Vous êtes sur la page 1sur 16

SPE-175441-MS

Detailed Approach for the Assessment of Accumulated Wellhead Fatigue


Kingsley Sunday, Phil Ward, Cary Griffin, 2H Offshore Engineering; Peter Lorenson, Apache North Sea

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Offshore Europe Conference & Exhibition held in Aberdeen, United Kingdom, 811 September 2015.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the
written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to a n abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
This paper presents a case study of the use of an enhanced analytical approach, combined with the use of structural
monitoring, as an enabling technique in the planning of sidetrack operations on a well in 110m water depth in the Northern
North Sea.

Subsea wellheads are fatigue sensitive structures due to their exposure to dynamic loading transferred from connected riser
systems. Intervention and workover operations over the life span of the well each contribute to the total fatigue damage
accumulated in components that cannot be readily inspected. Accurate assessment of fatigue damage accumulation in ageing
wells using advanced analytical techniques is often necessary to quantify the residual life of these components. The ability to
reliably estimate this life can greatly affect the planning and viability of operations to enhance the productivity of ageing
wells, particularly in the North Sea.

A simplified screening approach to wave and current fatigue assessment, typically applied to new-drill wells, with
comparatively fatigue resistant hardware is unlikely to be sufficient to provide the level of confidence required when
planning intervention operations on a brownfield development featuring less fatigue resistant hardware. Further
sophistication, more involved than the typical approach, is necessary to avoid over-conservatism and to provide confidence to
proceed with the planned operations. To thoroughly assess the fatigue accumulation in an ageing well, a detailed operational
history including hindcast or measured weather data is usually sourced. The assessment is conducted to ensure that the actual
operational conditions during periods when risers were connected are as accurately represented as possible. However, this
may not be sufficient to bring the resulting fatigue damage within the allowable limits set using the traditional code-defined
factor of safety approach. In this scenario further analytical methods are required.

The use of pre and post-failure analysis and the development of a monitoring strategy to maximise both the likelihood of
identifying a fatigue failure during the operations and the ability to subsequently calibrate the analytical models are
discussed. The applicability of these techniques to other operations and geographical locations is also described.

Introduction
The need for extending operational and production life of ageing subsea wells in the North Sea can require intervention
operations and the reconnection of the well to a marine drilling riser and subsea blow out preventer (BOP) stack. Depending
on the age of the well, several previous BOP-landed operations may have been conducted, with each causing potentially
significant accumulation of fatigue damage in the wellhead. Further, older wellhead equipment was typically not designed to
enhance fatigue resistance and thus may have accrued fatigue damage at faster rates than seen in more modern designs. Harsh
subsea environments, such as in the Northern North Sea can add further severity to fatigue damage accumulation.

To confirm the residual fatigue capacity of older wells, a historical fatigue accumulation analysis must be completed. This is
particularly important where the planned intervention operations are to be completed using newer vessels, especially 4th to 6th
generation semisubmersibles, which typically carry larger and heavier subsea BOP stacks required to enhance their
deepwater capabilities.
2 SPE-175441-MS

The case study presented in this paper describes analytical techniques employed for the fatigue assessment of a stand-alone
tieback well located in the northern North Sea in a water depth of 110 m (361 ft). The objective of the study is to calculate the
accumulated wellhead system fatigue damage and remaining fatigue life to enable planned intervention operations to proceed
using a 3rd generation semi-submersible vessel. The well was originally drilled in 1998-1999 using a 2nd generation semi-
submersible vessel.

The iterations in refinement of the analytical approach, required due to the insufficient residual fatigue lives calculated in the
first pass approach are presented in this paper, alongside a description of the preliminary calibration of the final analysis
model using in service structural monitoring data.

First Pass Analytical Approach


Initial Model Description
A model of the drilling riser configuration connected to the well during the initial drilling and completion operations is
generated using industry standard Finite Element Analysis (FEA) software.

The response of the riser system to wave and current loading can be significantly affected by the joint arrangement. For
example, placing buoyant joints rather than slick joints in the wave zone can increase the drag diameter of the riser in the
location of peak wave energy and hence increase the loading in the riser. In deeper water, placing a long length of slick joints
in a region of high current speeds can exacerbate any vortex-induced vibration (VIV) lock-on effects. Therefore, an
understanding of the actual riser joint arrangement used during previous operations is important to ensure the fatigue damage
rates calculated are appropriate for those operations. Obtaining the necessary data from historical drilling reports can be
difficult as this information may not have been recorded in all cases, or the information may have been lost during handover
of the asset between license holders. In the case considered here, information of the in-place riser configuration used for the
operations is unavailable, hence the riser stack-up has been assumed to include buoyant joints and 222kN (50kips) base
tension at the lower marine riser package (LMRP).

The riser system model consists of the following:


Two 21inch (0.53m) diameter slick joints of standard 50ft lengths and pup joints to make-up stack length;
Two 50ft buoyancy joints;
Drill-Quip SS-10 subsea wellhead configuration with stick-up height of 4m (13ft);
Tension applied to the riser system in order to maintain an overpull of 222kN (50kips) at the base of the LMRP.

The subsea wellhead, conductor and surface casing are modelled using a pipe-in-pipe approach to capture the interaction and
load sharing between the conductor and surface casing strings. The well design incorporates a 30inch OD conductor with
1.5inch wall thickness in the uppermost joint swaging to 1.0inch lower down. The surface casing is 20inch OD by 0.635inch
wall thickness.

Wellhead Modelling and Critical Locations


Accurate modelling of the motion tolerances between the low pressure housing (LP housing) and high pressure housing (HP
housing) of the wellhead system is critical in defining the fatigue damage rates in individual stress-raisers or hotspots within
the wellhead. In this case the interaction between the LP housing and HP housing is modelled using a non-linear stiff spring
which does not permit relative movement between both wellhead components. The stiff connection is modelled with an
equivalent node at the top of the LP housing connecting the HP housing, together with stiff springs along the contact surface
between the LP housing and HP housing as shown in Figure 1.

The system critical locations established following initial system response checks include the following:
BOP or subsea tree to wellhead connector;
The 18 10K HP housing body towards the bottom section;
HP housing to surface casing weld;
30 LP housing body towards the bottom section;
LP housing to conductor pipe weld.
SPE-175441-MS 3

Figure 1 Schematic of Subsea Wellhead System with Stiff Connection for Rigid-lock

Fatigue Data
BS 7608 (1993) and DNV-RP-C203 (2012) define a range of fatigue design curves for seawater cathodic protected
components that are used in the fatigue damage assessment. In lieu of component specific details from manufacturers, these
codes are employed as guidance for the selection of appropriate fatigue curves for specific locations. The fatigue design data
as given in the code are generated from fatigue testing to failure of weld specimens and components under consideration, and
the curves are based on the mean-minus-two-standard-deviation curves (97.7% probability of survival) for relevant
experimental data. Local stress concentration due to geometriccreated by joints and the weld profile is accounted for by the
joint or weld classification and corresponding S-N curves by means of application of Stress Concentration Factor (SCF) to
the cyclic stress for fatigue life calculation.

The applicable design fatigue curves and applied SCFs for the wellhead system are presented in Table 1.

Fatigue Detail Fatigue Safety


Detail Factor
Class SCF
Wellhead to Subsea Tree Connector B(2) 6.0
LP Housing (Body) B1(1) 3.0
HP Housing (Body) B1(1) 3.0 10
LP Housing Weld F(2) 1.1
HP Housing Weld F(2) 1.3
Note:
1/ S-N curve information from DNV-RP-C203 (2012)
2/ S-N Curve information from BS 7608 (1993)
Table 1 Fatigue Design Data and Safety Factors
4 SPE-175441-MS

Hindcast Metocean Wave and Current Data


Assessment of accumulated fatigue damage from previous drilling, completion and intervention operations is best estimated
using measured seastate data recorded during the historical operations. Where this is not available, hindcast metocean data
can be used instead. Using hindcast data removes the inaccuracy associated with applying long-term average conditions to a
short duration campaign over a specific time period. Hindcast data that includes the directionality of wave loading can be
used in combination with the specific vessel heading used in historical operations to ensure the correct vessel response is
modelled throughout the assessment.

Hindcast weather data including distribution of significant wave height and associated peak period by direction and current
profiles for the actual periods and duration of the operations are applied in this case. The operations with the BOP stack
connected occurred from February to June 1999. The hindcast weather information was derived from considering several
ocean sources recorded from 1979 to 2013, including the following:
Wave directionality: assessment performed using directional wave data which ensures that the stresses leading to fatigue
damage are not concentrated or localised at a single location around the circumference of the riser pipe. This is achieved
by using the corresponding vessel heading properties.
Wave seasonality: seasonal weather variations are modelled by sourcing specific data only for periods and durations for
which the riser system was connected and exclude times of disconnect due to harsh weather conditions, deployment of
the subsea tree, equipment repair or maintenance.

The selection of an appropriate background current for first order fatigue analysis is required to give a realistic representation
of the average hydrodynamic damping from current present through the operation. 95% exceedance to 80% exceedance
current profiles are typically considered depending on the duration of the operation. DNV-2011-0063(2011) recommends a
P10 percentile profile as background current for fatigue analysis. This current speed is exceeded 90% of the time and will
generate a conservative low damping level in the analysis. The measured current profile corresponding to this exceedance
level is applied in the current work.

Through depth current data is used for VIV fatigue analysis to simulate the current loading along the riser system. Scatter
diagrams of surface current speeds for the field have been combined with the extreme current profiles to generate the through
depth exceedance profiles for the analysis using the Weibull distribution.

Soil Profile Data, Interpretation and Modelling


As with most of the ageing well locations, the available soil information is not detailed enough for capturing the actual
interaction between the conductor pipe and the surrounding soil along the entire length of the model below mudline. To
progress with the assessment, assumption of undrained shear strength, angle of internal friction, and submerged unit weight
have been interpolated and extrapolated to cover the entire pipe column based on the available data for specific points along
the conductor length. The scenario may potentially represent a less stiff or overly stiff soil profile. Description of the soil data
used for the case study is presented in Table 2.

Submerged Undrained Angle of


Depth Below Seabed (m) Lithology Unit Weight Shear Strength Internal
(kPa) (kPa) Friction (deg)
0 - < 0.5m Silty SAND (Holocene) 14.0 - 20
Firm to very stiff sandy CLAY
< 0.5 74m 15.5 40.0 70.0 -
and clayey SAND
Dense SAND and very stiff to
74m + 15.5 70.0 -
hard clay (Shackleton Formation)
Table 2 Description of Soil Data for the Well Location

Based on the available soil description, the API RP 2A-WSD (2000) recommended approach is used to generate the pressure-
displacement curves along the conductor depth for simulating the conductor pipe-soil stiffness interaction. The soil resistance
is modelled using non-linear springs.
SPE-175441-MS 5

Other Considerations and Uncertainties


The presence of uncertainties in the available design data, or as is more common when assessing historical operations, the
unavailability of certain design information, leads to assumptions in order to progress with the assessment. These
assumptions can result in significant under-prediction or over-prediction of system response and fatigue damage.

A number of uncertainties in the design data used for this analysis include the following:
The level and condition of the cement level both internal and external to the conductor. It is assumed that cement is
present up to the mudline between the LP housing/conductor and HP housing/casing section;
The presence of any scour at the seabed. It has been assumed that no scour has occured.

First Order Fatigue Analysis Methodology


Wave induced fatigue assessment is conducted using a time domain Jonswap seastate approach. Structural damping of 1% for
buoyant drilling riser joints and 0.3% for riser slick joints is applied in the analysis (Padelopoulos et al. 2012). Background
current is applied co-linearly with wave loading. An 80% exceedance current profile is applied in the case study based on
hindcast current data for the actual duration of operations conducted. Care should be taken in the application of higher
background current profiles as improved but unrealistic fatigue damage may be predicted.

The results of the first order fatigue damage along the wellhead system for the first pass analytical approach are presented in
Figure 2. The allowable fatigue damage utilisation is 1.0 and accumulated fatigue damage exceeding 1.0 indicates no
remaining fatigue life. The unfactored fatigue damage predicted at the HP housing weld is 2.12, which is the most critical
location. This is followed by the wellhead to subsea tree connector and LP housing weld that have predicted fatigue damage
of 0.60 and 0.31 respectively. Initial available wellhead system information suggests that the HP housing weld elevation is
below the LP housing weld elevation. This model approach induces more stress at the HP housing weld as the transmitted
motions are shared between the LP housing weld which is suituated above the HP housing weld location.

Figure 2 First Order Fatigue Damage along Wellhead System First Pass Analytical Approach

Vortex Induced Fatigue Analysis Methodology


The case study considers a number of current profiles representative of historical conditions. Fatigue damage from each
current profile is factored according to the probability of occurrence of the profile to determine the VIV accumulated fatigue
damage covering a range of exceedance levels and extreme events for the operations. Palmgren-Miner rule summation for the
calculation of fatigue damage, similar to the first order fatigue damage calculation, is also applicable for current-induced
fatigue. VIV suppression devices such as strakes and fairings are not accounted for in the case study.

Result of the VIV fatigue damage along the wellhead system for the first pass analytical approach is presented in Figure 3.
The HP housing weld and wellhead to subsea tree are predicted to have fatigue damage of 0.018. The predicted VIV fatigue
6 SPE-175441-MS

damage is insignificant relative to the corresponding wave-induced fatigue damage; hence VIV damage does not change the
overall combined fatigue results due to both wave fatigue and current fatigue.

Figure 3 VIV Fatigue Damage along Wellhead System First Pass Analytical Approach

The resulting combined fatigue damage due to wave and current effects is presented in Table 3. The implication of the
predicted accumulated fatigue damage is that the allowable fatigue utilisation of 1.0 has been exceeded and the wellhead
system has no fatigue life remaining to allow planned operations to proceed, Hence model refinement and fatigue life
improvement is required as described in the next section of this paper.

Fatigue Damage
Combined Fatigue
Critical Location First Order Vortex-Induced Damage
Fatigue (FOF) Vibration (VIV)
Wellhead to Subsea Tree Connector 0.602 0.018 0.620
LP Housing Weld 0.307 0.018 0.325
HP Housing Weld 2.121 0.009 2.13
LP Housing Body 0.007 0.043 0.050
HP Housing Body 0.005 0.020 0.025
Table 3 Combined FOF and VIV Fatigue Damage First Pass Analytical Approach

Fatigue Life Improvement


Model Refinement
The refinements performed in order to improve the system response and fatigue life are described below:
Wellhead System Mesh Refinement: Investigation into the mesh at the wellhead system, especially at the hot-spots
revealed that an optimal mesh was yet to be achieved for the initial model approach. Increased mesh refininement in the
regions of load transfer between the housings was found to be necessary to remove artificially high stress variations
resulting from load distribution between the housings;
Housing Weld Profile Refinement: The initial modelling approach considered the LP housing and HP housing, together
with their transition welds, as consistent pipe sections with constant diameter and thickness. The model was refined to
capture the gradual transition of the throated section from the main housing cross section to the weld. Shorter elements
than originally included are added to the model in this region. The representation of the transition region in this manner
enables a more gradual distribution of stress into the weld location as supposed to abrupt and aggressive stress
transmission from a thicker to a thinner section. The housing throated section to the weld discretisation is demonstrated
in Figure 4.
SPE-175441-MS 7

Figure 4 Schematic of Subsea Wellhead System Showing Discretisation of Housing Throat Leading to Welds

Compressive Stress Reduction Factor


DNV-RP-C203 (2012) allows a reduction factor to be applied to any stress cycle that is wholly or partly in compression, if
the base material is not significantly affected by residual stresses due to welding. The applicable stress ranges are reduced by
the factor fm, derived from the following equation, before entering the S-N curve fatigue calculation.

+ 0.6
=
+
where
t maximum tension stress
c maximum compression stress

Of the five identified hotspots of concern, three qualify for the reduction factor approach. These three locations are the LP
housing, the HP housing and the wellhead to tree connector as these components are all forged and are not affected by
residual stresses due to welding. The stress reduction factor is applied only to the wellhead to tree connector as the most
critical of the three hotspots to which the stress reduction factor is applicable. The same approach may not be used for either
of the fatigue critical welds.

Wave Spreading
Wave energy at a given location has an angular distribution as well as a distribution over a range of frequencies. The
phenomenon wave spreading is the angular distribution of the wave energy from the quoted dominant direction. Previous
study indicates that spectral models without spreading can over-predict significant wave height refraction by as much as 20
percent when compared to directional spectral models (CETN 1985). Proper application of wave flow velocities is dependent
on the correct representation of the directional spread wave energy and neglecting the spreading effect results in over-
prediction of the flow velocities.

A single-peaked directional spectrum described by the product of two functions is the basis for all parametric spreading
functions. The functions are:

(, ) = (). (, )
8 SPE-175441-MS

where
S (f, ) = directional spectra; density function
S (f) = one-dimensional energy spectral density function
D (f, ) = angular spreading function
f = frequency in hertz
= direction in radians

The equation describing the angular spreading function used in this case is as follows:

2 2 (

() = { 0 ), ( + 0 ) < < ( + 0 )
2 2
0,
where
0 = mean wave direction in radians

However, careful understanding of the wave behavior is an important aspect in the use of wave spreading in structural
assessments. A schematic of a typical wave spreading function and the number of equally segmented directions is shown in
Figure 5.

Figure 5 Schematic of Wave Spreading Function and Directions

The base case fatigue assessment did not account for wave spreading around the riser pipe as wave spreading analysis
requires increased run time, which depends on the number of wave spreading directions as presented in Table 4. Application
of wave spreading in analysis using the directional spectra approach is limited to short crested seas. Wave spreading over 9
directions is applied in the case study. Appropriate selection of wave spreading parameters depends on the quality of the
available metocean data and information on short crested seas for the region of interests.

Wave Spreading Anaysis Run Time % Increase in Run


Direction (Hour) Time Compared to No
Wave Spreading
Direction = 0 0.50 -
Direction = 3 1.40 180%
Direction = 7 2.50 400%
Direction = 9 3.00 500%
Direction = 11 3.40 580%
Table 4 Analysis Run Time Comparison for Wave Spreading

Result of Improvements on Fatigue Life


The results of the wellhead fatigue life following application of model refinement and the improvements described above are
presented in Figure 6. The following considerations have contributed to the improvement of the wellhead fatigue life:
Wellhead system model refinement and discretisation of throat section leading to weld;
Application of compressive stress reduction factor;
Wave spread application short crested wave.

The HP housing weld and tree connector experienced fatigue life improvement, while the LP housing fatigue life reduced due
to improved stress sharing relation between the HP housing weld and LP housing weld. The HP housing weld fatigue life
improves by 67% which corresponds to reduction in fatigue damage from 2.121 initially to 0.692. The wellhead to tree
SPE-175441-MS 9

connector fatigue life improves by 50%, which corresponds to reduction in fatigue damage from 0.602 initially to 0.301. The
reduction in the LP housing weld fatigue life due to improved stress sharing relationship with the HP housing weld is
predicted to be -20% which corresponds to an increase in fatigue damage from 0.307 initially to 0.369.

Figure 6 Fatigue Damage Results after Application of Improvement Technique

The fatigue life presented in Figure 6 is unfactored first order fatigue life. Applying DNV-RP-C203 (2012) recommended
code factor of safety of 10 will cause the fatigue damage to exceed the allowable utilisation limit of 1.0 which indicates no
fatigue life remaining for planned operations. To enable operations to proceed, failure assessment (what-if approach) is
required to understand the likelihood of failure and also define the sequence of failure during operations in order to quantify
the implications of fatigue failure.

Failure Analysis What-If Approach


Fatigue failure analysis is performed for the planned operations to ascertain the effects of failure on the wellhead system and
the subsea tree connector. The objective is to determine the effects and sequence of failure should either the HP housing weld
or the LP housing weld fail within the wellhead during planned operations. The residual strength capacity of the well for a
failed wellhead to subsea tree connector scenario is considered to be zero for connected operations. Identification of failed
wellhead system due to fatigue is by in-service structural monitoring of motions which is discussed in the next section of this
paper.

The failed HP housing weld scenario considers the wellhead and conductor system but with the HP housing weld and
attached surface casing removed from the model to simulate failure of the HP housing weld. Damage rates along the LP
housing and conductor system are extracted assuming the HP housing weld has failed after reaching a cumulative damage
utilization of 1.0. This gives an understanding of how much loading and for how long the system can sustain ongoing
operations if there is no resistance contribution from the HP housing and casing system. A further model is developed in a
similar manner to represent failure of the LP housing weld. In this case, the LP housing and conductor pipe is removed,
leaving the HP housing and casing system as the only structural load bearing system below the BOP stack. Demonstration of
the failed HP housing weld and failed LP housing weld scenarios are presented in Figure 7.
10 SPE-175441-MS

Figure 7 Fatigue Failure Analysis: Failed HP Housing Weld and LP Housing Weld Scenarios

Figure 8 presents the predicted failure sequence for a failed HP housing weld scenario. The wellhead to subsea tree connector
is predicted as the next component to fail in the event of a failed HP housing weld, with the fatigue life reducing from 292
days for the non-failed case to 284 days in the failed HP housing weld scenario. The fatigue damage rates at the wellhead to
subsea tree connector and LP housing weld are predicted to be 1.5 times faster after failure of the HP housing weld.

Figure 8 Fatigue Failure Analysis Result: Failed HP Housing Weld Scenario

The failure sequence in the event of a failed LP housing weld is shown in Figure 9. In this scenario, the HP housing weld is
predicted as the next failure component with a fatigue life reduced from 307 days in the non-failed case to 214 days in the
failed LP housing weld case. The loss of integrity in the LP housing weld results in damage accumulating in the HP housing
weld 5 times faster after the failure than before.
SPE-175441-MS 11

Figure 9 Fatigue Failure Analysis Result: Failed LP Housing Weld Scenarios

In both failure cases, operations can continue for a period of at least four days after the initial failure before the second failure
would be expected to occur. Therefore, if the initial failure could be identified, through a change of motion response of the
system for example, sufficient time to permit safe disconnection from the well would be available. However, for the more
critical HP housing weld failure case, for which the secondary failure is predicted to be the subsea tree to wellhead connector,
the failure may not be readily noticeable through visual inspection of the subsea equipment. Therefore, the use of structural
monitoring is recommended to improve the likelihood of detection of a change in motion response. The use of in-service
structural monitoring also allows the system response due to external loads to be captured and compared to the response
predicted by the analytical model. Such comparison, of both the magnitude and frequency components of the motion
responses, allows for calibration of the model to reduce differences between predicted and measured response, and hence
enable improvement of the fatigue estimates through re-analysis using the calibrated model.

In-Service Structural Monitoring


System Design
In-service structural monitoring of the riser system is achieved through the use of motion measurement sensors secured to the
riser and subsea BOP stack. The motion sensors measure acceleration in 3degree of freedom (3DOF) and angular rates in 2
degree of freedom (2DOF) for the entire duration of the operations. The devices are retrieved every fortnight for processing
and assessing current integrity of the wellhead system. The primary sensors for the wellhead motion measurement were
installed at the base of the BOP stack as accessibility to the wellhead system for sensor installation is not practicable, as this
is very close to the mudline. The system is designed to capture the structural responses at the fatigue/strength critical
locations (LP housing weld, HP housing weld, and connectors). The sensor configuration and as-installed arrangement along
the riser system during operations is presented in Figure 10. The reason and purpose behind the arrangement of the
monitoring sensor assembly are given in Table 5.
12 SPE-175441-MS

Sensor Purpose
Qty Capability
AssemblyLocation Measurement Design Objective
3D Acceleration To simulate the vessel response
Vessel 1 Vessel motions in 6DOF
3D Angular Rate motions
Top of riser motions due to To calculate the riser fatigue
Top of 5ft Riser Pup Joint 1 wave, current and vessel damage due to wave and
motions current
Base of riser motions due to To calculate the riser fatigue
Base of 50ft Riser Slick
1 wave, current and vessel damage due to wave and
Joint
motions current
To calculate the wellhead
Top of BOP stack motions due fatigue damage due to
BOP Top 1 3D Acceleration to wave, current and transmitted motions, wave and
2D Angular Rate transmitted vessel motions current, also also to compare
against primary sensors
Base of BOP stack motions
BOP Bottom 2 due to wave, current and
To calculate the wellhead
transmitted vessel motions
fatigue damage due to
For redundancy and to enable
transmitted motions, wave and
continuous data recording at
Spares at BOP Bottom 2 current
base of BOP stack during
retrieval points
Table 5 Design Purpose for As-Installed Sensors Assembly for Monitoring Operations

126.00 m 1 Motion Sensor


Assembly:
16.5 m Vessel
3D Acceleration
2D Angular Rate

93.00 m

2 Motion Sensor Assemblies:


110.0 m Placed at the top and
bottom of the Riser
3D Acceleration &
2D Angular Rate

17.50 m
3 Motion Sensor Assemblies:
1 at the top of BOP
2 at the bottom of BOP
11.50 m 2 spares for changing and
retrieval of sensors at the bottom
6.00 m of BOP
3D Acceleration &
2D Angular Rate
Mudline

Notes:
3D: 3 Dimension
2D: 2 Dimension
Modelled to -50m

Figure 10 Sensors Assembly Configuration for Monitoring Operations


SPE-175441-MS 13

Response Comparison
The analytical system response from the refined global analysis model is compared with the measured system response as
presented in Figure 11. The variation in standard deviation of acceleration at the base of the BOP, in the vessel fore-aft/port-
starboard direction, throughout the entire monitoring period is compared for the analytical and measured response. The
significant wave heights experienced during the operations are overlaid. Both the analytical and measured accelerations
increase in periods of higher waves, as expected, with a clearer correlation in magnitude between the response and the
measured wave height seen in the analytical and measured response for the duration of operations. The predicted and
measured response differed by a factor of 2, with greater motion predicted by the analytical model for the entire duration of
operations. The riser was disconnected between the periods of 9th December to 18th December, which represents the period
with no data recorded.

Figure 11 Predicted Motions vs. Measured Motions Comparison

Comparison of the analytical and measured frequency response spectra of the riser system for a measured significant wave
height of 5m and peak period of 8.6sec is presented in Figure 12. The measured response spectrum shows two primary peak
regions: the wave excitation region between 6.0 seconds to 9.0 seconds (0.17 Hz to 0.11 Hz) and the wellhead system natural
frequency region which is between 2.0 seconds to 3.3 seconds (0.50 Hz to 0.30 Hz). The predicted system primary and
secondary frequency responses are 3.9 seconds (0.26 Hz) and 2.8 seconds (0.36 Hz) respectively, which overlaps with the
lower portion of the wave or current excitation zone and is likely to result in high predicted accumulated fatigue damage in
the wellhead system.

Further study is required to better understand and calibrate the global analysis model against the measured system response.
Calibration of the model is required to match the predicted system modal frequency reponse with the measured frequency
response and obtain comparable magnitudes of response. This process also is expected to lead to shifting of the BOP natural
frequency away from the wave or current excitation zone, hence reducing the wellhead system fatigue damage. Initial model
calibration of the system is described below.
14 SPE-175441-MS

Figure 12 Predicted Frequency vs. Measured Frequency Response Comparison

Initial Calibration
Data from the monitoring campaign is used to improve the analytical model response by calibrating the key input parameters.
This process enables the removal of many uncertainties and evaluation of conservatism built into the initial model. Only after
the model calibration is achieved can transfer functions to convert measured motions into stress at the hotspots of concern be
relied upon to give a reasonable estimate of fatigue damage accumulation in each hotspot. Depending on the quality of the
input data, the assumptions and level of uncertainties built into the initial model and the initial offset between analytical and
measured responses, the calibration process may require several iterations.

Recent studies have explained how uncertainties in soil strength and modelling can contribute to differences between
predictions and measurements and the importance of matching the frequency response of the BOP in order to avoid undue
wave or VIV excitation (Howells et al. 2015). The initial calibration is focused on soil interpretation and model refinement
and the result is presented in Figure 14. The original soil model and non-linear springs have been generated using the API RP
2A-WSD (2000) recommended method as described in the section of soil profile data of this paper. Other soil modelling
techniques such as presented by Reese and Cox (1971) for stiff clays and Jeanjean (2009) suggest that API RP 2A-WSD
(2000) underestimates soil stiffness, hence the initial approach to soil calibration is to use an alternative soil modelling
technique as presented by Jeanjean (2009). The result of using the Jeanjean (2009) modelling technique in the case study
generates a soil stiffness profile that is more than 2 times stiffer than the API RP 2A-WSD (2000) stiffness profile, when the
same soil input parameters are used as demonstrated in Figure 13. The significant benefit of improving and stiffening the soil
profile is observed in the shift in peak frequency of the predicted model towards the measured system natural frequency and
away from the wave excitation zone.
SPE-175441-MS 15

Figure 13 Comparison of Soil Stiffness Profile Generated Using API RP 2A-WSD (2000) and Jeanjean (2009)
Approach

Figure 14 Predicted Frequency (Initial and Calibrated) vs. Measured Frequency Response Comparison

Further study is required to completely calibrate the global model to achieve a good agreement in both motions and
frequency response between the predicted and measured system by considering: top tension, mud weight, BOP stack
hydrodynamic properties, riser and BOP stack damping coefficients, vessel response and detailed soil study. Detailed model
calibration is expected to further improve the system response and predicted fatigue damage of the system as demonstrated
by the initial model calibration process.

Conclusions
The conclusions of the work presented in this study are:
Fatigue assessment of historical operations on ageing wells often requires a more detailed approach than is typically
employed for new-drill equipment due to the lower fatigue resistance of older equipment, compared to modern designs;
This enhanced detail can include consideration of measured or hindcast seastate data, including directionality, from the
actual periods of historical operations;
16 SPE-175441-MS

Application of wave spreading in analysis can improve the predicted fatigue life. However, the appropriateness of
applying this technique for individual geographical locations, and the appropriate spreading constants to be selected, may
be difficult to confirm;
The failure analysis (what-if approach) technique presented herein can be employed for the most critical wellhead fatigue
scenarios to understand the effects of an initial failure on the fatigue damage rates in the remaining hotspots of concern,
and the sequence of failure in those hotspots for planning and mitigation purposes;
In-service structural monitoring of the riser and wellhead system during operations is a recommended approach to allow
changes in the motion response during an operation to be identified, and to reduce uncertainties in the global model
through comparison of the analytical and measured system responses;
Depending on the level of uncertainty in the analytical model, accurate calculation of fatigue damage using measured
structural responses and a transfer function approach may only be valid following calibration of the analytical model
using those measured responses.

Acknowledgements
The authors acknowledge the owner of the data presented in this paper, Apache North Sea Ltd, and are grateful for their
permission to publish.

Many thanks to all personnel at 2H Offshore that worked on the case study described herein, including John Orsak, Helena
Carr, Paul Sykes, along with the contributions and assistance of Hugh Howells and Mike Campbell.

Nomenclature
BOP Blow-out Preventer
DOF Degree of Freedom
Fatigue life Time required for a flaw to grow to a size at which failure is deemed to have occurred
FEA Finite Element Analysis
FOF First Order Fatigue
Hotspot Area of concentrated stress within a component
HP Housing High Pressure Housing
LMRP Lower Marine Riser Package
LP Housing Low Pressure Housing
SCF Stress Concentration Factor
VIV Vortex-Induced Vibration
WH Wellhead

References
1. API RP 2A-WSD, Recommended Practice for Planning, Desiging and Construction of Fixed Offshore Platforms
Working Stress Design. 2000. Washington, D.C: API. Errata and Supplement 1, 2002.
2. BS 7608, Fatigue Design and Assessment of Steel Structures. 1993. Norwich: BS.
3. Costal Engineering Technical Note (CETN), Directional Wave Spectra Using Cosine-Squared and Cosine 2S Spreading
Functions. 1985. Vicksburg, Mississippi.
4. DNV-RP-C203, Fatigue Design of Offshore Steel Structures. 2012. Hvik, Norway: DNV.
5. DNV 2011-0063, Report for JIP Structural Well Integrity - Wellhead Fatigue Analysis Method. 2011. Hvik, Norway:
DNV.
6. Howells, H., Baker, R., and Rimmer, A. 2015. Measurement of Wellhead Fatigue. Presented at Offshore Technology
Conference, Houston, Texas, USA, 47 May.
7. Jeanjean, P. 2009. Re-Assessment of P-Y Curves for Soft Clays from Centrifugal Testing and Finite Element Modelling.
Presented at Offshore Technology Conference, Houston, Texas, USA, 4-7 May.
8. Padelopoulos, P., Ritchie, M., Tognarelli, M. et al. 2012. Drilling Riser Structural Damping Test. Presented at ISOPE
Conference, Rhodes, Greece, 17-22 June.

Vous aimerez peut-être aussi