Oil Production Prediction of Multi-fractured Horizontal Wells Using Data Science Techniques

The oil and gas production from shale oil and shale gas reservoirs has increased rapidly over the last two decades. The combination of horizontal wells and hydraulic fracturing was one of the main reasons shale production became profitable. Multi-fractured horizontal wells have emerged as an advanced mean for enhancing well productivity in low permeability reservoirs. However, the economic viability of such projects is dependent on multiple parameters, especially with the oil price fluctuations over the last decade. Therefore, choosing the projects with the highest potential is essential to maximize the return on investment. To reduce the economic risks that hydraulic fracturing projects present, data science techniques can be used to choose the most promising projects. In particular, machine learning algorithms can be used to predict and optimize the well performance. The objective of this thesis is to predict the oil and gas production of several multi-fractured horizontal wells, using different machine learning models. These models will be trained using the performance of other wells that were already drilled and exploited in the same area. The most promising project can therefore be selected.