Autor: |
Ralda, Ismi, Hernomita, Desti, Erfando, Tomi |
Předmět: |
|
Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2945 Issue 1, p1-6, 6p |
Abstrakt: |
Since 50 years in the development of oil lifting at low permeability, fracturing is an alternative to improve recovery factor. Reservoir shale gas have a permeability is very small between 0.01 - 0.000001 mD. ANN is a deep learning method that uses hundreds of data as output which is later expected to obtain optimal prediction results from the Recovery Factor through trial and error on the number of covered up layer hubs. One reason of this investigate is to anticipate the victory rate of pressure driven breaking execution utilizing ANN. The strategy utilized in this think about could be a reenactment investigate strategy utilizing CMG Pearl for store recreation modeling and information affectability utilizing CMG CMOST with input shake mechanics criterion, shake mineral contexture, break half-life length, break dividing, break width and arrangement penetrability as well as yield on the frame factor recovery. The strategy in this ponder could be a reenactment utilizing CMG Jewel for supply modeling and information affectability utilizing CMG CMOST as input parameters for shake mechanics, shake mineral composition, break half-life length, break dispersing, break width and arrangement porousness with recuperation figure yield. manufactured neural systems with back proliferation strategies can deliver precise forecasts. Through 157 information with a comparison of 75% of the comes about of the calculation of the RF demonstrate from the CMG program for preparing and 25% of the comes about of the test show. The expectation comes about from Recuperation Figure utilizing the ANN strategy are ideal, so trial and mistake is carried out on the number of covered up layer hubs. The ideal and steady covered up layer hub is gotten at hub 10 with RMSE and MAPE values within the preparing information of 0.04; 0.5 and on the test information 0.08; 0.9. Other factual investigation values such as the coefficient of assurance R2 are 0.9 for putting away information and 0.9 for test information. The conclusion is thought about that the utilize of ANN in recovery factor forecast utilizing 10 covered up layer hubs demonstrated to be exceptionally great and fruitful. [ABSTRACT FROM AUTHOR] |
Databáze: |
Complementary Index |
Externí odkaz: |
|