SIMULATION IN MAPPING SAND DISTRIBUTION USING MULTIPOINT STATISTICS (MPS) WITH DIRECT SAMPLING ALGORITHM (DS)

• Main Author: JUANDI MANAF (NIM : 224 09 003); pembimbing: Ignatius Sonny Winardhi, Ph,D, ANDI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/17933
• Institution: Institut Teknologi Bandung

Mapping distribution of the sand is needed in reservoir characterization . The use of variogram models based on two-point statistical techniques to estimate the value of <br /> <br /> <br /> <br /> <br /> spatially only good for stationary data, but it still has a weakness for estimating nonstationary data. Reservoir has a complex geological structure and non-stationary. That is should be approached with multipoint statistics (MPS). MPS has the capability to modelling complex patterns and structures, such as the reservoir facies classification, by using the training image (TI) as a replacement of variogram model MPS algorithm developed basically aims to optimization of the algorithm capability in simulating the complex pattern of reservoir, the ability to simulate kategorial and <br /> <br /> <br /> <br /> <br /> continuous variables, the needs quite short of the computation time, with the possibility of developing algorithms towards multivariate process. The algorithm developed at the moment, have a weakness for satisfying those needs. Direct sampling algorithm with a pattern-based approach, can be an option the algorithm. Simplicity of the algorithm, but with the ability to simulate complex patterns with a short computing time to be excess of this algorithm. Ability is based on the concept of value different approach of distance for <br /> <br /> <br /> <br /> <br /> kategorial and continuous variables, and the presence of threshold parameters that are not owned by another similar algorithm. Applications on synthetic data for the univariate case, suggesting that the threshold value <br /> <br /> <br /> <br /> <br /> of 0.1 has been able to produce the optimum solution. While application on the synthetic data for the bivariate case, indicating that a distance weighting parameter that must be considered to produce the optimum simulation. Optimum realization on synthetic data obtained in this thesis by weighting wkn= 0,8 and wkk = 0,2. Realization optimum pruduced by wkn= 0,415 ; wkn= 0,39 and additional variabel distance wkCRR= 0,195