频变黏弹性流体因子叠前地震F-AVA反演方法

Estimating frequency-dependent viscoelastic fluid indicator from pre-stack F-AVA inversion

关键词:  黏弹性流体因子  频变AVO/AVA反演  频散属性  贝叶斯估计  流体识别

Key words:  viscoelastic fluid indicator  frequency-dependent AVO/AVA  dispersion attributes  Bayesian estimation  fluid identification

摘要:

地震波在含油气储层中传播时会发生不同程度的振幅衰减和速度频散现象,考虑介质黏弹性可更好地模拟地震波在复杂介质中的传播过程,构建频变黏弹性流体因子Fω*来定量表征孔隙流体引起的频散程度。依据Futterman近似常Q模型推导黏弹性频变AVA(F-AVA)反射系数近似方程,该方程与黏弹性介质精确Zoeppritz方程吻合度较高,系统地查明该方程的精度和F-AVA反演的可行性;此外,联合连续小波多尺度谱分解、贝叶斯估计框架和先验模型正则化,推导黏弹性介质叠前地震F-AVA反演的目标泛函,利用反复重加权最小二乘算法优化目标泛函。模型和实际测试结果验证了该反演方法的抗噪性和实用性,与常规频散属性的应用效果相比,该频变黏弹性流体因子Fω*的反演结果具有更少的“虚假亮点”干扰,该方法可以更有效地应用于储层孔隙流体识别中。

Abstract:

The phenomenon of velocity dispersion and amplitude attenuation will occur when seismic wave propagates in oil and gas reservoir. The viscoelasticity of subsurface media is introduced to simulate the attenuation and dispersion effects of seismic waves. Considering the viscoelasticity of medium, the propagation process of seismic wave in complex medium can be better simulated. Frequency-dependent viscoelastic fluid factor is constructed to quantitatively characterize the dispersion magnitude caused by pore fluids. With the Futterman approximate constant Q model, the approximate equation of viscoelastic frequency-dependent AVA (F-AVA) reflection coefficient is derived, which keeps a good agreement with the viscoelastic Zoeppritz equation. The accuracy and feasibility of the proposed equation are systematically ascertained. Furthermore, the mapping equation and objective function of F-AVA inversion are proposed by jointing continuous wavelet multiscale decomposition, Bayesian estimation framework and priori model regularizations together. And, the iterative reweighted least squares (IRLS) algorithm is utilized to solve the objective function. Model tests and a field case illustrate the anti-noise and practicability of the inversion method. Compared with the conventional dispersion attributes, the inversion result of the frequency-dependent viscoelastic fluid factor has less 'false bright spot interference, which proves the effectively in reservoir pore fluid identification.