Application of Multivariate Statistical Methods to Reservoir Fluid Characterization in Unconventional Liquid-Rich Basins
- Dayo Akintan Afekare (Louisiana State University)
- Document ID
- Society of Petroleum Engineers
- SPE Liquids-Rich Basins Conference - North America, 7-8 November, Odessa, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2019. Society of Petroleum Engineers
- Reservoir Fluid Characterisation, Regression modelling, Liquid-rich basins, Multivariate Analysis, Shale PVT
- 54 in the last 30 days
- 54 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 9.50|
|SPE Non-Member Price:||USD 28.00|
The objective of this work is to demonstrate the application of multivariate statistical tools in unconventional reservoir fluid characterization using publicly available PVT data and design a fluid characterization workflow. Several oil and gas properties were predicted for fluids in Eagle Ford groups, namely Eagle ford 1 (ef1), Eagle Ford 2 (ef2), Eagle Ford Shale (efs) and Eagle Ford (ef).
Principal component analyses (PCA) was first used for dimensionality reduction and selection of components that account for large variation of the dataset. Next, multivariate regression analyses were used to predict PVT properties as a function of reservoir fluid composition using gathered reports. Major steps in regression are trends observation, correlation of trend parameters with composition, model reconstruction and calibration.
PCA results show that virtually all original variables are required to account for substantial variation within the PVT datasets. It is also noticed that the first 3 principal components account for 84% of dataset variability. Regression models of different fluid properties including gas-oil ratio (GOR), oil formation volume factor, gas density, gas viscosity, API, saturation pressure developed for Eagle Ford Shale formation fluids demonstrate a superior level of accuracy on average (R sq. = 0.8-0.94) and it was revealed that most properties can be determined directly from mole percentage of hydrocarbon fractions. A strong correlation between initial reservoir pressure and formation true vertical depth (TVD) was identified from the PVT data. Consequently, oil formation volume factor and GOR maps covering areas where analysed data were collected from with R sq. above 0.8 were designed. A user-friendly reservoir fluid characterization workflow was proposed and can be applied to any type of reservoir fluid window- black oil, dry gas or condensate - in a liquid-rich basin. Exceptions were identified and discussed.
Reliable estimation of reservoir fluid properties has a tremendous impact on different facets of unconventional reservoir field development: drilling, completion, reservoir management, and economic analyses. Current understanding of shale PVT is inadequate due to difficulty in obtaining representative fluid samples before (low permeability) and after (large produced water volumes) hydraulic fracturing. Thus, this study builds on recent efforts in understanding shale PVT behavior and potentially eliminating need for rigorous fluid sampling using multivariate statistical methods.
|File Size||1 MB||Number of Pages||16|
KHANAL, A. New forecasting method for liquid rich shale gas condensate reservoirs with data driven approach using principal component analysis. Journal of Natural Gas Science and Engine e ring, v. 38, p. 621-637, 2017/02/01/ 2017. ISSN 1875-5100. Disponivel em; < http://www.sciencedirect.com/science/article/pii/s1875510017300148 >.