ArticlePDF Available

A Hybrid Artificial-Intelligence Predictive Model for Crude Oil Demand: A Case Study for a High Producer and a High Consumer

Authors:

Abstract

This paper develops a rigorous and advanced computational model to describe, analyze, and forecast global crude oil demand. The paper deploys a hybrid approach of artificial intelligence techniques: artificial neural network and genetic algorithms, to devise a methodological framework for developing forecasting models of global crude oil demand. We piloted two country cases of a high oil producer (Saudi Arabia) and a high oil consumer (China) to illustrate the effectiveness and applicability of the proposed methodology for developing oil demand outlook using artificial intelligence. The input variables of the neural network models include gross domestic product, population, oil prices, gas prices, and transport data, in addition to transformed variables and functional links. The artificial intelligent predictive models of oil demand were successfully developed, trained, validated and tested using historical oil-market data yielding excellent predictions of oil demand. The performance of the intelligent models of Saudi Arabia and China were examined for generalization attribute, predictability, and accuracy. Oil demand models for Saudi Arabia and China achieved a high prediction accuracy of a correlation coefficient of 0.975 and 0.996, respectively. The intelligent outlook models show that crude oil demand for both Saudi Arabia and China will continue to increase for the outlook period (2018-2022) but with mildly declining growth. This falling growth of oil demand can be attributed to the increase in energy efficiency, fuel switching, conversion of power plants from crude to gas-based plants, and an increase in the utilization of renewable energy such as solar and wind for electric generation and water desalination. The methodology proposed improves and enhances the conventional process of developing the oil demand outlook. It also improves and enhances the predictability and accuracy of current forecasting models of oil demand. In this study, features selection techniques are applied to identify and understand the endogenous and exogenous factors that influence global energy markets, particularly those factors that impact and drive global oil demand.
1
A Hybrid Artificial-Intelligence Predictive Model for Crude Oil Demand:
A Case Study for a High Producer and a High Consumer
Saud M. Al-Fattaha1
aSaudi Aramco, Dhahran 31311, Saudi Arabia. E-mail: saud.fattah@aramco.com
Abstract
This paper develops a rigorous and advanced computational model to describe, analyze,
and forecast global crude oil demand. The paper deploys a hybrid approach of artificial
intelligence techniques: artificial neural network and genetic algorithms, to devise a
methodological framework for developing forecasting models of global crude oil demand.
We piloted two country cases of a high oil producer (Saudi Arabia) and a high oil consumer
(China) to illustrate the effectiveness and applicability of the proposed methodology for
developing oil demand outlook using artificial intelligence.
The input variables of the neural network models include gross domestic product,
population, oil prices, gas prices, and transport data, in addition to transformed variables
and functional links. The artificial intelligent predictive models of oil demand were
successfully developed, trained, validated and tested using historical oil-market data
yielding excellent predictions of oil demand. The performance of the intelligent models of
Saudi Arabia and China were examined for generalization attribute, predictability, and
accuracy. Oil demand models for Saudi Arabia and China achieved a high prediction
accuracy of a correlation coefficient of 0.975 and 0.996, respectively.
The intelligent outlook models show that crude oil demand for both Saudi Arabia and
China will continue to increase for the outlook period (2018-2022) but with mildly declining
growth. This falling growth of oil demand can be attributed to the increase in energy
efficiency, fuel switching, conversion of power plants from crude to gas-based plants, and
an increase in the utilization of renewable energy such as solar and wind for electric
1 This paper was prepared for presentation at the 2018 IAEE Eurasian Conference held on Oct. 18-20, Baku, Azerbaijan.
Electronic copy available at: https://ssrn.com/abstract=3378041
2
generation and water desalination.
The methodology proposed improves and enhances the conventional process of
developing the oil demand outlook. It also improves and enhances the predictability and
accuracy of current forecasting models of oil demand. In this study, features selection
techniques are applied to identify and understand the endogenous and exogenous factors
that influence global energy markets, particularly those factors that impact and drive global
oil demand.
Keywords: Oil demand, Artificial intelligence, Demand forecasting, Saudi Arabia, China
Overview
The development of domestic and global oil demand outlooks is a crucial task for energy
planning, formulating strategies and recommending energy policies. Undertaking this
mission is an important and challenging endeavor as it can impact the economic trajectory
of countries or the bottom line of energy companies and related industrial sectors. The
purpose of this paper is to develop a rigorous computational model to describe, analyze,
and forecast crude oil demand using an artificial intelligence approach for a high oil
producer (Saudi Arabia) and a high oil consumer (China).
Intelligent models of oil demand were developed with data from 1970 to 2016. The
input variables of the neural network models include gross domestic product (GDP),
population, oil prices, gas prices, and transport data, in addition to transformed variables
and functional links. The artificial intelligent predictive models of oil demand were
successfully developed, trained, validated and tested using historical oil-market data
yielding excellent predictions of oil demand. The results of the intelligent oil demand
models were also compared with published industry forecasts. The neural network models
of Saudi Arabia (a high oil producer) and China (a high oil consumer) were successfully
developed. The performance of the intelligent models of Saudi Arabia and China were
examined for generalization, predictability, and accuracy. Oil demand models for Saudi
Electronic copy available at: https://ssrn.com/abstract=3378041
3
Arabia and China achieved a high prediction accuracy of a correlation coefficient of 0.975
and 0.996, respectively, indicating that the intelligent models captured very well the
dynamics of oil demand for both countries.
Methodology
This paper implemented a hybrid approach of artificial intelligence techniques: artificial
neural network and genetic algorithms, to develop a methodological framework for
developing forecasting models of global crude oil demand. This paper piloted two country
cases of a high oil producer (Saudi Arabia) and a high oil consumer (China) to illustrate
the effectiveness and applicability of the proposed methodology for developing oil demand
outlook using artificial intelligence. The intelligent models of oil demand were developed
with data from 1970 to 2016. The input variables of the neural network models are selected
using the features selection techniques of genetic algorithm and stepwise selection. These
input variables include gross domestic product (GDP), population, oil prices, gas prices,
and transport data, in addition to transformed variables and functional links.
Results
The neural network predictive models of oil demand were successfully developed, trained,
validated and tested using historical oil-market data yielding excellent predictions of oil
demand. The results of the intelligent oil demand models were also compared with
published industry forecasts. The neural network models of Saudi Arabia (a high oil
producer) and China (a high oil consumer) were successfully developed. The performance
of the intelligent models of Saudi Arabia and China were examined for generalization,
predictability, and accuracy. Oil demand models for Saudi Arabia and China achieved a
high prediction accuracy of a correlation coefficient of 0.975 and 0.996, respectively.
The artificial intelligent model of Saudi Arabia was constructed with the following
network design and architecture: multilayer perceptron (MLP) architecture,
backpropagation learning algorithm, three-layer network with 15 input nodes and one
Electronic copy available at: https://ssrn.com/abstract=3378041
4
output node. We used ensemble model for Saudi Arabia oil demand with a varying hidden-
layer nodes between 8 and 11 nodes. A logistic function was used for the hidden activation
and an exponential function for the output activation.
Similarly, the China oil demand model was designed and developed as an artificial
neural network model having: MLP architecture, backpropagation algorithm, three-layer
network with 18 input-layer nodes, 14 hidden-layer nodes, and one input-layer node. A
tanh function was used for the hidden activation and an exponential function for the output
activation. The error function of sum of squares was used for all developed models of
Saudi Arabia and China. Figure 1 shows the results of the artificial intelligent outlook model
compared to the actual oil demand of Saudi Arabia. The AI outlook model of Saudi Arabia
predicts very well the oil demand from 2013 to 2016. Figure 2 depicts the forecast of China
oil demand from the AI outlook model compared to the actual oil demand. Similar to Saudi
Arabia’s model performance, the China oil demand model shows excellent performance
of predicting oil demand from 2013 to 2016. It worth noting that these predictions are
independent and apart of the model development.
MBD
MBD
Figure 1- Saudi Arabia Oil Demand Outlook Model
4000 Actual Data
Outlook Model
3000
2500
2000
1500
1000
500
0 2020
Figure
2- China Oil Demand Outlook Model
12,000 Actual Demand
Outlook Model
10,000
8,000
6,000
4,000
2,000
0 2020
Fig. 1 & Fig. 2- Predictions performance validation of the hybrid AI predictive model compared to actual demand of
Saudi Arabia
, and China, respectively. Source: Author
Electronic copy available at: https://ssrn.com/abstract=3378041
5
Conclusion
Crude oil demand predictive models were developed using advanced analytics of artificial
intelligence. The intelligent outlook models for Saudi Arabia and China showed excellent
performance in prediction of crude oil demand. The models show that crude oil demand
for both Saudi Arabia and China will continue to increase for the outlook period but with
slightly declining growth. This falling growth of oil demand can be attributed to the increase
in energy efficiency, fuel switching, conversion of power plants from crude to gas-based
plants, and an increase in the utilization of renewable energy such as solar and wind for
electric generation and water desalination.
This paper provides an advanced, rigorous, and cutting-edge modeling approach to
our knowledge base of forecasting methods of crude oil demand. The methodology
proposed improves and enhances the conventional process of developing the oil demand
outlook. It also improves and enhances the predictability and accuracy of current
forecasting models of oil demand. In this study, genetic algorithms are also deployed to
identify and understand the endogenous and exogenous factors that influence global
energy markets, particularly those factors that impact and drive global oil demand.
Moreover, we recommend further application and use of the outlook models of oil demand
developed in this study to perform sector and regional analysis, scenario-based analysis,
energy policy analysis, and sensitivity analysis of influential factors.
References
Al-Fattah, Saud M. 2019. Artificial Intelligence Approach for Modeling and Forecasting
Oil-Price Volatility. SPE Reservoir Evaluation & Engineering Journal. DOI:
10.2118/195584-PA.
Al-Fattah, S.M., and Al-Nuaim, H.A. 2009. Artificial-Intelligence Technology Predicts
Relative Permeability of Giant Carbonate Reservoirs. SPE Reservoir Evaluation &
Engineering Journal (February 2009) 96-103.
Electronic copy available at: https://ssrn.com/abstract=3378041
6
Al-Fattah, S.M. and Startzman, R.A. 2003. Neural Network Approach Predicts U.S.
Natural Gas Production. SPE Production & Facilities Journal (May 2003) 84.
Mohaghegh, S., Al-Fattah, S.M., and Popa, A. 2011. Artificial Intelligence and Data
Mining Applications in the E&P Industry. Society of Petroleum Engineers: Richardson,
Dallas.
Electronic copy available at: https://ssrn.com/abstract=3378041
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Determination of relative permeability data is required for almost all calculations of fluid flow in petroleum reservoirs. Water-oil relative permeability data play important roles in characterizing the simultaneous two-phase flow in porous rocks and predicting the performance of immiscible displacement processes in oil reservoirs. They are used, among other applications, for determining fluid distributions and residual saturations, predicting future reservoir performance, and estimating ultimate recovery. Undoubtedly, these data are considered probably the most valuable information required in reservoir simulation studies. Estimates of relative permeability are generally obtained from laboratory experiments with reservoir core samples. In the absence of the laboratory measurement of relative permeability data, empirical correlations are usually used to estimate relative permeability data. Developing empirical correlations for obtaining accurate estimates of relative permeability data showed limited success, and proved difficult, especially for carbonate reservoir rocks. Artificial neural network (ANN) technology has proved successful and useful in solving complex structured and nonlinear problems. This paper presents a new modeling technology to predict accurately water-oil relative permeability using ANN. The ANN models of relative permeability were developed using experimental data from waterflood core tests samples collected from carbonate reservoirs of giant Saudi Arabian oil fields. Three groups of data sets were used for training, verification, and testing the ANN models. Analysis of results of the testing data set show excellent agreement with the experimental data of relative permeability. In addition, error analyses show that the ANN models developed in this study outperform all published correlations. The benefits of this work include meeting the increased demand for conducting special core analysis, optimizing the number of laboratory measurements, integrating into reservoir simulation and reservoir management studies, and providing significant cost savings on extensive lab work and substantial required time.
Artificial Intelligence and Data Mining Applications in the E&P Industry
  • S Mohaghegh
  • S M Al-Fattah
  • A Popa
• Mohaghegh, S., Al-Fattah, S.M., and Popa, A. 2011. Artificial Intelligence and Data Mining Applications in the E&P Industry. Society of Petroleum Engineers: Richardson, Dallas. Electronic copy available at: https://ssrn.com/abstract=3378041