PRODUCTION

Using 4-D seismic to monitor and improve steamflood efficiency

Time-lapse, 3-D seismic was tested in a pilot project, then applied in full-scale monitoring of a large portion of a huge steamflood. Improved oil recovery and better utilization of steam energy should result

M. W. Waite, Rusdinadar Sigit, S. D. Jenkins, and M. F. Bee, Caltex Pacific Indonesia

his article describes a successful 4-D seismic pilot and the subsequent application of a full-scale reservoir management project in Duri field, Indonesia, the site of the world’s largest steamflood. The pilot demonstrated that, for the Duri steamflood, the horizontal and vertical distribution of steam can be tracked over time, and that a large increase in the value of the steamflood could be realized if seismic monitoring were applied on a large scale. The pilot study led to the implementation of a larger, time-lapse, 3-D seismic monitoring program, also discussed below, which will result in more efficient and improved oil recovery from the field.

Located on the island of Sumatra, Duri field is operated by PT Caltex Pacific Indonesia, a Chevron and Texaco affiliate. The field produces about 300,000 bpd of high viscosity oil. Since primary methods will recover only a small fraction of original oil in place, continuous steamflooding is employed to reduce the oil’s viscosity and drive it toward producing wells. As a result, recovery should improve from 8% to nearly 60%.

4-D Pilot Study

Steamflood monitoring tools, such as time-lapse seismic (4-D) data, are needed to improve oil recovery and minimize energy utilization. A pilot project was implemented to evaluate the use of 4-D seismic as a monitoring tool. The pilot area (Fig. 1) in Duri field included a central steam injection well, surrounded by six production wells at the corners of a hexagon, and two observation wells in which direct temperature measurements were taken (for calibration against seismic observations).

The study included relatively shallow (350–750 ft), early Miocene, deltaic sandstone reservoirs with excellent porosity (30–38%) and permeability (> 1,500 md). However, the sandstones’ heterogeneity can cause problems — substantial oil can be left in unswept intervals and steam "cycling" between injector and producer, along high permeability pathways, can waste energy. If 4-D seismic could help map the steamfront’s migration, these and other problems might be better managed.

Data observations. A baseline and six monitor surveys were recorded at various intervals over 31 months. Vertical sections from seven data volumes are shown in Fig. 2. The steam injection interval is between 135 and 220 ms. Dramatic changes can be seen after only two months of injection. A large time structure develops from within and below the injection interval due to systematic changes in reservoir properties. This time structure grows in successive surveys and reaches a maximum after 31 months. Note that waveforms and amplitudes also change between surveys.

The growth of the travel-time structure is illustrated by comparing the reflection times from the base of the steam injection interval at successive calendar times, Fig. 3. This cross-sectional view of the travel-times shows what appear to be two different effects taking place simultaneously. After two months of steam injection, there appears to be an overall shortening of travel-times relative to the baseline survey (the baseline reflection times are shown in light blue).

In addition, a synclinal-shaped time structure forms near the injector well. This travel-time "push-down" grows in magnitude in the later monitor surveys of Fig. 3. After 31 months of steam injection, the reflections around the injector are delayed 12 ms from the baseline state. This travel-time delay indicates an apparent interval velocity decrease.

The spatial distributions of these pull-ups and push-downs are shown in Fig. 4. The difference in interval travel-time from the top to the bottom of the steam zone between baseline and monitor surveys are shown in color. Green implies the same interval travel-time (and velocity) in both baseline and monitor surveys. Blue represents pull-ups in the monitor data, and yellow and red represent push-downs.

After two months, the development of a pull-up is observed around the injection well. After five months, the pull-up extends over the entire pattern and a push-down is developing around the injector. The depth and radius of this push-down increases in successive surveys (radius = 165 ft after 31 months). On later surveys, push-downs develop at the corners of the study area, close to injectors located just outside the pilot pattern.

4-D modeling and steamflood anatomy. Fluid flow simulations and time-lapse seismic models were generated to help understand these seismic observations.

A detailed 3-D reservoir model was constructed using log data from pilot wells and known properties of the field’s pore fluids. This model was subjected to thermal fluid-flow simulation consisting of a 10-year primary production cycle, followed by a 3-year steam injection cycle. The former was necessary to simulate reservoir conditions immediately prior to steamflooding.

Reservoir data (temperature, pressure, porosity and fluid saturations) at time steps corresponding to the acquisition of each seismic survey were converted to acoustic velocity models using Gassman-Biot relationships and lab core velocity measurements. CMP gathers were ray-traced through the 3-D acoustic model using the pilot survey acquisition geometry, and then processed to simulate stacked seismic data. These synthetic data were compared with the simulated reservoir states and the recorded seismic data.

Cross sections through the thermal fluid-flow simulation (Fig. 5) show results after two months and 19 months of steam injection. Color-coding indicates regions that contribute differently to seismic response. The steam-saturated zone (red) is characterized by high pore pressure (350 psia) and temperature (430°F) and very low residual oil saturation (15%). In thicker layers, steam tends to override or rise to the top due to gravitational forces. Steam propagates outward from the injector at different rates due to differences in permeability, heat loss rates and gravity overrides. Beneath and out in front of the steam-saturated area is a zone of hot liquid (yellow) composed of condensed steam, heated formation water and heated oil. Beyond the hot liquid zone is a pressurized zone of cold liquid (green), having higher remaining oil saturation (up to 60%).

Before steam injection, and in the initial months afterward, the thermal simulator predicts a zone that is partially saturated with evolved hydrocarbon gas (magenta). This zone has measurable (2–10%) gas saturations and is characterized by ambient reservoir temperatures and low pressures (25 psia). The evolved gas is generated by the draw down in pressure during the 10-year primary production cycle, which lowers reservoir conditions below bubble point. Note that after two months of injection, hydrocarbon-evolved gas near the injector has been forced back into solution due to an increase in reservoir pressure from steamflooding.

Cross sections through the acoustic velocity model after two months and 19 months of steam injection are shown in Fig. 6. After two months, faster velocities (red) are due to passage of the pressure front and dissolution of evolved gas. Note the very good correspondence between the low velocity zones (blue) in Fig. 6 and the location of the steam chest on the thermal simulation cross section (Fig. 5).

Synthetic stack sections show the seismic distortions resulting from changes in reservoir velocity and density after two and 19 months of injection, Fig. 7. Note the pull-ups in reflector times associated with the higher velocity pressure wave and the push-downs associated with the lower velocity steam chest. The shapes and magnitudes of both pull-ups and push-downs are in close agreement with monitor data. Amplitudes are generally brighter in the steam chest region due to an increase in the impedance contrast between the very low impedance steam chest and the higher impedance of the unswept sands and adjacent shales. Phase changes also can be seen in the synthetic sections, which are due to reflection coefficient sign reversals and waveform interference. These amplitude and phase changes were also observed on the pilot seismic data.

Steam thickness mapping. Estimates of steam thickness after 31 months of steam injection were made for each of the three reservoir intervals, Fig. 8. Maps were generated by converting changes in interval slowness to steam thickness using a relationship established from time-lapse modeling. The blue areas on these maps have not been swept by steam, and are characterized by low reservoir temperature and high remaining oil saturation (around 100°F and 60% So).

The steam-saturated zones surrounding the injector are characterized by high temperature and low oil saturation (up to 375°F and 5% S_o). Note that the Lower Pertama interval is acting as a thief zone in the pilot area, and is taking the greatest amount of steam of the three layers shown in Fig. 8. Dramatic differences in the sweep patterns of the three intervals are apparent from these maps.

Economic feasibility. The quality and detail of reservoir information from the pilot study generated interest in large-scale application of seismic monitoring to improve reservoir management. A team of engineers, earth scientists and operations personnel assessed the economic feasibility of large-scale monitoring at Duri. After reviewing the practical benefits, the team determined that time-lapse, 3-D seismic could significantly improve injector profile management. Benefits include shutting off injection in swept zones and putting steam into cold zones. Time-lapse data could locate observation wells in the right places and could eventually reduce the need for them.

The team took into account cost of seismic data and the probability (risk) of various outcomes. The study examined several operating scenarios for the Duri steamflood and concluded that the largest net present value is obtained by aggressively managing the steamflood using 4-D seismic. The team therefore proposed time-lapse seismic monitoring on a large scale at Duri.

Large-Scale Seismic Monitoring

Following the completion of the successful pilot study, multipattern time-lapse seismic surveys have been acquired over many areas of the Duri steamflood to help manage steamflood conformance and identify and exploit regions of bypassed oil. These surveys are shot between 6 and 18 months apart, depending on steamflood maturity and operational need. The following case history illustrates the added-value of seismic data to the reservoir management process.

Area of interest. The area of interest (AOI) consists of sixteen, 15.5-acre, inverted nine- and five-spot well patterns, Fig. 9. Steam was injected into a 150-ft interval consisting of three major flow units. These are, from shallowest to deepest, the Upper Pertama, Lower Pertama and Kedua. Producing wells were completed as open hole gravel packs across the entire flood zone. Three observation wells were drilled in the AOI to monitor steamflood performance.

Two 3-D seismic surveys were acquired over the AOI to monitor steam flow and to assist in conformance management. The first survey was acquired before steam injection began, the second was acquired seven months after beginning injection.

Conventional interpretation. The relative distribution of heat production (as determined from flowline temperatures) in the AOI seven months after steam injection is shown in Fig. 9 as the diameter of each producer well symbol, which is proportional to flowline temperature recorded at the wellhead. Most producers register temperatures near ambient reservoir conditions of 100°F. However, five producers (P1-P5 in Fig. 9) had high flowline temperatures ranging from 224° to 294°F, which operational experience indicates to be caused by steam breakthrough.

Under ideal conditions with symmetrical areal conformance and high vertical sweep efficiency, steam breakthrough at the nine-spot pattern is expected after 4 or 5 years of steam injection. Thus, the much earlier-than-anticipated high flowline temperatures indicate suboptimal steamflood conformance. The keys to choosing the best remedial technique for improving conformance include:

• Identifying the steam breakthrough flow units
• Identifying the source injector(s)
• Making reasonably accurate assumptions of areal and vertical conformance from available data.

While a reservoir management review of each pattern in the AOI is beyond the scope of this article, a summary of conventional monitoring serves to highlight the uncertainty involved in making these determinations:

• A temperature survey in observation well OB2 near hot producers P2 and P3 shows steam-related heat in the Upper Pertama and ambient conditions in the other flow units.
• Injector profile data indicate that, on average, an unbalanced allocation of steam is occurring in favor of the Lower Pertama flow unit.
• None of the available data reveal which injector wells are sourcing steam to the hot producer wells.

The ambiguity presented by these data presents a reservoir management dilemma. Inferences drawn by extrapolation, such as using observation well temperature surveys to determine which layer to isolate in a nearby steam breakthrough producer, can lead to inappropriate actions and unsuccessful results.

Areal conformance mapping. Time-lapse monitoring of Duri field involves the continual processing and analysis of vast quantities of data. To streamline the effort and minimize cycle time to satisfy demands of such a large undertaking, an efficient method was used to transform the meaningful aspects of seismic data into a model of conformance that is consistent with the known or assumed reservoir-state near well control. A multivariate statistical analysis technique, called discriminate analysis, was used to classify seismic feature sets corresponding to each flow unit as characteristic of either the presence or absence of steam. This technique is described in detail in reference 2 and was further refined using an interconnectivity analysis, which is also described.

Conformance analysis. Areal conformance maps for the Upper and Lower Pertama flow units are shown in Fig. 10. A meaningful classification of the Kedua unit was not possible because of the absence of coherent steam indicators at this level.

The Upper Pertama conformance map reveals that steam in the layer has not reached any of the producers. It does show steam at well OB2, which is consistent with the findings of the temperature survey acquired in the well. Note that despite well OB2’s proximity to the hot producers P2 and P3, the OB2 temperature survey does not yield information relevant to remedial action for these producers. This underscores the inherent risk in drawing conclusions about interwell reservoir relationships from borehole information alone.

The Lower Pertama conformance map shows highly asymmetrical, channeled fluid flow with no overall directional preference. The map reveals that hot producers P1 through P4 are receiving steam from injectors I1 through I4, respectively. This information is critical for successful design of workover and conformance-improvement programs.

The most conspicuous of these anomalies is the highly directional steam channel from I3 to P3. Steam flow between these wells circumvents nearby well OB2, thus evading detection by an OB2 temperature survey. In addition to the nonuniform areal conformance in this layer, vertical sweep efficiency is adversely affected by gravity overrides.

A comparison of the two conformance maps reveals that the Lower Pertama flow unit exhibits the greatest areal sweep in the AOI. The ratio of the total area classified as steam in the Lower Pertama to the total area classified as steam in the Upper Pertama is 1.61. Interestingly, this number compares closely with the ratio of the total pore volumes (PV) injected for the two layers (1.66).

The Kedua flow unit was steamflooded in seven of the AOI patterns. The seismic interval associated with the Kedua is absent of steam indicators. A possible explanation is that, as suggested by the low surface-steam quality and the calculated heat loss through the uninsulated tubing, fluids injected into the Kedua are mostly in the liquid phase, and therefore, will generate a reduced or negligible seismic response. The absence of Kedua steam indicators supports, along with other data, the need to incorporate insulated tubing in future dual string injector completions.

Implementation. A work program to improve pattern conformance in the AOI was designed on the basis of all available geologic, geophysical and engineering data, Fig. 11. Elements of the program include the following:

• Cyclic steam stimulation of producer wells to increase oil mobility and "draw" steam into cold areas
• Chokeback of producer casing pressure to "push" steam into unswept areas
• Isolation of steam breakthrough intervals at producers to improve vertical conformance
• Profile modifications at injectors to improve vertical conformance.

The conformance maps from seismic data were critical in determining the amount of heat used in cyclic steam stimulation jobs at producers, selecting wells to choke back producer casing pressure, and deciding which flow units to shut off at producer wells for steam breakthrough isolation.

Source Literature For This Article

¹ Jenkins, S. D., M. W. Waite and M. F. Bee, "Time-lapse monitoring of the Duri steamflood: A pilot and case study," The Leading Edge, September 1997.

² Waite, M. W., R. Sigit, A. V. Rusdibiyo, T. Susanto, H. Primadi and D. Satriana, "Application of Seismic Monitoring to Manage an Early-Stage Steamflood," SPE Petroleum Reservoir Engineering, November 1997.

The authors

Mike W. Waite is a geophysicist with Texaco currently working in Kuwait. He has held positions involving seismic acquisition and processing, development and application of subsurface imaging techniques, prospect evaluation and reservoir characterization with 3-D/4-D seismic data. Mr. Waite holds a BS degree in physics from the University of New Orleans. He is a 1997 SPE Distinguished Lecturer on time-lapse seismic monitoring.

Rusdinadar Sigit is a geophysicist with Caltex Pacific Indonesia. He previously worked as a geophysicist for Amoseas Indonesia in various exploration fields. Mr. Sigit holds a BS degree in geology from ITB University in Indonesia.

Steven D. Jenkins received a bachelor’s in geology (1979) from the University of Tennessee, and a master’s in geophysics from Indiana University in 1981. Since 1992, he has worked as a senior development geophysicist responsible for 3-D seismic interpretation at Caltex Pacific Indonesia in Rumbai. He is currently working on detailed reservoir characterization, seismic sequence analysis and EOR monitoring using time-lapse seismic.

Michel F. Bee received a bachelor’s degree (1974) in general science and a master’s (1976) in civil engineering from the University of Poitiers, France, and a master’s (1979) and a Ph.D. (1984) in geophysics from Oregon State University. He is currently a staff geophysicist for Chevron Nigeria Ltd. in Lagos. His applied research interests include all areas of geophysics with a current emphasis on 3-D designing, multi-component analysis and EOR monitoring. He is a member of SEG and EAEG.

Copyright © 1999 World Oil
Copyright © 1999 Gulf Publishing Company