Reservoir Fluid Identification From Low-frequency Seismic Wave

With the development of oil and gas exploration, reservoir prediction accuracyincreased greatly as well. The ultimate goal of reservoir prediction is not onlydiscriminant reservoir if there is, but also need to determine the type and property ofthe fluid contained in the reservoir. It is necessary to carry out new technologies forreservoir fluid identification to improve the accuracy of reservoir prediction. Changesand anomalies in the low frequency part of seismic reflection wave recorded wealth ofinformation in the pores of underground rock containing fluid composition. With theimprovement of the acquisition device, technology, processing technology have beengradually improved, there have been some new techniques and applications in thelow-frequency seismic wave for detecting the reservoir and identifying oil and gas,which make the potential significance and application value of low-frequency seismicreflection wave more and more subject to the attention of the geophysical explorationindustry.Based on the theoretical and application achievement which predecessors studied,this article researches the mechanism of forming low-frequency seismic waves, thevalidity of the low-frequency information for fluid identification, the change patternof the characteristic parameters (such as amplitude, attenuation, etc.) of differentfrequency components (especially the signal component at low frequent end) whichare obtained from seismic effects made on reservoir rock under different physicalparameters and fluid conditions. Furthermore, this article found out the frequency orfrequency band that is sensitive to fluid, effective methods of extractinglow-frequency information and characterizing reservoir and fluid, and the keyprocessing approach of retaining effective seismic wave. The achievement of thisarticle is able to guide the effective use of seismic information for fluid identification,laid a solid foundation for the application of seismic information to fluid identification. In this paper, the main research work and progress is as follows:(1) Mechanism of forming low-frequency seismic waves that has relation to oil andgas reservoir. Based on Biot’s theoretical model that predicts the attenuation, themesoscopic mechanism and dispersion in a poroelastic medium and White’spatchy-saturation model, and assumed that the overburden medium is non-dispersive,we designed all kinds of models with different petrophysical parameters, includingfluid saturation, permeability, fluid viscosity, and so on, and analyzed the changecharacteristics of attenuation and phase velocity versus angular frequency in seismicfrequency range, and researched the law of reflection coefficient and phase angleversus angular frequency, when P-wave propagates normal to an interface between anon-dispersive overburden and a dispersive reservoir rock. For numerical modeling ofreservoir seismic response and analytic method of frequency response at present, weemployed the viscous-diffusive-velocity dispersion wave equation to simulate seismicresponse characteristics for different petrophysical parameters, such as viscouscoefficient, diffusive coefficient, and quality factor, and analyze and conclude thechange characteristic and law of seismic wave field in medium filled with fluid whichchanges versus the change of viscous coefficient and diffusive coefficient and qualityfactor. The results of the study showed that diffusive coefficient mainly effects onamplitude of seismic reflection wave, and fluid viscous coefficient mainly causes thedominant frequency to move to lower frequency, and velocity dispersion that causedby quality factor mainly effects on time delay and phase distortion of seismicreflection wave, and what’s more important, the quantity of frequency diminution isas a three powers function of the product both fluid viscous coefficient and reservoirthickness. According to both filtration theory and the theory of poroelasticity, westudied the relation of between the low-frequency seismic reflection coefficient andfluid viscous and permeability in a macroscopically homogeneous elasticfluid-saturated porous medium. We established the direct link between petrophysicalproperty and seismic response characteristic of reservoir from the above theoreticalresearch and analysis, and we employed the law and tendency of amplitude andattenuation of seismic wave versus frequency to determine frequency or frequencyband that is sensitive to reservoir fluid, providing the theoretical basis for theapplication of low-frequency seismic reflection wave to identify the fluid.(2) Methods of extracting low-frequency seismic wave. Firstly deep researched andexplored the advantages and disadvantages of Short-time Fourier transform, wavelettransform, time-frequency continuous wavelet transform, S transform, and various modified S transform. By comparison and analysis, we found that Short-time Fouriertransform is still an effective method for information extraction in individual orlimit-band frequency, and under the premise that the adjustment parameters thatcontrol time-frequency analysis is reasonable, both time-frequency continuouswavelet transform and S transform can obtain the same result, whereas the significantdifference is that the calculation of S transform and various modified S transform canuse the existing fast Fourier transform calculation method, and can meet thelarge-scale seismic data processing and analysis. Compared with otherstime-frequency decomposition, the matching pursuit decomposition has significantadvantages, but due to its too low computing efficiency, it is difficult to apply in theactual seismic data processing. To overcome this problem, an adaptive fast matchingpursuit decomposition is introduced, which construct time-frequency basis functionset by employing the wavelets bases with five parameters that can better comply withseismic signal, and in which utilizes a hybrid optimization algorithm that is composedof particle swarm optimization and BFGS. This method can separate thetime-frequency spectrum of the effective signal of the seismic reflection from thetime-frequency spectrum of nose, and has no boundary effect, making it is no need tospecifically consider the boundary problem and processing method in the course ofsignal processing and analysis.(3) Methods of the fluid identification using low-frequency seismic wave. Themethods of the fluid identification and characterization were explored fromlow-frequency seismic wave based on the selective or constructive high-precisiontime-frequency spectral decomposition methods.①On the basis of velocitydispersion and attenuation effect on amplitude and phase of reflection coefficient, theamplitude-versus-frequency curves might be divided loosely into three classes, andthe criteria and steps of the fluid identification using the abnormal characteristics thatare caused by velocity dispersion and attenuation were established.②Through thecomparative analysis of different frequency instantaneous spectral profile, it ispossible to identify the low-frequency shadow that has relation to hydrocarbonreservoir, in the reservoir below. When high-frequency energy is obviously decreased,and low-frequency energy is still intense, and its dominant frequency continuouslymoves to the end of lower frequency with the propagation time of seismic wave,automatic detection of low-frequency shadow can be carry out by utilizing statisticalregression methods.③The physics experiments have proved the high absorption inthe low frequency part may predict the water (oil) saturation. According to this phenomenon, fluid identification can be achieved. This method is still effective forreservoir with weak amplitude.④According to the linking between low-frequencyamplitude and fluid mobility, depend-frequency imaging and fluid mobility attributeswere constructed from the low-frequency seismic wave. These attributes can be usedfor oil and gas reservoir of direct imaging. Combined with darcy’s law, oil and gasproduction rate can be calculated from low-frequency seismic wave.(4) Noise suppression for preserving low-frequency seismic wave.①Accordingto time-frequency property of seismic data and feature of time-frequencydecomposition methods, following the thinking of the frequency slice filtering, andcombining with the advantages of all kinds of time-frequency analysis, a new filteringmethod is proposed, which use generalized S transform that has good time-frequencyconcentration criterion to transform seismic data from time-spatial domain totime-frequency-spatial domain (T-f-x), then in T-f-x domain apply EMD (EmpiricalMode Decomposition) on each frequency slice, and clear those IMFs(Intrinsic ModeFunction) that noise dominant, so as to suppress coherent noise and random noise.The theoretical model and real data processing proved that The EMD filtering methodin T-f-x domain after generalized S transform can effectively suppress random noiseand coherent noise of steep dip. What’s more important, and the distortion of eachfrequency component is less.②For the diffusion equation theory and thedistribution characteristics of noise in different frequency band of seismic data, anadaptive filter method is introduced, which considers multi-band and anisotropicstructure orientation of seismic data. The application of the actual data shows that themethod improves signal-to-noise ratio, at the same time, can effectively preserve thelow-frequency seismic waves.