Sensitivity Runs, Analysis, and Screening

studioSL offers classical and novel, intuitive approaches to sensitivity analysis and screening. In all cases the basic workflow is to have studioSL create multiple input files automatically, based on the parameters you would like to change. Each input file is then automatically run by the simulator and then results are displayed through various plots.  This workflow is ideally suited to the multiple simulations required for field-level history matching, where global parameters such as PVT, relperms, and grid permeability multipliers are modified.

Features

  • Reservoir modeling workflows include:
    • flow simulators such as 3DSL, Eclipse, PSim, and Powers,
    • Geomodeling software such as Petrel
  • Define parameters for flow simulation and geomodeling
    • Parameters can include continuous values as well as discrete, such as a series of include files

Latin hypercube sampling and one-at-a-time (OAT) sampling for efficient parameter selection

Sensitivity Analysis

Perform standard sensitivity analysis. Determine the heavy-hitter parameters with:
  • Tornado plots
  • Scatter plots
Use novel, intuitive sensitivity analysis using metric space methods.
  • Visualize model uncertainty
  • Intuitive, visual understanding of parameter impacts on model response

 

Novel graphics provide alternative, complimentary technique to sensitivity analysis

  • Relies on clustering analysis in metric space
  • Sensitivities can be applied to any parameter, continuous or discrete
  • Accounts for effects of stochastic noise

 

Screening

Clustering in metric space provides for screening based upon dissimilarity distance
  • Identify runs which are similar (in same cluster), and dissimilar (different clusters)
  • Cluster centroids can serve as representative models for diverse sub-selection of models
  • Historical data can also be placed in metric space
    • Visual identification of runs which are close to history
  • Visually compare diversity of runs with historical data