Você está na página 1de 13

IADC/SPE 59121 When Rock Mechanics Met Drilling: Effective Implementation of Real-Time Wellbore Stability Control

I.D.R. Bradford, SPE, Schlumberger Cambridge Research, W.A. Aldred, Schlumberger, J.M. Cook, SPE, Schlumberger Cambridge Research, E.F.M. Elewaut, Netherlands Institute of Applied Geoscience TNO, J.A. Fuller, Schlumberger Holditch Reservoir Technologies, T.G. Kristiansen, SPE, BP Amoco Norge and T.R. Walsgrove, Consultant

Copyright 2000, IADC/SPE Drilling Conference This paper was prepared for presentation at the 2000 IADC/SPE Drilling Conference held in New Orleans, Louisiana, 2325 February 2000. This paper was selected for presentation by an IADC/SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the IADC or SPE, their officers, or members. Papers presented at the IADC/SPE meetings are subject to publication review by Editorial Committees of the IADC and SPE. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract A new concept and process for real-time monitoring and control of wellbore stability establishes the drilling parameters required to optimize the drilling process and thereby reduce the potential for wellbore instability and subsequent unscheduled events or lost rig time. Surface and downhole measurements, recorded while drilling, are used to make regular updates to a model of the wellbore and to revise the drilling plan accordingly. The first step in the process is the generation of a mechanical earth model (MEM) using information obtained in offset wells and field and regional data. The proposed well trajectory for a new well is projected into the MEM and a set of stability parameters is generated for a given initial drilling plan. The product identifies potential danger zones within a well plan. During drilling, real-time data, including logging-whiledrilling (LWD), measurements-while-drilling (MWD), surface mechanical measurements, and fluids and solids monitoring information, are used to diagnose the state of the wellbore. Any significant hole instability is detected and a warning is given to the driller. The state of the wellbore is compared to the model, and any revision required to align the predicted with the actual state is made. This real-time update of the mechanical model is then used to predict the future state of the wellbore, in front of and behind the bit, for the given drilling plan. If the drilling plan can be improved, a revision will be recommended; for instance, reduction in the rate of penetration, increase in mud weight and circulation, and

change in hole direction. The drillers can independently evaluate their own recommendations for changes to the drilling plan and then decide on the best course of action. The process also provides a record of wellbore stability information that can be input to the field description for use in future wells and continuous improvement of the drilling process. Use of this concept was validated on the Valhall field in the Norwegian sector of the North Sea. Extended-reach drilling (ERD) to downflank targets has been problematic in recent years; there is a high risk that wells will be suspended or abandoned because of problems associated with wellbore instability in this very weak overburden. The Real-Time Wellbore Stability Control (RTWBSC) project team produced an MEM for the Valhall field, working closely with the drilling engineers to develop a well plan for a proposed ERD well. Implementation involved providing wellsite support to coordinate monitoring and detection of wellbore instability from real-time data, and on-line support in the drilling office to interpret data, update the MEM and revise the well plan. Through this process the team proposed and implemented a strategy of drilling the well in controlled states of failurenot a conventional drilling approach. The well successfully reached its target ahead of schedule and a planned string of intermediate casing was not required, mud losses (a previous problem contributing to instability and cost) were minimal and the well was cased to below the unstable overburden intervals. Introduction Wellbore instability is a major problem during the drilling of many oil and gas wells. Often quoted as costing the industry between 0.6 and 1 billion dollars per year,1 it currently leads to major difficulties in such diverse areas as the North Sea, Argentina, Nigeria and the Tarim basin.2,3 A recent, welldocumented spectacular example of the cost savings available from improved handling of wellbore instability is available for the Cusiana field operated by BP Amoco and partners in Colombia. Wellbore instability was very severe there, leading to costs per well of tens of millions of dollars. An integrated approach to the problem led to large reductions in these

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

costs.4,5 A fundamental aspect of this approach was to accept that wellbore instability was inevitable and to manage it rather than to eliminate it. The drilling industry historically addresses wellbore instability issues in two ways. The first approach treats the problem on an ad hoc basis; for specific problem formations, data and cores are collected and the drilling history is analyzed, allowing the formulation of a set of empirical rules. In the Valhall field of the North Sea, for example, wells drilled through the Middle Eocene formation at inclinations exceeding 65 are at high risk. These rules do reduce nonproductive time. However, they do not identify the underlying instability mechanism and do not appropriately relate it to drilling operations so that the full benefit of this knowledge is realized. Furthermore, many of these empirical rules apply to a well, and all need to be taken into account to determine the drilling parameters (e.g., mud flow rate, rate of penetration, pump pressure, trajectory). Techniques exist to solve this type of problem, but these are often not applied and can lead to an inadequate set of drilling parameters that can trigger wellbore instability. The second approach is based on log interpretation methods that estimate the safe mud weight window using rock strength and in-situ stress state predictions based primarily on sonic logging. The calculations are made, however, within the framework of classical rock mechanics where it is assumed that the maximum and minimum mud weights are governed by the onset of breakouts and fractures, respectively. Several common modes of wellbore instability (e.g., fractured shales, fault reactivation) are not amenable to this classical approach. The description of wellbore stability is, therefore, generally incomplete. Both approaches can be applied before or after, but not during, drilling. Any lessons learned from data or experience gathered on a well can therefore only be applied on subsequent wells in the same field. As a result, several wells can be drilled before the minimum cost construction technique is found. This significantly increases both the capital required for field development and the cycle time. Managing borehole instability in real time would potentially allow learning to be implemented on the current well so that the optimal construction technique is achieved over the minimum number of wells. Such an approach has not, however, been possible until recently because of technical constraints. The following developments now make it feasible: 1. There is increasing availability of MWD data.6 2. Wellbore deformation and failure mechanisms, and their relation to stress state, are better understood.7 3. There is improved understanding of how drilling practices (e.g., frequency of wiper trips, swab and surge pressures) influence instability and of how, in turn, instabilities of different kinds influence drilling. The RTWBSC concept uses real-time measurements and interpretation to manage wellbore instability (real-time here means essentially during drilling of the well; some real-time data arrive immediately as a formation is being drilled, but other data can be delayed by up to a few hours). Although

wellbore instability can be classified as either mechanical (e.g., failure of the rock around the hole because of high stresses, low rock strength, or inappropriate drilling practice) or chemical (damaging interactions between the rock, generally shale, and the drilling fluid), the integration of understanding of chemical and mechanical damage remains problematic despite intensive efforts throughout the oil industry. Accordingly, the RTWBSC process (a) determines whether a particular drilling problem is mechanical or chemical in origin, (b) deals with the mechanical aspects and makes recommendations, based on known rules of thumb, if the problem is chemical in origin. The four main components of this process are described in the next section. The first component is a wellbore model consisting of the trajectory, in-situ stress state, rock constitutive parameters and all types of instability mechanisms, together with a description of the drilling practices. It is constructed through the two approaches by which wellbore instability is currently addressed, and it uses offset well data, drilling experience and in some cases a seismic survey to define the geological structures. The accuracy of the model depends on the information available, but it always provides a framework against which real-time observations and interpretations are judged. The second component is the data acquisition program, which defines the types of data and sampling rate necessary to provide a reliable diagnosis of the instability mechanisms, their severity and the conditions under which they occur. The third component is a software tool that accepts data from a wide range of sources and manages the data flow, diagnoses the instability mechanisms, and quantifies both their severity and corrective drilling practices. A key part of this third component is the refinement of the subsurface model. The fourth component is a communication tool, such as an intranet Web site, that acts as a data repository and enables rapid dissemination of information and recommendations. The RTWBSC process was validated on an ERD well in the Valhall field of the North Sea (see Valhall field test section). This field, operated by BP Amoco Norge, is located in offshore blocks 2/8 and 2/11 in the Central Graben area of the southern part of the Norwegian North Sea. It was discovered in 1975, when the exploration well 2/8-6 encountered over 100 m of hydrocarbon-bearing section in Late Cretaceous chalk formations. Production began in 1982 from the highly porous Tor and Hod chalk formations.8 Valhall was originally developed to recover reserves of 250 million barrels. There are ongoing projects to increase recoverable reserves to 1000 million barrels.9 One project involves accessing downflank reserves in the far northern and southern parts of the field through ERD wells. Although this is economically attractive, because of the potential for significant gains in recoverable reserves over a relatively small cycle time, wellbore instability is a major problem: There is a high risk that wells will be abandoned or suspended before reaching their target. This factor, together with the availability of a comprehensive data set, meant that Valhall was suited to

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL 3

demonstrate the value and viability of real-time detection and control of wellbore instability. Real-time wellbore stability control process The process uses four main components: the MEM, a data acquisition program, data management software and a communication system. The implementation of the process and its components, for drilling optimization, is shown schematically in Fig. 1 and has three phases: 1. In the design phase, relevant data are gathered and the MEM is constructed. The wellbore stability and drilling plans are then formulated: these are taken into account during the design of the data acquisition program. 2. In the execution phase, the drilling process is monitored and data are aquired to detect instability. 3. In the evaluation phase, which also occurs during drilling, real-time data are interpreted, the MEM is updated as necessary and recommendations relating to drilling practices are made to the rig crew. Interpretation of realtime data should be made within the context provided by the MEM and the wellbore stability predictions: assessments of the validities of the interpretation and/or the MEM will be more reliable. The four components are discussed further in the following paragraphs. The implementation of the design-executeevaluate cycle is discussed in the Valhall field test section and is illustrated using events that occurred during the drilling of the well. Planning. Before drilling, the optimal, or least damaging, well construction techniques are identified through prognoses of the geology and instability mechanisms likely to be encountered and estimates of the conditions, including the stress state, that trigger the mechanisms. In areas where drilling has occurred, the geology can be characterized using offset well data such as logs and geological reports, perhaps combined with a seismic survey. In areas where no exploration has occurred (the case in Cusiana), it is necessary to rely on a geological prognosis only, albeit one now aided by geological modeling software tools.10,11 The process of analyzing the likely instability mechanisms and estimating their trigger conditions is described in the following paragraphs. Review of offset well construction. This review should include the drilling phase, with trips and casing runs. Attention is typically focused on (a) mud losses, cavings rates and morphology, geological reports and any (partial or full) stuck pipe incidents and (b) relating instability issues to the operation (tripping, backreaming) and comparing the mud density and/or equivalent circulating density (ECD) to the predicted stable mud weight window. The product of this review includes the instability mechanisms and their severity, indexed to true vertical depth (TVD) or, more generally, incorporated within an earth model. Any key factors influencing the instability, such as well or bedding inclination, should also be noted.12 The instability

mechanism at a given depth is categorized as either breakouts, sloughing, natural fractures, weak planes, drilling-induced fractures, faulting, undergauge hole, interbedded sequence, overpressured formation, unconsolidated formation, mobile formation, permeable formation or chemical activity. This list is not exhaustive; further categories can be envisaged. The severity of the instability is categorized as low, medium or high. 1. A low severity problem is one for which symptoms exist, but no remedial action is required. 2. An instability of medium severity has noticeable symptoms; minor action is required either to inhibit the problem or to deal with its consequences. An example is minor breakouts manifested by an increased cavings rate, or perhaps even a partially stuck pipe. The hole cleaning could be emphasized (to deal with breakout debris without stopping breakouts) or the mud weight could be increased by a small amount, thus inhibiting the problem. 3. A problem of high severity is a potential well-stopper. Without major remedial action (running casing), a total loss of borehole integrity is highly likely and will result in a sidetrack or abandonment. Density, sonic and gamma ray logs. Data can be constructed using logs from several offset wells. The sonic log should ideally consist of compressional and shear slownesses. In many cases, however, only compressional slowness is available: an empirical correlation is then needed to derive the shear wave speed. These data form the primary input for the MEM, which consists of the in-situ stress state, the formation constitutive parameters and the failure mechanisms. The accuracy of the MEM can be enhanced by correlating (a) the log-derived results to point data, such as information from cores or leakoff tests, and (b) quantities such as sonic velocities to constitutive parameters such as formation strength.13 The MEM and proposed well trajectory may then be used to predict the safe mud weight window.14 The instability evaluation must be combined with other factors considered during well planning, such as mud hydraulics, hole cleaning, torque and drag calculations, and casing programs. A discussion of how the relevant factors are integrated exceeds the remit of this paper. It is evident, however, that many iterations are required before the final trajectory and drilling practices are decided. Planning in Valhall. The geological structure of Valhall is dominated by a central uplift, elongated about a NorthNorthwest axis.8 Otherwise, the stratigraphy is relatively uniform, with formations varying a little in thickness and dipping away from the center of the field at an angle of approximately 5o to the horizontal. Figure 2 shows a generic stratigraphic column. Owing to the relatively uniform geology, 1D mechanical earth models (where the properties are only a function of TVD) are adequate for wellbore stability purposes in this field. The structure of Valhall is not, however, entirely axisymmetric, so it was necessary to construct MEMs that are

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

locally valid for the northern and southern parts of the field. Since the RTWBSC concept was validated in an ER well drilled in a northwesterly direction (Figs. 3 and 4), attention is restricted in the remainder of this paper to the MEM constructed for northern Valhall. The MEM derived prior to drilling is shown in Figs. 5 and 6.15 The associated mud window, derived using an undrained linear elastic-brittle model, is shown in Fig. 7.16 It is important to note, however, that in cases where the geological structure and/or rock behavior is more complicated (e.g., a salt diapir), fully numerical techniques, such as finite element analyses, are necessary to model the in-situ stress state and derive the mud weight window. The classical rock mechanics approach just described determines the risk of breakouts and mud losses. It is, however, increasingly recognized that many wellbores, especially those drilled at higher inclinations, fail because of instability mechanisms that are not amenable to this approach. Examples of such mechanisms include fractured zones, mobile formations and faulting. Practical quantitative or semiquantitative modeling of these instabilities requires development. Currently, issues pertaining to them are handled in a soft manner: drilling histories are analyzed to identify the location and severity of nonclassical failure. The dominant instability mechanisms for the discussed well are shown in Fig. 8. Medium and high severity instabilities are denoted by the thick vertical dotted and solid lines on the right side of the figure, respectively. Experience indicates that the naturally fractured zone lying between 2000 and 2200 m TVD [4160 and 4570 m measured depth (MD)] poses the most severe risk, particularly if the well inclination through this zone exceeds 65o. The region from 1510 to 1850 m TVD (2370 to 3680 m MD) contains rock with weak bedding planes; it becomes more unstable with time. Drilling strategy. The combination of the mud window (Fig. 7) and analyses of other hazards (Fig. 8) indicated it was impossible to drill the well without continuous rock failure because simultaneous remedies to all the instabilities did not exist: 1. The mud weight needed to be high to avoid both breakouts and underbalanced drilling. 2. The mud weight needed to be less than the minimum insitu horizontal stress to prevent fluid loss, particularly into the fractured zone between 2000 and 2200 m TVD. To formulate a strategy for drilling the well, it was necessary to assess the risk posed by each instability: 1. Breakouts are a controllable failure. This type of failure is either self-stabilizing (breakouts tend to stop growing after reaching a certain size) or can be controlled by remedial actions (increasing mud weight prevents breakout development), or both. 2. Destabilized fractured zones are an uncontrollable failure. This type of failure, once initiated, cannot be stopped easily and is expected to become ever more severe. Thus, the strategy for the well was to prevent destabilization of the fractured zone between 2000 and 2200 m TVD. This

approach is contrary to conventional drilling practices, which emphasize breakout control. This strategy involved the following: 1. A relatively low mud weight. It was accepted that this would induce breakouts. The resulting cavings were dealt with using hole-cleaning procedures and rate of penetration (ROP) control. The mud weight was increased in steps of 0.1 lbm/gal only if the rate of cavings influx into the annulus overwhelmed hole-cleaning capabilities. 2. Specific attention, within the monitoring program (discussed below), to cavings and mud losses to provide a warning of a destabilized fracture zone. Recommendations for drilling parameters, such as ROP, could only be quantified as drilling progressed and trends for parameters such as ECD became established. Data acquisition. A reliable diagnosis of the instability mechanisms, their severity and their trigger conditions requires a combination of MWD and LWD measurements, mud analysis, geological/micropalaeontological analysis and other surface information such as hookload and mud flow rate. The variety of data is notable and necessary because (a) wellbore instability and the influence of operations, together with the relationship between them, are very complex, and (b) the process cannot rely on any single source of information. Thus, sensible interpretations require integration of all available information. It is also important that the sampling rates are such that interpretations can be provided on an appropriate timescale. Clearly, data acquisition programs are designed on an individual well basis, taking into account the nature and risk posed by the anticipated hazards, together with other factors such as budget constraints, formation evaluation requirements and contingency plans. The benefits provided by acquiring specific types of data and desirable sampling rates are summarized below: use and flow of the data are discussed in the following paragraphs. LWD measurements can include annular pressure, caliper, gamma ray, resistivity (phase and attenuation; i.e., shallow and deep, respectively) and compressional slowness: 1. Annular pressure is an important measurement. It can be used to (a) determine the risk of mud losses or shear failure, (b) assess hole-cleaning effectiveness, and (c) evaluate annular cuttings/gas loading. 2. Resistivity measurements can be used to evaluate mud invasion into fractured or permeable zones and faults. 3. Compressional slowness can be used to determine formation strength or flag overpressured domains. The evolution of time-dependent instabilities can be assessed using the appropriate time-lapse data. MWD and surface measurements must include deviation, inclination, ROP, pump pressure, rotation rate in revolutions per minute, downhole torque, downhole weight on bit, surface torque and hookload, possibly combined with turbine revolutions per minute. The data are principally used to determine the risk of stuck pipe and hole-cleaning

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL 5

effectiveness. The combination of ROP and ECD data enables annular cuttings and/or gas loading to be managed. Mud logging data should, for safety reasons and loss control, consist of mud flow rate in and out, total active tank volume, change in the total active tank volume, average background gas and maximum background gas. Periodic mud measurementssuch as rheological parameters and fluid loss, and the percentages of oil, water and solidsare also desirable, not least to aid interpretation of annular pressure data. A cavings analysis greatly reduces the ambiguity in instability diagnoses; rate (i.e., volume), size range, average size, morphology, lithology, and source depth are desirable measurements. It should also be noted if the cavings are old (in a cuttings bed for several days) or are new (just become detached from the wellbore wall). This is discussed in Appendix A. LWD, MWD and surface information should be monitored continuously during drilling and also while tripping, provided the driller is pumping out of hole at a sufficiently high flow rate. It is advisable to conduct cavings analyses at 30-min intervals, with periodic mud logging data gathered every few hours. All data should be indexed to date, time, hole depth and bit depth to identify the effect of specific operations. Last et al. correlated greatly increased cavings volumes with trips and back-reaming.4 An appropriate selection of these measurements forms the basis of any data acquisition program that is part of a real-time wellbore stability control process. It is not an exhaustive list; other key data may be required depending on the nature of the instability. For example, if swelling shales are a severe problem, further mud analysis may be required. It is also not a must have list; the approach to real-time detection and control must be flexible so that no measurement is critical. The data acquisition program for the Valhall field test consisted of surface measurements, mud and cavings analyses, and extensive MWD and LWD measurements. The benefits provided by this program are discussed in the Valhall field test section. Decision support software. The process summarized in Fig. 1 is embodied, to a significant extent, in the decision support software shown in Fig. 9 and is designed for use on a Pentium laptop computer. This package contains data manipulation, evaluation and visualization algorithms that help the user make efficient, effective real-time decisions. It is not intended to be an automated drilling optimization tool. The package supports the user in five main areas: predicting instability mechanisms and their trigger conditions, diagnosing the wellbore state using real-time data, updating the earth model to ensure consistency between the predicted and the diagnosed states, providing recommendations to the driller, and visualization. Predicting the instability mechanisms and their trigger conditions has been discussed. Algorithms enable users to build trajectories and MEMs; safe mud weight windows are

calculated with an undrained elastic-brittle theory. Diagnosing the wellbore state using real-time data involves the integration of a number of disciplines; namely, geological analysis, drilling mechanics, formation evaluation, wellbore stability and mud logging (mud analysis and palaeontology). This is a complex process requiring human judgment, particularly to distinguish wellbore instability and poor hole cleaning. Diagnoses made within the context provided by the MEM and the planning analysis are more reliable than those made using only the real-time data. After the diagnosis is completed, the current wellbore state is compared to the model; human judgment determines if the two are consistent. If inconsistencies exist, it is necessary to update the MEM. When the predicted and diagnosed wellbore states agree adequately, recommendations either to suppress the instabilities or minimize their consequences can be made to the driller. For example, increasing mud weight will reduce the amount of breakouts, whereas decreasing the ROP will reduce the rate at which breakouts are exposed, resulting in less debris in the annulus given constant flow and rotation rates. The recommendations should apply over the entire open-hole interval or a specified subsection of it. The aim is to optimize the condition of the complete open-hole section and not to focus on remedial actions required just at the bit. Visualization is a key component of the support tool; the quality of the real-time decisions depends strongly on the ready and unambiguous assimilation of the output of the RTWBSC process. For example, Fig. 10 shows the predicted damage zone around a borehole resulting from shear failure. It is immediately evident that the failure is extensive enough to warrant increasing the mud weight to suppress the failure; hole-cleaning procedures would not be able to cope with the debris that would fall into the annulus. Communications. Decisions on well construction are made at the wellsite and in the office. The influence exerted by each location varies according to the operator, the level of actual/anticipated risk and the maturity of the field development program. The distribution of wellsite data and the procedures for implementing decisions resulting from the RTWBSC analysis must be compatible with working practices; there should be particular attention on communication. During the Valhall field test, the RTWBSC process was managed in the office by wellbore stability specialists working with an existing team of drilling engineers. A Schlumberger engineer trained in drilling risk management was at the wellsite to ensure (a) the necessary measurements were taken correctly and (b) the data flowed efficiently to the relevant people at the wellsite and in the office. This engineer was also responsible for communicating recommendations for wellbore stability at the wellsite and for conducting the cavings analysis. Although these recommendations are usually made by office-based personnel, a suitably trained engineer can make recommendations independently in some situations.

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

The acquired data are generally analyzed by personnel of differing disciplines (geologists, drilling engineers, mud loggers, formation evaluation and wellbore stability specialists), both at the site and in the office. This joint evaluation requires a reliable link with sufficient bandwidth between rig and office and a readily accessible repository for the information. The experience gained from Valhall and from the BP Amoco ETAP field has shown that a Web site can fulfill this requirement. Training classes on wellbore stability in general and cavings monitoring in particular were given to all drilling and mud logging crews going offshore on Valhall. The crews responded positively to these classes, which focused on avoiding, rather than reacting to, instability problems. Valhall field test The field test began with the drilling out of the 13 3/8-in. casing shoe at 1610 m (Fig. 8) and continued until the reservoir was penetrated at 5602 m (Point C). The section between the casing shoe and Point A was drilled using a rotary steerable assembly with a 12.25-in. bit and a 14-in. three-arm stabilized reamer. Sections AB and BC were drilled with conventional steerable assemblies having 12.25-in. bits. Drilling from casing shoe to Point A (1610-3832 m MD). The 13 3/8-in. casing shoe was drilled out using a mud weight of 14.2 lbm/gal; a leakoff indicated that fluid loss occurred at pressures exceeding 15 lbm/gal. During drilling, ECD data indicated that a safe lower bound to the minimum horizontal stress was 15 lbm/gal over the interval 1610 to 2040 m MD. The mud weight had to be raised to 14.6 lbm/gal by 2200 m MD to reduce background gas levels from 20% (gas peaks of 35% were observed). These high gas levels were consistent with the drilling hazards prognosis (Fig. 8) and resulted from matrix gas being released into the annulus as rock was crushed beneath the bit. The necessity for further mud weight increases, which would have led to the destabilization of the critical fractured zone between 4160 and 4570 m MD, was eliminated by slowing the ROP to below 30 m/h (Fig. 11). This action reduced the rate at which gas was released into the annulus and, combined with the mud weight increase of 0.4 lbm/gal, eventually led to background gas levels decreasing to less than 5%. Wellbore stability in this section was controlled following the strategy outlined previously. A mud weight of 14.2 lbm/gal prevented significant breakouts after the shoe was drilled out (Fig. 7). Subsequent mud weight increases resulted solely from the overpressure problems described, as hole cleaning coped with the levels of debris in the annulus caused by breakouts. Cavings analysis indicated no failure had occurred as a result of weak bedding planes while drilling this section (Fig. 8), although the instability mode became active during one trip (discussed below). The cavings rate is shown in Fig. 12. During the drilling of this section (0 to 100 hr approximately) the cavings rate remained reasonably steady, although there

was a reduction at around 3650 m MD caused by a packoff. The steady cavings rate resulted from the use of a rotary steerable tool and the absence of severe wellbore instabilities. The ECD was constrained by ensuring the ROP did not exceed 30 m/h: this rate controlled the cuttings loading and gas levels in the annulus. The ROP limit was deduced by correlating annular pressure while drilling and ROP data. Figure 13 shows a typical case. During the period 33 to 36 hr, the ROP exceeded 30 m/h and the ECD increased gradually as the cuttings loading in the annulus increased. Partial packoffs then occurred, causing the ECD to become highly erratic. Subsequently, the ROP was reduced to below 30 m/h and the hole was cleaned more effectively by increasing both the revolutions per minute and flow rate. The ECD became more stable and decreased gradually to 15.1 lbm/gal, indicating the ECD effects were a result of inadequate hole cleaning rather than continued wellbore instability. As drilling proceeded, mud weight rose to 14.6 lbm/gal and the ECD increased above the estimated minimum horizontal stress (Figs. 6 and 7) to between 15 and 15.2 lbm/gal, without mud losses. The minimum horizontal stress was therefore assumed to be 15.2 lbm/gal in the section 1610 to 3832 m. Although this value is a lower bound of h , it is more accurate than the previous h estimate. Figure 14 shows the refined model of the in-situ stress state. A severe problem occurred at 3649 m, where a fault was encountered. This fault was diagnosed using resistivity, gamma ray and mud loss data, as shown on Fig. 15. It can also be inferred from this data that a packoff occurred below the LWD resistivity tool where the ECD sensor is housed. The surface pump pressure increased significantly while the ECD remained constant. The reason for the packoff is uncertain, but it is due to either fault movement or rubbilized rock, which can occur around faults, blocking the annulus. This incident caused seal failure on the rotary steerable system, leading to lubricant loss. The assembly had to be pulled out of hole after drilling to 3832 m MD (Point A on Fig. 8). Specific procedures for wellbore stability control were developed for these trips and are discussed separately. The other key problem encountered in this zone was the presence of limestone stringers at 2943, 3258, 3290, 3305, 3330, 3350, 3546, 3508, 3550, 3596, 3645, 3650, 3668 and 3795 m MD. When the bottomhole assembly (BHA) was pulled back through these stringers, there was a tendency to pack off. It is thought that while the limestone stringers remained in gauge, hole enlargements either side of them resulted in lower mud velocities, which led to the formation of cuttings beds. Accordingly, during circulation periods the BHA was positioned away from these stringers. At the same time, to limit damage in the weakest formations, the MEM was used to select the strongest zones for rotation of the BHA (Fig. 5). Drilling from Point A to the reservoir (3832-5602 m MD). This section was drilled in two stages (AB and BC on Fig. 8) with conventional steerable assemblies having 12.25-in. bits.

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL 7

Wellbore stability control in this section consisted of ensuring the ECD did not exceed 15.2 lbm/gal (this initial constraint was later relaxed to 15.35 lbm/gal) to avoid destabilizing the naturally fractured zone. This was difficult given the large amount of sliding that occurred, and there was a strong emphasis on hole cleaning and ROP control procedures. The stability of caving beds was also a source of concern. These beds tend to avalanche down the well at inclinations around 60, causing pipe and BHAs to stick. In Section AB, it was found, unfortunately, that holding angle was difficult. Drilling was therefore halted at 4306 m MD (Point B) for the following reasons. 1. If drilling had continued, there was a risk the well would have penetrated a partially drained section of the reservoir, which is to the left of the fault shown on Fig. 8. 2. The wellsite engineer observed a caving produced through destabilization of the naturally fractured zone. The proximity of the planned trajectory to the fault (Fig. 8), made it necessary to trip out of hole to change out the BHA. The cavings analysis dictated the trip should occur without further drilling so as to limit damage to the key fractured zone. During the trip back into the hole, 12 bbl of mud were lost when the ECD exceeded 15.35 lbm/gal at 4120 m MD. The minimum horizontal stress in the MEM was therefore revised to 15.35 lbm/gal from 1610 to 4306 m MD. The refined model of the in-situ stress state is shown in Fig. 16. Figure 17 shows the strength profile of the overburden (to Point B) updated using LWD compressional slowness data. This data verified the rock strength profile constructed using offset well data (Fig. 5) and therefore no significant changes were made in the drilling strategy. The updated mud window is shown in Fig. 18. In Section BC, the necessity to maintain the ECD at 15.35 lbm/gal or less meant that breakouts were a severe problem (Fig. 18). The difficulty of cleaning hole with such severe breakouts can be seen in Fig. 12. The cavings rate varied greatly and, in particular, there were sudden bursts of solids over the shaker. The reservoir was, however, penetrated (Point C) ahead of schedule. Limestone stringers were again encountered at 4000, 4075, 4150, 4700, 4740, 4780, 4830, 4930, 4985, 5024, 5160, 5170, and 5310 m MD. Tripping procedures. To prevent problems associated with swabbing as the downhole assembly was pulled out of hole, the mud weight was increased from 14.6 to 14.8 lbm/gal during the first two trips out from Points A and B (Fig. 8). The procedure required the heavier mud to be circulated into the well after pulling 10 stands. The increase in mud weight was deduced from an analysis of pressure while drilling data from offset wells. Conversely, during the trips in to Points A and B, the mud weight was reduced from 14.8 to 14.6 lbm/gal to minimize problems associated with surging. The procedure required mud gels to be broken, after tripping 10 stands into the hole, by increasing the revolutions per minute. Lighter mud was then circulated into the well. It was also found that

increases in ECD during trips into hole were reduced by shearing the mud on the surface prior to circulating it downhole. The trip out of hole following the reservoir penetration (Point C) required the mud weight to be increased from 14.6 to 14.8 lbm/gal at the start of the 12.25-in. hole (Point A). This ensured the mud had sufficient carrying capability in the 14-in. hole section while keeping the effective mud weight in the entire open-hole section to a minimum. This was an important consideration for the casing operation; as the casing is run, large surge pressures destabilize the naturally fractured zones. During the trip out of the hole after the reservoir penetration, the hole was accidentally swabbed, causing the well to collapse at 3500 m MD. The wiper trip to clean this damage unfortunately initiated a sidetrack from around 3600 m MD. This sidetrack also penetrated the reservoir using the same wellbore stability strategy described previously in the Planning section, with further emphasis on hole cleaning. The casing string then re-entered the original track, which had been open through the fractured zone for several weeks, and was landed below the fractured zone. It could not, however, quite reach the bottom of the hole. The well as a whole cannot, therefore, be called a success. However, since the reservoir was penetrated ahead of schedule and the casing could still be installed in the troublesome fracture zone after several weeks of open-hole exposure to drilling fluid, the original wellbore stability strategy and the real-time approach can still be considered a success. Conclusions Real-time monitoring and control of wellbore stability systematically reduce the drilling risks associated with wellbore instability and other geological hazards. This realtime process treats such instabilities and hazards as conditions that impose constraints on the drilling parameters (mud weight, ROP, revolutions per minute, etc.) and then provides recommendations on the drilling practices most likely to ensure the entire hole section is maintained in the best, or least damaged, state. The concept and process discussed here have been validated on an ER well drilled in the Valhall field of the North Sea. The well reached its target ahead of plan and with much lower mud loss to the formation than usual (around 10% of the typical value for a well such as this on Valhall) and negligible activation of the fracture zones. The well was cased to below the unstable overburden intervals. Nomenclature UCS = Rock uniaxial compressive strength = Rock friction angle h = Total in-situ minimum horizontal stress H = Total in-situ maximum horizontal stress V = Total in-situ overburden stress P P = Pore pressure

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

Acknowledgments The Real-Time Wellbore Stability project was partly funded by the European Commission, under the THERMIE initiative (contract number OG-0199-95). During the Valhall field test, Schlumberger personnel located offshore (Paul Benoit, Ruth Bertelsen, Gael Boche, Andy Foster, Caroline Hatch, Vidar Haugen and Al Pattillo) were responsible for the data acquisition program. They contributed greatly to the success of the field test through their initiative and dedication. The assistance provided by Charles Jenkins of Schlumberger Cambridge Research is also gratefully acknowledged. Appendix A - Cavings monitoring An analysis of cavings can provide a signal that the borehole is failing and indicates both the nature of the instability and the troublesome formations. Cavings dimensions range from a few millimeters to 10 cm or more, with larger examples rising to the surface while lodged in the BHA. There are four main types of caving: tabular, angular, splintered and those that cannot be characterized. Examples of the first three types are shown in Figs. 19 to 21. Tabular cavings, shown in Fig. 19, are the result of natural fractures or weak planes. In the case of natural fractures, the fluid pressure in the annulus exceeds the minimum horizontal stress, resulting in mud invasion of fracture networks surrounding the wellbore. This can result in severe destabilization of the nearwellbore region (resulting from movement of blocks of rock), leading rapidly to high cavings rates, lost returns, stuck pipe and tools lost in hole. The blocks of rock are bounded by natural fracture planes and therefore have flat, parallel faces (Fig. 19). The other characteristic is that bedding, if any, will not be parallel to the faces of the caving. In the case of weak planes, the combination of low mud weight and a borehole axis that is within approximately 15o of the bedding direction can induce massive failure along the planes of weakness, leading to the symptoms described above.12 Cavings resulting from weak planes are characterized by having flat, parallel faces. The bedding direction is also parallel to the faces. Figure 20 shows angular cavings, which are a consequence of breakouts. These cavings are characterized by curved faces with a rough surface structure. The surfaces intersect at acute angles (much less than 90o). Splintered cavings are shown in Fig. 21. These cavings have two nearly parallel faces with plume structures. This type of caving is due to tensile failure occurring parallel to the borehole wall and commonly occurs in overpressured zones drilled with a small overbalance. The cavings rate can indicate the severity of failure, coupled with the efficiency of hole cleaning. It is measured every 30 min by the time required to fill a bucket placed underneath the shakers. This method may seem crude, but it is versatile (in terms of the number of different models of rig that it can be applied to) and reliable; more sophisticated solids measuring devices have been tried on a number of rigs, but very few have been satisfactory.

Micropalaeontological analyses determine the geological age of cavings. During the field test, an analysis of tabular cavings indicated that they originated from the upper section of the open hole, where the exposure time was longest, rather than from the dangerous naturally fractured zone. References
1. Santarelli, F.J.: Rock mechanics characterization of deep formations: a technico-economical overview, paper SPE 28021 presented at the 1994 Eurock Rock Mechanics in Petroleum Engineering Conference, Delft, August 29-31. 2. Charlez, P.A., Bathellier, E., Tan, C. and Francois, O.: Understanding the present in-situ state of stress in the Cusiana field Columbia, paper SPE/ISRM 47208 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 3. Charlez, P.A. and Onaisi, A.: Three history cases of cases rock mechanics related stuck pipes while drilling extended reach wells in North Sea, paper SPE/ISRM 47287 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 4. Last, N., Plumb, R.A, Harkness, R., Charlez, P., Alsen, J. and McLean, M.: An Integrated Approach To Evaluating and Managing Wellbore Instability in the Cusiana Field, Colombia, South America, paper SPE 30464 presented at the 1995 Annual SPE Techical Conference and Exhibition, Dallas, Oct 22-25. 5. Last, N., Plumb, R.A and Harkness, R.: From theory to practice: evaluation of the stress distribution for wellbore stability in an overthrust region by computational modelling and field calibration, paper SPE/ISRM 47209 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 6. Rosthal, R.A., Best, D.L. and Clark, B.: Borehole caliper while drilling from a 2-MHz propagation tool and borehole effects correction, paper SPE 22707 presented at the 1991 Annual SPE Techical Conference and Exhibition, Dallas, Oct 6-9. 7. Bradford, I.D.R. and Cook, J.M.: A semi-analytical elastoplastic model for wellbore stability with application to sanding, paper SPE 28070 presented at the 1994 Eurock Rock Mechanics in Petroleum Engineering Conference, Delft, Aug 29-31. 8. Munns, J.W.: The Valhall field: a geological overview, Marine and Petroleum Geology (1985), February, p. 23-43. 9. Kristiansen, T.G., Mandzuich, K., Heavey, P, and Kol, H.: Minimizing drilling risk in extended-reach wells at Valhall using geomechanics, geoscience and 3D visualization technology, paper SPE 52863 presented at the 1999 SPE/IADC Drilling Conference, Amsterdam, March 9-11. 10. Bryant, I.: Cybergeologist: 3D reservoir modelling using digital geological analogs, GasTIPS, Spring 1998, GRI-98/0144-001. 11. Bryant, I., Kaufman, P.S., McCormick, D.S. and Tilke, P.G.: Knowledge capture and reuse in geological modelling, paper presented at Gulf Coast Section of Society of Economic Mineralogists and Paleontologists Annual Meeting, December 1999. 12. Okland, D. and Cook, J.M.: Bedding-related instability in highangle wells, paper SPE/ISRM 47285 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 13. Plumb, R.A.: Influence of composition and texture on the failure properties of clastic rocks, paper SPE 28022 at the 1994 Eurock Rock Mechanics in Petroleum Engineering Conference, Delft, August 29-31.

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL 9

14. Bradford, I.D.R., Fuller, J., Thompson, P.J. and Walsgrove, T.R.: Benefits of assessing the risk of solids production in a North Sea reservoir using elastoplastic modelling, paper SPE/ISRM 47360 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 15. Kristiansen, T.G.: Geomechanical characterization of the overburden above the compacting chalk reservoir at Valhall, paper SPE/ISRM 47348 presented at the 1998 Eurock Rock Mechanics in Petroleum Engineering Conference, Trondheim, July 8-10. 16. Fjaer, E., Holt, R.M., Horsrud, P., Raaen, A.M., and Risnes, R.: Petroleum related rock mechanics, Elsevier, Amsterdam (1992).

SI Metric Conversion Factors bbl x 1.589873 ft x 3.048* gal (U.S. liq) x 3.785412 in. x 2.54* lbm/gal x 1.198264 psi x 6.894757 * Conversion factor is exact.

E-01 = m3 E-01 = m E-03 = m3 E+00 = cm E+02 = kg/m3 E-03 = MPa

Fig. 2. Generic stratigraphic column for the Central Graben. (Extracted from Kristiansen et al.9)

Fig. 3. A plan view of the Valhall field and ER well. (Extracted from Munns8)

Fig. 1. The design-execute-evaluate cycle for real-time wellbore stability control. The starting point is at the top, with initial data gathering and construction of the first MEM in the planning phase. The remainder of the cycle occurs as the well is being drilled.

Fig. 4. Trajectory of the ER well.

10

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

Fig. 5. Uniaxial compressive strength and friction angle in the Valhall overburden, estimated before drilling the ER well. Fig. 8. Anticipated instability mechanisms and their severities. The thick vertical dotted and solid lines on the right of this figure denote medium and severe instabilities, respectively.

Fig. 6. The in-situ stress state in the Valhall overburden, estimated prior to drilling the ER well.

Fig. 9. The flow of information and decisions through the prototype system. Ellipses represent data input, diamonds are decision or comparison points, and rectangles are processes. The starting points are the two upper ellispes, and the finish point is the lower left corner.

Fig. 7. Mud weight window, estimated prior to drilling the ER well.

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL11

Fig. 12. Cavings data. The cavings rate is 1440 x the reciprocal of the time, in seconds, taken to fill a 4.5-L bucket placed under the shakers. Fig. 10. A schematic of shear-induced borehole failure.

Fig. 13. Time-based data acquired during drilling of the interval between the 13 3/8-in. casing shoe and Point A. Fig. 11. The influence of mud weight and ROP on gas levels (shown as squares).

12

BRADFORD, ALDRED, COOK, ELEWAUT, FULLER, KRISTIANSEN, WALSGROVE

IADC/SPE 59121

Fig. 14. The in-situ stress state, refined following the drilling of the interval between the 13 3/8-in. casing shoe and Point A.

Fig. 16. The in-situ stress state, refined following the trip into hole to Point C.

Fig. 15. Identifying a large fault at 3649 m MD. PUMP and TVCA denote surface pump pressure and volume change in active mud tanks, respectively. TVCA takes account of the increase in hole volume during drilling.

Fig. 17. Strength parameters calculated using LWD compressional slowness data.

Fig. 18. Revised mud window calculation.

IADC/SPE 59121WHEN ROCK MECHANICS MET DRILLING: EFFECTIVE IMPLEMENTATION OF REAL-TIME WELLBORE STABILITY CONTROL13

Fig. 19. Tabular caving.

Fig. 20. Angular caving.

Fig. 21. Splintered caving.

Você também pode gostar