The objective of this work is to study the applicability of various Machine Learning algorithms for prediction of some rock properties which geoscientists usually define due to special lab analysis. We demonstrate that these special properties can be predicted only basing on routine core analysis (RCA) data. To validate the approach core samples from the reservoir with soluble rock matrix components (salts) were tested within 100+ laboratory experiments. The challenge of the experiments was to characterize the rate of salts in cores and alteration of porosity and permeability after reservoir desalination due to drilling mud or water injection. For these three measured characteristics, we developed the relevant predictive models, which were based on the results of RCA and data on coring depth and top and bottom depths of productive horizons. To select the most accurate Machine Learning algorithm a comparative analysis has been performed. It was shown that different algorithms work better in different models. However, two hidden layers Neural network has demonstrated the best predictive ability and generalizability for all three rock characteristics jointly. The other algorithms, such as Support Vector Machine and Linear Regression, also worked well on the dataset, but in particular cases. Overall, the applied approach allows predicting the alteration of porosity and permeability during desalination in porous rocks and also evaluating salt concentration without direct measurements in a laboratory. This work also shows that developed approaches could be applied for prediction of other rock properties (residual brine and oil saturations, relative permeability, capillary pressure, and others), which laboratory measurements are time-consuming and expensive.Keywords machine learning · routine and special core analysis · reservoir properties · salted formations · porosity and permeability alteration

No abstract

Summary We studied the applicability of a gradient-boostingmachine-learning (ML) algorithm for forecasting of oil and total liquid production after hydraulic fracturing (HF). A thorough raw data study with data preprocessing algorithms was provided. The data set included 10 oil fields with more than 2,000 HF events. Each event has been characterized by well coordinates, geology, transport and storage properties, depths, and oil/liquid rates before fracturing for target and neighboring wells. Each ML model has been trained to predict monthly production rates right after fracturing and when the flows are stabilized. The gradient-boosting method justified its choice with R2 being approximately 0.7 to 0.8 on the test set for oil/total liquid production after HF. The developed ML prediction model does not require preliminary numerical simulations of a future HF design. The applied algorithm could be used as a new approach for HF candidate selection based on the real-time state of the field.

We are presenting a new, highly intelligent AI-based ranking system for selecting the most appropriate candidates for well treatment. The system is trained to predict flow rates after hydraulic fracturing (HF) and rank wells by the expected effect of the event with machine learning techniques. We demonstrate a significant effort for preprocessing the available field data to create a dataset for training machine learning (ML) models. The dataset included information about geology, transport and storage properties, depths, oil/liquid rates before fracturing for target and neighboring wells. Each ML model has been trained to predict monthly production of oil and liquid right after fracturing and after flow stabilization. Also, confidence intervals of the prediction have been provided. To study the dynamics of future oil rate decline after HF on a stable regime, we have trained several regression models to make predictions at each future point (6 next months after fracturing). To estimate the effect due to HF, we defined expected production "without fracturing." Typically, wells behave with a stable decline trend of production that is approximated by Arps function. The function is defined before HF, then extrapolated to the period after the event where it shows expected production without fracturing. One may conclude about the effectiveness of HF by calculating areas difference under the extrapolated curve (cumulative production without HF), and ML predicted cumulative production for future six months. Reservoir engineers could calculate these differences for each well and create a ranking list from the highest effect to the lowest. The developed system does this automatically for the required oilfield or its part. Therefore, one may easily define the list of best candidates for HF. Gradient Boosting algorithm has been applied to obtain results. Feature selection and tuning of hyperparameters have been provided with the application of cross-validation technique. To test the developed approach, we have divided the dataset from 8 conventional oil fields at a ratio of four to one. The total dataset included 700+ well interventions. Then we have trained and validated models for flow rate prediction on the major part and tested on the holdout part. For different oil field determination coefficients (R2) and normalized root mean square errors (n-RMSE) for oil rate predictions were around R2=0.8 and n-RMSE=0.35 correspondently. The proposed technique is a new approach for fast, accurate, and objective selection of the candidates for hydraulic fracturing based on real-time state of a field. Such AI-based system could become very handy assistant for reservoir engineer in addition to hydraulic fracturing and hydrodynamic simulators. The presented solution computationally efficient and does not require detailed information about HF design.

No abstract