Applied Transfer Learning for Production Forecasting in Shale Reservoirs

There are documented cases of machine learning being applied to different segments of the oil and gas industry with different levels of success. These successes have not been readily transferred to production forecasting for unconventional oil and gas reservoirs because of sparsity of production data at the early stage of production. Sparsity of unconventional production data is a challenge, but transfer learning can mitigate this challenge. Application of machine learning for production forecasting is challenging in areas with insufficient data. Transfer learning makes it possible to carry over the information gathered from well-established areas with rich data to areas with relatively limited data. This study outlines the background theory along with the application of transfer learning in unconventionals to aid in production forecasting. Similarity metrics are utilized in finding candidates for transfer learning by using key drivers for reservoir performance. Key drivers include similar reservoir mechanisms and subsurface structures. After training the model on a related field with rich data, most of the primary parameters learned and stored in a representative machine or deep learning model can be re-used in a transfer learning manner. By employing the already learned basic features, models with sparse data have been enriched by using transfer learning. The approach has been outlined in a stepwise manner with details. With the help of the insights transferred from related sites with rich data, the uncertainty in production forecasting has decreased, and the accuracy of the predictions increased. As a result, the details of selecting a related site to be used for transfer learning along with the challenges and steps in achieving the forecasts have been outlined in detail. There are limited studies in oil and gas literature on transfer learning for oil and gas reservoirs. If applied with care, it is a powerful method for increasing the success of models with sparse data. This study uses transfer learning to encapsulate the basics of the substructure of a well-known area and uses this information to empower the model. This study investigates the application to unconventional shale reservoirs, which have limited studies on transfer learning.

FASAL concept in meeting the requirements of assessment and forecasting crop production affected by extreme weather events

The research in remote sensing application in India started first in agriculture way back in 1969. With the improvement in satellite sensors, data processing algorithms, models and computational power over time, this research culminated into development of operational projects of CAPE and FASAL, tackling an important issue of operationally providing pre-harvest crop production forecast to stakeholders. This review paper details the sequential developments in the use of remote sensing data for crop production forecasting. The scientific developments in the use of single and multi-temporal optical and microwave satellite images for crop identification and yield estimation in India have been reviewed. The case studies on use of remote sensing data for crop assessment under extreme weather events are also presented. These include the assessment of crop damage due to extreme weather events of floods, drought, and hailstorm. Examples on use of remote sensing for crop damage assessment due to pest and diseases and forecasting their incidence using satellite derived weather parameters are reviewed.

Establishing an operational system for assessment and forecasting the impact of extreme weather events on crop production

Crop production forecasting is essential for various economic policy and decision making. There is a very successful operational programme in the country, called FASAL, which uses multiple approaches for pre-harvest production forecasting. With the increase in the frequency of extreme events and their large-scale impact on agriculture, there is a strong need to use remote sensing technology for assessing the impact. Various works have been done in this direction. This article provides three such case studies, where remote sensing along with other data have been used for assessment of flood inundation of rice crop post Phailin cyclone, period operational district/sub-district level drought assessment and understanding the impact of recent hailstorm/unseasonal rainfall on wheat crop. The case studies highlight the great scope of remote sensing data for assessment of the impact of extreme weather events on crop production.

Production Forecasting with the Interwell Interference by Integrating Graph Convolutional and Long Short-Term Memory Neural Network

Summary Accurate production forecasting is an essential task and accompanies the entire process of reservoir development. With the limitation of prediction principles and processes, the traditional approaches are difficult to make rapid predictions. With the development of artificial intelligence, the data-driven model provides an alternative approach for production forecasting. To fully take the impact of interwell interference on production into account, this paper proposes a deep learning-based hybrid model (GCN-LSTM), where graph convolutional network (GCN) is used to capture complicated spatial patterns between each well, and long short-term memory (LSTM) neural network is adopted to extract intricate temporal correlations from historical production data. To implement the proposed model more efficiently, two data preprocessing procedures are performed: Outliers in the data set are removed by using a box plot visualization, and measurement noise is reduced by a wavelet transform. The robustness and applicability of the proposed model are evaluated in two scenarios of different data types with the root mean square error (RMSE), the mean absolute error (MAE), and the mean absolute percentage error (MAPE). The results show that the proposed model can effectively capture spatial and temporal correlations to make a rapid and accurate oil production forecast.

Modelling Annual Cocoa Production Using ARIMA Time Series Model

Cocoa is the most valuable tropical agricultural commodity, comes next to oil; a major target in Nigeria’s export diversification strategies. Cocoa production forecasting is important to the Nigerian agricultural transformation agenda. This study attempts to forecast Nigerian cocoa production between 2019 and 2025 using the ARIMA model. The automated analytical procedure implemented in the R software package indicated that ARIMA (0, 1, 1) is the combination with the least Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) and hence, the most appropriate for forecasting. The results revealed that cocoa production would fall by more than 20% in 2025 in comparison with 2018. Thus, to guard against the fall, cocoa farmers in the country should be incentivized through adequate financial and technical assistance.

Development of Production-Forecasting Model Based on the Characteristics of Production Decline Analysis Using the Reservoir and Hydraulic Fracture Parameters in Montney Shale Gas Reservoir, Canada

This study developed a production-forecasting model to replace the numerical simulation and the decline curve analysis using reservoir and hydraulic fracture data in Montney shale gas reservoir, Canada. A shale-gas production curve can be generated if some of the decline parameters such as a peak rate, a decline rate, and a decline exponent are properly estimated based on reservoir and hydraulic fracturing parameters. The production-forecasting model was developed to estimate five decline parameters of a modified hyperbolic decline by using significant reservoir and hydraulic fracture parameters which are derived through the simulation experiments designed by design of experiments and statistical analysis: (1) initial peak rate ( P hyp ), (2) hyperbolic decline rate ( D hyp ), (3) hyperbolic decline exponent ( b hyp ), (4) transition time ( T transition ), and (5) exponential decline rate ( D exp ). Total eight reservoir and hydraulic fracture parameters were selected as significant parameters on five decline parameters from the results of multivariate analysis of variance among 11 reservoir and hydraulic fracture parameters. The models based on the significant parameters had high predicted R 2 values on the cumulative production. The validation results on the 1-, 5-, 10-, and 30-year cumulative production data obtained by the simulation showed a good agreement: R 2 > 0.89 . The developed production-forecasting model can be also applied for the history matching. The mean absolute percentage error on history matching was 5.28% and 6.23% for the forecasting model and numerical simulator, respectively. Therefore, the results from this study can be applied to substitute numerical simulations for the shale reservoirs which have similar properties with the Montney shale gas reservoir.