Abstract Production data represent a source of information about the ongoing dynamic flow process in the reservoir. A proper analysis of these data might therefore give an indication of the relationships governing the fluid flow. Exploratory data analysis is one tool which can be used to extract the required information and to establish a statistical significance to the results. This paper presents a simple approach to examine the interwell communication and interference of a mature waterflood in order to identify and rank areas of potential improvement. Using this as a screening process we can identify areas with the best potential for further and more detailed studies. In addition some of the problems and limitations associated with analysis of production data are discussed.
Introduction Is there an easy way of evaluating waterflood performance based solely on surface measurements in the form of rates and pressures? Although production data have always been a key part of any waterflood evaluation, no good answer has been given to what we can expect to get from such an analysis.
Exploratory data analysis of production data is aimed at organizing and representing the data in an easy interpretive way. The analysis can take place at two levels, global and local. The global level will examine large scale trends (multiple wells), whereas the local analysis will focus on single well behavior and interwell relationships (well pairs). The use of production data as an analytical tool depends upon the quality and density of the data. More data of higher resolution and quality will yield significantly more information than irregular and poorly measured data. The quality of the analysis is therefore a direct function of the quality and the density of the measurements.
The analysis of production data poses a particular challenge since we have a 3-dimensional data set mixing both the spatial and the temporal (time) dimensions. Stationarity assumptions along the spatial plane do not translate into stationarity along the temporal axis. Special care is therefore necessary when dealing with both the spatial dimensions and the temporal dimension simultaneously. In addition, the surface measured production data constitute a 2-dimensional spatial plane, which represents the dynamic behavior of a 3-dimensional reservoir. Another challenge is the size of the data set which comprises the production data. An efficient handling and representation is therefore important to obtain a clear interpretation.
The primary objective of the methodology presented in this paper is to screen the field data to identify and rank areas of potential production improvement for a mature waterflood, in the form of changes to the individual wells, operational changes to the water injection allocation, infill drilling or pattern reorientation. This identification process depends upon the ability to extract information about the connectivity, sweep and interference among injection and production wells based on production data. Having identified potential areas of improvement, these areas can be further evaluated based on the geological knowledge and a more detailed reservoir study.