Transylvanian

Review
Vol 30, No. 2, 2022

Transylvanian Review
Centrul de Studii Transilvane| str. Mihail Kogalniceanu nr. 12-14, et.5, Cluj-Napoca
Email: [email protected]
Online Submission System: http://transylvanianreviewjournal.com/
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Computational Analysis for Optimum Multiphase

Flowing Bottom-Hole Pressure Prediction
Ugochukwu I. Duru1, Dennis Delali Kwesi Wayo3, Reginald Ogub2, Chindera Cyril1, Happiness
Nnani1
1Department of Petroleum Engineering, Federal University of Technology, Owerri, Imo State, Nigeria
2Department of Software Engineering, Federal University of Technology, Owerri, Imo State, Nigeria
3Department of Petroleum Engineering, School of Mining and Geosciences, Nazarbayev University, Kazakhstan

ABSTRACT

Computer intelligent models are the order of the day for the manipulation of data to better understand the trend
of complex situations under the questioned industry. The petroleum industry is faced with multiple datasets from
production logging tools, and predictive analysis without these computer intelligent tools can be devastating. Errors
of margins under these circumstances cannot be easily prevented, which may eventually lead to some biases in the
decision-making processes, thereby affecting the cost of operations and services in the industry. The proposed
computer intelligent models aimed at providing the best predictive analysis that was subjected through MATLAB
machine learning; model fitting-multilinear regression and deep learning; artificial neural network under the
feedforward function. This study however used an open-source dataset from a production well logging tool to evaluate
and affirm the accuracy of a computer intelligent model, suitable for processing complex problems. The predictive
analysis whiles considering the coefficient of determination ( R2) for these two models resulted that, the artificial neural
network feedforward function was better in predicting the flowing bottom-hole pressures at an R2 valued at 0.9787
than the multilinear regression whose R2 was valued at 0.6830. For a more accurate assessment of the performance
of the suggested models, the validation of this study took into account certain prior literature under a comparison data
analytical outcome. Multiphase flowing bottom-hole pressures considering Navier-Stokes equations can further be
computed using CFD under a finite element analysis to determine pressure drop differentials, predicting flow patterns
and geometry to enhance prudent decision-making analysis.

Keywords: MATLAB, Artificial Neural Network, Multilinear Regression, Machine Learning, Multiphase Fluid Flow,
Flowing Bottom-Hole Pressure

Introduction 2018). Several authors have used different approaches

to predict petroleum production in different oil and gas
The future yearns for more sustainable energy, fields. Information or data from existing well can be
and fossil fuel is still seen as the darling of fueling the used to predict the production performance of both
world, though the search for stable and safe energy is existing and new wells (Cao et al., 2016), but (Makinde
still on the front burner of almost all energy importers & Lee, 2019) used production data from simulated fluid
or consumers (Karakurt et al., 2012) for socioeconomic production to forecast production in unconventional
development. Carbon capture and storage application reservoir sources using Principal Component
are making decarbonization a possibility, thus reasons Methodology (PCM). Long-short-term memory (LSTM)
to ensure reservoir engineers and simulators predict an neural network has been used (W. Liu et al., 2019) to
accurate flow of fluids into the wellbore for further predict petroleum production in a China oil field.
production to the surface. The surface production Data-driven approaches (W. Liu et al., 2020; Nait
handling equipment must be adequately sized for Amar & Zeraibi, 2020; Sanusi et al., 2021) have taken
optimum production. the center stage in oil and gas industry for over two
For the accurate and optimum design of decades. This has been applied in different areas of the
production equipment, multiphase flow correlations industry (Abou-Sayed, 2012; Ali, 1994; Anderson, 2017;
(empirical, mechanistic or numerical) have been Hajizadeh, 2019; Shirangi, 2012; Sircar et al., 2021a;
employed for pressure drop prediction (Abdul-Majeed Syed, AlShamsi, et al., 2021; Teixeira & Secchi, 2019),
& Al-Mashat, 2000; Falcone et al., 2001; Tariq et al., some researchers do consider data analytics approach


Corresponding author: [email protected]
16010
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

as a preferred alternative to first principal equations available data when compared with commonly used
for predictions purposes. Machine Learning (ML) correlations in the petroleum industry. More realistic
algorithm has been used (Bhandari et al., 2015; multiphase flow characteristics using ML algorithms
Castiñeira et al., 2018; Dunlop et al., 2011; Noshi & compared with existing correlations has been used in
Schubert, 2018) in improving drilling operations, while predictions (Seong et al., 2020), (Simcenter STAR-
(F. Anifowose et al., 2017; F. A. Anifowose et al., 2017a, CCM+, 2021).
2017b; Esmaili & Mohaghegh, 2016; Hassanvand et al., Computational analysis to the prediction (Abdul-
2018; K. Liu et al., 2021; Ossai & Duru, 2021; Putcha Majeed et al., 2022) of the flowing bottom-hole pressure
& Ertekin, 2018; Salem et al., 2019; Sina et al., 2016) are by far seen to be relevant by the use of neural
applied it in the area of petroleum reservoirs, as well networks and multi-linear regressions (MLR)
as in petroleum exploration (Syed, Alnaqbi, et al., (Mamudu et al., 2021; E.-S. A. Osman, 2001;
2021). It is evident that, this approach can be employed Ponomareva et al., 2021; Sami & Ibrahim, 2021) based
to access information for decision-making in reservoir on sampled field data employed from carbonate
management (Rathnayake et al., 2022a). reservoirs. A decrease in flowing bottom-hole pressure
Down-hole conventional or smart tools used for is an increase in multiphase (oil, water and gas) flow,
gauging and measuring pressures in oil and gas wells however this depends solely on the flow geometry and
are sometimes unanimously inaccurate. More so, the properties of the fluid understudy (Firouzi &
analytics of these data gives a concise usage of the Rathnayake, 2019). In comparison to conventional
variables relating it to another for a meaningful wells, unconventional wells multiphase flow has a
prediction of flowing bottom-hole pressures. significant variation in terms of interfacial interactions
These gauges are as well very expensive to employ of fluids, depth and time accounts for various flow
and needs further expertise (Patel et al., 2014) to work regimes (Firouzi & Rathnayake, 2019; Wu et al., 2019).
and analyze results (D’Almeida et al., 2022). Being a Data is essential for analytical predictions
multiphase fluid flow, chances of producing fine grains (Barrufet et al., 1995), the computational intelligent
(sand) along with the reservoir fluid exists. This application had been seen creeping (Andrianova et al.,
scenario could cause sand erosion and/or partial 2018; Nikitin et al., 2022) into the petroleum industry
blockage of the pipe due to these fine grains and (Kern & Nicholson, 1965), solving complex problems
adversely affect the down-hole measuring tools related to operations and management. In the early
resulting in low accurate readings (Marshall & 21st century, the data compiled from fields of seismic
Thomas, 2015). The use of these gauges affects operations, EOR, and IOR (Tariq et al., 2020a) were
petroleum production negatively (Awadalla et al., handled simply by super computers with suitable
2016; Feng et al., 2016), suggesting more realistic graphic user interface (Makhotin et al., 2022). The
method of predicting FBHP for optimum production. forecast by this intelligence (Memon et al., 2015),
With available adequate production information reduces the equational errors that poses threats
from well test, according to (Economi, n.d.), flowing (Nwachukwu et al., 2018) to the production of flowing
bottom-hole pressure (FBHP) can be predicted from the fluids in the wellbore (Luo et al., 2018). More so, the
wellhead pressure. But, as an important factor as analytics of these data gives a concise usage of the
FBHP in multiphase flow pressure drop predictions, its variables relating it to another for a meaningful
accuracy cannot be compromised. The work by (E. S. A. prediction of flowing bottom-hole pressures.
Osman & Aggour, 2002) confirmed ML can actually Computational simulation tools (Khudaier et al.,
predict multiphase pressure drop with very high 2020) such as artificial neural networks are used in
accuracy. Multiphase flow predictive models developed solving non-linear problems, originating from the
by different authors using ML in the recent time gave theory of biological neurons (Graves et al., 2009). The
better performance and realistic result than existing study, (Bahaa et al., 2019) predicted accurate flowing
empirical or numerical models; (Ristanto, 2018) bottom-hole pressures without having to use expensive
considered ML performance in dynamic environment tools for measurement and analysis. Prediction from
like multiphase flow more realistic than physical artificial neuron network (ANN) were considered from
models, another study (Babanezhad et al., 2020) used a data set obtained from production logging tools, the
adaptive neuro-fuzzy inference system (ANFIS) due to model yielded good results (Ahmed & Ayoub, 2014;
its capability in training complex relationships. Sircar et al., 2021b). Accordingly, (Tariq, 2018),
Different multiphase flow algorithms have been used considered the use of ANN for prediction in vertical
to predict two-phase flowrate against challenges with wells by comparing previous predicted results
individual phase measurements in the past (Wang et emanating (Hørsholt et al., 2019) from production data.
al., 2020). Some researchers (Lin et al., 2020), The survey from further research conducted in
predicted preferred flow pattern for upward inclined literature and methods (al Shehri et al., 2020;
pipes with deep learning neural network using Ponomareva et al., 2021; Rathnayake et al., 2022b),
16011
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

demonstrated how to forecast and evaluate flowing Materials and Methods

bottom-hole pressure using artificial neural networks
and machine learning algorithms (Tadjer et al., 2021). Data Source and Analytical Models
The objective of this current study is to propose The actual production data came from down-hole
computational models that best fits an effective measurements that were recorded. These naturally
predictive analysis tool, and to validate and affirm the flowing bottom-hole pressure measurements (Ahmadi
results of the findings so as to reduce time-cost index & Chen, 2019; Martyushev, 2021) in the carbonate
regarding the usage of complex and expensive reservoir were taken above the perforated production
machines for predictive analysis. Fluid flow dynamics tubing without the use of any artificial lift systems
for this study made considerable assumptions to the (Asheim, 1986; Tariq et al., 2020a). The quality of the
records of the data that were obtained from intelligent 206 data point as shown in Table 1 & 2 with predictor
or smart production well logging tools to be in its variables; oil, gas and water flow rates, depth, internal
accurate original form (Sircar et al., 2021b). As a result, diameter of the tubular, well head and bottom-hole
the scope of this investigation was confined to surface pressures, temperatures, oil gravity were imported into
measured data that was readily available. MATLAB for the predictive analysis of flowing bottom-
hole pressure using ANN and Multilinear Regression.

Table 1: 206 Data Point Source Sampled (Asheim, 1986)

4600 2693.37 11000 4 6621 32.6 90 212 175 2804

700 411.69 1300 2.441 6271 32.6 90 212 230 2368
8616 4230.46 2500 3.813 6294 36.5 156 208 285 2343
2983 1023.09 905 3.958 6345 32.8 90 180 220 2161
3792 1749.74 3796 4 6340 32.6 90 212 210 2289
1079 593.45 900 1.995 6565 36 156 211 230 2641
14624 8057.82 1256 3.958 6518 36 157 210 250 2565
3500 1968.61 5800 3.958 6653 32.6 90 212 320 2615
4600 2693.37 11000 4 6621 32.6 90 212 175 2804
…
13629 5162.91 458 3.958 6285 32.6 90 212 280 2349

Table 1: Predictor Variables

Variables Field Units
Flowing oil rate (𝒒𝒐 ) bbl/day
Flowing gas rate (𝒒𝒐 ) Mscf/day
Flowing water rate (𝒒𝒐 ) bbl/day
Production tubing internal diameter Inches
Input (𝒙) (ID)
Well perforation depth (Depth) Ft
Oil gravity (API) API
Surface temperature (STM) oF

Well bottom-hole temperature (BTM) oF

Well-head pressure (Pwh) Psia

Output (𝒚) Flowing Bottom-Hole Pressure Psia
(FBHP)

Multilinear Regression
The complexity of the dataset requires a
multilinear relationship of the nine independent
variables (x) and a dependent variable (y) as shown in
Table 1. The relationship of these two; actual variables
x was used to predict the values of y, flowing bottom-

16012
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

hole pressure. The regression is mathematically

modelled with Eq. 1 and Eq. 2 (Firouzi & Rathnayake,
2019);
𝒚 = 𝜷𝒐 + 𝒙𝑻 𝜷 + 𝝐 (1)

(2)

Where 𝑥 is a predictive vector, 𝜖, is an error term, independent variables, x (input) through to a series of
𝛽𝑜 and 𝛽 are simulation of estimated parameters based hidden layers to a dependent variable, y (output).
on the data provided (Firouzi & Rathnayake, 2019; Therefore, the relationship to this is given by Eq. 3; y
Rathnayake et al., 2022b). Ordinary least squares = f(x1𝜃)
(OLS) is a method for estimating model parameters Eq. 3 is for the case of a multilayer function, and
that minimizes the sum of squares differences between Eq. 4 illustrates the output of the 1st layer, where g, bj
observed and fitted values. However, a technique to and wj are hyperbolic tangent, bias and the inputs of x
improve bias-variance, automatic variable selection, respectively.
using predictive analysis in Eq. 2 can be used to further
penalize or evaluate OLS results (Tibshirani, 1996). (3)
The illustration in Fig. 1 is the structural
Artificial Neural Networks framework or flow chat of how the ANN-feedforward
Flowing bottom-hole pressures (FBHP) were not function worked which contained nine inputs, one
only analyzed by multilinear regressions but a hidden layer with 30 neurons and an output. The
feedforward neural network was applied. The resilient neural network relationship of the input layer and the
network had a complex data to analyze, this multilayer hidden layer is the sigmoidal transfer function, and the
function sort to perform and handle non-linear network function between the hidden layer and the
relationships. In simple terms, the neural network of output is purely linear.
feedforward function processes or models the
Layer 2: 30 neurons, hidden
layers
Gas flow rate

Oil flow rate

Water flow rate

Tubing ID
Flowing Bottom-Hole Pressure
Depth (FBHP)
Oil gravity

Surface Temperature
nth
Layer 3: Output layers
Bottom Temperature

Well head Pressure

Layer 1: 9 inputs
(a)

(b)
Fig. 1 FBHP Structural Neuron Networks (a) ANN-Flow Chart (b) ANN-Matlab
16013
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Comparative Predictive Analysis takes the distance from points to the regression line
The assessment of the best model for predictions (the errors) and squares them. The steps involved in
were considered based on; Mean Square Error (MSE), calculating the mean square errors were:
Root Mean Square Error (RMS), Mean Absolute Error
(MAE), Median Absolute Percentage Error (MAPE), i. Find the regression line
and Coefficient of Determination ( R2), while the ii. Insert the X values into the regression equation
emphasis of the comparative analysis was based on to find the new Y values
Coefficient of Determination. iii. Subtract the new Y value from the original to
get the error
Mean Square Error (MSE): iv. Square the errors
Mean square error also known as the mean square v. Add the errors
deviation is the average squared difference between vi. Find the mean
the estimated value and the actual value (Eq. 5). It Mathematically,
𝟏
𝐌𝐒𝐄 = 𝐧
∗ (𝐚𝐜𝐭𝐮𝐚𝐥 − 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭)2 (4)

Root Mean Square Error (RMSE):

Root mean square error is the square root of the mean square error of all the errors (Eq. 6). Mathematically, it
can be written as
𝟏
𝐑𝐌𝐒𝐄 = ∗ (𝐚𝐜𝐭𝐮𝐚𝐥 − 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭) (5)
𝐧
2

Mean Absolute Error (MAE): It is the absolute value of the difference between an
This is the absolute error measured between observed value of a quantity and the true value (Eq. 7).
paired observations expressing the same phenomenon. Mathematically,
𝐧
𝟏
𝐌𝐀𝐄 = ∗ |𝐩𝐫𝐞𝐝𝐢𝐜𝐭𝐢𝐨𝐧 − 𝐭𝐫𝐮𝐞 𝐯𝐚𝐥𝐮𝐞| (6)
𝐧
𝐢=𝟏

Median Absolute Percentage Error (MAPE):

This is a measure of how accurate a forecast system is and measures the accuracy as a percentage (Eq. 8).
Mathematically,
𝐧
𝟏 𝐚𝐜𝐭𝐮𝐚𝐥 − 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭 𝐯𝐚𝐥𝐮𝐞
𝐌= ∗ (7)
𝐧 𝐚𝐜𝐭𝐮𝐚𝐥 𝐯𝐚𝐥𝐮𝐞
𝐭=𝟏

Coefficient of Determination (R2) model; this determines the extent of accuracy of the
This examines how the difference in one variable data. However, when the relationship of the data is
can be explained by the discrepancy in a second said to be closer to one, a higher degree of accuracy is
variable when predicting the outcome of any given achieved.
event (Eq. 9) and it is the base for every regression

𝐬𝐮𝐦 𝐨𝐟 𝐬𝐪𝐮𝐚𝐫𝐞𝐬 𝐨𝐟 𝐫𝐞𝐬𝐢𝐝𝐮𝐚𝐥𝐬

𝐑𝟐 = 𝟏 − (8)
𝐭𝐨𝐭𝐚𝐥 𝐬𝐮𝐦 𝐨𝐟 𝐬𝐪𝐮𝐚𝐫𝐞𝐬

Results and Discussions using the coefficient of determination as a measuring

tool, the results obtained determined which model was
Data drives all analysis, and for this study, the best fitted for the analysis.
open-source data from production logging tools were
used to predict the flowing bottom-hole pressures Regression (MLR)
based on the predictor variables. All variables for the Model fitting under multilinear regression was
inputs were imported and plotted for an outcome, and used to analyze the data. The predictor variables were
16014
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

all inclusive as against the outcome, where the actual from pressure of 2000 psi to 3,000 psi; this was an
FBHP was used against the predicted FBHP. The indication of an increase in flow rate due to an increase
scattered diagram from Fig. 2 demonstrates a in pressure. But the desired pressure was spotted to be
multiphase flow of liquid and gas with the influence of centered around 2300 psi to 2,800 psi under a high
tubing size, depth, pressure, temperature and gravity. dispersity rate.
However, the flow pattern on the graph was seen rising

Fig. 2 Actual and Predicted FBHP (MLR)

The coefficient of determination (adjusted R2) was measure of accuracy as shown by MATLAB statistical
valued at 0.6830; this explains the rate of predictive measure in Table 3.

Table 2: MLR Coefficients from MATLAB

Artificial Neural Network (ANN- demonstrated in Fig. 3. It could be seen that the
FEEDFORWARD) dispersion of the flow is closely narrowed on the
This predictive computational method was used to diagonal line, and flow is also spotted increasing at an
analyze flowing bottom-hole pressures using the increase in pressure. The central mass flow rates were
feedforward function. The analysis was drawn based on positioned between 1,900 psi to 2,700 psi. However, any
the predictor variables under same wellbore pressure beyond this point would reduce the
conditions. The actual flowing bottom hole pressures production.
were matched against the predictive values as
16015
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Fig. 3 Actual and Predicted FBHP (ANN)

The coefficient of determination ( R2) for this demonstrated in Fig. 4. This further explains the rate
trained and tested data was valued at 0.9787 as of predictive measure accuracy in Table 4.

Fig. 4 The repercussions of training, validation,

and testing FBHP dataset

Table 3: Collation of Predictive Model's Coefficient of

Determinations
Models R2 Adjusted R2
Multilinear Regression 0.6970 0.6830
Artificial Neural 0.9787 -
Network

As the anticipated and actual values were difference between two calculated numbers was
compared at the output layer, an error defined as the however conveyed back to the layers in order to fine-
16016
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

tune the weights and biases in order to improve the with progressive training runs, ANN amends
system's predictive competence as shown in Fig. 5. iteratively and improves.
Epoch is the name given to this iteration process, and

Fig. 5 Iterated output layer

Also, whiles the datasets were being trained and value of -8.517 on the x-axis. The yellow line in the
tested, the feedforward function from the neural error histogram is the zero error, and as the training
network plotted an error distribution histogram in Fig. dataset rises to 50 on the y-axis, the validation and
6. This was the difference in errors between the target testing fall within 50 and 60 respectively, which
values and the predicted or output values; meaning means, the error was within the range of the sampled
that, the values predicted were drifting away from the dataset.
targeted values and thereby resulting in a negative

Fig. 6 Error Histogram

EMPIRICAL, MLR AND ANN MODELS in order to determine which model was better for
Several authors have attempted to develop predicting flowing bottom-hole pressure. The stream of
appropriate coefficients of determination using a models employed utilizes the vigorous coefficient of
mathematical and logical approach of training, determination (R2) for analysis based on same datasets
validating, and testing datasets. This study's proposed used by these authors, as shown in Table 5.
models were compared and assessed for discrepancies

16017
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Table 4: Comparative Analysis of Models for R2

Year Authors/Model R2
1986 Kabir and Hasan (Kabir & Hasan, 1986) 0.7502
1994 Ansari (Ansari et al., 1994) 0.8178
1996 Chokshi (Chokshi et al., 1996) 0.8836
1999 Gomez (Gomez et al., 1999) 0.8324
1967 Orkiszewski (Orkiszewski & Houston, 0.9015
1967)
2020 Tariq (PSO_ ANN) (Tariq et al., 2020b) 0.9830
MLR_ Proposed 0.6830
ANN_ proposed 0.9787

Coefficient of Determination
1/2
1
0/8
0/6
0/4
0/2
0

ANN_ proposed
MLR_ Proposed
Chokshi

Gomez
Ansari

Orkiszewski
Kabir and Hasan

Tariq

1986 1994 1996 1999 1967 2020 Current Current

Fig. 7 Coefficient of Determination History and its predictive efficiency

An empirical approach for determining the operations. The potency of these employed computer
coefficient of determination in 1986 was 0.7502 (Kabir intelligent models has by far exhibited a quick review
& Hasan, 1986) as shown in Table 5, which was an of what to expect in the wellbore, reducing the cost of
improvement on (Orkiszewski & Houston, 1967) of an engaging well pressure surveyors in the industry.
R2 of 0.9015. In comparison as demonstrated in Figure The objectivity of reaching the desired result and
7 (Orkiszewski & Houston, 1967), the statistical affirming which computer intelligent model was
analysis of this study using MLR yielded a suitable for the predictive analysis of the production
substantially lower R2 value of 0.6830, and ANN at a logging tool was noted to have been the feedforward
most efficient value of 0.9787. The result obtained in function from the artificial neural network, with a
the work (Tariq et al., 2020b) using the particle swarm coefficient of determination R2 of 0.9787, however, the
optimization ANN (PSO_ANN) model, achieved the multilinear regression gave a less accurate adjusted R2
highest and most efficient R2 of 0.9830. The disparity value of 0.6830 in predicting FBHP. The result
between this study’s MLR, ANN, the PSO-ANN and obtained from the feedforward ANN was very close to
empirical models, on the other hand, reveals the most Tariq’s PSO_ANN model result of 0.9830 and supports
successful forecasting model for FBHP. the claim of (Ossai & Duru, 2021) of ANN performance
over other AI techniques dues it’s high accuracy.
Discussion and Conclusions The cost effectiveness to running sophisticated
logging tools had been greatly achieved by the
This study using big data from the production application of an accurate predictive computer
wells considered a computational analysis approach to intelligent model using a real time production logging
monitor and predict the likely event of maximizing oil data.
production by predicting flowing bottom-hole pressures We do recommend further research on the
to support and eliminate the biased decision taken influence of each flowrate on fluid dynamics with
during well completion and petroleum production
16018
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

respect to the temperature differences using same Abou-Sayed, A. (2012). Data Mining Applications in
production logging data in the wellbore. the Oil and Gas Industry. Journal of Petroleum
Flowing bottom-hole pressure can further be Technology, 64(10), 88–95.
exploited through a 3D surface plot and from a https://doi.org/10.2118/1012-0088-jpt
computational fluid dynamic outlook, identifying and Ahmadi, M. A., & Chen, Z. (2019). Machine learning
validation pressure drops using Navier-Stokes models to predict bottom hole pressure in multi-
equation. phase flow in vertical oil production wells.
Canadian Journal of Chemical Engineering,
Acknowledgements 97(11), 2928–2940.
https://doi.org/10.1002/CJCE.23526
We acknowledge all the authors whose scientific Ahmed, M. M., & Ayoub, M. A. (2014). A
research work was cited in this study for their Comprehensive Study on the Current Pressure
contribution to the body of knowledge. Drop Calculation in Multiphase Vertical Wells;
Current Trends and Future Prospective. In
Declaration Journal of Applied Sciences (Vol. 14, Issue 23, pp.
3162–3171).
Funding https://doi.org/10.3923/jas.2014.3162.3171
The authors declare that this study did not receive al Shehri, F. H., Gryzlov, A., al Tayyar, T., & Arsalan,
funds from any governmental, private, or non-profit M. (2020). Utilizing machine learning methods to
funding bodies. estimate flowing bottom-holepressure in
unconventional gas condensate tight sand
Conflict of Interest fractured wells in Saudi Arabia. Society of
We hereby state that there are no known Petroleum Engineers - SPE Russian Petroleum
competing financial interests or personal ties that Technology Conference 2020, RPTC 2020.
could have influenced the research presented in this https://doi.org/10.2118/201939-MS
study. Ali, J. K. (1994). Neural networks: A new tool for the
petroleum industry? Society of Petroleum
Code availability Engineers - European Petroleum Computer
The code used in this study is available on request. Conference 1994, EPCC 1994, 233–242.
Credit Authorship Contribution Statement https://doi.org/10.2523/27561-ms
Ugochukwu I. Duru: Conceptualization, Anderson, R. N. (2017). “Petroleum Analytics Learning
Methodology, Formal analysis, Writing - review & Machine” for optimizing the Internet of Things of
editing and Supervision. Dennis Wayo: Methodology, today’s digital oil field-to-refinery petroleum
Data curation, Formal analysis, Writing - original system. Proceedings - 2017 IEEE International
draft. Reginal Ogu: Methodology, Data curation, Conference on Big Data, Big Data 2017 , 2018-
Formal analysis and Investigation and Writing - Janua, 4542–4545.
original draft. Chidera Cyril and Happiness C. Nnani: https://doi.org/10.1109/BigData.2017.8258496
Conceptualization, Coding, Investigation and Writing Andrianova, A., Simonov, M., Perets, D., Margarit, A.,
– original draft. Serebryakova, D., Bogdanov, Y., Budennyy, S.,
Volkov, N., Tsanda, A., & Bukharev, A. (2018).
References Application of machine learning for oilfield data
quality improvement. Society of Petroleum
Abdul-Majeed, G. H., & Al-Mashat, A. M. (2000). A Engineers - SPE Russian Petroleum Technology
mechanistic model for vertical and inclined two- Conference 2018, RPTC 2018.
phase slug flow. Journal of Petroleum Science and https://doi.org/10.2118/191601-18rptc-ms
Engineering, 27(1–2), 59–67. Anifowose, F. A., Labadin, J., & Abdulraheem, A.
https://doi.org/10.1016/S0920-4105(00)00047-4 (2017a). Ensemble machine learning: An
Abdul-Majeed, G. H., Kadhim, F. S., Almahdawi, F. H. untapped modeling paradigm for petroleum
M., Al-Dunainawi, Y., Arabi, A., & Al-Azzawi, W. reservoir characterization. Journal of Petroleum
K. (2022). Application of artificial neural network Science and Engineering, 151, 480–487.
to predict slug liquid holdup. International https://doi.org/10.1016/J.PETROL.2017.01.024
Journal of Multiphase Flow, 150(February), Anifowose, F. A., Labadin, J., & Abdulraheem, A.
104004. (2017b). Hybrid intelligent systems in petroleum
https://doi.org/10.1016/j.ijmultiphaseflow.2022.10 reservoir characterization and modeling: the
4004 journey so far and the challenges ahead. Journal
of Petroleum Exploration and Production
16019
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Technology, 7(1), 251–263. Resources Symposium.

https://doi.org/10.1007/S13202-016-0257-3 https://doi.org/10.2118/180984-ms
Anifowose, F., Khoukhi, A., & Abdulraheem, A. (2017). Castiñeira, D., Toronyi, R., & Saleri, N. (2018).
Investigating the effect of training–testing data Machine learning and natural language
stratification on the performance of soft processing for automated analysis of drilling and
computing techniques: an experimental study. completion data. Society of Petroleum Engineers -
Journal of Experimental and Theoretical Artificial SPE Kingdom of Saudi Arabia Annual Technical
Intelligence, 29(3), 517–535. Symposium and Exhibition 2018, SATS 2018.
https://doi.org/10.1080/0952813X.2016.1198936 https://doi.org/10.2118/192280-MS
Ansari, A. M., Sylvester, N. D., Sarica, C., Shoham, O., Chokshi, R. N., Schmidt, Z., & Doty, D. R. (1996).
& Brill, J. P. (1994). A Comprehensive Experimental Study and the Development of a
Mechanistic Model for Upward Two-Phase Flow in Mechanistic Model for Two-Phase Flow Through
Wellbores. SPE Production & Facilities, 9(02), Vertical Tubing. All Days.
143–151. https://doi.org/10.2118/20630-PA https://doi.org/10.2118/35676-MS
Asheim, H. (1986). MONA, An Accurate Two-Phase D’Almeida, A. L., Bergiante, N. C. R., de Souza
Well Flow Model Based on Phase Slippage. SPE Ferreira, G., Leta, F. R., de Campos Lima, C. B.,
Production Engineering, 1(03), 221–230. & Lima, G. B. A. (2022). Digital transformation: a
https://doi.org/10.2118/12989-PA review on artificial intelligence techniques in
Awadalla, M., Yousef, H., Al-Hinai, A., & Al-Shidani, drilling and production applications.
A. (2016). Prediction of Oil Well Flowing Bottom- International Journal of Advanced Manufacturing
hole Pressure in Petroleum Fields. Proceedings of Technology, 119(9–10), 5553–5582.
the International Conference on Industrial https://doi.org/10.1007/s00170-021-08631-w
Engineering and Operations Management, 8-10 Dunlop, J., Isangulov, R., Aldred, W. D., Arismendi
March, 3007–3017. Sanchez, H., Sanchez Flores, J. L., Alarcon
Babanezhad, M., Taghvaie Nakhjiri, A., Rezakazemi, Herdoiza, J., Belaskie, J., & Luppens, J. C. (2011).
M., Marjani, A., & Shirazian, S. (2020). Functional Increased rate of penetration through automation.
input and membership characteristics in the SPE/IADC Drilling Conference, Proceedings, 1,
accuracy of machine learning approach for 442–452. https://doi.org/10.2118/139897-MS
estimation of multiphase flow. Scientific Reports, Economi, M. J. (n.d.). Michael J Economi des.
10(1), 1–15. https://doi.org/10.1038/s41598-020- Measurement, 139–151.
74858-4 Esmaili, S., & Mohaghegh, S. D. (2016). Full field
Bahaa, M., Shokir, E., & Mahgoub, I. (2019). Soft reservoir modeling of shale assets using advanced
computation application: Utilizing Artificial data-driven analytics. Geoscience Frontiers, 7(1),
Neural Network to Predict the Fluid Rate and 11–20. https://doi.org/10.1016/J.GSF.2014.12.006
Bottom Hole Flowing Pressure for Gas-lifted Oil Falcone, G., Hewitt, G. F., Alimonti, C., & Harrison, B.
Wells. Society of Petroleum Engineers - Abu Dhabi (2001). Multiphase Flow Metering: Current
International Petroleum Exhibition and Trends and Future Developments. Proceedings -
Conference 2018, ADIPEC 2018. SPE Annual Technical Conference and Exhibition,
https://doi.org/10.2118/193052-MS 1291–1303. https://doi.org/10.2118/71474-MS
Barrufet, M. A., Rasool, A., & Aggour, M. (1995). Feng, Q.-H., Gao, S.-C., Zhang, J.-C., & Ye, X.-C.
Prediction of bottomhole flowing pressures in (2016). Well bottom-hole flowing pressure
multiphase systems using a thermodynamic evaluation method in 48 Block of S Gas field .
equation of state. Society of Petroleum Engineers Iceep, 186–189. https://doi.org/10.2991/iceep-
- SPE Production Operations Symposium 1995, 16.2016.31
POS 1995, 1995-April, 369–380. Firouzi, M., & Rathnayake, S. (2019). Prediction of the
https://doi.org/10.2523/29479-ms flowing bottom-hole pressure using advanced data
Bhandari, J., Abbassi, R., Garaniya, V., & Khan, F. analytics. SPE/AAPG/SEG Asia Pacific
(2015). Risk analysis of deepwater drilling Unconventional Resources Technology Conference
operations using Bayesian network. Journal of 2019, APUR 2019, November.
Loss Prevention in the Process Industries, 38, 11– https://doi.org/10.15530/ap-urtec-2019-198240
23. https://doi.org/10.1016/J.JLP.2015.08.004 Gomez, L. E., Shoham, O., Schmidt, Z., Chokshi, R. N.,
Cao, Q., Banerjee, R., Gupta, S., Li, J., Zhou, W., & Brown, A., & Northug, T. (1999). A Unified
Jeyachandra, B. (2016). Data driven production Mechanistic Model for Steady-State Two-Phase
forecasting using machine learning. Society of Flow in Wellbores and Pipelines. All Days.
Petroleum Engineers - SPE Argentina https://doi.org/10.2118/56520-MS
Exploration and Production of Unconventional
16020
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Graves, A., Liwicki, M., Fernández, S., Bertolami, R., Using Bottom-Hole Pressure Data. Journal of
Bunke, H., & Schmidhuber, J. (2009). A novel Energy Resources Technology, 142(10).
connectionist system for unconstrained https://doi.org/10.1115/1.4047304
handwriting recognition. IEEE Transactions on Luo, G., Tian, Y., Bychina, M., & Ehlig-Economides, C.
Pattern Analysis and Machine Intelligence , 31(5), (2018). Production optimization using machine
855–868. https://doi.org/10.1109/TPAMI.2008.137 learning in bakken shale. SPE/AAPG/SEG
Hajizadeh, Y. (2019). Machine learning in oil and gas; Unconventional Resources Technology Conference
a SWOT analysis approach. Journal of Petroleum 2018, URTC 2018, 2011.
Science and Engineering, 176(December 2018), https://doi.org/10.15530/urtec-2018-2902505
661–663. Makhotin, I., Orlov, D., & Koroteev, D. (2022). Machine
https://doi.org/10.1016/j.petrol.2019.01.113 Learning to Rate and Predict the Efficiency of
Hassanvand, M., Moradi, S., Fattahi, M., Zargar, G., & Waterflooding for Oil Production. Energies, 15(3).
Kamari, M. (2018). Estimation of rock uniaxial https://doi.org/10.3390/en15031199
compressive strength for an Iranian carbonate oil Makinde, I., & Lee, W. J. (2019). Principal components
reservoir: Modeling vs. artificial neural network methodology – A novel approach to forecasting
application. Petroleum Research, 3(4), 336–345. production from liquid-rich shale (LRS) reservoirs.
https://doi.org/10.1016/J.PTLRS.2018.08.004 Petroleum, 5(3), 227–242.
Hørsholt, S., Nick, H., & Jørgensen, J. B. (2019). A https://doi.org/10.1016/j.petlm.2018.08.002
hierarchical multigrid method for oil production Mamudu, A., Khan, F., Zendehboudi, S., & Adedigba,
optimization. IFAC-PapersOnLine, 52(1), 492– S. (2021). Dynamic risk modeling of complex
497. https://doi.org/10.1016/j.ifacol.2019.06.110 hydrocarbon production systems. Process Safety
Kabir, C. S., & Hasan, A. R. (1986). A Study of and Environmental Protection, 151, 71–84.
Multiphase Flow Behavior in Vertical Oil Wells: https://doi.org/10.1016/j.psep.2021.04.046
Part II-Field Application. Society of Petroleum Marshall, C., & Thomas, A. (2015). Maximising
Engineers of AIME, (Paper) SPE, 2, 479–491. economic recovery - A review of well test
https://doi.org/10.2118/15139-MS procedures in the North Sea. Society of Petroleum
Karakurt, I., Aydin, G., & Aydiner, K. (2012). Sources Engineers - SPE Offshore Europe Conference and
and mitigation of methane emissions by sectors: A Exhibition, OE 2015.
critical review. Renewable Energy, 39(1), 40–48. https://doi.org/10.2118/175518-ms
https://doi.org/10.1016/j.renene.2011.09.006 Martyushev, D. A. (2021). Experimental study of the
Kern, C. P., & Nicholson, F. R. (1965). Practical influence of bottomhole pressure of producing
Application of Calculated Multiphase Flowing wells on reserve production from complicated
Bottom-Hole Pressures. Journal of Petroleum carbonate reservoirs. Bulletin of the Tomsk
Technology, 17(12), 1373–1378. Polytechnic University, Geo Assets Engineering,
https://doi.org/10.2118/1218-PA 332(5), 110–119.
Khudaier, Y. abbas, Kadhim, F. S., & Yousif, Y. K. https://doi.org/10.18799/24131830/2021/05/3190
(2020). Using Adaptive Neuro Fuzzy Inference Memon, P. Q., Yong, S. P., Pao, W., & Pau, J. S. (2015).
System to Predict Rate of Penetration from Dynamic well bottom-hole flowing pressure
Dynamic Elastic Properties. Journal of prediction based on radial basis neural network.
Engineering, 26(7), 45–61. Studies in Computational Intelligence, 591, 279–
https://doi.org/10.31026/j.eng.2020.07.04 292. https://doi.org/10.1007/978-3-319-14654-6_17
Lin, Z., Liu, X., Lao, L., & Liu, H. (2020). Prediction of Nait Amar, M., & Zeraibi, N. (2020). A combined
two-phase flow patterns in upward inclined pipes support vector regression with firefly algorithm
via deep learning. Energy, 210, 118541. for prediction of bottom hole pressure. SN Applied
https://doi.org/10.1016/j.energy.2020.118541 Sciences, 2(1). https://doi.org/10.1007/S42452-019-
Liu, K., Xu, B., Kim, C., & Fu, J. (2021). Well 1835-Z
Performance from Numerical Methods to Machine Nikitin, N. O., Revin, I., Hvatov, A., Vychuzhanin, P.,
Learning Approach: Applications in Multiple & Kalyuzhnaya, A. v. (2022). Hybrid and
Fractured Shale Reservoirs. Geofluids, 2021. automated machine learning approaches for oil
https://doi.org/10.1155/2021/3169456 fields development: The case study of Volve field,
Liu, W., Liu, W. D., & Gu, J. (2019). Petroleum North Sea. Computers and Geosciences, 161.
production forecasting based on machine learning. https://doi.org/10.1016/j.cageo.2022.105061
ACM International Conference Proceeding Series, Noshi, C. I., & Schubert, J. J. (2018). The role of
124–128. https://doi.org/10.1145/3373419.3373421 machine learning in drilling operations; a review.
Liu, W., Liu, W. D., & Gu, J. (2020). A Machine SPE Eastern Regional Meeting, 2018-October.
Learning Method to Infer Inter-Well Connectivity https://doi.org/10.2118/191823-18ERM-MS
16021
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Nwachukwu, A., Jeong, H., Pyrcz, M., & Lake, L. W. Salem, K. G. S. K. G., Abdulaziz, A. A. M. A. M., &
(2018). Fast evaluation of well placements in Dahab, A. S. D. A. S. A. (2019). Prediction of
heterogeneous reservoir models using machine hydraulic properties in carbonate reservoirs using
learning. Journal of Petroleum Science and artificial neural network. Society of Petroleum
Engineering, 163(December 2017), 463–475. Engineers - Abu Dhabi International Petroleum
https://doi.org/10.1016/j.petrol.2018.01.019 Exhibition and Conference 2018, ADIPEC 2018.
Orkiszewski, J., & Houston, T. (1967). Predicting Two- https://doi.org/10.2118/193007-MS
Phase Pressure Drops in Vertical Pipe. Journal of Sami, N. A., & Ibrahim, D. S. (2021). Forecasting
Petroleum Technology, 19(06), 829–838. multiphase flowing bottom-hole pressure of
https://doi.org/10.2118/1546-PA vertical oil wells using three machine learning
Osman, E. S. A., & Aggour, M. A. (2002). Artificial techniques. Petroleum Research, 6(4), 417–422.
neural network model for accurate prediction of https://doi.org/10.1016/j.ptlrs.2021.05.004
pressure drop in horizontal and near-horizontal- Sanusi, S., Omisore, A., Blankson, E., Anyanwu, C., &
multiphase flow. Petroleum Science and Eremiokhale, O. (2021). Estimation of Bottom
Technology, 20(1–2), 1–15. Hole Pressure in Electrical Submersible Pump
https://doi.org/10.1081/LFT-120002082 Wells using Machine Learning Technique. Society
Osman, E.-S. A. (2001). Artificial Neural Networks of Petroleum Engineers - SPE Nigeria Annual
Models for Identifying Flow Regimes and International Conference and Exhibition 2021,
Predicting Liquid Holdup in Horizontal NAIC 2021. https://doi.org/10.2118/207122-MS
Multiphase Flow. All Days. Seong, Y., Park, C., Choi, J., & Jang, I. (2020).
https://doi.org/10.2118/68219-MS Surrogate model with a deep neural network to
Ossai, C. I., & Duru, U. I. (2021). Applications and evaluate gas-liquid flow in a horizontal pipe.
theoretical perspectives of artificial intelligence in Energies, 13(4), 1–12.
the rate of penetration. Petroleum, 8(2), 237–251. https://doi.org/10.3390/en13040968
https://doi.org/10.1016/j.petlm.2020.08.004 Shirangi, M. G. (2012). Applying Machine Learning
Patel, P., Odden, H., Djoric, B., Garner, R. D., & Vea, Algorithms to Oil Reservoir Production
H. K. (2014). Model based multiphase metering Optimization. Stanford University, 1–5.
and production allocation. Proceedings of the Simcenter STAR-CCM+. (2021). Segregated Flow
Annual Offshore Technology Conference , 4, 3336– Guidelines. STAR CCM+ Help Guide, 2.
3344. https://doi.org/10.4043/25457-ms Sina, R. G., Ani, M., Oluyemi, G., & Petrovski, A.
Ponomareva, I. N., Galkin, V. I., & Martyushev, D. A. (2016). Reservoir uncertainty analysis: The trends
(2021). Operational method for determining from probability to algorithms and machine
bottom hole pressure in mechanized oil producing learning. Society of Petroleum Engineers - SPE
wells, based on the application of multivariate Intelligent Energy International Conference and
regression analysis. Petroleum Research, 6(4), Exhibition. https://doi.org/10.2118/181049-MS
351–360. Sircar, A., Yadav, K., Rayavarapu, K., Bist, N., & Oza,
https://doi.org/10.1016/j.ptlrs.2021.05.010 H. (2021a). Application of machine learning and
Putcha, V. B., & Ertekin, T. (2018). A hybrid integrated artificial intelligence in oil and gas industry.
compositional reservoir simulator coupling Petroleum Research, 6(4), 379–391.
machine learning and hard computing protocols. https://doi.org/10.1016/j.ptlrs.2021.05.009
Society of Petroleum Engineers - SPE Kingdom of Sircar, A., Yadav, K., Rayavarapu, K., Bist, N., & Oza,
Saudi Arabia Annual Technical Symposium and H. (2021b). Application of machine learning and
Exhibition 2018, SATS 2018. artificial intelligence in oil and gas industry.
https://doi.org/10.2118/192368-ms Petroleum Research, 6(4), 379–391.
Rathnayake, S., Rajora, A., & Firouzi, M. (2022a). A https://doi.org/10.1016/j.ptlrs.2021.05.009
machine learning-based predictive model for real- Syed, F. I., Alnaqbi, S., Muther, T., Dahaghi, A. K., &
time monitoring of flowing bottom-hole pressure of Negahban, S. (2021). Smart shale gas production
gas wells. Fuel, 317(February), 123524. performance analysis using machine learning
https://doi.org/10.1016/j.fuel.2022.123524 applications. Petroleum Research, xxxx.
Rathnayake, S., Rajora, A., & Firouzi, M. (2022b). A https://doi.org/10.1016/j.ptlrs.2021.06.003
machine learning-based predictive model for real- Syed, F. I., AlShamsi, A., Dahaghi, A. K., &
time monitoring of flowing bottom-hole pressure of Neghabhan, S. (2021). Machine Learning
gas wells. Fuel, 317(February), 123524. techniques to Model Geomechanics and
https://doi.org/10.1016/j.fuel.2022.123524 Petrophysical Properties of Shale Reservoirs – A
Ristanto, T. (2018). Machine Learning Applied to Systematic Literature Review. Petroleum, 8.
Multiphase Production Problems. June, 1–71. https://doi.org/10.1016/j.petlm.2020.12.001
16022
Ugochukwu I. Duru et al. Transylvanian Review: Vol 30, No. 2, August 2022

Tadjer, A., Bratvold, R. B., Hong, A., & Hanea, R. vertical well having multiphase flow using
(2021). Application of machine learning to assess functional network technique. Society of
the value of information in polymer flooding. Petroleum Engineers - PAPG/SPE Pakistan
Petroleum Research, 6(4), 309–320. Section Annual Technical Conference and
https://doi.org/10.1016/j.ptlrs.2021.05.006 Exhibition 2018, PATC 2018.
Tariq, Z. (2018). An automated flowing bottom-hole https://doi.org/10.2118/195656-ms
pressure prediction for a vertical well having Teixeira, A. F., & Secchi, A. R. (2019). Machine
multiphase flow using computational intelligence learning models to support reservoir production
techniques. Society of Petroleum Engineers - SPE optimization. IFAC-PapersOnLine, 52(1), 498–
Kingdom of Saudi Arabia Annual Technical 501.
Symposium and Exhibition 2018, SATS 2018. https://doi.org/10.1016/J.IFACOL.2019.06.111
https://doi.org/10.2118/192184-MS Tibshirani, R. (1996). Regression Shrinkage and
Tariq, Z., Mahmoud, M., & Abdulraheem, A. (2020a). Selection Via the Lasso. Journal of the Royal
Real-time prognosis of flowing bottom-hole Statistical Society: Series B (Methodological),
pressure in a vertical well for a multiphase flow 58(1), 267–288. https://doi.org/10.1111/J.2517-
using computational intelligence techniques. 6161.1996.TB02080.X
Journal of Petroleum Exploration and Production Wang, H., Zhang, M., & Yang, Y. (2020). Machine
Technology, 10(4), 1411–1428. learning for multiphase flowrate estimation with
https://doi.org/10.1007/s13202-019-0728-4 time series sensing data. Measurement: Sensors,
Tariq, Z., Mahmoud, M., & Abdulraheem, A. (2020b). 10–12(July), 100025.
Real-time prognosis of flowing bottom-hole https://doi.org/10.1016/j.measen.2020.100025
pressure in a vertical well for a multiphase flow Wu, B., Firouzi, M., Rufford, T. E., & Towler, B. (2019).
using computational intelligence techniques. Characteristics of counter-current gas-liquid two-
Journal of Petroleum Exploration and Production phase flow and its limitations in vertical annuli.
Technology, 10(4), 1411–1428. Experimental Thermal and Fluid Science , 109,
https://doi.org/10.1007/s13202-019-0728-4 109899.
Tariq, Z., Mahmoud, M., Abdulraheem, A., Al-Shehri, https://doi.org/10.1016/J.EXPTHERMFLUSCI.20
D., Khan, M. R., & Janjua, A. N. (2018). An 19.109899
intelligent solution to forecast pressure drop in a

16023