INTERPRETER S CORNER Coordinated by Rebecca B. Latimer Imaging complex structure with crosswell seismic in Jianghan oil field QICHENG DONG and BRUCE MARION, Z-Seis, Houston, Texas, U.S. JEFF MEYER, Fusion Petroleum Technologies, The Woodlands, Texas, U.S. YIWEI XU and DILONG XU, The Bureau of Jianghan Oil Field, Guanghua, Hubei Province, China We conducted a crosswell seismic project in April 2002 in the Ma production area of Jianghan oil field in central Hubei Provice, P.R. China. This part of the Jianghan fault basin contains complex faults that dominate the reservoir zone, and they significantly affect flow paths and fluid distribution. Therefore, it is critical to understand the fault structure distribution in this area at the reservoir scale in order to develop and enhance production. The Ma area has been surveyed with 3D surface seismic and most wells are extensively logged. This area is known to be a seismically poor data area due to its extremely slow weathering velocities, highly dipping formation, and complex structure. As a result, surface seismic techniques do not provide high resolution, and have limited ability to image the faults or to provide independent quantification of the various critical reservoir properties. On the other hand, log measurements deliver high-resolution sampling and fine precision, but only sample a very small volume of the reservoir. In this very complex reservoir, the wellbore information is an undersampled representation because of the intrinsic geologic heterogeneity of the reservoir. Crosswell seismic imaging, a relatively new technique for generating high-resolution images of the reservoir between wells, can directly image detailed reservoir features and provide high-resolution information in seismically difficult areas where surface imaging is challenged by highly dipping formations and complex faulted structures in the shallow formations above the reservoir interval. The objective of the crosswell project in the Ma area was to generate both a velocity image and a reflection image to characterize the lithologic detail of the formation and the fault distribution in the reservoir interval. This information will be used by the geologist and reservoir engineer to adjust injection and to pinpoint infill-drilling locations. In this paper, we demonstrate that crosswell data can produce high-resolution images, which can better define the faulted reservoir intervals and fault distribution between the wells. Presurvey modeling and data acquisition. The surface conditions in the Ma area are such that most wells are drilled with large deviation and extended spacing, typically 500-700 m apart. Figure 1 shows the crosswell seismic survey map of the Ma area in Jianghan. The steeply dipping contours of the reservoir formation with crossing fault lines indicate the complexity of the reservoir. The map, sketched from drilling, surface seismic, and logging information, has numerous uncertainties because of the information gap that exists between the poor quality surface seismic data and the logging data. We acquired two crosswell profiles between three wells Ma57, Ma52-1, and Ma56-1. As indicated on the map, Ma52-1 was the receiver well, Ma57 and Ma56-1 the source wells. According to the coordinates and well deviation information, the surface spacing is 685 m and bottom-hole spacing is 650 Table 1. Lithologic units Unit Lithology number 1 2 3 4 5 Clay and argillaceous chalk Pure gypsolith Shaly sands with thin argillaceous chalk Shaly sands with organic shale Clay shale Average velocity (m/s) 3500 4700 3800 3700 3500 Figure 1. The crosswell seismic survey map of the Ma area in the Jianghan oil field. The faults were plotted based on drilling, surface seismic, and logging information. Well Ma56 is deviated to a maximum of 28 from the vertical. The largest spacing, between wells Ma56 and Ma57, is about 700 m. This map also shows formation dip of about 17 in this area. m for profile Ma52-1 to Ma57 and the surface spacing is 568 m and the bottom-hole spacing is 460 m for profile Ma52-1 to Ma56-1. Well Ma56-1 is highly deviated with a maximum angle of deviation of 28 at a depth of 900 m. The apparent dip of the formation between the wells in the Ma area is about 17 from the horizontal (based on log information). There are five major units in the reservoir in the Ma area: Xinyishang, Dagao, Youzhu I, Youzhu II, and Nige. Xinyishang and Dagao contain mainly clay shale and pure gypsolith. Youzhu I and II, from which much of the past oil production is believed to be derived, contain shaly sands with thin argillaceous chalk. The wireline logs recorded in the three wells were gamma ray, sonic, density and resistivity, and neutron porosity. The lithologies and average velocity interpreted from logs for each unit are shown in Table 1.
Figure 2. Side view of reflection rays in 3D presurvey modeling. The velocity model is derived from the sonic logs interpolated with a simple dipping structural assumption. The structural model is built with simple planar surfaces based on the formation tops picked from gamma ray and sonic logs in the three wells. Figure 3. The crosswell seismic data processing flow contains two parts, the tomography and reflection data processing. Figure 4. The primary result of traveltime inversion for both profiles. This result is based on the initial planar surface structural model using formation tops picked from the gamma ray and sonic logs in the three wells.
Figure 5. The direct arrival and reflections predicted by standard anisotropic raytracing through the velocity model for the Jianghan crosswell data. Both the direct arrival and reflection trajectories reasonably follow the real data. Figure 6. Jianghan crosswell reflection results with upgraded tomogram overlay. The upgraded tomogram is based on an updated structural model interpreted from the initial reflection image.
Figure 7. Comparison of crosswell reflection images with surface seismic data lines that cross Ma 57, Ma 52-1, and Ma 56-1. The surface seismic data in the black box represents the same depth sections as the crosswell data. Survey planning can be complex when dealing with highly deviated wells and steeply dipping formations so, before the survey was shot, transmission and reflection wavefield modeling was conducted. Figure 2, a side view of our three-dimensional model, shows reflected rays traced through the model. Since we had no reliable structural information, we derived the velocity information between the wells from a sonic log interpolation and a simple dippinglayer assumption. The objective of this modeling was to develop an acquisition strategy that provided good raypath coverage between the wells for the range of expected geologic scenarios. This model uses the actual well deviation, coordinates, and sonic logs. The modeling results in Figure 2 show that the anticipated reflection energy propagates outside of the plane of well Ma52-1 to well Ma56-1 below 900 m. When the crosswell survey was planned in this area, we considered reflection coverage, tomographic coverage, well deviation, out-of-plane effects, and multiple-profile integration for better control of structural uncertainty, and the ability to update the survey design during acquisition (Meyer et al., 2002). Crosswell seismic field operations access two wells simultaneously, in a manner similar to standard wireline operations. A piezoelectric downhole seismic transmitter was lowered into Ma57 or Ma56-1 and a 10-level hydrophone receiver array with 3-m spacing between levels was lowered into the Ma52-1 well to cover the depth interval 500-1400 m. The receiver array was fixed in one well, while the source was pulled slowly up the other well. The source was set to fire a quick sweep over a frequency range of 150 to 700 Hz every time the source moved another 1.5 m up the well. This enabled continuous data acquisition without stopping the wireline, and greatly accelerated the speed of acquisition. Throughout this process, the seismic data are collected for energy propagating between the wells, directly across the reservoir or other zone of interest. In the Ma area, 123 000 traces were recorded for two profiles, with a sample interval of 0.25 ms in each trace. The raw data signal-to-noise ratio (SNR) was fair to good. Data processing. As with standard processing of surface seismic data, a wide range of methods was used to process the seismic data into a structural image of the interwell space. Figure 3 shows the workflow. After data quality control and editing, the first processing step employed traveltime inversion to yield a velocity image between the wells with depth directly referenced. This image also served as the base model for reflection imaging. The direct arrival traveltimes were picked for 75 000 traces on both profiles to run P-wave traveltime tomography. The bin size of the grid for the velocity model in the inversion process was 6 m horizontally and 1.5 m vertically. The primary 3D structural model for inversion processing was built with tops from the Xiyishang, Dagao, and Nige units that were picked from each well s sonic or gamma logs. The picked tops of each unit were connected to create several planar surfaces. This ensures that structure in the model initially fits what is known about the geology near the wellbore. The model can be augmented by interpretation of the reflection data to refine the model s surfaces in an iterative manner. The traveltime inversion used for the crosswell data for the Ma area in Jianghan employs a continuation strategy that has been shown to decrease data-matching errors significantly when compared to traditional unconstrained inversion (Washbourne and Bube, 1998). The idea is to first resolve the lowest-frequency component in the model, and then to increase successively the spatial resolution in the model, generally by decreasing the weight of penalty terms that force model smoothness. The continuation approach has been shown to reduce the global traveltime residual, thereby producing a more accurate model. Figure 4 is the primary result of traveltime inversion for both profiles from the structural model generated from log inference. The gamma ray logs from each of the three wells
Figure 8. Three-dimensional view of the reflection image with spatially interpreted fault interfaces for the two profiles. (a) to (c) are the side views; (d) is the top view. F1, F2 and F3 are three major faults interpreted from crosswell data. are plotted with the tomograms for quality evaluation. The tomograms exhibit gently dipping structure that agrees with the dips inferred from the well-log ties. The tomographic velocities have similar structural trends to those inferred from the gamma ray logs and, we can identify several different velocity layers within the five formation units. These tomograms were updated later with the structural information extracted from the reflection image. The subsequent processing steps necessary to produce detailed reflection images were wavefield separation, mapping (which might be viewed as a special case of a Kirchoff depth migration), and postmap migration. The wavefield separation included tube wave, direct arrival, and downgoing reflection removal (Rector et al., 1992). After we isolated upgoing reflection arrivals from the noise (both coherent and incoherent), we then converted the data from the acquisition time domain to the image domain in depth. In the image domain, the source depth and receiver depth were replaced by the reflector position and the incidence angle, and the traveltimes were transformed into subsurface depth. Figure 5 shows the reflection traveltimes predicted by standard anisotropic raytracing of the Jianghan crosswell data. Clearly, both the direct arrival and reflection trajectories reasonably followed the actual seismic direct and reflected arrivals. The velocity model used for raytracing was derived via tomography, which solves for transversely isotropic (TI) anisotropy parameters in this shale prone area. After the upgoing reflection wavefield was mapped with a VSP-CDP mapping algorithm, the mapped data were transformed to produce constant-angle stacks. The final reflection image was produced by stacking reflection events with reflection angles in the range of 50-70 from the vertical. A special migration step was also included in the Jianghan crosswell data processing because of its complex structure. This postmap migration can improve the lateral resolution of the mapped data by applying a migration algorithm to collapse diffractions that are caused by formation discontinuities such as faulting (Byun and Rector, 1997). Imaging results. Figure 6 shows the reflection results overlying updated tomograms for the two profiles. The upgraded tomograms were generated with a new 3D nonplanar surface structural model that was built with additional tops picked in depth domain from the critical crosswell reflection result. In this figure, the gamma, sonic logs, and synthetics are plotted with the reflection images. The blue curves are the gamma ray logs, the black curves are the sonic logs, and the red curves are the tomographic velocity values extracted from the tomogram at the two wells. In Figure 6 most formations interpreted from the logs follow the reflection events laterally. The apparent discontinuities that exist in reflection data indicate the fault positions. The apparent throw also can be estimated from the reflection data. Generally speaking, the upgraded tomograms are structurally consistent with the reflection data but the reflection data have higher lateral and vertical resolution. Figure 7 compares the crosswell reflection image with surface seismic data from lines which cross wells Ma57, Ma52-1, and Ma56-1. The black dashed lines on the crosswell data represent fault interpretation. The surface seismic data in the black box represents the same depth section as the crosswell data. The structure in both seismic images has some similarities. However, the surface seismic data can be insufficient for detailed reservoir analysis due to lower res-
olution; the crosswell results have 5-10 times the resolution of the surface seismic. Figure 8, a three-dimensional view of the reflection images, shows the spatial distribution of the faults with interpreted fault interfaces for the two profiles. Figures 8ac illustrate side views and Figure 8d illustrates the top view. The violet interface represents a major reverse fault. From the 3D seismic data display, we can see that Ma52-1 is on the upthrown block of reverse fault F1. Two additional small faults, F2 and F3, are normal faults with the green and red fault interfaces. Conclusions. The Ma area of the Jianghan fault basin is known to be complex based on surface seismic information, logs, and well production data. Imaging from the surface is difficult due to highly dipping and complex faulted structures in the shallow formations above the reservoir interval. In addition, existing data showing the intrinsic geologic heterogeneity of the reservoir indicate the need for data with higher resolution than surface seismic data to better understand the reservoir architecture, including the fault structure distribution. Crosswell seismic was employed to produce high-resolution crosswell profiles between three wells in the Ma area. Interwell distances were 600-700 m. The resulting tomographic and reflection images reveal a complex distribution of subseismic faulting. Interpretation of the crosswell data in 3D yielded new understanding of the reservoir architecture. Based on the initial crosswell survey and additional high-resolution data planned to be acquired, reservoir connectivity between wells will be better characterized allowing the reservoir engineers to optimize injection patterns and increase production. The complex fault distribution may also require additional crosswell information to better pinpoint infill-drilling locations. Based on this initial survey, the value of high-resolution crosswell information in enhancing the understanding of reservoir architecture has been demonstrated. Use of crosswell data may lead to enhanced economic value for Jianghan Oilfield based on increased production and enhanced efficiency in developing the reservoir. Suggested reading. Crosswell seismic: review of field operation, survey planning and case history by Meyer et al. (West Texas Geological Society Fall Symposium, 2002). 3D high resolution imaging from crosswell seismic data by Washbourne and Bube (SPE 49176, 1998). Extraction of reflection from crosswell wavefields by Rector et al. (SEG 1992 Expanded Abstracts). Post-MAP migration of crosswell seismic data by Byun and Rector (SEG 1997 Expanded Abstracts). TLE Corresponding author: qdong123@yahoo.com