Inversion of Multi-Phase Petrophysical Properties Using Pumpout Sampling Data Acquired With a Wireline Formation Tester

摘要

Modern pumpout wireline formation testers (PWFTs) canrncollect a wide array of data during the pumpout phase of fluidrnsampling. Both flow rate and pressure are sampled in timernduring the pumping process and are used to infer apparentrnpermeabilities under the assumption of single-phase fluidrnflow. Numerical simulation of multi-phase flow has beenrnsuccessfully used to describe filtrate invasion and the resultingrnpumpout contamination as a function of pumping time andrnrate. Recent developments in invasion modeling also allowrnone to simulate the invasion profile for either water-base orrnoil-base filtrate invasion when the mud properties are coupledrnto the invasion process. Because of these developments, it isrnnow possible to determine more complex multi-phasernpetrophysical properties of rock formations. In this paper, wernreport on a new inversion technique used to estimaternpetrophysical formation properties from data acquired byrna PWFT.rnFirst, a 3D numerical sensitivity study of PWFT data isrncarried out over a wide range of formation propertiesrnincluding variations of permeability, anisotropy ratio, andrnporosity. Results from this sensitivity analysis are used as testrncases for the inversion algorithm to estimate formationrnparameters and their uncertainty in the presence of noisyrnmeasurements of pressure and flow rate. Inversion isrnperformed making use of a neural network approach. Wernappraise the robustness and efficiency of the inversionrnalgorithm with actual field data. The estimated formationrnparameters are further compared to core and wireline data.