“…In the ET method, a group of trees are formed randomly, and their results are aggregated. 49 The overall flowchart of all models is shown in Figure 2 .…”
Predicting asphaltene onset pressure (AOP) and bubble
point pressure
(Pb) is essential for optimization of gas injection for enhanced oil
recovery. Pressure-Volume-Temperature or PVT studies along with equations
of state (EoSs) are widely used to predict AOP and Pb. However, PVT
experiments are costly and time-consuming. The perturbed-chain statistical
associating fluid theory or PC-SAFT is a sophisticated EoS used for
prediction of the AOP and Pb. However, this method is computationally
complex and has high data requirements. Hence, developing precise
and reliable smart models for prediction of the AOP and Pb is inevitable.
In this paper, we used machine learning (ML) methods to develop predictive
tools for the estimation of the AOP and Pb using experimental data
(AOP data set: 170 samples; Pb data set: 146 samples). Extra trees
(ET), support vector machine (SVM), decision tree, and k-nearest neighbors
ML methods were used. Reservoir temperature, reservoir pressure, SARA
fraction, API gravity, gas–oil ratio, fluid molecular weight,
monophasic composition, and composition of gas injection are considered
as input data. The ET (
R
2
: 0.793, RMSE:
7.5) and the SVM models (
R
2
: 0.988, RMSE:
0.76) attained more reliable results for estimation of the AOP and
Pb, respectively. Generally, the accuracy of the PC-SAFT model is
higher than that of the AI/ML models. However, our results confirm
that the AI/ML approach is an acceptable alternative for the PC-SAFT
model when we face lack of data and/or complex mathematical equations.
The developed smart models are accurate and fast and produce reliable
results with lower data requirements.
“…In the ET method, a group of trees are formed randomly, and their results are aggregated. 49 The overall flowchart of all models is shown in Figure 2 .…”
Predicting asphaltene onset pressure (AOP) and bubble
point pressure
(Pb) is essential for optimization of gas injection for enhanced oil
recovery. Pressure-Volume-Temperature or PVT studies along with equations
of state (EoSs) are widely used to predict AOP and Pb. However, PVT
experiments are costly and time-consuming. The perturbed-chain statistical
associating fluid theory or PC-SAFT is a sophisticated EoS used for
prediction of the AOP and Pb. However, this method is computationally
complex and has high data requirements. Hence, developing precise
and reliable smart models for prediction of the AOP and Pb is inevitable.
In this paper, we used machine learning (ML) methods to develop predictive
tools for the estimation of the AOP and Pb using experimental data
(AOP data set: 170 samples; Pb data set: 146 samples). Extra trees
(ET), support vector machine (SVM), decision tree, and k-nearest neighbors
ML methods were used. Reservoir temperature, reservoir pressure, SARA
fraction, API gravity, gas–oil ratio, fluid molecular weight,
monophasic composition, and composition of gas injection are considered
as input data. The ET (
R
2
: 0.793, RMSE:
7.5) and the SVM models (
R
2
: 0.988, RMSE:
0.76) attained more reliable results for estimation of the AOP and
Pb, respectively. Generally, the accuracy of the PC-SAFT model is
higher than that of the AI/ML models. However, our results confirm
that the AI/ML approach is an acceptable alternative for the PC-SAFT
model when we face lack of data and/or complex mathematical equations.
The developed smart models are accurate and fast and produce reliable
results with lower data requirements.
“…However, the number of nodes and fuzzy roles for the formed ANFIS model by four inputs were set to 57 and 5, respectively. Figure 1 The schematic of the applied ANFIS model 19 …”
Section: Model Development and Performance Assessmentmentioning
confidence: 99%
“…During the past two decades, artificial intelligence (AI) has drawn increasing attention in petroleum engineering and geosciences owing to its capability and robustness in modeling complicated phenomena, including reservoir fluid and rock properties 19 – 22 , hydrocarbon-bearing potential of source rocks 23 , rock failure behavior 24 – 28 , soil behavior 29 , 30 and seismic characterization 31 , 32 . Predictive models thus got a boost with these new techniques.…”
Shear sonic wave velocity (Vs) has a wide variety of implications, from reservoir management and development to geomechanical and geophysical studies. In the current study, two approaches were adopted to predict shear sonic wave velocities (Vs) from several petrophysical well logs, including gamma ray (GR), density (RHOB), neutron (NPHI), and compressional sonic wave velocity (Vp). For this purpose, five intelligent models of random forest (RF), extra tree (ET), Gaussian process regression (GPR), and the integration of adaptive neuro fuzzy inference system (ANFIS) with differential evolution (DE) and imperialist competitive algorithm (ICA) optimizers were implemented. In the first approach, the target was estimated based only on Vp, and the second scenario predicted Vs from the integration of Vp, GR, RHOB, and NPHI inputs. In each scenario, 8061 data points belonging to an oilfield located in the southwest of Iran were investigated. The ET model showed a lower average absolute percent relative error (AAPRE) compared to other models for both approaches. Considering the first approach in which the Vp was the only input, the obtained AAPRE values for RF, ET, GPR, ANFIS + DE, and ANFIS + ICA models are 1.54%, 1.34%, 1.54%, 1.56%, and 1.57%, respectively. In the second scenario, the achieved AAPRE values for RF, ET, GPR, ANFIS + DE, and ANFIS + ICA models are 1.25%, 1.03%, 1.16%, 1.63%, and 1.49%, respectively. The Williams plot proved the validity of both one-input and four-inputs ET model. Regarding the ET model constructed based on only one variable,Williams plot interestingly showed that all 8061 data points are valid data. Also, the outcome of the Leverage approach for the ET model designed with four inputs highlighted that there are only 240 “out of leverage” data sets. In addition, only 169 data are suspected. Also, the sensitivity analysis results typified that the Vp has a higher effect on the target parameter (Vs) than other implemented inputs. Overall, the second scenario demonstrated more satisfactory Vs predictions due to the lower obtained errors of its developed models. Finally, the two ET models with the linear regression model, which is of high interest to the industry, were applied to diagnose candidate layers along the formation for hydraulic fracturing. While the linear regression model fails to accurately trace variations of rock properties, the intelligent models successfully detect brittle intervals consistent with field measurements.
“…Barati-Harooni et al 24 employed different ML and AI frameworks to predict minimum miscibility pressure (MMP) in enhanced oil recovery (EOR) process by N2 flooding. Amiri-Ramsheh et al 25 conducted an study about modeling of wax disappearence temperature (WDT) using different AI and ML methods. Mohammadi et al 26 employed a powerful ML technique to model hydrogen solubility in hydrocarbons.…”
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
