Comparison of Shale Oil Production Forecasting using Empirical Methods and Artificial Neural Networks

Abstract

The objective of this work is to evaluate the efficacy of empirical models in forecasting oil production in shale reservoirs, bycomparing and analyzing their fit and effectiveness to our dataset. The following three modelswere considered: A Conventional Decline Curve Analysis (CDC), an Unconventional Rate Decline (URD) Approach, and a Logistics Growth Analysis (LGA) method. A comparative study is performed to evaluate the use of Artificial Neural Networks (ANN) for production forecasts and to reinforce the thinking that it is imperative to include physical parameters in mathematical models to predict accurateforecasts. For this project, we used non-linear regression to fit empirical models to the dataset obtained from North Dakota Industrial Commission (NDIC). We evaluated the fit of modelswith the help of coefficient of determination. Physical parameters, such as porosity, saturation, shale volume, etc., and log data from sonic logs, gamma ray logs, etc., were selected as input to the ANN model andwere aided by Analysis of Variances (ANOVA). Amongst the empirical models for shale play, URD method is the most commonly used since it is idealfor fractured reservoirs with extremely low permeability. URD model did fit the cumulative production profiles, but could not accurately fit the monthly production profile. The CRD approach was overallunsuccessful in generating accurate future production profiles. Values forecasted from the ANN show less than 10% error in estimation. The inclusion of physical parameters has proven to be extremely promising in the forecast production from fields that do not have sufficient history for statistical fitting. Through aselection of physical properties from different sources, we have built an ANN model that fits with the production data in wells that have adiverse production history. Our work has shown the importance of including physical parameters into a process that was heretofore seen as a time series regression problem. In general, our new ANN-based method generated the best results.

Intelligent Production Prediction of Deep Offshore Hydrocarbon Reservoir: A Case Study of Niger-Delta Region of Nigeria

“…In these examples, ANN was used to forecast cumulative output volume or production rate over a given time in the future. Suhag et al [16] presented this technique, which involves estimating the three-month and six-month oil production rate using well logs and engineering completion data. Ahmadi et al [17] introduced ANN to estimate the oil flow rate in Iranian oil fields in the northern Persian Gulf.…”

Modeling and forecasting United States oil production along with the social cost of carbon: conventional and unconventional oil

“…In another study to predict US shale oil production using a new combination of nonlinear gray model and linear ARIMA residual correction, Wang et al (2018) stated that this new NMGM-ARIMA method can significantly improve the predictive effectiveness. Suhag et al (2017) in a study to predict shale oil production using experimental methods and artificial neural networks disclosed that the predicted values of the ANN network show less than 10% error in estimation. Velasco et al (2021) foreseen the US compacted oil production using the moving boundary approach.…”

New Insights into Fracture Porosity Estimations Using Machine Learning and Advanced Logging Tools

“…However, this is an expensive and time-consuming process, and it can have a significant impact on the conclusions drawn if there are insufficient cores available for examination. On the other hand, empirical approaches, while easy to apply, are limited to wells from which data have been collected, resulting in a considerable level of uncertainty when combined with extrapolated or anticipated geological data [11]. To overcome these limitations, a cost-effective, rapid, Fuels 2023, 4 and reliable model is required for reservoir evaluations and characterizations to describe porosity, using well logs and existing core data to certify the results.…”