PVP-Vol.

471, Fitness for Service, Life Extension, Remediation, Repair,

and Erosion/Corrosion Issues for Pressure Vessel Components -- 2 0 0 4
July 2 5 - 2 9 , 2004, San Diego, California USA
PVP2004-2239

FAILURE ANALYSIS OF WELD REPAIRED TURBINE ROTOR

Fred V. Ellis Blaine W. Roberts

Tordonato Energy Consultants, Inc. B. W. Roberts Engineering Consulting

4156 South Creek Road 4715 Rocky River Road
Chattanooga, Tennessee 37406 Chattanooga, Tennessee 37416
(423) 622-7555 (423) 892-3597
FAX (423) 622-7522 [email protected]
E-Mail; [email protected]

ABSTRACT corrosion properties for the turbine rotor weld repair application. The
The fast stageof an ]P turbine rotor failed after approximatelyeight elevated temperature strength of Modified 9Cr-lMo steel is
years of service. The rotor had been weld repaired using a 360 degree all approximately equal to that of TP304H stainless steel up to temperatures
weld metal buildup. The rotor material was specified as ASTM A-470 of 6000C. The high temperature strength is achieved by controlled
Class 8, ICr-IMo-I/4V forging material, and the weld metal was miero-ailoying additions of Cb, V, and N, coupled with a normalizing
Modified 9Cr-lMo known as Grade 91 in the wrought product forms. and tempering heat treatment. Extensive testing has been performed on
The weld repair procedure used a controlled weld bead deposition Modified 9Cr-lMo, and its metallurgical and mechanical properties are
process to refine the grain size of the HAZ. A failure analysis was fully docomented (4, 5).
performed includingmetallurgical examination and stress analysis. For Weldability studies have found no major concerns with joining
the failure analysis, the primary features of interest were the fracture Grade 91 steel, provided low hydrogen welding consumables and proper
eppearanee, cracking/failure location, and failure mechanism. The welding practices are used. However, Modified 9Cr-lMo weldments are
fracture appearance was =lack of fusion" where the fracture surface susceptible to cracking at the edge of the HAZ, Type IV cracking which
nominallyconforms to the weld bead shape. The primary cracking and can lead to premature stress rupture failures under conditions of tensile
failure was located at the fusion interface of the weld repair. The failure loading normal to the weld fusion interface (5, 6, 7, 8). Although initial
mechanism was creep rapture. Based on the thermal and stress analysis, laboratory studies indicated that hot-cracking was not a problem with the
the probable root cause of failure was improper location of the weld Grade 91 alloy, subsequent weld qualification tests for thick-section
fusion line during the design stage. Both the operating stress and welds have shown that hot-cracking may occur for certain heats of filler
temperature were higher than design. The operating temperature was metal (7).
higher than the design temperature because of the lower cooling of the In order to insure adequate fracture toughness of the Heat-Affected-
rotor experienced in serviee than predicted, Thus, the HAZ was located Zone (HAZ) and to reduce the potential for stress relief cracking in the
at a radial position having higher than allowable temperatures and CrMoV rotor material, specialized welding techniques are used such as
stressesfor the creepweak HAZ. temper bead welding to refine the grain size of the HAZ. Several studies
have shown premature stress rupture failures for both CrMoV and
NiMoV rotor repair welds using Modified 9Cr-IMo weld metal (1, 9).
INTRODUCTION Because the creep rapture strength of the repair welds is less than that of
Weld repairs of high temperature steam turbine rotors have been the base material, a stress analysis is usually performed for all turbine
routinelyperformed for many years (1, 2,3 ). The major concerns for the rotor weld repairs. The objective oft.he stress analysis is to ensure that
weld repair are the weldabilty and strength of the weld metal and the the HAZ is positioned in a low stress region, i.e.; at stress level below the
fi'aeture toughness and creep strength of the rotor HAZ/weldment. allowable stress for the HAZ at the operating temperature. The allowable
Typically, the turbine rotor base material is CrMoV steel and the weld stress of the repair weldment is usually based on uniaxial cross weld test
metal is Modified 9Cr-lMo. Modified 9Cr-lMo steel, or Grade 91 data and is a lower bound value.
material, is a ferritie alloy that has attractive mechanical, physical, and

51 Copyright © 2004 by ASME

1 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab

 bead deposition process (cold wire GTAW) to refine the grain size of the
Heat-Affected-Zone(HA,Z). The objectives of the HAZ grain refinement
were to minimize the potential for stress relief cracking, and to develop
good fracture toughness. Typically, the Post Weld Heat Treatment
(PWHT) is performed at temperatures of approximately650°C in order to
maintain the creep strength of the rotor base material.
The focus of the current study was the portion of the failed rotor
repair weld that is located remote from the major failure position,
approximately 90 ° circumferenfially. Because the cracking and damage
is less advanced at this position, it may be possible to more accurately
characterize the nature of the creep damage and creep cavitation at the
earlier stages of failure.
Photographs of the rotor repair"weldment removed from the IP rotor
Figure I. Drawing of the IP turbine showing the first is shown in Figure 2. The cross section of the ring sample has a
stage blades. trapezoidal shape with nominal dimensions of 48 mm in the axial
direction, by 19.7 mm on the leading edge of blade, and 21.6 ram on the
trailing edge of the blade. The thickness of the weld repair build-up on
the nng sample is approximately 9.5 rnm. This section is about half of the
total ring section that was removed for failure analysis. The principal
fracture was located in the other half of the ring and the results of the
metallurgical examination at the major fracture were given by Pratt (10).
Visual examination was performed. As seen in the photograph, the
axial entry blade attachments had been removed by machining.
Ultrasonic inspection had been performed and the regions containing
indications/refleetors were marked by the lines seen in Figure 2. The
indications were located on the trailing edge of the blade. The damaged
Figure 2. Documentation of failed IP rotor weld repair region was approximately 12 nun wide in the axial direction, and the
weldment in as-received condition. Cross-hatched area indications were found at all circumferential locations inspected. The
has indications of linear discontinuities detected by cracks adjacent to the fusion interface were visible on the saw cut ends of
ultrasonic testing. the ring sample.
Based on these results, specimens were taken from each end of the
This paper describes the results of a metallurgical evaluation ring sample for metallurgical examination. The sections were located
performed for a failed IP turbine rotor repair weld that was removed from nominally 90° from the major fracture and were approximately 25 mm
a fossil power plant. Metallurgical techniques utilized for this study were and 225 nan in length. Metallurgical mounts were made at both ends,
visual examination, fractography, hardness testings and optical designated A (25 mm long section) and B (225 mm long section). For
metallography. At the time of the failure, the rotor had accumulated location A, the following three mounts were made a'ansverse to the
approximately eight years of service. The objectives were the following: welding direction: (1) full width, (2) cracked half at trailing edge, and O)
(1) to determine the contribution/role of the weldment microstrueture in apparently sound leading edge region. For location B, the following
the cracking and failure, and (2) to document the creep damage remote mounts were made: (1) transverse for cracked half at a'ailing edge, (2)
from the major fracture. This paper gives results of the metallurgical transverse for leading edge region, and (3) parallel to welding direction
examinations, discussions, and conclusions. near the centerline of the crack. In addition, a fractography specimen was
prepared by grinding on both sides of the crack and fi-acturing the
remaining ligamant.
M E T A L L U R G I C A L EXAMINATION RESULTS Metallographic specimens were prepared using mechanical
The history of the fossil unit and IP turbine/rotor is of interest. The polishing techniques to obtain surfaces suitable for etching. Multiple
fossil power plant began service in 1959. In 1976, a new Intermediate polish/etch cycles were utilized to enhance creep cavitation detection.
Pressure/Low Pressure (IP/LP) replacement rotor was installed. Because For the etching process, either 2% nital alone or 2%nital followed by
of fatigue cracking and rotor steqale damage, the rateau (or first stage) of picric acid-HCl was used.
The microstructural features of interest are the microconstituents,
the IP rotor was weld repaired in 1992. The unit was designed for 566°C
hardness, grain size, base metal inclusion content, and cracking. The
operation but had been derated to 552°C, implying nominally lower
weld bead pattern and rnacro-sa'ueture of the (HAZ) are shown in Figures
operating temperatures for the repaired rotor compared to the original.
3 and 4. The primary cracking is at the fusion interface but there are
After approximately eight years of service, the weld repaired rateau or
secondary cracks in the HAZ remote from the fusion interface. There is a
first stage failed in 2000.
region consisting of a mixture of diluted weld metal and unmixed base
The weld repair performed in 1992 was a 360 degree all weld metal
material adjacent to the fusion interface in the weld metal. Based on the
buildup. Figure 1 is a drawing of the IP turbine showing the location of
composition limits provided by etching with nital alone, the maximum
the first stage turbine blades. The rotor was specified as ASTM A-470
chromium content of the diluted weld metal is estimated to be
Class 8, 1Cr-IMo-1/4V forging material, and the weld metal was
approximately 5%.
Modified 9Cr-IMo. The repairprocedure consisted of a controUed weld

Copyright © 2004 by A S M E 52

2 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab

 Figure 3. Cross section parallel to welding direction
showing continuous crack at fusion interface.

The HAZ slrueture is periodic as seen in Figures 3 and 4. The

periodic structure is the result of the multi-pass, overlapping weld beads,
used in the GTAW controlled deposition welding process. The
mierostrueture at the fusion interface is primarily fine grained martensite
but there are regions of coarser grained martensite. Details of the various
miero-struetures and doettmentation of the creep rupture cracking are
provided in Figures 5 and 6.
Hardness measurements were taken for the base metal, weld metal
and at various HAZ zones to detect possible soft zones soeh as
decarburization of the HAZ adjacent to the fusion interface and towards
the edge of the HAZ for Type IV failures. Three measurements were
taken at each site using a diamond pyramid hardness indenter with a 20
Kg load, converted to the Rockwell scale, and the average value is
reported. The hardness values for the weld metal are from 27 HRC to 30
HRC. The HAZ hardness values are from 25 HRC to 27 HRC adjacent
to the fusion interface and 93 HRB at the edge of the HAZ. The lower
hardness at the edge of the HAZ is an indication of thermal softening in
the over-tempered base material. The hardness of the rotor base material
is 97 HRB, below the typical value of 22 to 25 HRC for CrMoV rotor
forglngs (11).
The grain size of the HAZ and rotor base material was determined
by comparison with ASTM standard charts. The mieroslrueture of the
Figure 4. Cross secUons transveme to welding direction
base material is shown in Figure 7. Values of the ASTM grain size were
showing cracks at fusion interface and HAZ.
as follows: (1)1-2 for the base material, (2) 6-7 for the coarse grained
I-IAZregions adjacent to the interface, and (3)12-13 for the fine grained The ASTM inclusionclassification seherue is based on a qualitative
I-IAZ adjacent to the interface. Thus, the grain size of the rotor base analysis which consists of a number from I to 4 indicating the nominal
material was significantly coarser than the ASTM 4 to 7 typical for rotor density of inclusions, and a rating based on the inclusion thickness of
either thin or thick series. The inclusion content classifications of the
steels (11) and significant grain refinement was achieved by the welding
process even in the coarse grained HAL With regard to the failure, the rotor base material were the following: (1) 1 thin for sulfide type using
Chart A, (2) 2 thin for the alumina type using Chart B, (3)1 thin for
creep properties of fine grained material are usually below those of
silicate type using Chart C, and (4) 2 thin for the globular type oxides
coarse grained material.
The inclusioncontent of the IP rotor base materialwas characterized using Chart D. Thus, the rotor base material is considered relatively
using the ASTM standard rating. The motivation for this characterization clean based on the observed inclusion elassificatious.
is based on the results of creep rupture studies. In several instances, With regard to documentation of the failure and cracking, the
uniaxial creep testing of weldments has shown that the creep duetilityand principal features of interest include the following; fracture appearance,
life of the intereritical/fine-gralned HAZ of low alloy steels correlates the cracking-failure location, and failure mechanism. The failure
mechanism is determined to be creep rupture based on the numerous
with the inelusinncontent (8).
intergranular creep cavities and cracks shown in the figures. The fracture
appearance was "lack of fusion" with the fi'acture surface nominally

53 Copyright © 2004 by ASME

3 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab

 Figure 6. Creep cavities in HAZ.
Figure 5. Creep rupture cracks at fusion interface and HAZ.
oriented normal to the radial direction. However, cracks oriented normal
conformingto the weld bead shape. It should be empasized that the "lack to the ho.op direction were also detected during examination ofthe mount
of fusion" (LOF) appearance is due to the close alignmentof the franture parallel to the welding direction. This cracking pattern is not un-
path to the fusion line and does not imply an initial failure to fuse. The expected based on the result of a typical stress analysis for a weld
fracture surface is shown in Figure 8. The fracture surface is oxidized and repaired rotor at steady state (1). In that ease it was fotmd that the
it is difficult to preserve features during cleaning. Thus, no detailed SEM resulting calculated stresses are biaxial with the radial component sfightly
examination was used. This same LOF fracture appearance was found greater than the hoop stress. In addition, it was found that the stress is
for the primary fracture (10) and is typically found for dissimilar metal relatively uniform in the axial direction, except at the edges of the
weld failures, longitudinal seam weld failures, and Type IV failures. buildup where slightly lower values were calculated.
In the axial direetiun, the cracks are only found through a portion of
the total thickness of 48 turn. The cracks were approximately 9.5 mm to
DISCUSSION 16 men in length and do not extend to the trailing edge; i.e., subsurface
The key aspects of the cracking and failure location are the by approximately 1.2 mm to 2.4 rnrn. The major cracks adjacent to the
following; (1) geometry; i.e., the axial/radial/eiremnferential position, fusion interface were oxidized, indicatingthat they were connected to the
relative to applied stress and operating temperature, and (2) external surface.
microstroetural; i.e., the zone of HAZ material that is cracked relative to As seen in the macro- and micrographs, the primary ereop
material ereap strength/ductility contributions. The major cracks were cavitation/cracking is located at the fusion interface. The cracking at the
located at the trailing edge of the turbine blade and parallel to the fusion fusion interface is present for both the fine grained HAZ and the coarse
interface; i.e., in the axial-hoop plane. The cracks are predominantly grained HAZ. There are also creep cavities at the edge of the refined

Copyright © 2004 by ASME 54

4 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab

 100

.......... ~ ............ ~, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ ...............

tO

(
14.1

LU

0:1
4go 500 5'I0 520 530 54.0 550 560
TEMPERATURE - C

Figure 9. Calculated rapture life of the rotor repair

Figure 7. Microstructure of rotor base material. weldment as a function of stress and temperature.

design of the weld repair. The IP turbine uses steam cooling to reduce
the operating temperature of the rotor. The thermal analysis indicated
that the operating temperature depends strongly on the amount of steam
cooling. Thus, a series of lives are predicted dependingon the amount of
steam cooling. For the ease of the initial as.designed steam cooling~the
calculated operating temperature is approximately 20°C less than the
temperature used in the original design life assessment. For the ease with
nozzle exit steam recirculation, the calculated operating temperature is
approximately 70C less than the temperature used in the initial design.
The service life of eight years is also shown in the figure. The shortest
life is for the ease of no cooling. The metal temperature is the steam
temperature and the expected life is approximately one half year. Based
Figure 8. Lack of fusion fracture appearance. on the stress analysis study, the following conclusions were made; (I) the
weldment properties were as expected, (2) the stress levels are higher
HAZ. However, no evidence of creep damage was found atthe Type IV than expected, (3) the design bulk operating steam temperature is sightly
position at the edge of the HAZ. Interface fractures are typically lower than expected, and (4) the temperature in repair area of the fusion
associated with inclusionsand/or second phase particles at the interface interface is higher than expected and is the more critical parameter than
and creep strength mismatch. stress. Thus, the probable root cause of failure is design of the repair
The variables that determine the service life for failure by creep weldment. The primary factor is the position of the HAZ at a radial
rupture are the operating temperature, the applied stress, and the rupture position having higher than designed operating temperature and stresses.
strength of the component material; i.e., CrMoV weldment. Thus, to fully As a corrective action, a temporary repair and refurbishment
evaluate the significance of the results, integration of the damage and approach was taken. Since the rotor base material beneath the repair
failure position with the results of thermal and stress analysis using the weld was not cracked or damaged and the principal function of the failed
creep rupture properties of weldmcnts is necessaW. stage was alignment, it is possible to remove this stage fi'om the rotor
Following the failure, a detailed stress aad thermal analysis was without significant effect on operation. The principal cost is a decrease
performed by the OEM and the results given to the utility (12). The in efficiency of approximately 8%~ Thus, a ring of material containing
calculated stress and temperature are used to predict the life of the weld the failure was removed and the rotor was returned to servicewithout this
repaired rotor. For the life prediction, the rupture lives of the HAZ stage following machining and balancing.
material used in the analysis are based on the OEM's d~ta base and are
significantly below the rapture lives for minimumstrength CrMoV rotor
base material at the stress and temperature conditions of interest. The life CONCLUSIONS
prediction results of the post failure analysis were compared to those for Based on the results of the metallurgical examination and evaluation
the initial weld repair design. of the failed IP rotor weld repair, the following conclusions were made:
Figure 9 shows the calculated rupture life of the rotor repair
weldment as a function of stress and temperature. A single life is given The cracking was located towards the trailing edge of the
at a stress of $0 for the initial design analysis performed in 1992. Results turbine blade and parallel to the fusion interface in the
of the current stress analysis indicated that the calculated operating stress, axial-hoop plane.
$1, is approximately 30 MPa greater than the value used in the initial

55 Copyright © 2004 by ASME

5 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab

 The primary cracking is located at the fusion interface and Conference on Advances in Life Assessment and Optimization of Fossil
adjacent rotor HAZ - both fine grained and coarse grained Power Plants, EPRI, Pale Alto, CA.
H A Z regions. There were secondary creep rupture cracks
and cavities at the edge of the fine grained HAZ. 10. Pratt, S. E., 2000, "IP Turbine Rotor Failure," TVA Report No. 20-
135, Tennessee Valley Authority, Chattanooga, TN.
The creep cracks are primarily oriented normal to the
radial direction. Cracks oriented normal to the hoop 11. NIMS, 1979, "Data Sheets on the Elevated Temperature Properties of
direction are also observed indicating that the applied 1 Cr-lMo-0.25V Steel Forgings for Turbine Rotors and Shafts (ASTM
stresses are biaxial. A470-8)," NIMS Creep Data Sheet No. 9A, National Institute for
Materials Science, Tokyo, Japan.
The fracture appearance is "lack of fusion" with the
fi'acture surface conforming to the weld bead shape. 12. Roberts, B. W., 2001, Private Communication, Chattanooga, TN.

The failure mechanism is creep rupture.

The probable root cause of failure is design. Both the

operating stress and temperature were higher than design.

REFERENCES
1. Bain, B., 1994, "PEPCO Morgantown Unit 1 HP/IP Rotor Weld
Repair," Welding and Repair Technology for Fossil Power Plants,
Conference Proceedings, EPRI, Palo Alto, Ca.

2. Driver, T. L., Amos, D. R., and Clark, R. E., 1994, Welding of

Combustion Turbine Rotors and Disks," Welding and Repair Technology
for Fossil Power Plants, Conference Proceedings, EPRI, Pale Alto, Ca.

3. Roberts, S. R., Boyles, P. R., and Silveri, P. J., 1994, "Repair of a

Turbine Rotor for 1050 F Service by Inserting a Rotor Section with Weld
Build-Up, ~ Welding and Repair Technology for Fossil Power Plants,
Conference Proceedings, EPRI, Pale Alto, Ca.

4. DiStefano, J. R. and Sikka, V. K., 1986," Summary of Modified 9Cr-

1Me Steel Development Programs: 1975-1985", ORNL-6303, Oak
Ridge National Laboratory, Oak Ridge, TN.

5. Brinkman, C.IL, Sikka, V. tC, Horal¢,L A., and Santella, M. L., 1987,
"Long-Terra Creep-Rupture Behavior of Modified 9Cr-lMo Steel Base
and Weldment Behavior," ORNIJTM-10504, Oak Ridge National
Laboratory, Oak Ridge, TN.

6. B. W. Roberts and D. A. Canonico, 1988, "Candidate Uses for

Modified 9Cr-lMo Steel in an Improved Coal-Fired Power Plant,"
Conference Proceodings: First Intema~onal Conference on Improved
Coal-Fired Power Plants, EPRI Report CS-5581-SR, Electric Power
Researeh Institute, Palo Alto, CA.

7. Ellis, F. V., Henry, J. F., and Roberts, B. W., 1990, "Welding,

Fabrication and Service Experiences with Modified 9Cr-lMo Steel,"
New Alloys for Pressure Vessels and Piping, ASME, NY, NY, pp. 55.-63.

8. EPRI, 1997, "Review of Type IV Cracking in Piping Welds," TR-

108971, EPRI, Pale Alto, C&

9. Ellis, F. V., Roberts, B. W., and Zielke, W. H., 2002, "Creep-Rupture

Properties of Turbine Rotor Repair Weldments," EPRI International

Copyright © 2004 by ASME 56

6 Copyright © 2004 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/pvp2004/71377/ on 03/30/2017 Terms of Use: http://www.asme.org/ab