IBP3328_10 IMPROVING HORIZONTAL WELL PLANNING AND EXECUTION USING AN INTEGRATED PLANNING APPROACH AND GEOSTEERING A CASE STUDY EXAMPLE IN BRAZIL DEEP WATER Rajeev Samaroo 1, Gregory Stewart 2, Paolo Ferraris 3, Gunnar Holmes 4, Ana Beatriz Guedes 5, Lee Stockwell 6, Fabricio Bezerra 7 Copyright 2010, Brazilian Petroleum, Gas and Biofuels Institute - IBP This Technical Paper was prepared for presentation at the Rio Oi & Gas Expo and Conference 2010, held between September, 13-16, 2010, in Rio de Janeiro. This Technical Paper was selected for presentation by the Technical Committee of the event according to the information contained in the abstract submitted by the author(s). The contents of the Technical Paper, as presented, were not reviewed by IBP. The organizers are not supposed to translate or correct the submitted papers. The material as it is presented, does not necessarily represent Brazilian Petroleum, Gas and Biofuels Institute opinion, nor that of its Members or Representatives. Authors consent to the publication of this Technical Paper in the Rio Oil & Gas Expo and Conference 2010 Proceedings. Abstract The optimum placement of wells within a reservoir is one of the most important challenges in modern day drilling operations (Ayodele 2004). Well planning is an essential step in drilling a well. This step becomes even more important when the well is a high-angle or horizontal one, and the reservoir structure or reservoir architecture (sand and shale distribution) is not fully defined by seismic measurements. Shell Brasil Ltda. wanted to develop the Argonauta B-West deep water field using two horizontal wells. The initial well design was based on subsurface data from two vertical offset wells, 1-SHEL-01 and 3-ARGO-01, and 3D seismic volumes. Shell s subsurface team used well placement technology supplied by Schlumberger to plan, steer and collect subsurface data which was immediately used to increase the net reservoir volume encountered in the development wells. The design of the first well (BW-1) placed the net-to-gross estimate at 64% and subsequent drilling achieved 54% net-to-gross from the horizontal section. Shale encountered in the first well highlighted the probability that the netto-gross estimate along the second well, BW-2 could be as low as 50%. The advanced suite of logs from BW-1, together with the original information was used to design a new well path that resulted in a final 78% net-to-gross estimate for BW-2. We present a case study in which real-time well data were incorporated into planning and drilling a subsequent well and possibly increased the net-to-gross reservoir in the second well from 50% to 78%. The increase in net-to-gross estimate was realized with the aid of active planning and proactive well placement, which led to a successful well completion. 1. Introduction As more high-angle wells are drilled, the complexity of the well planning process also increases. However, what exactly is well planning? For this paper, well planning is defined as the process of bringing together all available reservoir and drilling data and using these data to design new wells that will not only maximize the reservoir potential of that one well, but to also apply the knowledge gained from previously drilled wells. Regardless of the type of horizontal well being drilled, the basic requirements for drilling a successful well are: 1) Planning the optimum well path based on the current integrated geological and geophysical models. 2) Monitoring the progress through real-time data updates. 3) Continuous remapping the drilling data to identify the true stratigraphic 1 Mechanical Engineer Schlumberger 2 Master, Geologist Shell 3 Master, Electronics Engineer Schlumberger 4 Geophysicist Shell 5 Electronics Engineer Schlumberger 6 Master, Business Administration Shell 7 Electronics Engineer - Schlumberger
position of the bit relative to the reservoir. 4) Timely reporting of the data updates and providing recommendations necessary for drilling ahead (Mottahedeh 2005). These four basic requirements fulfill the primary goal of any reservoir exploitation venture in generating an optimum well plan to maximize hydrocarbon production (McLennan 2006). The Argonauta B-West Field is located offshore Brasil at water depths near -1,700 m sub-sea and produces from Paleogene turbidite sands. B-West forms part of the cluster development called Parque das Conchas, located in the BC-10 block. The block is operated by Shell Brasil Ltda. (50% share) on behalf of the BC-10 Joint Venture which includes Petrobras (35%) and ONGC (15%). Shell Brasil embarked upon a campaign to develop this deepwater field using two horizontal producers. These wells were initially designed to penetrate the reservoir near the crest of the structure. The landing points for the two wells were about 230 m apart at their heels, while the azimuths were to deviate approximately 25º according to the drilling plan and then trend parallel near the toes (Figure 1). Figure 1. Map showing planned wells and offset wells, both projected onto the Reservoir Top. Because this was a deepwater project, with associated high costs, the need for optimum well placement, which begins with the planning phase, cannot be over-emphasized (Ayodele 2004). Partnering with a Schlumberger team of drilling engineers, well placement engineers, and petrophysicists, the Shell sub-surface team initiated the planning phase. 2. Available Data Prior to development, two vertical wells defined the field. The closest well (1-SHEL-01) was located about 120 m from the first planned horizontal producer (BW-1) and roughly 380 m from the second planned well (BW-2). The second offset well (3-ARGO-01) was located around 1,500 m from the nearest planned horizontal well and defined the flank of the hydrocarbon accumulation (Figure 2). The log suites included gamma ray, resistivity, density, neutronderived porosity, compressional and shear seismic, plus drilling parameters and performance data. Reservoir interval log data is presented in the correlation panel in Figure 3. Net-to-gross values in the pay sections ranged from 64% - 69% (excluding the thick shale in 3-ARGO-01), while net-to-gross values in the aquifer sections were above 85%. The anticipated net-to-gross for the horizontal producers was in the range of 50% - 75% with the target net-to-gross values between 55% and 65%. 2
Figure 2. Seismic section along the line projected between 1-SHEL-01 and 3-ARGO-01 based on Figure 1 map. Good quality 3D seismic field data were also available. Various subsurface layers were interpreted using the seismic data. The seismic data were also inverted and used to identify various formation facies within a 3D geologic model but overlap of the acoustic impedance properties of the individual facies and the discontinuous nature of the reflectors led to a degree of uncertainty in both the reservoir structure and facies that might be encountered along the well paths. Hence, initial well paths were planned based on the geologic model with the knowledge that the paths may need to be modified as the wells were being drilled. In addition, understanding the drilling behavior of the various lithologies aided in directional drilling management (Stockwell et al. 2010). To limit mud handling changes and save time, a batch drilling approach was used. The upper portion of the two wells was drilled, followed by the build-up and landing sections. Both wells were landed in the top of the reservoir and intermediate casing was set before drilling the reservoir sections. This made the real-time data analysis more valuable as there was little time to apply the learning from drilling the reservoir in BW-1 to drilling the reservoir in BW-2. During the landing phase it was confirmed that all of the seismic soft events in the Paleogene section did not correlate to sand. At the entry of BW-1 a dual-soft event occurred at the top of the reservoir (Figure 4). It was found that the first soft event was a soft silty-shale similar to one discovered in offset wells (Figure3). It was also found that flattening the well path in soft Paleocene sands was difficult and 7 m TVD was lost in BW-2, relative to the planned trajectory. 3
Figure 3. Log section with formation markers for both offset wells. 3. First Horizontal Well - BW-1 To ensure a full gravel pack across the screened completion both horizontal wells were initially designed with straight horizontal sections at a constant 89.5º inclination. Figure 4 shows the planned trajectory for BW-1 along the seismic well intersection, together with 1-SHEL-01. The BW-1 path was designed, as much as possible, to stay within a soft amplitude event near the top of the reservoir. Figure 4. BW-1 planned trajectory along its seismic intersection and 1-SHEL-01, with resistivity and porosity data along its seismic intersection. The main uncertainties involved in planning the wells included the occurrence of thin or small non-reservoir bodies undetectable in seismic data, multiple facies giving similar seismic response, depth placement of the seismic and the actual location of the well itself. All of these features carried a degree of uncertainty, which needed to be evaluated, and calibrated if possible. 4
With this in mind, two well placement models were constructed. The first model was a simple layer-cake model defined using the top and base reservoir horizons, the resistivity log layers from 1-SHEL-01, and the oil/water contact (OWC). Figure 5 shows the model generated in the Schlumberger real-time geosteering software. The planned trajectory in this model would yield net-to-gross in the order of 62%. The model is used to simulate the resistivity response of the tools at bed interfaces and to estimate the distance detectable away from the bed-boundary mapping tool, given the expected resistivity variations. Additional models were built to capture a larger range of possible tool responses and capture slightly more geologic complexity. Figure 5. Simple layer-cake real-time geosteering software model for BW-1. The blue and red dots indicate the simulated bed boundary mapping tool response for detecting upper and lower conductive boundaries, respectively. A second model, took less of a layer-cake modeling approach. Using an amplitude cut-off, the shale layer observed in 1-SHEL-01 was defined within the model as a discontinuous, lens-like feature (Figure 6). The planned trajectory in this model gave an estimated net-to-gross in the vicinity of 64%. Currently, the 2D resistivity modeling package is limited in its ability to capture the complex stratigraphy found in turbidite channels and lobe systems. Further work is being performed to make the modeling package more flexible in capturing complex geology. Figure 6. Revised real-time geosteering software model for BW-1 generated by correlating shale events in 1-SHEL-01 along the seismic well intersection. The blue and red dots indicate the simulated bed boundary mapping tool response for detecting upper and lower conductive boundaries, respectively. The BHA used for drilling the reservoir included formation evaluation tools, as well as, well placement hardware. For formation evaluation, a multifunction logging-while-drilling sonde (array resistivity, azimuthal density, azimuthal gamma ray, neutron-derived porosity, photoelectric factor, sigma, and spectroscopy), and real-time reservoir steering service (lithology independent porosity, pore size distribution, and permeability estimation) tools were used. Well placement requirements were fulfilled through the use of a point-the-bit rotary steerable system, azimuthal laterolog resistivity including a near-bit resistivity measurement, laterolog resistivity imaging, azimuthal gamma ray and imaging, and the bed boundary mapping tool. With the aid of the bed boundary mapping data and the imaging-while-drilling data, real-time adjustments were made to the planned trajectory. The real-time changes resulted in turning the well slightly and dropping the well path 5
early to exit a thick shale and to avoid other non-reservoir zones. The final net-to-gross estimate in the reservoir zone was 54%, with the post-drill model still holding several similarities to the initial pre-drill models (Figure 7). Although actual results were slightly below expectation, executing a failure-free operation meant that BW-1 had seen reasonably good drilling and geological performance. Figure 7. Final real-time geosteering software model for BW-1 as defined by the bed boundary mapping tool s real-time distance-to-bed boundary information, imaging-while-drilling image and dip interpretation, and seismic data. Note that the thick shale sections probably have inter-bedded sands. 4. Second Horizontal Well - BW-2 Even though BW-1 added limited information regarding vertical depth (10 m), having additional knowledge about the reservoir facies distribution and bed structure in real-time allowed for applying 700 m of lateral and 10 m of vertical knowledge to the second horizontal well path. Correlating events though the seismic volume confirmed the probable presence of a significant non-reservoir event that could separate one side of the field from the other and could extend across much of the BW-2 well path. As shown in Figure 8, the thick shale that intersected BW-1 can be correlated to the hard seismic amplitude. If this correlation holds true, BW-2 could intersect 50% shale over the lateral length of the well (not taking into account thin sands that may occur in the shale or thin shale below seismic resolution or slits possibly occurring within the reservoir sands). Figure 8. Correlation of the shale event from BW-1, across to BW-2, and highlighting the possible shale section that could be encountered by the initial planned trajectory of BW-2 (red). The revised trajectory (green) is expected to encounter a shorter section of shale by building inclination shortly after drilling out the casing shoe. Resistivity logs are overlaid in 3D along the wellbore. As a result, a new well plan was proposed. Due to losing critical TVD upon landing BW-1, the new well plan required at least one of Shell Brasil s BC-10 rules for horizontal producers be broken. The well needed to be drilled uphill at an inclination above 90º. The azimuth was also changed to steer farther from 1-SHEL-01 and to stay in softer amplitudes. To maintain an opportunity for a successful well, other design elements of the well path were maintained, such as minimal changes in azimuth and dogleg severities less than 2º/30 m. 6
The revised well plan for BW-2 is also shown in Figure 8. The net-to-gross estimate for the new well path was estimated at 80%, above the original expected range of 50% to 75%. Figure 9 illustrates this model in the real-time geosteering software. Figure 9. Revised real-time geosteering software model for BW-2 created by correlating shale events in BW-1 along the seismic well intersection. The blue and red dots indicate the simulated bed boundary mapping tool response for detecting upper and lower conductive boundaries, respectively. During the actual drilling operations, shale was indeed met where it was expected based on the model, but it was not all shale. Some sands were also encountered within the hard amplitude interval and helped to increase the netto-gross. Figure 10 highlights this section of the well. Farther into the section, within a region containing soft amplitudes, a significant section of shale was encountered. Only a slight decrease in the soft amplitudes suggested shale could be within this region. Figure 11 highlights this section of the well, which subsequently led to a reduction of the overall net-to-gross estimates. Figure 10. The expected shale section was less than expected as the interval also contained sections of reservoir sand. The resistivity log is overlaid in 3D along the wellbore. Figure 11. The unexpected shale section encountered in the later part of the well. The resistivity log is overlaid in 3D along the wellbore. The overall net-to-gross for BW-2 was 74%, slightly short of the revised expectation. Nevertheless, it was 50% more than deemed possible, with the initial after landing well plans. Figure 12 illustrates the final geological model of the reservoir along the well path, as interpreted from seismic data, formation dips derived from laterolog resistivity images, and the bed boundary mapping tool distance-to-bed boundary inversions. 7
Figure 12. Final real-time geosteering software model for BW-2 as defined by the bed-boundary mapping tool s distance-to-bed boundary information, imaging-while-drilling image and dip interpretation, and seismic data. 5. Summary and Conclusions Through the use of proper planning techniques and a thorough examination of all the available data, optimized well placement can indeed begin in the initial phase of any drilling campaign. Through interactive trajectory adjustments using a bed boundary mapping tool and imaging-while-drilling data, the net-to-gross estimate encountered in BW-1 was within the expectation range. During the planning phase for BW-2, the well plan was redesigned based on the log data acquired in BW-1. A major shale section encountered in BW-1 was correlated using seismic data and raised concerns that the initially planned well path for BW-2 could intersect up to 50% shale. As a result, the trajectory was revised beyond the traditional drilling practices and offered improved net-to-gross expectations from the well. In the end, not only was the drilling phase a success, but also the completion phase, which saw a full gravel pack at inclinations as high as 92.5º. In the case of Argonauta B-West, traditional drilling practices for horizontal wells required that the well s inclination not exceed 90º, the dogleg severity not be greater than 2º/30 m, and changes in azimuth be minimized. By exceeding just one of these criteria, an estimated 50% increase in net-to-gross expectations was accomplished. As drilling and completions technologies continue to advance to keep pace with today s drilling challenges, simple and small changes in traditional mindsets can lead to the realization for even better overall well performance. 7. Acknowledgements The authors would like to thank Shell Brasil Ltda, Petrobras and ONGC for permission to publish this paper. Special thanks also go to Santiago Zambrano of Schlumberger Drilling & Measurements for his cooperation with the Well Placement Team in searching for better well design solutions. 8. References AYODELE, O. R. Optimization of Well Placement and/or Borehole Trajectory for Minimum Drilling Cost (A Critical Review of Field Case Studies). Paper presented at the Petroleum Society s 5th Canadian International Petroleum Conference, Calgary, Alberta, Canada, June 2004. McLENNAN, J. A. Well Plan Optimization in the Presence of Uncertainty. Paper presented at the Petroleum Society s 7th Canadian Petroleum Conference, Calgary, Alberta, Canada, June 2006. MOTTAHEDEH, R. Horizontal Well Geo-Navigation: Planning, Monitoring, and Geosteering. Paper prepared for presentation at the Petroleum Society s 6th Canadian International Petroleum Conference, Calgary, Alberta, Canada, June 2005. STOCKWELL, L., vankonijnenburg, J. H., HOLMES, G., STEWART, G., ZAMBRANO, S. Parque das Conchas (BC-10): Subsurface Challenges in Developing a Deepwater Shallow Geologically Complex Field. Paper prepared for presentation at the Offshore Technology Conference, Houston, Texas, USA, May 2010. 8