0 evaluări0% au considerat acest document util (0 voturi)
35 vizualizări0 pagini
This paper presents the computational-geomechanics technology developed and implemented to assist drilling into the highly depleted and compacted crest from the new water-injection platform at Valhall. The technology is based on a history-matched full-field finite-element-based geomechanics model to calculate stresses, strains, and displacements. The results are exported from the model to a geological modeling software to perform wellbore-stability calculations using BP best practices.
This paper presents the computational-geomechanics technology developed and implemented to assist drilling into the highly depleted and compacted crest from the new water-injection platform at Valhall. The technology is based on a history-matched full-field finite-element-based geomechanics model to calculate stresses, strains, and displacements. The results are exported from the model to a geological modeling software to perform wellbore-stability calculations using BP best practices.
Drepturi de autor:
Attribution Non-Commercial (BY-NC)
Formate disponibile
Descărcați ca PDF, TXT sau citiți online pe Scribd
This paper presents the computational-geomechanics technology developed and implemented to assist drilling into the highly depleted and compacted crest from the new water-injection platform at Valhall. The technology is based on a history-matched full-field finite-element-based geomechanics model to calculate stresses, strains, and displacements. The results are exported from the model to a geological modeling software to perform wellbore-stability calculations using BP best practices.
Drepturi de autor:
Attribution Non-Commercial (BY-NC)
Formate disponibile
Descărcați ca PDF, TXT sau citiți online pe Scribd
Drilling Risk at Valhall Using Computational Geomechanics Tron Golder Kristiansen, SPE, and Roar Egil Flateb, BP Norge Copyright 2010 Society of Petroleum Engineers This paper (SPE 119509) was accepted for presentation at the SPE/IADC Drilling Conference and Exhibition, Amsterdam, The Netherlands, 1719 March 2009, and revised for publication. Original manuscript received for review 21 June 2009. Revised manuscript received for review 10 April 2010. Paper peer approved 26 April 2010. Summary The Valhall field is an overpressured, undersaturated Upper Creta- ceous chalk reservoir located in the Norwegian sector of the North Sea. The reservoirs consist of weak chalk. The weak chalk results in significant reservoir compaction, exceeding 10 m in places, and corresponding seafloor subsidence, exceeding 6 m above the center of the field. Previous papers have documented the increasing drill- ing challenges in the shale overburden at Valhall because of the subsidence. This paper presents the computational-geomechanics technology developed and implemented to assist drilling into the highly depleted and compacted crest from the new water-injection platform at Valhall. The work focuses on handling the stress changes occurring in the overburden above a compacting reservoir and not the more traditional situation associated with depletion-induced frac- ture-gradient changes in the reservoir itself. The technology is based on a history-matched full-field finite-element-based geomechanics model to calculate stresses, strains, and displacements. The results are exported from the finite-element model (FEM) to a geological modeling software to perform wellbore-stability calculations using BP best practices for prediction of wellbore stability. In this environ- ment, one can also easily use as supporting data 4D seismic from the permanent life-of-field seismic array at Valhall. By calculating the operational mud-weight window over a high risk interval in the overburden, a good correlation to historical nonproductive time (NPT) was found. The methodology is used in detailed well plan- ning where moving the well 50 m in one direction is the difference between problem-free drilling or huge drilling challenges. The paper presents the first application of this technology on a new water injector delivered 60 days ahead of schedule with reduced costs of approximately USD 20 million and with avoidance of a potential train wreck that could have cost USD 60 million. Introduction The Valhall field is an overpressured, undersaturated Upper Creta- ceous chalk reservoir located in the North Sea approximately 290 km offshore southern Norway in 69 m of water (Fig. 1). Valhall was discovered in 1975, and the field was originally developed to recover 250 million bbl of oil; but it has to date produced more than twice this amount, and work is ongoing to recover more than 1 billion bbl of oil from the Valhall structure. A three-platform complex with 24 slots was installed in 1981 on the crestal part of the field. The 24 slots were extended to 30 on the original drilling platform around 1990. In 1996, a new 19-slot wellhead platform was installed next to the existing central complex for infill and extended-reach drilling (ERD). Because of severe drilling problems during ERD, a third wellhead platform was installed in the south flank in 2002, and a similar wellhead platform was installed in the north flank in 2003. In 2004, a new drilling and injection platform was installed on the crestal part to accommodate waterflooding of the field. The reservoir is at a depth of approximately 2400 m subsea and consists of two oil-bearing chalk formations, Tor and Hod. Approxi- mately two-thirds of the oil and the majority of the productivity are located in the Tor. The effective overburden stress in the crest of the field was approximately 500 to 1,000 psi at discovery, owing to significant overpressure (0.82 psi/ft). The chalk is relatively weak (unconfined compressive strength between 500 to 2,000 psi), and because reservoir pressure has been reduced during primary deple- tion from approximately 6,500 psi to 2,500 psi in the crest, the chalk matrix has compacted more than 10 m in places, resulting in severe seafloor subsidence. For more details on the reservoir and field development, see Barkved et al. (2003). The current seafloor subsidence is approximately 6 m at a rate of 14 cm/yr after a rela- tive constant rate around 25 cm/yr for most of the production his- tory. The compaction and subsidence have also resulted in casing deformations, as observed in other compacting and subsiding fields. For more details on casing deformations and design, see Kristiansen et al. (2000) and Pattillo and Kristiansen (2002). Drilling in the overburden at Valhall has been more challenging over time because of varying degrees of subsidence in the field. In the late 1990s, ERD wells to the south and the north of the field were unsuccessful, and an evaluation of the situation indicated that a more cost-effective solution would be to drill the flank wells from platforms in the south and the north of the field. Drilling from these platforms resulted in sustained best-in-class drilling performance (Kristiansen 2004). While the ERD program could be exchanged with the flank platforms in the north and south part of the field, the off-take in the crest continued and the crestal part of the Tor formation was depleted from initial pressure of 6,500 psi to below 2,000 psi in places; this is well below the water gradient. This depletion results in the maximum reservoir compaction and, hence, the largest risk for subsidence-induced changes in the overburden before entering the much-depleted reservoir. Drilling experience in the crest at Valhall during the last few years has indicated considerably varying drilling conditions over fairly short lateral distances. It was decided to develop a methodology to reduce the risk of wellbore instabil- ity before landing wells in the depleted crestal area of the field. This paper presents the computational-geomechanics technology developed and implemented to drill into the highly depleted and compacted crest from the new water-injection platform at Valhall. The technology is based on a history-matched full-field finite- element-based geomechanics model to calculate stresses, strains, and displacements. The results are exported from the FEM to a geological modeling software in order to perform wellbore stabil- ity calculations using BP best practices for prediction of wellbore stability. A Simple Analytical Model for Compaction, Subsidence, and 4D Seismic Before we start on the field case, it is worthwhile to look into the process of compaction and subsidence using a simple analytical model. The model for compaction and subsidence is based on work by Geertsma (1973). The rock-physics model for the shales is based on work by Holt and Fjr (2003) and Bakk et al. (2006). The mech- 2seis software we use was first presented by Barkved et al. (2005) and recently, in more detail, by Fjr and Kristiansen (2009). Fig. 2a shows the vertical displacement over a compacting elliptical disk-shaped reservoir. The model is three dimensional. One can see that at the top of reservoir, this corresponds to 3.4 m; December 2010 SPE Drilling & Completion 545 it will later be shown that the compaction in the Valhall reservoir is more than 10 m in some locations. The corresponding change in seismic two-way travel time (TWT) (in milliseconds) is shown in Fig. 2b, and the associated change in shear stress is shown in Fig. 2c. One can see from Fig. 2c that the change in shear stress occurs fairly close to the reservoir and is more pronounced at the transition between the compacting part of the reservoir and the noncompacting part. Fig. 2c also indicates three potential well Fig. 1The Valhall field is located 290 km off the cost of Norway in 69-m water depth in the Norwegian part of the North Sea. Fig. 2Output from the semianalytical mech2seis (a) shows vertical displacement above a reservoir compacting 3.4 m at 2400- m depth; (b) estimates two-way-time shift on the basis of a shale rock-physics model; (c) shows the shear-stress distribution around the compacting reservoir with three well locations, A, B, and C; (d) shows the stress changes at Well Location A; (e) shows the stress changes at Well Location B; and (f) shows the stress changes at Well Location C. The stresses shown are the vertical and two horizontal (x- and y-direction). Also note that the compaction of the reservoir increases the velocity in the reservoir caused by reduction of porosity. 546 December 2010 SPE Drilling & Completion locations named A, B, and C. Location A is a well location going into the compacting part of the reservoir, Location B is toward the edge of the compacting part of the reservoir, and Location C is in the noncompacting part of the reservoir. Figs. 2d, 2e, and 2f show change in the three principal effective stresses along vertical sec- tions in the three locations. As one can see, the changes in stresses are quite dependent on the location relative to the compacting part of the reservoir. Fig. 3 illustrates this change in stresses by showing results of a wellbore-stability analysis in the caprock, just before entering into the reservoir rock in Locations A, B, and C. The results are pre- sented as minimum mud weight required to avoid causing break- outs while drilling. The plots represent wells drilled in any azimuth and with any inclination at the three locations. The three figures in the top row of Fig. 3 show the initial conditions before compac- tion and subsidence. They are equal and show a normal-faulting stress regime, where the largest principal stress is vertical. In this situation, the mud-weight requirement increases with increasing wellbore inclination. A vertical well will plot in the center of the plot, while a horizontal well will plot at the outer circle and require the highest mud weight. The difference between the two horizontal stresses is insignificant (essentially zero), but the direction of the largest horizontal stress is N120E, as illustrated with the black arrows. Because the difference between the horizontal stresses is zero, the mud-weight requirement should be a function only of inclination and not azimuth. In the lower row of figures in Fig. 3, one can see that the stress situation, and hence the mud-weight requirements, is com- pletely changed. In Locations A and B, the stress regime suddenly becomes one of reverse faulting, with the vertical principal stress being the least. In Location A, the largest-horizontal-stress direc- tion is rotated. In Location C, the maximum-horizontal-stress direction is rotated; although the stress regime is still normal faulting, the difference between the two horizontal stresses has increased. The vertical stress has increased. This is because the noncompacting reservoir below is acting as a stress pillar (i.e., taking up some of the vertical load transferred from the overburden in the compacting reservoir area). Some of the predicted stress changes in this simple analytical model in response to 3.4-m vertical displacement (compaction) at the top of the reservoir will be lower in reality. This is mainly because of fault reactivation (i.e., shear displacement of favorably oriented faults). In reality, the reservoir geometries are not as simple as in this example; therefore, one tends to perform full-field model- ing using an FEM, which is very effective in dealing with complex geometries, elastoplastic deformations, and fault reactivation. The simple example here illustrates the mechanisms one can expect to encounter in the compacting field (i.e., introducing significant local stress variations). In the field example that we present in this paper, the well will be drilled from a central platform over the most-com- pacting subbasin in the Valhall field, similar to Location A. Then, the well will be drilled toward the edge of this basin (similar to Location B) and over a noncompacting reservoir area (similar to Location C) and then will enter a new compacting subbasin (similar to Location B) before continuing in the reservoir. Full-Field FEM For a complex compaction and subsidence case such as Valhall, the best approach is to construct a full-field geomechanical computa- tional model based on the FEM. The starting point in developing a numerical model for a compacting and subsiding field such as Valhall is an adequate characterization of the rock formations in the reservoir and the overburden; see Kristiansen (1998). The mod- eling approach used here is the one by Kristiansen and Plischke (2010) and is similar to the one presented by Pattillo et al. (1998) for Valhall earlier. The model is built in special purpose-made software developed to handle various geomechanical problems in the oil and gas industry. The main difference between our work and that of Pattillo et al. (1998) is that we use a more advanced elastoplastic constitutive model for calculating the reservoir com- paction. This new model includes the effect of water weakening, repressurization, and creep. In addition, our model also includes major faults in the reservoir and the overburden. In addition, the increased computer power has been used to model the field with the same element size as the reservoir flow model, which is down to 50 50 m in the crestal area. A grid of the full-field model is Fig. 3Wellbore-stability calculations in terms of minimum required mud weight to avoid wellbore collapse for the three differ- ent well locations in Fig. 2. The top row shows the initial conditions for all well locations, and the lower row shows the results after the compaction and subsidence that are shown in Fig. 2a. In addition to changes in minimum mud-weight requirements, the normal stress state has changed to also include strike/slip and reverse faulting and the stress directions are rotated. December 2010 SPE Drilling & Completion 547 shown in Fig. 4. The model consists of 11 layers in the overburden, 14 layers in the reservoir, and 5 layers in the underburden. The model is then populated with rock deformation properties for the advanced elastoplastic constitutive model, the transverse isotro- pic overburden, and the isotropic elastic underburden. The underbur- den could have been treated as transverse isotropic, but because of insignificant impact on the results, the underburden was simplified. The sideburden is extended layers laterally from the main model, and these have the same constitutive behavior as the layers in the main model. The model is then initialized with the initial pore pressure and the initial stress state as boundary conditions. The next step is to predict the reservoir compaction at various locations in the reservoir with time. This is achieved by reducing the pore pressure from initial pore pressure according to the reservoir-flow-model prediction. The water weakening is determined by the change in water saturation from the initial water saturation. The water weakening results in more compaction at constant stress. If the repressurization in front of the water front is strong enough, the water weakening will not take place before a potential blowdown of the reservoir. Because we are mostly interested in the historical production time scale and the near future, we did use only a one-way coupling between the reservoir flow model and the FEM. This was assessed to be adequate because the flow model closely matches the observed field pressures in the history-matched time period. The reservoir-flow-model compaction tables were, however, based on the geomechanics-model stress/strain response. More-advanced methods have to be used for forward prediction of reservoir pressures and production and will provide a more straightforward history match in the future, as demonstrated by Pettersen and Kristiansen (2009). On the basis of the changes in pore pressure and water satura- tion as received from the reservoir flow model, the FEM calculates changes in stress and resulting strain and displacements. Examples of the compaction predicted at top Hod and Tor reservoirs and the subsidence at the seafloor in January 2008 are shown in Fig. 5. 645,631 nodes 176,460 elements 1,936,893 degrees of freedom 4200m Fig. 4Grid of the full-field FEM used to generate the input data to the workflow. The grid extends from the seafloor to 4200-m depth, with the top of reservoir located near 2400 m. The grid is also extended far from the field outline in the densely meshed region to avoid boundary effects. Fig. 5Compaction maps for the top of Tor and Hod reservoirs. Tor hence includes both Hod and Tor compaction. The maps are from the full-field FEM shown in Fig. 4. 548 December 2010 SPE Drilling & Completion A Geological-Modeling-Software Model At selected timesteps (yearly or biennially) ASCII files were written as x-, y-, z-coordinates and property values for import into a geological model for wellbore-stability analysis. The x, y, and z are the universal transverse mercator coordinates and depth, respectively, with attributes of stresses, strains, and displacements. An example grid is shown in Fig. 6. Fig. 7 shows the vertical displacement between production startup in 1982 and 2006. The vertical displacement at the top reservoir central in the crest exceeds 8 m, while at the seafloor it approaches 6 m. One can also note the local variations at the top reservoir level. Fig. 6Overview of the geological-modeling-software grid based on the full-field FEM grid. The reservoir model is covisualized inside the grid for reference. Fig. 7Example of property grid in a geological modeling software based on the full-field finite-element-based geomechanics model. The reservoir model is displayed with vertical sections showing the large-scale vertical displacement between production startup in 1982 and 2006. One can see that the local compaction exceeds 8 m in the crest, and at the seafloor below the central platforms, subsidence approaches 6 m. December 2010 SPE Drilling & Completion 549 Wellbore-Stability Analysis BP has developed several workflows within a geological modeling software, and one of them is the BP best practices for prediction of wellbore stability that includes a wellbore-stability analysis for any trajectory in the 3D volume. The wellbore-stability predictions are for designing wells and determining optimum mud weights. A typical result from the wellbore-stability calculation is shown in Fig. 8. This figure shows the recommended mud weight between wellbore collapse and formation fracturing (lost circulation) for a vertical well in the crest of the field. The trends are realistic and based on field experience (i.e., the intervals with narrow mud-weight window correspond to challenging intervals), but the absolute values seem to be a bit low when considering wellbore- stability and lost-circulation events in the field. This should be better calibrated in new full-field geomechanics models. Application of Workflow on Well G-14 The first well on which the new workflow was applied turned out to be a very challenging one, according to the model predictions. The G-14 well is a well going from the central water-injection platform to the northern basin to start the pressure support of the producers in that basin (Fig. 9). The well did not stand out as a high-risk well on paper. It had a sail angle of 57 that is normally acceptable for Valhall (Kristiansen 2004). The stresses and pore pressure from the full-field geomechanics-model prediction were transferred into the geological software for wellbore-stability analysis. When the wellbore-stability calculation was performed, the mud-weight window [defined by the difference between mini- mum mud weight, or equivalent circulating density (ECD), to avoid wellbore collapse and the maximum mud weight (or ECD) to avoid lost circulation] was very small compared to what was expected for this well profile. In fact, the wellbore-stability calculation pro- duced a nonexisting mud-weight window for a significant portion of the last hole section before entering the reservoir (Fig. 10). If one investigates the changes along the proposed trajectory, it was clear that the narrow window was a result of a stress concentration around a noncompacting ridge between the central and north basin. This noncompacting ridge resulted in increased shear stress above it, as illustrated in Fig. 11. One option would be to try to find a less stressed location along the ridge, but that was impractical with the given target box for the well. Another option was to reduce the local shear stresses around the wellbore by dropping the sail angle. This option was selected, although this would result in an increased sail angle in the preceding section. The window in this section is much wider, WELLBORE STABILITY ANALYSIS Fig. 8Typical wellbore-stability plot from a vertical well in the Valhall crest just below the platforms, extracted from the well- bore-stability calculation. Also note that the absolute values of the predictions are not accurate, but the trends are. Fig. 9Illustration of the G-14 trajectory (typical build and hold, see upper right) leaving one of the central platforms (injection platform) and going into the northern basin. Compaction in a deeper reservoir layer is used for illustration. 550 December 2010 SPE Drilling & Completion and drilling experience and the model indicated this to be a minor and acceptable risk increase. Another risk would be the increased length of the 13 3 8-in. section (2300 m) that would need to be run, close to the limit of the rig floor, but that was found to be a manageable risk also. The modified trajectory is shown in Fig. 11. The new mud- weight window is shown in Fig. 12. It should be pointed out, again, that the absolute values of the predicted mud-weight window are close to field-experience data, but they are not absolutely accurate. More work on future full field FEM models during the history- matching stage could potentially give more-accurate absolute val- ues, but for relative comparison between various trajectory options, these results are acceptable. Verifying Model Predictions Using Field Drilling Data Because this was the first well on which the new workflow was applied, it was felt that more-extensive verification of the model predictions was needed. To do so, we started to examine the his- torical drilling performance of wells at Valhall. We then had to use several of the timesteps used in the full-field FEM model of the Valhall field to perform wellbore-stability analysis for historical wells. These results take into account both the temporal and the spatial changes in stress state in the Valhall overburden in response to the large-scale reservoir compaction below. Because the absolute values may not be accurate, we calculated an average mud-weight window across the most critical part of the last hole section before entering the reservoir. When plotting this value for a number of historical wells drilled at Valhall, we could define three groups of wells: one green group with accept- able NPT related to wellbore instability, one yellow group with higher NPT, and one red group with unacceptable NPT including train wrecks. The results of these calculations are shown in Fig. 13. This was a very useful figure for the experienced drillers because it gave a clear verification of the predictions from the workflow. Another interesting feature with the workflow is that one can check the change in drilling risk with time for one specific well trajectory and target box. This is illustrated with Well A-4B, which is the offset well to G-14. It is clear that it would be best to drill this well early in field life and that later drilling will be progressively more challenging. The well was drilled in 1995 and was clearly in the yellow-to-red group, as can be seen in Fig. 14. WELL BORE STABILITY ANALYSIS 0 500 1000 1500 2000 2500 3000 8 10 12 14 16 18 Stress Logs (ppg) Z
T V D B R T
( m ) Overburden (Sv) Pore Pressure Fracture Gradient Minimum Required Mud Weight Collapse Mud weight MW Window: No Go WELLBORE STABILITY ANALYSIS Fig. 10The original wellbore-stability results based on the build-and-hold G-14 trajectory, indicating a nonexisting mud- weight window at a critical interval approaching the reservoir. The zig-zag curve pattern is because of the use of a nonoptimal grid resolution in the geologic modeling tool. Deviatoric stress Fig. 11Modified G-14 trajectory entering the shear stress pillar extending up into the overburden. The horizontal plane is the reservoir pressure at top of the reservoir simulator covisualized as a reference. The yellow lines are faults. December 2010 SPE Drilling & Completion 551 On the basis of the encouraging results, we also looked into the correlation between the average mud-weight window and the reported days drilling the section. The correlation was not very good. The reason for this is probably inconsistent decision making around each individual well incident (i.e., the same operational action was not taken at the same time for the same type of incident). This may be because of different interpretation and decision-mak- ing processes for the various drilling teams. For the same average mud-weight window, one chain of decisions within a certain time window may result in limited impact on NPT, while another chain of decisions may result in severe NPT. More work is needed to also decipher the impact of decision making and potential small differences in drilling practices on each well. Therefore, the tradi- tional drilling-engineering approach using offset drilling data in a probabilistic analysis may be sufficient, although impact of local stress changes is currently not included as an impact function. Verifying Model Predictions With 4D Seismic The worlds first permanent seismic array was installed over the Valhall field in 2003 (Barkved et al. 2003). The cables are used to acquire seismic surveys at regular intervals. To date, 12 surveys have been acquired since 2003. Comparing the two first 4D-seis- mic surveys at Valhall, the streamer surveys from 1992 and 2002, it was clear that the compaction was inducing a strong impact on the seismic. This impact was not only present at the reservoir level, but the deformations above the reservoir could be followed many hundreds of meters in the form of seismic two-way time shifts (Barkved et al. 2005). The 4D seismic could be used to verify the full-field FEM results in the area of the well, as shown in Fig. 15. Fig. 16 also shows synthetic time-shift data based on the full-field FEM using an advanced module of mech2seis, presented in an earlier section (Barkved et al. 2005) and in more detail recently by Fjr and Kris- tiansen (2009). In this study, the verification of the geomechanics model consisted in a qualitative comparison of the synthetic time shifts with the observed ones to make sure that the geomechanics model was consistent with the local area of the well as seen by the seismic. If the two did not agree, one would need to modify the geologic model, the reservoir flow model, or the geomechanics models until a better agreement could be achieved. WELLBORE STABILITY ANALYSIS 0 500 1000 1500 2000 2500 3000 8 10 12 14 16 18 Stress Logs (ppg) Z
T V D B R T
( m ) Overburden (Sv) Pore Pressure Fracture Gradient Minimum Required Mud Weight Collapse Mud weight MW Window: OK Fig. 12The wellbore-stability results based on the modified build- and-drop G-14 trajectory, indicating a wider mud-weight window at a critical interval approaching the reservoir compared to Fig. 9. -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 G - 1 6 ( P E 7 ) r e v 0 0 T S N G 1 ( I S 2 ) r e v 0 0 P K A - 4 B
( 2 0 0 6 ) G - 1 4 r e v 2 6 p k G - 1 9 G 1 ( P E 8 ) r e v 0 0 P K G - 1 1 G 1 ( I E 2 ) r e v 0 0 P K G 1 ( R F 1 2 ) r e v 0 0 T S N A - 4 B
( 1 9 9 6 ) F - 2 F - 4 A - 4
B G - 1 6 ( P W 3 4 ) r e v 0 6 P K G - 1 6 ( R A 2 2 ) r e v 0 0 T S N G - 1 3 A - 4 B
( 1 9 9 4 ) F - 1 8 A - 4 B
( 1 9 9 2 ) F - 1 5 A - 3 C F - 1 F - 1 6 F - 3 G - 2 4 G - 1 9 r e v 1 0 p k G 1 ( P E 8 B ) r e v 0 0 P K F - 3 A G 1 ( R A 1 1 ) r e v 0 0 P K G - 1 9 r e v 1 4 p k G - 1 6 ( P E 3
N ) r e v 0 0 P K G - 6 A - 2 2 A G 1 ( W 9 ) r e v 0 0 P K G 1 ( P E 4 ) r e v 0 0 P K G 1 ( R D P E ) r e v 0 0 H I B G - 1 9 r e v 1 1 p k G 1 ( I S 3 ) r e v 0 0 P K F - 1 9 G 1 ( R A 1 6 ) r e v 0 0 T S N G 1 ( W 1 1 ) r e v 0 0 P K G 1 ( P E 6 ) r e v 0 0 P K G 1 ( R F 1 3 ) r e v 0 0 P K G - 2 2 G 1 ( W 1 0 ) r e v 0 0 T S N G - 6
( P W 1 ) r e v 0 1 P K F - 6 G - 1 8 F - 1 0 G - 1 5 G - 6
( R F 6 ) r e v 1 0 P K G - 1 0 Well E C D
W i n d o w
( p p g ) Red Yellow Green I n c r e a s i n g N P T -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 G - 1 6 ( P E 7 ) r e v 0 0 T S N G 1 ( I S 2 ) r e v 0 0 P K A - 4 B
( 2 0 0 6 ) G - 1 4 r e v 2 6 p k G - 1 9 G 1 ( P E 8 ) r e v 0 0 P K G - 1 1 G 1 ( I E 2 ) r e v 0 0 P K G 1 ( R F 1 2 ) r e v 0 0 T S N A - 4 B
( 1 9 9 6 ) F - 2 F - 4 A - 4
B G - 1 6 ( P W 3 4 ) r e v 0 6 P K G - 1 6 ( R A 2 2 ) r e v 0 0 T S N G - 1 3 A - 4 B
( 1 9 9 4 ) F - 1 8 A - 4 B
( 1 9 9 2 ) F - 1 5 A - 3 C F - 1 F - 1 6 F - 3 G - 2 4 G - 1 9 r e v 1 0 p k G 1 ( P E 8 B ) r e v 0 0 P K F - 3 A G 1 ( R A 1 1 ) r e v 0 0 P K G - 1 9 r e v 1 4 p k G - 1 6 ( P E 3
N ) r e v 0 0 P K G - 6 A - 2 2 A G 1 ( W 9 ) r e v 0 0 P K G 1 ( P E 4 ) r e v 0 0 P K G 1 ( R D P E ) r e v 0 0 H I B G - 1 9 r e v 1 1 p k G 1 ( I S 3 ) r e v 0 0 P K F - 1 9 G 1 ( R A 1 6 ) r e v 0 0 T S N G 1 ( W 1 1 ) r e v 0 0 P K G 1 ( P E 6 ) r e v 0 0 P K G 1 ( R F 1 3 ) r e v 0 0 P K G - 2 2 G 1 ( W 1 0 ) r e v 0 0 T S N G - 6
( P W 1 ) r e v 0 1 P K F - 6 G - 1 8 F - 1 0 G - 1 5 G - 6
( R F 6 ) r e v 1 0 P K G - 1 0 Well E C D
W i n d o w
( p p g ) Red Yellow Green I n c r e a s i n g N P T Fig. 13Calculated average mud-weight window (average minimum horizontal stressaverage collapse pressure) for a critical interval at Valhall for historical wells drilled at Valhall. The wells are divided into green, yellow, and red, where green indicates an acceptable operational mud-weight window, yellow is challenging, and red is nonacceptable mud-weight window. The coloring and well fits very well with experienced NPT. 552 December 2010 SPE Drilling & Completion -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 G - 1 1 A - 4 B
( 1 9 9 6 ) F - 2 F - 4 A - 4
B
( 1 9 9 5 ) G - 1 3 A - 4 B
( 1 9 9 4 ) F - 1 8 A - 4 B
( 1 9 9 2 ) F - 1 5 A - 3 C F - 1 F - 1 6 F - 3 G - 2 4 F - 3 A Well E C D
W i n d o w
( p p g ) Red Yellow Green Fig. 14A closer look at some of the wells in Fig. 13. This figure illustrates the changing drilling conditions in the subsiding overburden at Valhall with time. Well A-4B was the closest offset. One can clearly see that the trajectory becomes more and more risky to drill with time from 1992 to 1996. A-4B was drilled in 1995 and experienced problems and required a very high mud weight for wellbore stability. Results of Drilling the Well The well was spudded after the trajectory changes had been made and the well had been replanned. The drilling went very well in the shallower sections. In the high-risk 12-in. section, an abnor- mally high pore pressure was experienced. It was necessary then to weight up quite substantially over the planned mud weight. It is questionable whether this well could have been controlled with a mud weight lower than the fracture gradient if a higher sail angle had been used in this section. If a higher sail angle had been used, a higher mud weight would be needed to keep the well stable. The time-vs.-depth curve for the well is shown in Fig. 17. Here one can see that the well came in 60 days ahead of schedule, largely as a result of the integration of the new workflow in the well planning and execution. The shorter drilling time was estimated to have Fig. 15This shows the 4D static changes (indicating the change in two-way time shifts) in the seismic showing the changes across the ridge or the horst that Well G-14 had to go over and thus verifying the full-field finite-element-based geomechanics model-prediction. The two-way time shifts are reduced across the ridge (red) compared to a reduction above the basins (blue). Also compare with the synthetic results in Fig. 16, although using a reverse color scale. December 2010 SPE Drilling & Completion 553 reduced the cost of the well by approximately USD 20 million. It was also estimated that the potential-train-wreck avoidance had a value of USD 60 million. After this first application, the workflow has been applied on other wells with continued success. No wells have had the same estimated high risk as the G-14 well, however, when using this approach to wellbore stability. The workflow is improved continuously. The next history-matched full-field FEM will most likely be more accurate in terms of quantitative values for local wellbore-collapse pressure and fracture pressure because there will be a greater focus on this during the history matching. Another improvement that we are working on is the ability to also predict probabilistic well life, including risk of fault reactivation for each trajectory. The methodology will be based on the work by Fossum and Fredrich (2007). Another challenge will be to model the pore pressure more accurately, not so much in the shales, but in flaws, fractures, sand, and silt where free-migrating hydrocarbons dictate the local pressure. This will require a close integration with 4D seismic and rock physics (i.e., mech2seis and other seismic-technology developments). Conclusions A novel approach using computational geomechanics for predict- ing local stress changes in the overburden induced by reservoir compaction and subsidence has been presented. The workflow prediction was verified by qualitatively calculating drilling performance for a large population of historical wells drilled at Valhall using an average mud-weight window (ECD window) over the most critical interval before entering the res- ervoir. The calculations show good correspondence to the NPT experienced. The theory behind the approach has been demonstrated using a simple semianalytical model with integrated 4D seismic and geomechanics based on a shale rock-physics approach. The application potential for this technology is significant also in reservoirs with less compaction than Valhall, as in high-pressure/ high-temperature fields and other highly deforming reservoirs in various settings. The first application of the developed stress probing technol- ogy for wellbore stability on a Valhall well resulted in cost savings of approximately USD 20 million and an estimated train-wreck avoidance of approximately USD 60 million. In the near future, results will improve in terms of the quantitative aspects of stress changes and (we hope) pore-pressure prediction including the presence of free hydrocarbons in the overburden. The workflow will also be extended to provide a probabilistic well-life prediction for fault reactivation for individual trajecto- ries during well planning. Acknowledgments The authors wish to thank BP and the Valhall license partners, Hess, Shell, and Total, for permission to publish this work. The conclusions and views presented in this paper are the authors and do not necessarily reflect the view of the Valhall license. We also need to acknowledge the excellent work by Bertold Plischke, ISAMGEO GmbH on the full-field FEM. References Bakk, A., Fjr, E., and Holt, R. 2006. Simple Model for Lithological Anisotropy of Shale. Presented at the 68th EAGE Conference & Exhi- bition, Vienna, Austria, 1215 June.
3000 2500 2000 1500 1000 500 500 1000 1500 2000 2500 D e p t h
[ m ] Horizontal position [m] -10.0 -9.5 -9.0 -8.5 -8.0 -7.5 -7.0 -6.5 -6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 3000 2500 2000 1500 1000 500 500 1000 1500 2000 2500 D e p t h
[ m ] Horizontal position [m] -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 [m] [ms] Fig. 16This is mech2seis results from the G-14 area. To the left is a plot showing the vertical displacements from the full-field finite-element-based geomechanics model. The model shows the highly compacting central basin to the left in the figure, then the noncompacting to lower-compacting ridge, and then the start of the more-compacting northern basin to the right. The right- hand figure shows the corresponding synthetic two-way time shifts using a shale rock-physics model. VALHALL IP - WELL 2/8-G-14 T2 Time vs. Depth (Days) 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 0 50 100 150 200 Days D e p t h Orig. Trgt. Dur AFE Dur Act Dur Fig. 17Time-vs.-depth curve for G-14 (Act Dur) vs. technical limit (Orig. Trgt. Dur) and most-likely estimate (AFE Dur). 554 December 2010 SPE Drilling & Completion Barkved, O., Heavey, P., Kjelstadli, R., Kleppan, T., and Kristiansen, T.G. 2003. Valhall FieldStill on Plateau After 20 Years of Production. Paper SPE 83957 presented at Offshore Europe, Aberdeen, 25 Sep- tember. doi: 10.2118/83957-MS. Barkved, O.I., Kristiansen, T.G., Fjr, E. 2005. The 4D seismic response of a compacting reservoirExamples from the Valhall Field, Nor- way. Presented at the 75th Annual Meeting of SEG, Houston, 611 November. Fjr, E. and Kristiansen, T.G. 2009. An Integrated Geomechanics, Rock Physics and Seismic Model. Presented at the 71st EAGE Conference and Exhibition, Amsterdam, 811 June. Fossum, A.F. and Fredrich, J.T. 2007. Probabilistic analysis of borehole closure for through-salt well design. Acta Geotechnica 2 (1): 4151. doi:10.1007/s11440-007-0023-2. Geertsma, J. 1973. Land subsidence above compacting oil and gas reservoirs. J Pet Technol 25 (6): 734744. SPE-3730-PA. doi: 10.2118/3730-PA. Holt, R.M. and Fjr, E. 2003. Wave velocities in shale: A rock physics model. Proc., EAGE 65th Conference and Exhibition, Stavanger, 25 June, Vol. 1, 7996. Kristiansen, T.G. 1998. Geomechanical Characterization of the Overburden Above the Compacting Chalk Reservoir at Valhall. Paper SPE 47348 presented at SPE/ISRM Rock Mechanics in the Petroleum Industry (EUROCK 98), Trondheim, Norway, 810 July. doi: 10.2118/47348-MS. Kristiansen, T.G. 2004. Drilling Wellbore Stability in the Compacting and Subsiding Valhall Field. Paper SPE 87221 presented at the IADC/SPE Drilling Conference, Dallas, 24 March. doi: 10.2118/87221-MS. Kristiansen, T.G. and Plischke, B. 2010. History Matched Full Field Geo- mechanics Model of the Valhall Field Including Water Weakening and Re-Pressurisation. Paper SPE 131505 presented at the SPE EUROPEC/ EAGE Annual Conference and Exhibition, Barcelona, Spain, 1417 June. doi: 10.2118/131505-MS. Kristiansen, T.G., Barkved, O., and Pattillo, P.D. 2000. Use of Passive Seismic Monitoring in Well and Casing Design in the Compacting and Subsiding Valhall Field, North Sea. Paper SPE 65134 presented at the SPE European Petroleum Conference, Paris, 2425 October. doi: 10.2118/65134-MS. Patillo, P.D., Kristiansen, T.G., Sund, G.V., and Kjelstadli, R.M. 1998. Res- ervoir Compaction and Seafloor Subsidence at Valhall. Paper SPE 47274 presented at SPE/ISRM Rock Mechanics in Petroleum Engineering (Eurock 98), Trondheim, Norway, 810 July. doi: 10.2118/47274-MS. Pattillo, P.D. and Kristiansen, T.G. 2002. Analysis of Horizontal Cas- ing Integrity in the Valhall Field. Paper SPE 78204 presented at the SPE/ISRM Rock Mechanics Conference, Irving, Texas, USA, 2023 October. doi: 10.2118/78204-MS. Pettersen, . and Kristiansen, T.G. 2009. Improved Compaction Modeling in Reservoir Simulation and Coupled Rock MechanicsFlow Simula- tion, With Examples From the Valhall Field. SPE Res Eval & Eng 12 (2): 329340. SPE-113003-PA. doi: 10.2118/113003-PA. Tron Golder Kristiansen is BPs Subsurface and Wells Geomechanics Advisor. He has worked in several positions in BP, both within operating and technology groups, and he has also acted as Geomechanics Network Leader in BP. Kristiansens responsibilities include quality assurance support to BP assets and field studies across the globe, and his areas of expertise include a wide spectrum of geomechanics applications includ- ing wellbore stability, large scale geomechanics modeling, res- ervoir management, well management, completions, and spe- cial casing design for hostile subsurface conditions. He joined Amoco Norway in 1993 after graduating with a MS degree in petroleum engineering from University of Stavanger. Kristiansen has served on committees and acted as chairman for SPE con- ferences and forums. He was awarded SPE Young Engineer of the Year in 2000 in Norway. Roar Egil Flateb is a rock mechanics engineer with BP. His responsibilities include quality assurance sup- port to BP assets and field studies across the North Sea. Flatebs areas of expertise include a wide spectrum of geomechanics applications including wellbore stability, well management, and completions. He joined BP in 2003 after graduating with a PhD on chalk liquefaction from University of Stavanger.