Well production forecasting based on ARIMA-LSTM model considering manual operations

Dongyan Fan, Hai Sun, Jun Yao, Kai Zhang, Xia Yan and Zhixue Sun

Energy, 2021, vol. 220, issue C

Abstract: Accurate and efficient prediction of well production is essential for extending a well’s life cycle and improving reservoir recovery. Traditional models require expensive computational time and various types of formation and fluid data. Besides, frequent manual operations are always ignored because of their cumbersome processing. In this paper, a novel hybrid model is established that considers the advantages of linearity and nonlinearity, as well as the impact of manual operations. This integrates the autoregressive integrated moving average (ARIMA) model and the long short term memory (LSTM) model. The ARIMA model filters linear trends in the production time series data and passes on the residual value to the LSTM model. Given that the manual open-shut operations lead to nonlinear fluctuations, the residual and daily production time series are composed of the LSTM input data. To compare the performance of the hybrid models ARIMA-LSTM and ARIMA-LSTM-DP (Daily Production time series) with the ARIMA, LSTM, and LSTM-DP models, production time series of three actual wells are analyzed. Four indexes, namely, root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and similarity (Sim) values are evaluated to calculate the prediction accuracy. The results of the experiments indicate that the single ARIMA model has a good performance in the steady production decline curves. Conversely, the LSTM model has obvious advantages over the ARIMA model to the fluctuating nonlinear data. And coupling models (ARIMA-LSTM, ARIMA-LSTM-DP) exhibit better results than the individual ARIMA, LSTM, or LSTM-DP models, wherein the ARIMA-LSTM-DP model performs even better when the well production series are affected by frequent manual operations.

Keywords: Production forecasting; Hybrid model; ARIMA; LSTM; Daily production time series (search for similar items in EconPapers)
Date: 2021
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (34)

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544220328152
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:220:y:2021:i:c:s0360544220328152

DOI: 10.1016/j.energy.2020.119708

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().