Syawaldriyansah, Refiandi Reza (2023) Prediksi Dead Oil Viscosity Menggunakan Superlearner Algorithm. Other thesis, Universitas Islam Riau.

Text 163210666.pdf - Submitted Version Restricted to Registered users only Download (6MB) | Request a copy

Abstract

Reservoir fluid properties are very important for production and transportation optimization. Viscosity is defined as a fluid's resistance to flow. Dead oil viscosity is the viscosity of crude oil at atmospheric pressure and does not contain gas. Dead oil viscosity is very important in its application in petroleum engineering where dead oil is a function in each correlation to calculate the viscosity of crude oil. Viscosity values other than reservoir temperature can usually be predicted using empirical correlations if there is no laboratory data. The use of empirical correlation on the viscosity of dead oil is one of the properties that is difficult to correlate because the type of oil (depending on paraffin, aromatic, naphthene and asphalthene) has a large effect on the value of the viscosity. Therefore it is very important to know the value of viscosity and develop new methods of predicting viscosity when laboratory data are not available. In this study, 104 samples were used to predict dead oil viscosity using the Super learner algorithm using random forest and XGBoost as the base learner. The prediction results obtained will then be compared with previous studies using artificial neural networks (ANN).

Actions (login required)

View Item