Atlas of Gas Chimney Occurrences Associated With Oil and Gas Fields and Dry Holes: Case Studies From Deepwater Gulf of Mexico
- David Connolly (dGB Earth Sciences)
- Document ID
- Society of Exploration Geophysicists
- 2015 SEG Annual Meeting, 18-23 October, New Orleans, Louisiana
- Publication Date
- Document Type
- Conference Paper
- 2015. Society of Exploration Geophysicists
- Gulf of Mexico, visualization, interpretation, neural networks, integration
- 1 in the last 30 days
- 102 since 2007
- Show more detail
Many hydrocarbon producing basins of the world are dominated by vertical hydrocarbon migration. This vertical hydrocarbon migration is often directly detected in the seismic record as zones of vertically chaotic, low energy data, or “gas chimneys”. Geophysicists have noted that successful oil and gas wells are frequently less well imaged than wells over barren structures. This poor imaging can be due to a diffuse gas cloud over the fully saturated oil or gas reservoir. However, the poor imaging can also occur beneath the reservoir, indicating vertical charge into the reservoir. Chimneys have often been observed near producing fields. However, the relationship between the reservoir body and the associated chimneys has not been systematically documented. Thus there is a need for a comprehensive atlas of the chimneys associated with producing reservoirs, to provide as analogs for exploration and development drilling. This atlas can also provide metrics for risking the presence of effective hydrocarbon charge and top seal. Examples will be shown from the North Sea and Gulf of Mexico.
Many noncommercial wells drill valid structural closures containing effective reservoir objectives. Thus, they fail for a lack of adequate hydrocarbon charge, or a lack of effective top seal or lateral seal. In basins, dominated by vertical hydrocarbon charge, the morphology of vertical chimneys directly beneath the reservoir may provide clues to the effectiveness of this charge. Traps can be classified for charge effectiveness based on the abundance of chimneys in direct communication with the reservoir body. Similarly the morphology of chimneys directly above the reservoir may provide clues to the top seal integrity (Ligtenberg, 2006). Work in the North Sea (Heggland, 2013) over 100 structural traps shows that traps overlain by gas clouds have a high probability of success, while traps with point sourced or fault related top seal leakage have a higher chance of being breached or having limited hydrocarbon column heights. Traps with chimneys on the flanks of the structure will often have hydrocarbon column heights corresponding to this spill point. Traps can be classified for top seal effectiveness, based on the morphology of the chimneys above the reservoir. These classifications can then be a guideline for risking untested structures.
The diffuse nature of chimneys makes them difficult to map with 3D or 2D seismic data. Thus a method was developed to highlight and visualize these gas chimneys in normally processed seismic data (Connolly, et al., 2013, Aminzadeh, et al., 2006, Meldahl et al., 2001). Gas chimneys are detected using a supervised neural network trained on reliable examples of gas chimneys. The first step in the chimney processing workflow is to determine if the seismic dataset has the resolution to detect chimneys. Poorly imaged intervals below diapiric salt, karst topography, complex thrusting, or widespread volcanics may be too noisy to distinguish true hydrocarbon migration pathways. The second step is to review the seismic cube to select lines or cross lines which display the suspected vertical hydrocarbon migration pathways or gas chimneys most clearly. The third step is to pick example locations of suspected chimneys and non-chimneys which will be inputs into the neural network training. The fourth step is to choose a set of seismic attributes which will highlight the chimney. The fifth step is to calculate the value of those attributes at the locations. The calculated attributes are fed into a neural network and displayed on the key lines. Results are evaluated on key lines with possible updates of the chimney picks and attribute set until the results are optimized. At this stage the neural network can be applied to the entire seismic volume. The output is a “chimney probability volume” which can be displayed on seismic lines, time or depth slices, horizons, and as 3D geo-bodies.
|File Size||967 KB||Number of Pages||5|