Benefits of Maximum Likelihood Estimators in Predicting Bulk Permeability and Up-scaling Fracture Networks

Roberto E. Rizzo*, David Healy, Luca De Siena

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

The success of any predictive model is largely dependent on the accuracy with which its parameters are known. When characterising fracture networks in rocks, one of the main issues is accurately scaling the parameters governing the distribution of fracture attributes. Optimal characterisation and analysis of fracture lengths and apertures are fundamental to estimate bulk permeability and therefore fluid flow, especially for rocks with low primary porosity where most of the flow takes place within fractures. We collected outcrop data from a fractured upper Miocene biosiliceous mudstone formation (California, USA), which exhibits seepage of bitumen-rich fluids through the fractures. The dataset was analysed using Maximum Likelihood Estimators to extract the underlying scaling parameters, and we found a log-normal distribution to be the best representative statistic for both fracture lengths and apertures in the study area. By applying Maximum Likelihood Estimators on outcrop fracture data, we generate fracture network models with the same statistical attributes to the ones observed on outcrop, from which we can achieve more robust predictions of bulk permeability.
Original languageEnglish
JournalJournal of Structural Geology
Volume95
DOIs
Publication statusPublished - 2017

Keywords

  • Fractures
  • Maximum likelihood
  • Fracture networks
  • Up-scaling
  • Permeability tensor

Fingerprint

Dive into the research topics of 'Benefits of Maximum Likelihood Estimators in Predicting Bulk Permeability and Up-scaling Fracture Networks'. Together they form a unique fingerprint.

Cite this