Header image

PS2: Uncertainty Quantification and Optimization - Geostatistics and Reservoir Characterization

Monday, September 5, 2022
6:50 PM - 8:00 PM
Foyer & Room 1.2

Speaker

Dr Bogdan Sebacher
PhD
Military Technical Academy "Ferdinand I"

Conditioning facies probability fields to facies observations with convex optimization procedure in an element-free Galerkin framework

6:50 PM - 6:55 PM

Summary

The facies distribution of a reservoir is one of the biggest concerns for geologists, geophysicists, reservoir modelers, and reservoir engineers due to its high importance in the setting of any reliable decision making/optimization of field development planning. Since our knowledge about the geological characterization of a reservoir is limited, consequently, the facies distribution is not certainly known. The approach for parametrizing the facies distribution as a random variable comes naturally through the use of probability fields. The probability field of each facies gives the first information about the facies position in the reservoir domain. Since the prior probability fields of facies come either from a seismic inversion or from other sources of geologic information, they are not conditioned to the data observed from the cores extracted from the wells. In a previous study, we presented a regularized element-free Galerkin (EFG) method for conditioning the facies probability fields to facies observation. However, the method independently applies for each facies type, and, in some cases in a few grid-cells, the updated values neither stay between 0 and 1 nor are sum up to 1. The methodology shown here relies on the same regularized element-free Galerkin method and resolves these shortcomings by considering a single cost function for all probability and not different functions, one for each facies type. The robustness of the method relies upon the fact that the conditioned probability fields have values between 0 and 1 and all the updated fields sum up to 1 in all grid cells. This result is obtained with a convex optimization procedure involving an equality-inequality constrained problem solved with the Gradient projection method. The method is tested considering different cases of a reservoir model with three facies types having various geostatistical properties. As the application of the presented methodology, we have introduced the probability fields of the facies in the adaptive pluri-Gaussian simulation (APS) method. We performed a comparison between the results obtained with the APS and probability fields generated with EFG and the facies fields generated with the truncated pluri-Gaussian simulation model conditioned to facies observations. The comparison showed that the ensemble generated with APS better quantifies the uncertainty of global facies proportions than the ensemble generated with TPS. In addition, the prior variability of the facies distribution is higher for the APS case than the TPS case. This fact can be seen comparing the prior facies probability fields of the two methods.
Mr Muhammad Rifaldo Luthfan
Postgraduate student
Universitas Indonesia

3D Facies Modelling of Tuban Formation, North East Java Basin

6:55 PM - 7:00 PM

Summary

Based on the BP Statistical Review of World Energy 2021, at the end of 2020 Indonesia's proven oil reserves are still around 2.4 thousand million barrels. Indonesia's average daily oil production is 743 thousand barrels, while the average daily demand for oil is 1449 thousand barrels. In fact, the oil and gas reserves have not met the energy needs in Indonesia. Therefore, it is necessary to optimize the utilization of one of the productive marginal oil and gas fields by building a 3D facies model approach as a reference to increase production in S Field.
3D facies model is a computational depiction of the earth's crust based on petrography analysis, electrofacies analysis, well correlation, facies association analysis, structural modelling and variogram analysis. The aim is to determine the type of platform, lithofacies, depositional environment and facies distribution in S Field. Geologically, S Field is an oil field located in the Tuban Formation, North East Java Basin. The Tuban Formation is a carbonate build-up that has grown since the Early Miocene with a depositional location in the form of an Isolated Platform. Based on petrographic analysis, the Tuban Formation consists of 3 lithofacies including Skeletal Grainstone, Skeletal Packestone, and Skeletal Wackestone which are physiographically associated with Fore Reef and Inter Reef (Enos and Moore, 1983).
3D modeling of carbonate rock facies was built using the Truncated Gaussian simulation (TGS) with trend method. TGS method is a stochastic facies modeling method that is suitable for modeling reservoir units or facies that have natural sequences (Matheron et al., 1987). This sequence can be either a reservoir quality sequence or a stratigraphic sequence, either vertically or laterally.
Based on the results of 3D modeling of carbonate rock facies, the facies association of Fore Reef is in the middle as an elongated shelf in the west-east direction, which is dominated by Skeletal Grainstone and Skeletal Packstone as carbonate peaks in the S Field, while the facies association of Inter reef is the result debris or detritus from the Fore Reef in the slope transition area.

loading