Ind. Eng. Chem. Process Des. Dev. 1983, 22, 410-424
410
Thermophyskai Properties of Coal Liquids. 1 Selected Physical, Chemical, and Thermodynamlc Properties of Narrow Boillng Range Coal Liquids James A. Gray” Gulf Research and Development Company, PmSbwgh, Pennsylvenla 75230
C. Jeff Brady, John
R. Cunnlngham, James R.
Freeman, and Grant M. Wllson
Wttec Research Company, Incorporated, Provo, Utah 84807
Liqukl products from the SRC-I1 processing of a high volatle bituminous coal were distilled into narrow boiling range fractions with average bdhg points ranglng up to 800 K. Liquid density, thermal conductivity, and viscosity were measured at temperatures to 505 K, and specific heat was determined to 600 K. Other property determinations
included molecular weight, pour point, elemental analyses, water content and solubility, and hydrocarbon types. Both organic oxygen content and water solubility showed a pronounced maximum for the fraction having an average boiling point of 470 K and significantly influenced density. A secondary distillation on SIX different fractions spaced over the entire boiling range yiekled heart cuts spanning 4-26 K that were used for measurement of surface tension, heat of vaporization, and vapor pressure at temperatures to 755 K. Properly data for all the fractions are compared
with other publlshed data and existing property correlations.
Introduction The design of operable and efficient processing steps for large-scale coal liquefaction plants requires a knowledge of the physical, chemical, and thermodynamic properties of coal liquid fractions up to the most extreme conditions encountered in the plant. For example, the SRC-I1 processing of coal into distillable products involves conversion and separation steps that operate at high temperatures (to 733 K) and pressures (to 13.9 MPa). It is now widely recognized [Recon Systems, Inc. (1981); Kidnay and Yesavage (1980); Brule et al. (1982); Watanasiri et al. (1982); Hwang et al. (1982); and Wilson et al. (198l)l that estimation of physical and thermodynamic properties from existing data on petroleum fractions can introduce large errors for two reasons: first, available data on petroleum fractions often do not extend to high enough temperatures; and second, coal-derived liquids are significantly more aromatic and have much greater heteroatom contents than conventional petroleum fractions. Thus, the many correlations that have been developed for use in refinery process calculations are frequently not directly applicable and have to be modified or extrapolated into regions where very little information is available. Coal liquids generally boil over a wide range of temperatures and contain so many chemical compounds that the only practical way to characterize the composition for process design purposes is to fractionate the coal liquid into narrow boiling cuts (approximately 28 K) and consider these as pseudocomponents. This approach has been used extensively in the past for complex multicomponent mixtures, such as petroleum. Once the pseudocomponents are defined by minimal data, such as normal boiling point and specific gravity, other properties, such as molecular weight, critical temperature and pressure, and acentric factor, can be estimated from correlations. The latter information can then be used in vapor/liquid equilibrium and enthalpy calculations that serve as the basis in the design of most of the key processing steps in coal liquefaction. 0196-4305/83/1122-0410$01.50/0
This report discusses the results of the first phase of a program to measure selected properties of narrow boiling fractions of SRC-I1 distillate; liquid density, thermal conductivity, viscosity, molecular weight, pour point, elemental analyses, water content and solubility, hydrocarbon types, surface tension, vapor pressure, and heat of vaporization. Experimental Section A. Preparation of Distillate Fractions. The number of property measurements listed above required the distillation of a large quantity of coal liquid in order to obtain sufficient quantities (as much as 15 L in some cases) of the individual fractions. Several large batches (568 L) of full boiling range distillate and recycle slurry (bottoms from atmospheric flashing of hot separator slurry product), generated from SRC-I1processing of Powhatan No. 5 Mine coal (hvAb) on Process Development Unit P-99 (located at Gulf’s Research Center, H a r m m e , PA) were prepared and distilled into 19 narrow boiling fractions having midboiling point temperatures from 340 to 794 K and boiling ranges varying from 11to 103 K in width (except for one deliberately wide fraction having a 170 K boiling range). Distillate was used to produce fractions boiling below 672 K, and recycle slurry was used, after partial solids removal, to yield fractions boiling above 672 K. Reflux control of the distillate fractionation allowed production of 11to 46 K wide cuts, whereas the side-arm method used on the slurry fractionation yielded much wider cuts (83 to 103 K). Side-arm distillation was needed because of the 616 K pot temperature limit to avoid cracking. Secondary fractionationswere performed on six cuts that were spaced about evenly over the full boiling range, in order to obtain 2 to 3 L of six “heart cuts” specificallyfor use in vapor pressure, heat of vaporization, and surface tension measurements. Vapor pressure measurements require rather narrow boiling fractions to prevent significant changes in composition with degree of vaporization from affecting results. The secondary fractions had boiling 0 1983 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
Table I. Basic Characterization of Narrow Boiling Coal Liquid Fractions vapor temp, K a t 101.3 KPa SP gr, 5wt % 50wt % 95wt % cut 288.7 K/ cut no. off off off width, K 288.7 K 1 31 5 340 346 31 0.7234 2 363 37 2 389 26 0.7701 3 387 398 11 39 4 0.7696 4 405 41 0 421 16 0.8125 5 434 440 449 15 0.8956 5HC" 433b 436 439 b 6 0.8827 6 463 47 7 14 469 0.9538 7 48 8 49 3 509 21 0.9622 8 520 525 544 24 0.9761 8HC 521 525 4 523 0.9718 9 542 548 5 60 18 0.9802 10 554 588 34 573 0.9972 11 599 615 64 5 1.0392 46 llHC 610 623 13 616 1.0359 12 639 65 0 1.0793 666 27 13 662 67 1 68 4 1.0918 22 14 480 1.0184 57 2 650 170 15 605 64 2 705 100 1.0773 16 641 67 6 724 83 1.0973 16HC 651 660 669 1.0910 18 17 668 711 7 58 1.1195 90 17HC 694 69 9 705 1.1204 11 18 708 1.1733 750 811 103 19 751 794 852 101 1.1950 19HC-A 766b 779 791 25 1.1792 19HC-B 791 801 1.1971 811 20 4501650 502 548 616 111 0.9951 6501900 618 67 6 787 1.09 169 Heart cut from secondary fractionation.
Pour point, K
Figure 5. Measured and calculated vapor pressures of coal liquid fractions.
5. Vapor Pressure. Vapor pressure measurements were performed on the six narrow boiling heart cuts shown in Table IV at temperatures up to 755-783 K or the critical point. Figure 5 illustrates the data and shows the consistency between the static and flow measurements and between the various samples. The curves in Figure 5 are
derived from a modification of the Grayson-Streed (1963) correlation and indicate excellent agreement with the experimental data. The slight discrepancies that do exist appear to be related simply to the small mismatch between the average boiling points of the samples and the fractions specified in the Grayson-Streed correlation.
416
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
Table IV. Characterization Properties of Secondary Fractions fraction number boiling point,a K specific gravity, 288.7 K/288.7K molecular weight Wikon b WiltecC eq 2 critical temperature, K WiIsonb Wiltec critical pressure, MPa Wilson& Wiltec acentric factor Wilson b Wiltecd
5HC 433 0.8827
8HC 520 0.9718
llHC 612 1.0359
16HC 659 1.0910
17HC 692 1.1204
19HC-A 776 1.1792
19HC-B 801 1.1971
112 126 116
156 169 158
215 214 212
239 258 237
264 275 258
326 350 315
332
645 646"
757 748 a
866d 863
930 919d
969 958d
1068 1058d
1094 1078'
3.25 3.51 a
3.05 3.24"
2.69 2.90'
2.65 2.85'
2.52 2.73'
2.41 2.62'
2.32 2.52'
0.322 0.340
0.396 0.465
0.481 0.531
0.530 0.575
0.563 0.605
0.620 0.679
0.641
-
-
a Determined from vapor pressure measurements. From correlation of Wilson (1981). Determined as adjustable parameter in calculating heat of vaporization from the slopes of the vapor pressure curves. Back-calculated parameter from vapor pressure equation. e Assumed values based on correlation of Wilson et al. (1981)and results for cuts 5HC and 8HC.
Following the same procedure as described by Lee and Kesler (1975), Wilson et al. (1981) used the Riedel equation to fit vapor pressure data on eleven polynuclear aromatic compounds and six coal liquid fractions. The resulting equation was In
( P R ) = f(0)
+ wfi')
(3)
g
+
P
where Po) = 5.671485 - 5.809839/TR- 0.867513 In TR 0.1383536TR6 and f ( l ) = 12.439604 - 12.755971/TR 9.654169 In TR + 0.316367T~~. This equation was used to correlate the vapor pressure data in Figure 5 yielding the critical temperature and acentric factor as the adjustable parameters to fit the data as discussed previously. The predicted critical temperatures in Table IV are in reasonable agreement with those derived from the data by using eq 3. On the other hand, the predicted acentric factors were consistently lower than the experimentally derived values. The average deviations between measured and predicted vapor pressures ranged from 1.0 to 4.7% for the six data sets. 6. Heat of Vaporization. Heats of vaporization for the six heart cuts and two wide boiling samples are shown in Figure 6 as a function of temperature. The heats of vaporization were also calculated based on the slopes of the vapor pressure curves and the Clapeyron equation as d(ln Po) (4) dT where AHv is the latent heat on a molar basis, T is the absolute temperature, A2 is the compressibility factor of vapor minus that for the liquid, and d(ln PO)/dT is the slope of the vapor pressure curve. The compressibility factor of the vapor was calculated from a modified Redlich-Kwong equation of state as described by Wilson et al. (1981), and the compressibility factor of the liquid was calculated by the procedure of Hwang et al. (1982) and Wilson (1981). Since the calculated heats of vaporization are on a mole basis, while the measured heats of vaporization are on a weight basis, the ratio of the two values gives the molecular weight of the fraction. When molecular weight is used as an adjustable parameter to obtain the best match of measured and calculated results, the average deviation ranges from 1.3 to 3.4% for five of the data sets. Only a single data point could be obtained on cut 19HC-A. Comparison of these molecular weight values with$those calculated by the two correlations in Table IV indicates AHv = R P A Z -
m3 . 1
250-
4a
m-
'
150-
8 c
I
1M-
50-
O m
350
450
4w
5w
550
Bw
LL o
7w
1
750
800
1
850
TEMPERATURE K
Figure 6. Heat of vaporization of coal liquid fractions.
good agreement. Thus, it is possible to predict accurately the heat of vaporization from measured vapor pressure data. The molar heat of vaporization at the normal boiling point, AElmp, can also be fitted to an equation based on Routon's rule, similar to the Kistiakowsky [see Smith and Van Ness (195911 equation. Using the molecular weight calculated from eq 2 and interpolating to obtain MVmp from the experimental data, the resulting equation is
--
P"
- 13.176 In Tb + 2.931
Tb
where ARwp is in J/mol and Tb is in K. This equation gives an absolute average deviation of 1.4%. Heats of vaporization at temperatures other than the normal boiling point are calculated from the Watson (1931) relationship
where T, is the critical temperature. 7. Liquid Density. Two general approaches, empirical correlations and equations of state, have been used to predict liquid density at various temperatures and pressures. Chapter 6 of the API Technical Data Book covers most of the empirical approaches, while any number of equations of state are available in the literature. Brule et al. (1982) have had good success with a modified Bene-
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
Table V. Summary of Measured Surface Tension Data 366 K 450 K fraction Ua Pb U P no.
5HC 8HC llHC 16HC 17HC 19HC-B 450/650
23.3 27.9 27.5 29.4 31.0 30.6
86.2 86.2 689 86.2 86.2 86.2
28.4
86.2
-
21.7 25.9 25.6 26.2 27.4 28.8 28.4 27.1
-
533 K P 1896 1379
U
21.5 24.6
1206 86.2 689 86.2 86.2 86.2 86.2 86.2
616 K
25.0 26.3 25.7 19.8d 23.3
c
18.5
86.2 86.2 86.2 86.2 689
672 K
P 5860 41 37
U
21.6 23.6 23.7 16.2d 17.9
417
1379 689 ' 689 86.2 1379
P
U
-
-
16.2
8274
18.5 21.2 20.5
2068 1379 689 86.2 6205
e
16.4
(I N/m2. kPa. 19.1 x lo-* N/m2 at 561 K;plugging occurred at higher temperatures. These two points are questionable because of measurement difficulties thought to be due to solid d-posits on the drop tip. e A point at 672 K was not possible due to coking problems.
,
130
,
,
,
,
,
,
,
I
,
I
,
,
0 POUR POINT
1 20
0 CUTSHC \
OZOL
' m
'
'
4W
'
' 5M
'
'
Mo
'
m'
'
' e a
'
'
8m
'
i: 0' %
01
A CUT11HC '\ 0 CUTlGHC \\ 0 CUT17HC 0 CUT 19HC-8 U R . 0552 ( l - T R ) --- U R - 1% (1-TR) 232 FROM WILSON (1981)
-
'\ '\
'
TEMPERATURE K
Figure 7. Density of coal liquid fractions vs. temperature.
03
dict-Webb-Rubin equation of state developed specifically for coal fluids,while Hwang et al. (1982) have had equally good success with a reduced density correlation. In the density prediction method proposed by Hwang et al. for coal liquids, the Watson and Nelson (19331, Riedel(19541, and Lydersen (1955) procedures are combined into a single correlation that was shown to predict the densities of coal liquids to temperatures of 730 K and pressures of 22.1 h4Pa with an average deviation of 1.4%. The liquid density data for most of the primary fractions listed in Table I are shown in Figure 7 as a function of temperature. None of the pressures exceeded 4.24 MPa during the measurements, and most were much lower in order to approximate a saturated liquid condition. Each set of data was fitted with the Rackett (1970) equation =A.~(~-TR)~/~
(7)
where A and B are constants for each data set derived from regression analysis. Critical temperature was estimated by using the correlation of Brule et al. (1982). The curves in Figure 7 represent eq 7 and are useful in extrapolating the data to higher temperatures. The absolute average deviation of the various data sets ranges from 0.02 to 0.22%. Because eq 7 is not general, the procedure of Hwang et al. (1982) was evaluated for accuracy in predicting the experimental data. The reduced density is calculated at a reference temperature and at the desired temperature by the Riedel equation p~
=1
06
07
08
REDUCED TEMPERATURE, T R
Figure 8. Reduced surface tension of coal liquid fractions vs. reduced temperature.
By use of the reference density (typically from the specific gravity at 288.7 K), the density at the desired temperature is calculated from the Watson and Nelson (1933) relationship P
= P*d(
2)
Equations 8 and 9 yielded an absolute average deviation of 0.80% over the measurement temperature range and a maximum deviation of -5.77 % . As would be expected, the deviation was least at the first point beyond the reference temperature and increased as the temperature increased. The densities of the heaviest fractions, cuts 12-16, agreed best with predictions. A slight modification of the coefficients in eq 8 with simultaneous solution of eq 9 should significantly improve the general method. 8. Surface Tension. Measured surface tension data on the six heart cut coal liquid fractions and a single 505-616 K wide boiling fraction are listed in Table V and plotted in Figure 8 in terms of reduced properties. Consistent trends with temperature are apparent, and the data for the wide boiling fraction overlap the other data as would be expected. The data correlation is excellent (less than 5% deviation) by using the simple equation UR
+ 0.85(1 - TR)+ (1.6916 + 0.9846~)(1- TR)1/3 (8)
05
04
where
UR
= 0.0552(1 - TR)o'4
(10)
is reduced surface tension, defined as
u/
418
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
0 311 K 0 366K
A 422 K
100
-
o 478 K V533K
I
i
10
50 WT % OFF TEMPERATURE, K
Figure 10. Viscosity of coal liquid fractions vs. boiling point.
01 1 0 3 1 ~K-'
Figure 9. Viscosity of coal liquid fractions.
(T,'J3P,2J3), u is surface tension in dyn/cm, T,is in O R , and P, is in psia. Equation 10 is represented by the solid line in Figure 8. The dashed line in Figure 8 represents the surface tension-temperature correlation recommended for light hydrocarbons and petroleum fractions by the API Technical Data Book. It is based on the equation of Wilson (1981), i.e.
where u is surface tension in N/m, Kw is the Watson characterization factor, and TRis the reduced temperature. Although most literature data on low molecular weight nonpolar compounds can be fitted using (1- TR)1.252 as the temperature variable, the coal liquid fraction data do not follow this relationship. Hwang et al. (1982) used eq 11 for predicting surface tension on wide boiling fractions with limited success. 9. Viscosity. Viscosity data on the primary coal liquid fractions are summarized in Figure 9 at temperatures to 505 K and at pressures approximating saturated liquid conditions. Calculation of the viscosity from the capillary tube pressure drop and flowrate required the density data in Figure 7. The viscosity data appear consistent and increase in a predictable manner with decreasing temperature and increasing fraction boiling point. The resulta are in general agreement with data published for EDS liquids. Figure 10 illustrates the same data (interpolated at 56 K intervals from Figure 9) plotted vs. 50 w t % temperature and clearly indicates the effect of average coal liquid fraction boiling point. The viscosities of several of the fractions in Figure 9 overlap, as would be expected, because their boiling ranges overlap. A good method of correlatingthe viscoeity of coal liquids is not yet available. Two methods of estimating the vis-
cosity of coal liquids were evaluated by Hwang et al. (1982), and a third method was proposed by Starling and coworkers (1980). One of the correlations evaluated by Hwang et al. was empirically developed for petroleum fractions, while the second correlation was based on the application of corresponding states concepts to a large amount of experimental pure component liquid viscosity data. Hwang et al. found that the petroleum fraction correlations predicted substantially higher viscmities than the experimental results for coal liquid fractions at temperatures below 590 K. Measurement temperatures for the coal liquids went as high as 700-755 K [Exxon (198011. The absolute average deviation was 45% for measurements in the absence of hydrogen and 28% under a high-pressure hydrogen atmosphere. The corresponding states correlation reduced the deviation by half. The correlation proposed by Starling and co-workers based on a three-parameter corresponding states concept was not any better than the latter correlation. Amin and Maddox (1980) have reviewed a large number of empirical equations for correlating liquid viscosity data with temperature that were found useful in extrapolating the data in Figure 9 to somewhat higher temperatures. 10. Thermal Conductivity. Thermal conductivity measurements were performed on eight coal liquid fractions at temperatures to 505 K and at a pressure of 5.62 MPa. The results are illustrated in Figures 11 and and 12 for the distillate fractions. The isotherms in Figure 12 are based on values interpolated from the data in Figure 11. The thermal conductivities of the coal liquid fractions exhibited unusual trends as shown in Figure 12, and were generally not in agreement with the petroleum fraction correlations. Thermal conductivity decreased with increasing temperature, as would be expected. However, unlike petroleum fractions, thermal conductivity increased with increasing specific gravity except for fractions 6,8, and 10, which gave results that overlapped somewhat. As Figure 12 illustrates, a region of maximum thermal conductivity exists near a b o i i point of 470 K corresponding
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983 419 Table VI. Specific Heat Data (J/kgK) fraction no. temp, K
2 1910 2020 2120 2230 2370 2470
29 8 32 3 348 37 3 398 423 448 47 3 49 8 523 548 57 3 598
4 2040 2130 2220 2300 2400 2520 2600 2740
8 1910 1980 2060 2150 2230 2320 2420 2520 2620 2750
6 1980 2080 2180 2280 2320 2430 2530 2640
10 1750 1840 1930 2020 2110 2210 2310 2420 2540 2670 2800
12 1700 1770 1840 1910 1980 2060 2130 2220 2300 2410 2510 2640 2800
14 1760 1840 1920 2000 2089 2160 2260 2350 2450 2560 2680 2800
15 1740 1810 1900 1980 2050 2120 2190 2260 2310 2340 2410 2550 2740
17 1660 1760 1840 1940 2020 2100 2180 2230 2260 2300 2400 2580 2760
19 1640 1740 1840 1920 2010 2090 2180 2220 2280 2330 2370
Table VII. Vapor Pressure Data for Secondary Coal Liquid Fractions
0140
-
1
1
1
t
~
1
1
0
I
I
0 140
0 0 0130
-
0 120
-
0 0 130
y
i
Y
3
€
E2 z
8
i 0110-
5
vm
0100-
0 0
0 .
om-
0080-
*CUT2 CUT4 0 CUT 6 A CUT8 0 CUT 10 D C U T 12 OCUTl6 0 CUT 18
0070-1
I
02
03
I
0110
El CI
8 3 J
0100
5 E
0
t: E
0120
0 090 0
0 0
I
I
04
05
I
I
OB
1
1
'
07
08
1
I
09
'
0 070
10
300
350
400
450
500
550
600
650
7M)
750
420
I d . Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
Table VIII. Heat of Vaporization (kJ/kg)for Secondary and Wide Boiling Coal Liquid Fractions cuts temp, K 366.5 422.0 477.6 533.2 588.7 644.3 699.8 755.4 a
5HC 341.2 312.8 296.4 238.6 179.6
----
8HC
llHC
16HC
--
--
17HC
19HC-A
450/650
--
--
--
296.8 276.7 257.9 224.4 175.4
--
299.4 279.2 257.1 234.9 210.6
650/950
---
--
266.7 251.2 224.4 196.4 142.4
272.1 252.5 225.7 211.8
Questionable point.
Table IX. Liquid Density Data for Primary Coal Liquid Fractions density, cut no. temp, K press., kPa lo3 kg/m3 cut no.
1
2
3
4
5
6
7
8
300.1 340.7 382.6 422.4 461.2 504.4 299.3 345.7 380.5 422.8 460.1 503.9 299.8 340.2 382.5 423.9 461.1 502.4 297.4 339.3 276.7 418.8 460.4 498.8 297.4 339.1 376.6 415.9 460.2 498.6 299.6 340.7 380.6 419.0 458.8 499.7 292.2 339.1 378.1 421.8 459.4 503.4 299.1 340.3 380.1 418.8 458.7 499.6
446.1 790.8 1136 1480 1825 2170 446.1 1136 1136 1480 1825 2170 446.1 790.8 1136 1480 1480 1825 446.1 790.8 1136 1480 1480 1825 446.1 446.1 446.1 790.8 790.8 790.8 790.8 1480 2170 2859 3549 4238 466.1 446.1 446.1 790.8 790.8 790.8 790.8 1480 2170 2859 3549 4238
0.71636 0.67704 0.63603 0.58968 0.53617 0.44542 0.74418 0.70893 0.67784 0.63806 0.59890 0.54565 0.75975 0.72694 0.69056 0.65387 0.61565 0.56793 0.80288 0.77282 0.74429 0.70489 0.66405 0.61922 0.88866 0.85644 0.82481 0.78729 0.74953 0.70862 0.94526 0.91495 0.88337 0.85289 0.81816 0.78021 0.95877 0.92949 0.90066 0.86833 0.83547 0.79398 0.97006 0.94008 0.91251 0.88493 0.85479 0.82232
creases linearly with increasing temperature and specific gravity. The API TechnicalData Book Figure 12A3.1 gives a graphical correlation of thermal conductivity vs. tem-
9
10
11
12
13
15
16
density, temp, K
press., kPa
296.7 341.4 378.5 421.1 457.8 490.2 295.6 341.8 380.7 420.6 456.9 496.3 339.5 374.9 402.8 437.5 459.7 499.2 340.7 368.5 414.7 456.9 492.9 340.2 368.7 415.3 456.6 494.0 303.1 338.3 380.9 420.0 459.7 499.4 303.1 338.4 381.2 418.8 461.9 497.4 302.8 338.9 349.2 381.2 419.7 461.4 498.6 302.2 338.7 380.4 419.5 460.2 498.5
790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 790.8 790.8 790.8 446.1 446.1 790.8 790.8 790.8 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1 446.1
lo3 kg/m3 0.97239 0.94392 0.91520 0.88725 0.86158 0.83108 0.99428 0.96688 0.94039 0.91373 0.88791 0.85729 1.00782 0.98384 0.96641 0.94288 0.92691 0.89928 1.04812 1.02754 0.99831 0.96781 0.94087 1.05920 1.03861 1.00979 0.98028 0.95191 1.07337 1.04957 1.01940 0.99195 0.96832 0.94054 1.07452 1.05186 1.02183 0.99324 0.96802 0.94024 1.09604 1.07575 1.06584 1.04501 1.01527 0.99194 0.96571 1.10146 1.07741 1.05066 1.02155 0.99624 0.97068
perature that is independent of specific gravity. A graphical correlation is also provided for pressure corrections at pressures above 3.45 MPa. Starling and co-workers
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983 421
Table X. Viscosity Data for Primary Coal Liquid Fractions viscosity, cut no. temp, K press., kPa MPa -s 1 ,
2
3
4
5
6
7
8
300.3 339.6 381.2 424.5 395.5 341.1 382.8 421.3 297.8 340.4 386.5 424.3 295.0 341.5 381.6 421.4 294.8 340.2 387.6 423.6 463.8 501.8 295.8 341.5 381.9 423.4 465.1 502.O 296.8 341.5 381.5 420.9 463.0 505.0 295.8 343.2 380.4 422.1 456.9 503.7
790.8 790.8 1480 1480 790.8 790.8 790.8 1480 446.1 790.8 1136 1480 790.8 790.8 790.8 1480 790.8 790.8 790.8 790.8 790.8 1480 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 1480 2170 2859 3549 4238
0.35450 0.25571 0.18521 0.13424 0.59991 0.37452 0.25226 0.20798 0.56880 0.36672 0.25478 0.19250 0.75028 0.44553 0.31264 0.24605 1.6841 0.79633 0.44784 0.32106 0.24175 0.20062 3.4439 1.3057 0.70837 0.45872 0.31348 0.23776 4.5858 1.5635 0.83440 0.53266 0.36048 0.25490 7.7350 2.2782 1.2294 0.74667 0.54274 0.38296
(1980) have derived a general thermal conductivity equation based on corresponding states principles and a large data bank based on pure compounds found in coal liquids. There were no thermal conductivity data available on coal liquid fractions at that time to evaluate their correlation. Although measurements were limited to a maximum temperature of 505 K, Figure 11 indicates that extrapolation of the thermal conductivity data to high temperature may be possible, because most of the data are grouped in a rather narrow band when plotted vs. reduced temperature and extend over the range 0.3 < TR< 0.9. Critical temperature was estimated from the correlation of Brule et al. (Table 111). The data for cut 6 lie above the other results, probably as a result of the high organic oxygen and dissolved water content of this fraction compared to the other fractions. Thermal conductivity measurements were also performed on recycle slurry and coal feed slurry samples from SRC-11processing of Powhatan No. 5 Mine coal. The data (included in Tables W-XI) show that both slurry samples have about the same thermal conductivity and that the values are significantly higher than those obtained for the distillate fractions. The ash content of the slurry appears responsible for these observations. The recycle slurry contained 8.99 w t % ash compared to 8.65 wt % in the coal feed slurry. Slurry thermal conductivities reported by Droege et al. (1980) are somewhat lower than the above
viscosity, cut no.
temp, K
press., kPa
9
296.8 340.6 382.6 422.6 463.9 504.8 341.4 381.9 422.2 462.8 500.2 296.9 342.2 383.8 424.2 504.0 342.6 426.3 464.5 506.9 346.8 342.9 380.2 420.4 459.9 496.3 340.4 382.3 423.3 463.1 501.5 341.9 379.6 420.4 460.2 500.0
1480 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 446.1 790.8 790.8 790.8 790.8 4238 4238 4238 4238 2859 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 790.8 5617 5617 5617 5617 5617
10
11
12
13
15
16
MPaes 7.9321 2.3230 1.1421 0.71945 0.47894 0.36320 3.1902 1.5161 0.89413 0.57978 0.40763 72.769 8.7558 3.0455 1.6645 0.81819 21.007 2.0987 1.2207 0.75116 17.039 39.145 7.2242 2.7561 1.3837 0.90457 14.700 4.0711 1.7477 1.0129 0.66059 53.460 8.7925 3.1453 1.5671 1.0300
results, probably because leas ash was present in the slurry. 11. Specific Heat. Specific heat measurements on the 19 primary coal liquid fractions were attempted at temperatures up to their normal boiling points or the onset of decomposition, by use of a differential scanning calorimeter (DSC) as described by Smith et al. (1980) and O’Neill (1966), except that aluminum sample pans were used. This method gave erroneous results because of sample vaporization and loss from the pan. Therefore, the DSC method is not recommended for specific heat measurement, unless appropriate design changes are made to overcome the problems. The specific heats of ten fractions, measured in a conventional calorimeter at approximatkly saturatkd conditions, are listed in Table VI. The specific heats of cuts 14,15, and 17 overlap somewhat the values for cuts 10 and 12, as would be expected, because of the boiling range overlap. Surprisingly, the specific heat of cut 2 is lower than for cut 4. It is also lower than cut 6 at temperatures below 387 K. This could be a result of cut 2 being composed predominantly of a few compounds which do not follow the trends for the other cuts exactly. Cuts 15,17, and 19 all exhibit a phase transition region starting at about 478 K, as evidenced by a slope change in the specific heat-temperature plot. This is probably a result of finely stratified crystalline material going into solution, a typical phenomenon observed previously when distilling narrow
422
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983
Table XI. Thermal Conductivitya of Coal Liquid Fractions and Slurries thermal cond., cut no. temp, K W/m-K cut no.
-
2
4
6
8
10
302.1 347.9 395.5 438.9 475.1 510.4 302.0 348.2 395.5 440.2 481.6 510.5 305.6 347.6 386.6 428.9 476.7 510.5 302.8 347.3 381.8 436.8 471.8 512.2 301.7 349.0 394.8 430.6 474.0 511.1 302.4 347.6 409.6 436.2 471.6 512.2 300.5 349.0 394.9 430.2 473.4 510.9 297.5 352.8 402.6 449.5 482.3 509.0
0.1168 0.1003 0.09472 0.08732 0.07992 0.07712 0.1199 0.1092 0.1003 0.09066 0.08548 0.08304 0.1370 0.1293 0.1225 0.1141 0.1046 0.1005 0.1338 0.1281 0.1208 0.1126 0.1054 0.1022 0.1294 0.1220 0.1164 0.1107 0.1054 0.1021 0.1284 0.1241 0.1175 0.1103 0.1039 0.1039 0.1284 0.1221 0.1139 0.1097 0.1050 0.1029 0.1287 0.1213 0.1135 0.1104 0.1064 0.1039
12
16
18
recycle slurry b
coal feed slurry
temp, K
thermal cond., W1m.K
298.4 344.7 398.4 428.6 464.9 512.2 299.9 348.9 394.8 430.8 473.4 511.1 298.3 345.9 396.5 440.8 469.6 511.6 466.4 510.8 433.3 356.9 391.2 363.2 493.0 404.3 359.7 402.9 440.0 475.3 509.1 359.7 402.6 440.3 475.2 509.5 364.0 397.3 423.2 453.0 432.0 506.2 490.5
0.1287 0.1255 0.1198 0.1186 0.1154 0.1106 0.1256 0.1233 0.1182 0.1142 0.1122 0.1103 0.1334 0.1296 0.1237 0.1211 0.1130 0.1180 0.1153 0.1137 0.1247 0.1315 0.1321 0.1304 0.1168 0.1283 0.1585 0.1532 0.1485 0.1406 0.1374 0.1574 0.1559 0.1497 0.1448 0.1424 0.1588 0.1546 0.1471 0.1505 0.1440 0.1449 0.1443
a All measurements performed at 5617 kPa. Specific gravity = 1.246 (294 K/294 K), 17.1 wt % pyridine insolubles, 52.6 wt % ash on pyridine insolubles. Specific gravity = 1.270 (294 K/294 K), 35.6 wt % pyridine insolubles, 24.3 wt % ash on pyridine insolubles.
boiling fractions. Cut 19 was a solid when charged to the cell and probably melted a t a temperature below 372 K. The data were compared to specific heats calculated from a modified Watson-Nelson correlation developed by Exxon (1980) and to a graphical correlation developed by Gulf using the data of Kidnay and Yesavage (1980) and other unpublished Gulf data. The values predicted by the former correlation gave an overall absolute average deviation of 4.3% and a bias of 2.590, while the latter estimates had an absolute average deviation of 3.4% and a bias of -2.2%. In both cases, cut 2 specific heat was predicted with the least accuracy of all the fractions with absolute average deviations of 6.1% and 6.7%, respectively. Conclusions Chemical analyses on the 19 coal liquid fractions demonstrated the vastly different nature of coal liquids compared to petroleum fractions. Heteroatom content is very
large, especially in the boiling range 425 to 550 K, where organic oxygen content is as high as 6.0 wt % Phenolic compounds were shown to account for most of this oxygen. Water solubility in these samples is also very high, as much as 2.7 wt 7'6 a t saturation, indicating that increased water solubility is a result of association with phenolic groups. This has important implications in the design of fractionation equipment in which the water will be liberated and condensed overhead resulting in larger than usual condenser heat duties. Much more data are needed on water solubility as a function of temperature. Basic nitrogen reaches levels of 0.8 to 1.0 wt % in some fractions, confirming a severe environment for catalytic upgrading. Distillate fractions boiling above 430 K contain more than 50% aromatic compounds, and this reaches 90% by a boiling point of 475 K. Polar aromatic compound concentration attains a maximum in the same boiling point range as does organic oxygen. The significant composi-
.
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 3, 1983 423
tional changes in the 425 to 550 K boiling range influenced the specific gravity and thermal conductivity of the fractions in this region. An equation was developed for calculating molecular weight from boiling point and specific gravity that fits the experimental data well and is in good agreement with other published data. However, available methods for estimating critical properties and acentric factor for coal liquids yield significantly different values for these parameters. More work needs to be done on obtaining a consistent set of parameters that can be used in a variety of thermodynamic correlations. Vapor pressure results were in good agreement with predictions using a modified Grayson-Streed (1963)correlation, indicating that this method is adequate for coal liquids. Several additional fractions should be checked, however, in the boiling range where organic oxygen is most concentrated. Heats of vaporization can be estimated from the slopes of the vapor pressure curves, offering an alternative calculation procedure to empirical correlations. Liquid density measurements yielded results in good agreement with correlations developed for other published data for coal liquids. Surface tension and thermal conductivity data were drastically different from petroleum fraction correlations and these data need to be confirmed. Thermal conductivity increased with increasing fraction specific gravity and increasing boiling point, which is just the opposite for petroleum fractions. Viscosity data are comparable to results published for EDS products and exhibited the expected trends. However, available petroleum fraction viscosity correlations and pure compound correlations appear to have only marginal applicability to coal liquids, and a new, more accurate, correlation method is needed. Specific heats of coal liquid fractions can be accurately calculated by use of petroleum correlations that have been modified for application to coal liquids. The above property data were obtained on coal liquids produced from a high volatile bituminous coal from the Pittsburgh seam that is typical of coal used in the SRC-I1 process. Other types of coals may yield liquid products with somewhat different properties. This work should therefore be extended to low rank coals in order to assess the magnitude of the differences. Tabulations of vapor pressure, heat of vaporization, liquid density, viscosity, and thermal conductivity data are given in Tables VII-XI. These results should be useful to other investigators in developing new correlations and modifying existing ones. Acknowledgment The density, viscosity, and thermal conductivity measuremenh were performed by Drs. Robert Maddox and Ruth Erbar at FPRI. The specific heat measurements were performed by Mr. Andrzej Brzezinski at Dynatec R/D Company. Evaluation of the DSC method for specific heat measurement was done by Dr. Chaur Wen a t Gulf. The guidance of R. G. Sperhac and D. M. Jewell of Solvent Refined Coal International, Inc., is kindly acknowledged as well as the support provided under the U.S. Dept. of Energy Contract No. DE-AC01-79ET10104. Nomenclature A , E = constants in Rackett equation (eq 7)
Po) = simple fluid term
in vapor pressure equation (eq 3) correction term in vapor pressure equation (eq 3) AHv = heat of vaporization, J/mol (eq 4) AHw = heat of vaporization, kJ/kg (eq 6)
f(l) =
Mmp = heat
of vaporization at normal boiling point, J/mol
(eq 5 )
A H w p = heat of vaporization at normal boiling point, kJ/kg
(eq 6) = (1.8Tb)’/3/sG;Watson characterizationfactor (eq 11) MI = molecular weight Po = vapor pressure, kPa PRO = reduced vapor pressure, P / P C P, = critical pressure, MPa R = gas constant SG = specific gravity, 288.7 K/288.7 K T = temperature, K Tb = mid boiling point of distillate cut, K T, = critical temperature, K TR = reduced temperature, TIT, V, = critical volume, m3/mol 2, = critical compressibility factor A2 = compressibility factor of vapor minus that of liquid at a given temperature and vapor pressure (eq 4) Greek Letters y = orientation parameter as defined by Starling et al. p = liquid density, kg/m3 pref= liquid density at reference temperature (usually 288.7 K), k / m 3 (eq 9) pR = reduced liquid density, pV,/MI PR,ref = reduced liquid density at reference temperature (eq
Kw
9)
u = surface tension, dyn/cm or N/m (eq 10, 11) UR = T;l3 Pc’l3);reduced surface tension in eq
(41
10 where
is in dyn/cm, T, in OR, and Pc in psia (eq 10) w = acentic factor (eq 3) Registry No. Water, 7732-18-5. Literature Cited u
American Petroleum Institute “API Technical Data Book-Petroleum Refining”, 3rd ed.; Washington, DC, 1976; Vol. I and 11. Amin, M. B.; Maddox, R. N. Hyckocerbon Process. 1980, 59(12), 131. Bruie, M. R.; Lin, C. T.; Lee, L. L.; Starling, K. E. AIChE J. 1982, 28. 616. Chung, K. E.; Anderson, L. L.; Wiser, W. H. Fuel 1979, 58, 847. Droege, J. W.; Stickford. G. H.; Chauhan, S. P. “Thermophyslcal Properties of Coal Liquids”, 3rd Quarterly Technical Status Report, Apr l-June 30, 1980 US. DOE Contract No. AC01-79ET14941, Report No. BMI-2063, June 1980. Exxon Research and Engineering Co. “EDS Coal Liquefaction Process Development”, Phase IV Quarterly Technical Progress Report for the Period Oct l-Dec 31, 1979; U.S. DOE Contract No. EF-77-A-01-2893, FE2893-45, Apr 1980. Gray, J. A. “Selected Physical, Chemical and Thermodynamic Properties of Narrow Boiling Range Coal Liquids from the SRC-I1 Process”, Report for the Period Mar 1960-Feb 1981, US. DOE Contract No. DE-ACO179ET10104, Apr 1981. Grayson, H. G.; Streed, C. W. “Vapor-Liquid Equilibria for High Temperature, High Pressure Hydrogen-HydrocarbonSystem”, 6th World Petroleum Conference, Frankfurt/Main, Section VII, Paper 20-PD7. 1963. Grassman, P.; Straumann, W.; Wldmer, F.; Jobst, W. “Measurement of Thermal ConductivHy of Liqulds by an Unsteady State Method”, Progress in International Research on Thermodynamic and Transport Properties, ASME, Academic Press: New York, 1962. Hwang. S. C.; Tsonopouios, C.; Cunningham. J. R.; Wilson. G. M. Ind. Eng. Chem. PrOcessDes. D e v . 1982. 21, 127. Kern, D. Q. “Process Heat Transfer”; McGraw-Hili: New York, 1950; p 803. Kidney, A. J.; Yesavage, V. F. “Enthaiphy Measurement of Coal Derived Liquids”, Final Report for the Period June 1975-Mar 1979, U.S. DOE Contract No. EX-764-01-2035, Jan 1980. Lee, B. I.; Kesier, M. G. AIChE J . 1975, 21, 510. Lydersen, A. L.; Greenkorn, R. A.; Hougen, 0. A. University of Wisconsin Engineering Experiment Station Report No. 4, Madison. WI, Oct 1955. McIivried, H. G.; Gail, W.; Tsai, S. C. “SRC-I1 Processing of Pbburgh Seam (Powhatan No. 5 Mine) Coal in Process Development Unlt P-99”, Interim Report for the Period Dec 1978-June 1979, U.S. DOE Contract No. DEAC01-79ET10104, FE1496, Apr 1980. O’Neiii, M. J. Anal. Chem. 1988, 38, 1331. Perry, J. H., Ed. ”Chemical Engineers’ Handbook”, 4th ed.;McGraw-Hili: New York, 1963; pp 3-225. Rackett, R. 0. J. Chem. Eng. Deta 1970, 15, 514. Recon Systems, Inc. “Fundamental Data Needs for Coal Conversion Technology”, Final Report No. TID-28152, U.S. DOE Contract No. EY-76C-02-4059, Jan 1981. Riazl, M. R.; Daubert, T. E. Hydrocarbon Process. 1980, 59(3). 115. Riedei, L., Chem.-Ing.-Tech. 1954, 26, 259. Smith, J. M.; Van Ness, H. C. “Introduction to Chemical Engineering Thermodynamics”, McGraw-Hill: New York, 1959; p 132. Smith, N. K.; Lee-Bechtoid, S. H.; Good, W. D. Report No. TPR-79/2, Bartiesviile Energy Technology Center, US. Department of Energy, Jan 1980. Starling, K. E.; Bruie, M. R.; Lin, C. T.; Watanasiri, S.; Ajlan. M. H.; Chung, T. H.; Kumar, K. H.; Lee, L. L.; Li, M. H. “Coal-Caic Probct Report to De-
424
Ind. Eng. Chem. Process Des. Dev. 1983, 22, 424-429
partment of Energy, Coal Conversion Systems Technical Data Book Project", Institute of Qas Technology, Report OUIIGTIS-1436&1, published by the School of Chemlcal Engineering and Materials Science, Universlty of Oklahoma, Norman, OK, Aug 1980. Suatoni, J. C.; Swab, R. E. J . Chromatogr. Sc/. 1076, 14, 535. Swansiaer. J. T.: Dlckson. F. E.; Best. H. T. A n d . Chem. 1074. 46. 730. Wa"l, Suphat. Bule, M. R.; Starling, K. E. A I C h E J . 1082, 28, 626. Watson. K. M. Ing. Eng. Chem. 1031, 23, 380. Watson, K. M.; Nelson, E. F. Ind. Eng. Chem. 1033, 25 880.
Wilson, 0. M.; Johnston, 0. H.; Hwang. S. C.; Tsonopoulos, C. Ind. Eng. Chem. Process Des. Dev. 1081, 20. 94. Wilson, G. M. In "Fowrdatkns of Computer-Aided Chemical Process Design", Mah. R. H.; SeMer, V. E., Ed., Engineering Foundation: New York, 1981; Vol. 2, p 31.
Received f o r review February 5, 1982 Accepted October 12, 1982
Thermaphysical Properties of Coal Liquids. 2. Correlating Coal Liquid Densities Gerald D. Holdert and James A. Gray' Gulf Research and Development Company, Piftsburgh, Pennsylvania 75230
An evaluation of several correlations for predictlng saturated densities of coaklerhred liquids from boillng point and specific gravity indicates that fhe best correlatlng equation is a one-constant modification of the Rackett equation given by p, = 0.2841241-rn)2' . This correlation predlcts SRC-I1 liquid densitles with a 0.4% average error and EDS liquid densfies with a 1.1% average mor. The correlations appear to be valid to at least 95 % of the critical temperature. A new correlation for predicting specific gravity from the boiling point allows the liquid densities to be predicted from the boiling point alone.
Introduction Accurate determination of the density of liquid streams from coal conversion plants is important in any process design. Many methods have been developed for estimating densities of well-characterized hydrocarbons (Chiu et al., 1973; Lu et al., 1973; Rackett, 1970; Riedel, 1951; and other methods reviewed by Reid et al., 1977),but little has been done in determining the suitability of these methods for predicting the density of coal liquids. In the present study, several methods for estimating the densities of coal liquids are examined and compared. The primary criteria for making judgments about the suitability of a particular correlation are: (1)the ability of the method to predict densities accurately from a minimum of experimental characterization; boiling point, Tb,and specific gravity, SG, should be sufficient; (2) the generality, or the ability, of the correlation to predict the densities of a wide variety of coal liquids; (3) the ability of the correlations to extrapolate to densities near the critical density. With these criteria as a guide, a wide variety of correlations are examined. Data The correlations were evaluated by using pycnometer density data obtained for narrow boiling coal liquid fractions. The fractions were distilled from full boiling range product generated from SRC-I1 processing of Powhatan Mine No. 5 (Pittsburgh Seam) coal in the Gulf 1ton/day process development unit. The liquids have boiling ranges spanning over 11K (20 O F ) to 99 K (179 O F ) intervals with nominal boiling points, Tb,ranging from 340 K (152 OF) to 676 K (757 OF). Detailed descriptions of the experimental procedure used in obtaining these cuts and their 'Chemical and Petroleum Engineering Department, University
of Pittsburgh, PA 15261.
0198-4305/83/1122-0424$01.50/0
physical and chemical makeup are given by Gray et al. (1983) and Gray (1981), but a few of the properties important to the present study are given in Table I. Values for the boiling point are experimental 50 wt % off temperatures or the temperature in an atmospheric distillation where one-half of the cut is distilled. The molecular weight and specific gravity are experimental in that they are based upon correlations for smoothing actual experimental data for each cut, but the critical properties, acentric factor, and orientation parameter are based solely upon the correlations of Brule et al. (1982) and Wilson (1981). The densitylspecifw gravity values at 288.7 K were obtained by smoothing the density data for each cut using a three-constant Rackett equation and extrapolatingto this temperature. This technique appeared to give more accurate values of specific gravity than the API hydrometer measurements reported by Gray et al. (1983). The experimental data used in developing the correlations below were reported in Table IX of Gray et al. (1983). The pressure at which the densities were measured was that necessary to keep the fraction entirely in a liquid state at the experimental conditions, and all densities are taken as saturated liquid densities. Correlating Parameters The correlations used in this study require that the specific gravity and boiling point of the fluid be known, although, as shown later, the specific gravity alone can be used if necessary. These properties are then used to estimate the critical temperature using the correlations developed by Wilson (1981) or by Brule et al. (1982); the two critical temperatures given for each cut in Table I are so obtained. For some correlations, a second parameter, either the acentric factor (Wilson) or the orientation parameter (Brule) is needed. This second parameter is also derived from the specific gravity and boiling point. Values for the 0 1983 American Chemical Society