Statmin Stastical Program

Statmin is a statistical program that calculates lithology, mineralogy, and/or porosity. The program is based on a technique first presented at the 1988 Society of Professional Well Log Analysts Symposium and later reprinted in the June, 1991 edition of "World Oil", "A Practical Approach to Statistical Log Analysis", by William K. Mitchell and Richard J. Nelson. Mr. Mitchell was instrumental in Statmin's development and continues to assist PETCOM in its support.

Statmin is very flexible and can adapt to the needs of the analyst. Any well log or computed curve that responds to lithology/mineralogy can be used as an input. Conversely, any mineral that can be determined by the input data can have its volume computed.

The core of the program is the matrix, which consists of rows of log inputs and columns of the minerals and porosity that you are computing. Basically you input the corresponding endpoint value for each along with a standard deviation, which serves as a confidence factor:

The program is designed to solve balanced, over-determined, and under-determined models:

• Over-determined models - More logs than necessary are used to calculate the mineral volumes. In this case the program will compute the answer that best fits all the logs by using the "Maximum Likelihood Estimate" of the mineral fractions.
  
• Under-determined models - The more common case where not enough logs are available to compute all the mineral fractions present at one time. The program provides the following techniques to address this problem:
  
  • Sequential Models (Prior Matrix)
    Alternate Minerals
    Alternate Matrix
    
  • Computed Data Input

Statmin uses linear response equations, which allows it to run much faster than non-linear models. These methods are available when it is necessary to deal with non-linear responses:

• Inputs can be computed curves:
  
  • If you want to use a non-linear DT response, you can compute and use sonic porosity (e.g. from Hunt-Raymer) instead of DT.
    
  • Instead of non-linear PEF or UMA apparent, use the linear UMA, which can be calculated in Basic Log Functions or in Mathpack.
    
  • If a resistivity input is desired use BVW, BVWSXO, etc. instead of ILD, MSFL, etc.
    
  • If non-linear clay responses are desired, compute VCL externally using transforms such as Clavier, Steiber, etc. and use that input curve.
  
  
• Endpoints can be values or curves. For instance, a pre-processing pass can be made (from the Endpoints tab of Statmin) which computes curves that serve as roving neutron endpoints for quartz and dolomite that are based upon porosity.

Petcom Inc., 5646 Milton Street, Suite 540, Dallas, TX 75206. Tel: 214-368-2191; Fax: 214-368-5281.