Broadband Electrical Spectra of Hydrate-Bearing Sediments and Saturation Evaluation of Hydrate and Gas: A Cross-Scale Numerical Study

Hydrate and gas may coexist in natural gas hydrate (NGH) reservoirs. Electrical conductivity has been used for evaluating hydrate saturation, but it is challenging to distinguish between hydrate and gas. To achieve a reliable assessment of the gas resources of NGH reservoirs, it is desirable to develop saturation evaluation models for both gas and hydrate. In this work, the broadband electrical spectra ranging from 10^–2 to 10⁶ Hz covering both the polarizations of electrical double layer (EDL) and hydrate were utilized for predicting the saturations. First, cross-scale numerical models were built based on the molecular dynamics (MD) and finite element (FE) methods. The ion diffusion coefficient and mobility from the MD model were transferred to the FE model to obtain the broadband electrical response of the hydrate-bearing sediment (HBS). Second, the saturation evaluation models of hydrate and gas were established based on the Maxwell-Garnett (MG) theory and high-frequency electrical spectra for different hydrate distribution modes (i.e., pore-filling PF, grain-coating GC, and PF-GC mixed modes). It has been demonstrated that the ion diffusion coefficient and mobility at the nanoscale and the broadband electrical spectra of HBS at the microscale can be obtained from the cross-scale numerical models. The ion diffusion coefficient and mobility are influenced by the pore-water salinity, temperature, and clay mineral. In the frequency range of 10⁴–10⁶ Hz, the quadrature conductivity is sensitive to the variations of hydrate and gas saturations, providing a physical basis for establishing saturation models. In the range lower than 1 kHz, the low-frequency polarization does not include the contribution of EDL for the GC hydrate cases, while the EDL polarization plays a dominant role for the PF hydrate cases. With the numerical solution as a reference, the relative and root-mean-square errors of model-predicted saturations are located within ±10.00% and lower than 7.00%, respectively.