Loading...

Assisted History-Matching for Fractured Reservoir Characterization

Rezaei Kalat, Alireza | 2018

698 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51489 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Ayatollahi, Shahab; Masihi, Mohsen
  7. Abstract:
  8. Fracture reservoirs are highly heterogeneous. This heterogeneity makes the process of adjusting model parameters to match both the static geological and dynamic production data challenging. For this reason, the characterization of the fracture network of these reservoirs, which is achieved by finding the appropriate probability distributions of the fracture properties in the discrete fracture network model, requires the use of an integrated workflow for the process of history-matching.This thesis, presents an integrated workflow for the process of history-matching of naturally fractured reservoirs with field-scale performance capability. In this methodology, first, multiple discrete fracture network models are generated and upscaled to dual-porosity models. Subsequently, the amount of objective function (the difference between the observed field data for bottom-hole pressure with the simulations results) is calculated by running the simulation for each dual-porosity model. Then, the artificial neural network is constructed as a proxy model with a set of discrete fracture network parameters as an input and the value of their corresponding objective function as output. Finally, with the optimization of the neural network using the genetic algorithm and the final examination of the results by simulation, several matched models and their corresponding discrete fracture network models is obtained. In this methodology, available FMI well-logs are used to infer the probability distributions of relevant fracture parameters (including intensity, aperture, length, dip and azimuth) and an analytical method is used to upscale the discrete fracture network models to their corresponding dual-porosity models.This method has been applied successfully on two different synthetic reservoirs and implementation of this assisted history-matching, resulted in multiple equally probable discrete fracture network models and their upscaled dynamic models that honor geological data and match the dynamic production history for each reservoir
  9. Keywords:
  10. Fractured Reservoirs ; History Matching ; Dual Porosity Model ; Upscaling ; Artificial Neural Network ; Optimization ; Discrete Fracture Model

 Digital Object List

 Bookmark

No TOC