An approach to forecast production profiles, oil-gas ratio and water contamination probabilistic assessment
An approach to probabilistic assessment of field development forecast parameters for gas-condensate reservoir using multivariant simulation was studied. The necessity of such decision may be explained by the aim to level the influence of large number of uncertainties which occur during the work wit...
שמור ב:
| Main Authors: | , , |
|---|---|
| פורמט: | Article |
| שפה: | רוסית |
| יצא לאור: |
North-Caucasus Federal University
2024-10-01
|
| סדרה: | Наука. Инновации. Технологии |
| נושאים: | |
| גישה מקוונת: | https://scienceit.elpub.ru/jour/article/view/691 |
| תגים: |
הוספת תג
אין תגיות, היה/י הראשונ/ה לתייג את הרשומה!
|
| סיכום: | An approach to probabilistic assessment of field development forecast parameters for gas-condensate reservoir using multivariant simulation was studied. The necessity of such decision may be explained by the aim to level the influence of large number of uncertainties which occur during the work with reservoir simulation models. «tNavigator» software was used as the main instrument since it provides wide functionality in the sphere of interest. The variants of variables implementation were studied, the review of experimental designs and optimization algorithms was done. At the first step, the simulation model was history matched using Differential Evolution algorithm, since its initial version had problems with phase withdrawals and pressure dynamics. Corresponding history matching quality was controlled by specially generated objective function values. At the second step a series of production forecasts based on the best history matching cases was calculated; cumulative distribution functions for field development parameters under consideration were received to get the necessary probabilistic assessment. As a result, the workflow for values of interest getting was provided; also, the variants of further modifications for studied approach were formulated: the number of simulation runs can be decreased through the choice of three base variants and use of multi-dimensional scaling which provides the opportunity for realizations of equal probability clustering with further choice of representative cases. |
|---|---|
| ISSN: | 2308-4758 |