Content of PetroWiki is intended for personal use only and to supplement, not replace, engineering judgment. SPE disclaims any and all liability for your use of such content. More information
Because a waterflood project spans several decades, it is monitored continuously and routinely by engineers who are responsible for its operations, as well as periodically using more-detailed and specialized technical studies (e.g., full-field numerical-reservoir-simulation studies). There are many opportunities to modify and improve the waterflood as data are acquired and analyzed.
The basics of a waterflood analysis center on material balance concepts. Applying material balance concepts means that initially there is "reservoir fill-up" if the reservoir previously had some years of primary production. During this period, the reservoir is repressured to its original reservoir pressure because the injected-water volumes will be substantially greater than the produced-fluid volumes. Thereafter, the waterflood will be operated as a voidage-replacement process.
The earliest waterflood monitoring techniques were developed soon after the first field applications of waterflooding; they were based on simple plots, maps, and calculations. Among these were the plots published by Dyes et al. that estimated the water breakthrough and post-breakthrough behavior of various waterflood pattern configurations. Figs. 1 and 2 are examples of these plots for a five-spot pattern and a direct-line-drive pattern, respectively, and can be used to make "first-estimate" waterflood calculations.
Fig. 1 – Effect of mobility ratio on oil production for five-spot pattern. (Ψs = fraction of total flow coming from the swept portion of the pattern.)
Routine data gathering
Waterflood monitoring begins with the acquisition of the routine data that are necessary for engineering calculations. The routine data include:
- Well-by-well daily oil-, gas-, and water-production rates
- Well-by-well water-injection rates
- Injection wellhead pressures
- Production-well pressure data
Often, the well-by-well daily rate data are back-calculated from the gathering center’s total produced volumes of these three phases, and then are allocated back to the individual wells on the basis of periodic individual well tests.
Next, the production- and injection-well data are allocated to the individual reservoir intervals, if multiple reservoir intervals are commingled. This well-by-well data allocation requires that spinner surveys (or their equivalent) be run periodically in the individual wells to determine how much of the fluid is coming from each of the perforated intervals. The spinner surveys should be run both with the well flowing and with the well shut in. Data from these surveys, along with pressure-buildup and -falloff data, help in the estimation of the reservoir pressure in the various reservoir intervals. Also, a variety of production logs should be run to estimate changes in fluid saturations in the near-wellbore region as the waterflood progresses.
The field engineers can use all these data in the types of calculations that are described below.
Special data acquisition
During the waterflood, there are likely to be opportunities to gather additional data away from the injection and production wellbores. These data can take several forms.
Infill wells are likely to be drilled in locations where oil should not have been displaced by the injected water. Consequently, these locations are not good locations for determining how effectively the various portions of the reservoir intervals are being swept by injected water. After drilling, the hydrocarbon distribution actually present at an infill well location is evaluated using openhole or cased-hole logging. Where formation waters have a sufficiently high salinity, pulsed-neutron thermal-decay time and resistivity logging may be used to evaluate the residual oil saturations. In lower water-salinity conditions, carbon/oxygen and resistivity logging are used. Using these techniques, the locations of fully flooded, partly flooded, and unflooded reservoir intervals can be determined in new wells and in existing producing wells. Multiple logging runs over time in a producing well allow the monitoring and management of a waterflood. If wells are drilled later during the waterflood, then the residual oil saturation distribution can be obtained by use of special coring procedures or special tracer tests.
Special observation wells sometimes are drilled at a location in the oil reservoir where the water/oil flood front should be detectable as it passes. Historically, most of these wells were cored but had steel casing, such that standard logging methods could not be used; however, before the development of through-steel-casing resistivity logging, fiberglass casing together with induction-resistivity logging occasionally were used to observe the water/oil displacement process over time.
Recently, the 4D-seismic technique has been developed to determine in what directions the water is moving from the injection wells. The 4D-seismic technique compares 3D-seismic data that were obtained before the start of waterflooding to a second or third 3D survey that was conducted some years later. This allows an areal visualization of where there are high-pressure areas caused by water injection and where there are low-pressure areas caused by production (typically from the presence of some free-gas saturation near the production wellbores). Also, the 4D picture might show which portions of the reservoir pay intervals are well connected and which are not.
Simple waterflood analysis techniques
The initial engineering analysis of waterflood performance is done to ensure that the water is being injected into the wells at the desired well-by-well rates and, if several oil-reservoir intervals are perforated in each injector, that each of those reservoir intervals is taking its appropriate share of the injected water from each of the injection wells. This may appear trivial, but many waterfloods have had significant problems in this regard. Before the waterflood begins, the engineers must estimate how much water should be flowing into each injector. Those estimates are based on fill-up and voidage-replacement calculations for that area of the oil reservoir and for the reservoir as a whole.
A "bubble map" can be used for visualizing the advance and relative volumes of the injected water. Fig. 3 shows the injection wells’ bubble map for one of the staggered line-drive pattern elements of one of the oil reservoir intervals of the Long Beach Unit (LBU) area of the Wilmington oil field in California, US Bubble maps are created for each injector by dividing the volume of water injected by the movable oil per vertical reservoir interval [thickness × porosity × (1-Swc-Sorw) ] to calculate the area that should have been swept by that volume of injected water; then presenting these areas on a map as circles of various sizes centered on each of the injection-well locations.
Fig. 3 – Layer injection-bubble map for Voidage Block 3 FO sand interval – LBU of the Wilmington oil field in California, US
Another aspect of waterflood monitoring is to track the performance of the production wells. As noted earlier, some oil reservoirs will have had a considerable period of primary production during which the reservoir pressure was drawn down below the bubblepoint pressure and a gas saturation developed. These reservoirs first will "fill up," which means that they will go through a period during which the injection water displaces mobile gas and increases the reservoir pressure to force free gas back into solution in the oil. During the early part of the fill-up period, the production wells will see minimal response because of water injection. When the reservoir pressure has returned to its original value, the water injection becomes a voidage-replacement operation. During the voidage-replacement period, the engineer must make sure of two things:
- That the volume of injected water equals the reservoir voidage of oil, gas, and water production
- That the various areas or patterns of the reservoir are being balanced to maximize the oil recovery and to minimize water-handling operating costs.
A number of graphs of the production and injection data can be prepared to help analyze the waterflood performance. For example, a conformance plot is a plot of cumulative oil recovery (or oil recovery efficiency) vs. net displaceable hydrocarbon PV injected. A waterflood performance envelope is defined by drawing an obtuse triangle that is bounded by recovery at the start of waterflooding, the maximum oil recovery at 100% of the x-axis, and a third point, Fpvg, which is related to the net injection required to displace the existing gas saturation at the start of the waterflood and is defined as:
where Sg = gas saturation, fraction.
Because actual performance cannot fall outside the performance envelope, this plot is a check on fluid allocation for a pattern. This plot also can indicate when injection water is being lost to thief zones. When slope changes are noted in this plot, the possible causes should be investigated. Fig. 4 is a conformance plot for the Kuparuk River oil field on Alaska’s North Slope.
Fig. 4 – Example of a conformance plot.
Other plots that typically are made after water breakthrough are the X-plot and its close cousin, the log(WOR) vs. cumulative-oil-production plot. The X-plot technique, developed by Ershaghi and coworkers, is based on the leaky-piston-displacement concepts of Buckley and Leverett and assumes that the plot of log (krw/kro) vs. Sw is a straight line. The X-plot is a graph of recovery, ER vs. X, where . This plot’s underlying assumption suggests that its usefulness theoretically is limited to higher water cuts.
The "cut-cum" plot, or plot of log(WOR) vs. cumulative oil production, easily can be made using production data for each well and for reservoirs as a whole. These plots generally are useful predictors of future waterflood performance because there is a considerable period of straight-line behavior on these plots for many wells and reservoirs when the waterflood is fully developed without major variations in field operations. Analysis of log(WOR) vs. cumulative-oil-production plots for waterflood analysis has been conducted using numerical reservoir simulation. These simulations show that the linear trends on this type of plot are found even at low WOR values and also are found for a variety of reservoir layering, flood configurations, and operational changes.
Decline curves also have been used for waterflood analysis. Where waterfloods were initiated in depleted sandstone reservoirs, empirical correlations were developed to estimate the likely oil rate increases during fill-up and while the oil bank moves toward the production wells, and then to estimate the oil rate declines as the WOR increased later in the life of the waterflood. Fig. 5 shows the production response for this situation. For the oil rate increase and decline periods, the plot has exponential oil rate vs. time characteristics.
Fig. 5 – Production response of a fully developed waterflood with water input to all injection wells at the beginning of the flood.
Hall plots and Hearn plots often are used to monitor injection wells. Use of these plots helps to maximize water-injection rates, which accelerates oil production from offsetting producers. On a Hall plot, the bottomhole injection pressure is plotted vs. cumulative water injection to monitor reservoir fill-up and average-reservoir-pressure increase. On a Hearn plot, the inverse injectivity index is plotted vs. cumulative water injection. Both types of plots also can be used to determine whether the water-injection rates are being kept below those allowed by the fracture-parting pressure.
A combination of these simple plot and calculation techniques typically is used for waterflood analysis. By plotting both oil rate vs. time and log(WOR) vs. cumulative oil, the engineer can better understand how well the waterflood is performing compared to original estimates and can determine what changes, if any, are needed.
Keep in mind that most of these techniques are premised on continued current operations. If there are significant changes in the allocation of injection water or if there are well workovers or pattern realignments, then the trends on these plots might not remain straight lines from which future waterflood performance is easy to predict.
Sophisticated waterflood-monitoring techniques
The modern numerical reservoir simulator is the best tool for performing waterflood analysis, including history matching of past performance and projection of future performance for continued current operations or for various operational and well changes. However, this tool generally is used with an updated history match only every 5 to 10 years.
Between major studies, the field engineers typically use simpler surveillance methods that have been upgraded by the availability of the notebook computers that have sizeable hard disks and rapid computing capabilities. New software packages have been developed that analyze trends and can handle large amounts of electronically acquired data.
One example of the use of this approach is the set of techniques used for the LBU area of the Wilmington oil field, which has 1,200 wells, multiple oil reservoirs, and 27 years of waterflood history. Fig. 6 shows the logic used for the surveillance calculations. As Fig. 6 shows, a large waterflood involves massive amounts of data that must be handled logically and consistently so that the engineers can obtain useful results.
Fig. 6 – Surveillance overview logic for a layered waterflood.
The routine waterflood-analysis calculations that are used for the Kuparuk River oil field are another example of this approach. The Kuparuk River oil field covers more than 200 square miles and contains 600 patterns, with two separate reservoir intervals. Again, a massive amount of well data has been gathered over more than a decade of waterflooding and primary production. All of the simple calculation procedures discussed earlier in this article [decline curves, log(WOR) vs. cumulative-oil-production plots, and X-plots] are used to evaluate well-by-well and pattern performance. Figs. 7 and 8 show relationships among the various calculation and database modules within the material balance shell and the object relationships.
Fig. 7 – Relationship of relational database management system (RDBMS) tables to material-balance shell logic. (HCPV = hydrocarbon pore volume, and SQL = structured query language.)
Fig. 8 – Material-balance object relationships.
These two examples are of very large oil fields. Oil fields with fewer injection and production wells will require less-extensive waterflood-monitoring calculations; however, all the types of material-balance calculations will be required to determine whether the waterflood is performing as expected and whether significant operational changes are needed.
The usefulness of all these calculations depends greatly on the quality of the input data. The original field data must be reviewed for completeness and accuracy before entering them to these types of calculations. Where gaps in the data or clearly erroneous numbers are found, the engineers must judge how to edit the data or adjust calculations for such data problems. Where total field data has been allocated to the individual wells, the allocation procedures should be checked, including whether they have changed over time and thus caused changes in the slopes of some of plotted data—changes that are not real, but that are artifacts of the data-allocation procedures.
|fw||=||fractional flow of water|
|Fpvg||=||fraction of displaceable pore volume that is gas saturated|
|Sorw||=||residual oil saturation to waterflooding, fraction PV|
|Swi||=||initial water saturation, fraction PV|
- Dyes, A.B., Caudle, B.H., and Erickson, R.A. 1954. Oil Production After Breakthrough—as Influenced by Mobility Ratio. Trans., AIME 201: 201. http://dx.doi.org/10.2118/309-g
- Craig Jr., F.F. 1971. The Reservoir Engineering Aspects of Waterflooding, Vol. 3. Richardson, Texas: Monograph Series, SPE.
- O'Donovan, A.R., Smith, S.G., and Kristiansen, P. 2000. Foinaven 4D Seismic—Dynamic Reservoir Parameters and Reservoir Management. Presented at the SPE Annual Technical Conference and Exhibition, Dallas, 1–4 October. SPE-63294-MS. http://dx.doi.org/10.2118/63294-MS
- Staples, R., Hague, P., Cooke, G. et al. 2002. Integrating 4D seismic to optimize production. Presented at the European Petroleum Conference, Aberdeen, 29–31 October. SPE-78346-MS. http://dx.doi.org/10.2118/78346-MS.
- Woodling, G.S., Taylor, P.J., Sun, H.H. et al. 1993. Layered Waterflood Surveillance in a Mature Field: The Long Beach Unit. Presented at the SPE Western Regional Meeting, Anchorage, 26–28 May. SPE-26082-MS. http://dx.doi.org/10.2118/26082-MS
- Chapman, L.R. and Thompson, R.R. 1989. Waterflood Surveillance in the Kuparuk River Unit With Computerized Pattern Analysis. J Pet Technol 41 (3): 277–282. SPE-17429-PA. http://dx.doi.org/10.2118/17429-PA
- Buckley, S.E. and Leverett, M.C. 1942. Mechanism of Fluid Displacement in Sands. Trans., AIME 142: 107–116. SPE-942107-G. http://dx.doi.org/ 10.2118/942107-G
- Ershaghi, I. and Omorigie, O. 1978. A Method for Extrapolation of Cut vs Recovery Curves. J. Pet Tech 30 (2): 203–204. SPE-6977-PA. http://dx.doi.org/10.2118/6977-PA
- Ershaghi, I. and Abdassah, D. 1984. A Prediction Technique for Immiscible Processes Using Field Performance Data (includes associated papers 13392, 13793, 15146 and 19506 ). J Pet Technol 36 (4): 664–670. SPE-10068-PA. http://dx.doi.org/10.2118/10068-PA
- Lo, K.K., Warner, H.R., Jr. , and Johnson, J.B. 1990. A Study of the Post-Breakthrough Characteristics of Waterfloods. Presented at the SPE California Regional Meeting, Ventura, California, USA, 4–6 April. SPE-20064-MS. http://dx.doi.org/10.2118/20064-MS
- Bush, J.L. and Helander, D.P. 1968. Empirical Prediction of Recovery Rate in Waterflooding Depleted Sands. J Pet Technol 20 (9): 933–943. SPE-2109-PA. http://dx.doi.org/10.2118/2109-PA
- Willhite, G.P. 1986. Waterflooding, Vol. 3. Richardson, Texas: Textbook Series, SPE.
- Jarrell, P.M. and Stein, M.H. 1991. Maximizing Injection Rates in Wells Recently Converted to Injection Using Hearn and Hall Plots. Presented at the SPE Production Operations Symposium, Oklahoma City, Oklahoma, USA, 7–9 April. SPE-21724-MS. http://dx.doi.org/10.2118/21724-MS
- Currier, J.H. and Sindelar, S.T. 1990. Performance Analysis in an Immature Waterflood: The Kuparuk River Field. Presented at the SPE Annual Technical Conference and Exhibition, New Orleans, 23–26 September. SPE-20775-MS. http://dx.doi.org/10.2118/20775-MSfckLR
- Hedges, P.L. and Scherer, P.W. 1996. Group Oriented Software Solution for Pattern Material Balance of an Areally Extensive Field, Kuparuk River Field, Alaska. Presented at the Petroleum Computer Conference, Dallas, 2–5 June. SPE-35987-MS. http://dx.doi.org/10.2118/35987-MS
Noteworthy papers in OnePetro
Use this section to list papers in OnePetro that a reader who wants to learn more should definitely read
Use this section to provide links to relevant material on websites other than PetroWiki and OnePetro