In closing the discussion of his paper, the author describes a logical approach to three-dimensional flow. The total number of inclusions and their locations could be simulated and the flow could be computed by the network model. However, the volume of computation would be very large. The difference between this and the two-dimensional model is discussed. The number of simulations required for a given degree of accuracy depends on the problem. For the problem under consideration, calculations show that the mean and variance stabilized between 500 and 700 simulations. The number of inclusions should reflect known subsoil conditions. The mean value determined from available data has been used in this problem. A more refined approach would assume that the presence of an inclusion is a Poisson process with given average rate per volume of subsoil. Then the number of inclusions is a random variable. The variance of the computed flow should be greater. The approach used is approximate. Comparison with the results of finite element calculations for trial sections show that the average flow may differ by ten to 14 percent; however, the head may differ considerably.

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00080528
  • Record Type: Publication
  • Report/Paper Numbers: Proc. Paper 10897
  • Files: TRIS
  • Created Date: Feb 27 1975 12:00AM