#TITLE_ALTERNATIVE#

• Main Author: Sudaryanto (NIM 22290006), Bagus
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/7377
• Institution: Institut Teknologi Bandung

ABSTRACT: <br /> <br /> <br /> One reason to use stochastic technique in describing reservoir is the incomplete nature of the available data. But, not all of the available data is used for describing reservoir, such as well test data. Currently, the only method of conditional simulation which is able to incorporate well test data is Simulated Annealing Method (SAM). This study investigates the possibility of implementation of SAM with well test data to describe reservoir properties, i. e. permeability distribution. For this purposes, the computer program of SAM which can handle well test data has succesfully been developed and used. Results of the study show that if the data that is used in the simulation contain no errors, SAM with well test data gives better predictions than SAM without well test data, especially if the position of well test wells are distributed over the simulation area. The quality of SAM prediction shows a strong dependence upon the variability of the variable to be described. The quality will be better for lower variability of the variable. For practical purposes, it is not necessary to use additional data but well test data to describe low variability variables. We still, however, need another additional data for describing variables of medium or high variability. The result of sensitivity analysis shows that the quality of SAM prediction is not sensitive to uncertainty of well test data. It is relatively not affected by error in radius investigation value and slightly affected by error in efffective property (permeability) value. The quality of SAM prediction is sensitive to uncertainty in semivariogram data. It is strongly affected by error in intermittency exponent and nugget effect value, fairly affected by error in sill value, and slightly affected by error in correlation range value.