This paper presents an integrated approach to data interpretation during pool or field development. The methodology is based on the integration of data at different levels of resolution, multivariate statistical analysis, and advanced computer graphics.
Statistical techniques compress and organize large amounts of data into a small set of information. They aid in the identification of the most significant factors, any valid relationships, and patterns hidden in geological databases. Statistical processing estimates the degree of uncertainty surrounding exploration events, while visualization techniques provide a presentation and interpretation tool. Special attention has been given to the application of scientific visualization for presenting and analyzing multidimensional data sets.
The paper explains how to convert multidimensional geological data sets into a single parameter defined as the production probability. This parameter is visualized together with the three-dimensional properties of the formation. In addition, the paper presents a method for overlaying different multivariate data sets representing non-overlapping sparse matrices.
The techniques presented in this paper improve the testing of geological hypotheses and lead to advances in the understanding of "cause and effect" relationships between formation properties, field activities, and well performance.
The methods and techniques described in this paper are useful in identifying any valid relationships and patterns hidden in exploration and development databases. In general, geological data is multidimensional, often noisy, non-reproducible and worst of all, the majority of geological samples are mixtures of simpler components. Statistical and numerical methods can help to un-mix these samples, find the original compositions of sample data, find variance and patterns, and define and test geological hypotheses1. These techniques help to summarize the multivariate data and relate it to geological events. In addition, they help to identify the most important factors, which are then used to develop the predictive models. These models can be quantified, and used to direct exploration and development efforts in areas with the highest possible potential. Computer graphics and specifically, scientific visualization, improve the interpretation process of the data or results of the statistical analysis2,3,4.
Statistical analysis and scientific visualization were applied to estimate the production capacity or production probability based on available geological, petrophysical, and engineering data in a naturally fractured reservoir. Results of the analysis were used to provide answers and build hypotheses in the exploration process. Statistical models allowed for the selection of regions of interest or specific sites (wells) with a higher potential than the surrounding locations.
The following is a list of steps in the analysis:
development of the integrated data base
univariate analysis with the necessary data checks
estimation of pair wise correlations
estimation of correlations between groups (canonical correlations)
development of mathematical models to predict production rates
development of models which discriminated between good and poor performers
visualization of data and models.
The next section presents the most important steps of the analysis.
The project data base contained geological, petrophysical, DST, and completion parameters. The geological and petr