Capillary Pressure Prediction from Rock Models Reconstructed Using Well Log Data

Abstract

Capillary pressure (P c ) is one of the main factors governing the hydrocarbon distribution within a reservoir. Its determination usually requires expensive, time-consuming laboratory experiments on a restricted number of core samples, while the continuous P c profile of a well is practically derived from the nuclear-magnetic-resonance (NMR) downhole logging measurements. This paper presents a robust and inexpensive method of predicting the continuous P c profile of a well from rock models reconstructed using various well log data. The approach first generates a representative rock model for the formation at each given depth of interest. The rock model is constrained by formation parameters derived from the logging data and accounts for diagenetic processes such as compaction and precipitation of carbonate and clay minerals. Simulations of fluid flow and primary drainage are then performed on rock models to determine the P c curve and absolute permeability.To test and validate our modeling approach, we select 16 sandstone core samples from various geologic settings to perform laboratory measurements and numermical simulations. Rock models are reconstructed using the measured grain-size distributions and grain mineralogy from core samples. The drainage P c curves derived from rock models match well with laboratory measurements on the corresponding core samples, while P c curves converted from NMR T 2 distributions using the simple relationship of ξ = 2 T P c show differences in shape. Furthermore, the computed permeability of rock models show good agreement with the core permeability, mostly falling within the ± 2 times measurements. We have also applied the rock modeling technique to predict the continuous P c and permeability profiles of a well. Formation grain-size distribution and mineralogy at each depth are derived from downhole measurements and used to generate rock models. Generally, our computed permeability falls within the same order of magnitude as the measurements on core samples from the same depth. The simulated P c curves differ in shape from those converted from NMR T 2 distributions. However, in this case it is unknown which one represents the real P c curve due to the absence of laboratory core measurements.

Retracted: A review of the nuclear magnetic resonance physics and application in petroleum industry

“…Jin et al 60 developed a geometrical algorithm to construct a rock model using NMR T 2 relaxations, by assuming spherical grains with variable size distributions. Grains were generated from the NMR T 2 distribution by applying a predefined grain size distribution, and a simulator was later used to determine capillary pressure and permeability.…”

Genetic units averages of kappa for capillary pressure estimation from NMR transversal T2 distributions (Niger Delta as field case study)

“…Applying NMR T 2 cutoff to delineate the bulk volume irreducible (BVI) from the bulk volume moveable (BVM) presents an improved calibration of NMR-derived saturation to measured data, over existing works by Jin et al (2012). This approach is based on the assumption that the smaller pore spaces are occupied by the wetting phase alone, and the irreducible water (S wirr ) saturates the pores with T 2 less than a threshold relaxation time (T 2cutoff ), whereas the fluids in pores with T 2 > T 2cutoff are movable.…”

“…Additional curve fitting procedure is required for accurate capillary pressure modelling. This can be obtained by either using a fitting procedure that gives an average scaling factor for matching with measured data or the proposed modification after Jin et al (2012). 5.…”