Premium
Determining permeability of tight rock samples using inverse modeling
Author(s) -
Finsterle Stefan,
Persoff Peter
Publication year - 1997
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/97wr01200
Subject(s) - permeameter , permeability (electromagnetism) , soil science , geology , inverse problem , porosity , inverse , uncertainty analysis , geotechnical engineering , petroleum engineering , mathematics , statistics , hydraulic conductivity , geometry , mathematical analysis , genetics , membrane , biology , soil water
Data from gas‐pressure‐pulse‐decay experiments have been analyzed by means of numerical simulation in combination with automatic model calibration techniques to determine hydrologie properties of low‐permeability, low‐porosity rock samples. Porosity, permeability, and Klinkenberg slip factor have been estimated for a core plug from The Geysers geothermal field, California. The experiments were conducted using a specially designed permeameter with small gas reservoirs. Pressure changes were measured as gas flowed from the pressurized upstream reservoir through the sample to the downstream reservoir. A simultaneous inversion of data from three experiments performed on different pressure levels allows for independent estimation of absolute permeability and gas permeability which is pressure‐dependent due to enhanced slip flow. With this measurement and analysis technique we can determine matrix properties with permeabilities as low as 10 −21 m 2 . In this paper we discuss the procedure of parameter estimation by inverse modeling. We will focus on the error analysis, which reveals estimation uncertainty and parameter correlations. This information can also be used to evaluate and optimize the design of an experiment. The impact of systematic errors due to potential leakage and uncertainty in the initial conditions will also be addressed. The case studies clearly illustrate the need for a thorough error analysis of inverse modeling results.