Search

Vortex induced vibrations (VIV) may be developed in the sag-bend area of steel catenary risers (SCRs) and lazy wave risers (SLWRs) induced by the top motions. There is lack of reliable tools to predict VIV responses in such oscillatory flow conditions in the riser design. A time-domain VIV load model has been implemented in software VIVANA-TD. This load model adds a vortex shedding force term with a synchronization model to the conventional Morison load formulation. The synchronization model introduces a phase-coupling between the force and response, which effectively captures how the local vortex shedding frequency may increase or decrease to obtain lock-in. This time domain VIV load model makes it possible to account for time-varying flow speed. In the present study, the vortex shedding process and response characteristics of VIV in oscillatory flow conditions have been reviewed. The capability and limitations of the existing time domain VIV load model have been investigated. The empirical parameters have been derived from model test data. The validation study using the Equinor SCR and Deepstar SLWR test data shows that the predicted maximum fatigue damage is within a factor of 5 compared to the measurement for 85% of the cases. This demonstrates that the present VIVANA-TD can describe the main characteristic of VIV responses in the oscillatory flow conditions.

This paper presents results from a case study of vortex-induced motions (VIM) of a floating, spar type offshore wind turbine. VIM response can lead to increased motions of the wind turbine tower as well as contribute to fatigue of moorings and cables. Reliable predictions of VIM response are therefore needed for safe and cost-effective design. One option is to extend the use of analysis methods developed for vortex-induced vibrations (VIV) of slender structures. Three different analysis methods are compared in the present work: • Linear frequency domain analysis with coefficient-based VIV loads. • Nonlinear time domain simulation with a harmonic load found using a VIM design curve. • Nonlinear time domain simulation with the recently developed time domain VIV load model. The nonlinear time domain simulations can include the dynamic wave and wind loads together with the VIM loads. Nonlinear interaction effects can then be included, and fatigue damage can be calculated from the total response instead of combining damages calculated separately. Only the analysis with time domain VIV loads gives in-line VIM response. The results show that if in-line motions occur, they may give significant contributions to the mooring line tensions and fatigue. The effects of the higher Reynolds number and the low length to diameter ratio of floating spars must be investigated. Comparisons with full-scale measurements and adjustment of load coefficients are necessary before a time domain VIM load model is established.

An empirical time-domain (TD) vortex-induced vibration (VIV) prediction model has been implemented in a software called VIVANA-TD based on its earlier development by Thorsen at NTNU. It models the synchronization of VIV loads and structural responses with a set of empirical parameters generalized from model tests. Combining this time domain hydrodynamic load model with a non-linear finite element structural model makes it possible to account for structural non-linearities and time-varying flow. A joint industry project (JIP), i.e., Lazy Wave Riser JIP has been organized to improve the design basis for SLWRs. This JIP is executed by SINTEF Ocean with support from NTNU. The industry participants are Equinor, BP, Subsea7, Kongsberg Maritime and Aker Solutions. The overall objective of this JIP is to systematically validate VIVANA-TD, in order to establish it as an industrial tool for VIV prediction. It is also aimed to improve the empirical basis and methods for calculation of VIV of deep-water steel lazy wave risers (SLWRs). In the present paper, the validation study is presented for selected model tests in constant flow conditions with uniform and sheared profiles. The test model includes bare pipe, pipe with partial strake coverage and riser model with staggered buoyancy elements. The empirical parameters have been generalized based on extensive model test data. Limitations and improvement of the model have been also been explored. The results show that the present TD model can represent reasonably the VIV loads and that the prediction has good agreement with measurements in general.

Model tests of a top tensioned riser (TTR) model were carried out as a part of a joint industry project, with the purpose of better understanding the dynamic behaviour of drilling riser and verifying the calculations of the riser analysis tools. Sinusoidal motion in one direction was imposed at the top end of the riser model to simulate vessel motion. The tests were carried out in still water, accelerations and bending strains were measured along the riser model. Numerical simulations were performed using RIFLEX and the predicted global responses were compared with the model tests. This paper discusses interesting aspects of this comparison as well as the general dynamic behaviour of the top tensioned riser. It was found that the dynamic responses of a TTR with vessel motion can consist of not only the in-line (IL) responses due to vessel motion at the riser top end, but also cross-flow (CF) vortex-induced vibrations (VIV) under conditions when Keulegan-Carpenter ( K C ) number is relatively small. CF VIV response is estimated using a time domain VIV prediction model and compared to the measured response. The main conclusion is that the IL global dynamic responses and CF VIV responses are predicted sufficiently well.

A structured technology development process targeting to combine industry and Statoil’s experience has produced an engineering approach for wellhead fatigue analysis that is verified against measurements of load and load effects in actual subsea wells. This paper outlines Statoil’s wellhead fatigue analysis approach, which is based on the new industry standard for wellhead fatigue analyses, DNVGL-RP-E104, ref. [1]. Parts of the methodology has been presented in previous papers. The present paper provides a birds eye view, putting all the pieces together into one coherent methodology. The development and validation of an engineering approach for estimating the bending moment in the surface casing, between the wellhead housing and top of cement, will be presented in detail; this has previously been referred to as load sharing between wellhead and conductor. The wellhead fatigue analysis approach is based on a “coupled model”, which in this case means that the conductor with PY-soil springs are included in the model, compatible with industry recommendations [1], with the following main characteristics: • The lower boundary condition is modelled as a conductor in soil with a bending stiffness equivalent of the well system. • Soil and template interaction is modelled by discrete springs. • The global riser load analysis is run with long crested waves and head sea. Directionality of the waves are handled by reduction factors applied to the damage rate. Alternatively, directionality effects may be included by running multiple wave directions with short crested waves. • Fatigue capacity of the hotspots in the well system is represented by ΔM-N curves generated from detailed FE models. Typically, ΔM-N curves are established for connectors, welds between housings and casings, and for the wellhead housings. The paper includes validation against full scale measurements for a wellhead of preloaded type. In addition, it is demonstrated how the approach can be used for wellheads where the high-pressure housing may rotate inside the low-pressure housing. For this case, the validation is performed against a full 3D solid element model. The analysis approach presented is computationally effective and it will hence enable increased focus on sensitivity analyses. Analysis work is moved from time consuming local- and global analysis, to effective post-processing of data. Uncertainty in the input parameters has been found to significantly influence the fatigue estimate. Understanding these effects is considered vital for making conscious decisions on the fatigue life of a well. See e.g. [8], [10] and [20]. As pointed out already in 1985 by Valka et.al., ref. [5], and also by Milberger et. al., ref. [6], the cement level, and the relative motion of the two housings, represent large uncertainties. Macke et. al, ref. [10], showed that the additional uncertainty due to cement level and friction between housings exceeds the levels covered by the traditional fatigue safety factor of DFF = 10. A method is proposed to handle this in a consistent manner.

Over the last decades, the complexity and duration of offshore drilling operations have steadily increased. The size and weight of the risers and BOP stack has grown significantly. These factors have led to an increase in fatigue loads imposed on the wellhead structures during drilling and completion operations. Wellhead fatigue might ultimately lead to loss of well structural integrity and pressure containment and therefore safe and reliable drilling of subsea wellheads has gained high priority in the global oil and gas industry. This paper presents two of the most complex real time instrumentation campaigns for drilling operations. Analyses of a connected drilling riser system including the well structure are complex and involve several engineering disciplines. In addition, there are many unknowns going into the equations when accumulated fatigue damage of the wellhead is estimated. Therefore, assumptions need to be made, very often on the conservative side. A typical example are the global drilling riser analyses where the environmental conditions, actual rig motion and riser / BOP behavior are uncertain. With the duplex scope of accurately documenting the wellhead fatigue status during drilling operations and of achieving a better understanding of the actual risk level of wellhead fatigue, Statoil decided to start a very comprehensive monitoring campaign. Two MODU representing very different generations of rigs in terms of weights and types of equipment were instrumented from topside to BOP connector. Strain gauges were installed around the BOP connector as close as possible to the wellhead in order to capture wellhead response as accurately as possible. Due to the large number of sensors, high accuracy requirement and high sampling frequency of data to be shown live, a cabled solution was selected vs remote battery operated sensors transmitting via acoustic. Double set of cables, sensors and topside equipment were installed in order to make the instrumentation system fully redundant and suited for permanent installation. All data were additionally made available real time onshore to allow the full overview of the operation. To author’s knowledge, these two instrumentation systems are the most comprehensive and complex of this type installed on a drilling riser as of today. The first of the two system was installed approximately three years ago and it is still in operation. This paper describes the instrumentation systems installed and gives an extract of the quality data extracted and already used in already published studies [1, 2, 3].

The issue of wellhead fatigue has been given significant attention during recent years. The complexity of the wellhead system in terms of interactions betwe[en conductor (low pressure) and wellhead (high pressure) housings leads to various analysis methods being proposed to evaluate fatigue damage in the system. Most of the critical base material hotspots are located in the wellhead housing region where the loads start to distribute between the high- and low pressure housing and the load paths are highly complex varying for different input parameters. Due to this, detailed fatigue analyses are typically performed on a project to project basis for the same wellhead geometry. This paper proposes an approach that simplifies the analysis of the base material hotspots in the housings and makes it independent of where the specific type of the wellhead system is used. It is suggested to consider the housings of the wellhead system as one component with a single characteristic M-N curve, or a few M-N curves if complexity requires so. The M-N curve is a specialization of the standard S-N curves provided in rules and standards. They are generated by combining the calculated load-to-stress curve at a given hotspot with the applicable S-N curve. The load used as a reference is typically a cross-sectional moment at the top of high pressure housing. For these purposes 3D FE models have been developed for two principal wellhead types, rigid lock and non-rigid lock. The models are used to investigate the effect of different boundary conditions and applied loading on M-N curves for each hotspot analysed. Sensitivity studies have been performed for several parameters that are considered important in wellhead fatigue analysis. Based on the sensitivity studies, the effect of each parameter on typical base material hotspots is presented. This paper provides estimates for the spread in the M-N curves for each individual base material hotspot in the wellhead housings. Results indicate that a single characteristic M-N curve per wellhead system type can be selected to represent the wellhead housings. In addition, based on results from the analyses carried out, recommendations regarding generalized boundary conditions to obtain a characteristic M-N curve for a specific wellhead type have been given.

Fatigue of subsea wellhead systems due to wave-induced loads from riser and rig motions has been subjected to increased attention in recent years. It is expected that calculated riser loads are conservative, as both input parameters and methodology are associated with some uncertainty. However, it is difficult to quantify the degree of conservatism in analytical results unless reference to measurements can be made. Statoil has conducted drilling riser load instrumentation campaigns in several locations around the world over the last few years in order to gather high quality data for accurate assessment of the fatigue loads imposed on the subsea wellheads (see e.g. ref. /1/, /2/, /3/, and /17/). Four (4) of these measurements campaigns have been studied in more detail, with the intention to quantify the degree of conservatism to be expected from drilling riser analysis. Three (3) of these cases have direct bending moment measurements from strain sensors at the BOP connector elevation, giving high confidence in the results. For one (1) of the cases, the bending moments at the WH have been established by use of indirect methods (ref. /2/). The campaigns have been anonymized; They are from the Norwegian Continental Shelf (NCS), with water depths ranging from 110 to 400m. This paper presents the findings from our comparison of measurements and analytically derived drilling riser loads from these 4 campaigns. The analysis models have not been “tuned” to match the measurements. The goal is rather to get a measure of how well riser analyses are able to predict the real world. The conclusion in this paper is that the global drilling riser analyses accurately predict the cyclic loads on the subsea wellheads, provided that the input data are known with high degree of detail, including e.g. riser tension setting; drill pipe tension variation over time; and hydrodynamic loads. We found that scatter in the results is due to the uncertainty inherent to several of the input parameters. It is also shown that the accumulated fatigue damage from a full drilling campaign, can be established with sufficient degree of accuracy with somewhat lower requirement to the level of detail of the input, like e.g. using unidirectional waves instead of short crested waves. Directionality and spreading of the wave field can be handled by use of factors on the damage rate. A directionality factor is proposed, to enable comparison of directionality of the measured response to the predictions from global analysis.

In wellhead fatigue analysis, results are quite often sensitive to input parameters that are associated with considerable uncertainty. Such parameters may, e.g., be the cement level in the annulus between surface casing and conductor, the soil and template support, friction and down weight on the landing shoulder, etc. A common approach in deterministic design to deal with such problem is to use worst case assumptions on sensitive parameters. However, this may result in unreasonable short fatigue lifetime estimates. In this paper, it is shown how existing fatigue design equations using the concept of design fatigue factors can be augmented with characteristic design values for additional uncertain variables such that the overall safety level obtained is comparable to the recommended target levels for typical uncertainties in load effect calculations. Results from sensitivity analysis and SRA are compared to investigate to what extent sensitivity analysis results can be utilized as part of the design analysis. The intention of the proposed approach is to be reasonably conservative but to avoid being overly conservative.

Marine drilling riser is subject to complicated environmental loads which include top motions due to mobile offshore drilling unit (MODU), wave loads, and current loads. Cyclic dynamic loads will cause severe fatigue accumulation along the drilling riser system, especially at the subsea wellhead (WH). Statoil and BP have carried out a comprehensive model test program on drilling riser in MARINTEK's Towing Tank in February 2015. The objective is to validate and verify software predictions of drilling riser behavior under various environmental conditions by the use of model test data. Six drilling riser configurations were tested, including different components such as upper flex joint (UFJ), tensioner, marine riser, lower marine riser package (LMRP), blow-out preventer (BOP), lower flex joint (LFJ), buoyancy elements, and seabed boundary model. The drilling riser models were tested in different load conditions. Measurements were made of microbending strains and accelerations along the riser in both in-line (IL) and crossflow (CF) directions. Video recordings were made both above and under water. In this paper, the test setup and test program are presented. Comparisons of results between model test and RIFLEX simulation are presented on selected cases. Preliminary results show that the drilling riser model tests are able to capture the typical dynamic responses observed from field measurement, and the comparison between model test and RIFLEX simulation is promising.

Understanding well conductor–soil interaction mechanism plays an important role in well integrity assessment. Evaluation of field data obtained from monitored offshore wells can provide valuable insights into the matter. This paper presents a comparison of the numerical results obtained from a full three-dimensional (3D) finite element (FE) analyses model of the blowout preventer (BOP), wellhead (WH), conductor, and surface casing versus the field measured data obtained during drilling operations. Sensitivity studies were also performed for several parameters that were considered important in the local well response analysis under observed sea state conditions. The conductor–soil analyses were simulated using the Winkler spring p–y curves obtained by two different approaches: American Petroleum Institute (API) RP2GEO and a recently developed model by Zakeri et al. The results of bending moments in conductor and surface casing have indicated good agreement between FE model results and the field measurements.

Marine drilling riser is subject to complicated environmental loads which include top motions due to Mobile Offshore Drilling Unit (MODU), wave loads and current loads. Cyclic dynamic loads will cause severe fatigue accumulation along the drilling riser system, especially at the subsea well head (WH). Statoil and BP have carried out a comprehensive model test program on drilling riser in MARINTEK’s Towing Tank in February 2015. The objective is to validate and verify software predictions of drilling riser behaviour under various environmental conditions by use of model test data. Six drilling riser configurations were tested, including different components such as Upper Flex Joint (UFJ), tensioner, marine riser, Lower Marine Riser Package (LMRP), Blow-Out Preventer (BOP), Lower Flex Joint (LFJ), buoyancy elements and seabed boundary model. The drilling riser models were tested in different load conditions: 1. Forced top motion tests 2. Regular wave test 3. Combined regular wave and towing test 4. Irregular wave test 5. Combined irregular wave and towing test 6. Towing test (VIV) Measurements were made of micro bending strains and accelerations along the riser in both In-Line (IL) and Cross-Flow (CF) directions. Video recordings were made both above and under water. In this paper, the test set-up and test program are presented. Comparisons of results between model test and RIFLEX simulation are presented on selected cases. Preliminary results show that the drilling riser model tests are able to capture the typical dynamic responses observed from field measurement, and the comparison between model test and RIFLEX simulation is promising.

Enabling safe and reliable operations of subsea wellheads has a high priority in the global oil and gas industry. The objective of the current paper is to provide a novel method for bending moment estimates at the wellhead based on indirect moment measurements; this moment, together with fatigue properties are then used for fatigue damage estimation. Indirect bending moments are based on inclinations and accelerations measured by motion reference units (MRU) attached to blowout preventer (BOP), lower marine riser package (LMRP) and lower riser joint (LRJ) immediately above the lower flexible joint (LFJ). Also, required is the tension time history in the same period at the LRJ. The proposed methodology here can be implemented and integrated into a portal for data acquisition and visualisation. In order to validate the proposed method for indirect bending moment estimation, strain gages have been attached to a BOP and marine riser during drilling operations offshore Norway. Strain gage readings are transformed to bending moment which is used as reference (the so-called direct moment). The proposed method is used to calculate the moment indirectly, the so-called indirect moment. The resulting indirect moments agree very well with the direct moments.

Wellhead systems in the North Sea are frequently subjected to large bending loads, due to shallow waters, heavy seas and semi-submersible drilling rigs. This makes fatigue of wellhead systems a challenge. The fatigue estimates of such wells are often conservative when based on global FE riser simulations. To reduce the conservatism from such simulations, measurements of the actual wellhead loads have been performed on certain occasions. This is often a costly and intricate process that usually require a subsea cable from the rig. This paper presents an alternative method for measuring WH loads by the use of autonomous motion sensors (IMUs) on the BOP and lower end of the drilling riser. Measured inclinations from the IMU on the riser are used to capture the loads acting from the riser onto the BOP. The inclinations from the sensor on the BOP are used to capture the dynamic response of the BOP and well system. The combination of the measured excitation load and the measured dynamic behavior, makes it possible to reliably estimate the WH bending moment, without the use of riser simulations. The accuracy of the proposed methodology is demonstrated in a FE riser simulation. The model acts as a controlled environment where both input parameters and WH loads are known. The sensor data is extracted from the analysis at the relevant locations, run through the proposed methodology and then compared with the WH loads in the model. As a final verification, the method is tested on data from an actual measurement campaign where both IMU and strain gauge measurements are available. The WH loads will be calculated based on the proposed method and IMU data, and the results will be compared to the data from the strain gauge based load sensors.

The growing size of BOPs, longer drilling campaigns on wells, and operations in harsher environments has resulted in increased challenges in properly documenting wellhead fatigue during planned or executed drilling operations. The industry has started directing its efforts toward the calibration of analytical tools which are typically adopted for predicting wellhead fatigue. The ultimate goal for achieving this ambitious scope is to identify a benchmark set of analytical results that will predict field measurements. Early on Statoil identified a major obstacle: the absence of a good and comprehensive dataset of field measurements to serve as point of reference. Statoil and Aker Solutions cooperated on a pilot project with the intent of collecting a dataset of full scale measurements during drilling operations to be used to validate and calibrate the theoretical wellhead fatigue calculation methodologies. The main objective of the instrumentation campaign was to measure sectional forces as close as possible to typical wellhead hotspots by the use of three sets of strain gauges installed on the outside surface of the conductor and on the outside of the surface casing. With the objective of collecting an exhaustive dataset of measurements, accelerometers and inclinometers were installed on the BOP, the riser adapter, the riser below the upper flex joint and on the rig. An additional set of six strain gauges was installed on the riser to record riser tension variations. Environmental conditions were logged on board the rig and by the hindcast data provider. Operational events were carefully logged. This paper presents the following: • Data processing used for quality assurance and calibration of the measured data and the associated data challenges • Highlights of the instrumentation system capabilities to capture salient events of a typical drilling campaign and of ad-hoc performed rig operations to calibrate and validate the measured data • Effect of a controlled rig cross motion test, performed to evaluate quasi static loads on the well and calibrate strain gauge sensor orientations • A riser pull test, performed to validate strain gauge functioning • Several landing and disconnecting of the LMRP • Manipulation of the preload between the high pressure housing and the low pressure housing to investigate the effect of the preloading on the load sharing between the casings Since King and Soloman [2], the industry is still lacking quality field data to be used in order to validate the various analytical models used in the analyses of subsea conductor and wellheads. The results will confirm the quality of the measured data and will represent a first data point of comprehensive measured field data. This data will be used for future required work in calibrating the different building blocks pertaining to the analytical tools dedicated to well head fatigue predictions [3].

Large scale model tests have been conducted in a towing tank facility for the determination of the hydrodynamic coefficients of subsea blowout preventers. A subsea blowout preventer (BOP) is a large, complex device 10–15 [m] tall, weighing 200–450 [ton]. The BOP stack consists of two assemblies, the ‘lower marine riser package’ (LMRP) connected to the riser string and the BOP itself, connected to the wellhead. Together they represent a large lumped mass, which directly influences the natural frequencies and vibration modes of the riser system, particularly those of the BOP-wellhead-casing assembly. Large uncertainties in the estimates of the hydrodynamic coefficients (added mass, lift and drag or damping) result in large uncertainties in the fatigue damage predictions of the riser and wellhead system. The trend toward larger and heavier BOPs, which could place BOP-wellhead-casing oscillation frequencies in the range of wave frequencies, has motivated Statoil and BP to start a new research project on this subject. The project involves a large scale model test for experimental determination of hydrodynamic coefficients. Two different BOP designs were tested in a towing tank at model scale 1:12. The models weighed about 50 [kg] in air and were about 1.2–1.5 [m] tall. A six-degree-of-freedom oscillator was mounted under the carriage of the towing tank for oscillation of the models in different directions. Static tow tests and forced oscillation tests with and in the absence of steady current were carried out. Keulegan-Carpenter (KC) numbers ranged between 0.2 and 2.0, while the Sarpkaya frequency parameter β was in the range from 4,000 to 50,000. The Reynolds numbers of the static tow tests ranged between 50,000 and 150,000. This paper focuses particularly on tests in the surge direction with and in the absence of a steady current. Results indicate that the hydrodynamic coefficients for BOP stacks are quite different from those of simpler geometries like a circular cylinder. In addition, they provide new insight for analytical modeling of global hydrodynamic forces on BOPs in many configurations and scenarios.

Subsea wellhead systems are exposed to wave induced cyclic loading when a drilling unit connects to the well with a marine riser and a BOP. When connected, access is provided to the well and reservoir, and allows for operations such as further drilling, side tracking or workover. Once the operation is completed, the BOP is disconnected from the well, and the wellhead system is not exposed to cyclic loading any longer. Over the lifetime of a well, a number of such operations take place. A wellhead system is perhaps exposed to a total duration of fatigue loading of up to a year, which comprises a sequence of operations of different durations in different seasons. Fatigue predictions for offshore structures are typically based on statistical average of environmental conditions over a large number of years. This is appropriate for permanent installations exposed to continuous wave loading over the lifetime which is often 20 years or more, since variations in the environmental conditions from one year to another is equally represented in the statistics and experienced by the structure. However, for an operation of short duration, the uncertainty in the environmental conditions for that particular period in that particular year needs to be addressed. The weather during October this year is unlikely to be the same as in October last year, and can also be significantly different from average October weather. Although there exists no standard way of doing wellhead fatigue analysis, a commonly applied approach is to do the analysis in a single plane. This is obviously conservative since the wave direction will vary over time, and the fatigue loading will be distributed more around the circumference of the pipe sections in the wellhead system. Furthermore, the environmental conditions are typically based on statistical average for the month or season when the operation is to be executed, sometimes with some conservatism of including the adjacent more severe month or using annual data. Long crested waves are often assumed. This paper address the effect of the uncertainty in the environmental conditions on the accumulated fatigue damage for single and sequences of operations of different durations at different times of the year. A drilling rig in the North Sea has been analyzed using 56 years of hind cast data of significant wave height, peak wave period and main wave direction. Statistics of the fatigue damage rates are calculated and used in a structural reliability analysis in order to estimate reasonably but not overly conservative factors that are to be multiplied with the fatigue damage estimated in a conventional design analyses. Results based on long crested and short crested sea are calculated. An annual variation factor is proposed to account for the variability from one year to another. Secondly, a directional effect factor is proposed to account for the directional variations and its uncertainty on fatigue. Both factors are first estimated considering a single operation only, where the duration is varied between 3 days and up to a year. Thereafter, a sequence of operations of different durations at different times of the year is analyzed, and it is proposed how to consider the accumulated duration of such sequences compared to a single continuous operation. The expected result is an annual variation factor which is greater or equal to unity and a directional effect factor which is less than unity, both with lower values the longer the duration. The product of the two is a quantification of the degree of conservatism associated with a deterministic design analysis using long crested head sea and statistical average omnidirectional weather for the planned drilling operations.Copyright © 2015 by ASME

During the past few years, there has been a focus on increased oil recovery, which has increased the demand for drilling operations on subsea wells. Furthermore, the BOPs have become significantly larger in size. As a consequence, well head fatigue has been a growing concern for offshore drilling in the North Sea. In order to assess accumulated fatigue damage on a wellhead system, it is necessary to have some knowledge of the load history. This can be established e.g. by use of global riser analyses. However, this will include some model uncertainty, and is known to be conservative in many cases. Improved confidence can be achieved by measuring the actual loads on the system, thus removing the uncertainty inherent in a numerical analysis. It is challenging to measure strains on a wellhead directly, however in some cases it is possible to measure the strains close to the Wellhead, thus allowing for direct measurements of the bending moments and axial forces. Statoil has been able to measure strains close to the WH on 3 different drilling MODUs. In one case this was accomplished by using strain gauges directly on a spool piece below the BOP stack, in a second case by measuring the strains on the conductor casing and the surface casing below the housings, and in a third case in the form of flange face gap variations on the BOP connector flange. The direct method uses strain gauges or displacement measuring pins to assess bending strains. These systems are valuable in a sense that they provide an accurate measure of the bending moments. However, in some situations it is not possible to measure these loads directly, e.g. due to space limitations, or when the BOP support structure complicates the load path in the stack. Furthermore, it is difficult to measure bending moments on the wellhead when the BOP has landed on a horizontal XT for completion. In these cases indirect methods must be utilized instead. This paper presents a method for indirect measurements of the bending moments on a wellhead. The indirect method uses inclinometers on the lower riser adapter and BOP stack and allows for calculation of the bending moments in any cross-section below the lower flex joint. The benefit is that this automatically corrects for the height difference between measuring stations and the well head datum. Previous work on this topic has focused on horizontal displacement on the BOP stack, see e.g. [5]. Our results indicate that BOP displacement is not a good indicator of well head moments. However, a high correlation is observed between BOP stack inclination and well head moments. Statoil is currently preparing papers concerning the theoretical background of these methods [6]. Statoil is instrumenting and using both direct and indirect methods on several offshore drilling campaigns. Direct methods have been used to verify the indirect method, and the results show a good agreement. The results are used to monitor BOP stack behavior and well head fatigue, giving important input to the operational decision process. A measurement campaign on a semisubmersible drilling rig is used as an example case in this paper.

Fatigue of subsea wellhead systems due to wave-induced loads from riser and rig motions has been subjected to increased attention in recent years. Major accidents due to fatigue failure has not been experienced up to now, but with increasing size of drilling rigs and BOPs in combination with longer drilling campaigns on wells, proper documentation of adequate fatigue capacity for a planned operation is becoming increasingly important. Achieving this has turned out to be a challenging issue for some wells, where application of existing fatigue analyses methodology provides rather short fatigue lives. It is expected that such calculated fatigue life results are somewhat conservative due to conservatism in the assumptions commonly applied as both input parameters and methodology are associated with some uncertainty. However, it is difficult to quantify the degree of conservatism in such analytical results. One area of conservatism relates to wellhead fatigue analyses typically being performed under the assumption that the environmental actions and response is unidirectional (in one plane only) while assuming the most unfavorable direction. This is clearly not the case in real life, and one would like to take some benefit of distributing the loading more accurately around the circumference of the pipe section. For offshore structures subjected to continuous loading over many years, one may use directional scatter diagrams for wave height and wave period based on average over years of data. However, for wellhead fatigue, the duration of a drilling or well intervention operation may be relatively short, perhaps weeks or a few months, and it is more likely that there one prevailing environmental direction may dominate such short term exposures. The environmental conditions will vary from one year to another, i.e. the environmental conditions in one particular year may be less or more severe than the average represented by the scatter diagram. Both these variation effects are studied in the present paper by doing fatigue analysis for different durations; 3 days, 10 days, 1 month, 3 months, 6 months and one year. The same period is taken from each of the 56 years of environmental data, and individual results for that period within each year are calculated. Statistics quantifying the directional effects is determined. The directional effect is here defined as the ratio of the fatigue life where the directional information is taken into consideration and the fatigue life using an omnidirectional scatter diagram considering one direction only. Directional effect with reference to both head sea and beam sea are considered. Similarly, statistics of the fatigue life variation between years are obtained. Statistics are estimated separately for the six different durations. Finally, the statistics derived are applied in a structural reliability analysis (SRA). The SRA results are then used to propose reasonable factors that can be applied in conventional wellhead fatigue analysis to account for directional effects and for variation in environmental conditions from one year to another, depending on duration of the planned operation.

Subsea wellhead mechanical fatigue can potentially result in a gross structural failure of barrier elements in the upper part of the well, potentially resulting in loss of well control. Several major E&P operators have acknowledged the importance of wellhead fatigue and are participating in the JIP “Structural Well Integrity”. It is within the scope of this JIP to develop a recommended practice for wellhead fatigue analysis methodology. The analysis methodology currently being investigated by the JIP is a decoupled approach, with modifications of the lower boundary to account for the stiffness of the conductor, soil and template interface. A detailed local wellhead model is used to generate the lower boundary condition for a decoupled global riser load analysis model. This lower boundary condition definition is intended to capture the overall non-linear stiffness of a site specific well in order to achieve best possible global riser loads estimate. In this article the effect of varying the lower boundary conditions on a global load estimate is studied. Global load estimates are generated from a typical North Sea case and various lower boundary conditions are introduced as the only change to the global riser model. A fixed lower boundary condition is used as a reference and load estimates generated from riser models with various lower boundary conditions are compared. The different lower boundary conditions selected for comparison in this study has been derived from the following cases: 1. Fixed at WH 2. As per ISO 13624-2 3. As per JIP “Structural Well Integrity” -Current 4. As per JIP “Structural Well Integrity” -Modified Comparing the analysis results gives indications that the lower boundary condition modelling approach affect global riser load estimate. The fixed lower end boundary conditions did not yielded the most conservative load history in a fatigue context. Modelling well specific flexibility at the riser lower end increased the total number of wellhead fatigue load cycles. This finding support the current approach suggested by the works of the JIP “Structural Well Integrity”. Ensuring that riser load results are still conservative places a higher importance on precise local modelling of the well system.