CN-122017991-A - VSP pre-drilling stratum speed prediction and driving speed modeling method

Abstract

The invention discloses a VSP pre-drilling stratum speed prediction and driving speed modeling method. Aiming at the problems of fuzzy underground structure, insufficient precision of a velocity model and the like of a pre-drilling area in the prior art, the method comprises the steps of firstly correcting the acoustic logging speed by utilizing high-precision longitudinal wave speed obtained by a zero-well source distance VSP, secondly establishing an initial velocity model of the pre-drilling area by combining a rock physical relation between the velocity and the density of a target layer before drilling in a neighboring area and pre-drilling first arrival travel time prediction and combining VSP multi-wave information to construct a time-depth constraint, then obtaining an optimal seismic wavelet based on a VSP corridor and a corrected acoustic wave curve, further establishing a target function suitable for the inversion of the velocity before drilling of the VSP, and obtaining an inversion result of the velocity before drilling by utilizing VSP corridor data. And finally, under the constraint that ground seismic data are interpreted to obtain a stratum structure, updating a ground seismic velocity model by utilizing VSP zero bias and the velocity obtained by inversion before drilling, so that the velocity model is more suitable for the underground real velocity structural characteristics.

Inventors

• CAI HANPENG
• Hou Yuehao
• ZONG JINGJING
• DAI JINGYUN
• YAO XINGMIAO
• LU CAI
• CHEN WEI
• HU GUANGMIN

Assignees

• 电子科技大学

Dates

Publication Date

20260512

Application Date

20260222

Claims (4)

1. 1. A VSP pre-drilling stratum speed prediction and driving speed modeling method is characterized by comprising the following steps: s1, restraining the speed before drilling by utilizing information of a VSP multi-wave junction and adjacent channels, predicting the speed before drilling by a method of fitting a first arrival curve, and constructing a reasonable initial speed model before drilling; s2, combining known speed density data of the adjacent region, fitting a relation between the two based on a petrophysical model, and then extracting seismic wavelets for inverting a target section by utilizing VSP corridor and logging information to lay a foundation for pre-drilling speed inversion; S3, inverting the pre-drilling speed by taking the corridor waveform as a target, and finally updating the offset speed.
2. 2. The method of predicting formation velocity and modeling driving velocity before drilling of VSP of claim 1, wherein step S1 comprises: s11, extracting first arrival time of a zero offset VSP seismic section, and fitting the first arrival time before drilling by using a curve fitting method; S12, multi-wave junction points are extracted from the zero-offset VSP seismic section, the adjacent seismic channel speed information is combined, the fitted first arrival curve is restrained, the pre-drilling speed is calculated according to the pre-drilling VSP first arrival, and a pre-drilling initial speed model is obtained.
3. 3. The method of predicting formation velocity and modeling driving velocity before drilling of VSP of claim 1, wherein step S2 comprises: s21, establishing a speed-density empirical formula according to the speed and density data of a known well section, establishing a speed-density relation, selecting a Gardner relation formula to establish a function mapping relation between speed and density, and converting the relation between a reflection coefficient and wave impedance into a direct relation between the reflection coefficient and speed; S22, assuming wavelets contained in VSP corridor seismic data are in a Rake wavelet form, reasonably setting the Rake wavelet frequency and the phase range, traversing the Rake wavelets combined by different frequency phases in the range and convolving with reflection coefficients of known well sections to generate seismic records, then carrying out similarity comparison analysis on each synthetic record and VSP corridor waveforms, selecting wavelets with the highest similarity as wavelets for inversion at a subsequent speed on the basis of the principle of maximum correlation coefficients, and finally determining sub-wave energy coefficients according to the energy ratio of the synthetic records and corridor waveforms so as to ensure the energy consistency of the synthetic records and corridor waveforms.
4. 4. The method of predicting formation velocity and modeling driving velocity before drilling of VSP of claim 2, wherein step S3 comprises: S31, the difference between VSP pre-drilling speed inversion and ground seismic speed inversion is that the accurate stratum speed of a shaft bottom adjacent section is used as a hard constraint of pre-drilling speed inversion, an initial speed model participating in pre-drilling speed inversion is built, the upper part is a section of real stratum speed above the shaft bottom, and the lower part is the initial speed built by the step S12; S32, minimizing VSP corridor records and synthetic seismic records into inversion loss functions, constructing a loss function containing TV regularization constraint by taking an initial velocity model as an inversion starting point, carrying out segment inversion iteration on the stratum velocity by utilizing a sliding window technology, and carrying out iterative inversion by minimizing the loss function, thereby realizing high-precision prediction of the pre-drilling velocity; S33, according to the known stratum structure obtained by utilizing the ground seismic data interpretation, combining the zero offset VSP speed and the inverted pre-drilling speed, and updating a speed model along the stratum structure to obtain the accurate speed model containing the pre-drilling area.

Description

VSP pre-drilling stratum speed prediction and driving speed modeling method Technical Field The invention belongs to the field of oil-gas seismic exploration, and particularly relates to a VSP pre-drilling stratum velocity prediction and driving velocity modeling method. Background In oil and gas exploration and development, a stratum velocity model is an important foundation for depth migration imaging, construction of a map, reservoir prediction and drilling engineering deployment, and especially in a pre-drilling stage, the accuracy of the stratum velocity model is directly related to the reliability of key engineering decisions such as borehole trajectory optimization, casing program design, stratum pressure prediction, drilling risk control and the like. With the continuous expansion of exploration targets to deep, ultra-deep and fracture-fracture complex reservoirs, the pre-drilling speed model provides higher requirements in terms of timeliness, accuracy and geological structure consistency. However, in the pre-drilling stage, because of lack of sufficient underground priori information, the formation interface morphology, fault distribution and velocity mutation positions are difficult to accurately master, so that an initial velocity model constructed based on empirical layered assumption or simple trend extrapolation often has significant deviation from an underground real medium, and the requirement of inversion on a high-precision background model is difficult to meet. Zero-well source distance VSP data has irreplaceable important roles in speed modeling and speed calibration due to high signal-to-noise ratio, strong vertical resolution and direct measurement of the formation speed beside the well. Although zero-well source distance VSPs can effectively provide high-precision longitudinal speed information, the utilization mode of the VSPs in the prior researches is still limited, and the VSPs focus on speed calibration or quality control, and particularly have insufficient potential in the aspect of pre-drilling speed prediction. Pre-drilling velocity predictions based on VSP corridors are typically made using seismic inversion methods, but suffer from lack of pre-drilling prior velocity information, which is strongly ill-conditioned. In order to alleviate the pathogenicity of VSP pre-drilling speed inversion, the conventional inversion technology generally adopts regularization to improve stability. Although regularization constraints can suppress noise to some extent, they are not selective to formation boundaries and abrupt velocity changes, often weakening velocity gradients, blurring fault locations, screeding critical formation interfaces, and thus deviating the final inversion result from the true geologic structure. The problems are overlapped, so that the conventional method is difficult to adapt to VSP pre-drilling speed prediction, and further the accuracy of a speed model is affected, and the offset imaging quality, pre-drilling track planning and the like are reduced. In order to meet the requirement of pre-drilling prediction under complex geological conditions, development of a novel pre-drilling speed inversion method capable of fusing multi-source information such as VSP, ground earthquake and logging, reducing dependence on unreliable prior models and retaining structural boundaries and speed abrupt change characteristics is needed. In order to cope with the challenges, students at home and abroad put forward to use a multiple regression model to model formation speed prediction problems, use a deep convolutional neural network to construct a substitute semblance characteristic expression, automatically complete speed estimation and other various improvement strategies through the cyclic neural network, remarkably improve the accuracy of a speed model, and greatly improve the intelligent degree and adaptability of the speed modeling. The existing research results show that the pre-drilling speed inversion method still faces the following problems in practical application: 1. In the VSP pre-drilling speed inversion stage, the seismic wavelet morphology of the pre-drilling target horizon is difficult to predict due to the lack of measured well calibration. The traditional method depends on statistical estimation or adjacent well reference, so that the matching degree of wavelet phase and frequency components and an actual stratum is low, and the reliability of a VSP pre-drilling speed inversion result is directly weakened; 2. the corridor superimposed data is only one seismic data, can be regarded as zero-incidence-angle seismic data, can be used for acoustic impedance inversion only, and is required to be stratum velocity in engineering practice. The existing impedance-speed secondary conversion flow not only depends on an uncertain density empirical formula, but also can cause the accumulation and superposition of prediction errors in a calculation chain; 3. In pre-drilling en