Fluid Phase Equilibria, 51 (1989)

87-100 87
Elsevier SciencePublishers B.V.,Amsterdam-PrintedinThe
Netherlands

GENERALIZED RACKETT-TYPE CORRELATIONS TO PREDICT THE DENSITY OF

SATURATED LIQUIDS AND PETROLEUM FRACTIONS

Jose 0. Valderrama* and Base1 F. Abu-Sharkh

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

ABSTRACT

Two new generalized correlations for the prediction of

saturated liquid densities at temperatures between the triple point
and the critical point are presented. The correlations use pure
component parameters which are commonly available: the critical
properties, the molecular weight, and the normal boiling
temperature. The correlations have been tested for many substances
including polar and non-polar fluids, and organic and inorganic
compounds. Results are compared to predictions by other generalized
correlations and to experimental data from the literature. The
proposed equations predict the density of saturated liquids within
3% deviation for most fluids and with a maximum deviation of 5%.
This accuracy is better than that obtained with other available
generalized correlations. Another advantage of the new correlations
is that they are generalized in terms of pure component properties
readily available in the literature for most substances. It is also
shown than the proposed correlations give good predictions of
liquid density of petroleum fractions.

IN!l!RODUCTION

Liquid density values are required in many design problems

and in liquid metering calculations in the industry. The design of
condensers and reboilers, the sizing of storage vessels, and the
calculation of tower heights, material and energy balances
involving liquids, and vapor-liquid equilibrium, all require
accurate liquid density values. Also, repetitive calculations in
process simulation require a good generalized correlation which
uses pure component parameters. Although several correlations for
density estimation have been proposed in the literature, they fail
in fulfilling one or more of the basic features which a good
correlation should possess, namely: 1) mathematical simplicity; 2)
acceptable accuracy for a wide variety of substances; 3)
applicability over a wide range of temperature and pressure; and

* Author to whom correspondence should be addressed.

Present address: Cole-Co10 1498, Linares, Chile.

0378-3812/89/$03.50
0 1989Elsevier Science Publishers B.V.
88

4) they should be predictive, so they can be used in those cases

where experimental data are not available.

Good reviews on methods for liquid density estimation are

those of Spencer and Danner (1972) and of Spencer and Adler (1978).
The main conclusion of these reviews is that the most simple
equation to estimate the density of pure saturated liquids with
reasonable accuracy was a modification of Rackett's equation
(Rackett, 1970). This modified version of Rackett equation is known
as the Spencer-Danner correlation (Spencer and Danner, 1972) and it
uses one single adjustable parameter. Shah and Yaws (1972) proposed
an improved equation which includes two adjustable parameters for
each compound. More recently, Campbell and Thodos (1985) have
modified the Spencer-Danner correlation by making the single
empirical parameter a function of the temperature. In all these
approaches, adjustable parameters estimated from experimental data
are required. A good generalized correlation which uses available
pure component properties has not yet been proposed.

In this work, we have analyzed two empirical correlations

which have been shown to give accurate density predictions for many
classes of fluids. These are: the modified Rackett equation
proposed by Spencer and Danner (1972) and the two parameter
equation developed by Shah and Yaws (1976). These equations have
served as bases to obtain generalized correlations in terms of pure
Component properties.

GENERALIZED SPENCW-DANNEFl CORRELATION

Rackett (1970) proposed the following equation for the density

of a liquid (PL) as a function of the temperature:

-[1+(l-TR)2'7]
nL = (MRc/RTc)Zc (1)

The equation is predictive in nature and can be used for many

simple, non-polar liquids.

However, in their review, Spencer and Danner (1972) have found

significant deviations in density estimation using Rackett's
equation. They proposed a simple modification of Rackett's equation
 89

which makes the correlation more accurate, but at the expense of

losing its generalization. Spencer and Danner replaced the critical
compressibility factor Zc by an adjustable empirical parameter ZRA
as follows:

-[l+(l-T,)2'7]
PL = (MPC/RTC)ZRA

For 84 liquids, including hydrocarbons, other organic compounds,

and some inorganic substances, Spencer and Danner found average
deviations below 1% (0.5% for hydrocarbons, 0.6% for other organic
fluids, and 0.74% for inorganic compounds). More recently, Danner
and Daubert (1983) and also Reid et al. (1987) provide values of
2RA for many compounds. These include simple light hydrocarbons up
to C8' heavier hydrocarbons up to C,6, polar fluids such as
alcohols, ketones, esters, amines, ethers, and organic acids, more
complex fluids such as some polyols, polyfunctional compounds,
inorganic acids and other inorganic substances. Average percent
deviation between experimental and predicted values (using the
given ZRA) are usually below 1%. For only 13 fluids, deviations
greater than 2% were found with some as high as 4% (for
triethylene glycol), 5% (for nitric oxide), and 6.7% (for
hydrogen). We have not considered these last three substances in
our study.

To obtain a relation between the ZRA parameter in Spencer-

Danner correlation and other pure component properties we start
with the equation for the density ( pb) at the normal boiling point
(Tb) as:

-[l+(l-TbR)2'7]
= M/Vb = (MRC/RTC)ZRA (3)
nb

where TbR--Tb /Tc is the reduced temperature at the normal boiling

point and Vb is the molar volume at the normal boiling point. For
Vb we use an equation of the form given by Tyn and Calus (1975)
which relates the volume at the normal boiling point (Vb) to the
critical volume (Vc) through the following relation:

V b = 0; (4)
90

with Vb and Vc expressed in (cm3/gr mole). Tyn and Calus proposed

to use CY = 0.285 and B = 1.048. In our work, we have considered .
and 8 as two adjustable parameters, a and b. Rearranging eqn. (3)
we fi nd:

2/l]
Il+( I-TbR)
= (aPc/RTc)V; (5)
'RA
or
2/7]
W+(l-TbR)
ln(Tc/Pc)ZRA = ln(a/R) + blnVc (6)

The values of ZRA given by Danner and Daubert (for 116 fluids) were
used in this work. Other data needed, such as critical properties,
normal boiling point, and molecular weight were obtained from Reid
et al. (1987).For a and b,the following values were found through
regression analysis: a = 0.3445 and b = 1.0135. Thus, the correla-
tion for ZRA in terms of the normal boiling temperature (Tb), the
critical temperature (Tc), and the critical volume (Vc) is:

1/[1+(1-TbR)2'7]
= 1 (0.3445P,/RT,)V;'01351 (7)
'RA

-1

I I I I I I I I

4.0 4.5 5.0 5.5 6.0 6.5 7.0

Fig. l.- The parameter ZRA as given in eqn. (6) -vs.- lnV,.
 91

Figure 1 shows the correlation as given by eqn. (6). We obtain

the linear behavior suggested by eqn. (6) with only small
deviations in the case of a few fluids. The final expression for
the density using the generalized expression for ZRA is given in
Table 1.

GENERALIZED SHAH AND YAWS CORRELATION

Shah and Yaws (1972) improved upon the Rackett and Spencer-
Danner proposals by convenient regrouping of the terms and the
introduction of two empirical parameters. Their proposed equation
was:

(1-TR)2’7
=A/B (8)
pL

.6

.5

.4
4
I.4
2 .3
8
E
d .2

.l

cl _
0 .l .2 .3 .4 .5 .6
(W/vC)

Fig. 2.- The parameter A -vs- the critical density (M/V=)

92

Here, A and B are adjustable empirical parameters. The authors

provide values of A and B for several compounds calculated by
regression of experimental data of liquid density. In most cases
the data were available at temperatures from the triple point to
the critical point. The deviations between predicted and
experimental values of density were below 1% for most fluids.
Water, helium, and the alcohols are the exception. For these
components, deviations between 1% and 2% were found. The fluids
considered to test the correlation included some halogens, sulfur,
carbon and nitrogen oxides, hydrogen halides, nitrogen hydrides,
organic oxides, alcohols, and many classes of hydrocarbons.

To generalize Shah and Yaws's correlation, the parameters A

and B were treated in a fashion similar to the SRA parameter in
Spencer-Danner equation. We have considered A as a function of the
critical density ( PC = M/Vc, M being the molecular weight) and
plotted the parameter A -vs- M/Vc. As obtained from Fig. 2 the
parameter A can finally be expressed in terms of the critical
density (M/Vc), as follows:

A = 0.01256 + 0_9553(M/Vc) (9)

5.5

5.0

1 I I I I I I
3.5 4.0 4.5 5.0 5.5 6.0 6.5
In VC

Fig. 3.- The parameter B as given by eqn.(lO) -vs- lnV,

 93

We should note here that with the use of this correlation we cannot
reproduce the exact value of the density at the critical point.
However, better density predictions are obtained at lower
temperatures.

The parameter B is not clearly related to a specific property

of the fluid. We have used Tyn and Calus's relation for the volume
at the normal boiling point (eqn. 4) and proceeded as was done for
the case of ZRA. Arranging the equations, we finally obtained:

In [ MB(1-TR)2'~[0.01256+0.9553(M/Vc)] = lna' + b'lnVc (IO)

Values of A and B were obtained from Shah and Yaws (1976), and the
parameters a' and b' in eqn. (IO) were calculated by regression
analysis of the given A and B values. The optimum values of a' and
b' were found as: a'=0.3126 and b' =1.0333. The correlation for B
can be finally expressed as:

l/(l-TbR)2'7
B = 1 (0.0039/M+0.2987/Vc)Vc1~033 ] (‘1)

Figure 3 shows the proposed correlation as given by eqn. (IO).

According to that equation, this should give a straight line which
is what the figure actually shows. The final correlation for the
liquid density using the generalized equations for A and B is given
in Table 1.

APPLICATION OF THE CORRELATIONS TO PURE FLUIDS

The proposed correlations have been used to predict the

density of many fluids and the results were compared to data given
by Schlunder (1983). Figure 4 shows calculated and experimental
values of saturated liquid densities for a number of selected
fluids. Calculated values were obtained using the generalized
equation for ZRA The degree of accuracy of the predictions using
the proposed correlation for ZRA in different ranges of temperature
can be observed in the figure. The components for which density
values are plotted are the following: (I) propane, (2) isobutane,
(3) n-hexane, (4) n-heptane, (5) n-nonane, (6) acetylene, (7) I,2
butadiene, (8) t-butyl alcohol, (9) I-propanol, and (IO) toluene.
94

650

450 -

350
I I I I I I
250 300 350 400 450 500

TMPERATURE (K)

800 -

600-

500-

400 -

200 250 300 350 400 450 500

TEMPERATURE (K)

Fig. 4.- Density predictions for some selected fluids

using the generalized correlation for ZRA.
 95

Four other correlations, which use critical properties and

acentric factor as generalizing parameters, have been used for
comparison. The correlations are those of Yen and Woods (19661,
Chueh and Prausnitz (1967), Rackett (1970), and Gunn and Yamada
(1971) and are shown in Table 1. The density of several fluids has
been calculated using these correlations. Table 2 shows the percent
deviation in liquid density estimation using the four correlations
mentioned above and the two proposed in this work. As seen in the
table, there are some cases in which other equations give better
predictions than our new correlations, although our equations still
give deviations below 5% for all fluids. In some cases other
correlations give deviations of over 20%. The absolute percent
deviation %D for a set of N experimental data points is defined as:

%D = (100/N) ~i[PLLexp)-PL(cal)llPL(exp)li (12)

1
The numbers in the Table 2 correspond to this definition of percent
deviation.

APPLICATIONS TO PETROLEUM FRACTIONS

One important application of liquid density correlations is

the estimation of the density of petroleum fractions. Experimental
data on density and other properties of three petroleum fractions
were taken from Saeed (1986). Table 3 gives some characterization
properties of these petroleum fractions and the predicted values of
ZRA (from eqn. 7), A (from eqn. 9), and B (from eqn. 10). The
critical properties and the molecular weight of the petroleum
fractions required to evaluate the genaralized parameters ZRA, A,
and B were estimated using the correlations proposed by Riazi and
Daubert (1980). Riazi and Daubed's correlations are simple and
give reasonable values of the critical properties and the molecular
weight for petroleum fractions in the range of temperature
considered here. Table 4 shows the values predicted by the
correlations proposed here (see Table 1) for the three petroleum
fractions described in Table 3. The predicted densities are within
4% accuracy using the correlation for ZRA and within 1.5% accuracy
using the correlations for A and B.
96

TABLE 1: Generalized correlations (which use critical propertie

and acentric factor only) used in this work.

Generalized ZRA (this work)

-[1+(l-T,)2'71/[l+(1-TbR)2'7]
pL = (MP,/RT,)[(0.3445PC/RTc)V;-033]
Generalized A and B (this work)

= (0.01256 + 0.9533M/Vc)[(0.0039/M+0.2987/VC)V~-0331 Y(T)

pL
Y(T) = -[(l-TR)/(l-T,R)12'7

Yen and Woods (1966)

4
i/3
PL = P,[I + Z $(I-TR) 1
1
kl = 17.4425 - 214.578Zc + 989.6252; - 1522.062:

k2 = -3.28257 + 13.6377Zc + 107.484422 - 384.2112: for (Zc < 0.26)

k2 = 60.2091 - 402.063Zc + 501.02~ + 641.0Zc3 for (Zc > 0.26)

= 0 and k4 = 0.93 - k2
k3
Chueh and Prausnitz (1967)

PL = P,/(Vo +wv, + W2V2)

'i
=ai+b.T
1R
+c iT; + d i T i + ei/TR + filn(l-TR)

= 0.11917
;oo = 0.009513 ;2 1 -0"-;;;;3"
= 0.21091 c 2 = -I:01601
:oo = -0.06922 2
d2 = 0.34095
e = 0.07480
fo" = -0.084476 ;2 z _;.;63;;;8
2 -
Racket (1970)

-[l+(l-TR)2/7]
PL = (MP~/RT~) zc

Gunn and Yamada (1971)

PL = vo/[v1(1-~v2)l

vO
= P /RTc(0.2920 - 0.0967w)
2

vl = 0.33593 - 0.33953TR + 1.51941T; - 2.02512T; + 1.11422T;

for 0.2~T~c0.8, and

v1 = 1.0 + 1.3(l-TR)"2 log(l-TR) - 0.50879(1-TR) - 0.91534(l-TR)2

for 0.8<TRc1.0

v2 = 0.29607 - 0.09045TR - 0.04842T; (for TR>0.2)

 97

TABLE 2: Percent deviation in the prediction of liquid densities of

pure fluids using several generalized correlations.

Yen- Chueh- Rackett Gunn- --This work--

Woods Prausnitz Yamada A&B z
RA

Methane 1.5 1.6 3.0 1.9 4.3 3.6

Ethane 1.1 1.0 3.5 4.0 1.3 1.1
Propane 0.5 1.5 2.1 3.0 0.3 0.9
iso-Butane 3.9 2.9 4.0 2.8 1.2 0.7
nButane 2.3 2.6 1.1 4.2 0.8 1.6
neo-Pentane 3.2 2.3 3.9 2.2 1.3 2.1
iso-Pentane 0.2 0.4 0.7 3.1 1.7 3.1
nPentane 4.3 5.6 3.6 4.8 3.7 4.9
nHexane 0.2 1.3 2.5 5.4 0.6 1.2
nHeptane 1.4 2.1 2.1 5.4 1.8 1.3
noctane 1.7 0.8 1.8 5.6 1.6 1.5
nNonane 1.9 1.4 4.2 6.3 0.9 1.5
nDecane 0.6 2.0 0.5 8.4 1.4 0.9
cycle-Pentane 0.4 0.9 0.2 2.2 0.8 2.0
cycle-Hexane 0.2 0.5 0.5 2.8 0.7 2.2
Benzene 2.1 2.4 1.0 5.4 2.1 0.9
Toluene 0.5 1.0 0.7 7.4 0.7 0.9
mXylene 1.6 1.8 1.8 5.8 1.8 1.9
0Xylene 1.7 0.9 0.1 5.3 0.9 0.9
pXylene 3.5 0.7 7.3 2.0 0.3
Ethyl Benzene 0.4 0':: 0.7 5.1 0.5 1.3
Acetylene 2.7 4.2 5.4 3.8 3.6
Ethylene 3.3 i:: 3.5 2.7 3.7
Propylene 2.3 1.7 1.5 3.1 3:: 2.8
I,2 Butadiene 0.5 1.1 0.4 5.1 0.5 0.8
I,3 Butadiene 1.4 2.2 2.0 6.3 1.7 2.3
Methanol 1.2 5.2 3.0 12.5 3.7 4.3
I-Propanol 2.4 3.7 2.7 1.4 1.0
Ethanol 2.0 E 2.6 4.2 3.8 3.6
n-Butanol 0.7 9:4 12.1 15.7 1.1 3.5
t-Butylalcohol 1.7 10.1 1.8 2.4 0.4 1.4
Ethyl Ether 1.3 0.5 0.8 6.2 0.7 1.8
Ethylene Oxide 1.1 4.5 1.1 12.6 3.6 3.9
Propylene Oxide 9.3 4.4 15.9 14.7 4.5 4.8
Ethyl Acetate 1.2 1.7 8.3 1.4 0.6
Chloroform 9.6 9:: 10.5 2.7 4.4 4.9
Aniline 6.8 6.2 9.7 3.6 4.4 4.2
Acetic Acid 1.8 3.6 17.6 24.5 0.6 2.7
Acetone 2.0 2.4 9.6 15.7 0.8 0.9
Ammonia 1.3 3.8 3.2 16.6 3.2 4.1
Carbon Dioxide 1.4 1.3 1.0 2.7 1.9 2.5
Carbon Monoxide 1.9 0.6 2.6 1.0 3.9 4.5
C. Tetrachloride 4.7 4.7 0.3 4.2 1.3 1.4
Chlorine 0.9 1.0 1.6 5.4 0.4 0.6
Fluorine 6.3 5.7 3.2 2.6 4.3 4.5
Nitrogen 3.4 3.0 0.5 0.4 4.2 4.6
Oxigen 1.1 6.1 6.0 2.5 4.2 4.3
98

TABLE 3: Characterization Properties and other calculated

properties of three petroleum fractions obtained from
a medium Arabian Crude Oil (Experimental values are
from Saeed, 1986)

Property Fraction 1 Fraction 2 Fraction 3

Boiling Temp. Range (K) 368-478 478-533 618-728

API gravity 55.2 42.4 26.7

50% Boiling Temp. (K) 423 508 669

Specifiz Gravity (60°F) 0.758 0.814 0.899

Average Molec. Weight 124 178 317

Critical Pressure (bar) 23.9 18.5 12.3

Critical Temperature (K) 604.1 696.0 833.5

Critical Volume (lt/Kmol) 496 710 1275

SRA (eqn. 7) 0.245 0.237 0.226

A (eqn. 9) 0.251 0.252 0.250

B (eqn. 10) 0.261 0.255 0.233

TABLE 4: Experimental and calculated values of density (Kg/m3)

for the three petroleum fractions described in Table 3.
(Experimental values are from Saeed, 1986)

Fraction 1 Fraction 2 Fraction 3

T(K) exp. A&B exp. A&B exp. SRA A&B

=RA 'RA

303 746 784 756 803 835 804 889 886 901

323 731 761 740 789 820 791 876 874 889

343 714 743 724 774 805 777 863 861 876

373 690 721 698 753 781 755 844 841 856

393 673 701 680 738 765 740 832 828 844

423 647 671 652 716 740 717 812 808 823

443 628 649 632 700 723 701 800 794 810

473 597 614 599 677 696 676 781 773 788

503 568 576 563 652 667 650 762 751 766
 99

CONCLUSIONS

Our new correlations predict densities of liquids with a

deviation of below 5% for many classes of fluids. We have also
tested several correlations published in the literature and found
some deviations of the order of 10 to 20%. Only pure component
properties (critical temperature and volume, molecular weight, and
normal boiling temperature) are required to predict the density of
saturated liquids in the temperature range from the triple point to
the critical point. For any fluid in this range of temperature, the
correlations proposed here predict liquid densities with deviations
of the order of 2 to 3% with maximum deviations of 5%. Good density
predictions for petroleum fractions are also obtained.

NOTATION

a,b: parameters in the correlation for ZRA (dimensionless)

a',b': parameters in the correlation for B (dimensionless)
A,B: parameters in Shah and yaws's equation (dimensionless)
M: molecular weight
N: number of experimental points in a set of data
PC: critical pressure (atm)
R: ideal- gas constant (atm cm '/gr mole K)
T: absolute temperature (K)
T: normal boiling temperature (K)
b
reduced normal boiling temperature (TbR=Tb/Tc)
TbR'
Tc: critical temperature (K)
T: reduced temperature (TR=T/Tc)
R
v: volume at the normal boiling temperature(cm3/gr mole)
b
vc: volume at the critical point (cm3/gr mole)
ZC: critical compressibility factor (dimensionless)
parameter in Spencer-Danner correlation (dimensionless)
'RA'
Greek symbols
a, 8: constants in Tyn and Calus's correlation (dimensionless)
Pb: density at the normal boiling point (gr/cm3)
P z: density at the critical point (gr/cm3)
PD: density of a saturated liquid (gr/cm3)
PL(cal): calculated liquid density (gr/cm3)
PL(exp): experimental liquid density (gr/cm3)

W: acentric factor (dimensionless)

100

REFERENCES

Campbell, S.W. and G. Thodos, 1985. Prediction of Saturated Liquid

Densities and Critical Volumes for Polar and Nonpolar Substances.
J. Chem. Eng. Data, 30: 102-111.

Chueh, P.L. and J.M. Prausnitz, 1987. AIChE J., 13: 1099. (Taken
from Reid et a1.,19871.

Danner, R.P. and T.E. Daubert, 1983. Manual for Predicting Chemical
Process Design Data. Ch. 4: pp l-5 AIChE, New York - USA.

Gunn, R.D. and T. Yamada, 1967. AIChE J., 17: 1341. (Taken from
Reid et a1.,19871.

Katz, M., P. Lobo, A. Sancho, and H. Solimo., 1971. Viscosities,

Densities, and Refractive Indices of Binary Liquid Mixtures. Can.
J. Chem., 49 (15): 2605-2609.

Orrit, J.E., 1983. Orthobaric Liquid Densities of Natural - Gas

Component Mixtures. Fluid Phase Equil., 12: 253-281.

Perry, R.H. and C.H. Chilton, 1974. Chemical Engineers' Handbook,

5th edition. McGraw-Hill Co., New York USA.

Rackett, H.G., 1970. Equation of State for Saturated Liquids. J.

Chem. Eng. Data, 15 (4) 514-517.

Reid, R.C., J.M. Prausnitz, and B.E. Poling, 1987. The Properties
of Gases and Liquids. McGraw-Hill Book Co., New York - USA.

Riazi, M. R. and T. E. Daubert, 1980. Simplify Property Prediction.

Hydroc. Process., 59: 115-116.

Saeed, M. R., 1986. Density Behavior of Saudi Arabian Crude Oils

and their Fractions. M-SC. Thesis, University of Petroleum &
Minerals, Dhahran-Saudi Arabia.

Schlunder, E.U. (editor), 1983. Heat Exchanger Design Handbook,

vol. 5. Hemisphere Pub. Corp., New York-USA.

Shah, P-N. and C.L. Yaws, 1976. Densities of Liquids. Chem. Eng.
Oct. 25. [Taken from Yaws, 19771.

Spencer, C.F. and S.B. Adler, 1978. A Critical Review of Equations

for Predicting Saturated Liquid Density. J. Chem. Eng. Data, 23(l):
82-89.

Spencer, C.F. and R.P. Danner, 1972. Improved Equation for Predic-
tion of Saturated Liquid Density. J. Chem. Eng. Data, 17: 236-241.

Tyn, M.T. and W.F. Calus, 1975. Estimating Liquid Molar Volume.
Processing, 21(4): 16-17.

Yaws, C.L., 1977. Physical Properties. McGraw-Hill, New York-USA.

Yen, L.C. and S.S. Woods, 1966. A Generalized Equation for Computer
Calculation of Liquid Densities. AIChE J., 12(l): 95-102.