LEE- Petrol ve Doğal Gaz Mühendisliği Lisansüstü Programı

Bu topluluk için Kalıcı Uri

http://hdl.handle.net/11527/19422

Gözat

Yazar "Artun, Emre" ile LEE- Petrol ve Doğal Gaz Mühendisliği Lisansüstü Programı'a göz atma
  • Öge
    Forecasting the performance of shale gas wells using machine learning
    (Graduate School, 2023) Shedaiva, Mohammed ; Artun, Emre ; 824796 ; Petroleum and Natural Gas Engineering Programme
    Utilization of real data to develop data-driven models in the petroleum industry has gained momentum in the past decade. The challenges related to modeling unconventional reservoirs has been recognized as the driving force behind this change in approach. Data-driven models help to enhance operations, increase efficiency and save time. In the meantime, several numerical reservoir simulators are used for modeling and forecasting the performance of shale gas wells. However, these models are computationally expensive and the simulators could indirectly face with difficulties in forecasting performance for the unconventional shale reservoirs comparing to conventional ones. This study employs a data analytics approach to investigate and gain understanding into the main driver parameters that influence the gas production performance in unconventional reservoirs (i.e. cumulative gas production after one year). The dataset utilized in this study is acquired from SPE Data Repository and consist of 53 wells (SPE, 2021). The study essentially utilized two primary methods, namely exploratory data analysis (EDA) and predictive data analytics modeling. Through the utilization of exploratory data analysis (EDA), the correlation between each reservoir and operational parameter with the cumulative gas production (Gp) is clearly identified. A number of reservoir and operational parameters display a strictly monotonic relationship with the gas production. Out of all variables, gas saturation was the variable, which demonstrated the strongest correlation. Furthermore, predictive data-analytics models based on statistical and machine learning algorithms were developed to forecast the cumulative gas production after 1 year. Among the five conducted models, extreme gradient boosting machine (XGBoost) proved to be the optimal technique for forecasting gas production, as it yielded the highest Coefficient of Determination (R2) and the lowest Root Mean Square Error (RMSE). Finally, an analysis of variable importance was conducted to determine the key variables, which have the highest predictive power in forecasting gas production performance in unconventional shale reservoirs. The operational parameters such as the number of stages, lateral length and bottom perforation along with reservoir properties such as gas saturation, porosity and thickness are more dominant than the other reservoir and operational parameters. Gas saturation is the most critical parameter, which is considered as the key driver of forecasting the gas production. The findings of this study will be beneficial in the design and development similar forecasting modelling projects.
  • Öge
    Machine learning based selection of candidate wells for extended shut-in due to fluctuating oil price
    (Graduate School, 2024-07-10) Lobut, Beyza ; Kulga, Burak ; Artun, Emre ; 505211502 ; Petroleum and Natural Gas Engineering
    Fluctuations in oil prices adversely affect decision making situations in which performance forecasting must be combined with realistic price forecasts. In periods of significant drops in the prices, shutting in wells for extended durations such as 6 months or more may be considered for economic purposes. For example, prices during the early days of the Covid-19 pandemic forced operators to consider shutting in all or some of their active wells. In the case of partial shut-in, selection of candidate wells may evolve as a challenging decision problem considering the uncertainties involved. In this study, a mature oil field with a long (50+ years) production history with 170+ wells is considered. Reservoirs with similar conditions face many challenges related to economic sustainability such as frequent maintenance requirements and low production rates. It is aimed to solve this decision-making problem through unsupervised machine learning with the help of the data obtained during production. Average reservoir characteristics at well locations, well performance statistics and well locations were used as potential features that could characterize similarities and differences among wells. After a multivariate data analysis that explored correlations among parameters, clustering algorithms were used to identify groups of wells that are similar with respect to aforementioned features. Using the field's reservoir simulation model, scenarios of shutting in different groups of wells were simulated. 3 years of forecasted reservoir performance was used for economic evaluation that assumed an oil price drop to $30/bbl for 6, 12 or 18 months. Results of economic analysis were analyzed to identify which group of wells should have been shut-in by also considering the sensitivity to different price levels. It was observed that well performances can be easily characterized in the 3-cluster case as low-, medium- and high-performance wells. Analyzing the forecasting scenarios by considering "NPV per active well" and "NPV from Cash Flow" parameters showed that shutting in all low-, high- and medium-performance wells altogether during the downturns results in better economic outcomes for "NPV per active well". However, shutting in high- and medium- performance wells altogether and operating only low- performance wells during the downturns results in better economic outcomes for "NPV from Cash Flow". The results show that the "NPV from Cash Flow" parameter is most sensitive to the oil price during the high price period, while the "NPV per active well" parameter is most sensitive to the number of wells shut-in during the low oil price period. This study demonstrated the effectiveness of unsupervised machine learning in well classification, particularly for the problem studied. Operating companies may use this approach for selecting wells for extended durations of shut-in in periods of low oil prices.