Measurement of Dependent Variables
Fundamental to the development of porosity or permeability prediction models is the creation of a calibration database which is used to both develop and test the predictive models. To create a database, it is necessary to obtain quality data for the dependent variable, porosity and/or permeability, and for all of the independent variables, such as pressure, temperature, time, and composition. In empirical modeling, the "operational" definitions of the independent variables and the procedures followed for their measurement are highly flexible. An empirical equation can be developed using an independent variable which is a useful statistical tool but is not a direct measure of a porosity controlling process itself, like depth.
The only constraint upon the operational definitions for independent variables is that they must always be measured the same way. This liberty decreases as the models become progressively more process-oriented or theoretical and become measures of real-world processes containing externally defined variables such as pressure and temperature. Though great flexibility is possible in the definition and meaning of the independent variables, the “operational” definition of the dependent variable, porosity for example, is highly important. The measure and type of porosity used as the dependent variable in the developed model will govern the type of porosity which that model predicts. This is the measure of porosity that will be applied to the real world and form the basis for economic decisions. Thus, if a porosity prediction model is developed which predicts, for example, average porosity, then the model cannot be assumed
Figures & Tables
Reservoir Quality Assessment and Prediction in Clastic Rocks
This course is designed to emphasize the following topics: (1) Historical perspective on previous and current empirical, and geochemical methods of reservoir quality prediction; (2) Overview of diagenetic processes which significantly impact reservoir quality and those factors which act as major controls on those processes; (3) Proper design of a comprehensive or limited-focus predictive analysis of reservoir quality; (4) Methodologies for the accurate measurement of all major dependent and independent variables; (5) Data analysis techniques involved in quality control and the assessment of variability prior to performing multivariate regression; (6) Steps involved in the generation of a multivariate regression to insure that the model developed provides maximum accuracy using a minimum number of independent variables; (7) Case histories from a variety of settings illustrating application of the recommended approach to reservoir quality prediction.