On the Test Parameters for Flexible

Riser in-plane Bending Fatigue Testing

E. Bendiksen, G. Agustsson C. Bergenstof Nielsen
National Oilwell Varco, Denmark1 (NOV)
www.nov.com

ABSTRACT
The fatigue design of the steel armour layers is an important factor in flexible riser design.
Typically, the riser must be able to endure the loads of a field condition for a period of 20
years. In flexible pipe design, this is analysed in global and local flexible pipe fatigue
modelling. However, modelling needs to be supported by proper testing. One of the
widespread test methods is the so-called in-plane bending fatigue testing which is one of the
important qualification tests for flexible pipes. In spite of this, literature contains only limited
guidance about the test procedures. Some guidance is found in API 17B/ ISO 13628-11. It is
the aim of the present paper to provide more specific guidance on the selection of test
parameters. A result of the review of the test parameters is a clear picture of the coverage of a
specific test. This is important when judging whether a new in plane test is required for a
specific pipe / project. Since there is no unique relation between field data and test
parameters, one test programme will cover different fields and vice versa. In order to
intelligently transfer the knowledge obtained in one test to different pipe designs and different
fields it is necessary to understand the steps in the fatigue model, and these steps are
schematized here-in. The focus is on the calculation of armour wire stress. NOV Flexibles
has, over the past 12 years, built up knowledge on the modelling of the real strain response
during cyclic bending and tension. This has been done using fibre bragg gratings embedded in
the tensile armour layers. In this work it has been shown that NOV Flexibles is able to model
the tensile armour stresses at low and high angles which is necessary to conclude from test to
different field applications. The conclusions drawn in the paper are supported by examples
from actual fatigue testing.

Formerly known as NKT Flexibles

INTRODUCTION
A study with the aim to establish the current state-of-the-art on flexible pipe integrity has
been carried out by MCS Kenny in 2010 [3]. The study has compiled a global population
database of flexible pipe usage and failure damage statistics. This data base includes 1900
dynamic risers and 1400 static lines from over 130 production facilities. From the data base
315 individual damages or failure incidents were reported. The incidents have been split up in
damage mechanisms; see Figure 1. The histogram in Figure 1 shows 13 categories that
represent the sum of 315 incidents. One category designated Other contains 25 incidents
covered by 7 failure modes, one of which is armour wire failure. It is not possible to extract
further information from the report with regard to the distribution within this group; however,
if it is assumed that a seventh of the 25 incidents relate to armour wire failure, then of 315
incidents approximately 3 wire failures have been reported. This corresponds to a failure
percentage of approximately 1%.

Figure 1 Failure/Damage mechanism distribution (source: State of the Art Report on

Flexible Pipe Integrity, MCS Kenny, 2010, report number 2-4-5-013/SR01, [3].)
The reason for this low failure percentage is probably that the fatigue design methodology of
risers contains a number of steps that, depending on the level of knowledge within each step,
comprises implicitly a level of conservatism/prudency. This level of conservatism adds up for
each design step. It is of common interest to narrow the gap between the fatigue design
methodology and real life operation of risers. In this way we get lighter designs that mitigate
fatigue or, everything else being equal, we get increased lifetime.
This is why it is of great interest to look closely at each design step of the fatigue design
methodology.
In this paper we have chosen to present our ideas on what to consider in connection with full
scale in-plane bending fatigue testing (IPBT), and why we think that the results from such a
test can be used to qualify pipes to more than just one application.

FATIGUE ANALYSIS OF FLEXIBLE PIPES

Design methodology
The fatigue design methodology for a flexible riser is based on the requirements in API 17J
[1], the guidance in API 17B [2] and the ever evolving industry practice. The process is in
nature iterative, but for ease of understanding it is described here as a sequence. This process
may involve hundreds of engineering hours and computer simulation time. The description
here is a simplification of the actual process. The process, as shown in Figure 2, can be
described as a three step process:
1. The first step involves collection of the environmental data. The environmental load
on the flexible pipe comes through vessel movements and direct wave / current
loading. Wind obviously plays a role in the loading but only in-directly. The vessel
response is normally divided in a static off-set from its mean position (from wind
and current), and a first order movement (from wave action). The environmental loads
are given in statistical tables for the actual position of the vessel and the first step
consists of condensing these wave scatter diagrams for different load directions into a
table of load cases. In this table, load cases are described with a number of waves,
wave height and wave period. This load case table needs to also describe how wave
direction, off-sets, current and vessel loading are to be combined with the wave
loading.
2. The second step consists of global analyses of the riser. This can be done in
commercial programs like Orcaflex, Flexcom 3D, and Riflex. The inputs to these
analyses in addition to the load case table are:

Information about the vessel (RAO, vessel draught etc., geometry of interface
between vessel and riser system)
Information about the riser system itself (length of pipe, buoyancy, etc.)
Geometry and material data of the bend limiting device (bend stiffener or bell
mouth)
Properties of the flexible pipe (weight, diameter, stiffness, etc.)

The outcome of the global analyses is tables of corresponding values for tension and
angle or curvature of the flexible pipe. The values will normally be given for the
fatigue critical areas: Hang off, hog/ sag bend and touch down point.
3. The final step is a local analysis of the flexible pipe for these global loads. Local
refers to the fact that it is an analysis of a limited length of the flexible pipe. The
analysis can be more or less simplified, see for example [4], [5] and [6]. In the
analysis, the global loads are applied to a model of the pipe. This means that the
analysis needs input on the detailed geometry including material data for the flexible
pipe cross section and the bend limiting device (bend stiffener or bell mouth).
Furthermore, a very important parameter is the pressure that the pipe is exposed to.
Temperature also plays a role because the material properties are temperature
dependent.
Using these data, the angles for each load case are transformed to curvature.

Secondly the cross section(s) are analysed for the curvatures and the armour wire
deflections and stresses are determined (mean value and range).
The next step in the local analysis is to calculate the damage for the stresses. The link
between these two is the armour wire SN-curve. Derivation of SN-curves is a subject
on its own, see [7]. A very important parameter in SN-curve selection for flexible pipe
armour wires is the annulus environment. The SN-curve is found by testing in the
actual environment. Another aspect to consider is that there is a mean stress and a
variation over time and both have an influence on fatigue performance. The mean
stress can be taken into account through for example a Gerber correction of the stress
variation, [7], or by doing SN-testing at the actual mean stress.
The last step of the local analysis is to sum up the damage of the contribution from all
load cases. This is normally done using the Palmgren-Miner rule, ref API 17J [1].
The maximum allowable damage over the lifetime is 0.1, i.e. there is a safety factor of
10 on the design life.
The analysis can be carried out as a regular approach or an irregular approach. The
description above covers the regular approach. However, the scheme for the irregular
approach is very much the same except the analyses differ and some additional
complexity is added (mainly in selection of load case but also in counting of number
of cycles and computational efforts).
Vessel and mooring data

Environmental data

Hs/Tp 0-3 3-4 4-5 5-6

5

0-1
1-2
2-3
3-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
11-12
12-13
13-14
14-15
Sum

37

6-7

7-8

8-9

9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum

108 176

197

172

126

2 35 170
Hs/Tp 0-3 3-4
3
38
5
0-1
3
2
1-2
Hs/Tp
2-3
0-1
3-4
1-2
4-5
2-3
5-6
3-4
6-7
4-5
7-8
5-6
8-9
6-7
9-10
7-8
10-11
8-9
11-12
9-10
12-13
0
7 75 319
10-11
13-14
11-12
14-15
12-13
Sum 0
7
13-14

82

49

28

15

0
0

75

319 769 1265 1594 1637

1429

1084

722

426

224

Permeation
analysis

1010

402 585 612 506

353
218
123
65
32
16
7
3
2
1
1 3133
4-5 5-6 6-7 7-8
8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum
158 343 475 473
371
243
139
72
35
16
7
3
1
2377
37 108 176 197 172 126
82
49
28
15
8
4
2
1
1010
30 113 232 304
286
209
126
66
31
13
5
2
1
1421
35 170 402 585 612 506
353
218
123
65
32
16
7
3
2
1
1 3133
3
24
83
159
195
170
114
62
29
12
4
1
856
0-3 3-4 4-5 5-6 6-7 7-8
8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum
3
38 158 343 475 473
371
243
139
72
35
16
7
3
1
2377
3
18
56
99
113
90
55
27
11
4
1
477
5 37 108 176 197 172 126
82
49
28
15
8
4
2
1
1010
3
30 113 232 304
286
209
126
66
31
13
5
2
1
1421
2
12
35
57
59
43
23
10
3
1
245
2 35 170 402 585 612 506
353
218
123
65
32
16
7
3
2
1
1 3133
3
24
83
159
195
170
114
62
29
12
4
1
856
1
7
20
30
28
18
9
3
1
117
3
38 158 343 475 473
371
243
139
72
35
16
7
3
1
2377
3
18
56
99
113
90
55
27
11
4
1
477
1
4
11
14
12
7
3
1
53
3
30 113 232 304
286
209
126
66
31
13
5
2
1
1421
2
12
35
57
59
43
23
10
3
1
245
1
2
5
6
5
2
1
22
3
24
83
159
195
170
114
62
29
12
4
1
856
1
7
20
30
28
18
9
3
1
117
1
2
2
2
1
8
3
18
56
99
113
90
55
27
11
4
1
477
1
4
11
14
12
7
3
1
53
1
1
1
1
4
2
12
35
57
59
43
23
10
3
1
245
1
2
5
6
5
2
1
22
0
1
7
20
30
28
18
9
3
1
117
1
2
2
2
1
8
0
1
4
11
14
12
7
3
1
53
1
1
1
1
4
0
1
2
5
6
5
2
1
22
0
769 1265 1594 1637 1429 1084 722
426
224
106
43
17
4
1
1 9723
1
2
2
2
1
8
0
1
1
1
1
4
0
0
75 319 769 1265 1594 1637 1429 1084 722
426
224
106
43
17
4
1
1 9723
0

14-15
Sum

Operational data
Pipe

106

43

17

Selection of global fatigue

load cases in-house
programs

9723

Bend stiffener
SN-test

Configuration

Global analysis
ORCAFLEX

Local analysis
BFLEX

TDP
No

Height

Period

additional

No

Angle to curvature
Curvature to stress
Stress to damage
Mean stress
Damage summation

mean
Tension
Hog/ Sag
range
a_mean a_range Tension
Tension
Hang mean
Off
No
a_mean a_range Tension
mean range
Tension
No
a_mean a_range Tension range

off-set
direction
current

Figure 2 The fatigue design procedure schematically

TDP Damage = 0.1

Hog/
sag Damage
HO Damage
= 0.1 = 0.1
HO Damage = 0.1

FATIGUE TEST
Why make fatigue tests
Above it is explained how the flexible pipe is designed against fatigue. IPBTs are made in
order to demonstrate that the design procedure leads to a safe operation for the planned design
life. This validation can be done in several ways: 1. A validation of the methodology for
fatigue design or 2. A validation of an actual pipe for loads similar to the actual loading. API
17B defines the two types of testing:
1. For the service-life model validation, API 17B recommends the use of a single load
block using approximately 400,000 cycles of one load block to arrive at a fatigue damage
of 1.0 in the most fatigue susceptible layer.
2. For qualification of a certain pipe design (for a given application), a service simulation
is run to the actual calculated fatigue damage for the pipe in operation (or 0.1 if this is not
available). For the service life simulation, API 17B recommends 2-4 million cycles
divided in to load blocks based on the pipe application in the field.
Some modifications are often done relative to the above definitions. In order to demonstrate a
safety factor, the service simulation test can be extended to a higher damage than experienced
in-service. The service-life model validation test is often divided in to a number of load
blocks of different sizes. Finally, an IPBT is often a two stage test where the first stage is a
service simulation test and the second part is a service-life model validation test, this
possibility is also described in API 17B. The damage in the service simulation test is
representative for operating condition, i.e. it is run to a damage of 0.1 in the most fatigue
susceptible layer. Then the test is continued into a service life model validation test. This part
runs to a damage of 1.0 or until pipe failure, whichever comes first.

How are fatigue tests made

In an IPBT a pipe sample of approximately 15 m is tested. The pipe sample is complete with
all layers and end fittings. As shown in Figure 3, the pipe is pressurized, tensioned and bent
back and forth a number of times. The pipe sample is normally equipped with a bend limiter,
i.e. a bend stiffener or a bell mouth, in order to avoid very concentrated bending of the pipe.

Figure 3 Basic principle of an in-plane bending fatigue test

The testing is done in a dedicated test rig, see Figure 4. The test can be carried out with the
pipe in a horizontal or vertical layout.
Usually the same pressure is kept throughout the entire test.
For practical reasons, bending is typically applied at one end of the pipe (top), while the other
end is tensioned (bottom). Normally, the top end is fixed against the twist and the other end is
allowed to rotate freely.

Figure 4 A fatigue test rig, [2]

Usually, the bending plane is kept the same throughout the testing. This simplifies the test
setup and accelerates the test since fatigue damage is concentrated on a small section on the
pipe.
API 17B recommends that the pipe length between the bend limiter and top end-fitting should
be at least one pitch of the outer tensile armour wires, and between bend stiffener and bottom
end-fitting at least three pitches. These are rules of thumb; care should be taken and it should
be considered to analyse the effects on wire stresses from end-fittings if the sample is short
relative to these recommendations.
It is important to extensively monitor the loading, the pipe response and its integrity. This is
further discussed in a section on acceptance criteria. The monitoring is accompanied with
extensive high frequency logging of the data and software for presenting it in an easy and
accessible way.

TEST PARAMETERS, LOAD CASES AND ACCELERATION

OF TEST
General
The test programme for fatigue testing is a compromise between keeping loading realistic and
applying a sufficient load to obtain the specified fatigue damage within an affordable number
of load cycles. This acceleration of damage can be done in several ways. The field data of one
field can be transferred to a test programme in different ways so that different test
programmes can be representative of a certain field. The reverse is also true that a certain test
programme can be representative of several different fields, both with respect to production
data and environmental data.
The parameters involved in the test are:

The pipe design

The bend stiffener design
Temperature
The curvature of the pipe (mean value and amplitude)
Applied tension (mean value and amplitude)
Internal pressure
The number of cycles

The remainder of this section explains the selection of test parameters.

Pipe design/ test sample

A certain pipe design would normally be taken as covering another application when the ID is
within plus/minus 2 inches, the layer sequence is the same, and the wire geometry is the same
while wire size and grade may differ, ref. API 17B (Table 20 and section 9.4.4.1).

Bending restriction device; bend stiffener, bell-mouth

The bend limiting device serves the same purpose as during operation: Protection of the pipe
against excessive bend radii, especially close to the end fittings. The bend limiting device
plays a large role in the relation between test rig angle and pipe curvature.
On some project specific IPBT the test has been run with the actual bend stiffener. However,
this can be unpractical because the bend stiffener may be so big that the length limitations of
the test rig means that it can be difficult to have a free section of one pitch length above the
bend stiffener and a three pitch length below. Furthermore, the bend stiffener is one of the
links between the rotational angle of the pipe ends and the curvature of the pipe and thereby
between the angle and stress fatigue damage. Therefore, a very large bend stiffener will lower
the damage or a smaller bend stiffener can be a way of accelerating the test.

Temperature
API 17B does not recommend any specific bore temperature for the test and the temperature
is considered a secondary parameter whose effects can be analysed and tested separately. The
temperature distribution in the pipe influences the bend stiffness and therefore the relation
between applied angle and obtained curvature.

API 17B recommends that a thermal analysis is carried out in order to establish the
temperature increase in different layers due to friction between them.
Active cooling or heating can be coupled to the test specimen in order to control the
temperature of the bore and/or different layers.

Bore fluid
Water or oil can be used as a bore medium. The impact on SN-curve from diffusion of the
bore medium into the pipe annulus must be taken into account.

Frequency
Due to cost and time considerations, a test will usually be run at the highest possible
frequency. The frequency can be limited by the test rig itself and because friction may buildup temperature in the pipe wall.

Test program
The four remaining parameters are the test parameters entering the test program. It is
important that the loading is realistic in order to test for the true failure modes. However, we
want to apply a damage of say 1.0 in 3 million cycles where in the field a damage of 0.1 in
100 million cycles is normal (and in the field the SN-curve is normally worse due to corrosion
fatigue where the test is traditionally run with non-corrosive annulus environment). As a
consequence we need a damage rate in the test which is 333 times larger in the test than in
reality. Some acceleration of the damage is therefore necessary. The acceleration can be done
by increasing the load in terms of pressure, tension and/ or curvature. Most often a
combination is used.
Internal pressure
API 17B recommends that the highest expected operational pressure be used. However, bore
pressure is usually a means for acceleration of the test and it is therefore selected
conservatively, usually near the design pressure of the pipe.
Number of cycles
The number of cycles recommended in API 17B is in the range of 2-4 million cycles for a
service simulation test and 400,000 cycles for a service life model validation test. It is
difficult to argue against API 17B, although in the common situation where the two test types
are combined a distribution with half the cycles in the service simulation test and half in the
service life model validation test could result in a more reasonable loading pattern.
Tension
According to API 17B the tension should be selected conservatively from the global load
cases. It is often said that tension and pressure must be selected so that the loading of the
actual application is covered. It is not necessary to cover the extreme loading as this is not
representative for fatigue, but tension corresponding to the normal operation should be
represented, although it is the authors opinion that it is more relevant to represent the stress
and contact pressure levels. Tension can be used as an acceleration parameter for the test
since a higher mean tension will increase contact pressure and raise the mean stress in the
tensile armour wires and thus accelerate the fatigue damage.

Curvatures and angles

The angle is the load applied in the test rig, while the curvature of the pipe is the parameter
giving the armour wire stress. The link between the two is the bend stiffness of the pipe and
bend stiffener.
Non-zero mean angles during fatigue testing may result in pig tailing of the relatively short
test sample and should be applied with care.
A guidance in API 17B is that the extreme angles of the actual application are represented in
the test; however, this is not always practical because it may create very large loads in the
test. Furthermore, as explained above, the relevant parameter seen from the pipe is more the
curvature or the stress. The purpose of the fatigue test is fatigue validation and not extreme
load validation. As a consequence, the industry often relaxes this API 17B recommendation.
Furthermore, as explained above, API 17B allows a service life model validation test to be
run at one load case only, without any requirement for this load case to look like in-field
loads. Nevertheless, most people would consider a certain design qualified through service
life model validation tests.
During the traditional 20 year design life, a flexible riser will see in the order of magnitude
100 million load cycles. The majority of these will result in very low angles and curvatures.
Angles and curvatures which are difficult to represent with reasonable precision in a fatigue
test rig and if it was possible they would result in a fatigue damage corresponding to that of
the riser in-service (i.e. maximum 0.1).
Normally a distribution where there are many cycles of the small load cases and fewer of the
large ones is aimed at (similar to what is seen on the wave scatter diagram in Figure 2 upper
left corner). Angles are selected so that they are practical and result in the required damage in
a series of load cases. This is not to say that the angles should be selected far away from what
is experienced in normal service. In fact, one should aim for load cases that are in the same
order of magnitude as that of the field. In some examples later this is shown in terms of stress
or angle. One could have chosen curvature also.
API 17 B warns against artificially improving the fatigue performance of the armour wires by
applying large load cases early in the test. We recommend to verify that the largest load case
does not induce yielding of the armour wires but stays in the high cycle fatigue region and to
mix the load cases in a logical way and mix large and small load cases from the beginning.

ACCEPTANCE CRITERIA
API 17B suggests several acceptance criteria for the pipe in IPBT, which ranges from carcass
cracks to loss of up to 5% of the tensile armour wires. However, API 17B also emphasizes
that the purchaser and the manufacturer should agree on these, and therefore a wide variety of
acceptance criteria can be found in the industry for these kinds of tests. Here, unified criteria
are proposed. The main purpose of the IPBT is to verify fatigue performance and only
acceptance criteria for the armour layers are given here:

Test phase
Service simulation (damage up to 0.1)

Service-life model validation (damage up to

1.0 based on the design SN-curve)

Service-life model validation (damage up to

1.0 based on the mean SN-curve) or
destructive test

Acceptance criteria
Less than 5% broken tensile armour wires,
and a successful pressure test to 1.25 times
the design pressure with tensile load applied.
No loss of containment, and a successful
pressure test to 1.25 times the design
pressure with tensile load applied. Number
of broken armour wires in agreement with
statistical analysis, see below.
Number of broken armour wires in
agreement with statistical analysis, see
below.

Test stop criteria

Techniques to detect wire breaks during testing are improving. NOV is perfecting the wire
break detection via optical monitoring. Alternatives include acoustic monitoring, acceleration
response from pipe, and measurements of elongation of pipe during testing. Results for the
optical monitoring technique are very promising. It is obvious that such methods are very
interesting from a pipe integrity management point of view during in-service. The technology
is also beneficial for application in fatigue testing: Using this methodology it is possible to
run a service-life model validation test to failure consisting of a predefined number of wire
breaks. In this way the final failure of the pipe (loss of pressure containment) can be avoided,
which saves time and results in a dissection of a pipe that is un-damaged by the loss of
containment and thereby learnings from the IPBT can be increased.

Statistical method for determination of expected number of

failures.
A practical approach to estimate the number of broken wires at a certain calculated
Accumulated Fatigue Damage (AFD) is the following:
1) Estimate the actual number of wires in the fatigue zone that have a certain level of
AFD.
2) The AFD expresses the probability that a wire is broken. The number of expected wire
breaks in a fatigue region is therefore calculated as the number of wires times the AFD
times the used probability curve (0.5 for the mean curve and 0.024 for the design
curve).
An example: Using the mean SN curve, and running to AFD = 1.14 in a fatigue region
containing 10 wires, we get that the expected number of wire breaks is 10 1.14 0.5 = 5.7
wire breaks.
Further
3) The fatigue zone can be divided into regions of equal damage and the number of wire
breaks can be assessed
4) Ultimately the number of wire breaks can be summed.
5) In case of evaluating the fatigue damage level of a service simulation, normally less
than 0.5 wire breaks should be found using the above methodology.

The coverage of validation test for other pipes and

other applications
We have sometimes been faced with the question of whether one fatigue test covers another
pipe or the same pipe in another part of the world. As stated above the test parameters should
be selected to be in the same order of magnitude as those for a real application and the
distribution should be so that a curve of relative occurrence versus load is similar to reality,
see Figure 5 and Figure 7. However, the relative occurrence will not be identical; the slope
needs to be different so that the test can be performed within a reasonable number of load
cycles. Therefore, the resemblance between the actual field in terms of angles / curvatures and
the test loading is something that can always be argued but also something that can be
questioned. It cannot be used to determine whether a test covers a certain application. In our
opinion a test covers a pipe design when the internal diameter is within plus / minus two
inches of the test pipe, and when stresses and contact pressures of the test pipe covers those of
the in-field conditions for the pipe.

18

16

14

top angle (deg)

12

10
angle range, Kristin
angle range, Visund
8

0
0

10

20

30

40

50

60

70

80

90

100

accumulated occurence (%)

Figure 5 Top angle distribution (smooth blue curve is fatigue test and step wise
purple in-field.
Another aspect that one should consider when evaluating the validation of a certain pipe in a
certain field is the overall verification of the methodology. Other successful tests, although on
quite different pipes, add comfort to the methodology.
Furthermore, it is important to realise that since the fatigue testing is accelerated, it means that
the design methodologies, mainly the local analysis methodology must be equally well suited
to handle large and small load cases. For the BFLEX package this has been demonstrated
using fibre bragg grating as explained in [6].

Figure 6 Comparison between measured and predicted stresses in tensile armour at

small and large load cases, extract from [6].

FATIGUE TEST A CASE STUDY

Fatigue Modelling
National Oilwell Varco (NOV) has performed a number of IPBTs. In order to show the
conservatism in the NOV fatigue design methodology, one case story, the IPBT of an 8
production riser, is presented here.
The global analysis for this particular riser was done for a lazy wave configuration where a
spread moored FPSO was used. The water depth was 1700 m and the global dynamic
modelling was performed using Flexcom.
Based on metocean data the dynamic model provides, among other, fatigue load cases in
terms of tension angle data/plots for each load case. Knowing the pipe and bend stiffener
stiffness we get all of the tension, shear forces, and bending moment ranges for the hang-off
section for the whole range of waves with the corresponding number of cycles. The resulting
load cases are presented in Table 1. These load cases have been used as input to the IPBT
service life test program for the 8 riser.

Table 1 Fatigue load cases in hang-off as found in the global dynamic model.

The load cases are transferred to the local fatigue 3D model (BFLEX) of the hang-off
configuration that returns, among other, mean stress, stress range, mean stress corrected stress
range (using Gerber), number of cycles, and damage per cycle. The resulting load case
spectrum is presented in Table 2.
The normalized accumulated fatigue is further presented as a function of stress range in
Figure 7. The service life test program is based on relatively fewer stress ranges between 150
MPa and 250 MPa and relatively more stress ranges above 250 MPa in such a way that a
fatigue damage of 0.1 is obtained during 2 million cycles. Such a distribution is shown by the
red line in Figure 7.

Table 2 Load cases expressed as stresses, damage and cycles as expressed by the local
fatigue 3D model BFLEX, calculated in hang-off.

Figure 7 Accumulated fatigue damage as a function of stress range

The resulting load cases all based on the operating pressure of 30 MPa needed to reach a
Miner sum of 0.1 are shown in Table 3.
Table 3 Resulting load cases to obtain 0.10 fatigue damage.

N_cycle
Seq.
X 1000

Rig
Angle
mean
(deg)

Rig
Angle
range
(deg)

Tension
@ mean
rig angle
(kN)

Tension
range
(kN)

Stress
Range
(MPA)

Mean
Stress
(MPA)

Damage

1395

2.41

1.7

1372

135

164

493

2.51E-02

400

2.71

2.3

1372

183

222

492

3.95E-02

100

3.13

2.9

1372

230

253

485

1.51E-02

75

3.48

3.5

1361

278

269

484

1.53E-02

30

3.84

4.1

1337

326

282

481

7.89E-03

2000

0.10

The resulting test program based on not violating the maximum, 25% fatigue damage per load
case criterion is shown in Table 4.

Table 4 The service life test program in blocks, not violating the maximum of 25%
fatigue damage per load.
Internal pressure 30.0MPa

Seq.

Rigcase

N_cycle

Rig
Angle
mean
(deg)

Rig
Angle
range
(deg)

Tension
@ mean
angle_rig
(kN)

Tension
range
(kN)

174375

2.4

1.7

1372

135

50000

2.7

2.3

1372

183

12500

3.1

2.9

1372

230

9375

3.5

3.5

1361

278

7500

3.8

4.1

1337

326

9375

3.5

3.5

1361

278

12500

3.1

2.9

1372

230

50000

2.7

2.3

1372

183

348750

2.4

1.7

1372

135

10

50000

2.7

2.3

1372

183

11

12500

3.1

2.9

1372

230

12

9375

3.5

3.5

1361

278

13

7500

3.8

4.1

1337

326

14

9375

3.5

3.5

1361

278

15

12500

3.1

2.9

1372

230

16

50000

2.7

2.3

1372

183

17

348750

2.4

1.7

1372

135

18

50000

2.7

2.3

1372

183

19

12500

3.1

2.9

1372

230

20

9375

3.5

3.5

1361

278

21

7500

3.8

4.1

1337

326

22

9375

3.5

3.5

1361

278

23

12500

3.1

2.9

1372

230

24

50000

2.7

2.3

1372

183

25

348750

2.4

1.7

1372

135

26

50000

2.7

2.3

1372

183

27

12500

3.1

2.9

1372

230

28

9375

3.5

3.5

1361

278

29

7500

3.8

4.1

1337

326

30

9375

3.5

3.5

1361

278

31

12500

3.1

2.9

1372

230

32

50000

2.7

2.3

1372

183

33

174375

2.4

1.7

1372

135

After the completion of the service life test the pipe should be further tested to a Miner sum of
1.0 in order to illustrate the safety factor of 10 in the fatigue model. This was done using

316000 cycles of a specially designed load case with increased angle and tension range as
well as increased internal pressure. The test program is shown in Table 5.
Table 5 Test program to obtain a total Miner sum of 1.0
Internal Pressure 39.0 MPa

N
Cycles
X 1000

Rig
Angle
mean
(deg)

Rig
Angle
range
(deg)

Tension
@ mean
rig
angle
(kN)

Tension
range
(kN)

Stress
Range
(MPA)

Mean
Stress
(MPA)

Damage

316

5.79

1125

750

421

560

0.90

After completion of the service life test, the pipe should be further tested using a test-tofailure test program. The fatigue calculations were therefore performed using the mean SN
curve. In order to accelerate the fatigue damage the bore pressure was set to 39 MPa like the
extension test program. The accumulated fatigue damage (AFD) was at the onset of the
destructive test 0.09 using the mean SN curve. The remaining destructive testing program was
set to 0.91 so that the AFD would be 1.0 using the mean SN curve at the end of the
destructive test. The test program for the destructive test part is shown in Table 6.
Table 6 Test to failure test program
Internal Pressure 39.0
MPa
N_cycle
x 1000

Rig
Angle
mean
(deg)

Rig
Angle
range
(deg)

Tension
@ mean
rig
angle
(kN)

Tension
range
(kN)

Stress
Range
(MPA)

Mean
Stress
(MPA)

Damage

590

1.03

10

1125

750

438

503

0.91

The presented test programs derived from an actual field shows tension and angle ranges. The
test programs also show the stress range, mean stress and the internal pressure.
However, it is more interesting to observe the contact pressure (from tensile armour to antiwear tapes below) and stresses on wires combined, since they contain information on wire
movements and exposure to fatigue loading. High contact pressures and low stresses indicate
little movement of the wires, hence a low exposure to fatigue loading. High contact pressures
and high stresses indicate extensive loading and exposure to fatigue loading. Therefore these
parameters are directly comparable between different pipes, different wave fields and
different test programs and give more information of a pipe designs fatigue resistance.
An overview of these numbers as calculated from the three test programs is presented in
Table 7.

Table 7 Overview of some of the most important fatigue drivers, i.e. contact pressure
and stresses, divided in the test program load cases.

Fatigue Testing
During the year 2007 and 2008 the full scale dynamic fatigue testing of the NOV 8 flexible
riser according to the modelling and test program given above was carried out.
The phase 1 of the dynamic test program included a service life simulation test and an
extension test.
The service life simulation test program was designed to qualify the pipe design by simulating
a predicted cumulative damage of 0.1 based on a selected service loading for deep water risers
and the design SN curve for the tensile armour wires. The extension test was designed to
demonstrate the reserve capacity of the pipe design and to validate the safety factor implicit in
the design analysis methodology.
The extension test included a total of 316 000 cycles, composed to generate a cumulative
damage of 0.9 under service conditions based on the design SN curve for the tensile armour
wires. Thus on completion of the service life simulation test and the extension test, a total
cumulative damage of 1.0 was obtained. The successful test thus verified a safety factor of 10
for the design analysis methodology for the present flexible riser. The planned dynamic test
program was successfully completed without any observations of damage to the pipe, based
on measured test parameters and visual external inspections. A hydrostatic pressure test was
successfully performed at the end of the service life simulation test, demonstrating the
structural integrity of the sample with respect to pressure holding capacity. The hydrostatic
test was undertaken at a pressure of 1.25 times the design pressure of 345 bar, combined with
an axial tensile load of 1500 kN.

A phase 2 test program was initiated to further investigate the residual life of the riser as well
as possible failure modes and failure mode developments for the riser. The aim of the phase 2
test program was to apply a number of cycles corresponding to a total cumulative damage of
1.0 based on the mean SN curve for the tensile armour wire.
The phase 2 test was planned with a total of 590 000 cycles in order to get to a total
cumulative damage of 1.0 based on the mean SN curve for the tensile armour wire.
The planned phase 2 of 590 000 cycles was successfully completed without any observations
of damage to the pipe, based on measured test parameters and visual external inspections. For
the above reason the testing was continued. Failure by leak of the sample was observed after
an additional 92 000 cycles. The leak was noted through one of the venting holes of the top
end fitting. Measured elongation of the pipe sample during the final phase of the test program
indicated failure developments of some of the tensile armour wires.
At the end of testing, the sample had been subjected to a total cumulative damage of 1.14
based on the mean SN curve of the tensile armour wires.
Following the fatigue testing the dynamic riser was dissected. The dissection showed that a
number of tensile armours were broken in the fatigue zone. Further analysis of the test data
showed that the wires broke during phase 2, and that the number of wire breaks was as
expected when compared to the total cumulative damage of 1.14.
Thus it can be concluded that the NOV fatigue methodology in this test has proven to be
conservative.

CONCLUSIONS

The complex fatigue design procedure has been documented. Although complex, it is well
described and statistics show few in-service failures due to fatigue.

The In-plane bending fatigue test is a standard test for validation of the methodology and
for the validation of actual pipes for their use.

The In-plane bending fatigue test is defined in API 17B and the present paper provide
further guidance to the selection of test parameters, test acceptance criteria and the
common practice.

Fatigue tests are an accelerated test and a test for one pipe in certain field conditions will
cover for a range of other pipes and / or other applications as long as the internal diameter
is the same within plus / minus two inches and the armour wires stresses and contact
pressures are covered in the test.

Therefore the following parameters are important when making a fatigue test:

That the test pipe has the same ID as the actual pipe plus/minus 2 inch
That the test pipe has the same cross sectional shape of armour wires not necessarily
same size or material

That the load cases are selected so that contact pressure between armour wires, armour
wire stress and armour wire stress range are of the same size as those for the actual
application

An example of fatigue testing validating the methodology is given. The example

demonstrates the conservatism of the NOV fatigue design methodology.

REFERENCES
[1] Specification for unbounded flexible pipe, API 17 J, July 2008
[2] Recommended Practice for Flexible Pipe, API 17B, July 2008.
[3] State of the art report on flexible pipe integrity, report no. 2-4-5-013/SR01,
http://www.oilandgasuk.co.uk/publications/viewpub.cfm?frmPubID=152
[4] Pascal Estrier, Updated Method for the Determination of the Service Life of Flexible
Risers. Proceedings of the First European Conference Marinflex, 1992
[5] Svein Svik, On stresses and fatigue in flexible pipes, Dep. Of Marine Structures,
NTNU, 1992
[6] Svein Svik & Ragnar R. Igland, Calibration of a flexible pipe tensile armour stress
model based on fibre optic monitoring, OMAE 2002-28092
[7] S. Berge, E. Bendiksen, J. Gudme and R. Clements, Corrosion fatigue testing of flexible
riser armour procedure for testing and assessment of design criteria, OMAE 2003 37327.