Title: Production decline analysis of oil and gas resources with robust fit and time series analysis

Authors: Octavianus Wiliantoro; Di Niu; Huazhou Li

Addresses: Department of Civil and Environmental Engineering, School of Mining and Petroleum Engineering, University of Alberta, Edmonton, Alberta, T6G 2W2, Canada ' Department of Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, T6G 2W2, Canada ' Department of Civil and Environmental Engineering, School of Mining and Petroleum Engineering, University of Alberta, Edmonton, Alberta, T6G 2W2, Canada

Abstract: Production decline analysis is widely used in petroleum industry for reserve estimation, well/field life span prediction, and economic analysis. In this paper, a comprehensive study is conducted to evaluate, enhance and compare the performance of both parametric and non-parametric models in terms of their application in production decline analysis. Instead of using ordinary fit, we propose to apply robust fit to train the parameters in the parametric models. Based on the production data collected from 11 wells, we demonstrate that robust fit can create more accurate linearisation and better model the trend of production data than the conventional ordinary least-squares fit in Duong's model (2011), although it gives comparable performance as the ordinary least-squares in Arps' exponential decline model (1945). Compared to the original Duong's model (2011) and the enhanced Duong's model (2011) with robust fit, our proposed ARIMA model can provide much higher accuracy in terms of P50 predictions. [Received: January 22, 2016; Accepted: October 20, 2016]

Keywords: production decline analysis; robust fit; reservoir engineering; ARIMA model.

DOI: 10.1504/IJOGCT.2019.096492

International Journal of Oil, Gas and Coal Technology, 2019 Vol.20 No.1, pp.1 - 30

Received: 22 Jan 2016
Accepted: 20 Oct 2016

Published online: 05 Dec 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article