From 3D data to drill site: two views of lithology prediction
Velocity field interleaved with 3D depth migrated data. [Courtesy Landmark Graphics] Faults radiating from a salt dome are much more apparent in a continuity time slice than in the original data.(Courtesy Landmark Graphics)
Geco-Prakla's Lars Soenneland and Houston consultant Norman Neidell
Faults radiating from a salt dome are much more apparent in a continuity time slice than in the original data.(Courtesy Landmark Graphics)
3D seismic has undoubtedly revolutionized geophysics and greatly reduced the incidence of dry hole drilling over the last few years, but between the acquisition of 3D seismic data and the selection of drilling sites lies an arcane area of the petroleum industry that is little understood by anyone other than its geophysicists. The technology and its jargon are daunting, but the exercise of interactive processing and interpretation of the data to coax it into revealing the substrata it reflects is becoming both incredibly complex and, at the same time, more and more reliable. Computer power is now capable of handling enormous quantities of data in a relatively short period of time, and software is now being produced that can greatly facilitate not only the processing of the most intricate data, but the iterative modeling of that data, be it structure or velocities.
Once processed and modeled, however, it still falls to the skills of the geophysicists to look at the results and predict the lithology that will be encountered when drilling is done, including the target areas, and thereby recommend the drill site. But lithology prediction is not a set procedure; many different methods are employed, and new technologies now being developed are facilitating more accurate prediction.
The is considerable disagreement among geophysicists today about lithology prediction methods.
The most commonly used are:
- Seismic stratigraphy - in which the geometries of depositional sequences and internal reflection patterns are used to determine the location of the most important clastic source areas and the most probable locations and intervals where sand was deposited.
- Regional amplitude analyses - wherein variations in amplitudes of reflection in dominantly clastic settings are studied. High amplitudes are interpreted to indicate alternating sandstones and shales.
- Fault analysis - Fault angles change across surfaces representing changes in lithology, perhaps due to differences in competence of sandstones and claystones. The angle of a fault plane is often steeper in sandstones than in claystones. These conditions make it possible to predict where and in which intervals sand was deposited. Also they indicate that many faults have their downward termination at a particular stratigraphic level that may be associated with certain lithologies in which the fault movements are taken up.
- Velocity analysis - used to interpret lateral lithology variations along selected seismic sections. Often it is also used as a constraint on predictions derived from the other methods.
- AVO - used to determine hydrocarbon-bearing sands.
Rocks and seals
"In general, there are two common reservoir rocks, carbonates and sandstones, and there are other rocks that act as seals and have other properties, particularly shales and salt, says noted geophysicist and consultant, Norman Neidell, of Houston-based Neidell & Associates. "If you're going to drill wells, you prefer that the reservoir rocks be there, and you prefer that the geometry be favorable for trapping. Even if you have the geometry, however, you have no assurance that you have reservoir rock or that you have trap. Therefore, if you can predict lithology, both sands and shales, and if you can go beyond that and identify hydrocarbons from indicators, then you can reduce the risk of drilling substantially. That's why we want to identify lithology.
"We could know that we have a salt dome pushing up that can create very favorable geometric configurations, but if there are no sands terminating against the salt, it could be both positive and negative in reflection. That's probably the trickiest region for lithology prediction, the sand-shale crossover region. This phenomena is found in basins, and in some of the older basins, you don't have very much of the low acoustic impedance regimes, but in the Gulf of Mexico, which dumped a lot of sediments in a very short period of time, particularly offshore Louisiana, you have a very thick sediment of low impedance lands, and that's probably the primary difference between offshore Louisiana and offshore Texas - the top 10,000-12,000 ft offshore Louisiana is in that zone where you can have bright spots - what I call Zone 1 low acoustic impedance sands, but offshore Texas that thickness is only about 3,000 ft, and in offshore Texas, by the time you're at 10,000 ft, you're well into Zone 3, the consolidated sands.
"Of course, the carbonates are always higher in velocity and density than the shales. So no matter where they occur or at what age they occur, they distinguish themselves from shales with positive reflections and behave pretty much like hard consolidated sands. And then of course, when you have bedded salt, it tends to confuse the issue, because, compared to the carbonates, it can have high or low acoustic impedance depending on the age of the carbonate. The young carbonates will be lower in velocity and density than salt and the older will be higher, but we usually don't have that much confusion, because many times we can recognize salt by its distinguished geometry and flow. In addition, when we have good seismic displays, as we do now with enhanced dynamic visual displays and very good processing, you actually see a color representation of both the velocity changes and the geometry, and using the principals of seismic stratigraphy, we can identify what kind of rocks you have and what their ordering is, whether they're thick or thin, whether they have geometries that are favorable for accumulations or not, in most basins, even without well control. So the field of seismic lithology prediction is fairly well developed using some of these newer tools, especially in combination."
Neidell thinks the use of marine seismic shear waves has had very little impact on lithology prediction.
"The problem with shear is its extremely difficult to process," he says. "In the marine environment, you're using converted shear waves, and you can only measure these over the relatively thick intervals, whereas many of the paying zones that interest us are quite thin seismicly. They may not be thin in physical terms, but remember that if you have a bright spot, low velocity 6,700 ft/sec., and if you have a consolidated rock with 14,000 ft/sec. , its got twice the velocity, so twice the structural thickness will have twice the time thickness as half the thickness of a slower rock. So when you get to the consolidated rocks, they are very thin in time. Thus it's difficult to characterize them with shear waves. I would call anything that has to do with shear waves as experimental at this time and not really contributing in an operational sense.
"To me, velocity analysis is the key to stablize the imaging and getting good amplitudes. So if you want to get reliable seismic amplitudes, you have do to a lot of very precise velocity analysis.
"Some people are sold on AVO, too, but it has been most successful only in cases where you have low acoustic impedance sands - where you really don't need it. Where you have the high velocity rocks, the amplitudes you're looking for are about ten times as small as in the lower velocity rocks. So AVO has only been marginally successful.
"As a consequence, I'm more inclined to go with stratigraphy and inversion for lithology prediction. Probably the most useful application of AVO is when a rock is either terminated updip or thins. If there are hydrocarbons in it, you drop the velocity updip, if it thins you apparently drop the velocity updip, and we use the AVO to try to determine which situation we have. When you get to the hard rocks, the differences you're looking for are very small and tax the method. Virtually all published examples are in low velocity regions."
Lars Soenneland, of Geco-Prakla takes a different view of both shear waves' and AVO's role in lithology prediction. "We now have additional information with regard to marine seismic, in that we are now able to directly record shear waves," he says. "With shear waves, we have better discrimination possibilities with full elastic observation rather than only acoustic observations. Technology is providing us with better discrimination. Closely related to this is 4D, where we attempt to discriminate between fluid effects and lithology effects, with lithology now being description of the porous rock itself."
Soenneland says that geophysicists need to know which parameters in the seismic wave field are related to which of those effects, so that they can distinguish between the lithological effects and the fluid effects. If they are able to do that, he says, the science will have come a long way toward providing an answer of monitoring fluid distribution in a reservoir. Despite the advance of 4D for reservoir monitoring, it is still not always possible to discriminate between what is a lithology effect and what is a fluid effect with current technology.
"In my view, says Soenneland, "one of the challenges we have today is to quantify how accurately we need to repeat the seismic experiment to be able to attribute the changes seen over the production time of a reservoir. If we want to attribute the observed changes in the seismic response as a function of production time, we know we have to repeat the seismic experiment very accurately. If not, we might attribute what I consider noise to fluid effects.
"As an industry, we need to specify very clearly what the requirement is with respect to repeatability so that we don't misuse the term 4D seismic, because today the term is poorly or ill defined in our industry. Everything that is a new seismic experiment over a given reservoir at different production times has been called 4D, but I often disagree; because it does not fulfill some of the basic requirements to be 4D.
Where you see some new technology emerging, however, is in permanent sensors, because to be able to repeat the seismic experiment, you need to keep the receiver array at exactly the same position over time. BP is doing that now on its Foinaven Field, west of Shetland. So far, they have only acquired the baseline survey, so we don't know how well it will repeat. That type of new technology is will support the use of lithology prediction in 4D, however."
Neidell disagrees about the significance of 4D and lithology prediction's role in it. "I don't see lithology prediction and 4D really being related, he says. In the first place, 4D isn't usually done until after you have established production, so you pretty well know the lithology and know the geometry, and the purpose of doing the 4D is simply to monitor the reservoir. It's something done after quite a bit of information has been accumulated and production established. You wouldn't do 4D in an exploration situation probably ever."
Seismic or sequence stratigraphy serves as a kind of constraint for lithology prediction. It draws upon the termination patterns that are visible between stratigraphic sequence boundaries, making it possible to use seismic stratigraphic geometry frameworks as a constraint on lithology prediction, since they provide a framework for understanding how the depositional environment was paleotyped. Used in combination with what Soenneland calls lithology inversion methods, a constraint is added that facilitates better inversion.
Velocity analysis and AVO are the primary constraints to perfect lithology prediction, but with seismic or sequence stratigraphy, there is also the geometry constraint.
Soenneland says that, "Both primary effects and secondary effects are important because, if you want to apply AVO analysis, you need to have the macro velocity model in place. If you don't, the effort of doing AVO inversion is wasted, since the macro velocity model is used to insure that the AVO response is analysed as a function of the correct wave form. If it's the wrong velocity model, it might mean that you have the wrong wave form in your AVO analysis, so it's a must that you have the macro model in place before you can start your AVO inversion. But having said that, the classical AVO inversion is that exploiting the signal to establish a VP/VS ratio, so that ratio, you might call a property that is related to the geometry model. The geometry model is then how the different interfaces of your sedimentary rocks are organized. So, for example, you can think of the seismic stratigraphic model to be exactly that geometric framework."
Copyright 1996 Offshore. All Rights Reserved.