reduced.
ing. The marker is 100 µm.

5
where trepan specimens were 2. Chunky graphite forms early during
extracted from a finish machined eutectic solidification, and “is a
flange 100-200 mm thick. The two branched and interconnected net-
main components of the microstructure work within eutectic cells.”
were nodular and chunky graphite. 3. Low carbon equivalent (CE), chills,
After testing, each test bar was “non-excessive amount of rare
mounted and the nodularity was earth (RE) elements”, and low
measured using ASTM A247 criteria. supercooling can help eliminate
UTS and 0.2% YS are shown, but chunky graphite.
notice the UTS degrades with lower 4. “High purity charges in combination
nodularity, whereas the 0.2% YS with excessive amount of RE
remains constant. Percent elongation elements will promote chunky
has behavior similar to UTS. Design graphite formation.”
engineers will have to accommodate 5. In the presence of excessive
this behavior for real components, Cerium, Antimony (Sb) is the most
especially at connection flanges. suitable element to add to prevent
chunky graphite formation.
Fatigue behavior is also affected by 6. “Chunky graphite occurs after
chunky graphite. The illustration in nodular graphite when RE’s are
Figure 9 shows that final life for test added, but before nodular graphite
bars taken from an area with chunky in any other case [7]”.
graphite has a lower fatigue life. With 7. “High nodule counts reduces the
this type of data, designers are able to probability that other types of
model volumes of material with a graphite will form [8]”, and
known stress state, to predict if the 8. There is no single factor promoting
substandard material could reach full chunky graphite, but several factors
life. act together.

Elimination of Chunky Graphite in a Several of these observations bear


Nodular Ductile Iron Casting. This additional comments, which would
subject has been extensively studied, apply to a heavy-section ductile iron
recently by several teams from casting, where chemistry would be the
Sweden [5, 6]. Some important obser- controlling factor for the formation of
vations made by them includes: chunky graphite. This applies to the
thermal-center of the section, outside
1. UTS and %EL are lowered, while the reach of external chills.
YS and hardness remain unaf-
fected within a region of chunky The first is the use of too much pig
graphite. iron. While a certain percentage of pig
iron is usually necessary to “normalize”
heats, too much of it presents a base
alloy denuded of nucleation sites for
the primary graphite nodule that is the
first to precipitate from the melt. This
can produce a heat lacking in a good
nodule count. The same could be said
of a heat taken too high in
temperature. As a starting point, the
Figure 9, showing a reduction in low authors suggest 1/3 pig, 1/3 scrap
cycle fatigue life for chunky vs steel, and 1/3 recycled metal when
good ductile iron. formulating heats. Acceptable heavy-
section ductile iron castings can be

6
made with a nodule count as low as

Volume Percent of Chunky Graphite


50/mm2, but the preferred nodule
density would probably be within the
range of 100-150/mm2.
Tsumura, Sb/Ce, Area %

The second comment has to do with


“non-excessive amounts of rare earth
elements”. For a foundry in China,
which has abundant supplies of rare-
earth elements like Ce in the inoculant,
the definition of “non-excessive” will
differ from foundries in countries,
which may not have rare-earth
elements in their inoculation materials.
Worse yet, most foundries do not even Figure 10. Increasing the Sb/Ce ratio
report individual values of Ce. >1.5 can eliminate chunky graphite in
heavy-section ductile iron castings.
The third comment has to do with the [from references 9 & 10].
subject of inoculating with materials
containing Ce and Sb. Larranaga [9] Chemical/Process Drift. Another
and Tsumura [10] have produced a point to consider is when a CTQ like a
simple diagram that shows how dark spot “comes-and-goes” without a
chunky graphite can be eliminated if controlling reason. If one had a
the final Sb/Ce ratio is >1.5. Figure 10 grading scale for dark spots and
shows this relationship between the plotted this variable as a function of
amount of chunky graphite and the time as in Figure 11, one might see a
Sb/Ce ratio, and is mentioned because cyclic variation repeating itself every
of limited success in using it in at least year or ½ year.
one foundry making heavy section
ductile iron castings. This long-range order might then
suggest another variable to track,
A fourth comment has to do with low potentially yielding another cause for
superheat. Although reducing super- chunky graphite. That variable might
heat increases the chance that dross be how the Sb and Ce contents of the
can become entrapped within the recycle material have changed over
casting, doing so will reduce the time. The inoculation material must
supercooling and this is also then be adjusted to maintain the
characteristic of a casting free of desired final ratio of Sb/Ce.
chunky graphite, as described by
Larranaga [9].

Out-of-control/Unacceptable characteristic

Figure 11. The variable causing chunky graphite may also be related to long-
range order, especially when the effect "comes-and-goes” for no apparent reason.

7
Final Comments. The subject of dark order to produce a more uniform
spots and chunky graphite was nodular ductile iron casting.
discussed. This type of graphite within
a nodular ductile iron casting occurs Acknowledgments. The authors
infrequently due to several factors would like to acknowledge N.Jasinski,
including chemistry, inoculation G. Klack, O. Lopez, and C. Liu for
practice, pour temperature (super- helping to collect much information
heat), and solidification time. The area that was helpful during the preparation
of the casting that retains heat, i.e. a of this paper.
“hot spot” - a function of part geometry,
is a significant factor. Abbreviations:

Dark spots are easily recognized on >, greater than


saw-cut or finish-machined surfaces, %EL, percent elongation
and the microstructure within a dark CA, corrective action
spot was shown. The dark spots have CTQ, critical to quality
a microstructure that don’t meet the FMEA, failure mode and effects
intent of ASTM A395 or EN1563 and analysis
must be dealt with separately if it is of LCF, low cycle fatigue
concern to the OEM. Within the dark NDI, nodular ductile iron
spots, the microstructure contains OEM, original equipment manufacturer
chunky graphite, a type of graphite RCA, root cause analysis
degeneration. A degradation of tensile RE, rare earth
UTS, %EL, and LCF were shown to T, ton
exist within the volume of affected UTS, ultimate tensile strength
metal (the thermal center of the
casting section). The supply chain References
must understand that stress and
temperatures continue to increase in 1. H.Roedter and M.Gagne, “Ductile
advanced turbine applications. The Iron for Heavy Section Wind Mill
quality of thick-section iron needs to Castings: A European Experience”,
improve in order for it to be a viable 2003 Keith Mills Symposium on
candidate for future applications. Ductile Cast Iron.
2. G.K. Bouse, “A Review of the Wind
Finally, key references were cited, Energy Business, and Achieving
which the authors believe, have High Quality Ductile Iron Sand
narrowed the causes of chunky Castings for GE Energy”, American
graphite and removed some of the Foundry Society, St. Louis, MO,
conflicting evidence in the literature. 2005.
But then, as now, the challenge 3. T.R. Farrell, “The Influence of
remains to find a way to measure ASTM Type V Graphite Form on
solidification rates, temperatures, and Ductile Iron Low Cycle Fatigue”,
chemistries within the areas that 1983 AFS Transactions, pg 83-15.
present the greatest risk to the casting, 4. M.Gagne and D.Argo, “Heavy
and then be able to take actions within Section Ductile Iron – Part I
the steel-making, molding, or casting Structure and Properties, and Part
processes that would influence the II Formation of Chunk Graphite”,
outcome of the final component. Advanced Casting Technology,
Foundry personnel are especially Edited by J. Easwaren, ASM
encouraged to read those articles and International, 1986, pgs 231-256.
to incorporate more complex solid- 5. R.Källbom, K.Hamberg, and L-E.
ification models within their process, in Björkegren, “Chunky Graphite –

8
formation and influence on
mechanical properties in ductile
cast iron”, Competent Design by
Castings: Improvements in a
Nordic Proj. Vol. 237, pp. 63-86.
6. R.Källbom, K.Hamberg, and
M.Wessén, and L-E. Björkegren,
“On the Solidification sequence of
ductile iron castings containing
chunky graphite”, Materials
Science and Engineering A 413-
414 (2005), pp 346-351.
7. T.C. Xi, J. Fargues, M. Hecht and
J.C. Margerie, “Formation and
prevention of chunky graphite in
slowly cooled nodular irons”, Mat.
Research Society Symposium,
Proc. Vol. 34, 1985, p67-76.
8. P.C. Liu, C.L. Li, D.H. Wu, and
C.R. Loper, Jr, “SEM Study of
Chunky Graphite in Heavy Section
Ductile Cast Irons. Ube Steel Co.,
Japan. AFS Transactions 83-51,
1983, p119-126.
9. P. Larranaga, et. al., “Effect of Sb
and Ce on the formation of chunky
graphite during solidification of
heavy-section castings of near-
eutectic spheroidal graphite irons”,
Metallurgical and Materials Trans.,
40A, Mar. 2009, p654.
10. O. Tsumura, et. al., Effects of rare
earth elements and Antimony on
Morphology of spheroidal graphite
in heavy-walled ductile cast iron”,
Japan Castings, 67, no. 8, 1995,
p540-545

You might also like