Abstract

3D seismic waves contain a great deal of information about the location of oil and gas in subsurface rock formations. However, this hydrocarbon-location information is sometimes hidden in complex wave patterns. Fortunately, a branch of physics called Quantum Mechanics can help decode seismic wave patterns to better locate oil and gas in 3D seismic wave patterns.

The Heisenberg Uncertainty Principle of Quantum Mechanics states that an event can not be resolved with infinite precision in both Time and Frequency simultaneously. Hence, the Time-Frequency Uncertainty Principle can be useful in finding hydrocarbon pay in the subsurface when 3D seismic data is available. This is because the hydrocarbon pay exists at a Particular Depth corresponding to a Particular Time on a seismic trace. The Time-Frequency Uncertainty Principle therefore states that the frequency content, or frequency influence of the Time-Located rock matrix hydrocarbons on the seismic trace, must be distributed across multiple frequencies both above and below the pay zone of interest. Therefore, we can find the pay in the time zone by looking carefully at the frequencies above and below the pay zone.

Precision Wells, LLC has employed a new type of mathematical feature extraction software, based on this Time-Frequency Uncertainty Principle, to study the feasibility of using 3D seismic data to separate productive wells from dry holes in both carbonate and sandstone reservoirs in the Permian Basin and in other basins. This new 3D seismic software methodology is termed Hydrocarbon Event Resolution Imaging (ERI) and has been employed to separate Economic wells from Non-Economic wells by analyzing 3D seismic data in the vicinity of multiple boreholes. These studies have been carried out by using ERI to optimally combine 3D seismic waveform information in Simultaneous Time-Frequency Space, with well production information, to produce 3D Contrast Volumes and 3D Contrast Maps.

In Event Resolution Imaging, the individual seismic traces in the vicinity of existing bore holes are analyzed to discover useful patterns in the 3D seismic data which robustly discriminate between any two states; state A and state B. For example, state A could represent an Economic gas well and state B could represent a Non-Economic gas well. As yet another example, state A could represent a high production oil well and state B could represent a low production oil well, or dry hole. In this way, Event Resolution Imaging employs mathematical information found in 3D seismic data to separate two distinct states, state A and state B, based on the known characteristics and the known production levels of existing wells.

By conducting the feature comparison in the Simultaneous Time-Frequency Space termed a symplectic space, the patterns in the seismic data are simultaneously explicit in both time and frequency. Therefore, pattern discovery and pattern recognition algorithms are more easily able to discern and recognize the patterns in the 3D seismic data as viewed in the symplectic space so as to find the patterns that most robustly separate between State A and State B. These optimal separation patterns are encoded as an Interpretation Key which is a mathematical recipe of how to gather up the signal that separates State A from State B, and how to discard the noise that does not separate State A from State B. In this way, the interpretation key becomes the separation key responsible for separating State A from State B on 3D seismic data.

In some ways, Event Resolution Imaging behaves like a neural network in that it learns from the seismic traces around existing wellbores. However, strictly speaking, ERI does not employ neural network technology to do its learning because it does not use back propagation or gradient descent. In addition, ERI does not use traditional discriminate analysis, linear regression, non-linear regres-