Conference on High Strength Steels for Hydropower Plants - Takasaki

Qualification Program for the Rehabilitation of Cleuson-Dixence:

Requirements and Results
H. Cerjak1, O. Chene2, N. Enzinger3, E. Roos4
1

Dr. Professor Emeritus, Graz University of Technology, Austria

2
Dr., Cleuson-Dixence, EOS, Switzerland
4
Ass. Prof., Graz University of Technology, Austria
3
Prof. Dr.-Ing., MPA Universitt Stuttgart, Germany

Abstract

Introduction

After the catastrophic failure of the Cleuson-Dixence

Hydropower Plant shaft in 2000 an intensive investigation
program revealed the main cause in the poor production
quality and insufficient non destructive evaluation. Although
the applied high strength steel material S890 was found to
fulfill the requirements for that application, social and
political arguments led to the application of steel grades with
lower strength level type S690QL for the lower, high
pressure part of the scheme and S500ML for the upper part.
The overall qualification program for this material and its
weldments, was developed and carried out at TU Graz and
MPA Universitt Stuttgart, will be presented in this
contribution.
Based on different welding processes, planned to be applied
in different sections of the shaft represented by different wall
thicknesses and materials, a test matrix was defined.
Microstructure, hardness, strength and toughness were from
main interest. Additionally to the requirements according to
valid standards critical crack sizes were determined and
verified not only for welds but also for repair welds. Results
obtained from small scale tests have been verified using wide
plate tests also containing defined flaws, and justified the
calculated bearing capacity, using the European SINTAP
procedure taking into account welding residual stresses.
Based on these test results the process parameters to be
applied were defined, suggested processes or filler materials
were rejected and specifications for non destructive
evaluation procedures were developed.
Experiences with the implementation are reported in
additional papers [1][1].

The main goal of the qualification program was to generate

quantitative data about the loading capacity of the material as
well as for the welds, selected for the reconstruction of the
shaft. A detailed description of the reconstruction works is
given by O. Chne in [1]. In this paper the specifications for
the materials selected as well as for the welds are described in
more detail.
In the qualification program the mechanical properties of the
plate materials and welds planned to be utilized in the new
scheme, shall experimentally be determined. This, by
applying conventionally mechanical technological tests as
well as fracture mechanics investigations: for the fracture
mechanics investigations small scale samples as well as wide
plates based on international accepted approaches were used.
The results from that investigation finally should serve as
input data for the calculation of hypothetic critical failure
sizes and configurations. This with the goal to define the
safety margin between the failure sizes detectable by the
nondestructive testing methods (NDT) applied and the
maximal bearable failure size.
The approach selected for NDT and the result revealed are
described by C. Faessel in [1]. By using that approach,
failures which were made during the planning and fabrication
of the original schemes which finally led to the catastrophic
event should definitely be prevented [3][4][5].
Materials investigated were high strength steels S500ML
with 25 and 50mm thickness and S690QL with 70mm
thickness from Dillinger Htte.

Planning of the Qualification Program

The planning of the qualification program was a result of

2-1

Conference on High Strength Steels for Hydropower Plants - Takasaki

intensive discussions between the owner, the fabricator and

the research institutions and is based on the Qualification
Program S890, which were performed to explain the causes
of the catastrophic failure happened in the year 2000. [5]
In a first step a test matrix based on the materials and
processes used for the rehabilitation of the pressurized shaft
was established which is shown in Table 1.
TABLE 1: MATERIALS AND PROCESSES CONSIDERED IN THE
QUALIFICATION PROGRAM [6]

Processes
basemetal
SAW1
SAW2
GTAWHW
GTAWHWmanual
SMAW
FCAW

SAW 2submerged arc welding with two wires

GTAW HWsemi automatic gas tungsten arc
welding with hot wire
GTAW HW manualmanual gas tungsten arc
welding with hot wire
SMAWshielded metal arc welding
FCAWflux cored arc welding

Depending on the materials, the wall thickness and the

accessibility in some cases post weld heat treatment (PWHT)
was applied. The heating rate was limited to 50C/hour to a
maximum temperature between 530 and 580C and a holding
time of 2 minutes per millimeter wall thickness.
Table 1.

Materials
S500ML
S690QL
25mm 50mm 70mm
x
x
x
x+PWHT
x+PWHT
x
x
x
x
x
x+PWHT
x

The test welds were performed by the fabricator under

original shop, respectively site conditions and can be
evaluated as fully representative for the rehabilitation work.
TABLE 2: CHEMICAL COMPOSITION OF INVESTIGATED
MATERIALS [7]

Materials selected
As described in [1] two grades of materials were selected:
steel S500ML (according to EN10025-4) for the upper part
and quenched and tempered steel S690QL (according to
EN10024-6) for the lower, high pressure part of the scheme.
To maintain a high toughness level, in the specifications the
trace elements S and P were, in relation to the EN standard,
remarkably restricted, the requirements for Charpy V
toughness were dramatically increased and additionally for
all plates the requirements for quality Z35 according to
EN10164 for reduction of area on tensile samples tested
perpendicular to the plate surface was requested. To cover the
materials behavior of the most critical positions in the shaft
the plate thickness foreseen were selected 70mm for S690QL,
25 and 50mm for the S500ML, respectively. Table 2 shows
the chemical analysis of the plates investigated. The plates
were supplied by Dillinger Htte.
Welding
Different welding procedures were proposed by the fabricator
to weld the necessary 12 km welds to be performed in the
shop, in front of the tunnel at site, as well as in the tunnel.
The qualification program covered all processes planned to
be applied. In the case of non sufficient results of the
properties of the weld certain measures were introduced
either to change the filler metal used or to disallow certain
processes from application.
The following welding procedures were investigated:
SAW 1submerged arc welding, single wire

2-2

S500ML S690QL
Heat# 65440 285196
C
0.09
0.16
Si
0.33
0.24
Mn
1.42
1.12
P
0.013 0.008
Si
<0.0005 <0.0005
Cr
0.025
0.29
Mo
0.01
0.27
Ni
0.42
0.11
Al
0.019 0.024
Cu
0.32
0.022
Ni
0.0042 0.0021
V
<0.001 0.0014
Nb
0.0254 0.0246
Ti
0.0025 0.0013
B
0.0004 0.0016
Zr
<0.0010 <0.0010

Investigations
For each of the chosen materials and processes several tests,
using small scale samples [6] as well as wide plate tests [7]
were performed. The following list gives a short overview on
the required and performed tests from [6]and [7]:
Microstructure and hardness was determined by
metallographic documentation of welding sequence

Conference on High Strength Steels for Hydropower Plants - Takasaki

and determination of hardness in three different

zones across the weld.
Tensile test according [9],[10]: test specimens were
taken in longitudinal as well as in perpendicular
direction to the weld. All tests were performed at
0C. All data from elastic and plastic deformation up
to reduction of area were determined.
Charpy-V testing. Samples according [11][12]:
Testing was performed with 12 samples for each
condition at different temperatures covering lower
shelf and upper shelf regime. Evaluation of tests was
performed according [13]
Fracture mechanics: basically CT20 Specimens
were tested; CT 10 samples were only used for
25mm plate thickness. A fatigue crack according
[14] was established on the CT-sample which
afterwards was side-grooved. All fracture mechanics
tests were performed at 0C under COD control
using the unloading compliance technique.
Uncracked samples were heat tinted at 280C for
one hour and opened at liquid nitrogen temperature.
The fatigue crack was documented with a
measurement-microscope and blunting at the crack
tip (SZW) was quantified using scanning electron
microscope [15]. Difficulties in the evaluation of
ductile initiation prior to cleavage failure were
conservatively estimated using the secant method.
Although cyclic loading is not a significant part in
the load spectra of a pressurized shaft, an estimation
of crack growth rates was performed on the base of
fatigue pre-cracking of fracture mechanic specimens
in air.
Wide Plate Tests: To evaluate the integral strength
of a real weld several wide plate tests with initially
implemented cracks were tested.
Weldability: Additionally to the investigations of
real welds in [6] and [7], physically simulated base
materials,
using the GLEEBLE machine,
representing different areas in the heat affected zone
(HAZ) describing the weldability were examined
(metallography and hardness, tensile test, Charpy-V
toughness) [8]. Therefore both materials were
loaded by applying different thermal cycles
representing extreme conditions due to welding
processes applied. Two different heat inputs of 10.1
and 28.8 kJ/cm representing t8,5 times of 5 and 23
seconds were used in combination with peak
temperatures of 1300C, 1000C and a double cycle
with peak temperatures 1300 and 1000C.

Results
Because of the large amount of data collected only some
results will be discussed.
In general it has to be reported that the qualification program
revealed very important and selective results regarding the

2-3

question and helped to select the acceptable welding

procedures and to optimize the welding consumables mainly
regarding the toughness behavior of the weld deposit.
Microstructure and Hardness
Fehler! Verweisquelle konnte nicht gefunden werden.
represents the hardness distribution across S690 QL semi
automatic GTAW hot wire in the as welded condition.
Despite of the fact that no PWHT was applied the maximum
hardness measured is lower than 300 HV. As expected a
limited loss of hardness near the HAZ in the cross section
could be observed.

FIG. 1: S690QL,

HARDNESS IN
WELDED CONDITION

GTAW HW

WELD IN AS

Tensile Strength
Fig. 2 shows the stress-strain curves obtained on welds
performed by different weld procedures. All processes
applied fulfill the requirements. Some weld deposited shows
overmatching behavior.

FIG. 2: STRESS STRAIN CURVES FOR DIFFERENT WELDS.

Charpy V Testing
Fig. 3 reveals the Charpy V toughness observed on welds
performed by different processes as well as that of the base
materials. It can be observed that the weld metal produced by
the different welding procedures show significant differences
in the Charpy V behavior.
Also the HAZ Charpy V toughness is influenced by the

Conference on High Strength Steels for Hydropower Plants - Takasaki

different welding processes (see Fig. 4).

FIG. 3: S690QL, CHARPY TOUGHNESS FOR DIFFERENT WELD

DEPOSITES AND THE BASE MATERAL (GW)

FIG. 5: S500ML, J-INTEGRAL OF DIFFERENT WELDS AND BASE

MATERIAL (GW)
Wide Plate Tests

J/

V
y
p
ra
h
C

temperature T/C

FIG. 4: S690, CHARPY TOUGHNESS IN THE HAZ OF DIFFERENT

WELDS.
Fracture Mechanics testing
FIG. 5 represents crack resistance curves, as an example on
material S500ML, measured at the different weldments.
Similar to the observations made during the Charpy V impact
tests clear distinction between the toughness behavior of
different weld deposit could be observed.

The wide plates clearly proofed the findings and conclusions

drawn from the small scale specimens. For example from the
small scale specimen results it has been concluded that the
GTAW-HW weld of S500ML does not reach a sufficient
quality contrary to the S500ML weld SAW1. This is clearly
verified by the wide plate tests in full wall thickness.
Fig. 6 shows the results of three wide plate tests of two
different S500ML-welds. The specimen 8 WP1 is a wide
plate from SAW1 of S500ML with a cross section of 300 *
50 mm2, through wall crack (center crack) with a length of
121 mm. Wide plates 9WP1 and 9WP2 are GTAW-HW with
a nearly identical grosss section (300 * 47 mm2), where
9WP1 has a through wall crack with a length of 122 mm,
9WP2 has a semielliptical surface crack with 149,4 mm *
25,8 mm (2c * a). Wide plate 8 WP1 confirmed a satisfying
load-deformation capability of the SAW1 weld, where as
both specimens 9WP1 and 9WP2 failed in the linear elastic
range showing a clearly insufficient toughness behavior.

wide plate tests S500ML, s = 50 mm

7000

8WP1, SAW1, through wall crack

6000

Load F / kN

5000

4000

3000

9WP2, GTAW-HW, surface crack, cleavage

2000

9WP1, GTAW-HW, through wall crack, cleavage

1000

0
0

10

COD / mm

FIG. 6: S500ML, LOAD-CRACK OPENING DISPLACEMENT

BEHAVIOUR OF WIDE PLATE TESTS

2-4

Conference on High Strength Steels for Hydropower Plants - Takasaki

Weldability

Summary

As a matter of fact weldability investigation showed good

results for S500 ML, whereas the investigation of S690QL
showed some remarkable influence of the post weld heat
treatment on Charpy-V toughness.
While single cycles with low peak temperature and double
cycles, representing multi pass welding improved behavior
after PWHT, cycles with high peak temperature showed a
shift of transition temperature to higher values including a
low level of upper shelf.

With this Qualification Program, first time ever a consequent

investigation of materials and welding processes was applied
to evaluate the overall loading capacity as precaution of
construction work of a hydropower plant pressurized shaft.
Not only standards but also up to date scientific knowledge
and testing methods were considered. Based on the results of
the extensive experimental test program critical crack sizes
according to the R6 method were determined to define
absolute necessary requirements for the non destructive
evaluation procedure and determine the distance between
detectable and critical failure size.
Due to the good correlation and integrity of all tests,
especially the good agreement between Charpy V, fracture
mechanics, small sample tests and wide plate tests, the results
are judged to be reliable. Based on this high level of
trustworthiness the original foreseen welding procedures
SMAW for S690QL and GTAW HW for S 500 ML were,
due to inadequate values, excluded from production process.

Critical Crack Size

Based on the extensive investigation of materials and
production processes the fracture behavior described by crack
initiation values Ji, verified by means of wide plate test
results were used in the so called R6 approach [16] to
estimate critical crack sizes at different locations of the
pressurized shaft. Results with and without considering
welding residual stresses have been compared.
Fig. 7 gives an example of calculated crack sizes for different
crack shapes (a/t) on a certain position in the shaft. Compared
with the applied pressure at this location the critical crack
dimension (67mm) can be read.
Minimal critical flaws are predicted in longitudinal direction
in the size of 98mm (S500ML) and 67mm (S690QL).

References
[1]

[2]

[3]
[4]

a
P
M
/
er
u
ss
er
p
alc
it
ir
C

[5]

[6]

[7]
Flaw dimension /mm

FIG. 7. S690, CRITICAL FAILURE SIZE FOR LONGITUDINAL

[8]

CRACK

Considering welding residual stresses these theoretical values

are as low as 15mm in longitudinal direction of S500ML and
21mm in circumferential direction of S690QL areas.
Critical calculated crack sizes in longitudinal direction of
S690QL becomes larger than in circumferential direction due
to the fact, that longitudinal welds are stress relieve annealed,
leading to an assumed drop of residual stresses to 20% of
yield strength, and circumferential welds are always
performed on site, where PWHT is not possible and therefore
show maximum residual stress level.

2-5

[9]

[10]
[11]
[12]
[13]

[14]

O. Chene, Welding Processes for the Cleuson Dixence Shaft.

Conference on High Strength Steel for Hydropower Plants, Takasaki
2009
C. Faessel, Non Destructive Testing of Cleuson-Dixence shaft.
Conference on High Strength Steel for Hydropower Plants, Takasaki
2009
H. Cerjak, N. Enzinger (edts), High Strength Steels for Hydropower
Plants. Graz, 2005
Bllinghaus, T.; Cerjak, H.-H.; Enzinger, N.; Dorsch, T.; Saarinen, K.:
Results of Stress Corrosion Investigation on Steel and Weldments on
Steels S890 and S690. - in: High Strength Steels for Hydropower
Plants. (2005)
Cerjak, H.-H.; Dimmler, G.; Enzinger, N.: Steel S890 for pressurised
shaft CD: Review of the Metallurgical Investigations. - in: High
Strength Steels for Hydropower Plants. (2005)
Eisele, U., Silcher, H., Schuler, X., Roos, E.: Untersuchungsprogramm
zur Durchfhrung der Integrittsbewertung einer Druckrohrleitung,
Teil 1, , Untersuchungsbericht 901 2459 000 Dr. El/Dr. Sil
Materialprfungsanstalt Universitt Stuttgart. 11.04.2007
Stadtmller, W., Wiedemann, J., Schuler, X., Roos, E.:
Untersuchungsprogramm zur Durchfhrung der Integrittsbewertung
einer
Druckrohrleitung,
Teil2:
Groplatten-Zugversuche,
Untersuchungsbericht
901
2459
007
Dr.
Sr/Wd/Hel,
Materialprfungsanstalt Universitt Stuttgart. 11.04.2007
H. Cerjak, N. Enzinger, P. Mayr, Chr. Gollowitsch. Weldability of steel
types S500ML and S690QL, Expertise 42-06, Institute for Materials
Science and Welding, Graz, 30.1.2007
DIN 50125: Prfung metallischer Werkstoffe, Zugproben, Richtlinien
fr die Herstellung. Normenausschuss Materialprfung im DIN e.V.
Beuth-Verlag, 1991
DIN 10002, Teil 1: Prfung metallischer Werkstoffe, Zugversuch,
Prfverfahren Normenausschuss Materialprfung, Beuth-Verlag, 1991.
DIN EN 10045-1: Kerbschlagbiegeversuch nach Charpy
DIN EN ISO 14556:2000 Kerbschlagbiegeversuch nach Charpy
(V-Kerb), instrumentiertes Prfverfahren. DIN e.V.
G. Hofer, C. Hung, U. Gnes. Eine mathematische Funktion fr die
Beschreibung der Ergebnisse aus Kerbschlagbiegeversuchen.
Werkstofftechnik
ASTM E 1820-05 Standard Test Method for Measurement for Fracture
Toughness. Beuth Verlag GmbH, Auslands Normen-Service.

Conference on High Strength Steels for Hydropower Plants - Takasaki

[15] ISO 12135: Metallic Materials Unified Method of Test for the
Determination of Quasitstatic Fracture Toughness. 1st edt. 2002-12-01,
Geneva, Switzerland
[16] R6, Rev 4. R-Code Software for Assessment the Integrity of Structures
Containing Defects, Version 4.3, British Energy, 31.03.2006

2-6

Conference on High Strength Steels for Hydropower Plants - Takasaki

2-7