PVT Correlations for Unconventional Reservoir Fluids — Bridging Lab and Field Applications

The primary objective of this study is to develop pressure-volume-temperature (PVT) correlations to support machine learning (ML) efforts and routine reservoir engineering calculations/forecasting in unconventional reservoirs. PVT correlations are most often used to estimate fluid properties (saturation pressure (p_b or p_dew), solution gas-oil-ratio (R_sb), formation volume factors (for oil, gas, and water), fluid viscosities, fluid compressibilities, etc.) in cases where laboratory measurements are unavailable. In unconventional reservoir developments, laboratory analyses of fluid properties are generally limited to early development cases and situations where significant fluid property variations are expected / observed.

While numerous "black oil" correlations have been proposed and applied since the 1940’s (after Standing [1947]), these historical correlations generally provide poor predictions for the near-critical/volatile fluids encountered in many unconventional reservoirs (e.g., highly volatile oils, retrograde gas condensates, and wet gases). In this work we present a series of customized PVT correlations to address these deficiencies for unconventional reservoirs and to bridge the gap between abundant (but low-confidence) field-data with sparse (but high-confidence) laboratory PVT data.

For this work, as a test of the methodology, we specifically use a high-quality, single-operator database of PVT samples (n ≥ 127, based on the PVT function being considered) where these samples span multiple unconventional basins and fluid classifications (e.g., black oils through retrograde condensates). These data are used to validate and extend the methodology / relations given by Mejia-Martinez and Blasingame (2022) for p_sat, oil FVF, oil viscosity, and oil compressibility based on five input variables (stock tank oil gravity, separator gas gravity, reservoir temperature, and the gas-oil ratio — and the saturation pressure is included for correlations of FVF, viscosity, and compressibility). Correlations (both functional forms and coefficient values) which are tuned to the entire (anonymous) multi-operator/multi-basin PVT dataset are provided in this work for general applications.

Lastly, we proposed and demonstrated a practical workflow which ties these correlations to spatial and vertical fluid property distributions using readily available field data.