#TITLE_ALTERNATIVE#

• Main Author: Herdiana (NIM 124 03 010), Yudi
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/9696
• Institution: Institut Teknologi Bandung

Abstract: <br /> <br /> <br /> <br /> <br /> <br /> <br /> In thin layer cases - where top and base of the layer are seismically indistinguishable - a conventional inversion method would give a non-unique solution. To handle this problem, a method that is capable to give many potential solutions and simultaneously search for the best solution is needed. When enough well data are available and sparsely located, then a combination between Sequential Gaussian Simulation (SGS) and Simulated Annealing (SA) methods may be used to form the core of a geostatistical inversion scheme. <br /> <br /> <br /> <br /> <br /> <br /> <br /> In this study, the geostatistical inversion scheme was applied to a 2D-synthetic seismic data generated from a wedge or pinch-out model. A noise free and a noisy data (with 20percent additive noise) were analyzed. An experiment was performed using three constraining wells which are: (i) located at the fixed locations and (ii) located randomly in the model. The randomly located wells experiment was used to test the independency of the method in converging to the global minimum solution. <br /> <br /> <br /> <br /> <br /> <br /> <br /> Results show that the SGS and SA methods can actually be combined to form a capable geostatistical inversion tool. Both - the noise free and the noisy - synthetic data were inverted to acoustic impedances optimally. This method was capable to converge to a unique solution which is independent to the starting model. This is due to the Kriging method within SGS which honour the well data properly and the power of the SA method in searching the global minimum solution. <br />