CN117828458A - Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient - Google Patents
Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient Download PDFInfo
- Publication number
- CN117828458A CN117828458A CN202211176093.4A CN202211176093A CN117828458A CN 117828458 A CN117828458 A CN 117828458A CN 202211176093 A CN202211176093 A CN 202211176093A CN 117828458 A CN117828458 A CN 117828458A
- Authority
- CN
- China
- Prior art keywords
- resistivity
- gradient
- baoceng
- kong
- porosity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 239000011435 rock Substances 0.000 title claims abstract description 40
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 title claims abstract description 31
- 238000011158 quantitative evaluation Methods 0.000 title claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 43
- 230000002457 bidirectional effect Effects 0.000 claims abstract description 37
- 238000004364 calculation method Methods 0.000 claims abstract description 34
- 238000002790 cross-validation Methods 0.000 claims abstract description 20
- 238000010606 normalization Methods 0.000 claims abstract description 20
- 230000002441 reversible effect Effects 0.000 claims abstract description 18
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 6
- 238000013459 approach Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 abstract description 23
- 230000000704 physical effect Effects 0.000 abstract description 12
- 238000007781 pre-processing Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 37
- 230000004044 response Effects 0.000 description 20
- 230000000694 effects Effects 0.000 description 15
- 239000011148 porous material Substances 0.000 description 12
- 230000008569 process Effects 0.000 description 9
- 238000012512 characterization method Methods 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 6
- 238000005457 optimization Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000000877 morphologic effect Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000009897 systematic effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 230000002411 adverse Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000000052 comparative effect Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 239000010459 dolomite Substances 0.000 description 2
- 229910000514 dolomite Inorganic materials 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 235000019738 Limestone Nutrition 0.000 description 1
- 241000361919 Metaphire sieboldi Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 229910052925 anhydrite Inorganic materials 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- OSGAYBCDTDRGGQ-UHFFFAOYSA-L calcium sulfate Chemical compound [Ca+2].[O-]S([O-])(=O)=O OSGAYBCDTDRGGQ-UHFFFAOYSA-L 0.000 description 1
- 125000005587 carbonate group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 229910003460 diamond Inorganic materials 0.000 description 1
- 239000010432 diamond Substances 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000006028 limestone Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 125000006850 spacer group Chemical group 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradients, which comprises the steps of preprocessing a single well resistivity curve through interwell standardization, and then calculating a resistivity forward gradient and a resistivity reverse gradient to obtain a resistivity bidirectional gradient difference which is used as an important feature of model training; and finally, on the basis of normal normalization of all the characteristics, training by using a support vector regression method with a kernel function as a radial basis function and taking acoustic wave time difference, resistivity logarithm and resistivity bidirectional gradient difference as characteristics through a k-fold cross validation mode to obtain a quantitative calculation model with high Kong Baoceng porosity. According to the invention, the resistivity gradient related parameters are added into the nonlinear model training of support vector regression, so that the nonlinear calculation capability of the model is enhanced, the identification degree of high Kong Baoceng and the calculation accuracy of porosity are improved, and an important basis is provided for reservoir physical property characteristics and high-quality reservoir evaluation.
Description
Technical Field
The invention relates to the field of oil and gas logging interpretation and reservoir evaluation, in particular to a carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient.
Background
Oil and gas are usually present in the pores of a subsurface reservoir, the proportion of pore volume to the total volume of the rock, i.e. the porosity, is of great significance for oil and gas exploration and development: when the porosity of the reservoir is large, the reservoir has strong reservoir capacity for oil and gas, the rock contains oil and gas in high abundance, the exploration value of the oil and gas reservoir is larger, and meanwhile, the reservoir with high porosity generally has better permeability, so that the oil and gas is easier to extract from the reservoir, and the extraction cost is lower. In summary, high porosity reservoirs are a major concern for oil and gas exploration and development.
The porosity of a reservoir is typically calculated using the response of the log, which is often affected by various factors such as the borehole environment, the layer thickness, the longitudinal resolution of the log, etc., and deviations in the calculation result, especially when the layer Kong Chu is thin, the log response cannot completely exclude the influence of the adjacent layers, resulting in underestimation of the porosity. For carbonate reservoirs, such as a tidal flat-phase dolomite reservoir, the lithology and physical properties of the reservoir show strong heterogeneity in the longitudinal direction along with the periodical change of sea level, and a high Kong Chu layer usually appears in the form of thin strips and is limited by layer thickness, so that the reservoir is lost in the evaluation process or is underestimated in the whole or the porosity of the reservoir is underestimated, and therefore, the reservoir which has important significance for exploration and development is difficult to evaluate qualitatively and quantitatively accurately.
Methods commonly used for porosity calculation are all considered: the logging response at a depth point is related only to the petrophysical properties at that depth point, and not to the petrophysical properties at other depth points. I.e. the logging response is considered to have a unique correspondence (point at depth) and exclusivity (point at other depths) with the rock properties at depth. In practice, however, the logging instrument has a certain detection radius in the longitudinal direction, and when data is acquired at a certain depth point, rock information of adjacent depth cannot be avoided to be acquired, so that the logging response at the certain depth point is actually an overall response to the rock properties within a certain depth range. For a thick and heterogeneous layer, the rock properties of adjacent depth points are small in difference, the influence is negligible, but for a depth point where a high Kong Baoceng is located, the logging response is influenced by the rock properties of the dense adjacent layer and deviates from the real situation of the depth point obviously, the calculated porosity by using the method is low to different degrees, and great difficulty is brought to qualitative and quantitative evaluation of a high-hole thin layer.
Disclosure of Invention
The invention aims to overcome the problem that the logging response is influenced by the property of the tight adjacent layer rock and obviously deviates from the real situation of the depth point, and the calculated porosity is lower by using the prior art, and provides a carbonate rock high Kong Baoceng quantitative evaluation method based on resistivity gradient.
The height Kong Baoceng refers to a stratum with porosity of more than 5% and layer thickness of less than 2m.
In order to achieve the above object, the present invention provides the following technical solutions:
a carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient, which comprises the following steps:
step a, logging well and obtaining a resistivity curve, and carrying out interwell standardization on the resistivity curve to obtain a standardized resistivity logarithmic curve;
step b, calculating a resistivity forward gradient and a resistivity reverse gradient according to the normalized resistivity logarithmic curve to obtain a resistivity bidirectional gradient difference;
and c, taking the core porosity as a fitting object, and taking the acoustic time difference, the resistivity logarithm and the resistivity bidirectional gradient difference as characteristics to establish a carbonate rock height Kong Baoceng porosity quantitative evaluation model based on support vector regression.
According to the evaluation method, through interwell standardization of the resistivity curve, systematic errors caused by inconsistent performance, graduation, operation and the like of a logging instrument are eliminated, and the comparability of multi-well data and the multi-well logging interpretation precision are improved; the evaluation method uses the concept of resistivity gradient, takes the resistivity bidirectional gradient difference as the training characteristic of a machine learning algorithm, effectively introduces the resistivity information of the adjacent depth point into the porosity calculation of the current depth point, intuitively characterizes the resistivity morphological characteristic and the relation between the resistivity morphological characteristic and the porosity and the layer thickness, reduces the adverse effect of the disadvantage of insufficient longitudinal resolution capacity of a logging curve on the high Kong Baoceng porosity calculation, and improves the characterization capacity of the resistivity information on the high Kong Baoceng; according to the evaluation method, the rock core porosity is used as a fitting object, the acoustic wave time difference, the resistivity logarithm and the resistivity bidirectional gradient difference are used as characteristics, a carbonate rock height Kong Baoceng porosity quantitative evaluation model based on support vector regression is established, a complex nonlinear conversion relation between the acoustic wave time difference, the resistivity and the resistivity bidirectional gradient difference and the porosity is facilitated to be established rapidly and accurately, and the quantitative evaluation precision of the porosity of Kong Baoceng is improved.
Preferably, the step a is based on the high-resistance mark layer distributed in the region, and the original resistivity curve R 0 Carrying out interwell standardization to obtain a standardized resistivity logarithmic curve R, wherein the high-resistance mark layer is a mark layer with resistivity greater than 50000 Ω & m, and the correction formula is as follows:
wherein R is i The numerical value of the resistivity logarithmic curve R at the ith depth point after interwell standardization is dimensionless; r is R 0i For the original resistivity curve R 0 Values at the ith depth point, Ω.m;resistivity characteristic values of the high-resistance mark layers distributed in the areas are omega.m; r is R H To correct the resistivity characteristic value of the high-resistance layer of the well, omega.m.
The resistivity of carbonate rock greatly fluctuates among different reservoirs, the resistivity and the physical properties of the reservoirs often show an exponential relationship, and the nonlinear complexity of the resistivity and the physical properties of the reservoirs is reduced by taking the logarithm of the resistivity before interwell standardization, so that the fitting effect and the model training speed in the modeling process are improved.
Preferably, the resistivity forward gradient, reverse gradient and bi-directional gradient difference calculation method in the step b is as follows:
the method for calculating the forward gradient comprises the following steps:
G f ＝R d+Δs -R d
wherein G is f Is positive gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R after the standardization of the well is d, and the numerical value is dimensionless; Δs is the gradient spacing, m; r is R d+Δs The numerical value of a resistivity logarithmic curve R after interwell standardization at the depth of d+delta s is dimensionless;
the calculation method of the reverse gradient comprises the following steps:
G o ＝R d -R d-Δs
wherein G is o Is reverse gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d; Δs is the gradient spacing, m; r is R d-Δs The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d-deltas;
the calculation method of the bidirectional gradient difference comprises the following steps:
G d ＝G f -G o
wherein G is d Is a bidirectional gradient difference and is dimensionless; g f Is positive gradient and dimensionless; g o Is reverse gradient and has no dimension.
Preferably, the gradient interval Δs ranges from: 1m < 2 Deltas < 2m.
Preferably, the gradient interval Δs=0.7m.
According to the evaluation method, the carbonate rock height Kong Baoceng is quantitatively evaluated based on the resistivity gradient, a porosity calculation method of the traditional well logging curve and the calculated porosity, which corresponds to each other in depth, is broken through, the characteristic that the well logging response of the high Kong Baoceng, especially the carbonate rock, is affected by an adjacent high-resistance layer is fully considered, the degree of actually measured resistivity is higher by introducing the concepts of the resistivity forward gradient, the resistivity reverse gradient and the bidirectional gradient difference, the resistivity bidirectional gradient difference is used as the characteristic, and the porosity calculation accuracy of the high Kong Baoceng is improved.
Preferably, the step c includes the steps of:
c1. respectively carrying out normal normalization on each characteristic curve, and respectively obeying standard normal distribution by each characteristic sample for modeling, wherein the calculation formula is as follows:
wherein X is 0i 、X 0j Respectively a certain characteristic original sample set X 0 Values at the i and j sample points;an average value of the original sample set on a certain characteristic; m is the total number of training set samples; x is X s Is a standard deviation conversion coefficient, and is dimensionless; x is X i Ith in normalized sample set X for a featureThe value of the sample point;
c2. and fitting the sample after normal normalization by using a support vector regression model to obtain a quantitative calculation model with high Kong Baoceng porosity.
The distribution range of each characteristic sample is as close as possible by respectively carrying out normal normalization on each characteristic curve, and the numerical values of each characteristic of the sample are unified under the same standard; the support vector regression model is used for fitting the samples after normal normalization, partial middle and low hole and dense reservoir data points can be better removed, more attention is paid to the high Kong Baoceng with higher porosity fitting difficulty, the finally obtained model is not easy to be disturbed by sample disturbance, better characterization capability is provided for a reservoir with strong heterogeneity, and the quantitative porosity evaluation precision of the reservoir is improved.
Preferably, the support vector regression model in step c2 uses three features of acoustic wave time difference, resistivity logarithm and resistivity bidirectional gradient difference after normal normalization.
The acoustic time difference and the resistivity are functions of the porosity, the porosity of the reservoir can be represented to a certain extent, and the representation accuracy can be further improved by adding the resistivity bidirectional gradient difference into the fitting characteristic.
Preferably, the support vector regression model is trained using a k-fold cross validation approach.
Preferably, in k-fold cross validation, k=3.
The k-fold cross validation mode is used for optimizing the super parameters in the support vector regression model training process, and each group of super parameter combinations is trained by using the method, so that the super parameter combination with the highest regression accuracy is the optimal super parameter combination, and the variable which subjectivity brings to model accuracy in the super parameter selection is reduced.
Preferably, the kernel function used by the support vector regression model is a radial basis function.
The kernel function is used as a radial basis function, the optimization problem is mapped to the feature space in a kernel function mode and then solved, and the calculation amount for solving the support vector regression optimization problem is reduced.
Compared with the prior art, the invention has the beneficial effects that:
1. according to the evaluation method, the logarithm of the resistivity is taken before interwell standardization, so that the nonlinear complexity of the resistivity and reservoir physical properties is reduced, and the fitting effect and the model training speed in the modeling process are improved. By carrying out interwell standardization on the resistivity curve, the system error caused by inconsistent performance, graduation, operation and the like of the logging instrument is eliminated, and the comparability of multi-well data and the multi-well logging interpretation precision are improved.
2. The evaluation method uses the concept of resistivity gradient, takes the resistivity bidirectional gradient difference as the training characteristic of a machine learning algorithm, effectively introduces the resistivity information of the adjacent depth point into the porosity calculation of the current depth point, intuitively characterizes the resistivity morphological characteristic and the relation between the resistivity morphological characteristic and the porosity and the layer thickness, reduces the adverse effect of the disadvantage of insufficient longitudinal resolution capability of a logging curve on the high Kong Baoceng porosity calculation, and improves the characterization capability of the resistivity information on the high Kong Baoceng.
3. According to the evaluation method, the rock core porosity is used as a fitting object, the acoustic wave time difference, the resistivity logarithm and the resistivity bidirectional gradient difference are used as characteristics, a carbonate rock height Kong Baoceng porosity quantitative evaluation model based on support vector regression is established, a complex nonlinear conversion relation between the acoustic wave time difference, the resistivity and the resistivity bidirectional gradient difference and the porosity is facilitated to be established rapidly and accurately, and the quantitative evaluation precision of the porosity of Kong Baoceng is improved.
Drawings
FIG. 1 is a flow chart of a method for quantitatively evaluating carbonate rock height Kong Baoceng based on resistivity gradient;
FIG. 2 is a schematic diagram of the response of the high hole sheet resistivity logging in accordance with the present invention;
FIG. 3 is a conceptual diagram of the related gradient of resistivity according to the present invention;
FIG. 4 is a graph of the relationship between the bidirectional gradient difference of resistivity and the core porosity for different types of reservoirs, logarithms of resistivity and values of different deltas according to the present invention;
FIG. 5 is a histogram of the numerical distribution of each feature of sample points of an investigation region in the present invention;
FIG. 6 is a flow chart of model training and verification using the k-fold cross-validation method of the present invention;
FIG. 7 is a graph comparing the porosity calculated using the method of the present invention with the porosity calculated using method A;
FIG. 8 is a graph comparing the calculated porosity using the method of the present invention with the porosity calculated using method B.
Detailed Description
The present invention will be described in further detail with reference to test examples and specific embodiments. It should not be construed that the scope of the above subject matter of the present invention is limited to the following embodiments, and all techniques realized based on the present invention are within the scope of the present invention.
Example 1
A carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient, which comprises the following steps:
step a, logging and obtaining a resistivity curve, taking a high-resistance mark layer widely distributed in a region as a standard, and comparing an original resistivity curve R 0 Carrying out interwell standardization to obtain a standardized resistivity logarithmic curve R, wherein the high-resistance mark layer is a mark layer with resistivity greater than 50000 Ω & m, and the correction formula is as follows:
wherein R is i The numerical value of the resistivity logarithmic curve R at the ith depth point after interwell standardization is dimensionless; r is R 0i For the original resistivity curve R 0 Values at the ith depth point, Ω.m;the resistivity characteristic value of the high-resistance mark layer is omega.m, which is widely distributed in the area; r is R H To correct the resistivity characteristic value of the high-resistance layer of the well, omega.m.
The purpose of interwell normalization is to correct well-to-well log systematic errors. In general, for an oil and gas field, during long-term exploration and development, it is difficult to ensure that all well logging curves are of the same type or are calibrated in the same environment, so that errors mainly based on calibration factors are unavoidable between logging data of each well. Meanwhile, even if the instrument types are the same and the scale conditions are similar, it is difficult to ensure that the well bore conditions are the same in actual well logging, and systematic errors are caused in well logging curves between wells when the well bore is irregular or the used mud parameters are different. Therefore, when data analysis of multiple wells is involved, the logging data of different wells needs to be normalized among the wells, and the corrected curve can be considered to be measured under identical scale environment and experience conditions.
In this embodiment, because the acoustic wave instrument has a simpler measurement principle and a fixed instrument frequency, the possibility of obvious system errors between wells is relatively small, so that the acoustic wave curves do not need to be normalized among wells. For the resistivity curve, on one hand, the difference of the resistivity of water and rock is far greater than the difference of the acoustic wave time difference of water and rock, so that the difference of scale conditions or borehole environments easily causes abnormal disturbance of the resistivity curve, and on the other hand, the embodiment establishes a high Kong Baoceng porosity quantitative calculation model in the subsequent steps, and the 3 characteristics used for the resistivity have two characteristics, so that the accuracy of the resistivity is more required. Thus, there is a need for interwell normalization of resistivity curves.
In this embodiment, the study object is carbonate rock, in the carbonate rock section, dense limestone or anhydrite is generally used as a standardized marking layer, and because the porosity of the two is extremely low, the physical properties of different areas of the oil field cannot be obviously changed, and the difference of logging curves of different wells caused by the difference of the physical properties of the rock can be eliminated, so that the accuracy of the standardization between wells is improved. If the lithology of pore development is chosen as the marker layer, the differences in log curves between different wells may be either systematic errors or differences in the physical properties of the rock itself, which may not facilitate the determination of the standardized corrections.
The resistivity is logarithmic before interwell standardization, mainly because carbonate rock is influenced by the huge resistivity difference and strong heterogeneity of water and rock skeletons, the resistivity fluctuates greatly between reservoirs and non-reservoirs, good reservoirs and poor reservoirs, and the resistivity and the reservoir physical properties often show an exponential relationship. The method has the advantages that the log processing is carried out on the resistivity, the nonlinear complexity of the resistivity and the physical properties of the reservoir can be reduced, and the fitting effect and the model training speed in the modeling process are improved.
The calculation method of the well standardization is a mark layer ratio method, as described by a correction formula, the original resistivity curves R are respectively processed 0 Resistivity of high resistance mark layer widely distributed in regionAnd correcting the high-resistance layer resistivity R of the well H Taking the logarithm and then using the method of the mark layer ratio by plotting the original resistivity logarithm curve lgR 0 Multiplied by this ratio coefficient, and corrected to a logarithmic resistivity curve R meeting the unified criteria of the area under investigation.
And b, calculating a resistivity forward gradient and a resistivity reverse gradient according to the normalized resistivity logarithmic curve, and obtaining a resistivity bidirectional gradient difference on the basis.
The traditional porosity calculation method considers that the logging curve and the calculated porosity have a one-to-one correspondence in depth, namely the calculated porosity value with the depth point being iLogging response vector X corresponding to the depth point only i Related (X) i Is a vector, each element in the vector is the value of each log response of the depth point) and is associated with the log response of the adjacent depth point (such as X i+1 、X i-1 ) Irrespective of the fact that the first and second parts are. In practice, since the logging instrument is longitudinally of a certain detection radius, when the logging instrument detector is aligned to a depth iWhen the stratum is formed, the acquired signals are not only from the depth i, but also from the depths i-1 and i+1, and even from the stratum which is farther away from the depth i, so the conventional resistivity method is established +.>The one-to-one correspondence is actually inaccurate. Normally, the difference caused by the reasons is negligible, because the reservoir is thicker and is obviously larger than the longitudinal detection radius of the logging instrument, and for a certain depth point in the middle of the thick reservoir, the logging response of the adjacent depth point does not cause obvious interference to the logging response of the current depth point because the rock of the adjacent depth point is close to the physical property of the rock. However, for high Kong Baoceng, the above difference cannot be ignored because the layer thickness is close to or even smaller than the detection radius of the logging instrument, resulting in a thin layer of logging response being affected by an adjacent high-resistance layer, especially for carbonate rock, further resulting in a high Kong Baoceng logging response deviating from reality due to the large difference in physical properties between the dense and high pore layers.
Since the logarithm of resistivity is a value obtained by taking the logarithm of resistivity, the change trend and degree thereof have consistency with the resistivity, and in fig. 2, 3 and related descriptions, the "logarithm of resistivity" is collectively referred to as the resistivity for simplicity of description and convenience of understanding.
As shown in FIG. 2, where R is the longitudinal detection radius of the instrument, R i 、R j 、R k True resistivity of the formation at depths i, j, k, R' i Is the actual measured value (non-real value) of resistivity at depth i. When the midpoint of the logging instrument detector is opposite to the depth i, the measured resistivity R 'recorded by the midpoint is recorded' i In effect is a weighted sum of formation resistivity between the depth j (j=i-r) and the depth k (k=i+r), i.e. the light grey area in the figure. As can be seen from the figure, the low-resistance reservoir section of the normal thick layer, namely the concave section in the figure, has larger thickness and R j 、R k The difference in resistivity values between them is small, so that R 'is measured' i Record value and true value R i Proximity. The high-hole thin layer section in the figure has thin thickness due to the reservoir layer, namely the peak section in the figure, R j 、R k The measured resistivity R 'recorded at this point varies greatly' i Is significantly affected by the adjacent depth point, resulting in serious deviation from the resistivity true value R of the point i 。
In view of this, in calculating the porosity of a target depth point, it is necessary to introduce a logging response adjacent to the depth point. The introduction of the logging response of the adjacent depth point is not intended to further deepen the effect of the adjacent depth point on the resistivity of the target depth point, but rather to eliminate its effect on the target depth point as much as possible by modeling. For this embodiment, a method of gradient resistivity of the current depth point and the adjacent depth point is used to introduce the adjacent depth point log response into the model. Three concepts are introduced first: resistivity forward gradient, reverse gradient, bi-directional gradient difference. The method for calculating the forward gradient comprises the following steps:
G f ＝R d+Δs -R d
wherein G is f Is positive gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d; Δs is the gradient spacing, m; r is R d+Δs The value of the resistivity logarithmic curve R is dimensionless when the depth is d+deltas;
the calculation method of the reverse gradient comprises the following steps:
G o ＝R d -R d-Δs
wherein G is o Is reverse gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d; Δs is the gradient spacing, m; r is R d-Δs The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d-deltas;
the calculation method of the bidirectional gradient difference comprises the following steps:
G d ＝G f -G o
wherein G is d Is a bidirectional gradient difference and has no dimension.
As shown in FIG. 3, for a normal thick layer, the resistivity of adjacent depth points within the low-resistance reservoir is not greatly different, and the gradient is forward and reverseThe absolute value of the difference between the two-way gradient and the two-way gradient is smaller. For a high Kong Baoceng, the sheet resistivity is characterized by a spike-like characteristic with large absolute values of both forward and reverse gradients, for example, the resistivity value R at the depth point d d Resistivity value R smaller than delta s of upper and lower depth interval d-Δs And R is d+Δs Therefore, resistivity is positive with gradient G f Positive value, reverse gradient G o Negative value, bi-directional gradient difference G d And presents a positive value with a large absolute value. The larger bi-directional gradient difference in resistivity of the latter is a reflection that the measured resistivity due to the influence of the layer thickness is higher than the true resistivity, compared to the normal thick layer and the high Kong Baoceng.
The resistivity and the porosity are inversely related, and a higher measured resistivity can result in a lower porosity calculated using conventional methods. According to the concept and principle related to the resistivity gradient, the bidirectional gradient difference is capable of indicating the higher degree of the measured resistivity. The two-way gradient difference of the resistivity is taken as a characteristic to be added into a training sample, and the subsequent model training in the embodiment can enable the training sample to have resistivity correction capability, so that a certain compensation effect can be achieved on lower porosity.
In the calculation of the resistivity forward gradient and the reverse gradient, a gradient interval deltas needs to be determined, and the value plays an important role in representing the higher degree of the measured resistivity by the resistivity bidirectional gradient difference. An excessively large deltas may result in a loss of the ability of the resistivity bi-directional gradient difference to characterize the measured resistivity of the sheet, while an excessively small deltas may be excessively sensitive to curve changes. Since the longitudinal resolution of the resistivity logging tool is between about 1 meter and 2 meters, 1m < 2 Δs < 2m.
In actual calculation, it is necessary to determine a most appropriate Δs value in the interval to enhance the accuracy of quantitative evaluation as much as possible. The log resistivity versus core porosity is shown in fig. 4a, where diamond data points are high Kong Baoceng sample points and circular data points are low pore sample points, both of which are bounded by 5% porosity. The distribution interval of the low pore sample points on the resistivity feature is wider than the high Kong Baoceng sample points. The main section of the resistivity logarithm where the high Kong Baoceng sample point is located is set as a reference section, and as can be seen from the graph, the distribution range of part of the low-hole sample points on the resistivity logarithm characteristic is obviously different from the reference section, namely, the part of the sample points can be distinguished from the high Kong Baoceng sample points by only using the resistivity logarithm characteristic, and the type of the low-hole sample points is referred to as type 1 low-hole sample points; while the distribution range of the other part of low-hole sample points on the resistivity logarithmic feature almost completely falls in the reference interval, the sample points of the type are completely indistinguishable from the high Kong Baoceng sample points only by using the resistivity logarithmic feature, and the sample points of the type are called as the low-hole sample points of the type 2. Therefore, in order to improve the evaluation accuracy and exert the function of the characteristic of the resistivity bidirectional gradient difference to the greatest extent, a key principle of selecting a proper delta s value is to distinguish 2 types of low-pore sample points which are extremely easy to confuse with high-pore thin-layer sample points on the resistivity characteristic from high Kong Baoceng sample points as obviously as possible on the resistivity bidirectional gradient difference characteristic.
Taking the characteristics of delta s of 0.5m, 0.7m and 0.9m on the basis of 1m < 2 delta s < 2m, and testing the distinguishing degree of a high Kong Baoceng (core porosity > 5%) sample point and a class 2 low-hole sample point after using the characteristic of resistivity bidirectional gradient difference. Fig. 4b, 4c and 4d are the resistivity bi-directional gradient differences as a function of core porosity using Δs=0.5 m, Δs=0.7 m, Δs=0.9 m, respectively. The source of class 1 and class 2 low-hole sample points in the three figures is completely consistent with that of figure 4a, the solid line interval is the main distribution interval of class 2 low-hole sample points, and the dotted line interval is the main distribution interval of high Kong Baoceng sample points. As can be seen from the graph, when Δs=0.7m, the difference between the main distribution interval of the 2 types of low-hole sample points and the high Kong Baoceng sample points is the largest, and when Δs=0.5 m and Δs=0.9m, the difference is relatively small, that is, Δs=0.7m is considered as the extreme point with the resolution of Kong Baoceng, and the characteristic of the bidirectional gradient difference of the resistivity has the best effect on supplementing the logarithmic characteristic of the resistivity, so Δs=0.7m is taken.
And c, taking the core porosity as a fitting object, and taking the acoustic time difference, the resistivity logarithm and the resistivity bidirectional gradient difference as characteristics to establish a carbonate rock height Kong Baoceng porosity quantitative evaluation model based on support vector regression.
Still further, step c comprises the steps of:
c1. respectively carrying out normal normalization on each characteristic curve, and respectively obeying standard normal distribution by each characteristic sample for modeling, wherein the calculation formula is as follows:
wherein X is 0i 、X 0j Respectively a certain characteristic original sample set X 0 Values at the i and j sample points;an average value of the original sample set on a certain characteristic; m is the total number of training set samples; x is X s Is a standard deviation conversion coefficient, and is dimensionless; x is X i The values of the ith sample point in the sample set X after normalization for a certain feature.
The purpose of curve normalization is to make the distribution range of each characteristic sample as close as possible. Fig. 5 shows distribution diagrams of features of sample points of 5 wells from a research area, and it can be seen from the figures that the number range of acoustic wave time differences is more than 40-60, the number range of the logarithm of resistivity is more than 1-5, and the number range of the bidirectional gradient difference of resistivity is more than-3-2 for the features before normalization. If normalization is not performed, the information of the other features will be masked because the acoustic time difference value range is the largest, so that each feature needs to be normalized. As shown in fig. 5, the distribution of the samples in each feature is close to normal distribution, so the normalization method uses a normal normalization method, and after the above formula conversion, the samples obey normal distribution with a mean value of 0 and a variance of 1 in each feature, i.e. the values of each feature of the samples are unified under the same standard.
c2. And fitting the sample after normal normalization by using a support vector regression model to obtain a quantitative calculation model with high Kong Baoceng porosity.
Because the acoustic propagation speed and the conductivity of the pore fluid are different from those of the skeleton, the acoustic time difference and the resistivity are functions of the porosity, the size of the porosity of the reservoir can be represented to a certain extent, and meanwhile, the representation precision can be further improved by adding the resistivity bidirectional gradient difference into the fitting characteristic. Therefore, in this embodiment, the support vector regression model uses three features of acoustic wave time difference, resistivity logarithm and resistivity bidirectional gradient difference after normal normalization.
Support vector regression is an efficient way to fit data with complex nonlinear relationships, and the use of this method in this embodiment has the following reasons:
the relationship between the porosity and the acoustic wave time difference, the resistivity logarithm and the resistivity bidirectional gradient difference is not a simple linear relationship or a nonlinear relationship which can be described by a single formula, but an unknown complex nonlinear relationship, and when three features are used for regression of the porosity at the same time, the redundant relationship between the features cannot be described by using a simple mathematical relationship. Support vector regression can project data into a high-dimensional space, and the correlation between sample points is adaptively found in the high-dimensional space, so that the established model has higher accuracy and greater flexibility, and is suitable for representing complex nonlinear relations such as those in the embodiment.
A general fitting method requires calculation of errors for each data point, each error being used in the fitting to the parameters. The support vector regression algorithm is provided with a spacing band, when the difference between the fitting value and the true value is in the spacing band, loss is not counted, namely, errors of data points in the spacing band cannot be used in parameter fitting, so that a finally obtained model is not easy to be disturbed by sample disturbance, better characterization capacity is provided for a reservoir with strong heterogeneity, stability of the model is effectively maintained, and generalization and popularization capacity of the model is greatly enhanced.
In summary, since the main focus of the fitting is high Kong Baoceng porosity, and a significant portion of medium-low pore and dense reservoir data points exist in the sample points, the porosity of the reservoir is relatively easy to calculate due to the relatively large thickness, so that the porosity of the reservoir does not need to have too high fitting requirements, the fitting can be considered to be accurate when the fitting value is close to the target value in the model training process, and the error is not used for fitting the model parameters. The feature of support vector regression enables the model to pay more attention to the high Kong Baoceng with higher difficulty in porosity fitting, and is in principle helpful for improving the accuracy of quantitative evaluation of the porosity of the reservoir.
Support vector regression is in effect the finding of a hyperplane or hypersurface in a multidimensional space to achieve regression of data points. Taking the hyperplane as an example, its functional expression can be written as:
f(x)＝ω T x+b
wherein b is a constant term obtained by fitting, ω is a coefficient vector obtained by regression, i.e
ω＝(ω 1 ，ω 2 ，…，ω n )
Wherein each term in ω is a coefficient (slope) corresponding to each feature, and n is the total number of input features.
In the case of soft spacing, adding a regularization term, the optimization problem of support vector regression can be formalized as follows:
wherein ω is a coefficient vector, determining the slope of the fitting hyperplane on each feature; c is a regularization coefficient, and the influence degree of peripheral points on the model is controlled; e is the width of the interval band, and fitting losses are not counted in the width; l (L) ∈ Is a coefficient related to the width of the spacer tape; f (x) i ) Is a fitting value; y is i -sample true value.
According to the principle, the support vector regression algorithm can realize the advantages of being little influenced by peripheral data points, enhancing the stability and generalization capability of the model and the like through the control of parameters such as C, E and the like.
The process of support vector regression model training is actually the process of determining the kernel function, the optimal super-parameters and the model parameters omega.
The support vector regression optimization problem is usually solved directly by requiring a huge amount of computation, and in order to solve the problem, the optimization problem is usually mapped to a feature space by a kernel function mode and then solved. Thus, the kernel function is the first important variable of model accuracy. Radial basis functions and polynomial kernel functions are the two most commonly used functions. The expressions for the two kernel functions are:
K rbf (x i ，x j )＝exp(-γ||x i -x j || 2 )
wherein K is poly The polynomial kernel function is adopted, and d is the term number of the polynomial; k (K) rbf And gamma is a characteristic coefficient of the radial basis function. Which kernel function is used in an embodiment depends on the one hand on the accuracy of the regression and on the other hand on the efficiency of the regression.
In the machine learning algorithm, there are two general types of parameters, one type of parameter is a model parameter, such as each coefficient in a vector ω, and the other type of parameter is an output of model training; the other type of parameters are super parameters, such as C, E parameters in a support vector regression optimization formula and d and gamma parameters in a kernel function, wherein the parameters are manually provided for a model before training and are input for model training, and the super parameters are provided by the human body and have a certain subjectivity, so the super parameters are another important variable of model training precision of the kernel removal function.
In order to reduce the variable of model accuracy brought by subjectivity, the embodiment uses a k-fold cross validation mode to optimize the super-parameters, namely, a series of preset values are provided for each super-parameter, the preset values of different super-parameters are combined to form a plurality of super-parameter combinations, each combination is brought into the model to train and compare fitting accuracy until the super-parameter combination which enables regression accuracy to reach the highest is found, namely, the plurality of super-parameter combinations are input into the model before training, and the unique optimal super-parameter combination and the model parameter omega matched with the unique optimal super-parameter combination are obtained after training is finished.
As shown in fig. 6, the k-fold cross validation randomly and averagely divides the training data set into k parts, for each super parameter combination, k-1 parts are used as a training set, 1 part is used as a test set, one training is performed, then one part is used as a test set, one training is performed again, and the like until the k times of training are completed, namely the cross validation of one set of super parameter values is completed, and the average precision of the k times of training is recorded as the regression precision of the set of super parameter values. By using the method, each group of super-parameter combinations is trained, so that the super-parameter combination with the highest regression accuracy is the optimal super-parameter combination.
In this embodiment, if the radial basis function is used for training, the super parameters to be determined include C, e, γ; if a polynomial kernel function is used, the hyper-parameters to be determined include C, E and d. The optimizing of the super-parameters requires a plurality of cross-validation periods, because the interval where the optimal super-parameters are likely to appear is completely unknown before the first cross-validation period, a larger value range is usually given to the super-parameters during the first cross-validation, the range is gradually narrowed after the cross-validation period is carried out for several times, and the approximate range where the optimal super-parameters are likely to appear can be determined for different kernel functions before the last cross-validation period begins. Table 1 shows the pre-set values of the super parameters, the number of combinations and the effect comparison before the start of the last k-fold cross-validation (k=3) cycle.
Table 1 cross-validation (k=3) of hyper-parameters preset, number of combinations and effect comparison
As can be seen from Table 1, using the radial basis function as the kernel function, there are 3718 types of super-parameter combinations, the optimizing time is 6min54s, a set of optimal super-parameters and matched model parameters under the condition that the kernel function is the radial basis function are obtained, the optimal regression accuracy reaches 47.2%, using the polynomial kernel function, only 714 types of super-parameter combinations, the optimizing time reaches 53min54s, which is 7.8 times of the radial basis function, the optimal super-parameters and matched model parameters under the condition that the kernel function is the polynomial kernel function are obtained, but the optimal regression accuracy is only 43.25%. The radial basis function is used as a kernel function in this embodiment, since it has significant advantages over polynomial kernel functions in both training efficiency and accuracy.
In the above-mentioned super-parameter optimizing process, a k-fold cross-validation mode is used. In fact, k is also a super parameter, that is, the number of training sets into which the training set is split in the cross-validation process is required to be manually specified. In machine learning, the value of k is usually 3-5, in this embodiment, the values of k are respectively 3, 4 and 5, and a radial basis function is used, so that under the condition that the kernel function, the hyper-parameter value ranges of C, E, gamma and the like and the value number remain unchanged, the training efficiency and the training precision of the model when the values of k are different are compared as shown in table 2.
As can be seen from the table, the regression accuracy of the three k values is close, and k=3 is slightly higher; comparing the training time, it was found that the training time was significantly less for k=3 than for the other two values. In view of the above factors, for the present embodiment, when model training is performed using the k-fold cross validation method, the value of k is 3.
Table 2k model training efficiency and accuracy comparison table at different values
| k takes on value | Training time | Regression accuracy |
| 3 | 6min54s | 47.98％ |
| 4 | 11min8s | 47.73％ |
| 5 | 16min55s | 46.06％ |
Example 2
Taking a Sichuan basin earthworm gate mountain front area with a certain gas field dolomite reservoir as an example, taking a sample of a 5-well in a research area as a training data set and taking a Y1-well as a verification on the basis of the embodiment 1, so as to test the application effect of the evaluation method.
In this example, a conventional porosity calculation method a is provided as a comparative example, which uses directly the resistivity log, acoustic wave time difference, and resistivity log ratio as fitting features without using the resistivity gradient correlation concept; the modeling is directly performed by using a multi-element fitting mode without using a machine learning algorithm.
Fig. 7 shows the comparison of the porosity calculated by the method a (light grey dotted line in the figure), the porosity calculated by the evaluation method provided by the present invention (black solid line in the figure) and the core porosity (dark grey dots in the figure). In this example, the core porosity > 5% is used as a criterion for the high pore lamellae, and the figures are marked with the numbers (1) to (9), respectively.
As can be seen from FIG. 7, the porosities obtained by the evaluation methods of the present invention are close to those obtained by the method A in the layers (1), (2) and (8), while the porosities obtained by the evaluation methods of the present invention and the core porosities in the layers (3), (4), (5), (6), (7) and (9) have higher coincidence, accounting for 66.7% of the total layer number, and especially in the layers (5), (6) and (7), the accuracy advantage of the evaluation methods of the present invention is more obvious, accounting for 33.3% of the total layer number.
In this embodiment, a conventional porosity calculation method B is also provided as a comparative example, which uses directly the logarithm of resistivity and the acoustic wave time difference as fitting features without using the concept of resistivity gradient correlation; a support vector regression algorithm was used to build the porosity calculation model.
Fig. 8 shows the porosity calculated by method B (light grey dotted line in the figure), the porosity calculated by the evaluation method provided by the present invention (solid black line in the figure) and the core porosity (dark grey dots in the figure). In this example, the core porosity > 5% is used as a criterion for the high pore lamellae, and the figures are marked with the numbers (1) to (9), respectively.
As can be seen from FIG. 8, the porosities obtained by the evaluation methods of the present invention are close to those obtained by the B method in the layers (1), (2), (4), (8) and (9), while the porosities obtained by the evaluation methods of the present invention and the core porosities in the layers (3), (5), (6) and (7) have higher coincidence, accounting for 44.4% of the total layers, and especially in the layers (5), (6) and (7), the accuracy advantage of the evaluation methods of the present invention is more obvious, accounting for 33.3% of the total layers.
The resistivity gradient-based carbonate rock height Kong Baoceng quantitative evaluation method is compared with the conventional method A, and the difference is mainly as follows: firstly, the method A does not use the concept related to resistivity gradient as training characteristics; and secondly, the method A does not use a support vector regression machine learning algorithm. On one hand, the quantitative characterization capability of the model on the height Kong Baoceng is weakened, so that the calculated porosity is influenced by the layer thickness limitation, on the other hand, the complex nonlinear processing capability is also reduced, for example, the porosity calculated by the method A is lower than the porosity calculated by the evaluation method provided by the invention, and the quantitative characterization capability of the model on the height Kong Baoceng is not strong, so that the embodiment of the layer thickness influence and the complex nonlinear processing capability are lower. Compared with the method A, the method B adopts a support vector regression method, so that the complex nonlinear processing capacity is improved, and the calculated porosity is obviously improved in the layers (4) and (7), thus the characteristic is embodied. However, since the resistivity gradient was not used as a feature to reduce the layer thickness influence, the characterization ability of the partial heights Kong Baoceng such as (5) and (6) was still not as good as that of the evaluation method provided by the present invention, which also confirms the advantageous effects of the evaluation method provided by the present invention from the side.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the particular embodiments disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.
Claims (10)
1. The carbonate rock height Kong Baoceng quantitative evaluation method based on the resistivity gradient is characterized by comprising the following steps of:
step a, logging well and obtaining a resistivity curve, and carrying out interwell standardization on the resistivity curve to obtain a standardized resistivity logarithmic curve;
step b, calculating a resistivity forward gradient and a resistivity reverse gradient according to the normalized resistivity logarithmic curve to obtain a resistivity bidirectional gradient difference;
and c, taking the core porosity as a fitting object, and taking the acoustic time difference, the resistivity logarithm and the resistivity bidirectional gradient difference as characteristics to establish a carbonate rock height Kong Baoceng porosity quantitative evaluation model based on support vector regression.
2. The method for quantitative evaluation of carbonate rock height Kong Baoceng based on resistivity gradient according to claim 1, wherein said step a is based on a high-resistance mark layer distributed in a region, and is characterized by a raw resistivity curve R 0 Carrying out interwell standardization to obtain a standardized resistivity logarithmic curve R, wherein the high-resistance mark layer is a mark layer with resistivity greater than 50000 Ω & m, and the correction formula is as follows:
wherein R is i The numerical value of the resistivity logarithmic curve R at the ith depth point after interwell standardization is dimensionless; r is R 0i For the original resistivity curve R 0 Values at the ith depth point, Ω.m;resistivity characteristic values of the high-resistance mark layers distributed in the areas are omega.m; r is R H To correct the resistivity characteristic value of the high-resistance layer of the well, omega.m.
3. The method for quantitatively evaluating the carbonate height Kong Baoceng based on the resistivity gradient according to claim 1, wherein the resistivity forward gradient, the reverse gradient and the bidirectional gradient difference in the step b are calculated as follows:
the method for calculating the forward gradient comprises the following steps:
G f ＝R d+Δs -R d
wherein G is f Is positive gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R after the standardization of the well is d, and the numerical value is dimensionless; Δs is the gradient spacing, m; r is R d+Δs The numerical value of a resistivity logarithmic curve R after interwell standardization at the depth of d+delta s is dimensionless;
the calculation method of the reverse gradient comprises the following steps:
G o ＝R d -R d-Δs
wherein G is o Is reverse gradient and dimensionless; r is R d The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d; Δs is the gradient spacing, m; r is R d-Δs The numerical value of the resistivity logarithmic curve R is dimensionless when the depth is d-deltas;
the calculation method of the bidirectional gradient difference comprises the following steps:
G d ＝G f -G o
wherein G is d Is a bidirectional gradient difference and is dimensionless; g f Is positive gradient and dimensionless; g o Is in the opposite directionGradient, dimensionless.
4. A method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradients as claimed in claim 3, wherein the gradient interval Δs is in the range of: 1m < 2 Deltas < 2m.
5. The method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradient according to claim 4, wherein the gradient interval Δs=0.7 m.
6. The method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradient according to claim 1, wherein the step c comprises the steps of:
c1. respectively carrying out normal normalization on each characteristic curve, and respectively obeying standard normal distribution by each characteristic sample for modeling, wherein the calculation formula is as follows:
wherein X is 0i 、X 0j Respectively a certain characteristic original sample set X 0 The values at the ith and jth sample points;an average value of the original sample set on a certain characteristic; m is the total number of training set samples; x is X s Is a standard deviation conversion coefficient, and is dimensionless; x is X i Normalizing the value of the ith sample point in the sample set X after a certain feature;
c2. and fitting the sample after normal normalization by using a support vector regression model to obtain a quantitative calculation model with high Kong Baoceng porosity.
7. The method for quantitatively evaluating the carbonate height Kong Baoceng based on the resistivity gradient according to claim 6, wherein the support vector regression model in the step c2 uses three features of acoustic wave time difference, resistivity logarithm and resistivity bidirectional gradient difference after normal normalization.
8. The method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradients of claim 6, wherein the support vector regression model is trained using a k-fold cross-validation approach.
9. The method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradient of claim 8, wherein k=3 in k-fold cross validation.
10. The method for quantitative evaluation of carbonate height Kong Baoceng based on resistivity gradients of claim 6, wherein the support vector regression model uses a kernel function that is a radial basis function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211176093.4A CN117828458A (en) | 2022-09-26 | 2022-09-26 | Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211176093.4A CN117828458A (en) | 2022-09-26 | 2022-09-26 | Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117828458A true CN117828458A (en) | 2024-04-05 |
Family
ID=90504719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211176093.4A Pending CN117828458A (en) | 2022-09-26 | 2022-09-26 | Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117828458A (en) |
-
2022
- 2022-09-26 CN CN202211176093.4A patent/CN117828458A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2017084454A1 (en) | Stratum component optimization determination method and device | |
| Bauer et al. | Sources of uncertainty in shear stress and roughness length estimates derived from velocity profiles | |
| CN109736784B (en) | Sedimentary rock formations pore pressure prediction calculation method | |
| CN103336305B (en) | A kind of method dividing Sandstone Gas Reservoir high water cut based on gray theory | |
| CN113837501B (en) | Shale total organic carbon prediction method, system and equipment based on graph neural network | |
| US20240177033A1 (en) | Method, Device and Equipment for Selecting Key Geological Parameters of a To-Be-Prospected Block | |
| CN114490622A (en) | Automatic ocean temperature and salinity observation data quality control method and system based on climate state range threshold | |
| CN112145165B (en) | Microcrack-pore type reservoir dynamic and static permeability conversion method | |
| CN111598440A (en) | Multi-angle driven quantitative evaluation method and system for permeability of complex medium reservoir | |
| CN112796738A (en) | Stratum permeability calculation method combining array acoustic logging and conventional logging | |
| CN112836393B (en) | Method for analyzing reservoir heterogeneity based on multi-scale entropy | |
| CN117828458A (en) | Carbonate rock height Kong Baoceng quantitative evaluation method based on resistivity gradient | |
| CN116864026A (en) | Ocean dissolved oxygen concentration reconstruction model construction method based on Argo warm salt profile | |
| CN111963162B (en) | Lithology and water-based dual-parameter fluid property identification method | |
| CN114060015B (en) | Method and device for evaluating gas content of compact sandstone | |
| US20220342108A1 (en) | Method, device and medium for acquiring logging parameters | |
| CN113216945B (en) | Quantitative evaluation method for permeability of tight sandstone reservoir | |
| CN111122813B (en) | Water quality category evaluation method based on regional groundwater flow field direction | |
| Descalzi et al. | Synergetic log and core data treatment through Cluster Analysis: a methodology to improve reservoir description | |
| CN112346147A (en) | Reservoir evaluation method based on neutron density porosity difference | |
| CN116467555B (en) | Ocean profile observation data quality control method and system | |
| CN117077462B (en) | Method and system for optimizing gamma logging while drilling curve of deep oil gas accurate navigation | |
| US20230400598A1 (en) | Iterative well log depth shifting | |
| CN111929744A (en) | Kendall's coefficient-based multivariate information reservoir classification method | |
| CN115390152A (en) | Prediction method, equipment and medium of logging curve |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |