NO20150828A1 - Method for determining wellhead fatigue - Google Patents

Description

Introduction

The invention is within the field of determining fatigue of constructions, and more specifically to a method for determining possible fatigue damage on subsea wellhead systems due to being connected to risers.

Background

A lot of effort has been put into establishing fatigue status of subsea wellheads during the last decade. Time spent on detailed fatigue calculations has been reduced considerably, mainly due to more efficient processes for gathering input data and a more streamlined process for fatigue calculations. Despite this, a full detailed fatigue calculation for a subsea well may still take several weeks to complete, which may be challenging when planning future drilling operations on subsea wellheads.

The results of a fatigue analysis are sensitive to input data variations. It is thus not meaningful to perform a generic analysis. An analysis will be specific for each specific well and well operation.

A lot of effort has been put into making the methods for assessing fatigue more efficient. Since it is still a time consuming process to perform a detailed fatigue analysis, an operator handling and managing a plurality of existing subsea should single out and prioritize wells that might be exposed to fatigue, and then perform a detailed analysis on these.

It is thus a need for a screening tool that will enable an operator to single out and select subsea wells which should be prioritized for a detailed analysis.

A simplified methodology for comparing fatigue loading on subsea wellheads has been performed and is described "A simplified methodology for comparing fatigue loading on subsea wellheads", presented at International Conference on Ocean, Offshore and Arctic Engineering, OMAE2013-11529 by H.Holden, P.Bjønnes, M.Russo [ref. 1]. A simplified way of comparing the fatigue load on different subsea wells is presented. The simplest comparison is done by accumulating the number of days BOP has stayed connected to the wellhead. The wellhead fatigue load is however heavily dependent on the vessels used, water depth and weather while connected to the well. An equation for deriving a benchmark load factor for each operation phase for a subsea well is proposed. This benchmark load factor takes into account the water depth, metocean season of the operation, BOP height and weight, and the stiffness of the marine riser lower fl ex joint. This benchmark load factor will represent a standard number of days with a BOP connected, correcting for some known effects. The goal has been to define a measure of 'BOP days' that accounts for the water depth, operational season, and BOP particulars. A base case (one MODU, 100 m water depth, and all year operation), equating to one standard BOP day, has been chosen as the reference for all cases discussed.

The validity of the benchmark load equation will be shown through a comparison with 31 different global riser analyses intended for wellhead fatigue. For each of the 31 data sets, time domain load analysis is done for all sea states in the wave scatter diagram. The different analyses covers different rigs, water depths and two operational phases (with or without subsea XT installed). To enable a large scale comparison of the bench mark factor, an approach where the fatigue load is summarized using the bending moment standard deviation on the wellhead datum is presented. This methodology is then compared to four full fatigue calculations using a typical subsea wellhead fatigue capacity. Then the simplified fatigue calculation is performed for all 31 global riser analyses. The calculated damage is then compared with the corresponding bench mark formula in each case.

Finally it is shown how this benchmark load formula has been implemented into a database as a fatigue load criticality screening tool for the different Statoil subsea wells. The database is called Wellspot and is belonging to Statoil. It is further shown how this can be used as a tool during planning of future operations, and how to prioritize wells where a detailed fatigue analysis is recommended.

The method presented in ref. [1] did however only consider the load side of the fatigue and the method was unable to capture effects of different fatigue resistance for different wellhead systems. Further, a load equation was presented assuming a fixed slope of 3 for all cases for the fatigue S-N curve characterizing material performance. The load side equation thus overestimated the effect of water depth and environmental season for wells with a non-welded hot spot (HS) as the most critical hot spot.

The Wellspot database system is developed by 4Subsea to support wellhead fatigue calculation process within Statoil. The purpose of the database system is to store analysis input and results to increase the efficiency of the wellhead fatigue calculations performed. The work was presented at OMAE 2013 and was implemented into Wellspot, where fatigue criticality for each well for specified operations is calculated. The sum of this is then used to evaluate and prioritize wellhead fatigue analyses activities. After close to two years of experience with Wellspot a couple of challenges have been discovered. First of all, the implemented equation includes a factor for the seasonal variations of weather. This factor was established from the monthly wave scatter diagrams. The actual wave conditions during a single operation might deviate significantly from this. This has been further documented in "Wellhead fatigue, effect of uncertainty in the directional variation and environmental conditions for operations of short duration" presented at OMAE2014-24482 by T.Hørte, G.Grytøyr, M.Russo, M.Hofstad, L.Reinås. Comparisons between the calculated fatigue damage and the calculated fatigue criticality showed large deviations that could be related to the difference between the monthly wave scatter diagram and actual wave conditions.

Weather hind east data for the Norwegian continental shelf is stored in Wellspot. In addition to this, global riser analyses results are stored for all wellhead fatigue analyses performed. It is hence possible to include the actual weather in the estimate of the fatigue criticality in Wellspot. The process used for the estimate is illustrated in Figure 1.

The method according to the present invention also takes the resistance side of the equation into account. This resistance represents the well ability to resist dynamic riser loads. Introducing resistance to the screening method makes the results more accurate.

Short description

The invention comprises a method for determining possible fatigue in a wellhead system to be examined. The method is defined by choosing parameters contributing to load transitions through wellheads and establishing equations defining the relationship between said wellhead system and one or more pre-examined wellhead systems håving detailed load distribution results.

Establishing a parametric representation of a bending load distribution on the wellhead system to be examined by combining the existing load distribution results for said pre-examined wellhead systems with said parameters and equations.

Determining possible fatigue based on the resulting parametric representation of the bending load distribution of the wellhead system to be examined.

Other aspects of the method are defined in the claims.

Detailed Description

The invention will now be described in detail with reference to the figures.

Figure 1 shows the work flow in Wellspot for inclusion of observed weather; Figure 2 shows the relationship between Msumwtand Mstd,Sum ; Figure 3 illustrates the load paths for a wellhead; Figure 4 illustrates possible ways of transferring bending load from the wellhead to the ground; Figure 5 illustrates bending moment distribution along a conductor; Figure 6 illustrates portion of static applied bending moment through surface casing; Figure 7a, 7b show portion of static applied bending moment through conductor and surface casing; Figure 8a, 8b show the portion of static applied bending moment through conductor and surface casing for highest (red line) and lowest (blue line) load factor of variations in applied bending moment; Figure 9 shows a portion of applied bending moment through conductor for different wellhead system designs; Figure 10 shows a portion of applied bending moment through surface casing for different wellhead system designs; Figure 11 shows proposed shape of the Lf,Hsposition for the surface casing; Figure 12 shows proposed shape of the Lf,Hsposition for the conductor; Figure 13 shows linearity of the load to stress curves illustrated as the cumulative distribution of the r-squared value from a linear regression of the load to stress curve for loads between -750kNm to 750kNm wellhead bending load. Figure 14 shows response along the conductor collected from QA database; Figure 15 shows response along the surface casing collected from QA database.

The publication DNVGL-RP-0005, "Fatigue design of offshore steel structures", 2014-06 [ref. 2] provides a set of equations for calculating fatigue. By assuming that the main portion of the dynamic load in a wellhead system is bending load imposed on a wellhead from a marine riser BOP, the fatigue equations given in said publication can be rearranged for the purpose of wellhead fatigue analyses. The equation for fatigue calculations using the S-N method described in said publication and in DNV Reg No.: 2011-0063/ 12Q5071-26, "Wellhead Fatigue Analysis Method", Rev 01, 2011-01-18 [ref. 3] giving the number of cycles to failure for a given dynamic stress range is given in Eq. 1.

Where:

Nall- is the number of cycles to failure at the given load

log(a) - is a SN-curve constant

m - is the slope of the SN curve

Acr - is the dynamic stress range

t - is the wall thickness

tref- is reference wall thickness for pipelines and risers given as 25 mm k - is the thickness coefficient

For a pipe section under pure bending load, the fatigue damage due to this load could be rewritten as follows:

For thin walled pipes:

Where:

SCF - is the hot spot stress concentration factor

AM - is the bending moment range at the given hot spot y - is the distance from pipe center to the hot spot

/ - is the second moment of inertia of the pipe

D - is the pipe outer diameter

t - is the pipe wall thickness

Inserting Eqn. 3 into Eqn. 1 gives the following expression for the allowable number of cycles.

Where:

Damage - is the fatigue damage

N - is the number of cycles with the given load range Kfat, HS - is a fatigue factor given by the hot spot data only

Naii - is the allowable number of cycles to failure

log(a) - is a constant giving the position of the S-N curve [ref. 1]

m - is a constant giving the slope of the S-N curve [ref. 1]

t - is the wall thickness at the fatigue spot

tref- is the reference wall thickness, [ref. 1]

k - is the thickness coefficient, given by the selected S-N curve, [ref. 1]

The challenge with Eqn. 4 is that the bending moment range is different from hot spot to hot spot, depending on its elevation. A bending load factor defined in Eq. 6 may be introduced into Eq. 4.

Where:

Lf HSposition - is a load factor giving the portion of the bending moment load for the given location

AMHSPosition is the bending moment range at the hot spot location

AM^ - is the bending moment range applied on the wellhead connector

In Eq. 7 the load part and the resistance part of the fatigue damage is separated, where N- AM^ j represent the load side, while Kfatrepresents the hot spot resistance. The load factor Lf HSposition is an expression of the bending load distribution between the different load paths. This distribution will depend on several factors like: Hot spot elevation, Soil stiffness, Template stiffness, Wellhead/conductor housing support stiffness, Wellhead type, Casing down weight, Conductor and surface casing stiffness.

The value of this factor is further discussed later in this paper.

Eq. 7 in the previous section calculates the damage for one load range within a histogram. The total damage for an operation is the sum of the damage from all the cycles during the operation is presented in Eq. 8.

The expression N- AM^ represents the load side of the fatigue damage equation, and the expression is similar to the wellhead fatigue load benchmark (ref. [3]), where the fatigue load is presented as follows:

where:

BOPdstd- fatigue load criticality factor (standard BOP days connected)

«, - number of days connected for the actual operation, i Mfotj - Mfot, the total bending moment, for the actual operation, /', ref [3]

m - is the exponent in the SN curve

Mr - Constant, = 210.3Nm, ref. [3]

The load benchmark (BOPd8td) is a value that in WellSpot is calculated for each operation, while the ^ N- AM^ value needs a full time history of an analysis or measurement. For a one slope SN-curve, this expression can be replaced with the N • AM™^ where A<M>caris given by Eq. 10.

where:

AMcar- this is the characteristic bending moment range for one histogram. This is the moment range that with the total number of cycles in the histogram will give the same damage as the histogram itself for one slope SN curve N - number of cycles in the given histogram block

AMwjj - Mjot, the total bending moment, for the actual operation, i

m - is the exponent in the SN curve

The expression ^ N- AMj^ can then be substituted by the following expression:

where:

rid - is the number of days for the given operation Tz- is the characteristic zero crossing period

Ns - is the number of seconds per day

In ref. [1] the relationship between (BOPdsta) and the sum of the standard-deviation of the bending moment times the probability of occurrence (Mtd,sum) was established. It was shown that this was a good estimate of the fatigue loading on a wellhead system, and that the wellhead fatigue load criticality factor (BOPdstd) is an adequate estimator of the

.sum-

To investigate the relationship between M^umand AM^. a study using the marine riser response from one given riser analysis is performed. The following work was performed: - For each HS level in the yearly wave scatter diagram, one load histogram representing one day of operation is established. The weighting of the different wave peak periods is done according to the probability of occurrence given by the yearly scatter.

The weighted average of N ■ AM^ is then calculated for each histogram.

This work was done for m = 3, 4 and 5, and the result is summarized in Figure 2. The results of the study shows that the relationship between AM™rand M^sumis linear with a slope of 2.6 - 3.2 depending on the slope of the SN curve. As Mtd,sumis directly related to MTot- Mr (ref [1]), the simplified equation for fatigue damage then becomes as follows:

where:

D - fatigue damage for one operation

n - number of days connected for the actual operation

Kioad - constant related to the inclination of the MN curve. Kload= 0.3 • m +1.7

Mfot- fatigue load criticality factor (standard BOP days connected, from ref. [1])

Mr - constant = 210.3kNm

Figure 3 illustrates the load paths for a wellhead. The load driving the fatigue damage accumulation in a wellhead system is the bending load. The effect of tension variations on the fatigue damage is limited. Structurally a wellhead system is designed to transfer the loads from the BOP and riser down into the soil. Figure 4 illustrates possible ways of transferring bending load from the wellhead to the ground. A bending load on top of a wellhead system could go into the ground through one of the three load paths illustrated in the figure. The first path is: Wellhead - Conductor - Tailpipe - Template - Soil; the second path is; Wellhead - Conductor - Soil, and the third path is Wellhead - Cement - Conductor - Soil.

From a fatigue perspective it is normally preferred to move the dynamic loads away from the wellhead and its extensions, and try to take as much load as possible in the tailpipe and template structure.

A wellhead system is a complex structure consisting of a range of components with ability to carry bending load. The system might be considered as a set of springs connected together into different load paths as presented in Figure 4. The load will be distributed between these load paths based on the stiffness (ability to carry bending load) of each path. The stiffer the sum of the stiffness's along the load path is the more of the dynamic load is carried through that load path. It should be noted that one of the most important points of this transition is the interface between the conductor and wellhead or the conductor and tailpipe. When designing a wellhead system it is important to look at it as one system, i.e. chain of "springs" in each load path. It is for instance no use in håving a stiff template capable of carrying a large load if the interface between the conductor and the tailpipe is incapable of transferring the load.

The mechanical behaviour of the conductor will depend on the support in the soil and the template. The bending moment distribution through the conductor is summarized in Figure 5.

The applied bending moment at the top will decay down through the soil. The slope of the bending moment decay along the conductor for all the wells will depend on the stiffness profile of the soil springs. Looking at the moment diagram from BC 1-3, the major difference in the moment distribution when introducing the template support, is the steepness of the slope of the bending moment reduction. The stiffer the support, the steeper slope of the bending moment decay. If a rotational support is added (BC 4-5), a portion of the bending moment will be transferred to the template, but the overall behaviour of the moment decay is similar.

The mechanics in the surface casing is much simpler than for the conductor. The bending moment distribution will heavily depend on the cement shortfall between the conductor and the surface casing. In the area with full cement support between the surface casing and the conductor, both pipes will behave like one composite pipe. In the region without cement support the surface casing will experience a bending moment at the top. Most of this bending moment is transferred out to the conductor at the conductor/surface casing hang off interface. As the conductor is externally loaded by the soil springs, this casing will be more restricted against bending than the surface casing. This restriction will have to be applied to the surface casing through the cement. The area of the surface casing without cement will hence be a beam with an applied bending moment at the top and an opposite bending moment at top of the cement.

Figure 6 illustrates the bending moment in a typical surface casing for different cement shortfalls. The curves starts at the wellhead datum (3 m above mud line), where M=MWh, and the load portion is hence 1.0. Then the load increases due to the shear force on the wellhead to approximately 1.2 at the conductor/wellhead interface. The amount of load transferred into the surface casing will depend on the cement shortfall. The shorter cement shortfall, the more load in the surface casing. Then the inclination of the moment curve will be constant down to the cement termination where an opposite bending moment will apply. If the number of possible cement shortfalls is increased to infinite, a continuous line will give the load in the surface casing as function of cement shortfall.

The method developed according to the present invention is to be used for screening of fatigue criticality of wells. A trade-off between accuracy and simplicity will have to be made. The main driver with respect to accuracy is the load transition between the different components in a wellhead system. The methodology is to develop a

simplified estimate for the expression Lf,Hsposition described as Eqn. 6 above.

The work is based on numerous local finite element analyses prepared for a project where the target was to establish the fatigue status of an entire field development consisting of approximately 70 subsea wells. This field contained several different wellhead system designs, including both rigid lock and traditional non-rigid lock systems.

To establish the Lf,hsposition a 3D FE model of a non-rigid lock wellhead system is used as a base case. The FE model is built according to ref. [3] based on original drawings from the system vendor, template stiffness properties by an independent vendor of template analyses, soil properties from Wellspot and analysed with 15 different cement levels. The template is modelled with 3D solid elements in the interface towards the conductor housing while the template stiffness is represented by a spring. The model is loaded with a unit shear force at the top of the BOP. From the 3D FE model, bending moments is normalized with applied load as a function of cement level and elevation studied. This is the basis for the Lf;HSposition-The actual cement level between the conductor and surface casing is always uncertain; 15 different cement levels are analysed, and the highest load factor with respect to cement level at each elevation are extracted. The cement levels are located between the cement outlet and 40 meters below mud line. This reduce the number of variables in Lf;HSposition to elevation and applied bending moment and give the surface plots of the load portions as presented in Figure 7a and 7b showing portion of static applied bending moment through conductor, fig. 7a and surface casing, fig. 7b.

The surface plots show a non-linear behaviour of the load portion with respect to elevation. Figure 8a, 8b show the portion of static applied bending moment through conductor, fig. 8a, and surface casing, fig. 8b, for highest (red line) and lowest (blue line) load factor of variations in applied bending moment at each elevation. The blue line represents load factor for a typical bending moment of 500kNm. Variations in applied bending moment have negligible effect in the load transition between the wellhead components. The resulting load portion represents LfjHsPosition.

In addition to variations of distance to wellhead datum and applied bending moment three different wellhead system designs have been studied. The three designs are Non Rigid Lock (type I), which is base case, Non Rigid Lock (type II) and a Rigid Lock design. The main difference between the Non Rigid Lock designs is the length of the conductor housing. The load portions are presented in Figure 9 and Figure 10. The load transition in the rigid lock system has a slightly different behaviour in the areas of the locking between the components, but overall the differences are not severe.

All the analyses presented above were performed for one particular field on the Norwegian continental shelf (NCS). For this development, the soil stiffness/template support was similar for most of the wells. Similar curves of LfHSposition could in principle be developed for all different well type and location. The overall mechanics is however similar for most subsea wells. It was hence decided to test the methodology for this development on other subsea developments on the NCS by selecting some parameter variations. It was found that the most important parameter was the rotational support of the template. Based on this it was decided to include different rotational support from the template to further increase the accuracy.

A 4Subsea QA database, described below, was used to verify that the results are valid for other developments as well.

It should be noted that the methodology implemented in the method according to the invention is meant for screening purposes. A trade-off between simplicity and accuracy will therefore have to be assessed.

As described earlier, the template rotational support is important with respect to the load transition through a wellhead system. As the purpose of the method presented is to perform simple screening before a full fatigue analysis is done, a simple approach where the template support is divided into 3 different classes was selected. The first class is: No template support, i.e. satellite wells with support from the soil only. The second class is: Low template support, i.e. for templates with no or limited rotational support. The last class is high template support, i.e. template interfaces where the conductor is firmly hung off in a steep angled shoulder able to transfer the bending moment from the conductor into the template.

Plots of the selected Lf,hsposition for the different classes is presented in Figure 11 and Figure 12.

A plurality of wellhead fatigue analyses has been performed over the years. Analysis inputs and results from these analyses have been stored in an internal 4Subsea database used for QA purposes. The results from the local analyses of a wellhead system are stored in terms of load-to-stress curves mapping the local stress at each hot spot to the bending moment at the wellhead datum. These load-to-stress curves, one for each fatigue hot spot, is then given for a wide range of input parameters such as cement shortfall, operational phase and soil/template support stiffness. One analysis will typically contain 500 - 5000 load-to-stress curves depending on the number of parameter variations. The 4Subsea QA database contains close to 80000 load-to-stress curves from different analysis vendors, a variation of wellhead system vendors and different wellhead system designs. These data have been used to verify the proposed Lf,Hsposition and the proposed fatigue equation.

The first major assumption for the proposed equation is that Lf;HSpositionis linear with respect to the amount of load put on top of the wellhead. This is off course a simplification, as it is well known that e.g. stick slip behavior of the surface casing and non-linear soil support is giving non-linear response. However, the main portion of the load to stress curves is fairly linear in the typical load range driving the wellhead fatigue damage. The cumulative distribution of the r-squared value from a linear fit of all load-to-stress curves in the 4Subsea QA database is presented in Figure 13. The curve fit is done for bending loads in between +/- 750 kNm, corresponding to a bending load range of 1500kNm. The dashed lines only include the load-to-stress curves with a maximum stress of 1 .OMPa within the load range (only including the load-to-stress curves giving actual fatigue damage).

To verify the proposed equation for Lf;HSposition, all the load-to-stress curves in the QA database have been converted into bending moment at the given hot spot. For similar load cases with variations on cement shortfall, only the most critical shortfall was shown in the results. This is plotted for both the conductor and the surface casing together with the proposed shape of Lf,Hsposition in Figure 14 and Figure 15 for the conductor and the surface casing respectively.

The present invention establishes an efficient method for estimating possible fatigue in a wellhead system to be examined.

The method is defined by choosing parameters contributing to load transition through wellheads. Different parameters may be chosen, e.g. one or more of parameters defining rotational support of the template, cement shortfall, operational phase, soil/template support stiffness.

Based on this the next step is establishing equations defining the relationship between said wellhead system to be examined and one or more pre-examined wellhead systems håving detailed load distribution results as described in detail above. These relationships and equations were described in detail above.

The next step is establishing a parametric representation of a bending load distribution on the wellhead system to be examined by combining the existing load distribution results from 3D analysis for said pre-examined wellhead systems with said parameters and equations.

In one embodiment the bending load distributions are represented by a parametric curve describing axial distributions of bending load as a function of vertical distance to the top of the wellhead in the wellhead system.

By establishing this parametric curve it is possible to evaluate bending load at any depth for all wellheads with same geometry and soil conditions but at different vertical locations of for instance a welded joint that are exposed to fatigue.

With standard methodology it is not possible to establish said parametric curve. Instead the bending load is tåken from a set location directly form 3D-modell.

If the wellhead system to be evaluated has a different soil/template condition than the wellhead system with the existing wellhead results, the parametric curve for the bending load can modified with parameters describing the effect of the different soil condition.

Possible fatigue is then determined based on the resulting parametric representation of the bending load distribution of the wellhead system to be examined. By calculating the load distribution, the problem solved is evaluation of fatigue damage.

The accuracy of the method is within what could be expected for a simplified approach. It should be noted that the methodology should not be used as the only documentation of fatigue damage for a field, but it is applicable for cost effective screening of large field developments, for selection of the correct wells for a full fatigue analysis.

If the inventive method reveals possible fatigue of a wellhead system, a full fatigue calculation should be performed.

The equations described above can also be used to scale already established fatigue results between similar wells with known differences. Improved accuracy of the methodology can be achieved if a more accurate estimator of Lf,hsposition, can be established. The proposed shape of Lf, Hs position can be considered as a typical behavior in the North Sea, but particularly for satellite wells this soil stiffness might affect the load along the conductor.

Evaluation of load distribution in wellheads is both time-consuming and resource demanding if performed with exiting tools and methods. By first screening and single out wellheads with possible fatigue the cost-efficiency and saving of time is obvious. It will also contribute to safety aspects since wellheads systems with fatigue damages may be identified at an early stage.

Claims (8)

1. A method for determining possible fatigue in a wellhead system to be examined,characterized in: - choosing parameters contributing to load transition through wellheads and establishing equations defining a relationship between said wellhead system and one or more pre-examined wellhead systems håving detailed load distribution results; - establishing a parametric representation of a bending load distribution on the wellhead system to be examined by combining the existing load distribution results for said pre-examined wellhead systems with said parameters and equations; determining possible fatigue based on the resulting parametric representation of the bending load distribution of the wellhead system to be examined.

2. The method according to claim 1, where parameters selected are one or more of rotational support of the template, cement shortfall, operational phase, soil/template support stiffness.

3. The method according to claim 1, where the bending load distributions are represented by a parametric curve describing axial distributions of bending load as a function of vertical distance to the top of the wellhead in the wellhead system.

4. The method according to claim 1, by introducing a simplified linear expression, Lf HSposition , for a load factor giving the portion of the bending moment on the wellhead according to:

where: AMHSPosition - is the bending moment range at a hot spot location; AMjpjj - is the bending moment range applied on the wellhead connector.

5. The method according to claim 4, by introducing an expression for fatigue damage for both the load part and the resistance part according to:

where: N ■ AM^ jj represents the load part; Kfatrepresents the hot spot, HS, resistance part

6. The method according to claim 4, by introducing a simplified expression for fatigue damage according to:

where: D - fatigue damage for one operation n - number of days connected for the actual operation Kioad- constant related to the inclination of the MN curve. ^=0.3-111 + 1.7 Mfot- fatigue load criticality factor Mr - constant = 210.3kNm

7. The method according to claim 2, where if the wellhead system to be evaluated has different soil/template stiffness than the wellhead system with the existing wellhead results, the parametric curve for the bending load is modified with parameters describing the effect of the different soil condition.

8. The method according to claim 1, further performing a full fatigue calculation if possible fatigue is concluded.