A study established a quantitative fracture identification method for the igneous reservoir rock in the Xingcheng group of Xushen gas field, a field near Daqing in northeastern China.
The method uses a curve unit principle and an algorithm.
The field has much data available such as core descriptions, conventional logs, formation microimaging (FMI), and drilling data.
Compared with the identification from conventional well logs, FMIs have an advantage, in most cases, for characterizing fractures. But in some local low-resistivity formations, conventional well logs could assist in interpreting FMIs.
Practical experience and this study demonstrate that the combination of FMIs with conventional well logs can lead to a method for identifying fractures in igneous rock and shed some light on other complex reservoirs, such as metamorphic rock, carbonate rock, and tight-sand oil and gas reservoirs.
The oil and gas industry gradually has discovered igneous rock reservoirs, an unconventional formation containing oil and gas.
Some of the features of igneous rock reservoirs are a wide-ranged distribution, long period of geologic evolution, diverse lithology, and a complex pore structure. Fractures are a major part of evaluating these reservoirs.
Core observations
According to the fracture-development features of cores with different lithologies, fractures developed in welted pyroclastic rock represent 55% of total observed fractures. Their development is greater than fractures in lava and normal vocaniclastic rock.
In the welted pyroclastic rock, most developed fractures seen are in the rhyolitic crystal tuff, accounting for 45%. Among all of the welted rocks, rhyolitic rock has the highest fracture development level of 15%.
From the perspective of fracture occurrence, high-angled fractures account for 67% of the total number of fractures. Seldom seen are low-angled and horizontal fractures.
About 80% of the fractures are shorter than 50 cm. Fractures longer than 75 cm are rare.
About 67% of the fractures have a terminal number of 0, indicating that the observed core fractures are only a part of the whole fracture length. That is to say that the total fracture length is much longer than observed.
Compared with other fractures in the study area, some blocks have good fracture development whenever the planar development characteristic shows obvious heterogeneity.
Log characteristics
Mud filtrate always fills the high-angled fractures. The filtrate can serve as a low-resistance channel for lateral-log interpretation. Moreover, the effective conductive section of high-angled fractures will stay consistent in the radial direction, while the conductive section of pores will enlarge gradually.
In the test area, the shallow lateral log in the effective conductive section in the fracture and pores reads much less than that of deep lateral log (Rlld > Rlls).
In the segment of high-angled fractures in the core, where there is high resistance, there appears an abnormal phenomenon of low to middle resistance in the dual-water log curve with amplitude differences of 200-300 ohm-m and a positive separation. Observed also is an obvious increase in the photoelectric curve when the mud contains barite.
Because the P-wave travels faster in fluids than in consolidated rocks, the acoustic wave will slow down when it passes horizontal fractures and low-angle fractures. This leads to a larger time difference.
This is rarely seen in high angled and horizontal fractures. For these fractures, there is almost no response to the time difference of the S-wave when it passes the horizontal and low-angle fractures. High-angled and vertical fractures have a larger time difference for the S-wave (Fig. 1).
The vertical resolution of resistance-imaging logs could reach 5 mm with a full hole covered image. Processed with depth correction, image generation, balance adjustment, standardization, and image display, the original data transform into images that reveal all valuable formation characteristics around the borehole. Moreover, the images also provide quantitative information on fractures.
One can view fractures in space as planes. Any plane that is not perpendicular to the hole axis can intersect with the cylindrical borehole. The interaction has an oval shape.
Unfolded along the borehole wall, the fracture forms a sine curve in which the orientation of the lowest point stands for the fracture dip. The tangent value, θ, of fracture dip equals the ratio of the distance from the sine-wave peak, Dsin, to the bottom and the hole diameter, d, (tan θ = Dsin/d).
In Fig. 2, the fracture dip is 90°, indicating the fracture tilts eastward and has 180° mean strike in the south-north orientation. The amplitude of the sine curve is 50 cm, borehole diameter is 22.4 cm, and the calculated inclination is 65.9°.
Probability calculation
If one of the formation's physical properties responses on a log curve differs from that of the adjacent beds, then this part of log curve represents the well log unit or curve unit.
The log-response values of most sample points in a log unit are close to the mean value with only a minor deviation with a distribution out of the boundary of X:[a,b]. This indicates that the discrete log data conforms to normal distribution. For any log unit with omitted boundary points and if the intervals are from a to b, Equation 1 (see equation box), expresses the probability distribution function X.
Under this assumption, the mathematical expectation of variable X is E, where X = (x0, x1, …, xn), variance is σ, and the differential is J. Equation 2 defines the probability of fractures existing on a single curve.
In the equation ΔJ = ABS (J0 – J), ΔE = ABS (E0 – E). Δσ = ABS (σ0 – σ), J0 is the range of the log unit, E0 is the mathematical expectation of the log unit or the average of the normal distribution, σ0 is the variance of the log unit, J is the range of the log unit within a moving window, σ is the variance of the log unit within a moving window, and μ is the average of the log unit within the moving window.
Following the Bayes principle with multiple log curves for calculating the probability of the existence of a fracture, Equation 3 expresses the fracture AND probability while Equation 4 expresses the fracture OR probability.
In the equations, Pfi is the probability of fracture existence on a single curve, n is the log curve number or variable number, and Wi is the weighting for every log curve.
For calculating the AND-OR probability of fractures, the weighting of every curve is:
• Compression wave slowness (time difference) AC is 0.2.
• Photoelectric absorption index (Pe) is 0.2.
• Deep laterolog is 0.3.
• Shallow laterolog is 0.3.
The calculation uses five sample points for determining the length of the moving window. If the borehole enlargement is more than 1.3 times the borehole diameter, the calculation needs to include the borehole enlargement in the intervals processed. The lower limit of fracture probability in each log curve is 0.1.
Table 2 indicates that the OR probability is more accurate than the AND probability.
Decimal point attributes lead to the larger error and smaller value of the AND probability. Decimal point attributes have less effect on the OR probability; therefore, the OR probability is the dominant approach used for discriminating fractures during on site well logging operations.
Fracture identification
In this study, the conventional logging sample interval is 12.5 m. Observation of the core fracture shows a fracture length between one to several sampling intervals in the igneous rock reservoir from the Xingcheng group.
To identify fractures through the curve unit and conventional logging, the fracture length needs to span three sample points (two sample intervals). Three sample points can ensure that the fracture curve unit changes from large to small and large again. That is to say, to see the fracture characteristics response on a conventional log curve, the extended length of one fracture must be more than 25 cm, which constrains conventional logs from identifying fractures more than 25 cm in length.
From 3,741.5 to 3,743.1 m in Well Xusheng 8, the calculated fracture probability is greater than 0.6. Seven high-angle fractures are between 30 cm to 90 cm long. All of them are split (Fig. 3).
Several reasons contribute to the good fracture identification in this interval, as follows:
• Some local crushing in the core indicates that there are more than seven fractures in the formation.
• Most observed fractures in this interval are high-angled fractures, or even standing fractures with a long extended longitudinal length. Conventional logs show that the fractures are conductive.
• The drilling mud used in the volcanic rock interval of this well was water based and formed a low-resistance channel after the mud invaded the fractures. The low-resistance strip and rock matrix could make up a parallel circuit detected with conventional logs.
• Volcanic reservoir rock intervals have stable hole diameters that provide a high-quality log that allows seeing minimal differences in physical properties around the wellbore.
Image logs have the advantage in quantitative analysis of fractures but are not always reliable. For instance, when the fracture develops in a low-resistance interval, where there is a large difference between high and low resistivity in the borehole, the low-resistivity interval would mask the fracture.
Fig. 4 compares fracture identification with conventional and image logs in Well xs3 for the interval 3,990-4,010 m.
The conventional logs show only a 15 ohm-m resistivity while the image log curve has a 90,000 ohm-m resistivity. This is 6,000 times more than shown on the conventional log, and the images are dark over the entire interval. Barite in the drilling fluid causes this phenomenon.
This barite infiltration into fractures leads to an anomalously high Pe value. The interval 3,990-4,010 m clearly shows this abnormality. The interval has a Pe of 17.9 barns/electron compared with a 17.9 b/e in the normal interval. Also the S-P interval shows anomalously high readings.
According to the structural map, the well is encompassed by the major Songxi fault zone and a secondary fault, which is the dark area, indicates poor coherence in this formation. This structural setting also is conducive for the development of many fractures. Much evidence indicates that this interval is distributed with fractures but also probably has a complex fracture network.
The analysis shows the OR probability of fracture identification with conventional logging is 0.7 and the AND probability is 0.5, which could be identified as fracture response. Due to the masking of fractures by a conductive formation with low resistivity, the image log showed no fracture response.
Observations
The study found that:
• Fractures in igneous reservoir rock of the Xingcheng group are mostly high-angled fractures, which mainly developed with welted volcanic clastic rock that is rhyotaxitic crystallinoclastic, and rhyolitic. The observed fractures in cores are only a part of the fractures in the formation. The fractures observed are less then 50 cm in length.
• The dual-lateral log, S-P interval, and photoelectric cross-section log provide good fracture response in igneous rock. The degree of fracture development can be characterized more accurately by OR probability of fractures that are less influence by the coefficient of error.
• Conventional log identification of fractures requires that the drilling fluid before logging must be water based. From the perspective of application effects, the curve unit is a reliable approach in identifying fractures.
• In most cases, the FMI provides a good fracture characterization. Some local low-resistivity formations can mask fracture features on the FMI, but conventional well logs can serve in correlating FMI results to distinguish the low-resistivity caused by fractures.
• The technical ideas of comprehensive fracture-discrimination combined with core observation, conventional logging, and image logs could shed some light on fracture identification in other kinds of reservoirs.
Acknowledgment
The authors thank their colleagues who helped in the study, Liao Wang for providing the on site information of pilot tests in Daqing oil field, and the dean of the oil and gas engineering department at China University of Petroleum (Beijing) for his permission to publish this article. OGJ
Bibliography
Hawlander, H.M., "Diagenesis and reservoir potential of volcanogenic sandstones-Cretaceous of the Surat Basin, Australia," Journal of Sedimentary, Vol. 66 No. 3/4, 1990, pp. 181-95.
Mathisen, M.E., and McPherson, G.M., "Implications for hydrocarbon exploration in Volcaniclastic deposits," Sedimentation in Volcanic Settings special publication No. 45, 1991, pp. 20-27.
Secmann, U., and Schere, M., "Volcaniclastics as potential hydrocarbon reservoirs, " Journal of Clay Minerals, Vol. 19, No. 9, 1984, pp. 457-70.
Jin Huo, et.al., "The logging evaluation of volcanic rock in ancient 16 well field," Southwest Institute of Petroleum's Journal, (translated from Chinese), Vol. 25, No. 6, pp. 5-8.
Yuexia Dong, et al., "The evolutionary sequence research of Tertiary's volcanic rock in PU Nan depression," Journal of Petroleum Exploration and Development, (translated from Chinese) Vol. 30, No. 2, 2003, pp. 24-26.
Youjia, Wu, and Dalin, Liu, "The type and characteristic of volcanic rock reservoir of metamorphic rock in China, Southwest Institute of Petroleum's Journal, (translated from Chinese), Vol. 26, No. 6, 2004, pp. 5-6.
Zhengshun, Xun, et al., "The gas reservoir of volcanic rock's identification and evaluation of Xusheng gas field in Daqing," Journal of Petroleum Exploration and Development, (translated from Chinese), Vol. 33, No. 5, 2006, pp. 521-31.
Chenglin, Zhao, et al., The volcanic rock of Liaohe basin and hydrocarbons, The Oil Industry Press (translated from Chinese), 1999.
Fang Chen, et al., "The research on the form mechanism of highly porous deepseated volcanic rock," Southwest Institute of Petroleum's Journal (translated from Chinese), Vol. 27, No. 3, 2005, pp. 9-10.
Jiangguo Wang, et al., "The application of APS Logging in the low resistivity gas's identification of volcanic rock," The Gas Industry Press (translated from Chinese), Vol. 27, No. 8, 2007, pp. 35-37.
More Oil & Gas Journal Current Issue Articles
More Oil & Gas Journal Archives Issue Articles
View Oil and Gas Articles on PennEnergy.com
