Sie sind auf Seite 1von 12

Proceedings of the ASME 2015 Pressure Vessels and Piping Conference

July 19-23, 2015, Boston, Massachusetts, USA




Izumi Nakamura Naoto Kasahara

National Research Institute for Earth Science and The University of Tokyo
Disaster Prevention Bunkyo, Tokyo, Japan
Tsukuba, Ibaraki, Japan

ABSTRACT behavior of structures and to take adequate measures to prevent

After the accident at Fukushima Dai-ichi Nuclear Power the fatal failure mode.
Plant in the 2011 Great East Japan Earthquake, the International With the aim of clarifying the failure modes and the
Atomic Energy Agency (IAEA) requires to consider the design ultimate structural strength under extreme loadings, a research
extension conditions (DEC) for the safety management of project is conducted from 2012 to 2016 [1]. In this research
nuclear power plants (NPPs). In considering DEC, it is program, the very high temperature, pressure, and great
necessary to clarify the possible failure modes of the structures earthquake are considered as the extreme loads.
and their mechanism under the extreme loadings. Because In the investigation for seismic loads in this research
piping systems are one of the representative components of program, piping systems are selected as the possible failure
NPP, and there is a possibility to failure at seismic events, the position in nuclear power plants. Collapse is typical failure
authors conducted an experimental investigation on failure mode postulated by design rules [1]. On the other hand, it is
modes and their mechanisms of piping systems under excessive well known that the major failure mode of piping systems
seismic loads. under large seismic loads would be the fatigue failure from a lot
The experiments are categorized into the fundamental plate of previous studies [2-12]. Furthermore other failure mode such
tests and pipe component tests. In this paper, the results of the as ratchet-buckling failure had been reported [5], though it may
pipe component tests would be described. In the pipe be a rare case. In consideration of such background, a series of
component tests, the authors used piping specimens constituted experiments is planned to investigate the failure modes and the
with one steel elbow and a weight. Though the input process to failure of piping systems under excessive seismic
acceleration level was much over the allowable level to prevent loads. This paper summarizes the experimental results of the
collapse failure by the seismic design, the failure mode shaking table tests on steel elbow pipe specimens.
obtained by the excitation tests were mainly the fatigue failure.
The reduction of the dominant frequency and the increase of
the hysteresis damping were clearly observed in the high-level EXCITATION TESTS ON ELBOW PIPE SPECIMENS
input acceleration due to the plastic deformation, and they The experiments for seismic load in this research program
affected the specimens' vibration response greatly. are constituted mainly in two parts [1]. One is the fundamental
plate test using plate-type test pieces. The other is the excitation
tests on pipes, which are more realistic structural model. In
INTRODUCTION consideration of the DEC, it is necessary to evaluate the effect
After the accident at Fukushima Dai-ichi Nuclear Power of the elastic-plastic dynamic behavior of pipe structures and
Plant in the 2011 Great East Japan Earthquake, the importance the process to failure under the dynamic excitation input. So
of consideration for the design extension conditions (DEC) is shaking table tests were conducted on steel pipe elbow
widely recognized. But it is not clarified the failure processes specimens in 2014 (Hereinafter this experiment is referred as
and the failure modes of the components of nuclear power pipe component tests). The main purpose of the pipe
plants under extreme loadings such as severe accidents and component tests is to obtain the dynamic behavior of pipe
huge earthquakes which exceeds the design level. In order to structures including the elastic-plastic region, and the failure
treat the DEC, it is necessary to understand the ultimate mode under the dynamic alternate excitation load.

1 Copyright 2015 by ASME

The test specimen consisted of one steel pipe elbow, one the pipe is mainly 100kg. It can be increased up to 120kg by
weight, and two straight pipes. The configuration of the adding the adjustment plates to the weight. Figure 2 shows the
specimen is shown in Fig.1. The material used in the natural frequencies and the vibration modes of the specimen
experiments are Carbon Steel JIS STS410 (Japanese Industrial made by STS410 pipe with a 100kg weight. One-directional
Standards: carbon steel pipes for high pressure service) and excitation tests were conducted by a shaking table. The
Stainless Steel 304SS. The material properties by the inspection intended deformation of the specimen is the in-plane bending
certificate reports and the design stress intensity of each deformation, which is the second vibration mode of the
material [13] are listed in Table 1. The size of the pipe is 50A- specimen.
sch40, that is, 60.5mm outer diameter (Do) and 3.9mm nominal
wall thickness (t). Long elbows are used as the specimens, and
the elbow curvature radius (R) is 76.2mm. Thus, R/Do is 1.26
and Do/t is 15.51 for these specimens. The weight at the end of

(Unit: mm)
Weight: 100~120kg Shaking direction

(a) Dimension of the test specimen (b) Test setup

Fig.1 Configuration of the test specimen

Table 1 Material properties by inspection certificate

y[N/mm2] u[N/mm2] El.[%] Sm*[N/mm2]
STS410 309 468 55 137
SUS304 251 566 60 137
* Sm: Design stress intensity determined by the design code

Acc. [m/s2]

1st mode: 3.03Hz 2nd mode: 3.04Hz 0

(Out-of-plane bending) (In-plane bending) -5

0 5 10 15 20 25 30

(a) Sine wave 01

Acc. [m/s2]


0 5 10 15 20
3rd mode: 18.2Hz
(b) Sine-wave 03
Fig.2 Natural frequencies and vibration modes
(Weight: 100kg, STS410) Fig.3 Input sinusoidal waves used in the excitation

2 Copyright 2015 by ASME

Five specimens were used in the experiment. Four of them are for the fatigue failure tests, and SLE05 is for the high-level
were the specimens to obtain the fatigue failure mode, which is input tests. The input waveform in the experiment was mainly
assumed the major failure mode in a seismic event from the the sinusoidal wave as shown in Fig.3 (a). The waveform
various previous studies. The other specimen was a trial consists of 30 steady amplitude cycles part and 10 cycles
specimen to obtain failure modes except for the fatigue failure transient parts at the beginning and the ending. This waveform
by the excitation which exceeds the design limitation level to a is referred as Sine-wave 01. The waveform of which the
large extent. The test conditions of the pipe component tests are numbers of cycles at the steady amplitude parts changed from
summarized in Table 2. The specimens named SLE01~SLE04 30 to 60 was also used in the fatigue tests depends on the
Table 2 Test condition of the pipe component tests
Internal Hoop stress by Input Intended failure
Name Material Weight
Pressure internal pressure waveform*1 mode
STS410 100 kg
SLE02 9 MPa 69.8 MPa
Fatigue failure
SLE03S 304SS Sinusoidal wave 100 kg / 110 kg*2
SLE04 13.5 MPa 104.7 MPa 100 kg
SLE05 9 MPa 69.8 MPa 120 kg Collapse
*1: Wide-band random wave and sinusoidal sweep were also conducted to obtain the vibration characteristics of specimens
*2: Weight of SLE03S changed from 100kg to 110kg in the middle of the iteration of excitations. Its reason is described in the
section "Outline of the test results and obtained failure mode"

Table 3 Summary of the test results of the pipe component tests

Dominant frequency Number of Number of cycles
Input Acc.*1 Input
Name and damping ratio excitation to cause failure at Failure mode
[m/s2] waveform*2
(In-plane bending mode) iteration 5m/s2 excitation*3
1.5 Sine-wave 01 7
2.0 Sine-wave 01 1
Sine-wave 01 3
SLE01 2.63 Hz / 0.82% 3.0 59
Sine-wave 02 3
4.0 Sine-wave 01 5
5.0 Sine-wave 01 2
Fatigue failure
1.5~4.0 Sine-wave 01 1 for each with ratchet
SLE02 2.63 Hz / 0.97% 176 deformation
5.0 Sine-wave 01 6
2.57 Hz / 1.25% 1.5~4.0 Sine-wave 01 1 for each
(at 100kg weight)
SLE03S Sine-wave 01 4 776
2.47 Hz / 1.07% 5.0
(at 110kg weight) Sine-wave 02 11
1.5~4.0 Sine-wave 01 1 for each
SLE04 2.63 Hz / 1.04% 184
5.0 Sine-wave 01 7
1.5, 3.0, 4.0
SLE05 2.43 Hz / 0.78% sweep 1 for each - (No failure)
9.0 Sine-wave 03
*1: The target acceleration.
*2: The frequencies of the input sinusoidal wave were tuned depending on the input acceleration level in each specimen as listed
in Table 4.
*3: The number of cycles at the steady amplitude parts of the maximum input acceleration (5m/s2). The cycles in lower excitation
levels and at the increase/decrease parts are not included.

3 Copyright 2015 by ASME

progress of the excitation tests. This waveform is named as maximum input accelerations were determined in consideration
Sine-wave 02. In the high-level input experiment, the of the elastic-predicted response of the specimens and the
waveform named Sine-wave 03 as shown in Fig.3 (b) was limitation of the test facility. As for SLE03S, the response
used. Because the consecutive excitations by the sinusoidal acceleration at the weight and the measured strain became to
input under the large acceleration may cause the environmental decrease as continuing the sinusoidal input. In order to increase
vibration problem around the facility, Sine-wave 03 has less the vibrational response of the specimen to obtain a failure
steady amplitude cycles than Sine-wave 01 and Sine-wave 02, mode, the additional weight was applied. So the weight
instead of applying higher input acceleration. Sine-wave 03 condition became 110kg in the middle of the series of
consists of 15 cycles transient part at the beginning, three excitations.
cycles of steady amplitude, and two decreasing amplitude Prior to the excitations by the sinusoidal input, excitations
cycles at the ending. The frequency of the input sinusoidal by a wide-band random wave (frequency band: 0.1-20Hz) at
wave was tuned in the excitation tests. The detail of the input 0.2m/s2 input level were conducted to obtain the dominant
frequency is described in the "Test Results" section. In addition frequency of the specimens in the elastic region. Figure 4
to the excitation by the sinusoidal input, a wide-band random shows the transfer function of SLE01 by the wide-band random
wave was used to obtain the vibration characteristics of the wave excitation. As the result of the wide-band random wave
specimen at the low level acceleration. Sinusoidal sweep excitation, the natural frequency of SLE01 was estimated as
excitations were also conducted on SLE05. 2.63Hz (in-plane bending mode), and the damping ratio was
Four specimens were pressurized to 9MPa with room about 0.82%. The transfer functions of other specimens were
temperature water. The internal pressure caused about 70MPa similar to the result of SLE01. The dominant frequencies listed
of hoop stress on the specimen. This hoop stress is about 23% in Table 3 are the frequencies obtained by the wide-band
of the yield stress for the specimens made of STS410, and random wave excitation. As for SLE03S, which is made of
about 28% of the yield stress for the specimen made of 304SS. 304SS, the dominant frequency is a little lower than the other
One specimens made of STS410 was pressurized up to specimens with 100kg weight due to the difference of the
13.5MPa. The hoop stress by the internal pressure is about Young modulus between the carbon steel and the stainless steel.
105MPa. That is about 34% of the yield stress. The excitation
tests were conducted at ambient temperature (about 10 oC).
Response acc. / Input acc.


Elastic-Plastic Response Behavior of Specimens 6
The excitation test results are summarized in Table 3. The
maximum applied input accelerations to the specimens were
5.0m/s2 for SLE01~SLE04, and 9.0m/s2 for SLE05. These 2

180 2.35 2.4 2.45 2.5 2.55 2.6 2.65 2.7
Frequency [Hz]
Fig.5 Relation between the input frequencies and

the amplification ratio under the 1.5m/s2
-90 sinusoidal input (SLE01)
0 2 4 6 8 10 Table 4 Input acceleration and frequencies of the
Frequency [Hz]
sinusoidal input for each specimen
Input Frequency of input sinusoidal wave [Hz]
50 f=2.63Hz Acc.
h=0.82% [m/s2] SLE02 SLE03S SLE04 SLE05
Amplification ratio

1.5 2.6 2.5 2.6
2.0 2.5 2.4 2.5
3.0 2.4 2.4 2.4
4.0 2.3 2.3 2.3
0 2 4 6 8 10 2.3 / 2.2
Frequency [Hz] 5.0 2.3 2.3
2.1 (at 110kg)
Fig.4 Transfer function of SLE01 by the wide-band
random input 9.0 2.0

4 Copyright 2015 by ASME

The dominant frequency of SLE05 is also a little lower than the suitable input frequencies corresponding to the input
other specimens due to the heavier weight. The dominant acceleration level in the experiment. SLE01 was used for this
frequencies of SLE01, SLE02 and SLE04 are almost same. The trial excitation test. The excitation tests on SLE01 were
damping ratios of each specimen were estimated around 1%. conducted as below; after the excitation by the wide-band
Provided the specimens are in the elastic region, random wave, SLE01 was excited by the 1.5m/s2 Sine-wave 01.
excitations by a shaking table to resonate the specimen can be The frequency of the input wave was set as 2.4Hz. This input
achieved by sinusoidal inputs at the specimens dominant frequency was determined as about 90% of the dominant
frequencies. But when the specimens reached to the plastic frequency of SLE01 by reference to previous studies [11, 12].
region, the specimens would not response under the input Because the sinusoidal excitations in these previous studies
frequency used in the elastic region because the dominant were conducted after several high-level excitations and the test
frequency would be reduced due to the plastic deformation of models used in those studies had experienced plastic
the specimens. Because the reduction ratio would depend on deformation before the excitation by the sinusoidal input, the
the input acceleration level, it was necessary to investigate the suitable input frequency which caused a large response to the
test models was lower than the dominant frequency in the
Table 5 Sinusoidal sweep excitation conditions in elastic region. But as for SLE01, the response was not so
SLE05 amplified by the excitation by the 1.5 m/s2-2.4Hz sinusoidal
input. This is because SLE01 remained in the elastic region
Input Acc. Frequency range of Sweep Sweep rate
before this first 1.5 m/s2 excitation, and the discrepancy of the
[m/s2] sweep excitation time [sec] [Hz/sec]
input frequency and SLE01's dominant frequency in the elastic
0.3 From 3Hz to 2Hz 80 region resulted in the small vibrational response amplifier. So
the trial excitations to search the suitable input frequency to
1.5 From 2.6Hz to 1.6Hz 80
0.0125 cause the plastic deformation were conducted on SLE01 by
3.0 From 2.7Hz to 2.2Hz 40 changing the input frequencies of Sine-wave 01 at 1.5 m/s2
excitation. Figure 5 shows the relation between the input
4.0 From 2.4Hz to 1.9Hz 40 frequencies and the amplification ratio of these trial excitations.
Here, the amplification ratio was defined as the ratio of
maximum response acceleration to the maximum input
90 acceleration.

0 As shown in Fig.5, 2.6Hz seemed to be the suitable input

frequency to get a relatively large response of the specimen at
the 1.5 m/s2 input acceleration level. The suitable input
-180 frequencies were changed as the input acceleration level
1 1.5 2 2.5 3 3.5 4 4.5 5
Frequency [Hz] increased due to the inelastic behavior of the specimen, so the
similar trial excitations were conducted at higher input
70 Random 0.2m/s2 acceleration level. Based on the results of these trial excitations
60 on SLE01, the input acceleration level and the correspondent
Amplification ratio

50 2
frequencies for SLE02~SLE04 were decided. The input
Sweep 0.3m/s acceleration and the frequencies of each specimen are
30 2
summarized in Table 4.
Sweep 1.5m/s
0 12
Response acc. / Input acc.

1 1.5 2 2.5 3 3.5 4 4.5 5

Frequency [Hz] 10 SLE03S
20 SLE04
2 SLE05
Sweep 3.0m/s 6
Amplification ratio

Sweep 4.0m/s 4

5 0
0 2 4 6 8 10 12 14
Input acc. [m/s2]
1.5 2 2.5 3 Fig.7 Relation between the maximum input
Frequency [Hz]
acceleration and the amplification ratio under the
Fig.6 Transfer function of SLE05 by the wide-band sinusoidal wave input
random input and sinusoidal sweep excitations

5 Copyright 2015 by ASME

In addition to the trial excitations on SLE01, sinusoidal almost flattened at the sweep excitation by the 4.0m/s2 input
sweep excitations with relatively large input acceleration were acceleration. It means that the specimen became insensitive to
conducted on SLE05. The purpose of the sweep excitations was the input frequency as the input acceleration level increased.
to search the transition of the dominant frequency and to obtain The dominant frequency at 4.0m/s2~5.0m/s2 was seemed to be
the amplified ratio under the elastic-plastic region. The about 2.3~2.2Hz from these excitation tests. The reduction of
excitation condition of the sweep excitations are listed in Table the dominant frequency and the amplification ratio is
5. Figure 6 shows the transfer functions obtained by the wide considered due to the large hysteresis behavior of the
band random wave at 0.2m/s2 input acceleration and the sweep specimens. Figure 7 shows the relation between the maximum
excitations of SLE05. As shown in Fig.6, the amplification input acceleration and the amplification ratio for all specimens
ratios considerably reduced as the input acceleration level under the sinusoidal wave input. (Note that the input accelerations
increased. The dominant frequency also decreased as the input in Fig.7 were the maximum accelerations measured on the shaking
acceleration level increased, and the amplification ratio became table by the accelerometer during the tests, and most of them were
4 4

3 3

2 2

1 1

Load [kN]
Load [kN]

0 0

-1 -1

-2 -2

-3 -3

-4 -4
-200 -150 -100 -50 0 50 100 150 200 -200 -150 -100 -50 0 50 100 150 200
D01 [mm] D01 [mm]
(a) SLE02, 0.2m/s2 input (Wide-band random) (b) SLE02, 1.5m/s input
4 4

3 3

2 2

1 1
Load [kN]
Load [kN]

0 0

-1 -1

-2 -2

-3 -3

-4 -4
-200 -150 -100 -50 0 50 100 150 200 -200 -150 -100 -50 0 50 100 150 200
D01 [mm] D01 [mm]

(c) SLE02, 5.0m/s2 input (d) SLE05, 9.0m/s2 input

Fig.8 Load-deflection curve of SLE02 and SLE05
25 200

Response Acc. [m/s2]

Response Disp. [mm]

SLE03S 50
5 SLE03S
0 0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Input Acc. [m/s2] Input Acc. [m/s2]
(a) Input acc. - Response acc. (b) Input acc. - Response disp.
Fig.9 Relation between the maximum input acceleration and the maximum response acceleration & displacement
at the weight (sinusoidal wave input)

6 Copyright 2015 by ASME

larger than the input accelerations listed in Table 3. This is because the Failure mode obtained by the excitation tests
shaking table tends to overshoot especially under the high target input The failure mode obtained by the excitation tests was the
excitation.) As shown in Figs.6 and 7, the amplification ratio at fatigue failure and deformation for SLE01~SLE04. Fig.10
1.5m/s2 input was about 11~12, that at 3.0m/s2 input was about shows the typical failure mode obtained in the excitation tests.
5, that at 5.0m/s2 was about 4, and that at 9.0m/s2 was As for SLE05, no significant damage nor deformation was
decreased to about 1.6, though the amplification ratio in the observed after the 9.0m/s2 input excitation by Sine-wave 03,
elastic region was over 50 as shown in Fig.4. and the excitation tests was finished without failure. The outer
Figure 8 shows the load-deflection curves of SLE02 and diameters at the elbow center section in the in-plane direction
SLE05. In these figures, the load at the weight was calculated (intrados extrados) and in the out-of-plane direction (crown
by multiplying the measured response acceleration by the mass crown) were measured before and after the excitation tests.
of the weight, and the deflection was the measured Figure 11 shows the deformation of the outer diameters for
displacement at the weight by a laser displacement meter. The each specimen. As shown in Fig.11, the outer diameter of
solid line shows the load-deflection relationship in the elastic SLE01~SLE04 were increased after the excitation tests in both
region obtained by the wide band random wave excitation, and directions. The diameter in the out-of-plane direction for
the dashed line shows the twice of the gradient to the elastic
region. The load-deflection curves after the 1.5m/s2 input show
the hysteresis behavior. In the vibration systems which show
the hysteresis behavior as shown in Fig.8, the hysteresis
damping can be calculated as the equivalent viscous damping
(heq) by the following Eq. (1);

1 W Through-wall crack
heq (1) Crack penetration and
4 We leak of water
(a) Fatigue crack at the (b) Penetration check test
Where W: the dissipated energy in one cycle (The area of flank of the elbow result
one cycle of the hysteresis loop), and We: the equivalent Fig.10 Typical failure mode obtained in the pipe
1 component tests (SLE02)
potential energy ( ke 2 , ke: equivalent stiffness, : the
2 65
deflection of the vibration system). SLE01
heq in each excitation was estimated by Eq. (1) and the SLE02
Outer diameter [mm]

load-deflection curves. As the results, heq at the 1.5m/s2 input 63
acceleration was about 4.9%. heq at the 5.0m/s2 input 62
acceleration was about 14.2%, and heq was up to more than 61
20% at the 9.0m/s2 input acceleration. Considering that the
damping ratio in the elastic region was about 1%, the effect of 60

hysteresis damping is significant in the elastic-plastic region. 59

Though the input acceleration level was much larger than the
collapse level determined by the design procedure as described Before After
in the following section, the actual response of the pipe
specimens under the dynamic alternative input was not so large (a) In-plane direction (intrados extrados)
due to such a large amount of hysteresis damping. 65
Figure 9 shows the relation between the input acceleration SLE01
and the response acceleration & displacement at the weight of SLE02
Outer diameter [mm]

SLE01~SLE05 under the sinusoidal excitations. The data at 63 SLE03S

1.5m/s2 input acceleration about 8.0m/s2 response 62
acceleration of SLE01 in Fig.9 (a) is the result by 2.4Hz input
frequency. Though the response displacements increased as the
input acceleration increased, the response accelerations almost 60
plateaued in high-level input accelerations. The response 59
acceleration and displacement of SLE04, on which higher
internal pressure was applied, tend to be a little higher than 58
Before After
those of the other specimens.
(b) Out-of-plane direction (crown - crown)

Fig.11 Outer diameter deformation of elbows

before and after the excitation tests

7 Copyright 2015 by ASME

SLE05 was almost same before and after the excitation tests. the elbow (270o) shows about 80,000 (=8%) cumulative
For the specimens SLE01~SLE04, the increase of the mean strain. The strain range of the hoop strain at the crown of the
strain during the excitation was observed on the measured hoop elbow was about 10,000~12,000 for SLE02, SLE03S, and
strain at the crown of the elbow. An example of the typical SLE04.
strain behavior is shown in Fig.12 (SLE02, the first excitation From theses strain behavior and the deformation of the
by 5.0m/s2). Such strain behavior indicates that the ratchet outer diameter of the elbows, the failure mode of
deformation occurred at that point. Figure 13 shows the SLE01~SLE04 can be determined as the fatigue failure
cumulative strain at the crown of the elbow through the accompanied with ratchet deformation. The number of cycles
excitation tests for SLE02, SLE03S, and SLE04. As shown in until failure was about 60~200 for the specimen made of
Fig.13, the cumulative strains are remarkable in the hoop STS410 (SLE01, SLE02, and SLE04). The number of cycles
directions for those specimens, especially for SLE04. Though until failure was about 800 for the specimen made of 304SS
the strain gauge at the failure occurred side (90o) was damaged (SLE03S). The response acceleration and the response
during the experiment, the measured strain at the other side of displacement at the weight of SLE03S was almost same with
those of the carbon steel specimens, so the difference of the
4 10
fatigue life is considered due to the difference of the material
Strain [micro]

4 Axial
3 10
Hoop properties between the carbon steel and the stainless steel. As
2 10
for SLE04, the internal pressure was 1.5 times higher than other
1 10 specimens. Though the ratchet deformation was clearer on
0 SLE04 as shown in Figs.11 and 13, and the response
0 5 10 15 20 25 30 35 40
Time[s] accelerations tended to a little higher than the other specimens
through the excitation tests as shown in Fig.9, there was no
Fig.12 An example of the strain time histories at significant difference in the fatigue life or the failure mode
the crown of the elbow (SLE02, 5m/s2_#01) between SLE04 and SLE02.

4 4
8 10 8 10
Hoop (90o) Hoop (90o) Strain gauge was damaged
6 10
Hoop (270o) Strain gauge was damaged 6 10
Hoop (270o)
(Hoop 270o)
Axial (90o) (Hoop 270o) Axial (90o)
Strain [micro]

Strain [micro]

4 4
4 10 Axial (270o) 4 10 Axial (270o)

4 4
2 10 2 10

0 0

4 4
-2 10 -2 10
Pressure up

White noise











Pressure up
White noise
Test information Test information
Note: The suffix 'a' of test information means the excitation by Sine-wave
(a) SLE02 02, and 'b' means the weight is 110kg.
8 10
(b) SLE03S
Hoop (90o)
4 o
6 10 Hoop (270 )
Axial (90o)
Strain [micro]

4 10 Axial (270o)
Strain gauge was damaged
2 10
4 (Hoop 90o)

-2 10
Pressure up
White noise

Test information

(c) SLE04 (d) Strain measured location in the elbow section

Fig.13 Cumulative strains through the excitation tests for SLE02, SLE03S and SLE04

8 Copyright 2015 by ASME

COMPARISON OF THE TEST RESULTS WITH JEAC4601-2008 is not endorsed yet. The primary plus
SEISMIC DESIGN CODE secondary stress intensity, Sn, is limited to 3Sm, and the
limitation of the primary plus secondary stress intensity range is
Stress intensities and the design limitation in the determined as the usage factor not to exceed 1.0 for class 1 pipe
Japanese seismic design code under Level Cs, Ds [13].
The primary stress intensity of the class 1 pipe, Sprm
[N/mm2], are determined as the following Eq. (2) by the Calculation of primary stress intensities in the
Technical Code for Seismic Design of Nuclear Power Plants, excitation tests and comparison with the design
JEAC4601-2008[14]. The primary plus secondary stress limitation
intensity range, Sn[N/mm2], is determined as Eq. (3), the peak In order to understand the applied input level in the
stress intensity, Sp[N/mm2], as Eq. (4), and the alternating stress excitation tests described above, the experimental results were
intensity, Sl[N/mm2] as Eq. (5): compared with the primary stress limitation in the design code.
When comparing the test results with the design limitation,
PD0 M ip the stress intensity in the experiment should be determined. As
S prm B1 B2 (2) an example of estimation the stress intensity in the experiment,
2t Zi the fictitious elastic moment [16] or the pseudo elastic force
[17] is assumed in the static loading tests. On the other hand, it
M iz
Sn C2 (3) seems that the method to determine the stress intensity in the
Zi excitation tests to compare the design limitation is not
M iz established as far as the authors know. In this paper, the
S p K 2C2 (4) following two procedures are tried to calculate the stress
Zi intensity in the excitation tests;
Sl K e (5) (1) Assuming the elastic response under a high-level
2 input and estimating the stress intensity based on
the design procedure
Where B1, B2, C2, K2: stress indices, B1=0.5, B2=2.52, In this procedure, the resultant moment Mip in
C2,=3.78, K2=1.0 in case of the elbow used in the experiment Eq.(2) is calculated by the following Eq.(7).
[13], P: internal pressure [MPa], D0: outer diameter [mm], t:
nominal wall thickness [mm], Mip: resultant moment due to a M ip m ai Af l (7)
combination of mechanical loads, including the inertia load by
a seismic force [N-mm], Mis: Resultant range of moment due to Where m: the mass of the weight [kg], ai: the
the seismic load [N-mm], Zi: section modulus [mm3]. In this maximum input acceleration [m/s2], Af: amplification
experiment, D0=60.5[mm], t=3.9[mm], and Zi=9,223.707[mm3] ratio under the elastic region, l: the distance from the
according to the pipe configuration. center of the weight to the center of the elbow [m].
The value of Ke is 1 if Sn is less than or equal to 3Sm. If Sn As shown in Fig.4, Af is about 54 for SLE01. Af
is larger than 3Sm, Ke is calculated according to the design code for all specimens varied around 60 (the minimum is 54
of JSME [13]. The calculation procedure depends on the value (SLE01) and the maximum is 75 (SLE05)). l is
of K=Sp/Sn. In this experimental case, Ke is calculated by Eq. 1.263[m] in this case.
(6), considering that K=K2=1.0 and that the carbon steel and Taking SLE01 for example, Af is 54 and m is
austenitic stainless steel are used in the experiment. 100[kg]. Considering that the internal pressure was
9.0[MPa], the first term of Eq.(2) is to be
3S 34.9[N/mm2], which corresponds to 0.25 Sm. So the
Ke 1 q 1 1 m (6) primary stress would reach to 3Sm when the Mip
Sn become 1.378[kN-m] in the experiments. Provided
that the specimens keep the elastic behavior under a
Where q is determined by the design code of JSME [13], q=3.1 high-level input acceleration, the input acceleration to
in the case of carbon steel and austenitic stainless steel. cause Mip corresponding to 3Sm would be about
The primary stress intensity was limited to less than 3Sm 0.2m/s2. But as described in the previous section, the
according to the previous seismic technical guide, JEAG4601- failure mode obtained by the excitation tests was not
1987[15] in Japan. The limitation of the primary stress is collapse but the fatigue failure, though the input
assumed to prevent "collapse" failure of the pipe structures. acceleration was up to 5.0m/s2. Collapse failure did
Though the limitation of the primary stress caused by bending not occur at the 9.0m/s2 input acceleration with
load was deleted in JEAC4601-2008 when the Japanese heavier mass condition, which was more than 45 times
technical code was revised in 2008, the primary stress larger input compared with the Japanese primary stress
limitation is still in force in the Japanese regulatory, because limitation. The above reason is considered as the effect

9 Copyright 2015 by ASME

of the elastic-plastic behavior of the pipe specimens. Where L: inertia force [kN], : response
The decrease of the dominant frequency and the displacement at the weight[mm].
increase of the hysteresis damping in the elastic-plastic The maximum response displacement at the
region work to suppress the dynamic response of the weight by the 5.0m/s2 input acceleration excitation
pipe specimen. As a result, it is considered that the was 117.496[mm], so the fictitious elastic load Lf in
collapse failure mode would be hardly occur on the this case is 2.742[kN]. The fictitious elastic moment
ductile materials like steel due to the elastic-plastic Mf (=Mip in Eq.(2)) is calculated as 3.463[kN-m] by
effect, though the input acceleration level was much multiplying Lf by l, and then Sprm is calculated as
larger than the design assumed level. The reversal 945.1[N/mm2] by Eq.(2). The 945.1[N/mm2]
characteristics of input motion may also work to corresponds to 6.9Sm. The primary stress intensity for
suppress the collapse failure. SLE05 by 9.0m/s2 input acceleration excitation is also
calculated by this procedure. The result is
(2) Using the elastic load-deflection relationship and 1541.5[N/mm2], and it corresponds to 11.25 Sm.
the measured displacement at the weight, Though the calculated primary stress intensity by this
calculating the fictitious elastic moment, and then procedure is also over 3 Sm, the obtained failure mode
estimating the stress intensity based on the design was the fatigue failure, not collapse.
The load-deflection relationship such as shown Calculation of the alternating stress intensities in the
in Fig.8 is obtained by the measured response experiment and comparison with the design fatigue
acceleration and the response displacement at the curve
weight. The fictitious elastic load (Lf) would be The alternating stress intensities (Sl) in the experiment are
obtained as shown in Fig.14 from the elastic load- also calculated to compare the experimental results with the
deflection relationship and the measured displacement design fatigue curve. The procedure to obtain Sl is based on the
at the weight. The fictitious elastic moment (Mf) can second procedure described in the previous section (using the
be calculated by multiplying Lf by l. And then the elastic load-deflection relationship and the measured
experimental primary stress intensity can be obtained displacement at the weight). Mis in Eqs.(3) and (4) is replaced
by using Mf as Mip in Eq.(2). In this procedure, the by 2 Mf in the calculation. The calculated Sl and related values
elastic-plastic effect in the specimens' response to calculate Sl are listed in Table 6, and the comparison with the
behavior is included to some extent by using the design fatigue curves in JSME standard [13] and the calculated
measured response displacement to obtain Lf. Sl is shown in Fig.15. Figure 15 shows that the specimens in the
Taking SLE02 for example, the elastic load- experiment have seismic safety margins to some extent in
deflection relationship (which is shown by a solid blue comparison with the design fatigue curve. This is still a brief
line in Fig.8(a)-(c)) is expressed as Eq.(8) estimation, and it is necessary to estimate the effect of the
cycles by lower level input excitations and to calculate the
L 0.023334 (8) usage factors in the experiment as a future work in order to
estimate the margins for fatigue failure in the actual pipe
components under the dynamic seismic loads. In addition to the
comparison with the design fatigue curve, precious strength
Lf evaluation without safety margin by inelastic analysis and
average fatigue curve are planned.
Elastic load-deflection
In order to investigate failure modes and response
behaviors of piping systems under excessive seismic load,
excitation tests on the steel pipe elbow by a shaking table were
conducted. Obtained failure mode of the steel pipe elbow was
the fatigue failure with ratchet deformation. The reduction of
r dominant frequency and the amplification ratio was clearly
Actual load-deflection observed under the elastic-plastic region excitation. The
relationship by measured hysteresis damping was up to over 10% under the excitation
response acceleration and which exceeds the design limitation condition, so the response
displacement of the piping systems tends to be suppressed compared with the
r: measured response displacement elastically assessed response based on the design procedure.
The procedure to calculate the experimental stress
Fig.14 Schematic illustration to obtain the intensities by excitation tests is proposed, and those stress
fictitious elastic load, Lf intensities are compared with the design limitation. The results

10 Copyright 2015 by ASME

show that the current seismic design in Japan includes some catastrophic failure modes" (PD: Professor N. Kasahara at the
safety margins for both of the collapse failure and the fatigue University of Tokyo). The research project is sponsored by the
failure in comparison with the actual strength of pipe Ministry of Education, Culture, Sports, Science and Technology
components. of Japan (MEXT). The authors wish to extend their
From the results of the pipe component tests, it is appreciation to MEXT for funding and supporting the research
necessary to take a little ingenuity to obtain failure modes program.
except for the fatigue failure on the piping structures, for
example, changing the configuration of the test specimen, input REFERENCES
conditions, or the material of the test pipe. In the fundamental [1] Kasahara, N., Nakamura, I., Machida, H., and Nakamura,
plate test using a plate-type test pieces, the collapse failure was H., 2014, "Research Plan on Failure Modes by Extreme
obtained by the excitation tests on test pieces made of lead (Pb) Loadings under Design Extension Conditions", PVP2014-
[1]. The excitation tests on the pipe elbows made of lead were 28349.
planned as the next structural tests to obtain the "collapse" [2] Fujita, K., Shiraki, K., Kitade, K., and Nakamura, T., 1978,
failure mode [16]. Through the series of the excitation tests, the "Vibration Damaged Experiments of Curved Piping for
conditions to cause different failure modes would be Investigating the Seismic Ultimate Strength", JSME, 44 -
investigated. 386, pp. 3437 3445 (in Japanese).
[3] Fujiwaka, T., Endou, R., Furukawa, S., Ono, S., and
ACKNOWLEDGMENTS Oketani, K., 1999, Study on Strength of Piping
The research work described in this paper is a part of the Components under Elastic-Plastic Behavior due to Seismic
research project named "Study on failure mechanism of nuclear Loading, ASME PVP, 387, pp. 19 25.
components under ultimate loadings and prevention of [4] Touboul, F., Blay, N., and Lacire, M. H., 1999,
Table 6 Alternating stress intensities Sl in the experiment and
Elastic load-deflection relationship
L=0.023147 L=0.023334 L=0.022596 L=0.023147
(L:[kN], :[mm])
Maximum response displacement at the
weight by the 5.0m/s2 input 115.643 117.496 116.423 122.281
acceleration excitation [mm]
Mis [kN-m] 6.762 6.925 6.645 7.150

Sn [N/mm2] 2768.3 2835.4 2720.6 2927.2

Sp [N/mm2] 2768.3 2835.4 2720.6 2927.2
Ke 2.788 2.796 2.783 2.805
Sl [N/mm ] 3859.3 3963.3 3785.4 4105.6
Number of cycles to cause failure at
59 176 776 184
5m/s2 excitation (From Table 3)

4 4
10 10
Alternating stress intensity [N/mm ]
Alternating stress intensity [N/mm2]

1000 1000

100 Design fatigue curve 100

SLE02 Design fatigue curve
10 10
4 5 6 4 5 6
10 100 1000 10 10 10 10 100 1000 10 10 10
Number of cycles Number of cycles
(a) Carbon steel pipe (b) Austenitic stainless steel pipe
Fig.15 Design fatigue curve [13] and the experimental results

11 Copyright 2015 by ASME

Experimental, Analytical, and Regulatory Evaluation of Thinned Wall Piping Model and Damage Detection before
Seismic Behavior of Piping Systems, Transactions of the Failure", ASME PVP, #PVP2010-25839.
ASME, Journal of Pressure Vessel Technology, 121, pp. 388 [11] Nakamura, I., Otani, A., Sato, Y., Takada, H., Takahashi,
392. K, and Shibutani, T., 2011, "Investigation of The Seismic
[5] Tagart, S.W., Jr., Tang, Y. K.,Guzy, D. J., and Ranganath, Safety Capacity of Aged Piping System - Shake Table Test
S., 1990, "Piping dynamic reliability and code rule change on Piping Systems With Wall Thinning by E-Defense",
recommendations", Nuclear Engineering and Design, 123, ASME PVP, #PVP2011-57560.
pp. 373-385. [13] The Japan Society of Mechanical Engineers, 2005, "Codes
[6] Varelis, G.E., Karamanos, S.A., and Gresnigt A.M., 2013, for Nuclear power Generation Facilities Rules on Design
"Pipe Elbows Under Strong Cyclic Loading", Journal of and Construction for Nuclear Power Plants ", JSME S
Pressure Vessel Technology, 135, pp. 011207-1-011207-9. NC1-2005. (in Japanese)
[7] Yoshino, K., Endou, R., Sakaida, T., Yokota, H., Fujiwaka, [14] Japan Electric Association2009, "Codes for Aseismic
T., Asada,Y., Suzuki, K., 2000, "Study on Seismic Design of Design of Nuclear Power Plants", JEAC4601-2008 (in
Nuclear Power Plant Piping in Japan Part 3: Component Japanese)
Test Results", ASME PVP 407, pp.131-137. [15] Japan Electric Association1987, "Technical Guidelines
[8] Takahashi, K., Watanabe, S., Ando, K., Hidaka, A., for Aseismic Design of Nuclear Power Plants", JEAG4601-
Hisatsune, M., and Miyazaki, K., 2009, "Low cycle fatigue 1987 (in Japanese)
behaviors of elbow pipe with local wall thinning", Nuclear [16] Hasegawa, K., Miyazaki, K., Nakamura, I., 2008, "Failure
Engineering and Design, 239, pp. 2719-2727. Mode and Failure Strength for Wall Thinning Straight Pipes
[9] Nakamura, I., Otani, A., Shiratori, M., 2004, "Failure and Elbows Subjected to Seismic Loading", Transactions of
Behavior of Elbows with Local Wall Thinning under Cyclic ASME, Journal of Pressure Vessel and Technology, 130,
Load", PVP-486-2, pp.173-180. pp.011404-1-011404-8.
[10] Nakamura, I., Otani, A., and Shiratori, M., 2010, [17] Hinnat, C., and Paulin, T., 2008, "Experimental Evaluation
"Comparison of Failure Modes of Piping Systems With of the Markle Fatigue Methods and ASME Piping stress
Wall Thinning Subjected to In-Plane, Out-of-Plane, and intensification factors", ASME PVP, PVP2008-61871.
Mixed Mode Bending Under Seismic Load: An [16] Nakamura, I., Demachi, K., and Kasahara, N., 2015, "An
Experimental Approach", Transactions of the ASME, Experimental Investigation on Failure Modes of Piping
Journal of Pressure Vessel Technology, 132, pp. 031001-1 Components under Excessive Seismic Load", SMiRT23 (to
031001-8. be published)
[10] Nakamura, I., Otani, A., Sato, Y., Takada, H., and .
Takahashi, K., 2010, "Tri-Axial Shake Table Test on the

12 Copyright 2015 by ASME