2019 Fourth International Conference on Advances in Computational Tools for Engineering Applications (ACTEA)

Accurate Prediction of Gas Compressibility Factor

using Kernel Ridge Regression
Maher Maalouf∗ , Naji Khoury† , Dirar Homouz‡ and Kyriaki Polychronopoulou §
∗ Khalifa University, P.O. Box 127788, Abu Dhabi, UAE
Email: maher.maalouf@ku.ac.ae
† Notre Dame University - Louaize, Beirut, Lebanon

Email: nnkhoury@gmail.com
‡ Khalifa University, P.O. Box 127788, Abu Dhabi, UAE

Email: dirar.homouz@ku.ac.ae
§ Khalifa University, P.O. Box 127788, Abu Dhabi, UAE

Email: kyriaki.polychrono@ku.ac.ae

Abstract—The natural gas compressibility factor (z) is one of
the critical parameters in the computations used for the upstream
and downstream zones of petroleum/chemical industries. The
process of obtaining accurate value for physical and thermody-
namical properties of hydrocarbons is getting more challenging
in the case of multicomponent non ideal systems. The purpose of
this work is applying the kernel ridge regression (KRR) in the
form of the recently developed truncated regularized kernel ridge
regression (TR-KRR) algorithm to estimate z-factor. Compared
to the support vector machines (SVM), the KRR algorithm is
just as accurate as, but faster than SVM.
Index Terms—kernel ridge regression, z-factor, gas compress-
ibility factor, truncated-Newton method.
I. I NTRODUCTION
Natural gas compressibility factor (z) is one of the critical
parameters in the calculations that are used for the upstream
and downstream zones in petroleum/chemical industries and
other fields such as natural gas as the latter is being used for
electric power generation. In particular, natural gas is a mixture
of several components with methane as the major compound,
whereas other compounds can be found such as nitrogen,
carbon dioxide, ethane, propane, and heavier hydrocarbons
[1], [2], [3]. From an engineering perspective, handling pro-
cesses that involve natural gas mandates the knowledge of the
compressibility factor, which basically describes the deviation
between real gas and ideal gas. Usually is difficult to define
the critical parameters for a natural gas mixture. For this,
pseudo-critical properties are widely used in the petrochemical
industry. Critical properties values can be calculated based on
mixing rules. Given a known composition, Kay [4], SBV [5]
and SSBV (modified by Sutton [6]) are mixing rules that have
been widely used. In general it can be stated that the mixing
rule does not follow a simple average of the concentrations of
each component in the mixture. Also, Different components in
the natural gas can be classified to their geometry and polarity.
In particular, small, spherical molecules (e.g. CH4 ) are well
fitted by a two-constant law of corresponding states. On the
other hand, non-spherical and weakly polar molecules require
the development of correlations using a third parameter, e.g.,
the acentric factor, ω. Given the above mentioned criteria and
considerations, the selected composition of the natural gas for
the accurate prediction of Z was sweet dry natural gas. These
data were in the range of 35o to 250o F and 1, 000 to 10, 000
lb. per sq. inch [7].
The process of obtaining accurate value for physical and
thermodynamical properties of hydrocarbons is getting more
challenging in the case of multicomponent non-ideal systems
[1], [2], [3]. There are two families of models in thermodynam-
ics for phase equilibrium predictions and calculations, namely
the equation-of-state model and liquid activity coefficient.
Activity coefficient model can only be used for describing
mixtures of any complexity as long as the liquid is well below
its critical temperature. On the other hand any mathematical
formulation of V , P , T and composition is called equation
of state (EoS). Although many equations of state have been
proposed, almost all are empirical. The most common sources
of Z-factor computations are the experimental measurements,
equations of state (EoS) and empirical correlations [1], [2],
[3]. In the literature there are above twenty correlations with
two variables for calculating the Z-factor. However these
correlations are complex and require specifying an initial value
and long computational time.
There are several popular regression (prediction) algorithms
such as artificial neural networks (ANNs) [8], support vec-
tor machines (SVMs) in addition to the least squares (LS)
method. ANN is an approach that mimics biological neurons
to reproduce intelligent data processing techniques such as
pattern recognition, classification, regression, among others,
by using processing units that are called artificial neurons.
The SVM was first introduced by Vapnik [9] as a potential
alternative to conventional ANN and is a kernel-based method
for classification, regression and function approximation [10],
[11]. The popularity of this method has increased and has been
used in various research areas due to its superior stability and
efficiency superior way compared to other techniques.
Ordinary least squares (LS) algorithm is a simple and
powerful technique that can be used for linear regression. It
can be easily programmed and implemented with little effort.
Thus, one would benefit from extending LS to non-linear

978-1-7281-0130-9/19/$31.00 ©2019 IEEE

 2019 Fourth International Conference on Advances in Computational Tools for Engineering Applications (ACTEA)
regression problems. This can be done by utilizing the kernel
approach used in SVM and applying it to the LS method.
This combination gives rise to kernel ridge regression (KRR),
which is an easy to use and powerful nonlinear regression
method. Recently, Maalouf and Homouz [12] significantly
improved the efficiency of KRR by using truncated Newton
method for parameter estimation.
The objective of this work is applying the recently devel-
oped method of kernel ridge regression (KRR) to estimate z-
factor. The z-factor determination is both resource and time
consuming and hence it is important to utilize a fast and
accurate technique to estimate the z-factor.
II. T RUNCATED -R EGULARIZED K ERNEL R IDGE
R EGRESSION (TR-KRR) A LGORITHM
Our objective is to estimate a real-valued functional out-
come in the form of vector y using input data X through a
function y = f (X). The simplest form of this function is the
linear model and it can be expressed in matrix form as:
y = Xβ + , (1)
where β is the variable vector and  is a random error. The
data X in RN ×d consist of N rows and d variables withy as
an outcome vector, and each row is associated with an output
yi .
The least squares (LS) method is widely used for estimating
β [13], [14] by minimizing the sum of squared residuals
(RSS),
`
X Newton. Maalouf and Homouz [12] combined KRR with the
RSS = 2i = T  = (y − Xβ)T (y − Xβ). (2)
i=1
As long as the matrix (XT X) is non-singular, a solution exists
and is given by
β̂ = (XT X)−1 XT y. (3)
The LS method has some disadvantages such as poor esti-
mation of the regression coefficients with unstable and large
absolute values [15] and it is also limited to linear regression
models. Kernel methods such as Kernel Ridge Regression
(KRR) can overcome these difficulties in the LS method.
KRR introduces a non-linear function φ(.) that maps the
data from one dimensional space into a higher dimensional
one, such that
φ : x ∈ Rd → φ(x) ∈ F ⊆ RΛ . (4)
The φ transformation function usually maps nonlinear rela-
tionships between y and X into a linear relationship. Al-
though these transformations φ(.) are generally unknown,
the regression solution to the problem depends mainly on
the dot product in the feature space. The kernel function,
K = K(xi , xj ) = hφ(xi ), φ(xj )i takes the place of the dot
product.
Therefore, the regression model using the KRR method is
y = Kα + , (5)
where α is the unknown parameter in the KRR model and is
estimated by the following objective function minimization:
1 λ
f (α) = (y − Kα)T (y − Kα) + αT Kα. (6)
2 2
where λ ≥ 0 is the regularization (cost) parameter. The
regularization term λ solves the large unstable coefficient
problem discussed in the least-squares method.
Hence, the solution can now be rewritten with respect to α
as
α = (K + λIN )−1 y. (7)
Using direct and closed-form solution (7) to find the esti-
mate of α could be very slow if the matrix (K + λIN ) is
dense [11] as it involves N equations and N unknowns, and
a complexity of O(N 3 ).
Alternatively, iterative approaches such as the truncated-
Newton method can be used to the matrix inversion. The KRR
model above can be re-written as
(K + λIN )α = y, . (8)
The above model is a linear system of equations that consists
of K as the kernel matrix and y as the prediction vector. This
system can be solved iteratively using all kinds of numerical
methods, including the conjugate gradient (CG) method. The
CG method has a computational complexity of order O(N 3 ) at
worst if the solution converged in N steps. However, placing
a limit on the number of iterations leads to the truncated
truncated Newton method in a nearly linear CG algorithm
called Truncated Regularized Kernel Ridge Regression (TR-
KRR).
Interested readers should refer to Maalouf and Homouz [12]
for a complete description of the TR-KRR algorithm.
III. T HE Z-FACTOR DATA
The data used in this analysis is similar to the data in the
work of Kamyab et al. [16]. This data is based on digitizing
the Standing Katz chart [7], [2] for natural hydrocarbon gases.
The data shows the dependence of the z-factor on two vari-
ables; pseudo-reduced pressure (Ppr ) and the pseudo-reduced
temperature (Tpr ). The reduced pressure (or temperature) is
defined as pressure (or temperature) divided by the critical
value of pressure ( or temperature) for an individual gas. For
a mixture of gases such as natural gas, the critical value is
replaced with pseudo-critical one and thus the pseudo-reduced
pressures and temperatures.
IV. C OMPUTATIONAL R ESULTS & D ISCUSSION
In this study, we have used both the KRR and the support
vector regression (SVR) methods to predict the z-factor with
ten-fold cross-validation [17]. The n-fold cross-validation di-
vides the data into n-folds, leaving n−1 folds for training, the
remaining fold is used for testing the model. This procedure
continues iteratively until all the testing folds are used. The
 2019 Fourth International Conference on Advances in Computational Tools for Engineering Applications (ACTEA)

Algorithm 1: Linear CG for computing α̂. A = K+λIN , advantage of KRR method lies in its high computational
b=y efficiency over SVR. When it comes to accuracy, both methods
Data: A, b, α̂(0) produce very close accuracy. In addition, both models have to
Result: α̂ such that Aα̂ = b be optimized for their parameters. Now, given the fact that
1 begin KRR is faster that SVR, then those parameters can be found
2 r(0) = b − Aα̂(0) /* Initialize the residual much faster using KRR.
*/
3 c=0 TABLE I: Optimal parameters using ten-fold cross validation
4 while ||r(c+1) ||2 > ε2 and c ≤ Max CG Iterations do
for both TR-KRR and SVR, where γ is the RBF kernel width
5 if c = 0 then
6 ζ (c) = 0
and C is the regularization parameter of SVR
7 else
T(c+1) (c+1) TR-KRR SVR
8 ζ (c) = r rT(c+1)r r(c) /* Update
λ σ C γ 
A-Conjugacy enforcer */
0.01 0.35 100 4.5 0.1
9 d(c+1) = r(c+1) + ζ (c) d(c) /* Update the
search direction */
T(c) (c)
10 s(c) = − drT(c) Ad
r
c /* Compute the optimal
step length */ TABLE II: Ten-fold CV accuracy and CPU time (in seconds)
11 α̂(c+1) = α̂(c) − s(c) d(c) /* Obtain an for both TR-KRR and SVR.
approximate solution */
12 r(c+1) = r(c) − Aα(c) /* Update the TR-KRR SVR
residual */ R2 MSE CPU Time R2 MSE CPU Time
13 c=c+1 0.997 0.00039 3.33 0.997 0.00039 5.56

average of all the results is the overall accuracy. Furthermore,
the Gaussian Radial Basis Function (RBF) kernel:
2
(− 2σ12 ||xi −xj || )
K(xi , xj ) = e = e(−γ||xi −xj ||
is chosen for both methods, with a kernel width σ. The
accuracy of the algorithms is computed using the ten-fold cross
validation average of both the coefficient of determination (R2 )
and mean-squared error (MSE) values. The MSE measures
the prediction error and defined as the average of the squared
errors between the actual data and the predicted values of the
model.
The data in our experiments is normalized for each variable
vector such that each variable vector has a mean of zero and a
standard deviation of one. In addition, the normalization was
only applied on the independent variables. The KRR method
is implemented on a personal computer with 8GiB of RAM,
and for SVR the toolbox LIBSVM for MATLAB [18] is used
on the same machine.
The final and optimal parameters with their corresponding
accuracy and computational time are shown in Tables I and
II, respectively. Table I provides the optimal parameter values
for both regression models. As mentioned earlier, the param-
eters of both models could be either user defined or found
through a grid search algorithm. It should be noted here that
different parameter values, specially for SVR could affect the
computational speed and the accuracy of the algorithm. Table
II shows that the two methods produce very high accuracy
for the corresponding parameter values shown. Regarding
computational time, the computational speed of the TR-KRR
algorithm is almost double that of SVR, and this agrees with
the findings of Maalouf and Homouz [12] that the biggest
V. C ONCLUSIONS
We have presented the use of truncated regularized ker-
nel ridge regression (KRR) algorithm as an alternative to
2
)
(9) EoS and empirical correlation approaches of calculating gas
compressibility factor (z-factor). The KRR predicts the z-
factor by building an nonlinear regression model in terms of
the pressure and temperature. The performance of the KRR
algorithm is compared to that of the state-of-art data mining
method support vector machines (SVM). With the conclusions
that KRR is much computationally more efficient than the
support vector regression (SVR) method, while both methods
provide an accurate way for calculating the z-factor, it can be
concluded that KRR is a time efficient method for large data
sets.
R EFERENCES
[1] M. B. Standing, Volumetric and phase behavior of oil field hydrocarbon
systems: PVT for engineers. California Research Corp., 1951.
[2] D. L. V. Katz, Handbook of natural gas engineering. New York :
McGraw-Hill, 1959.
[3] J. W. Amyx, D. M. Bass, and R. L. Whiting, Petroleum reservoir
engineering: physical properties. McGraw-Hill College, 1960, vol. 1.
[4] W. Kay, “Gases and vapors at high temperature and pressure-density of
hydrocarbon,” Industrial & Engineering Chemistry, vol. 28, no. 9, pp.
1014–1019, 1936.
[5] W. Stewart, S. Burkhardt, and D. Voo, “Prediction of pseudo-critical
parameters for mixtures,” in AIChE Meeting, Kansas City, MO, May,
vol. 18, 1959.
[6] R. Sutton et al., “Compressibility factors for high-molecular-weight
reservoir gases,” in SPE Annual Technical Conference and Exhibition.
Society of Petroleum Engineers, 1985.
[7] M. Standing and D. Katz, “Density of natural gases,” Trans., AIME, vol.
146, pp. 140–149, 1942.
[8] C. M. Bishop, Neural networks for pattern recognition. Oxford
university press, 1995.
[9] V. Vapnik, The Nature of Statistical Learning. Springer, NY, 1995.
 2019 Fourth International Conference on Advances in Computational Tools for Engineering Applications (ACTEA)

[10] N. Cristianini and J. Shawe-Taylor, An Introduction to Support Vector

Machines and other kernel-based learning methods. Cambridge
University Press, 2000.
[11] J. Shawe-Taylor and N. Cristianini, Kernel Methods for Pattern Analysis.
Cambridge University Press, 2004.
[12] M. Maalouf and D. Homouz, “Kernel ridge regression using truncated
newton method,” Knowledge-Based Systems, vol. 71, pp. 339–344, 2014.
[13] J. M. Lewis, S. Lakshmivarahan, and S. Dhall, Dynamic Data Assimi-
lation: A Least Squares Approach. Cambridge University Press, 2006.
[14] D. C. Montgomery, E. A. Peck, and G. G. Vining, Introduction to Linear
Regression Analysis, 5th ed. Wiley, 2012.
[15] L. C. C. Montenegro, E. A. Colosimo, G. M. Cordeiro, and F. R. B. Cruz,
“Bias correction in the cox regression model,” Journal of Statistical
Computation and Simulation, vol. 74, no. 5, pp. 379–386, 2004.
[16] M. Kamyab, J. H. Sampaio, F. Qanbari, and A. W. Eustes, “Using
artificial neural networks to estimate the z-factor for natural hydrocarbon
gases,” Journal of Petroleum Science and Engineering, vol. 73, no. 3,
pp. 248–257, 2010.
[17] B. Efron and R. J. Tibshirani, An Introduction to the Bootstrap. Chap-
man & Hall/CRC, 1994.
[18] C.-C. Chang and C.-J. Lin, LIBSVM: a library for
support vector machines, 2001, software available at
http://www.csie.ntu.edu.tw/ cjlin/libsvm.