Fractured Carbonate
Learning objectives
- See why the fractures are the reservoir in a tight carbonate
- Run the Part 8 rock-physics chain to an azimuthal AVO response
- Read fracture strike from the ellipse, once the fill is known
- See why acoustic and isotropic-elastic both fail here
The Brief
The third capstone needs the most elaborate engine of all. A tight carbonate has almost no matrix permeability; it produces only where it is naturally fractured, so the fractures ARE the reservoir. The question is not merely whether fractures exist, but which way they run and how intense they are, because a well has to be steered to cross them. Characterise the fracture set from surface seismic.
The Build
This is the whole rock-physics chain of Part 8 in one shot. A carbonate background plus an aligned fracture set, a strike, a crack density, and a fluid, becomes an HTI medium, and its reflection at 30° incidence, drawn around the compass, is an ellipse with two axes, one along the fractures and one across them. Which axis is long depends on the fill: at a crack density of 0.070 the reflection is 0.0072 lower across the fractures than along them when the cracks hold gas, and 0.0051 higher when they hold brine. So the strike comes from the axes together with the sign of that swing, and the size of the swing measures intensity only once the fill is known. Set the fracture parameters and read the strike back.
The Debrief
Which engine? The azimuthal, anisotropic workflow, and nothing less. An acoustic model returns a single stacked amplitude with no fracture information. Even an isotropic-elastic model, with full AVO, is azimuth-blind: it gives an offset trend but no orientation. Only a model that carries the anisotropy and varies with azimuth can recover a strike and a density. Here the target IS the anisotropy, so a model that averages over azimuth throws away the entire answer.
This is the one capstone where the most complete engine is the only fit-for-purpose choice, the exact opposite of the fault dataset. It also exposes the method's limit: switch gas to brine and the long axis of the ellipse turns through 90°, because brine removes most of the normal weakness and the shear term takes over, while oil all but cancels at this angle. The P-wave response cannot say which way the fractures run without knowing what fills them, which is why shear-wave splitting, blind to the fluid, is kept as the backstop. The next capstone moves the problem into time, detecting a CO₂ plume between two surveys.