Proceedings of the ASME 2011 30th International Conference on Ocean, Offshore and Arctic Engineering

OMAE2011
June 19-24, 2011, Rotterdam, The Netherlands
Proceedings of the ASME 2011 30th International Conference on Ocean, Offshore and Arctic
Engineering
OMAE2011
June 19-24, 2011, Rotterdam, The Netherlands
OMAE2011-50026

OMAE2011-50026

WELLHEAD FATIGUE ANALYSIS METHOD

Lorents Reinås Torfinn Hørte
University of Stavanger/Statoil ASA Det Norske Veritas, AS
Forus, Stavanger, Norway Høvik, Oslo, Norway
[email protected] [email protected]

Morten Sæther Guttorm Grytøyr

Statoil ASA Det Norske Vertias (U.S.A.) Inc.
Vækerø, Oslo, Norway Houston, Texas, USA
[email protected] [email protected]

ABSTRACT iii) Fatigue damage assessment, where a mapping of the

Re-completion and re-drilling of existing wells and loads with the relevant load to stress curve is carried
introduction of new large drilling rig systems are elements that out together with subsequent fatigue damage
have led to renewed focus on the fatigue capacity for existing calculation. Appropriate S-N curve is applied together
and new subsea wells. Due to lack of applicable codes and with wave scatter diagrams for the relevant operations
standards for such fatigue calculations, a unified analysis and durations. The final result is the accumulated
methodology has been developed and described in a Wellhead fatigue damage.
Fatigue Analysis Method Statement (MS). The intention of this With a unified analysis methodology in place particular
work is to reflect the best practice in the industry and to provide attention is placed on a structured and specified analysis input
an important contribution to well integrity management. The and output. Results are suggested presented as a function of
analysis methodology is limited to fatigue damage from time and also as a function of key analysis input parameters that
dynamic riser loads present during subsea drilling and work are associated with uncertainty. These are prerequisites from a
over operations. The analysis procedure may be divided into well integrity management perspective in ensuring analysis
three parts. results that are comparable. This paper presents the essence of
i) A local response analysis that includes a detailed the Wellhead Fatigue Analysis Method that was developed in
finite element model from wellhead datum and below. cooperation between Statoil and DNV. Currently this analysis
Interaction between the structural well components methodology is under extension and revision in the joint
and soil structure interaction is properly accounted industry project (JIP) "Structural Well Integrity During Well
for. The main result from this analysis is the load-to- Operations". 11 operators participate in this JIP which also has
stress curve that describes the relationship between structured cooperation with equipment suppliers, drilling
the riser loads at the wellhead datum and the stress at companies and analysis houses. The aim is to form a wellhead
the fatigue hot spots. The analysis also provides the analysis recommended practice document.
lower boundary conditions of the global load analysis
model. BACKGROUND
ii) A global load analysis where the floating mobile Access in subsea wells from a MODU is provided
drilling unit (MODU) motions and wave loads on the through a temporal conduit, namely the riser. The well interface
riser are taken into account. The results are time series towards a removable riser is the subsea well head (WH).
or load histograms of the loads at wellhead datum, Subsea WH is subject to dynamic loads transmitted from the
with focus on the bending moment, in all relevant connected riser system. These loads can be combination of
environmental sea states. bending and tension (compression). This paper addresses
fatigue damage estimation in the upper part of the well

1 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 construction. A fatigue failure of a wellhead system may lead to wellhead fatigue loads and resistance for a number of subsea
a serious accident. Should the wellhead structurally fail it may wells. Most of the work has been done on existing wells with a
expose the wellbore to the environment leaving no efficient loading history and future operations in planning. The
remedies to hold back potential well pressure and for this experience of this work revealed a need for a unified analysis
reason WH fatigue is a threat to well integrity. methodology.
Hopper [1] reported that in 1981 a subsea WH parted from The analysis methodology has varied in the industry, and a
its surface casing during operations in UK waters and this number of input parameters are associated with a high degree
resulted in a cause investigation. Hopper concluded that the of uncertainty and are difficult to assess. Consequently, fatigue
failure was due to fatigue vortex induced vibrations (VIV) analysis is time consuming and results have been difficult to
loading from the drilling riser and that some modifications to compare and to verify. Analysis work needs to be precisely
details of the subsea wellhead were needed. Hopper’s work documented for it to be possible to reproduce. The analysis
shows that subsea WH exposed to dynamic loading exceeded methodology is therefore intended to be specific on these
the fatigue limit of the girth weld connecting the WH body to aspects, and aims to reflect the best practice in the industry, and
the surface casing extension. Other studies followed the work to contribute to an improved basis for comparisons. The
of Hopper. analysis methodology was described in detail by a Method
Dykes et al. [2] work on unified WH systems for subsea Statement (MS) document produced by an integrated team from
exploration drilling discusses design considerations of dynamic Statoil and DNV. The subsea WH suppliers have been involved
loading. Fatigue loading was identified as one design load case through workshop sessions and document review. It was
and a North Sea load histogram was deployed as a design load acknowledged from the start that this revision of the MS is not
on the systems discussed. A graphical comparison between the the final one, and more work will need to be performed in order
Dykes et al. load spectrum and a typical North Sea spectrum to further improve and extend the applicability of the analysis
from recent analysis work we have performed can be seen in method. A JIP by 11 operators in the Norwegian part of the
Figure 1. The loading is presented as moment range cycle North Sea is currently undertaking further development,
density functions for better comparison. The magnitude of improvements and extensions to this methodology whilst
loading in this North Sea case is significantly higher compared maintaining these objectives [4].
to what Dykes et al. presented in 1989.
Since then drilling activities in subsea wells has increased
in complexity, measured depth of subsea wells has increased Moment cycles density functions
and subsea wells are drilled at deeper waters. All these factors
are indicating that the duration of operations has increased per 10000
well. This development is partly identified by Osmundsen et al
[3] who analyzed the productivity parameter “meters drilled per 1000
North Sea
Cycles density function (cycles/Nm)

day” on 642 exploration wells on the NCS from 1965 till 2008. Dykes
In particular Osmundsen et al. results shows that increased 100
water depth has a negative effect on drilling productivity.
The comparison of loading (S) indicates an increase and 10
combined with a prolonged exposure (N) this will significant
increase the fatigue accumulation when applying an S-N 1
fatigue approach.
The typical subsea wellheads systems of today are in 0,1
principle similar to the system analyzed by Hopper and Dykes
et al. They differ in details, material and size, but are still 0,01
characterized by a welded connection between rigid wellhead
housing and the less rigid casing pipe extending down into the 0,001
well. Conventional wellhead systems still include some radial
tolerances between wellhead and conductor housing in order 0,0001
for it to be installable. The internal load path is in principal
similar. 0,00001
In 2005 Statoil experienced significant lateral BOP
movements on a subsea well during drilling operations from a 0,000001
MODU in the North Sea. This subsea well head had seen
lasting operations from drilling rigs, accumulating to 0 500 1000 1500 2000
approximately 1 year of operations. The abnormal movements Moment range (kNm)
were explained by a parted conductor casing extension weld,
caused by fatigue loading driven by drilling riser dynamic Figure 1 Comparison of moment cycles density functions
loads. This incident lead to a number of investigations of

2 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 SYSTEM DESCRIPTION MODU and riser  
A WH fatigue analysis is both complex and
The initial part of the drilling operations, installing the
multidisciplinary. The WH system itself is structurally complex
conductor and the surface casing with WH, is performed in
with several interacting components. The interactions between
open sea. Once the WH has been installed a marine drilling
these components are to a certain degree dependent on the
riser and BOP is connected to the WH to establish a conduit
installation during the process of drilling a well.
from well to MODU. The analysis methodology differentiates
The WH system interacts with the surrounding soil which
between modes of operation and denominates this mode of
it is cemented into. Thus, there is a structure-soil interface
operation as operational phase 1. When a BOP/marine riser is
which has to be included in the analysis. In addition to the soil,
connected to the re-entry hub of a XT this is named phase 2. A
the WH system might be connected to a template which adds
phase 3 is defined as open water work over riser C/WO
further boundary conditions to the analysis.
operations.
During drilling, work over/ well intervention or plugging
As long as the BOP or the Lower Work over Riser Package
the well for abandonment, a MODU is connected to the WH
(LWRP) with its associated riser system is connected to the WH
system through a riser. Loads are imposed on the WH system
there will be dynamic loads transferred to the WH. There is a
from both MODU motions and riser. These loads are dependent
need for three different riser models to perform global load
on the hydrodynamic properties of the MODU and riser and the
analysis of all 3 operational phases. A principle model sketch is
environmental conditions during these operations.
shown in Figure 7. The three different models are:
Thus, when assessing the fatigue damage of a WH system,
• Drilling model; BOP landed on the WH, phase 1
structural, hydrodynamic, geotechnical, metocean and
• Completion model; BOP landed on the XT, phase 2
operational knowledge is necessary.
• Work over model; LWRP landed on the XT, phase 3
The purpose of the global load analysis is to establish a
Well construction  load history on the WH system, to be used in a fatigue
A typical primary well construction serves two purposes, assessment. The load history is given as bending moment range
one as a structural foundation for external loads and secondly variation with the WH elevation as a reference.
as a pressure containing vessel for well pressures. During the
design process of a well a design WH pressure is established
and the pressure designs are done to cope with this pressure.
For WH and conductor housings the designs are standardized
and are not well specific. Casing selections are done well
specific to match the well design pressure and the designer
selects outer diameter, wall thickness, material grade and casing
connector type. The lower extension of the wellhead housing is
then sized to match the surface casing selection and the two
elements are joined by welding. Similar considerations are done
for conductor housing and conductor casing which also are
welded together (see Figure 4 for WH system components).
The industry standards applicable to design of these elements
do not put any specific dynamic load requirements on these
welds.[5]
The Petroleum Safety Authority (PSA) in Norway has
imposed a 2 barrier envelope requirement on any well
construction. Details can be seen in NORSOK D-010 [6]. This
applies not only to the finalized well construction but also to
any stage of the well construction work. During drilling the first
barrier is “replaced” by the presence of drilling fluid (drilling
mud) hydrostatic overpressure and the second barrier is
“replaced” by the BOP.
During initial drilling fatigue hotspots in the surface casing
and WH housing will be part of the second barrier envelope,
thus making them critical. If casing strings are combined into
one string the importance of these hotspots are upheld through
all operational phases.

 
Figure 2 System overview

3 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 Environmental loads  Soil support 
Each riser model will be analyzed under several The soil supports the well both vertically and horizontally.
environmental conditions. The resulting load response is The soil resistance depends on the soil characteristics that are
strongly dependent on the environmental conditions applied i.e. assessed by in situ testing at different layers. The stress
the weather or sea states. In a given sea state there is a certain distribution relevant for fatigue depends on the distribution of
distribution of wave heights and wave frequencies. Two horizontal support at the various soil layers. The soil resistance
important spectral parameters are the significant wave height, is given by characteristic of the pressure (P) versus
Hs, and the spectral peak period, Tp, given by a wave spectrum displacement (Y). This relationship is denoted P-Y curves. The
[7]. conductor casing is also supported vertically by the soil due to
There may be more parameters required to describe a friction, however, the distribution of vertical support is not
specific wave spectrum e.g. the peakedness factor, γ applies to important for fatigue calculations according to the present
the JONSWAP (JOint North Sea Wave observation Project) method. The top of a well including soil and cement with a
spectrum which is most commonly used in the North Sea. BOP on top is schematically shown in Figure 3. The riser loads
The long term distribution of sea states on a given location are represented by moment (M) and transverse (V) loads.
(typically an oil field) is compiled in a table listing the Variation in axial tension has a minor influence on the fatigue
probability of occurrence of each combination of Hs and Tp. stress compared to the bending effects, provided that the heave
This table is referred to as a scatter diagram and may be based compensation system works efficiently.
on year round or seasonal observations. When performing a
fatigue load analysis one has to cover all relevant sea states on OVERALL ANALYSIS PHILOSOPHY
the specific location of the actual WH. This implies running The suggested WH fatigue damage analysis method is
simulations where the riser is exposed to all combinations of Hs graphically illustrated in Figure 10. Fatigue damage on existing
and Tp. The final fatigue assessment is then found by doing a wells is to be calculated for operations carried out in the past
weighted sum of the contribution from all relevant short term and for planned operations. The analysis procedure may be
conditions. divided into three parts.
One may also use the above approach when assessing the
i) A local response analysis that includes a detailed non-
accumulated fatigue damage from a set of historical operations.
linear finite element model from wellhead datum and
Based on a listing of the measured weather at the specific
below. Interaction between the structural well
location a scatter diagram representative for the historical
components and soil structure interaction is properly
weather conditions can be made by counting the number of
accounted for. The main result from this analysis is
occurrences of each combination of Hs and Tp. This historical
the load-to-stress curve that describes the relationship
scatter diagram is then used in the fatigue accumulation.
between the riser loads at the wellhead datum and the
stress at the fatigue hot spots. The analysis also
provides the lower boundary conditions of the global
load analysis model.
ii) A global load analysis where the MODU motions and
wave loads on the riser are taken into account. The
results are time series or load histograms of the loads
at wellhead datum, with focus on the bending
moment, in all relevant environmental sea states.
iii) Fatigue damage assessment, where a mapping of the
loads with the relevant load to stress curve is carried
out together with subsequent fatigue damage
calculation. Appropriate S-N curve is applied together
with wave scatter diagrams for the relevant operations
and durations. The final result is the accumulated
fatigue damage.
Load-to-stress curves for a specific fatigue hotspot from
the local response analysis are combined with the global load-
time series to produce stress-time series for the hotspot in
question. Alternatively the same load-to-stress curve can be
combined with bending moment load range histograms to
Figure 3 Schematic of top of well with BOP produce hotspot stress histograms. The results may be
multiplied by a stress concentration factor if relevant. This will
be done per hotspot subject to analysis.

4 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 Fatigue damage is calculated using rainflow counting subsea WH for the only purpose of fatigue calculation using the
(RFC) in the case of stress time series or by summation of S-N approach. The model is not intended for any other
damage from each block in the case of stress histograms. All structural evaluations than fatigue. This assumption allows
relevant operational phases are to be considered. simplification where only the hotspots of interest needs to be
The results are sensitive both to analysis method and input modeled and meshed accurately; the rest of the WH
parameters. Some parameters may be difficult or impossible to components are intended to behave structurally correct. Details
assess accurately. As a result we have experienced that results important for the load transfer through the system must be
from different analysts have varied by orders of magnitude for detailed accurately and meshed accordingly. The FE WH model
the same well. Engineers tend to make conservative choices for shall extend to a depth such that the stresses are not influenced
parameters and methodology that are not uniquely defined. The by the model termination. As rule of the thumb this may be at
suggested analysis methodology is intended to be specific on 50 m below mud line, or to 20 m below the lowest relevant
analysis methodology. hotspot. This approach allows a transition from solid to beam
At present, the applied philosophy is to calculate the elements at the lower part of the model. The main result from
fatigue damage due to the “expected load history”, trying to the static local response analysis is
avoid unnecessary conservatism. However, to acknowledge the • A load-to-stress-curve describing the relation between
fact that the results are sensitive to parameters that are stresses in a hotspot and the bending moment applied
particularly uncertain, sensitivity results to some key on the WH datum (Figure 5)
parameters (i.e. casing hang-off weights and cement level) are • Lower boundary conditions input for the global load
to be reported. Other sensitivity results may also be necessary, analysis (Figure 6)
depending on degree of uncertainty in the input parameters. The • Geometrical Stress Concentration Factor (SCF)
user of the results then has the responsibility to select between The analysis methodology statement report gives detailed
the reported results, based on their own evaluation and guiding on selection of element types, mesh sizing, contact
judgments of the input parameters, compensating measures and definition, modeling of cement, soil spring representation,
associated risk. reference elevation, X,Y,Z convention, boundary conditions,
friction, model loading and solution.
DATA INPUT Components in WH systems are interacting and must be
A review of time spent on analyzing the fatigue modeled by use of contact surface definitions. The precision of
accumulation on several subsea wells, revealed that a majority the contact is essential for the quality of the results. Sliding
of time was spent on collecting, organizing and checking the between the contact surfaces in the housing parts affects the
input data rather than on actual analytical work. Performing load paths between the parts and thus the stress levels in the
WH fatigue analyses requires a lot of input in order to establish structure.
geometrically correct models and boundary conditions. For Contacts shall be modeled with surface-to-surface
existing wells a load history for prior operations and load definition with frictional sliding and deformable surface. The
prediction for future operations are needed to run the dynamic contact segments are located on the surfaces of the 3-D solid
global load analysis. For better efficiency attention has been elements and shall be compatible with the underlying 2nd order
placed on input data collection activities. elements. The same type of contact is used for the cement
Some input data needed will be specific to an individual contact. An isotropic Coulomb friction model is used in the
well. Other data are generic for field/area and others are contact faces. The contact stiffness in the tangential direction
specific to the MODU and its riser/BOP system. All of this should be stiff enough to avoid elastic slip that may influence
information has to be collected from a variety of parties and it the stress results. Full scale tests we have done indicate that the
takes dedicated and skilled personnel to assure the correct input static friction is about twice the dynamic, and two approaches
data. Statoil has established a data storage database for the have been studied to simulate this:
purpose of wellhead fatigue analysis and it includes QA and Stick-slip models have been tested in the local response
change control functionality. Data is stored in the format analyses, however, reliable and consistent results have been
needed for the unified analysis methodology. As this database difficult to obtain. It may also be argued that it is inconsistent to
has been populated it has eased the analysis effort and increased use such a model in a static analysis.
the efficiency in delivering analytical results. Analytical modification of the "static" load-to stress curve
One objective of the analysis methodology is to avoid based on a physical approximation of the behavior has also
confusion and therefore it defines naming conventions, units, been tested; however, this requires engineering judgment and
reference levels and signing conventions for all types of input. manual work from case to case. This introduces a uncertainty
Soil P-Y Curves, as an example, are specified to be generated that is in conflict with the objective of comparability of results.
according to ISO 19902 [8] A static friction factor of 0.25 has been commonly applied,
and is supported by tests, with a dynamic friction factor being
LOCAL RESPONSE ANALYSIS 50% of this value. Published results from full scale tests
The analysis methodology describes in detail how to build indicate a lower value of 0.1[9]. Sensitivity studies we have
a fully parametric 3D, 180 degree symmetric FE-model of a conducted show that the conservatism of using the dynamic

5 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 friction only, compared to a physically more correct curve, is
small in most cases (considering the surface casing weld
towards the WH). As a compromise a constant static friction
factor of 0.15 has been selected as default based on the above
considerations.
It should be noted that the local response analysis is used
to tune characteristic values input to the lower boundary
condition in the global load analysis, and this imposes the
requirement that the local analysis is done prior to the global
load analysis.
The analysis methodology defines parameters naming
conventions, and puts emphasis on the need to use a parametric
modeling approach. There is a need for input parameter
sensitivities studies and parametric models are a sound
Figure 5 Example of load-to-stress curve. (The moment
approach in such cases. With the upfront effort in QA of
range set relatively high in order to cover all relevant cases.)
standardized input data formats, new models can be generated
fast and with minimal possibility of human errors. Model
GLOBAL LOAD ANALYSIS
debugging is done by establishing the parametric model script,
The purpose of the global load analysis is to establish a
and with input data available the analyst can concentrate on
load history on the WH system, which essentially is the lower
being exactly that, an analyst. We believe this is an essential
boundary condition in the global load model. This load history
attribute to the suggested approach.
will then be combined with the results from the local response
finite element analysis in a fatigue assessment. The geometrical
data of the riser and WH system shall be based on input
information as described in MS. The required results from the
global load analysis are time series of the moment at the WH
datum. The analysis methodology gives detailed guiding on
modeling of upper and lower boundaries, rig motions, non
linear flex joint stiffness, heave compensation systems and
hydrodynamic coefficients. A guiding on the level of current
loading is also included. This dynamic analysis provides:
• Time series of bending moment at WH datum
(Figure 8)
• Bending moment load range histograms, presenting
the number of cycles for each load range by use of
RFC (Figure 9)

Boundary conditions lower end 
The lower end of the riser model shall be terminated with
lower boundary representation of the WH global stiffness.
Since the first version of the analysis methodology was
established the recommended lower end boundary condition
modeling approach has changed. The following is our current
and preferred approach.
The WH system is modeled as an elastic beam (length H
and stiffness EI) with one non-linear horizontal spring support
(spring stiffness K) at a small distance (Hst) below the top (see
Figure 6). The local response analysis of the complete WH
system will be used for quantification of these characteristic
values. This eliminates the need for a complex soil and contact
model in two different software systems. The error that is
introduced is not significant, and the benefit is a lower end
boundary condition that mimics the global dynamic behavior of
the WH. We would like to emphasize that the intension of this
Figure 4 Typical WH system and a mechanical “simplified” wellhead representation is to ensure a “tuned”
interpretation of the WH system

6 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 behaavior of lowerr end in the gllobal load anaalysis only, annd that The modeeling of riser fflex-joints maay serve as an example;
the model
m is not su
uitable for thee fatigue damaage accumulattion. thhe analytical methodologyy presented heere requires thhat a real
n
non-linear low
wer flex joint rrotational stifffness be includded in the
r
riser model. The T reason foor this requireement is driveen by the
inntent of the analysis, as this model more effectivvely will
trransfer bendding momennts into thhe subsea wellhead. w
T
Traditionally t flexjoint has been moodeled with a constant
the
r
rotational stifffness. This is usually defined
d as thhe secant
s
stiffness, i.e. the stiffness at the maxim mum deflectioon angle,
u
usually defineed at room teemperature an nd given as the static
v
value, i.e. after
a relaxatioon. Wellheadd fatigue looading is
s
simulated witth a mean fl flex-joint angel of 0 degrrees with
a
amplitudes farr below the mmaximum angeel of the flexjjoint. The
n linear rotaational stiffnesss model will predict higherr stiffness
non
v
values for sm mall amplituudes compareed to the traditional
t
c
constant rotattional stiffnesss approach. Applying a constant
r
rotational stifffness is non-cconservative and
a thus in-coorrect in a
w
wellhead fatigue context buut a conservative assumption n for riser
Figu ure 6 Lower ende boundarry condition: elastic beam m with inntegrity evaluuation. As illuustrated by thhis example we w require
one non-linear hoorizontal spriing at top thhe riser modeel to capture all a relevant ph hysical effectts that are
This method is different froom the approaach outlined inn ISO e
expected to occcur during driilling operatio ons.
136224-1 [10], whhich usually deecouple the global
g load annalysis
model from the conductor
c anaalysis model at the Lowerr Flex
Jointt axis. Achieeving a nonnlinear repressentation of lower 3 risser modelss 
bounndary conditioons for the global load analysis,a inclluding There is a need for three different riiser models inn order to
effeccts from potential wellheadd rotation in the conductoor and ccover all the operational
o phases as descriibed earlier. A principle
effecct of tolerancces (gaps) inn addition to the soil struucture m
model sketch ofo these 3 moddels is shown in Figure 7.
interraction is not easily
e done diirectly in a gllobal analysis.. With The first model
m “Drillinng” comprisess of a conventtional XT
the suggested
s appproach a detailed well support model iss only s
system, (a veertical XT), aand all drillin ng and comppletion is
neceessary in the local
l response analysis, annd this reducees the p
performed witth the BOP onn the WH.
channce of inconsiistency betweeen global annd local analyysis if Second model
m “Complletion” is valiid for a horizzontal XT
suchh modeling whhere included iin both. s
system, all drilling
d is pperformed wiith the BOP P on the
With increasiing BOP sizess and thereby increased atteention W
WH,(model 1)) but the prodduction tubinng is installedd with the
to BOP dynamicss, proper loadd prediction deepends on acccurate X on the WH
XT H, and the BO OP landed on the XT. This increases
bounndary conditions. Recentt experience shows thaat the thhe effective height of thee stack, and hence influeences the
dynaamic responsee of the BOP may contribuute significanntly to f
fatigue damage rate during operations.
o
the fatigue
f damag ge in some caases. BOP dynnamic responsse can Third moodel “Workover” relates too workover opperations,
W (conductor) and
be trriggered by the global flexibbility of the WH u
using a C/WO O riser in openn sea mode withw the riser connected
c
may give rise to a cantilever typpe eigenmodee of the BOP on o top o top of the XT
on X by use of aan LWRP.
of thhe WH. Drill pipee inside the marine
m riser shhall be includded in the
a
analysis modeel given a chhoice of pipee-in-pipe moddeling or
luumped mass techniques.
t
Boun
ndary conditions uppeer end 
The upper ennd of the risser model shhall as a minnimum
Simullation lengtth 
incluude the uppeer flex joint below the rotary. It maay be
conssidered to incllude the diverrter and the rootary in the model.
m A minimuum of 1 houur simulation length is req quired for
t upper end when
This will enable a more correctt interface at the eeach sea state and is based on the recom mmendations inn a recent
usingg a pipe-in-pip
pe model. s
study. Steinkjeer et al [11] cconcludes thatt the uncertainnty of the
f
fatigue damagge rate from each sea staate is attenuating very
s
slowly with siimulation lenggth. This is based on a stuudy of the
R
Relevant ph
hysical effeccts 
c
convergence o the damagee rate in a short term sea staate, using
of
When global load analysiis is done to check integrrity or a many as 300 simulations with differen
as nt seed numbeer for the
fatiggue life of deetails in the riser itself, modeling
m maay not r
random numbeer generator, aand of the unccertainty of thhe damage
alwaays be approppriate for W WH fatigue evvaluation andd vice r
rate based onn long term sstress distribuution and varriation of
versaa. Understandding this diffeerence in inteention is important s
simulation lenngth.
when n making anallytical choicess. The statistical parameeters, such as a standard deviation,
d
c
converges relaatively quicklly in each sinngle time series, for a

7 Copyrright © 2011 by
b ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 given seed number, but the variation between each case with
varying seed number may still be large.
The final fatigue estimate, which is obtained as a weighted
sum of the contribution from each sea state, is less sensitive to
Spectral Peak Period [s]
Hs 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
0-1 83 734 2300 3836 4332 3816 2857 1917 1194 707 404 225 124 67 36 20 11 6 7 22676
1-2 9 319 2435 7725 13998 17408 16665 13251 9215 5813 3413 1901 1019 531 271 137 68 34 33 94245

the simulation length, since the uncertainty from each case 2-3
3-4
4-5
5 161
2
1399
61
1
5181
620
26
10734
2660
301
14711
6162
1465
14911
9023
3679
12083 8279 4996 2738 1394
9330 7386 4756 2609 1263
5552 5624 4162 2402 1138
672
555
461
310
226
165
139
87
53
61
32
16
26
11
5
19
6
2
77819
44789
25052

tends to balance out. Blocking of sea states in the scatter

5-6 13 174 910 2343 3451 3251 2137 1051 409 132 37 9 2 1 13920
6-7 8 113 589 1457 1973 1638 911 366 112 28 6 1 7202
7-8 7 82 392 862 1009 702 318 101 24 4 1 3502
8-9 6 65 256 466 447 252 91 22 4 1 1610

diagram should be done with caution and no single block 9-10

10-11
11-12
7 51
7
1
155
37
7
222
83
23
168
90
37
75
53
31
21
19
14
4
4
4
1
1
1
704
294
118

should contribute to more than 7% of the total fatigue damage.

12-13 1 5 12 14 9 3 1 45
13-14 1 3 5 4 2 1 16
14-15 1 2 2 1 6
15-16 1 1 2
92 1058 4898 13022 24157 34932 42042 43811 40394 33181 24132 15325 8383 3942 1624 617 230 92 68 292000

Cycles Weighted sum

Drilling. Completion. Workover.
of short term
Phase 1 Phase 2 Phase 3
histograms

Surface
Drill floor stack Long-term
histogram

Rotary,
diverter, Moment range
flex joint (Nm)

Riser
Workover Figure 9 Principle sketch of calculation of long term
tensioner
Riser histograms as a weighted sum of short term histograms,
Telescopic
Flex joint using the probability of occurrence of each sea state as
Joint Stress weight. Note that the number of bins in the figure is very
Marine Joint
LMRP Riser low for illustrational purposes.
/BOP LWRP

XT FATIGUE DAMAGE ASSESSMENT

WH
datum XT The fatigue assessment is based on the conventional
principle using S-N curves and the Miner-Palmgren hypothesis
for fatigue damage accumulation.
Mudline
The damage assessment is done by mapping the time series
of the loads acting on the well (Figure 9) with the load-to-stress
curve (Figure 5) to obtain time series of the hotspot stress. If
Figure 7 Sketch of riser model main components relevant, additional hotspot stress concentration factors may be
applied before these stress-time series are subjected to rainflow
Moment
counting. Relevant S-N curve is then selected according to DnV
(Nm) recommended practice for fatigue calculations for offshore
structures [12]. This document is available for free download
Time (s)
from the publisher. The fatigue calculation is carried out for the
number of sea states as specified in the relevant scatter
diagram, and the accumulation reflects the probability of sea
Cycles
R.F.C state for the duration of the operational phases (1 - 3) that
contribute to the fatigue damage.

Alternative approach  
Instead of computing the time series of stress, one may as
an alternative use the global analysis results in terms of
Moment
(Nm)
range
histograms of the load ranges acting on the well, as illustrated
by Figure 9. In this case the load histograms obtained from
Figure 8 Principle sketch of calculation of short term rainflow counting of the load-time series are assessed for each
bending moment load histograms. Note that the number of individual sea state. Each load range in the histogram is then
bins in the figure is very low for illustrational purposes mapped with the load-to-stress curve with the same amplitude
to the positive and negative side. This provides a histogram of
stress ranges that is evaluated against the S-N curve for damage
accumulation.

8 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 Damage accumulation  database. Effective tools may be developed to speed up
calculations for a number of different load-to-stress curves and
The total accumulated fatigue damage  is a
potential new information about relevant scatter diagrams.
summation of the damage from all relevant phases that
Deviations from a perfectly vertical conductor housing
contribute, and may be written on the form:
and/or a perfectly vertical WH due to installation tolerances are
probable. Analysis of such a case would yield increased
∑   asymmetry in the load-to-stress curve. However, the average of
the tension and compression part of the load-to-stress curve
Where  is the accumulated damage during phase i. The would remain comparable to the vertical case, and the
accumulation comprises phases that occurred in the past, and consequence on fatigue life is therefore evaluated to be low. A
phases that are planned in the future. similar reasoning may be applied in the case of a constant mean
This equation is to be calculated a number of times at each load; e.g. due to current forces or MODU offset.
hotspot in order to cover: Pretension of the WH to one side (e.g. by MODU offset),
• Results for all cement levels as required causing the conductor housing and the WH to behave as a rigid
• Results with estimated hang-off weights of casings body, would, theoretically, lead to a potential reduction in
• Results with hang-off weights set to zero. fatigue loading. Such a situation is not considered in the present
This latter case may be of interest if the extent of down analytical methodology.
weight is uncertain or if temperature effects are significant.
Results are also calculated as a function of time, and the Documentation 
relative contribution from the various phases can then be seen.
The following assumptions should be noted: The analytical approach imposes specific requirements on
• Stress variation at the hotspot is uniquely expressed by the analysis report. This is done to achieve better basis for
the variation in moment and associated shear force at comparison between analysis on different wells or if the same
the WH datum; i.e. no other loads contribute to the well is analyzed by several parties. The results shall be
fatigue. documented both numerically and graphically. Any deviations
to methodology and assumptions shall be clearly stated. Level
• The weld and cross-sectional properties are constant
of conservatism shall be commented. The discussion should
around the circumference.
clearly identify if the deviation is uniquely to the analysis at
• Hotspot stress is calculated on the compression and on
hand or if it of a more general nature. This will be important
the tension side, and the load to stress curve may be
input to future updates of this analytical method.
unsymmetrical in compression and in tension.
• The load-to-stress curve is the same for loading and
FURTHER WORK
unloading; i.e. potential hysteresis effects are not
The following are under consideration for the next revision
accounted for.
of the analytical method:
• Preloaded WH systems
Special considerations   • Connectors
Current loading may be applied in the analysis (i.e. a lower • Fracture mechanics
10% fractile), and consequently the load-time series get a mean • VIV
value different from zero. The mean value of the load-time • Thermal effects
series must then be subtracted before the mapping with the • M-N Diagram
load-to-stress curve. This situation is implicitly covered when • Monitoring
the alternative approach using load histograms from rainflow
counting of the load time series is used.
The time series of hotspot stresses at 16 equally spaced
SUMMARY
points around the circumference is needed if out of plane
Subsea wellheads are dynamically loaded during drilling
motion of the riser is allowed for; e.g. due to short crested sea
operations with a riser connected to the wellhead. The
or application of specific directional information. If current
magnitude of these loads appears to be higher today than what
load is also applied with out of plane response, the mean value
has been previously been published [2]. The conventional
should be subtracted from the time series of the moment
subsea wellhead systems fatigue resistance is dependent on
component arising from the load corresponding to the current
well construction design and operational issues. The internal
direction. Then the resulting moment should be calculated, and
load sharing between conductor and surface casing will vary on
mapped with the load-to-stress curve. Note that this situation
an individual well basis and this calls for individual well
goes beyond the present scope.
analysis. A potential WH fatigue failure poses a threat to well
The load histogram approach is effective in case of in-
integrity.
plane loads only. Load histograms require very little storage
The present problem involves many steps, many analysis
space, are easy to visualize and are well suited for storage in a
choices and a number of uncertain parameters. Engineers used

9 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 to verification analysis of ultimate strength integrity tend to REFERENCES
make conservative choices for parameters and methodology
that are not uniquely defined. Integrity verification of fatigue [1] Hopper, C. T., 1983, "Vortex Induced Oscillations of Long
capacity may end up being overly conservative. The level of Marine Drilling Risers," Second Intl. Deep Offshore
conservatism, measured in terms of probability of failure, may Technology Conference Valletta, Malta 1, pp. 97-109.
be quantified by use of structural reliability analyses. Structural [2] Dykes, C. E., Hopper, H. P., and Jones, R. G., 1989,
reliability is a tool often used in modern code development, but "Development and Implementation of a Universal Wellhead
has not yet been applied to WH fatigue analysis. System," SPE/IADC Drilling Conference, New Orleans,
Dykes et al stated in their work on unified exploration Louisiana,USA, pp. 655-664.
subsea WH in 1989 that fatigue was a design load case [2]. [3] Osmundsen, P., Roll, K. H., and Tveterås, R., 2010,
Fatigue assessment of subsea WH is a limit state problem that "Exploration Drilling Productivity at the Norwegian Shelf,"
needs explicit consideration of the fatigue load history. Existing Journal of Petroleum Science and Engineering, 73(1-2), pp.
relevant codes are ultimate load based and do not provide 122-128.
proper guidelines for fatigue. [4] Dnv,2010,"Offshore Update",No 2,
Analysis of subsea wellhead accumulated fatigue damage www.dnv.com/offshoreupdate,
is multidisciplinary and input data collection and QA is [5] Iso, 13628, Petroleum and Natural Gas Industries -- Design
important. We have placed high importance on the input data and Operation of Subsea Production Systems Part 4: Subsea
collection process and Statoil has established a database for wellhead and tree equipment, 1999
analysis purposes. A unified analysis methodology has been [6] Norsok, D-010 Rev. 3 Well Integrity in Drilling and Well
developed and is described in the current MS document. A JIP Operations, 2004
has been formed between 11 operators to further develop this [7] Faltinsen, O. M., 1990, "Sea Loads on Ships and Offshore
analytical approach and the aim is to form a wellhead analysis Structures", Cambridge University Press, Cambridge, UK,
recommended practice document [4]. [8] Iso, 19902 First edition, Petroleum and Natural Gas
Industries- Fixed Steel Offshore Structures, Soil reaction for
piles under lateral actions 2007
NOMENCLATURE [9] Milberger, L. J., Yu, A., Hosie, S., and Hines, F., 1991,
MS Wellhead Fatigue Analysis Method Statement "Structural Requirements for the Effective Transfer of
JIP Joint Industry Project Environmental Loadings in a Subsea Wellhead System,"
BOP Blow Out Preventor Offshore Technology Conference, Houston, Texas, USA, pp.
MODU Mobile Offshore Drilling Unit 103-113.
S-N Curve Stress vs Number of Cycles resistance curve [10] Iso, 13624-1, Petroleum and Natural Gas Industries-
OD Outer Diameter Drilling and Production Equipment, Design and operation of
LWRP Lower Workover Riser Package marine drilling riser equipment, 2009
XT X-mas tree [11] Steinkjer, O., Sødahl, N., and Grytøyr, G., 2010,
Hs Significant wave height "Methodology for Time Domain Fatigue Life Assessment of
Tp Spectral peak period Risers and Umbilicals," OMAE, 29th International Conference
JONSWAP JOint North Sea Wave observation Project on Ocean, Offshore and Arctic Engineering, Shanghai, China
spectrum [12] Dnv, RP-C203, Fatigue Design of Offshore Steel
P-Y curves Pressure vs displacement curve Structures 2008
RFC Rain Flow Counting
WH Wellhead
C/WO Completion/Workover riser
VIV Vortex Induced Vibrations

ACKNOWLEDGMENTS
The authors would like to thank Statoil ASA and DNV for
the permission to publish this information.

10 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms

 Figure 10 Schematic of overall analysis philosophy

11 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/25/2014 Terms of Use: http://asme.org/terms