ﻃﺮاﺣﯽ ﻣﺪل ﮔﺮوﻫﯽ ﺗﺨﻤﯿﻦ ﺗﺮاواﯾﯽ ﻣﺨﺰن ﻫﯿﺪروﮐﺮﺑﻮري ﺑﺎ اﺳﺘﻔﺎده از ﻧﮕﺎره ﻫﺎي ﭘﺘﺮوﻓﯿﺰﯾﮑﯽ ﺑﺮ اﺳﺎس ﺗﻔﮑﯿﮏ ﻟﯿﺘﻮﻟﻮژﯾﮑﯽ

Designing an ensemble model for estimating the permeability of a hydrocarbon reservoir by petrophysical lithology labeling

ﺗﺮاواﯾﯽ ﯾﺎ ﻧﻔﻮذﭘﺬﯾﺮي، ﯾﮑﯽ از ﺧﺼﻮﺻﯿﺎت ﻣﻬﻢ ﻣﺨﺎزن ﻧﻔﺖ و ﮔﺎز اﺳﺖ ﮐﻪ ﭘﯿﺶﺑﯿﻨﯽ آن دﺷﻮار ﻣﯽﺑﺎﺷﺪ. در ﺣﺎل ﺣﺎﺿﺮ از ﻣﺪلﻫﺎي ﺗﺠﺮﺑﯽ و رﮔﺮﺳﯿﻮﻧﯽ ﺑﺮاي ﭘﯿﺶﺑﯿﻨﯽ ﺗﺮاواﯾﯽ اﺳﺘﻔﺎده ﻣﯽﺷﻮد ﮐﻪ ﺷﺎﻣﻞ ﺻﺮف زﻣﺎن و ﻫﺰﯾﻨﻪﻫﺎي زﯾﺎد ﻣﺮﺗﺒﻂ ﺑﺎ اﻧﺪازهﮔﯿﺮيﻫﺎي آزﻣﺎﯾﺸﮕﺎﻫﯽ اﺳﺖ. در ﭼﻨﺪ وﻗﺖ اﺧﯿﺮ، ﺑﻪ دﻟﯿﻞ ﻗﺎﺑﻠﯿﺖ ﭘﯿﺶﺑﯿﻨﯽ ﺑﻬﺘﺮ، از اﻟﮕﻮرﯾﺘﻢﻫﺎي ﯾﺎدﮔﯿﺮي ﻣﺎﺷﯿﻦ ﺑﺮاي ﭘﯿﺶﺑﯿﻨﯽ ﺗﺮاواﯾﯽ اﺳﺘﻔﺎده ﺷﺪه اﺳﺖ. در اﯾﻦ ﻣﻄﺎﻟﻌﻪ، ﻣﺪل ﮔﺮوﻫﯽ ﯾﺎدﮔﯿﺮي ﻣﺎﺷﯿﻦ ﺟﺪﯾﺪي ﺑﺮاي ﭘﯿﺶﺑﯿﻨﯽ ﺗﺮاواﯾﯽ در ﻣﺨﺎزن ﻧﻔﺖ و ﮔﺎز ﻣﻌﺮﻓﯽ ﺷﺪه اﺳﺖ. در اﯾﻦ روش، دادهﻫﺎي ورودي ﺑﺎ اﺳﺘﻔﺎده از اﻃﻼﻋﺎت ﻟﯿﺘﻮﻟﻮژي ﻻگﻫﺎ ﺑﺮﭼﺴﺐﮔﺬاري ﺷﺪه و ﺑﻪ ﺗﻌﺪادي از دﺳﺘﻪﻫﺎ ﺗﻔﮑﯿﮏ ﺷﺪه اﺳﺖ و ﻫﺮ دﺳﺘﻪ ﺗﻮﺳﻂ اﻟﮕﻮرﯾﺘﻢ ﯾﺎدﮔﯿﺮي ﻣﺎﺷﯿﻦ ﻣﺪلﺳﺎزي ﺷﺪ. ﺑﺮﺧﻼف ﻣﻄﺎﻟﻌﺎت ﻗﺒﻠﯽ ﮐﻪ ﺑﻪ ﺻﻮرت ﻣﺴﺘﻘﻞ روي ﻣﺪلﻫﺎ ﮐﺎر ﻣﯽﮐﺮدﻧﺪ در اﯾﻨﺠﺎ ﺿﻤﻦ ﻃﺮاﺣﯽ ﯾﮏ ﻣﺪل ﮔﺮوﻫﯽ ﺑﺎ اﺳﺘﻔﺎده از اﻟﮕﻮرﯾﺘﻢﻫﺎي ETR وDTR و GBR و دادهﻫﺎي ﭘﺘﺮوﻓﯿﺰﯾﮑﯽ، ﺗﻮاﻧﺴﺘﯿﻢ ﺻﺤﺖ و دﻗﺖ ﭘﯿﺶﺑﯿﻨﯽ ﻫﻤﭽﻨﯿﻦ ﺧﻄﺎي ﻣﯿﺎﻧﮕﯿﻦ ﻣﺮﺑﻌﺎت را ﺑﻪ ﻃﺮز ﭼﺸﻢﮔﯿﺮي ﺑﻬﺒﻮد ﺑﺒﺨﺸﯿﻢ و ﺗﺮاواﯾﯽ را ﺑﺎ دﻗﺖ 99/82 درﺻﺪ ﭘﯿﺶﺑﯿﻨﯽ ﮐﻨﯿﻢ. ﻧﺘﺎﯾﺞ ﻧﺸﺎن داد ﮐﻪ ﻣﺪلﻫﺎي ﮔﺮوﻫﯽ در ﺑﻬﺒﻮد دﻗﺖ ﭘﯿﺶﺑﯿﻨﯽ ﺗﺮاواﯾﯽ در ﻣﻘﺎﯾﺴﻪ ﺑﺎ ﻣﺪلﻫﺎي اﻧﻔﺮادي ﺗﺎﺛﯿﺮ ﻓﺮاواﻧﯽ دارﻧﺪ و ﻫﻤﭽﻨﯿﻦ ﺗﻔﮑﯿﮏ ﻧﻤﻮﻧﻪﻫﺎ ﺑﺮ اﺳﺎس اﻃﻼﻋﺎت ﻟﯿﺘﻮژي دﻟﯿﻠﯽ ﺑﺮ ﺑﻬﯿﻨﻪ ﻧﻤﻮدن ﺗﺨﻤﯿﻦ ﺗﺮواﯾﯽ ﻧﺴﺒﺖ ﺑﻪ ﺗﺤﻘﯿﻘﺎت ﮔﺬﺷﺘﻪ ﺑﻮد.

Permeability is one of the important characteristics of oil and gas reservoirs that is difficult to predict. In the present solution, experimental and regression models are used to predict permeability, which includes time and high costs associated with laboratory measurements. Recently, machine learning algorithms have been used to predict permeability due to better predictability. In this study, a new ensemble machine learning model for permeability prediction in oil and gas reservoirs is introduced. In this method, the input data are labeled using the lithology information of the logs and divided into a number of categories and each category was modeled by machine learning algorithm. Unlike previous studies that worked independently on models, here we were able to predict the accuracy of such a square mean error by designing a group model using ETR, DTR, GBR algorithms and petrophysical data. Improve dramatically and predict permeability with 99/82% accuracy. The results showed that group models have a great effect on improving the accuracy of permeability prediction compared to individual models and also the separation of samples based on lithology information was a reason to optimize the Trojan estimate compared to previous studies