Page 165 - Characterization and Properties of Petroleum Fractions - M.R. Riazi

T1: IML
P2: IML/FFX
P1: IML/FFX
QC: IML/FFX
14:23
June 22, 2007
AT029-03
AT029-Manual-v7.cls
AT029-Manual
3. CHARACTERIZATION OF PETROLEUM FRACTIONS 145
the knowledge of PNA composition for prediction of prop-
other parameters. As the complexity and the number of pa-
erties of heavy fractions is more useful than for light frac- laboratory data for any physical property in terms of some
tions. For wide boiling range fractions knowledge of complete rameters increases the accuracy of the correlation also in-
distillation curve is quite useful to consider nature of different creases with respect to the data used in the development of
compounds and their effects on the properties of the mixture. the correlation. However, the main problem with empirical
As it is shown in Chapter 4, for wide and heavy fractions such correlations is their limited power of extrapolation and the
as C 7+ fractions, distribution of carbon number in the frac- large number of numerical constants involved in the correla-
tion is the most useful information besides specific gravity. tion. For example, in Chapter 2, several correlations are pro-
Further analysis of minimum laboratory data for characteri- vided to estimate molecular weight of hydrocarbons in terms
zation of petroleum fractions is provided in our previous work of boiling point and specific gravity. Equation (2.50) derived
[84]. from Eq. (2.38) has only three numerical constants, which
Predictive methods of characterization must be used when are developed from molecular weight of pure hydrocarbons.
experimental data are not available. If possible, one should Tsonopoulos et al. [86] made an extensive analysis of various
make maximum use of available experimental data. A sum- methods of estimation of molecular weights of coal liquids.
mary of standard test methods for some specifications of Equation (2.50) was compared with several empirical corre-
liquid petroleum products is given in Table 3.31. For some lations specifically developed for coal liquids having as many
properties equivalent test methods according to the interna- as 16 numerical constants. They concluded that Eq. (2.50)
tional standards organization (ISO) are also specified in this is the most accurate method for the estimation of molecular
table [24, 85]. weight of coal liquids. No data on coal liquid were used in
development of constants in Eq. (2.50). However, since it was
developed with some physical basis and properties of pure
3.9 ANALYSIS OF LABORATORY DATA AND hydrocarbons were used to obtain the numerical constants
DEVELOPMENT OF PREDICTIVE METHODS the equation has a wide range of applications from pure hy-
drocarbons to petroleum fractions and coal liquids, which are
In Chapter 2 and this chapter the predictive methods in terms mainly aromatics. This indicates the significance of develop-
of readily available parameters are presented for estimation of ment of correlations based on the physical understanding of
various properties related to basic characteristics and quality the nature of the system and its properties. The main advan-
of petroleum fractions. Generally these methods fall within tage of such correlations is their generality and simplicity.
two categories of empirical and semiempirical correlations. The main characteristics of an ideal predictive method for a
In an empirical correlation the structure of the correlation certain property are accuracy, simplicity, generality, and avail-
is determined through fitting the data and the type of input ability of input parameters. The best approach toward the
parameters in each correlation are determined through anal- development of such correlations would be to combine phys-
ysis of experimental data. While in a semi-empirical corre- ical and theoretical fundamentals with some modifications.
lation, the structure and functionality of the relation is de- An example of such type of correlations is Eq. (2.39), which is
termined from a theoretical analysis of parameters involved an extension of Eq. (2.38) derived from physical basis. A pure
and through analysis of existing theoretical relations. Once empirical correlation might be quite accurate to represent the
the main functionality and nature of a correlation between data used in its development but when it is applied to other
various physical properties is determined, the correlation co- systems the accuracy is quite low. In addition characterizing
efficients can be determined from experimental data. The best the systems according to their degree of complexity is helpful
example of such a predictive method is development of Eq. to develop more accurate correlations. For example, heavy
(2.38) in Chapter 2, which was developed based on the un- fractions contain heavy and nonpolar compounds, which dif-
derstanding of the intermolecular forces in hydrocarbon sys- fer with low-molecular-weight hydrocarbons present in light
tems. This generalized correlation has been successfully used petroleum fractions. Therefore, in order to increase the de-
to develop predictive methods for a variety of physical prop- gree of accuracy of a predictive method for a certain prop-
erties. An example of an empirical correlation is Eq. (2.54) erty it is quite appropriate to develop one correlation for
developed for estimation of molecular weight of petroleum light and one correlation for heavy fractions. For heavy frac-
fractions. Many other correlations presented in this chapter tions because of the nature of complex compounds in the
for estimation of properties such as aniline and smoke points mixture three input parameters are required. As variation in
or methods presented for calculation of octane numbers for a properties of pure hydrocarbons from one family to another
blend are also purely empirical in nature. In development of increases with increase in carbon numbers (i.e., see Figs. 2.15
an empirical relation, knowledge of the nature of properties and 2.23), the role of composition on estimation of such prop-
involved in the correlation is necessary. For example, aniline erties for heavy fractions is more than its effect on the prop-
point is a characteristic that depends on the molecular type erties of light fractions. Therefore, including molecular type
of hydrocarbons in the fraction. Therefore, it is appropriate in the development of predictive methods for such properties
to relate aniline point to the parameters that characterize hy- of heavy fractions is quite reasonable and useful and would
drocarbon types (i.e., R i ) rather than boiling point that char- enhance the accuracy of the method.
acterizes carbon number in a hydrocarbon series. Once the structure of a correlation is determined from
Mathematical functions can be expressed in the form of theoretical developments between various properties, ex-
polynomial series; therefore, it is practically possible to de- perimental data should be used to determine the numerical
velop correlations in the forms of polynomial of various de- constants in the correlation. If the data on properties of
grees. With powerful computational tools available at present pure hydrocarbons from different families are used to
it is possible to find an empirical correlation for any set of determine the constants, the resulting correlation would be

--`,```,`,``````,`,````,```,,-`-`,,`,,`,`,,`---
Copyright ASTM International
Provided by IHS Markit under license with ASTM Licensee=International Dealers Demo/2222333001, User=Anggiansah, Erick
No reproduction or networking permitted without license from IHS Not for Resale, 08/26/2021 21:56:35 MDT