12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Ain Shams Engineering Journal

Volume 12, Issue 1, March 2021, Pages 917-926

Engineering Physics and Mathematics

Real-time prediction of rate of penetration while drilling

complex lithologies using artificial intelligence techniques
Salaheldin Elkatatny

Department of Petroleum Engineering, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi
Arabia

Received 16 January 2020, Revised 9 May 2020, Accepted 29 May 2020, Available online 5 August 2020, Version of
Record 5 March 2021.

Show less

Outline Share Cite

https://doi.org/10.1016/j.asej.2020.05.014 Get rights and content

Under a Creative Commons license Open access

Abstract
Predicting the rate of penetration (ROP) plays a key role in the success of the drilling operation. It
is not an easy task to predict the ROP with high accuracy as it depends on several factors such as;
drilling parameters, drilling fluid properties, and drilled formation characteristics. The objective
of this paper is to develop a new empirical equation for predicting the ROP in real-time using
different artificial intelligence (AI) techniques such as artificial neural networks (ANN), adaptive
neuro-fuzzy inference system (ANFIS), and support vector machine (SVM). For the first time, poly
https://www.sciencedirect.com/science/article/pii/S2090447920301350 1/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

diamond crystalline (PDC) bit design parameters, total flow area, in addition to mud density
(MWin), gamma ray (GR), and drilling parameters were used to build the AI models. Actual field
data was used to build the AI models (1000 data points from Well A) and another 972 data points
from Well B were used for validating the developed AI models.

The obtained results confirmed that the three AI techniques could be used to predict the ROP for
complex lithologies with high accuracy. The ANN outperformed the SVM and ANFIS for
predicting the ROP for the unseen data (972 data points of validation). The developed ROP-ANN
model could be used to predict the ROP with high accuracy (the root mean square error (RMSE)
was less than 0.659 for the available two wells). The developed empirical correlation was able to
predict the ROP with high accuracy, RMSE was 0.66. The new ROP equation can be used without
the need for the ANN Matlab code or special software.

Previous Next

Keywords
Rate of penetration; Complex lithologies; Artificial intelligence; Bit design parameters;
Real-time

Abbreviations
AI, Artificial intelligence; ANFIS, Adaptive neuro-fuzzy inference system; ANN, Artificial
neural networks; C, Regularization parameter; GPM, Pumping rate; GR, Gamma ray; M,
Matrix material type; MWin, Mud density; PDC, Poly diamond crystalline; Q, Pumping
ratexxx; R, Correlation coefficient; RMSE, Root mean square error; ROP, Rate of
penetration; RPM, Revolution per minute; S, Steel material type; SPP, Standpipe
pressure; SVM, Support vector machine; T, Torque; TFA, Total flow area

1. Introduction
Time is one of the critical parameters affecting the cost of drilling operations [1]. Therefore,
optimizing the drilling parameters such as the rate of penetration (ROP) will lead to saving an
extra cost. The drilling program design considers selecting the optimum drilling parameters
such as the drill bit and drilling mud. Also, during the rig drilling operation, real-time analysis
for the drilling parameters should be conducted to optimize the weight on bit (WOB), rotating
speed revolution per minute (RPM), pumping rate (Q), and other parameters [2], [3], since these
parameters have a great impact on the drilling performance that is represented by the ROP.
https://www.sciencedirect.com/science/article/pii/S2090447920301350 2/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Optimizing the drilling operations could be achieved by maximizing the ROP which will reduce
the drilling time and consequently the drilling cost per feet will be reduced [4]. Several models
have been developed to predict the ROP for drilling optimization, the accuracy of such ROP
prediction models is so important [5]. The main point behind that; is how several drilling
parameters control and affect the ROP [6]. Therefore, understanding the drilling data behavior is
considered to be a key factor for generating a good ROP prediction model. Several parameters
affect the ROP and can be categorized into controllable and uncontrollable factors [7]. The
controllable drilling parameters such as the WOB, RPM, Q, torque (T), and standpipe pressure
(SPP), while the uncontrollable drilling parameters such as the bit size, drilling mud type, density,
and rheological properties. The controllable parameters do not affect each other, while the
uncontrollable parameters affect each other, and it is not easy to determine the effect of one
parameter separately [8]. The bit penetrates the formation during the drilling operation by the
action of three forces which are the WOB, the string rotation speed measured by the RPM, mud
pumping rate in gallons per minute (GPM) [9]. The torque (T) is caused by the action of the WOB
and the RPM. The SPP is the pressure generated by the mud pumping and measured at the
standpipe.

1.1. Literature review

ROP prediction models can be classified into two categories, which are traditional models and
data-driven models [10]. The traditional models use the physics-based relationship between the
ROP and the drilling parameters. Several studies had been performed to predict ROP from the
drilling parameters using the mathematical functions. The first mathematical model to estimate
the ROP for rolling cutter bits was developed by Maurer [11], which estimates the ROP based on
the rock compressive strength, WOB, the diameter of the drill bit, and RPM. Bingham [12]
developed another analytical model for ROP estimation which is considered to be a well-known
ROP model. However, many limitations faced these analytical approaches [13]. Both Maurer [11]
and Bingham [12] correlations do not consider the effect of formation compaction, bit hydraulics,
differential pressure, and bit wear on ROP change; this decreases the prediction accuracy of ROP
using Maurer [11] and Bingham [12] correlations. Bourgoyne and Young [14] developed a
correlation for ROP estimation based on multiple regression analysis. In this model, the authors
improved ROP predictability by considering the effect of formation strength, depth, and
compaction, the pressure differential in the bottom hole, bit diameter, WOB, RPM, bit wear, and
bit hydraulic on the ROP.

1.1.1. Predicting the ROP using Artificial intelligence (AI)

The technology revolution along with the deep learning algorithms enabled the engineers to
deeply analyze the collected data and derive meaningful information. The predictive data analysis
was started by statistical techniques to intelligent computing techniques and lastly hybridize
computing techniques. Studies were performed to compare the accuracy of the three approaches
and the results showed that hybridized computing techniques are the best, then rough
https://www.sciencedirect.com/science/article/pii/S2090447920301350 3/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

computing techniques are better than statistical techniques whereas hybridized computing
techniques [15]. Currently, deep learning has been utilized with wide applications and helped to
solve many real-life problems [16], [17].

To overcome the weaknesses of the analytical models, a new approach for ROP estimation was
followed in which the artificial intelligence (AI) techniques were used. Generally, a lot of AI
applications in the petroleum industry have been developed [18], [19] to predict different
parameters such as the drilling bit wear from drilling parameters [20], [21], real-time prediction
of the alteration in the drilling fluid rheology [22], [23], [24], lithology identification and
petrophysical properties [25], [26], [27], prediction of the total organic carbon for evaluation of
unconventional resources [28], [29], [30], [31], estimation of oil recovery factor [32], [33], estimation
of the pore and fracture pressures [34], [35], evaluation of the static Young’s modulus [36], [37], [38],
rock strength failure parameters prediction [39], [40], and prediction of the formation tops [41].

Artificial neural networks (ANN) is an AI technique for data learning and it looks like the way that
the human brain uses its biological neurons for learning features [42], [43]. The ANN structure is
represented by the number of layers and each layer has basic elements called neurons. The
minimum number for the ANN construction is three layers, which are the input, hidden, and
output layer. Transfer functions are fundamentally used to link the ANN layers and appropriate
algorithms are used for data training [44]. The neurons are connected and that connections are
associated with weights and biases [45].

Bilgesu et al. [46] presented the first ANN study for ROP prediction. Many works were performed
to compare the results of the ROP predicted using the AI models versus the predictability of the
traditional analytical models [47], [48], [49], [50], [51], [52], [53]. The results of these studies showed
that AI models have higher accuracy compared with traditional models. Another study by Mantha
and Samuel [54] used AI coupled with statistical regression for ROP estimation, Ahmed et al. [55]
used support vector machine for ROP prediction, Hybrid models were also used by Hegde et al.
[56] to estimate the ROP, and neuro-fuzzy was used by Yavari et al. [57]. Recently, Gan et al. [58]
studied the ROP prediction presented a novel hybrid model by combining the wavelet filtering,
mutual information analysis, and support vector regression. The model was built and validated
using real field drilling data and showed good accuracy for ROP prediction as RMSE was 0.199
and the model outperformed other models from literature. Another study for the ROP prediction
was performed using a neural networks with a radial basis function that was optimized by the
improved particle swarm optimization and showed a good prediction for ROP [59].

The main goal of this paper is to develop new ROP models that can be used for complex
lithologies using different AI techniques such as ANN, ANFIS, and SVM. Actual field data was
used to build and validate the developed models. For the first time, the assessment of the effect of
including the poly diamond crystalline (PDC) bit design features on the ROP drilling
performance was performed and these features were included in the AI models as inputs. This
work studied in deeply the combination of bit design parameters, formation characteristics, mud
https://www.sciencedirect.com/science/article/pii/S2090447920301350 4/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

properties, and drilling mechanical parameters on building ROP models using different AI
techniques. Besides, the study presented a new empirical correlation to estimate the ROP while
drilling with easy use without the need for special software or the Matlab code.

2. Data description
The data used in this study was recorded during drilling operation in the Middle East. The
recorded data covered the 8 3/8″ hole section. The data was collected from two wells. Well A data
(1000 data points) were used for training and testing the AI models and Well B data (972 data
points) was used for validating the developed models. The input parameters are WOB, SPP, RPM,
GPM, MWin, T, GR, bit design (3 codes), and total flow area (TFA), while the predicted parameter
is the ROP. Four formations with different lithology were penetrated through drilling the 8 3/8″
section. All data were collected during the drilling operation, even the GR was recorded using
downhole sensors that installed with the downhole assembly [41].

Table 1 lists the statistical analysis for Well A data. It is clear that WOB rages from 5.07 to 31.24
klbs, SPP ranges from 2801.9 to 3438.9 psi, pipe speed ranges from 48 to 90 rpm, flow rate ranges
from 354.8 to 492.6 gpm, MWin from 90.91 to 110.63 pcf, T ranges from 4.93 to 10.43 klb. ft, GR
ranges from 19.98 to 165.36 API, TFA ranges from 1.23 to 1.47 in2 and the ROP ranges from 2.28
to 25.26 ft/h. Table 2 lists the statistical parameters for Well B (972 data points) and it is clear that
the data range for Well A data which is used for building the AI models covers the data range of
Well B data which is used for validating the developed AI models.

Table 1. Statistical analysis for Well A data (1000 data points).

Minimum Maximum Mean Range Standard Deviation Skewness Kurtosis

WOB, klbs 5.07 31.24 26.17 17.52 5.13 −0.54 0.30

SPP, psi 2801.9 3438.9 637.0 3197.4 159.7 −1.02 −0.54

RPM 48.00 90.00 42.00 61.90 8.34 4.27 1.89

GPM 354.80 492.60 137.80 431.56 32.36 −0.24 0.17

MWin, pcf 90.91 110.63 19.72 97.79 4.27 2.51 1.39

T, klb. ft 4.93 10.43 5.50 7.68 1.49 −0.95 −0.09

GR 19.98 165.36 145.38 79.77 34.06 −1.23 −0.11

Bit Code1 4.00 8.00 4.00 4.88 1.65 −0.17 1.35

Bit Code2 1.00 4.00 3.00 3.01 1.33 −1.29 −0.76

https://www.sciencedirect.com/science/article/pii/S2090447920301350 5/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Minimum Maximum Mean Range Standard Deviation Skewness Kurtosis

Bit Code3 2.00 9.00 7.00 4.17 3.12 −1.19 0.88

TFA, in2 1.23 1.47 0.25 1.38 0.11 −1.48 −0.62

ROP, ft/hr 2.28 25.26 22.99 8.79 5.73 0.03 1.08

Table 2. The statistical analysis for the validation data set (Well B, 972 data points).

Minimum Maximum Mean Range Standard Deviation Skewness Kurtosis

WOB, klbs 6.10 31.20 25.10 18.58 5.66 −1.00 0.30

SPP, psi 2817.6 3434.3 616.7 3173.9 165.7 −1.17 −0.30

RPM 48.00 90.00 42.00 64.22 9.91 1.68 1.54

GPM 355.60 492.60 137.00 424.70 32.79 −0.13 0.17

MWin, pcf 91.47 110.59 19.12 98.85 4.88 0.78 1.24

T, klb. ft 5.02 10.43 5.41 8.21 1.59 −1.17 −0.43

GR 20.31 165.36 145.05 84.09 33.72 −1.03 −0.36

Bit Code1 4.00 8.00 4.00 4.73 1.54 0.75 1.66

Bit Code2 1.00 4.00 3.00 3.37 1.22 0.04 −1.43

Bit Code3 2.00 9.00 7.00 3.47 2.85 0.04 1.43

TFA, in2 1.23 1.47 0.25 1.42 0.09 0.92 −1.70

ROP, ft/h 2.29 25.17 22.88 7.81 5.02 1.12 1.37

The relative importance between the input parameters and the ROP was determined by
calculating the correlation coefficient (R). Fig. 1 shows that the ROP is a strong function of TFA,
bit code 2 and 3, GPM, and MWin where R’s were −0.72, 0.74, −0.72, 0.67 and −0.59, respectively.
While the ROP is a moderate function of RPM, WOB, T, and GR where R values were −0.48, −0.51,
−0.49, and −0.44. The ROP is a weak function of standpipe pressure and bit code 1 where the R’s
were 0.29 and −0.36, respectively.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 6/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Download : Download high-res image (112KB) Download : Download full-size image

Fig. 1. The correlation coefficients between the input parameters and ROP.

For the first time, the drill bit design was considered for the model inputs to study its influence
on the penetration rate. The poly diamond crystalline (PDC) bit design features are coded by a
letter and three-bit numbers that described the bit features in terms of body material, hardness,
cutter size, bite profile [60], [61], [62]. Fig. 2 shows the schematic of the PDC bit. The letter
described the bit body material as it is a steel (S) or a matrix (M) type. The first code refers to the
hardness of the drilled formation and it ranges from 1 to 8, the number increases as the
formation hardness increases. The second-bit code is used to describe the bit cutter size as 2
stands for 19 mm cutter size and 3 for 14 mm cutter size. The third code is used to describe the
bit profile as 3 is a medium profile and 2 for the short profile. These bit features are very
important for bit selection and it is recommended to be selected by the technical designer [63].
The used bits while drilling this section are type M, and therefore, it is not involved in the model
inputs, while the used bits in that section had different codes based on the penetration
performance along with the drilled formations.

Download : Download high-res image (100KB) Download : Download full-size image

https://www.sciencedirect.com/science/article/pii/S2090447920301350 7/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Fig. 2. PDC bit design features.

2.1. Lithology description

The obtained recorded data covered four different formations as listed in Table 3. The drilled
formations are different in their lithological description. Formation-A is a composite lithology
and mainly contained limestone and dolomite and its thickness is 1234 ft. It has hard dolomitic
lithology with cryptocrystalline to microcrystalline size scale. Formation-B has sandstone
lithology with 680 ft thickness which has very fine to fine-grained quartz with silicious cement
and argillaceous matrix. Formation-C is a sandy shale formation and has a thickness of 275 ft.
Formation-D is 2433 ft thickness and it is a tight shale formation which is non-calcareous, silty
dark, and grey-coloured shale.

Table 3. Formations lithology description.

3. Methodology
Three different AI techniques were used to predict the ROP by using the eleven input parameters
which are WOB, SPP, RPM, GPM, MWin, T, GR, bit design (3 codes), and TFA. ANN, ANFIS, and
SVM are trained using 75% of Well A’s data while 25% of Well A’s data was used for testing. The
https://www.sciencedirect.com/science/article/pii/S2090447920301350 8/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

training data set was selected by random iteration and then checked to cover and represent the
different four formations penetrated during drilling. The three models were built by conducting
sensitivity analysis for each model’s design parameters and the model performance was evaluated
by calculating the root mean square error (RMSE).

Three AI tools (ANN, ANFIS, and SVM) were used to predict the ROP. The three methods were
widely used in the application of AI in the petroleum industry. ANN as a tool has main
controlling parameters like the number of layers, the number of neurons in the hidden layers,
training function, and transfer function. Optimizing these parameters will help to increase the
model accuracy. For ANFIS tool, cluster radius and number of iterations are the critical
parameters for optimizing the model performance. SVM was used as the third tool for prediction
models. The hyper-parameters for SVM as regularization parameter (C), lambda, epsilon, Kernel
option, and Kernel function were optimized through many iterations.

ANN model was built by the sensitivity analysis for the neural networks architecture parameters
such as the number of hidden layers, the number of neurons, network function, training
function, and transfer function. The analysis was conducted to optimize the model for getting the
highest accuracy for the training and testing data set in terms of RMSE. The number of hidden
layers was changed from 1 to 2 and the number of neurons was changed from 10 to 30 neurons.
Different training and transforming functions were used to evaluate the model.

ANFIS model was built based on the subtractive clustering technique. The cluster radius and
number of iterations (Ep size) are the main parameters to optimize the ANFIS model. Changing
the cluster radius from 0.9 to 0.25 was executed using different iteration numbers to check the
model prediction performance.

The SVM model was built by performing the sensitivity analysis for the model design parameters
such as regularization parameter (C), lambda, epsilon, kernel option, and kernel function.
Different values for the kernel option were used from 1 to 10, and many kernel functions were
checked to optimize the SVM prediction model and enhance the model accuracy.

4. Results

4.1. ROP-ANN model

The sensitivity results showed that after running the code with different scenarios, the best case
was optimized by using 1 hidden layer, 20 neurons, fitnet as a network function, trainbr as a
training function, tansig as a transfer function for the model architecture. Table 4 lists the
different scenarios that were applied to optimize the ROP-ANN model. The highest accuracy of
the selected variables showed that the ANN was able to predict the ROP with RMSE’s of 0.363 and
0.552 for training and testing data, respectively. Fig. 3 shows the high accuracy match between the

https://www.sciencedirect.com/science/article/pii/S2090447920301350 9/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

actual and predicted ROP for the training data set (750 data points) and testing data set (250 data
points).

Table 4. ANN model sensitivity analysis.

No of Hidden No of Network Training Transfer RMSE RMSE

Layers Neurons Function Function Function Training Testing

1 10 fitnet trainbr tansig 0.816 1.045

1 15 fitnet trainbr tansig 0.562 0.807

1 25 fitnet trainbr tansig 0.454 0.779

1 30 fitnet trainbr tansig 0.385 0.804

2 20 fitnet trainbr tansig 0.301 0.764

1 20 newcf trainbr tansig 0.409 1.265

1 20 newdtdnn trainbr tansig 0.454 0.779

1 20 newelm trainbr tansig 0.400 0.929

1 20 newfit trainbr tansig 0.489 0.890

1 20 fitnet trainb tansig 0.438 0.881

1 20 fitnet trainr tansig 0.454 0.779

1 20 fitnet trainscg tansig 6.776 6.938

1 20 fitnet trainrp tansig 3.578 3.590

1 20 fitnet trainlm tansig 1.738 1.768

1 20 fitnet trainbfg tansig 1.543 1.749

1 20 fitnet trainoss tansig 0.843 1.214

1 20 fitnet trainbr logsig 0.989 1.168

1 20 fitnet trainbr hardlims 1.906 1.882

1 20 fitnet trainbr purelin 0.454 0.779

1 20 fitnet trainbr hardlim 7.363 7.531

1 20 fitnet trainbr tribas 2.757 2.642

1 20 fitnet trainbr softmax 2.557 2.814

https://www.sciencedirect.com/science/article/pii/S2090447920301350 10/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

No of Hidden No of Network Training Transfer RMSE RMSE

Layers Neurons Function Function Function Training Testing

1 20 fitnet trainbr tansig 0.363 0.552

Download : Download high-res image (375KB) Download : Download full-size image

Fig. 3. ROP prediction using ANN technique for A) training data (750 data points) and B) testing
data (250 data points).

4.2. ROP-ANFIS model

Table 5 represents the different scenarios for the ANFIS technique. The ANFIS model resulted in
the highest accuracy when the cluster radius was 0.25 and Ep-Size of 300. Fig. 4 shows the high
accurate match between the actual and predicted ROP for the training and testing data using the
ANFIS technique. The RMSE’s were 0.425 and 0.446 for training and testing data, respectively.

Table 5. ANFIS model sensitivity analysis.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 11/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Cluster Radius “Ep_Size RMSE Train RMSE Test

0.9 200 1.682 1.836

0.8 200 1.725 1.825

0.6 200 1.168 1.352

0.4 200 0.910 1.188

0.3 200 0.642 0.885

0.25 100 0.515 0.877

0.25 200 0.515 0.877

0.25 300 0.425 0.446

Download : Download high-res image (344KB) Download : Download full-size image

Fig. 4. ROP prediction using ANFIS technique for A) training data (750 data points) and B) testing
data (250 data points).

https://www.sciencedirect.com/science/article/pii/S2090447920301350 12/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

4.3. ROP-SVM model

The sensitivity analysis for the SVM technique was performed using C range from 100 to 3000,
lambda [10−1 to 10−5], epsilon range from 10−5 to 1, and many Kernel functions. The results
showed that the high accuracy model was built using Kernel option of 10, C of 1000, lambda and
epsilon of 10−5, and gaussianslow as for Kernel as listed in Table 6. Fig. 5 shows that the RMSE was
very small for training (0.006) and high for testing data points (0.718) when applying the
optimized SVM to estimate the ROP.

Table 6. SVM model sensitivity analysis.

Kernel option C Lambda epsilon kernel RMSE Training RMSE Testing

1 1000 10−5 10−5 gaussian 0.000 4.974

5 1000 10−5 10−5 gaussian 0.000 2.311

10 1000 10−5 10−5 gaussian 0.036 1.540

10 100 10−5 10−5 gaussian 0.273 1.517

10 3000 10−5 10−5 gaussian 0.025 1.615

10 2000 10−5 10−5 gaussian 0.026 1.558

10 1000 10−5 10−5 gaussian 0.036 1.540

10 1000 10−5 10−4 gaussianslow 0.035 1.144

10 1000 10−5 0.1 gaussianslow 0.100 1.134

10 1000 10−5 1.0 gaussianslow 0.845 1.444

10 1000 10−1 1.0 gaussianslow 1.079 1.783

10 1000 10−3 1.0 gaussianslow 1.462 1.523

10 1000 10−5 10−5 gaussianslow 0.006 0.718

https://www.sciencedirect.com/science/article/pii/S2090447920301350 13/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Download : Download high-res image (382KB) Download : Download full-size image

Fig. 5. ROP prediction using SVM technique for A) training data (750 data points) and B) testing
data (250 data points).

5. Discussion
The three methods used in this study provided three different AI-ROP prediction models. The
presented models showed a high prediction performance for training and testing data sets by
optimizing the models’ main design parameters. Fig. 6 summarizes the comparison between the
three techniques in terms of RMSE. The highest RMSE is only 0.718 confirming the high
accuracy of the three models in predicting the ROP in real-time for Well A data. The elapsed time
was different for the three AI models. SVM took the least running time (4.44 s) followed by ANN
which took 10.08 s and ANFIS took the longest time (557.69 s).

https://www.sciencedirect.com/science/article/pii/S2090447920301350 14/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Download : Download high-res image (99KB) Download : Download full-size image

Fig. 6. Comparison of ROP prediction using different AI techniques in terms of Root mean
square error.

For industrial application, the three models were validated to assess the model accuracy with
unseen data (Well B) and the next section presented the validation in detail.

5.1. AI models validation

For further comparison between these techniques and to evaluate the possibility of applying
these techniques in unseen data, Well B data (972 data points) was used.

Fig. 7 shows that the ANN was able to predict the ROP with RMSE of 0.66 for the unseen data
while the ANFIS model predicted the ROP with RMSE of 0.76. The SVM technique predicted the
ROP with RMSE of 1.12. It is clear from these results that the ANN is the best AI technique that
could be used for predicting the ROP with lower RMSE as compared with ANFIS and SVM.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 15/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Download : Download high-res image (426KB) Download : Download full-size image

Fig. 7. ROP estimation for the unseen data (972 data points) using A) ANN technique, B) ANFIS
technique, and C) SVM technique.

5.2. New empirical equation for ROP

Based on the weights and biases of the optimized ANN technique, a new ROP empirical equation
(Eq. (1)) was developed that could be applied without the need for the Matlab code. It is a simple
equation that could be installed in any computer and be used in real-time on a rig site to estimate
the ROP. Table 7 lists the weights and biases of the optimized ANN model, these weights and
biases could be applied into Eq. (1) with the input parameters to calculate the ROP.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 16/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

(1)

where
N: No of neurons in the hidden layer

W1: Weight between the inputs and the hidden layer neurons

W2: Weight between the hidden layer neurons and the output

b1: Biases of the hidden layer

b2: Biases of the output layer

Table 7. Weight and biases for the hidden layer of the ANN-ROP model.

# Weight between inputs and hidden layer (W1) biases (w2) bia
of (b1) (b2)
N

WOB SPP RPM GPM MWin T GR Bit Bit Bit TFA,

klbs psi Pcf Code1 Code2 Code3 in^2

1 −0.626 1.930 −1.256 −1.173 −0.258 −0.188 0.576 1.575 −0.424 −0.055 −0.388 0.202 1.305 0.52

2 −1.905 0.167 1.419 −2.432 0.202 −1.130 −0.866 −0.714 0.554 −0.694 −0.459 0.002 −1.156

3 −1.429 1.017 −2.030 −1.917 1.432 2.687 −1.618 0.702 −0.096 −0.193 −0.271 −0.942 0.669

4 1.126 0.575 −0.412 −0.674 −0.742 −2.639 −2.654 −0.923 1.291 −1.234 1.913 −0.812 1.311

5 −0.881 0.234 1.458 −2.056 −0.046 −0.889 −0.997 −0.831 −0.356 −1.285 −0.695 0.296 2.934

6 −0.496 −1.074 1.460 −1.709 1.163 −0.986 −1.794 −1.010 −0.550 −0.477 0.960 0.226 −1.497

7 1.747 −0.796 1.745 −0.172 0.070 0.927 −1.393 0.963 −0.182 0.652 −0.929 0.578 0.662

8 −0.288 −0.468 1.524 −1.366 1.002 0.097 −0.919 0.354 0.054 0.310 −1.818 2.444 −2.672

9 −1.001 −0.284 0.054 0.158 0.147 −1.398 0.639 0.736 −0.525 0.586 1.267 −0.789 −2.604

10 0.040 0.911 −0.498 0.768 0.138 1.358 −1.305 0.176 0.805 0.081 −1.517 −0.853 −1.960

https://www.sciencedirect.com/science/article/pii/S2090447920301350 17/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

# Weight between inputs and hidden layer (W1) biases (w2) bia
of (b1) (b2)
N

WOB SPP RPM GPM MWin T GR Bit Bit Bit TFA,

klbs psi Pcf Code1 Code2 Code3 in^2

11 −0.565 0.170 2.100 −1.297 −0.913 1.818 −0.355 1.184 0.672 −0.482 −2.198 0.924 2.762

12 0.573 1.845 −0.599 −0.122 1.500 −1.004 −1.270 −0.857 2.159 −2.045 −0.085 0.362 −3.539

13 −0.291 0.501 0.763 −0.137 −2.966 −2.044 0.665 0.568 0.983 −0.459 −0.072 −1.173 −1.912

14 0.129 −0.040 0.820 −3.052 1.463 −0.800 −1.756 −0.562 0.156 −0.044 0.405 2.297 −1.995

15 −1.497 −1.204 −0.001 0.030 3.171 −0.531 −1.912 −0.829 −0.359 1.141 0.483 −0.310 −1.343

16 −0.870 −0.583 −0.872 1.359 1.725 −0.312 −1.435 −0.545 0.821 −0.829 0.423 −0.711 1.224

17 −0.978 −0.080 −1.312 1.737 2.715 0.204 −2.146 −0.391 0.411 0.785 0.218 −1.047 1.996

18 1.380 1.139 −2.099 0.880 0.429 2.295 −0.826 −0.599 0.259 −0.068 −0.237 −0.428 −1.465

19 0.918 −1.020 1.676 0.035 0.588 −0.660 −0.027 1.045 0.508 −1.193 −0.337 1.329 1.331

20 0.776 −1.013 −2.390 1.534 2.457 −3.571 0.405 −1.206 0.122 −0.397 0.444 −0.461 1.658

Applying the developed equation (Eq. (1)) on the unseen data (Well B, 972 data points), the RMSE
was 0.66 confirming the high accuracy of the developed equation, Fig. 8.

Download : Download high-res image (120KB) Download : Download full-size image

Fig. 8. ROP prediction using Eq. (1) for the unseen data points (972).

https://www.sciencedirect.com/science/article/pii/S2090447920301350 18/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

6. Conclusions
Different AI techniques were built to predict the ROP for complex lithologies. For the first time,
the bit design and TFA were included as input parameters for the AI models in addition to the
real-time data of the drilling parameters, mud density, and gamma ray. The following
conclusions could be drawn based on the obtained results:
1. ANN, ANFIS, and SVM could be used to predict the ROP for complex lithologies with high
accuracy

2. ANN outperformed the SVM and ANFIS for predicting the ROP for the unseen data (972
points of validation).

3. Statistical analysis showed that the best combination of ANN variable parameters was 1
hidden layer, 20 neurons, fitnet as a network function, trainbr as a training function, tansig as a
transfer function, which could be used to predict the ROP with high accuracy.

4. The ROP-ANN model predicted the ROP with RMSE of less than 0.66 for the available two
wells.

5. The developed ROP empirical correlation was able to predict the ROP with RMSE 0f 0.66. The
developed equation could be used without the need for the ANN Matlab code.

Recommended articles

References
[1] W.C. Lyons, G.J. Plisga
Standard handbook of petroleum and natural gas engineering
(2nd ed.), Gulf Professional Publishing (2004)
Google Scholar

[2] Eren T, Ozbayoglu ME. Real time optimization of drilling parameters during drilling
operations. Paper presented in the SPE Oil and Gas India Conference and Exhibition.,
Mumbai, India, 20-22 January. SPE-129126-MS. https://doi.org/10.2118/129126-MS; 2019.
Google Scholar

[3] Payette GS, Spivey BJ, Wang L, Bailey JR, Sanderson D, Kong R, et al. Real-time well-site
based surveillance and optimization platform for drilling: technology, basic workflows and
field results. Paper presented at the SPE/IADC Drilling Conference and Exhibition, The
Hague, The Netherlands, 14-16 March. SPE-184615-MS; 2017.
https://doi.org/10.2118/184615-MS.
Google Scholar

https://www.sciencedirect.com/science/article/pii/S2090447920301350 19/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

[4] L.F.F. Barbosa, A. Nascimento, M.H. Mathias, J.A. de Carvalho Jr

Machine learning methods applied to drilling rate of penetration prediction and
optimization-A review
J Petrol Sci Eng, 183 (2019), Article 106332, 10.1016/j.petrol.2019.106332
Article Download PDF View Record in Scopus Google Scholar

[5] C. Soares, K. Gray

Real-time predictive capabilities of analytical and machine learning rate of penetration
(ROP) models
J. Pet. Sci. Eng., 172 (2019), pp. 934-959, 10.1016/j.petrol.2018.08.083
Article Download PDF View Record in Scopus Google Scholar

[6] R.F. Mitchell, S.Z. Miska

Fundamentals of drilling engineering
Society of Petroleum Engineers, Richardson, TX (2011)
Google Scholar

[7] M.E. Hossain, A.A. Al-Majed

Fundamentals of sustainable drilling engineering
Scrivener Publishing, Austin (2015)
Google Scholar

[8] Osgouei RE, Özbayoğlu ME. Rate of penetration estimation model for directional and
horizontal wells. In: 16th international petroleum and natural gas congress and exhibition
of Turkey. http://old.earthdoc.org/publication/publicationdetails?publication=76983; 2007.
Google Scholar

[9] Al-AbdulJabbar A, Elkatatny S, Mahmoud M, Abdulraheem A. Predicting Rate of

Penetration Using Artificial Intelligence Techniques. Paper presented at the SPE Kingdom
of Saudi Arabia Annual Technical Symposium and Exhibition, Dammam, Saudi Arabia, 23-
26 April. SPE-192343-MS. https://doi.org/10.2118/192343-MS; 2018.
Google Scholar

[10] C. Hegde, H. Daigle, H. Millwater, K. Gray

Analysis of rate of penetration (ROP) prediction in drilling using physics-based and data-
driven models
J Petrol Sci Eng, 159 (2017), pp. 295-306, 10.1016/j.petrol.2017.09.020
Article Download PDF View Record in Scopus Google Scholar

[11] W. Maurer
The “perfect-cleaning” theory of rotary drilling
J. Pet. Technol., 14 (1962), pp. 1270-1274, 10.2118/408-PA
View Record in Scopus Google Scholar
https://www.sciencedirect.com/science/article/pii/S2090447920301350 20/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

[12] M.G. Bingham

A new approach to interpreting rock drillability
Petroleum Publishing Company (1965)
Google Scholar

[13] C. Soares, H. Daigle, K. Gray

Evaluation of PDC bit ROP models and the effect of rock strength on model coefficients
J. Nat. Gas Sci. Eng., 34 (2016), pp. 1225-1236, 10.1016/j.jngse.2016.08.012
Article Download PDF View Record in Scopus Google Scholar

[14] A. Bourgoyne, F.S. Young

A multiple regression approach to optimal drilling and abnormal pressure detection
Soc. Pet. Eng. J., 14 (1974), pp. 371-384, 10.2118/4238-PA
Google Scholar

[15] D. Acharjya, A. Anitha

A comparative study of statistical and rough computing models in predictive data analysis
Int J Ambient Comput Intell (IJACI), 8 (2) (2017), pp. 32-51, 10.4018/IJACI.2017040103
View Record in Scopus Google Scholar

[16] S. Hu, M. Liu, S. Fong, W. Song, N. Dey, R. Wong

Forecasting China future MNP by deep learning
Behavior engineering and applications, Springer, Cham (2018), pp. 169-210
CrossRef View Record in Scopus Google Scholar

[17] S. Chatterjee, S. Sarkar, S. Hore, N. Dey, A.S. Ashour, V.E. Balas

Particle swarm optimization trained neural network for structural failure prediction of
multistoried RC buildings
Neural Comput Appl, 28 (8) (2017), pp. 2005-2016, 10.1007/s00521-016-2190-2
View Record in Scopus Google Scholar

[18] O.E. Agwu, J.U. Akpabio, S.B. Alabi, et al.

Artificial intelligence techniques and their applications in drilling fluid engineering: a
review
J Petrol Sci Eng, 167 (2018), pp. 300-315, 10.1016/j.petrol.2018.04.019
Article Download PDF View Record in Scopus Google Scholar

[19] K. Nagy, E. Hajrizi

Beyond the Age of Oil and Gas-How artificial intelligence is transforming the energy
portfolio of the societies
IFAC-PapersOnLine, 51 (30) (2018), pp. 308-310, 10.1016/j.ifacol.2018.11.307
Article Download PDF View Record in Scopus Google Scholar

https://www.sciencedirect.com/science/article/pii/S2090447920301350 21/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

[20] R.A. Arehart

Drill-bit diagnosis with neural networks
SPE Comput Appl, 2 (4) (1990), pp. 24-28, 10.2118/19558-PA
View Record in Scopus Google Scholar

[21] M.A. Kamel, S. Elkatatny, M.F. Mysorewala, A. Al-Majed, M. Elshafei

Adaptive and real-time optimal control of stick–slip and bit wear in autonomous rotary
steerable drilling
J Energy Res Technol, 140 (3) (2018), 10.1115/1.4038131
Google Scholar

[22] K. Abdelgawad, S. Elkatatny, T. Moussa, M. Mahmoud, S. Patil

Real time determination of rheological properties of spud drilling fluids using a hybrid
artificial intelligence technique
J Energy Resour Technol (2018), 10.1115/1.4042233
Google Scholar

[23] S.M. Elkatatny

Real time prediction of rheological parameters of KCl water-based drilling fluid using
artificial neural networks
Arabian J Sci Eng, 42 (4) (2017), pp. 1655-1665, 10.1007/s13369-016-2409-7
View Record in Scopus Google Scholar

[24] A. Alsabaa, H. Gamal, S. Elkatatny, A. Abdulraheem

Real-time prediction of rheological properties of invert emulsion mud using adaptive
neuro-fuzzy inference system
Sensors, 20 (6) (2020), p. 1669, 10.3390/s20061669
View Record in Scopus Google Scholar

[25] X. Ren, J. Hou, S. Song, Y. Liu, D. Chen, X. Wang, et al.

Lithology identification using well logs: a method by integrating artificial neural networks
and sedimentary patterns
J Petrol Sci Eng, 182 (2019) (2019), Article 106336, 10.1016/j.petrol.2019.106336
Article Download PDF View Record in Scopus Google Scholar

[26] S. Elkatatny, Z. Tariq, M. Mahmoud, I. Mohamed, A. Abdulraheem

Development of new mathematical model for compressional and shear sonic times from
wireline log data using artificial intelligence neural networks (white box)
Arabian J Sci Eng, 43 (11) (2018), pp. 6375-6389, 10.1007/s13369-018-3094-5
View Record in Scopus Google Scholar

[27] T. Moussa, S. Elkatatny, M. Mahmoud, A. Abdulraheem

https://www.sciencedirect.com/science/article/pii/S2090447920301350 22/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Development of new permeability formulation from well log data using artificial
intelligence approaches
J Energy Res Technol, 140 (7) (2018), 10.1115/1.4039270
Google Scholar

[28] A.A. Mahmoud, S. Elkatatny, A. Ali, M. Abouelresh, A. Abdulraheem

Evaluation of the total organic carbon (TOC) using different artificial intelligence
techniques
Sustainability, 11 (20) (2019), p. 5643, 10.3390/su11205643
View Record in Scopus Google Scholar

[29] Mahmoud AA, Elkatatny S, Abdulraheem A, Mahmoud M, Ibrahim O, Ali A. New technique
to determine the total organic carbon based on well logs using artificial neural network
(white box). In: Proceedings of the 2017 SPE Kingdom of Saudi Arabia Annual Technical
Symposium and Exhibition, Dammam, Saudi Arabia, 24–27 April. SPE-188016-MS.
https://doi.org/10.2118/188016-MS; 2017.
Google Scholar

[30] Mahmoud AA, Elkatatny S, Ali A, Abouelresh M, Abdulraheem A. New robust model to
evaluate the total organic carbon using fuzzy logic. In: Proceedings of the SPE Kuwait Oil
& Gas Show and Conference, 13–16 October, Mishref, Kuwait. SPE-198130-MS.
https://doi.org/10.2118/198130-MS; 2019.
Google Scholar

[31] A.A. Mahmoud, S. Elkatatny, M. Mahmoud, M. Abouelresh, A. Abdulraheem, A. Ali

Determination of the total organic carbon (TOC) based on conventional well logs using
artificial neural network
Int J Coal Geol, 179 (2017), pp. 72-80, 10.1016/j.coal.2017.05.012
Article Download PDF View Record in Scopus Google Scholar

[32] A.A. Mahmoud, S. Elkatatny, W. Chen, A. Abdulraheem

Estimation of oil recovery factor for water drive sandy reservoirs through applications of
artificial intelligence
Energies, 12 (19) (2019), p. 3671, 10.3390/en12193671
View Record in Scopus Google Scholar

[33] Mahmoud AA, Elkatatny S, Abdulraheem A, Mahmoud M. Application of artificial

intelligence techniques in estimating oil recovery factor for water drive sandy reservoirs.
In: Proceedings of the 2017 SPE Kuwait Oil & Gas Show and Conference, Kuwait City,
Kuwait, 15–18 October. SPE-187621-MS. https://doi.org/10.2118/187621-MS; 2017.
Google Scholar

[34]
https://www.sciencedirect.com/science/article/pii/S2090447920301350 23/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Ahmed AS, Mahmoud AA, Elkatatny S. Fracture pressure prediction using radial basis
function. In: Proceedings of the AADE National Technical Conference and Exhibition,
Denver, Colorado, USA, April 9–10, 2019, AADE-19-NTCE-061.
Google Scholar

[35] Ahmed AS, Mahmoud AA, Elkatatny S, Mahmoud M, Abdulraheem A. Prediction of pore
and fracture pressures using support vector machine. In: Proceedings of the 2019
International Petroleum Technology Conference, Beijing, China, 26–28 March. IPTC-
19523-MS. https://doi.org/10.2523/IPTC-19523-MS; 2019.
Google Scholar

[36] A.A. Mahmoud, S. Elkatatny, A. Ali, T. Moussa

Estimation of static young’s modulus for sandstone formation using artificial neural
networks
Energies, 12 (11) (2019), p. 2125, 10.3390/en12112125
Google Scholar

[37] S. Elkatatny, Z. Tariq, M. Mahmoud, A. Abdulraheem, I. Mohamed

An integrated approach for estimating static Young’s modulus using artificial intelligence
tools
Neural Comput Appl, 31 (8) (2019), pp. 4123-4135, 10.1007/s00521-018-3344-1
View Record in Scopus Google Scholar

[38] S. Elkatatny, M. Mahmoud, I. Mohamed, A. Abdulraheem

Development of a new correlation to determine the static Young’s modulus
J Pet Explor Prod Technol, 8 (1) (2018), pp. 17-30, 10.1007/s13202-017-0316-4
View Record in Scopus Google Scholar

[39] Z. Tariq, S. Elkatatny, M. Mahmoud, A.Z. Ali, A. Abdulraheem

A new technique to develop rock strength correlation using artificial intelligence tools
SPE reservoir characterization and simulation conference and exhibition, Society of
Petroleum Engineers (2017), 10.2118/186062-MS
Google Scholar

[40] Z. Tariq, S. Elkatatny, M. Mahmoud, A.Z. Ali, A. Abdulraheem

A new approach to predict failure parameters of carbonate rocks using artificial
intelligence tools
SPE Kingdom of Saudi Arabia annual technical symposium and exhibition, Society of
Petroleum Engineers (2017), 10.2118/187974-MS
Google Scholar

[41] S. Elkatatny, A. Al-AbdulJabbar, A.A. Mahmoud

https://www.sciencedirect.com/science/article/pii/S2090447920301350 24/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

New robust model to estimate the formation tops in real-time using artificial neural
networks (ANN)
Petrophysics, 60 (06) (2019), pp. 825-837, 10.30632/PJV60N6-2019a7
View Record in Scopus Google Scholar

[42] E. Angelini, A. Ludovici

CDS evaluation model with neural networks
J Serv Sci Manage, 2 (1) (2009), p. 15, 10.4236/jssm.2009.21003
View Record in Scopus Google Scholar

[43] M. Monjezi, H. Dehghani

Evaluation of effect of blasting pattern parameters on back break using neural networks
Int J Rock Mech Min Sci, 45 (8) (2008), pp. 1446-1453, 10.1016/j.ijrmms.2008.02.007
Article Download PDF View Record in Scopus Google Scholar

[44] R.P. Lippmann

An introduction to computing with neural nets
IEEE Assp Mag, 4 (2) (1987), pp. 4-22, 10.1109/MASSP.1987.1165576
Google Scholar

[45] G.E. Hinton, S. Osindero, Y.W. Teh

A fast learning algorithm for deep belief nets
Neural Comput, 18 (7) (2006), pp. 1527-1554, 10.1162/neco.2006.18.7.1527
Google Scholar

[46] Bilgesu H, Tetrick L, Altmis U, Mohaghegh S, Ameri S. A new approach for the prediction
of rate of penetration (ROP) values. Paper presented at the SPE Eastern Regional Meeting,
Lexington, Kentucky, 22–24 October. SPE-39231-MS. https://doi.org/10.2118/39231-MS;
1997.
Google Scholar

[47] A. Al-AbdulJabbar, S. Elkatatny, M. Mahmoud, K. Abdelgawad, A. Al-Majed

A robust rate of penetration model for carbonate formation
ASME. J. Energy Resour. Technol., 141 (4) (2018), Article 042903, 10.1115/1.4041840
Google Scholar

[48] K. Amar, A. Ibrahim

Rate of penetration prediction and optimization using advances in artificial neural
networks, a comparative study
Proceedings of the 4th international joint conference on computational intelligence
(2012), pp. 647-652, 10.5220/0004172506470652
View Record in Scopus Google Scholar

https://www.sciencedirect.com/science/article/pii/S2090447920301350 25/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

[49] R. Arabjamaloei, S. Shadizadeh

Modeling and optimizing rate of penetration using intelligent systems in an Iranian
southern oil field (Ahwaz oil field)
Pet. Sci. Technol., 29 (2011), pp. 1637-1648, 10.1080/10916460902882818
View Record in Scopus Google Scholar

[50] Bataee M, Mohseni S. Application of artificial intelligent systems in ROP optimization: a

case study in Shadegan oil field. Paper presented in the SPE Middle East Unconventional
Gas Conference and Exhibition, Muscat, Oman. Society of Petroleum Engineers. 31
January-2 February. SPE-140029-MS. https://doi.org/10.2118/140029-MS; 2011.
Google Scholar

[51] Salaheldin Elkatatny

New approach to optimize the rate of penetration using artificial neural network
Arabian J Sci Eng, 43 (11) (2018), pp. 6297-6304, 10.1007/s13369-017-3022-0
View Record in Scopus Google Scholar

[52] C. Hegde, H. Daigle, K.E. Gray

Performance comparison of algorithms for realtime rate-of-penetration optimization in
drilling using data-driven models
SPE J., 23 (2018), pp. 1706-1722, 10.2118/191141-PA
View Record in Scopus Google Scholar

[53] Elkatatny SM, Tariq Z, Mahmoud MA, Al-AbdulJabbar A. Optimization of rate of

penetration using artificial intelligent techniques. Paper presented at the Rock
Mechanics/Geomechanics Symposium, San Francisco, California, USA, 25-28 June. ARMA-
2017-0429; 2017.
Google Scholar

[54] Mantha B, Samuel R. ROP optimization using artificial intelligence techniques with
statistical regression coupling. Paper presented at the SPE Annual Technical Conference
and Exhibition, Dubai, UAE, 26-28 September. SPE-181382-MS.
https://doi.org/10.2118/181382-MS; 2016.
Google Scholar

[55] Ahmed AS, Elkatatny Salaheldin, Abdulraheem Abdulazeez, Mahmoud Mohammed, Ali
Abdulwahab Z, Mohamed IM. Prediction of rate of penetration of deep and tight
formation using support vector machine. Paper presented at the SPE Kingdom of Saudi
Arabia Annual Technical Symposium and Exhibition, Dammam, Saudi Arabia, 23-26 April.
SPE-192316-MS. https://doi.org/10.2118/192316-MS; 2018.
Google Scholar

[56]
https://www.sciencedirect.com/science/article/pii/S2090447920301350 26/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Hegde Chiranth, Soares Cesar, Gray Ken. Rate of penetration (ROP) modeling using hybrid
models: deterministic and machine learning. Paper presented at the SPE/AAPG/SEG
Unconventional Resources Technology Conference, Houston, Texas, USA, 23–25 July.
URTEC-2896522-MS https://doi.org/10.15530/URTEC-2018-2896522; 2018.
Google Scholar

[57] Hossein Yavari, Mohammad Sabah, Rassoul Khosravanian, David A. Wood

Application of an adaptive neuro-fuzzy inference system and mathematical rate of
penetration models to predicting drilling rate
Iran J Oil Gas Sci Technol, 7 (3) (2018), pp. 73-100, 10.22050/ijogst.2018.83374.1391
View Record in Scopus Google Scholar

[58] C. Gan, W.H. Cao, M. Wu, X. Chen, Y.L. Hu, K.Z. Liu, et al.
Prediction of drilling rate of penetration (ROP) using hybrid support vector regression: a
case study on the Shennongjia area, Central China
J Petrol Sci Eng, 181 (2019), Article 106200, 10.1016/j.petrol.2019.106200
Article Download PDF View Record in Scopus Google Scholar

[59] C. Gan, W. Cao, M. Wu, K.Z. Liu, X. Chen, Y. Hu, et al.

Two-level intelligent modeling method for the rate of penetration in complex geological
drilling process
Appl Soft Comput, 80 (2019), pp. 592-602, 10.1016/j.asoc.2019.04.020
Article Download PDF View Record in Scopus Google Scholar

[60] Winters WJ, Senger J, Oliver MS. Application of the 1987 IADC roller bit classification
system. Paper presented at the SPE/IADC Drilling Conference, New Orleans, Louisiana,
15–18 March. SPE-16143-MS. https://doi.org/10.2118/16143-MS; 1987.
Google Scholar

[61] R. Feenstra
Status of polycrystalline-diamond-compact bits: part I development (includes associated
papers 19032 and 19900). SPE-17919-PA
J Petrol Technol, 40 (06) (1988), 10.2118/17919-PA
Google Scholar

[62] R. Feenstra
Status of polycrystalline-diamond-compact bits: part 2-applications. SPE-17920-PA
J Petrol Technol, 40 (07) (1988), 10.2118/17920-PA
Google Scholar

[63] Silva N, Boulton AJ. Setting a new benchmark in ROP and bit durability in the pre-salt
section carbonate rocks of the deepwater mero field in Brazil. Paper presented at the

https://www.sciencedirect.com/science/article/pii/S2090447920301350 27/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Offshore Technology Conference Brasil, Rio de Janeiro, Brazil, 29–31 October. OTC-29843-
MS. https://doi.org/10.4043/29843-MS; 2019.
Google Scholar

Cited by (11)

Prediction of rate of penetration in directional drilling using data mining techniques

2023, Journal of Petroleum Science and Engineering

Show abstract

Real-time porosity prediction using gas-while-drilling data and machine learning with
reservoir associated gas: Case study for Hassi Messaoud field, Algeria
2022, Marine and Petroleum Geology

Show abstract

Determining the difference of kick tolerance with single bubble and dynamic
multiphase models: Evaluation of well-control with water/synthetic based muds
2022, Ain Shams Engineering Journal

Show abstract

In situ strength profiles along two adjacent vertical drillholes from digitalization of
hydraulic rotary drilling
2022, Journal of Rock Mechanics and Geotechnical Engineering

Show abstract

A formation-based approach for modeling of rate of penetration for an offshore gas

field using artificial neural networks
2021, Journal of Natural Gas Science and Engineering

Show abstract

Reservoir production optimization based on surrograte model and differential

evolution algorithm
2021, Journal of Petroleum Science and Engineering

Show abstract
https://www.sciencedirect.com/science/article/pii/S2090447920301350 28/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

View all citing articles on Scopus

Dr. Elkatatny is an associate professor at King Fahd University of Petroleum & Minerals. He is an associate professor
at Cairo University, on leave. He worked as a senior geomechanical engineer at Advantek International Company for 5
years. He received his Ph.D. in Petroleum Engineering at Texas A&M University, College Station, Texas. He received his
B.Sc. and M.Sc. degrees from the Petroleum Engineering at Cairo University, Egypt. He worked with SUCO Oil
Company as a drilling engineer for one year. His areas of research include drilling fluid optimization, filter cake
removal, formation damage, oil-well cementing, geomechanics, and artificial intelligence.

Peer review under responsibility of Ain Shams University.

© 2020 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Ain Shams University.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 29/30
12/21/22, 10:58 AM Real-time prediction of rate of penetration while drilling complex lithologies using artificial intelligence techniques - ScienceDirect

Copyright © 2022 Elsevier B.V. or its licensors or contributors.

ScienceDirect® is a registered trademark of Elsevier B.V.

https://www.sciencedirect.com/science/article/pii/S2090447920301350 30/30