Abstract

This paper describes methods and techniques used to create and analyze an inversion model from a 3D seismic data volume which integrates geology, geophysics, and engineering. The specific goal of the study is to optimize development of the Foster-South Cowden Field (Grayburg and San Andres formation), in Ector County, Texas.

Interpreting geologic stratigraphy from conventional seismic data is complicated by a number of factors. Seismic reflections do not define lithology. They are largely a response to variations within a lithologic sequence and to variations in rock properties within a single lithology. For this reason, seismic reflections actually have no stratigraphic significance on their own. Analyses of reflection data (wiggles) are qualitative and relative, containing a little-understood wavelet and its many complexities. A reflection sequence (e.g., peak-trough-peak...) can be the response from sands and shales in south Louisiana or from a North Sea lithologic sequence, or from a variably porous dolomite sequence in West Texas.

The seismic tool is increasingly important for providing data needed to optimize the production of discovered petroleum reserves. Quantitative analysis of inverted seismic data is necessary to provide densely controlled lithologic information compatible with engineering data. Inversion models can be generated by using inexpensive PC-based software, and by an interpreter who can supply the correct geologic constraints to find a unique solution. The 3D interpretation, inversion modeling, rock property conversion and analyses, and mapping are done by an experienced interpreter using the Vest Exploration Services 3D Seis-WinSeis PC-based software package.

The goal of an interpreter is to integrate seismic data with subsurface geologic and petrophysical data, to develop a limited set of possible geologic solutions, and to ultimately propose a unique solution. Seismic inversion modeling is a process which incorporates subsurface geology, and which constrains the modeled seismic data to fit within the possible geologic solutions. Most importantly, analysis using a seismic inversion model yields quantitative values of the lithologic properties for the study area.

Accurate, seismically-derived maps of porosity for productive zones, calibrated to well log data, are an important tool in understanding reservoir performance. The goal of this field study is to derive densely spaced, sub-surface seismic-derived data to map the distribution of carbonate rock properties for vertically thin zones. These data are correlated to historic, current, and projected future oil production. The rock property that most affects seismic response is primary and secondary porosity.

A brief description of the methodology used in this project is offered here. The focus of the work is the Grayburg (400 ft ~ 50ms) through the upper San Andres (200 ft ~ 25ms) formations. Reviewing the processed seismic amplitude (wiggle trace) data set is a natural first step, whereby “first impressions” can be made of the structure, geology, reflection waveform changes, and variations of data quality. Determinations of basic seismic data characteristics of : 1) useful frequency bandwidth (signal-to-noise ratio at different frequencies); and 2) wavelet phase are made for quality control. Proper amplitude relationships and optimized CDP stacking profoundly affect inversion model results, and are a pre-requisite of the data processing done earlier. Synthetic seismograms were used to determine an approximate phase correction, applied to the seismic data, and to tie important formation boundaries to specific reflections.

Forward synthetic seismogram models, which portray a geologic model as an expected seismic response, demonstrated that the specific producing intervals needed for engineering work are not defined by seismic reflections. In fact, the reflection which is relatively close to the top of Grayburg in one area is continuous with a younger lower Queen rock unit in another area. The inversion model displays these and other differences, revealing critical stratigraphic relationships unrecognized on seismic wiggle data.

A significant conclusion of this study was made based on review of hundreds of feet of Grayburg to San Andres core. There are no seismically significant lithology boundaries within the entire carbonate sequence, and reflections therein are in response to rock property changes.

The inversion modeling process used for this project requires subsurface descriptions (constraints) in the form of seismic velocities, provided by sonic logs. Modeled traces to be analyzed are a function of sonic velocity and time. The log values are not “force-tied” during the inversion process, but the results of the model are checked in two ways. First, the model trace characteristics in profile view are compared to sonic log curves for similarity and other control logs for zone relationships. Second, analyses of thin sequences (averaged interval velocity) in map view are compared to gross average porosity values determined from control wells using a cross-plotting technique. Any number of other geologically significant parameters can also be related to the seismic model results, but a critical element in the accuracy of any comparison is the correct selection of zones comparable in vertical and horizontal rock space. Determination of the boundary positions for each zone analysis in this project is made from geologically-based log correlations using an isopach-to-isochron conversion technique. A final conversion of the averaged interval velocity map to map of porosity values is made (for each production zone) using software included in the Vest 3DSeis-WinSeis software package.