table of Contents
1. Double lateral logging identification
4. Acoustic time difference response characteristics to identify cracks
5. Integrated probability method to identify cracks
6. Gini coefficient to identify cracks
7. Fractal dimension method to identify cracks
8. Characteristic parameter method
Research Background:
Fracture is a very important type of storage space. Most of them are distributed in hard formations with high base rock resistivity. It not only has storage capacity, but is also the main channel for oil and gas seepage.
The amount of oil and gas contained in the fracture accounts for about 30% of the total amount of oil and gas that has been detected, which also shows how important and practical it is to study imaging logging technology.
For geophysical exploration of fractured oil and gas reservoirs, imaging logging technology is the most advanced and intuitive method, but its measurement cost is very high, resulting in a limited penetration rate.
The logging data of most domestic oil and gas fields are still based on conventional logging data. Therefore, in the absence of imaging logging data, the technology of using conventional logging data to effectively identify fractures becomes very important.
Ten methods:
1. Double lateral logging identification
2. Resistivity recognition
3. Porosity recognition
4. Acoustic time difference response characteristics to identify cracks
5. Integrated probability method to identify cracks
6. Gini coefficient to identify cracks
7. Fractal dimension method to identify cracks
8. Characteristic parameter method
9. Wavelet transform method
10. Neural network method
1. Double lateral logging identification
Fig.1 Response characteristics of cracks on double lateral curves
The occurrence of cracks is directly related to the "difference" between the deep and shallow sides:
High-angle (generally above 75°) cracks, showing "positive difference" in both lateral directions;
Low-angle (generally below 60°) cracks, with “negative differences” in both sides
There is a “negative difference” between the sides of the crack at 45°, and the difference is the largest; at 60°-70°, the difference between the sides is small or no difference.
The more developed the fractures, the greater the fracture opening, fracture density, fracture porosity, and the radial extension depth of the fractures, and the greater the decrease in the resistivity of the dual lateral logging resistivity compared with the matrix rock.
2. Resistivity recognition
For example, the resistivity of tight limestone and gypsum of the Jialingjiang Formation is above 2000-10000ohm.m. When there are cracks, the resistivity decreases due to the influence of mud intrusion. The more developed the cracks, the deeper the mud intrusion, the more obvious the decrease in resistivity.
algorithm:
Frd: crack development degree Rmax: tight layer resistivity 2000-8000ohm.m
Rmin: Resistivity of the most developed cracks is 30-50ohm.m
3. Porosity recognition
In carbonate formations, neutron logging mainly reflects the total porosity of the formation, and acoustic logging mainly reflects the primary porosity of the formation, and the secondary porosity is mainly caused by fractures.
algorithm:
4. Acoustic time difference response characteristics to identify cracks
Acoustic jet lag generally does not reflect vertical fractures, but it responds more obviously to horizontal cracks, low-angle fractures and abnormally developed high-angle fractures, manifested as an abnormal increase in acoustic jet lag, or even cycle jumps.
Figure 2 Identifying cracks using acoustic time difference response characteristics
Predecessors made regression analysis on GR and RT, DT of the reservoir
Figure 3 Use of natural gamma and acoustic time difference map to identify cracks
And calculate the fracture development index by the parameters obtained from the above formula
Use the above formula to calculate the fracture development index of all wells in the work area, and obtain the comprehensive fracture evaluation diagram of each well:
Figure 4 Using the crack index to identify cracks
5. Integrated probability method to identify cracks
1) Through the analysis of different fracture parameters, the parameter characteristics of the core well core layer are obtained, and then the development thickness hi that affects the fracture is obtained, where i=1,2,...,m (m is the number of characteristic parameters representing the fracture) ；
2) After calculating the characteristic parameters, the ratio pi of the crack thickness can be obtained;
3) Reflect the weight coefficient wi of the crack by calculating each characteristic parameter;
4) Find the characteristic parameters and consider the comprehensive index CWP of fracture development;
Taking the sonic time difference in a certain depth section of a certain well as an example (Figure 5), the fracture development sections predicted by the sonic change rate are basically where the sonic time difference changes drastically, which is more consistent with the core data.
Figure 5 Acoustic time difference and crack probability
6. Gini coefficient to identify cracks
Set the value of a certain logging curve in a certain depth section as t1, t2, t3,...,ti,...,tn, and arrange it from small to y1, y2, y3,...,yi,...,yn, which represents The depth intervals are △d1,△d2,...△d3,…,△di,△dn, then let:
The Gini curve function of this layer can be obtained as:
Where is:
The Gini coefficient reflects the heterogeneity of the formation: as shown in Figure 6, the abscissa represents the depth interval where the logging value is located, and its size can be taken from 0 to 1; the ordinate represents the contribution value of the logging value, which is also taken from 0 to 1.
The curve AEC can represent the Gini curve. The curve enclosed with the straight line AC represents the degree of heterogeneity, and the larger the area, the more uneven the logging value, and the heterogeneity will increase accordingly.
When the area between the curve and the diagonal AC is 0, it indicates that the rock formation is uniform. Given that S2=1/2, the Gini coefficient is:
Figure 6 Gini coefficient of logging curve
7. Fractal dimension method to identify cracks
In the formula, l is the size of the divided box; f(q) and a(q) are calculated for each given q value, and the f(q)-a curve is the fractal spectrum of the multifractal.
It can be seen that the fractal knife value in the structural fracture of the reservoir can express its development degree, and the larger the value, the more obvious the fracture development.
The way to calculate the dimensions of the logging curve is to process the logging curve planes that need to be studied, that is, to draw all the curves in the logging diagram on a two-dimensional plane paper, and to mark the scale and dividing lines to perform the grid After the transformation, the number of grids N(L1},N(L2),...,N(Li) passed by the statistical curve, then:
In the formula, n is the number of curve data points in the interval;
Li: Represents the number of grids obtained after the i-th pass through grid encryption, 1ijn;
V: Logging value, resistivity logging can generally be expressed by the logarithm of the measured value.
Statistical relationship:
The logarithm is:
This equation can be used to fit Equation 4 to obtain the fractal dimension value D.
Tang Xiaomei et al. analyzed conventional logging data from the 1888-1925m well section at the bottom of the Chang 7 oil layer group in Well Zhuang 30, Heshui area, Xifeng Oilfield (Figure 7), and used the integrated probability index method and fractal dimension method to explain the location and extent of fractures. It can be seen by comparing it with the fracture interpretation results of imaging logging.
Using conventional logging data to identify the development location and extent of fractures in shale oil reservoirs is in good agreement with the fractures described in imaging logging interpretation, with a coincidence rate of over 85%.
Using conventional logging data to identify the development location and extent of fractures in shale oil reservoirs is in good agreement with the fractures described in imaging logging interpretation, with a coincidence rate of over 85%.
Fig. 7 Comparison of fracture interpretation results from conventional logging and imaging logging interpretation results (Well Zhuang 30, depth 1888-1925m)
8. Characteristic parameter method
1). Rock pore structure index:
As a constant in Archie’s formula, m can directly reflect the degree of rock bending. The more the rock fractures, the smaller the value of m. Therefore, the m value of the formation calculated from logging data can identify the development of fractures.
A and B are empirical coefficients, obtained by statistical methods;
F is the stratigraphic factor;
φ is the porosity, which is obtained by acoustic velocity or other methods.
2) Stratum factor ratio method:
R w is the resistivity of formation water; tR is the deep lateral resistivity; φ is the porosity, which is obtained by acoustic velocity or other methods.
In fractured formations, the more developed the fractures, the larger the FFR, and vice versa.
The formation factor F is usually affected by the pore complexity, formation lithology and porosity. When the lithology is fixed, F mainly depends on two other factors.
It is possible to find out that the formation factors behave differently on different rocks through related logging methods, and the ratio of formation factors is introduced to measure.
3) Three porosity ratio method:
Where:
The larger the Rp, the more secondary pores, that is, the more dissolved pores or cracks.
9. Wavelet transform method
Wavelet transform is similar to Fourier transform. It decomposes the signal into a combination of a series of wavelets, thereby transforming the time domain signal into the wavelet domain.
Wavelet transform has two processes: forward transformation and inverse transformation. The general signal processing process is to first transform the time domain signal to the wavelet domain, then modify the combination coefficients in the wavelet domain, and finally perform the inverse transformation to achieve the purpose of signal processing. .
Wavelet transform is divided into many kinds according to the different wavelet functions. For example, Mallat wavelet basis function, its fast algorithm is:
Positive transformation (decomposition)
Inverse transformation (reconstruction)
In the formula, h0 and h1 are the coefficients calculated by the wavelet function, which are generally understood as filter coefficients; c and d are the combination coefficients, the c coefficient represents the energy information of the low-frequency components in the time domain signal, and the d coefficient represents the energy of the high-frequency components information.
Figure 8 is a comprehensive schematic diagram of the 1465-1515m wavelet analysis of the main hole of the Chinese Continental Scientific Drilling.
The first three tracks in the figure are conventional logging curves. The imaging logging data show that there are many fractures in this interval, see track 7.
Well diameter expansion is more obvious and there are many intervals. In the interval with varying well diameter, the density curve becomes smaller and the neutron curve becomes larger, which is an obvious fracture development interval.
Figure 9 Comprehensive schematic diagram of formation fracture zone detected by wavelet coefficients (1465~1515)
Reconstruct the density log using resistivity, borehole diameter and sonic time-difference logging curves. The fourth track is the fracture synthesis index curve rbio-2.2 wavelet decomposition of the highest frequency wavelet coefficients in 5 layers, and the fifth track is the high frequency wavelet coefficients. In the energy curve, the sixth track is the formation fracture zone detected by this method and the fracture density calculated by 9; the seventh track is the fracture zone and fracture density shown by the imaging logging data.
Table 1 is a comparison table of wavelet transform detection results and imaging logging data. The two have good consistency.
Table 1 Comparison table of imaging logging and wavelet detection fracture zone
10. Neural network method
Based on the model calculation of the logging curve, the neural network method is used to identify the pattern point by point, and the fracture description results of the core are used as learning samples and verification conditions to establish a fracture interpretation model.
Figure 10 Schematic diagram of neural network
Interpretation results from fracture logging:
1) Well 2 has a larger fracture density than Well 1. Core data and fracture gradients show that Well 2 mainly develops micro-fractures and capillary fractures (mainly diagenetic fractures), and Well 1 mainly develops significant fractures (mainly structural fractures);
2) Well 2 mainly develops micro-fractures and capillary fractures. Conventional logging has limited longitudinal resolution. The interpretation results of logging fractures are relatively inconsistent with the thin-section observation results, which are consistent with the actual fracture development of the formation.
Figure 12 Forecast results
Reprinted from the public account: Petroleum Technology