GNGTS 2018 - 37° Convegno Nazionale

GNGTS 2018 S essione 3.1 571 calibration, creation of a geological model and inversion. Well-seismic calibration consists in finding the optimal wavelet and well-position for every seismic section. A preliminary zero- phase wavelet is extracted from the seismic. Then, logs are used to apply a delay, a dephasing and a normalization coefficient to this wavelet. Optimal wavelet is the one which maximises the correlation between the synthetic trace at well-location and the observed ones. The geological model is built using some key horizons picked in the seismic to define the main geological units. A depositional mode is assigned to each unit. This model constraints the inversion and provides a basic, low-frequency, structure for the parameters to be inverted. Porosity estimation through EMT. The Effective Medium Theory (EMT) relies on a Representative Elementary Volume (REV), i.e. the smallest volume over which a property is considered as a representative value for the whole material. The EMT model used in this work is an homogenization approach based on Eshelby’s inclusion theory (Adelinet and Le Ravalec, 2015). A homogenized Eshelby’s model considers a medium composed by a matrix and a given amount of inclusions. Evaluating the elastic properties of this medium is evaluating the macroscopic elasticity at the REV scale. Using sonic, density and impedance logs (sonic is measured, density is estimated and impedance is calculated) five facies were modeled. These facies represent the main lithostratigraphic units recognizable in the wells, they have been interpreted integrating all the information reported in reports and analyzing all the available logs (not only sonic log but also resistivity, SP, γ-ray logs). Every modelled facies consisted of a solid matrix and a porous part. A mineralogic composition was supposed for the solid matrix and the correspondent elastic moduli (Bulk modulus and shear modulus) were calculated. For every facies in every well, ellipsoidal pores with a constant aspect ratio were considered and a mixed-fluid, totally saturating the pores, was assigned. Local information about gas content comes from well reports. From these facies distribution, IP logs were inverted into porosity. Two different inversions were performed, the first one assuming a fully water saturated medium and the second one considering a mixed-fluid. The accuracy of the minimization process was checked evaluating the minimum value reached by an objective function. Correlation with multi-attributes analysis. This technique allows prediction of petrophysical parameters along seismic lines, starting from well log information and using a wide family of seismic-derived attributes as a guide. The algorithm combines attributes with the target log through a generalised multiple linear regression (Coren et al. , 2001). At each time sample, the target property is modelled as a linear combination of several attributes and the related eigenvalue equation is solved by least-square minimisation of the total prediction error (RMS between the actual value and the predicted value). This process is repeated iteratively for all wells to optimise the order of polynomial and the number of attributes to consider. The result of the training is then applied to the seismic section. When the validation is completed, prediction between wells along the seismic profile can be performed. In this work, multi-attribute analysis was performed to predict 2D panels of P-velocity, resistivity and porosity, using the available sonic and resistivity logs at wells, and the calculated porosity pseudo-log estimated through EMT. Resistivity was estimated in two way: excluding or including frequency-related attributes, i.e. the ones considered more sensible to gas presence. Gas content quantification. Gas content was quantified with Archie’s second law, using resistivity and porosity obtained by multi-attributes analysis. The resistivity section estimated without the frequency-related attributes would represent the background resistivity, i.e. water- saturated sediments. The resisitivity section estimated including the frequency-related attributes would represent the total resistivity, i.e. partially water saturated, gas bearing, sediments. Gas saturation can be derived from these two resistivities, assigning a reasonable value to the saturation exponent n . The application of different approaches to predict the petrophysical properties along seismic profiles yields some important results in the studied case. The IP sections represent the lateral and in-depth heterogeneities of the lithological