Reservoir characterization combining elastic velocities and electrical resistivity measurements

The elastic and electric parameters of rocks that can be obtained from seismic and electromagnetic data depend on porosity, texture, mineralogy, and fluid. However, seismic data seldom allow us to accurately quantify hydrocarbon saturation. On the other hand, in the case of common reservoir rocks (i.e., sandstones and carbonates), resistivity strongly depends on porosity and saturation. Therefore, the recent progress of controlled-source-electromagnetic (CSEM) methods opens new possibilities in identifying and quantifying potential hydrocarbon reservoirs, although its resolution is much lower than that of seismic data. Hence, a combination of seismic and CSEM data arguably offers a powerful means of finally resolving the problem of remote sensing of saturation. The question is how to combine the two data sources (elastic data and electrical resistivity data) to better characterize a reservoir.;To address this question, we introduce the concept of P-wave impedance and resistivity templates as a tool to estimate porosity and saturation from well log data. Adequate elastic and resistivity models, according to the lithology, cementation, fluid properties must be chosen to construct these templates. These templates can be upscaled to seismic and CSEM scale using Backus average for seismic data, and total resistance for CSEM data.;We also measured velocity and resistivity in Fontainebleau samples in the laboratory. Fontainebleau formation corresponds to clean sandstones (i.e., low clay content). We derived an empirical relation between these P-wave velocity and resistivity at 40MPa effective pressure, which is around 3 km depth at normal pressure gradients. We were not able to test if this relation could be used at well or field data scales (once appropriate upscaling was applied), since we did not have a field dataset over a stiff sandstone reservoir.;A relationship between velocity and resistivity laboratory data was also found for a set of carbonates. This expression was quadratic, and not linear as in the case of Fontainebleau sandstones. There are other factors that influence this relationship in the case of these carbonates, which include pore geometry, and amount of micritic cement. We observed that the expression is almost linear, but it deviates as we approach lower resistivities. This deviation can be explained by the presence of stiff pores such as moldic or intra-granular pores, which causes high velocity but low resistivity values when water-saturated. In the same way, the effect of micrite cement on velocity is stronger than its effect on resistivity, and that also is responsible for some of the scatter that we observe.;We also modeled both velocity and resistivity using self-consistent approximation with the same pore or inclusion geometries in both carbonate and sandstone laboratory datasets. In the case of carbonates, we found that we had to include needle-like pores to explain the low resistivity but high velocities. Needle is one of the geometries that allow us to have connected stiff pores. However, we also found that a fraction of compliant pores also had to be included in order to explain the velocity measurements on the carbonate dataset. The self-consistent model also approximated well the velocity and resistivity laboratory measurements on the Fontainebleau sandstones, using similar aspect ratios for both the velocity and the resistivity.;As far as semi-empirical and empirical models, we observed how the stiff-sand model fit well the Fontainebleau data at 40MPa, including S-wave velocities. The Raymer-Hunt-Gardner relation also did a good job at predicting P-wave velocity. Archie's equation with cementation exponent between 1.6 and 2.1 fits the resistivity measurements on the Fontainebleau sandstones. These two relationships can be combined to create a resistivity---P-wave velocity transform for this dataset.;When we attempted to use CSEM data to limit the shallow and low-frequency acoustic impedance trend for seismic inversion, we found that appropriate elastic and resistivity models must be chosen in order to have a good prediction of acoustic impedance, given resistivity. These expressions can be calibrated using well data with particular emphasis to the overburden. If no well log data are available in the shallow section, using the CSEM-derived resistivity data and an adequate cross-property relation (for example, one based on soft-sand model and Archie's equation) can be a good approach to predict the initial low frequency shallow acoustic impedance model. Validation tests showed that using the background trend from CSEM data as a constraint in impedance inversion can give a better fit to the acoustic impedance.;As part of our analysis of gas hydrate bearing sandstones, we found that normalized resistivity versus P-wave impedance templates can also be useful to predict reservoir properties, such as porosity and saturation for a gas-hydrate reservoir at well log scale. Porosity and saturation prediction of the hydrate-bearing layer from seismic data alone is highly dependent on its thickness and the properties of the overburden, and requires well-control data that can point to appropriate models and properties to use for the overburden. However, it would be interesting to test, using a resistivity model obtained from seismic data as the initial input, a CSEM inversion on a gas-hydrate-bearing sandstone.