Limestone and dolomite reservoirs account for approximately 50% of oil and gas production worldwide, yet seismic responses in carbonate rocks are poorly understood. Development of a carbonate rock physics model is extremely difficult because pore systems are more complex in carbonates than they are in clastics. Carbonates can have a variety of pore types, such as moldic, vuggy, interparticle, and intraparticle. The complex pore system creates significant scatter in the porosity-velocity relationship, as indicated in experimental results (e.g., Anselmetti and Eberli, 2001). Pore shape appears to be the dominant factor in carbonate rock physics. Moldic, intraframe, and vuggy pores tend to be rounded and make the rock stronger (faster) than when the pores are interparticle. Micro pores (e.g., microcracks) tend to be flat and make the rock weaker. To effectively characterize carbonate reservoir rocks, it is critical to develop a rock physics model capable of handling different pore types.
In addition to pore types and pore shapes, other factors need to be included in a physics-based rock model. Some specific additional factors are lithology and grain shapes, multiphase fluids and wetting effects, rock-fluid interactions (poro-elasticity), stress effects, anisotropy, heterogeneity and scale effects, chemical changes to the framework, and corrections for environmental effects due to logging conditions. Any rock physics model should be calibrated and validated with controlled laboratory experiments, field measurements, and computational rock physics. In this paper, we demonstrate the impact and validation of several of these factors.
We prefer an analytical rock physics model rather than empirical ones due to both the predictive power that it provides without having to acquire so much analog data and its physics-based nature. Empirical rock physics models are widely used due to their simplicity in fitting a relationship between parameters. The advantage of an analytical model is that once the parameters describing the physical controlling factors are determined, the model can be applied anywhere that the controlling parameters can be estimated. This achieves better predictions and deeper understanding of the subsurface than empirical models with less data. Nonlinear physics-based inclusion models (e.g., Kuster and Toksoz, 1974) are attractive because they handle various factors that impact seismic response in an internally consistent manner.