Click here to load reader
Upload
luis-alberto-izarra
View
5
Download
1
Embed Size (px)
Citation preview
Copyright 2003, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Denver, Colorado, U.S.A., 5 – 8 October 2003. 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 Modern reservoir characterization-simulation studies integrate geology, seismic, petrophysics and engineering to build three-dimensional (3D) static geologic and dynamic fluid flow models. These models have repeatedly proven their worth through reduced risk and improved reservoir management decisions. Prior to the advent of modern computers and software, traditional reservoir engineering techniques served the industry well through analytic and graphical analysis techniques. It would seem that the modern characterization tools would make these traditional reservoir engineering techniques obsolete; however, this paper shows that integrating these traditional techniques can improve the quality of modern 3D models.
This paper describes a field example where traditional reservoir engineering techniques from 40 years of waterflooding history were integrated into the modern reservoir characterization-simulation workflow. Introduction The Beaver Lodge Devonian Unit (BLDU) is located in the Williston Basin, North Dakota. The productive dolomite is located at an average depth of 10,125 feet. Production of the 42oAPI oil began in 1957 on depletion drive and waterflooding was initiated in 1963 following unitization in 1962. There are currently 34 producing oil wells and 24 injection wells in an inverted 9-spot pattern. Cumulative production is 62 MMSTBO.
An integrated reservoir characterization and simulation study was completed to maximize the future recovery and profitability. The study was based on a non-linear iterative workflow that integrates the engineering, geology and seismic. Figure 1 shows an example of this type of workflow.1 The first phase of the reservoir simulation portion of the workflow from Figure 1 is an initial engineering review. The first section of this paper describes how traditional waterflood diagnostic plots were
used to improve the study results. The second section shows how injection well profile logs and produced water salinity were used to improve the history match. These are non-traditional history match parameters that are often commonly available in mature waterfloods but are seldom utilized. The last section demonstrates how the simulation results were presented to confirm infill well locations and workover candidates. Traditional Waterflood Diagnostic Plots There are many variations on traditional analytic waterflood reservoir engineering diagnostics. The technique used for this study uses water-oil ratio (WOR) versus cumulative oil plots to analyze production wells and Hall plots2 to analyze injection wells. Figure 2 shows examples of these plots.3 Production Wells. The example WOR versus cumulative oil plot on the left side of Figure 2 shows three example curves. Curve A represents an inefficient flood with early water breakthrough. Curve B represents normal waterflood performance and curve C represents an excellent waterflood. Line A’ illustrates what happens when a water filled fracture or thief zone in the well is corrected and a poorly performing well (curve A) shifts to a normal waterflood (curve B). Curve B’ shows a normal waterflood switching over to an excellent waterflood because an EOR project.
WOR versus cumulative oil plots were created for all BLDU producers. Three main types of well behavior were recognized, 1. Flat WOR’s, 2. Steadily increasing WOR’s and 3). Rapidly increasing WOR’s.
Wells with flat WOR’s. In addition to the individual well WOR diagnostics, simple, single well; conceptual models were used to match specific wells’ behavior. Figure 3 shows the theoretical WOR behavior of a single layer homogeneous well using BLDU fluid and rock properties. The theoretical WOR in Figure 3 shows that a single layer model will produce virtually no water until a recovery factor of about 35% and then experience a relatively rapid increase in WOR, reaching a WOR of 10 at a recovery factor of 54%.
Figure 4 shows a field example of a well with a flat WOR along with a conceptual model match. In order to have an extended period of fairly constant WOR as seen in Figure 4, there must be at least two layers, one layer with producing essentially 100% oil and the other producing mostly water. A thin, high permeability layer was necessary to match the single well conceptual model for this well. The high perm thief zone
SPE 84414
Case Study: Merging Modern Reservoir Characterization with Traditional Reservoir Engineering Brian Rothkopf, SPE, iReservoir.com; Steve Fredrickson, SPE, Amerada Hess Corp.; and Jeff Hermann, SPE, Amerada Hess Corp.
2 SPE 84414
is 1/19th the thickness and 15 times the permeability of the main reservoir layer. In conclusion, wells with long periods of flat, low WOR’s most likely have an oil productive layer and a separate, thin, high permeability water productive layer. Wells with steadily increasing WOR’s. Well F307 is an example of a well with a steadily increasing WOR (Figure 5). This well’s performance is similar to the theoretical recovery from the homogeneous model in Figure 3. This well was also matched with a conceptual model with two layers having a three to two height ratio and a 7.5 to 1 permeability ratio. In summary, wells with steadily increasing WOR plots reflect layered geometry but not thief zones. Wells with rapidly increasing WOR’s. Well B307 is an example of a well with a rapidly increasing WOR (Figure 6). This type of behavior usually indicates a mechanical failure (casing or tubing leak) in the producer or offset injector. Wells with this type are difficult to history match in the full field model. It is beneficial to identify these wells early in the simulation study so that the well history can be reviewed to determine the likely cause of the erratic behavior. Injection Wells. Injection wells were evaluated using Hall plots. Hall plots use rectangular coordinates with cumulative water injection on the x-axis and cumulative pressure on the y-axis. Cumulative pressure is the cumulative sum of the product of the injection pressure and time in days. The example hall plot on the right side of Figure 2 from reference 2 shows three typical curves. Curve A with a concave up shape indicates increasing resistance to injection, sometimes an indication of poor water quality or insufficient filtering causing plugging. Curve B is for a stable or normal injector and curve C is for a well allowing water to channel or move easily to a producer. Curve B’ shows a normal injector that suddenly has severe water breakthrough to a producer, fractures into a high perm channel or some other loss of effectiveness.
Hall plots were created for all BLDU injectors (Figure 7). The wells were placed into four groups according to the slope, ranging from steep (high resistance to injection) to shallow (low resistance to injection). Well Types. Mapping the different production and injection well types and looking for patterns best analyze results of the waterflood diagnostic plots. Figure 8 shows the production and injection wells divided into seven different well codes based on the plots discussed in the previous sections. Injection well codes were type 1, steep slope Hall plot; type 2, medium- steep Hall plot; type 3, medium slope Hall plot; and type 4, shallow slope Hall plot. Production wells with flat WOR curves were well type 5, steadily increasing WOR’s were well type 6 and rapidly increasing WOR wells were well type 7. One pattern evident from Figure 8 is that the injection wells along the southern edge and southwest flank all have steep slope Hall plots (type 1). This may be a result of the geology (poor rock quality at the edge of the field) or because these edge injection wells do not have offset production wells. Most of the interior pattern injectors have medium or normal Hall plot slopes although several wells have very shallow slopes. The shallow slope implies a type “C” well from Figure 2 that allows water to channel too easily to offset producers. Most of these very shallow slope injection wells line up in a northeast
to southwest line across the field. In additional, a review of the production wells’ Productivity Index (PI) values showed anomalously high PI’s along the same trend. In conclusion, the well performance data indicated a northeast to southwest linear feature of high permeability or fracturing. Integration of Well Types into Seismic Interpretation. The map of well types (Figure 8) was compared to the geologic and seismic data. Figure 9 shows the injection well Hall plot types overlain over the seismic semblance. There is a very good correlation between the shallow slope Hall plots (least resistance to injection) and the dark areas of semblance (least reflectivity, most fractured). This provides a geologic confirmation of the high permeability or fracture trend found in the production data analysis. This gives confidence in the seismic interpretation and allows predicting the high injectivity areas away from well control.
Non-Traditional History Match Parameters A traditional simulation study is history matched by fixing the historic oil and water injection rates and history matching by comparing the simulation predicted water rates and reservoir pressures to actual field measured values. In this study, two additional non-traditional history match parameters, injection well profiles and produced water salinity were used. Injection Well Profiles. Injection well profile logs (PLT’s) are a common diagnostic tool in waterflood projects. The water injection rate versus depth in an injection well is measured using a logging tool with a combination of spinners, radioactive check shots and temperature. The full feature simulator used in this study can create synthetic injection well profiles that can be compared to the field measured PLT’s. Extracting the individual layer injection rates can also create a synthetic PLT at the date of the field measured PLT. In the BLDU, total of 134 individual injection profiles were available from 21 injection wells. The measured layer water injection rates from the injection profile log was compared to the model layer injection rates. This was used to calibrate the geologic model and improve the overall history match. Figure 10 is an example of a good match between the simulator and the field measured log. Figure 11 is an example of a poor match because there is a zone at 8070 that showed high injectivity in the field measured PLT but does not show up in the simulator PLT. In cases like this, the well logs were reviewed to determine why the simulator did not match the measured PLT. In several cases, the high injectivity layer seen in the measured PLT’s correlated to a washout in the open-hole logs. The petrophysical evaluation software skipped the high porosity and high permeability in these zones because it was tagged as “bad hole”. In order to correct this oversight, the logs were hand-edited to correct for the missed zones. The 3D geologic model was updated, the simulator was rerun and the history match was improved. Produced Water Salinity. The BLDU formation water salinity is very high (320,000 ppm) while the waterflood injection water averages 7200 ppm. Modern simulators have a salinity or brine option to account for the different properties between the formation and injection water. This option is
SPE 84414 3
important in BLDU because the water viscosity varies from 0.60 cp for the formation water to 0.27 cp for the injection water. In addition to being important for obtaining the correct mobility, the brine option also allows the produced water salinity to be calculated and compared to the field measured salinity. Figure 12 shows “bubble” maps of the produced water salinity measured in the field and calculated in the full field model history match. In general, the trends of high and low salinity (large and small circles) matches well. The match is not rigorous because the field history has had numerous periods when produced water of intermediate salinity was reinjected whereas the model assumes that all injection water is constant salinity. Results Figures 13 through 14 show the traditional history match plots for the total field oil and water rate and cumulative production. The field has stayed above the bubble point for most of the history so the gas production is simply solution gas and therefore not a good history match test. Figure 15 shows the field pressure match. The individual match is also good for most wells as shown in the example of Figure 16.
Mobile oil-in-place maps were used to determine the best areas for future development. Traditional maps of remaining oil volumes can be deceiving since large areas of low oil saturations in high porosity swept zones can skew the results. This can be particularly problematic in mature waterfloods. One solution is to calculate mobile oil saturations by subtracting the residual oil to water saturation (Sorw). The mobile oil saturation can then be converted to mobile stock tank barrels per acre by adjusting for the simulator cell volume, area and the oil formation volume factor. Figure 17 shows these maps at the model initialization, at the end of the history match, and at the end of the future predictions. These maps can be used to find areas that could justify additional development. Figure 18 shows a cross section of remaining oil saturation that illustrate how workover/infill locations can be chosen after areas of interest are found from the maps of Figure 17. Conclusions The techniques described in this paper can be used to improve the quality of a 3D reservoir characterization-simulation model, history match and future development predictions. 1. This paper demonstrates how production data (production
well WOR plots and injection well Hall plots) can be integrated into the seismic semblance to improve the quality of a 3D geologic model.
2. This paper illustrates how to use water injection profile logs to flow test a 3D geologic/simulation model.
3. This paper demonstrates the use of produced water salinity in a full field model history match.
4. This paper demonstrates the use of remaining mobile oil volume maps to identify future development prospects.
Acknowledgement The authors would like to thank Amerada Hess Corporation for permission to publish this paper. We would also like to thank the iReservoir.com and Amerada Hess employees that contributed significant time and ideas to this project. Thanks
to Wayne Biberdorf, Jim Gilman, Hai-Zui Meng, Virgil Miller, Michele Simon and Brad Watts. References
1. Campanella, J.D. Sonnenfeld, M.D., Zahm, L.C., Gilman, J.R., Conrad, L., Siemens, C., Zaitlain, B.A.: A Case Study: Using Modern Reservoir Characterization to Optimize Future Development of a Mature Asset”, SPE 77671 presented at the 2002 SPE Annual Technical Conference and Exhibition held in San Antonio, Texas, 29 September-2 October 2002.
2. Earlougher, R.C., Jr.: Advances in Well Test Analysis, Monograph Series, SPE (Dallas), (1977), 5, 86.
3. Sloat, B.F.: “Measuring Engineered Oil Recovery”, JPT (January 1991) 8.
Figure 1. Integrated study workflow example
Production Well Injection Well
WOR versus Cumulative Oil
Hall Plot
Figure 2. Traditional waterflood diagnostic plots
4 SPE 84414
Figure 3. Theoretical waterflood recovery from conceptual model
Figure 4. Field example and model match for a well with a flat WOR
Figure 5. Field example and model match for a well with a steadily increasing WOR
Figure 6. Field example and model match for a well with a rapidly increasing WOR
Figure 7. Hall plots for injection wells showing wells grouped according to the slope of curve
High PIHigh PI
High PIHigh PI
Figure 8. Map of showing well types from production well WOR plots and injection well Hall plots
SPE 84414 5
Semblance (reflection continuity): black=low, white=high
Figure 9. Seismic semblance with injection well Hall plots showing correlation between shallow slope (blue dots) and dark semblance.
Water Injection Rate, BWPD
Figure 10. Water injection profile log with good match in simulator
High injectivity interval left out of
model
Water Injection Rate, BWPD
Figure 11. Water injection profile log with poor match.
Measure salinity, ppm (1000), Jan-Feb 2003Fm wtr=320, Inj wtr=60
185
234
189
132
210
174
155
236
225
92
54
51
88
157
57
80
130
225
71
29
185
128
226
128
80
29
158
201181
116
212
171
207
114
249
459360
464640
469920
475200
480480
485760
491040
496320
501600
1380000 1385280 1390560 1395840 1401120 1406400 1411680
A B C D E F G H I
Field average= 117
303
305
304
306
307
308
309
310
311
312
313
314
315
Model salinity, ppm (1000), Jan 2002Fm wtr=320, Inj wtr=60
258
95
124
240
102
149
130
211
268
91
236
135
130
91
250
80
237
138
270
95
110
127
255
198
150
156
184
168
216290
175
157
137
318
125
176
294
459360
464640
469920
475200
480480
485760
491040
496320
501600
1380000 1385280 1390560 1395840 1401120 1406400 1411680
A B C D E F G H I
Field average= 158
303
305
304
306
307
308
309
310
311
312
313
314
315
Field Measured
Produced Water SalinityModel Predicted
Produced Water Salinity
Figure 12. Field measured and model predicted produced water salinity
Figure 13. Oil rate and cumulative history match
Figure 14. Water rate and cumulative history match
6 SPE 84414
Figure 15. Reservoir pressure history match
Water injection rate and cumulative
Oil rate and cumulative
Reservoir Pressure
Water rate and cumulative
Figure 16. Example of an individual well history match
Mobile Oil-in-Place, BO/Acre0 30,000
1975
Initialization
2030
End of Prediction
2003
Current
Figure 17. Mobile oil-in-place maps for initialization, current, and end of prediction
Perforations
Workover Candidates
WorkoverCandidates
Figure 18. Cross-section through field showing remaining oil saturation and possible workover candidates