Winsy Weku, Henny Pramoedyo
This study offers a variogram model approach in addition to many other theoretical models, based on logarithmic and logistical functions. The development of this model was carried out following the CNSD rules and variogram behavior both at the origin and infinity points, for this reliability model then performed on two Wolfcamp and Pre-prosperous family datasets. The results obtained are that both models work relatively well when compared with two theoretical models of variograms, namely Gauss and Exponentials. The Gaussian Variogram model is used to approximate the variogram experimental model for the wolfcamp data where the gaussian model can be approximated using the logarithmic function. For prosperous pre-family data can be used Exponential Variogram model for experimental approach variogram and exponential variogram can be approached using logistics function. The proposed function has two additional parameters, that is, the base uses base 10 (replaces the number e) and the rank uses a flexible number (recommended ρ∈[0, 3]) depending on the data used in the study. The selection of parameters is very important, because each dataset will have different parameters and will not be the same in approaching an experimental variogram. The use of error assessment, RMSE and model selection, AIC is helpful in showing the performance of both functions that are capable of matching even beyond the exponential and gaussian model of work. © 2018 DAV College.
Department of Mathematics, Brawijaya University, Malang, Indonesia
