Abstract
Deterministic methods are still widely used for reservoir characterization and modelling. The result of such methods is only one solution. It is clear that our knowledge about the subsurface is uncertain. Since stochastic methods include uncertainty in their calculations and offer more than one solution sometimes they are the best method to use. This paper shows testing of the deterministic (Ordinary Kriging) and stochastic (Sequential Gaussian Simulations) methods of reservoir properties distribution in the Lower Pontian hydrocarbon reservoirs of the Sava Depression. Reservoirs are gas- and oil-prone sandstones. Ten realizations were obtained by Sequential Gaussian Simulation, which are sufficient for defining locations with the highest uncertainties of distributed geological variables. The results obtained were acceptable and areas with the highest uncertainties were clearly observed on the maps. However, high differences of reservoir property values in neighbouring cells caused the numerical simulation duration to be too long. For this reason, Ordinary Kriging as a deterministic method was used for modelling the same reservoirs. Smooth transitions between neighbouring cells eliminated the simulation duration problems and Ordinary Kriging maps showed channel sandstone with transitional lithofacies in some reservoirs.
Highlights
Application of mathematics in geology is relatively new approach in the interpretation of subsurface geology
This paper shows testing of the deterministic (Ordinary Kriging) and stochastic (Sequential Gaussian Simulations) methods of reservoir properties distribution in the Lower Pontian hydrocarbon reservoirs of the Sava Depression
It is clear that the true geological model in the subsurface is singular, but since the description of the subsurface is based on well data, it is not possible to be certain that the solution obtained with geostatistical methods is the correct one
Summary
Application of mathematics in geology is relatively new approach in the interpretation of subsurface geology. It is clear that the true geological model in the subsurface is singular, but since the description of the subsurface is based on well data (point data), it is not possible to be certain that the solution obtained with geostatistical methods is the correct one. This is why all geostatistical methods contain some uncertainty; they are a way of estimating the subsurface conditions and they can provide only the most probable solution, in other words the model which is the closest to the real conditions. Different deterministic and stochastic distribution methods were tested and analysed to choose the most acceptable one
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have