Estimating Reservoir Composition for Gas Condensates and Volatile Oils from Field Data

Abstract

AbstractThis work describes a method for estimating reservoir fluid composition for gas condensates and volatile oils from readily available field data. Required data are separator conditions; first stage separator gas specific gravity, non-hydrocarbon impurities content, and gas-liquid ratio; and stock tank API gravity. The well must be stabilized, and the flowing bottomhole pressure must be above the saturation pressure.A database comprising compositions for 23 wet gases, 33 gas condensates, and 19 volatile oils was taken from the literature. The data sets included compositions in mole fractions of methane through heptanes-plus and nonhydrocarbon impurities, along with molecular weight and specific gravity of the heptanes-plus fraction. The database was restricted to fluids with heptanes-plus content between 1 and 25 mole %, and with relatively low non-hydrocarbon impurities content (less than 4 mole % H2S, less than 6 mole % CO2, less than 11 mole % N2, and less than 18 mole % total nonhydrocarbons).The method predicts reservoir fluid composition through heptanes-plus along with molecular weight and specific gravity of the heptanes-plus fraction. The resulting reservoir fluid composition may be used with an untuned equation of state to obtain preliminary estimates of saturation pressure, maximum liquid dropout, and modified-black-oil (MBO) fluid property tables for simulation. The predicted composition may also be used as a cross check of laboratory-measured composition.The method has been tested on data sets for 16 gas condensates that were not part of the composition database. Predicted compositions were surprisingly accurate, with average absolute errors (AAE) for methane, 1.0 mole %; ethane, 1.2 mole %; propane, 0.7 mole %; butanes, 0.5 mole %; pentanes, 0.3 mole %; hexanes, 0.3 mole %, and heptanes-plus, 0.4 mole %. Predicted specific gravities and molecular weights of the heptanes-plus fraction had average absolute relative errors (AARE) of 1.3% and 9.7%, respectively. Saturation pressures calculated from the predicted gas-condensate compositions using an untuned equation of state had an average relative error (ARE) of −0.3% and an AARE of 6.9%, with a median relative error of -3.7% and a median absolute relative error of 4.5%.Because of the current commercial importance of liquids-rich gas plays, the method should find application for 1) obtaining preliminary estimates of composition before laboratory data are available, 2) estimating PVT properties such as saturation pressure, maximum liquid dropout, and MBO fluid property tables from field data when PVT measurements are not available, and 3) cross-checking laboratory-measured compositions.