SPE 63281 Maximizing Profitability in Reservoirs Using New Technologies For Continuous Downhole Pressure Systems James L. Buchwalter, SPE, Gemini Solutions, Inc., and Ray E. Calvert, Gemini Solutions, and Colin S. McKay, Wood Group, and Stephen J. Thompson, Wood Group Copyright 2000, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 2000 SPE Annual Technical Conference and Exhibition held in Dallas, Texas, 1–4 October 2000. This paper was selected for presentation by an 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 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 Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. 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 Continuous down hole data in conjunction with new reservoir analysis tools made to work with this data have the potential to revolutionize the accuracy of reservoir management. The economic value of continuous down hole pressure data and the array of available options justify the use of these systems in almost all petroleum reservoir developments. The greatest value of these gauges is that with new data analysis tools reservoirs can be accurately managed early in the producing life, thus optimizing both short and long term reservoir management strategies. Traditionally, the value of these systems has been for completion optimization using a small subset of downhole data. The full value of the complete data stream has been ignored due to the large volumes of data, and the lack of software systems for efficiently working with these data. Consequently the reservoir has not been fully understood. A system of software tools has been developed to capture the full value of the data from permanent down hole gauges. This new software system automates the filtering of these data in an intelligent fashion. The resulting filtered pressure data can then input into a variety of reservoir analysis tools, for example reservoir simulation programs can now have a continuous reservoir simulation. Reservoir and production engineers can always have the optimal production strategy for the reservoir based on the current data. These tools are very easy to use, so models can be developed quickly and continually maintained with minimal effort. Typically it takes less than a week to build the initial model, and only a few hours a month to update and maintain an accurate history match. The paper will include a brief review of the down hole technologies and the software system which makes the data accessible to reservoir simulation and other reservoir analysis tools. Applications for the filtered data in both Gulf of Mexico and North Sea reservoirs will be introduced. Description of the Data Analysis System The data analysis system used to perform these studies consists of a data filtering system co-developed by Wood Group and Gemini Solution Inc. This was used in conjunction with Gemini’s commercial PC simulation and mapping software- Merlin and Apprentice. Apprentice is a digitizing, geostatistic characterization, mapping package that can be used to generate maps, calculate volumetrics, and export grid data to various simulators. Merlin is a full-featured black-oil simulator with pre and post processors, and a transparent interface to Apprentice. Both packages run on the PC, and can effectively build models of 100000 cells or more on a high end Pentium PC. The filtering package combined with Merlin and Apprentice is called Prophet. Prophet provides operators the means to efficiently manage down hole data sets of any size, and then quickly build and maintain accurate continuous reservoir simulations in a “real time” environment for reservoir management. The filtering system is needed because pressures may be measured on a second by second interval, but only a small subset of these pressures are normally necessary to correctly characterize the reservoir. On the other hand, at selected times such as during a buildup or well choke change pressures are needed on a finer interval for proper reservoir analysis. In addition, most operators would embrace a system that can automatically update to include new pressure data, or even new wells that come on line. The Prophet filtering system provides tools for all of the needs and much more. 2 JAMES L. BUCHWALTER The Prophet filtering system works as follows. At the onset of production an operator creates a project that will contain a series of user defined filter files for all well in the reservoir. Examples of possible filters might include a pressure point per day, and a pressure point every 30 minutes unless the pressure changes by more than 5 psi during the 30 minutes. The initial filter files are created to allow the user to start viewing the data on different scales. After the filter files are created, the remaining pressure points can be smoothed, deleted, as added in more detail in selected intervals. The “add data” feature is particularly useful after the data sets become large. For example, after ten years a project will have read tens of millions of pressure points to create the filtered files. Since the project is typically updated weekly or monthly the time for each incremental update was small. Suppose however that after ten years it is realized that it would have been beneficial to add detailed pressure data within a two day interval in year 5. Recreating the entire data stream from scratch using a new filter might be an overnight process. Prophet allows the user to selectively add data only within the interval of interest using new filter constraints. As a result, instead of reading 10 years of data to recreate a new filter file, only the selected days of interest are read, and automatically inserted into the existing filter file. This is of particular usefulness in reservoir simulation where the entire data stream is of value. Each time a user opens the project any new dates which have pressure data – either earlier or later, are automatically imported. The update can easily be self automated - for example an operator may choose to run the project update at midnight daily or weekly so that the filtered data files are available for review at the commencement of the next business day. The other components of the Prophet system include an easy to use PC simulation and mapping package which allows models to be quickly built and maintained by typical reservoir and production engineers. In the past, due to the high cost associated with simulation, only the largest reservoirs were simulated. The costs were high due to three main reasons. First, Unix workstations and the geological and reservoir simulation software that run in the Unix environment are expensive to purchase. Second, Unix based hardware and software typically require a significant amount of support to keep the systems up and running. Third, since the interfaces between the user and the calculation modules do not allow one to rapidly build the initial model, run the cases and analyze the results, companies incur the significant expense of training and maintaining simulation experts. The first two reasons can be mitigated by shifting to the PC environment since high-end PCs now have the power to run large models in a reasonable time period. Both the PC itself and the geological and reservoir simulation software that run on the PC are far less expensive to purchase and support than the Unix workstation-based alternatives. The third reason for SPE 63281 expensive simulation studies, the lack of first-rate interfaces that will significantly lower the man-hours and experience required to complete a study, is an issue that has only recently been addressed satisfactorily by systems such as the one described here. Today production and reservoir engineers can use models built on PC’s to manage their assets in real time, and the right answer for the data collected any date is always available. Data Filtering Examples Let’s look briefly at some sample raw data sets that were filtered with the Prophet system to see the power of the filter tool. The filtered pressure files below were built from a raw pressure stream of about 120,000 pressures collected over 8 days. The total time required to import this data for multiple filters took less than a minute. About 5 minutes of additional time was required to clean up the data stream using the tools described below. The raw data source for all examples were quartz gauges. The first data set we will examine is from a well that mistakenly had both good and bad pressures simultaneously returned from the down hole quartz gauge. The filtered data corresponds to a 30 minute 5 psi filter (Fig. 1). Keep in mind that the original raw pressure file for this well contains about 120,000 entries. After applying a box delete tool the original filter file was reduced to a filtered file (Fig. 2). This data stream proved correct and was acceptable for input into other reservoir analysis systems. The data can be saved in a variety of formats, either in whole or within a user defined range. In a second example, data filtering, smoothing, and insertion within an interval will be illustrated. The original filtered data file was created using a 30 minute and 5 psi pressure constraint (Fig. 3). As a first step the erroneous pressures at the bottom of the plot were removed (Fig 4). Next the spike was examined in closer detail (Fig. 5). It was unclear at first glance whether the pressure spike was a result of a partial buildup or just some additional bad data. To study the spike in more detail, additional data was added to the spike interval corresponding to a 5 minute and 2 psi filter using the original quartz pressure files (Fig 6). The data lies almost at identical times and represents a 26 psi spike – adding these additional data showed the operator that this was also bad data and the spike was deleted (Fig. 7). The last tool applied to the data was a time weighted pressure smoothing algorithm to reduce the entire data stream to less than 60 points (Fig. 8). In summary a 2500 fold decrease in SPE 63281 MAXIMIZING PROFITABILITY IN RESERVOIRS USING TECHNOLOGIES FOR CONTINUOUS DOWNHOLE PRESSURE SYSTEMS the number of data points was accomplished in this example without an appreciable loss in the data character. These tools allow large complex down hole pressure (as well as other data streams) to be filtered quickly yield a continuous data stream with variable sample rates suitable for capturing all important events surrounding data changes. Application - Determining Gas Reservoir Size With Limited Down Hole Pressure Data Introduction. Historically it can take several years to have a reasonable description of the reservoir, in part because down hole pressures are typically measured yearly at best. The following application shows how a limited data stream of less than 90 days was used to accurately determine the size of a gas reservoir. Rapidly determining the reservoir size can allow operators to use this knowledge to significantly impact early life development decisions to maximize both production and economics, and minimize the need to use sketchy exploration data. In this case the operator was able to accurately and quickly determine the reserves around the well to be able to optimize the development strategy. In the field case presented, a down hole bottom gauge was run in the first well completion string right above the perforation depth. The gauge is a SDG digital 16K gauges from Wood Group. The reading of the gauges are accomplished through a computer on the platform. The local PC stores the data on a continuous basis and, through intranet, sends a daily file to the company offices. Instantaneous readouts of the data are also accessible from the office. Daily files are processed by the filtering software to visualize trend and when needed, transferred data to simulation software. A team effort was used to study the reservoir, with both the operator and Gemini Solutions Inc. building models of the reservoir. The operator later reconfirmed results using a commercial well test analysis package. Methodology for Determining Reserves. The following methodology was used for developing an accurate reservoir description and development strategy. 1. A simulation model was built based on static information available including seismic interpretations, logs, and cores. 2. The initial simulation model was used to create a production forecast. 3. The well was produced for 60 days and the model forecast was compared to the actual producing history. 4. The possible parameters to be changed in the model in order to match the recorded production were identified 3 5. A history match was accomplished after modifying the identified parameters and integrating the recorded production and filtered bottom hole pressure data. 6. The corrected simulation model was used to re-estimate the original hydrocarbons in place and develop an optimal production strategy before rig demobilization. Model Description. The reservoir shown is a gas condensate reservoir in the Gulf of Mexico lying deeper than 8000 feet (Fig. 9). The reservoir was originally delineated using seismic data, and the exploration well number 2 tested more than 23 MMCF/D. Two other offset wells shown on the maps were drilled after the discovery well, but the resolution of the seismic was unable to positively determine the extent of the reservoir limits around well number 2. With this uncertainty the original gas in place in the reservoir and an accurate development plan could not be accurately determined. History Match Procedure. One of the principal difficulties in analyzing continuous pressure data streams is the fact that while flowing bottom hole pressures are available on a second by second basis, well flow rates at best are probably known on a daily average basis. Consequently, if in the model, rate specified wells are used to control the wells then a technique is needed to calculate average daily pressures corresponding to these rates. This is very difficult. In addition, even if a satisfactory average pressure technique were devised – much of the important analysis ability for the continuous data stream would be lost due to data averaging. Prophet has a variety of averaging different filters – but only pressure and time interval filters were applied to the original data for this project. Averaging filters were not applied to the resulting data subset. Step 1 – Radial Model for Estimating Permeability and Skin. A simple radial model was constructed and the longest buildup period was modeled to estimate the approximate permeability and skin around the well. The resulting history match is shown in Fig. 10. The match above showed a permeability of 10 MD and a skin of 2. These numbers were input as starting values for the full field model. An exact match is not necessary at this stage because further adjustments in these parameters were made in the full field model. Step 2 – Full Field Model and Rate Specified Wells. In the next step a full field model was constructed using seismic derived structure and net sand maps, and the permeability and skin estimated in step 1. In order to obtain an acceptable history match boundaries were placed on both sides of the producing well approximately 900 feet from the well. The 4 JAMES L. BUCHWALTER resulting boundaries superimposed on a net sand map around the well are shown in Fig 11. The resulting gas in place within the area in communication with the well is approximately 6.2 BCF. It is now believed that additional prospective area around the well identified by seismic is nonproductive. The bottom hole pressure history match for the well is shown in Fig. 12 and Fig. 13. The model results are shown as a line, and the filtered historical data are shown as squares. The flowing pressures can never be exactly matched with rate specified wells because the measured well rates are daily averages, and the true rate history variation is much greater as evidenced by the variability in the pressure history. Nevertheless this match was suitable for concluding that the gas in place and reservoir permeability and skin were approximately correct. As this was a gas condensate reservoir the condensate yield curve was slightly adjusted to match the recorded condensate history. This curve is shown in Figure 14. Step 3 – Full Field Model and Pressure Specified Wells. As a confirmation for the model constructed in step 2 a pressure specified well type was assigned to each startup, and a very detailed rate history was predicted from the filtered subset of the recorded bottom hole pressure data. The skin on the well was slightly adjusted from 2 to 3 until the cumulative gas production curve was exactly matched. The resulting model is capable of forecasting the well rates on a continuous basis using the recorded bottom hole pressures. The pressure and production matches are shown in Figure 15. The adjusted model was used to forecast an accurate production profile for the reservoir, and a workover strategy for completions in the future. Accurately Characterizing an Oil Reservoir Using Limited Down Hole Pressure Data A depletion drive sandstone oil horizon with one producing well and about 215 days of production and downhole gauge pressure history is the next case study. The reservoir has reasonable dip and the primary producing mechanism is thought to be fluid expansion aided by rock compaction particularly during the early producing life. The goals of this study were to match the first 115 days of production and pressure history to establish the hydrocarbons in place and accurate reservoir description. Details for this study could not be released at the time of this printing due to confidentiality so only a general overview of the study will be presented. The reservoir in question is thought to have variable rock compressibility, and an unknown contribution of rock compressibility on permeability reduction. Reservoir size is SPE 63281 not certain – seismic data interpretation does not have the necessary resolution to precisely define the reservoir limits. The unknown variables made the history match technique uncertain, and two opposite sets of assumptions with respect to compressibility were used to attempt the history match. The goal for this study was not necessarily to obtain a unique history match, but rather, understand the error bars based on the current knowledge available, and develop a successful development strategy that would work for either reservoir description. The production and pressure histories showed more than 7 partial buildups, and 2 complete buildups over the first 215 days of production. Both sets proved very important. The partial buildups allowed an understanding of the general pressure trend, and the complete buildups help characterize permeability and its change as reservoir pressure declined and pore volume changed due to rock compressibility. The first history match assumed constant permeability and compressibility, and the second history match assumed permeability and compressibility versus pressure relationships. Acceptable history matches using both assumptions were obtained, and the resulting original oil in place and recoverable hydrocarbons were remarkably close for the two techniques in spite of the different history match techniques. Most importantly, the limited down hole pressure data completely changed the operator’s view of the reservoir and the forecast for future performance. For example a several fold change in the original oil in place was determined, and confirmed by a new seismic reinterpretation. Without the continuous pressure data it may have taken years to accomplish the detailed analysis performed here. Accurately Characterizing an Water Drive Gas Reservoir Using Limited Down Hole Pressure Data Another application that is of particular importance around the world is correctly characterizing the influence of water influx in gas reservoirs. Traditional techniques for identifying water influx such as material balance require very accurate average bottom hole pressure histories. Since in the typical reservoir bottom hole pressures may only be measured once every 12 to 18 months, and 4 or more average pressures at different stages of production are needed before the water influx can be characterized, the time required for identifying water influx in most gas reservoirs is 4 or more years. Unfortunately this delay in identifying water influx can result in the irreversible loss of reserves that lie behind in the water invaded gas zone. Outrunning the water by accelerating gas production can improve the water drive gas recovery factors by 5-30% in these reservoirs. This is accomplished by accelerating production from existing wells, and if economics justify, drilling additional wells. In this example the benefit of real time simulation was seen upon examining a continuous pressure data stream that had SPE 63281 MAXIMIZING PROFITABILITY IN RESERVOIRS USING TECHNOLOGIES FOR CONTINUOUS DOWNHOLE PRESSURE SYSTEMS been collected over a 10-year period (Fig 16). The operator had no efficient tool for examining the pressures as a continuous data stream, and for the first 8 years of production used short segments of the data for well optimization and well test analysis. The reservoir was thought to be a depletion drive gas reservoir. The reservoir was produced for about 9 months of the year, and shut in for the remaining 3 months due to insufficient gas demand. After 8 years the entire data stream was evaluated and from the fact that the shut in bottom hole pressures built up at a rate of 2-3 psi per week along with further analysis it was determined that the reservoir was waterdrive. The operator immediately changed the operating procedure accelerating gas production to outrun the water influx, which significantly increased the recoverable reserves. Interestingly, the water influx could have been detected almost 9 years earlier, where the benefit would have been much greater. With the Prophet system, data streams like this are now analyzed in less than 30 minutes of man effort. Overview of Down Hole Pressure Gauge Technology Permanent continuous downhole reservoir monitoring systems have only been applied to a small number of the world’s oil and gas wells. Cost, reliability, risk, environmental limitations, arduous data evaluation processes and the lack of awareness have limited the widespread use of these technologies. New technologies for deployment, data management and interpretation have lowered the cost of the entire well monitoring process. The various technologies available will be summarized below. Electronic Systems. The vast majority of downhole pressure and temperature systems installed today are electronic systems. The early electronic systems utilized strain gauge technology for pressure measurement; this technology had limited resolution and suffered from inherent drift problems over time at pressure and temperature. A number of other sensor technologies were tried and tested including quartz capacitance sensors. Today resonating quartz, which utilizes the inverse piezoelectric effect to induce the resonator to vibrate at its mechanical resonant frequency, has been the primary sensor technology utilized in downhole pressure measurement. This technology provides the highest accuracy and resolution of pressure measurement. In recent years improvements in the manufacturing process have improved both the reliability and ruggedness of the sensors. In addition, market demands have resulted in reduced diameter and higher pressure and temperature rated sensors. In addition the running of dual pressure sensors has the benefit of increased reliability over time. The redundant sensor can be used to verify readings or provide the bottom hole pressure in the event that one of the sensors becomes problematic Non–Electronic Systems. High temperature operation and long-term reliability of permanent downhole monitoring systems have been improved by using sensors that do not 5 require any downhole electronics. The two non-electronic technologies currently available are Fiber Optic and Electrical Resonating Diaphragm. Fiber Optic. The initial Fiber Optic Well Monitoring (FOWM) system was developed from a Joint Industry Project (JIP) specifically to address high temperature operation and long-term reliability. FOWM sensors are based on the micromachined resonator concept with the structure being micromachined in silicon. The pressure or temperature sensor is a resonator chip that is mounted in a silicon tube. Changes in pressure and temperature induce forces in the sensor structure that modify its natural resonant frequency. The sensor is so small that it can be brought into vibration by shining it with light of the correct wavelength and frequency. By exciting the sensor with a modulating laser, the resonance frequency can be measured by a second laser beam that is reflected back to surface along the fiber. The second laser beam is then measured on the surface to determine the resonant frequency of the sensor. This frequency is then used to derive the pressure or temperature surrounding the sensor. Electrical Resonating Diaphragm. Electrical resonating diaphragm sensors have been in commercial use over the last six years2. They have reliably operated up to 390 degrees Fahrenheit 2. The high reliability and temperature rating is accomplished by having the sensing electronics at surface rather than downhole. This technology uses a downhole electrical resonating diaphragm that is excited by an electrical pulse from the surface. The diaphragm then returns a signal related to the pressure or temperature of the downhole environment. A surface acquisition system then measures the responding signal, which is then converted to pressure or temperature. This system generally consists of two pressures and one temperature sensor. Each sensor requires a separate pair of electrical conductors back to the surface acquisition unit. Additional sensors would require additional conductor pairs. In order to accomplish this a standard openhole 7-conductor wireline is currently used to transmit the electrical signals. Success using this wireline can be attributed to the fact that constant voltage is not required to power downhole electronics. This eliminates the potential across the electrical insulation and therefore minimizes the problems associated with electrical leakage. Conclusions Early characterization of the reservoir with continuous down hole pressure data in conjunction with easy to use PC simulation software offers the potential to revolutionize the way reservoirs are developed and managed in the future. In 6 JAMES L. BUCHWALTER SPE 63281 fact, recent developments in these technologies make these systems economical for the majority of the wells drilled in the world today. Historically and even today this is not largely recognized because filtering and data analysis systems that can make real time reservoir analysis a reality are just entering the market place, and not recognized by most operators. The real benefits derived from these technologies lie not so much is applying down hole pressure data in the conventional work flow for reservoirs that have been developed over the past 40 years, but changing the work flow process to take advantage of the accurate reservoir descriptions that can be obtained early in the life, and using this knowledge to minimize costs and maximize production for reservoirs. With this knowledge it becomes difficult to imagine where this technology is not applicable. The case studies discussed here are representative of typical reservoirs around the world, and the potential value just in these reservoirs totals many millions of dollars. Acknowledgements We thank Mike Flecker formerly of Wood Group for his vision that helped make Prophet possible, and for his many important contributions to the development of the Prophet system. References 1. W.G. Hazlett, SPE, and J.L. Buchwalter, SPE, and R.E. Calvert, Gemini Solutions, Inc., and T.M. Campbell, SPE, and R.A. Molohon, SPE, Mariner Energy, Inc.: “TimeCritical Decision Making Using PC-Based Reservoir Simulation,” paper SPE 53980 presented at the 1999 SPE Latin American and Caribbean Petroleum Engineering Conference held in Caracas, Venezuela, 21–23 April 1999. 2. Michael J. Flecker/Wood Group, Stephen J. Thompson/Wood Group, Colin S. McKay/Wood Group, James L. Buchwalter/Gemini Solutions, Inc: “Maximizing Reservoir Production Using New Technologies for Permanent Continuous Downhole Sensors,” paper OTC 12153 presented at the 2000 Offshore Technology Conference held in Houston, Texas, 1–4 May 2000. Fig. 1–Well A8 raw filtered data stream using 30 minute and 5 psi filter. . Fig. 2–Well A8 corrected data stream for 30 minute and 5 psi filter SPE 63281 MAXIMIZING PROFITABILITY IN RESERVOIRS USING TECHNOLOGIES FOR CONTINUOUS DOWNHOLE PRESSURE SYSTEMS 7 Fig. 3–Well A11 raw data stream for 30 minute and 5 psi filter. Fig. 5–Well A11 showing the pressure spike Fig. 4 – Well A11 corrected data stream for 30 minute and 5 psi filter. Fig. 6–Well A11 with additional data added to the pressure spike 8 JAMES L. BUCHWALTER Fig. 7–Well A11 after correctly filtering out the pressure spike Fig. 8–Well A11 after smoothing the corrected data SPE 63281 Fig. 9–Gas reservoir structure map producing from well 2 Fig. 10 – Gas reservoir pressure buildup match SPE 63281 MAXIMIZING PROFITABILITY IN RESERVOIRS USING TECHNOLOGIES FOR CONTINUOUS DOWNHOLE PRESSURE SYSTEMS Fig. 11–Gas reservoir boundaries needed to match historical data Fig.12–Gas reservoir flowing bottom hole pressure match using rate specified wells 9 Fig. 13–Gas Reservoir pressure history match using rate specified wells. Fig.14–Gas Reservoir condensate yield curve needed for history match. 10 JAMES L. BUCHWALTER Fig. 15–Gas reservoir history match with pressure specified wells P erm an en t D o w n h o le P ressu re D ata 2700 2450 B HP (psi) 2200 1950 1700 1450 1200 950 700 D ec-99 Ju n -99 D ec-98 Ju n -98 D ec-97 Ju n -97 D ec-96 Ju n -96 D ec-95 Ju n -95 D ec-94 Ju n -94 D ec-93 Ju n -93 D ec-92 Ju n -92 D ec-91 Ju n -91 D ec-90 Ju n -90 D ec-89 Ju n -89 D ec-88 D a te Fig 16- Water driver gas reservoir continuous downhole pressure data stream SPE 63281