Molecular Weight-Physical Property Correlation for Petroleum Fractions Seberal currelatioils of moleciililr tceiglit w i t h phjsicaai properties a r e presented which co\ er the ordinary range of p e t r o l e u m fractions. For fractions of '70-300 molecular weight (gasolines, herosenes, light 1ubrica: iiig oils) boiling point a n d gravity are employed j t h e correlation gives good results for pure hydrocarbon4 of t a r i o u s types (aterage deviation 2.4% for 131 eompoiiiicls) as well as for petroletun cuts. For l u h r i c a ~t ~ fractions of 240-700 molecular weight correlations a r e described for viscosity w i t h gravity a n d for viscosity a t 100' w i t h viscosity a t 210' F.; the f o r m e r a r e n o t , i n general, applicable to p u r e hydrocarhons but g i i e good results for most lubricant
fractions, Mith t h e exception of Gulf Coast or CaiiluriitGi distillates above 350 molecular xceight. With the use of a correction, t h e viscosity-gratity correlations give goo*! results for t h e s e oils also. 'I'he correlation of viscositj a t 100' w i t h viscosity a t 210' ic i n approximate agreemciit with t h e d a t a for pure hjdroc-arboils of various typcs nn{l gixek good results for all t j p e q of' p e t r o l e u m fracliorrfor which d a t a a r e available. For p e t r o l e u m waxes nit1 I ing point a n d r e f r a c t i \ e inde\ a1 80' C. a r e eniplojecl. t h e correlation is i n good a g r e e m e n t with t h e d a t a r o t t h e n-paraffins as well : I & for p e t r d c u r n w a x e q of t C ~ r i o i i types.
T
typc prescrited by Smith (Figure I ) . Averagc molecular wcigiii p0int.s were spotted and t'hc lincs ol constant molecular vxiglit \yere drawn which would agroc \rc!ll lor most of the homologow series. The biggest deviation is found with t8henormal paraflitit> n.hich always boil considerably higher than the average for atiy grctiip of paraffin isomers. %'lie rno1t:cular weight lines :iw d to agree with tho average paraffin data; thercforc: ( k i t s correlation is not quite 80 good for normal paraffins as for 1nos1
HE: experimental determination of molccular weight is time
~
consuming and requires ratlicr highly trained personnel. Therefore, it is desirable to obtain molecular w i g h t s for petroleum fractions from correlations with othrr physical propertics which can be more easily and accuratcly dctc:rmincd. The correlations presented here arc of thrce typcs which cover t,he ordinary range of petroleum fractions: ( u ) niolecular weight'boiling point-dcnsity for gasolinc,s, kerosencs, and light oils (molecular weight range 70-300) ; ( b ) molecular weight-viscosity-density and molecular weight-viscosity a t 100 O F.-viscosity at 210" for oil fractions (molecular weight range 240-680); (e) molecular weight-refract.ive index-melting point Fni petroLeum waxes (molecular weiglit rnngc 240-560). \.IO I,I.:C[JLAR WEIGHT-BOILING POINT-DENSITY
T h e correlat,ion of molccuhr weight with specific gravity anti boiling point is based on a curve presented by Smith ($?), which s h o w tho relation betwcen reciprocal ah3olute boiling point and spccifir gravity for the various homologous series. Smith's work was primarily coucerncd with obtaining a coiistitution index for crude oil fractions. The relation of this graph to molec-, ular weight, however, is quitc clear. Watson (Sf) presented a correlation of boiling point and densiby with molecular might, but the derivation and statistical evaluation were not givcn. The boiling point-dciisity correlat,iondeveloped in this work is, for the most part, in close agreement with that of Watson. Our correlation, however, is bnsed on truc boiling points, and curves are given for correcting observed boiling points obtaincd in thr A.S.T.M. distillation. In our development of the correlation, pure compound data (2, 3, i?9) were plotten on a graph of the
3.4
3.0
n
-0x
2.5
: I-
-
z P 0 W
L 0 2.0
m
15
I
other compounds. The abovc method of spottiug constant molecular w i g h t linw n.ns adequate up to 150 molcoular weight; however, thc reliability of general pure compomid dnta above this xnolecului tho following mctluiii wc.iylit is somewhat doubtful; t~lic~rcl'orc
INDUSTRIAL A N D ENGINEERING CHEMISTRY
April, 1946
443
handle, the graph was shifted over to a boiling point-density plot aa shown in Figure 3. Since the correlation is based on pure compound data, the temperature coordinate is true boiling point at 760 mm. I n using the correlation, it is therefore necessary to correct boiling points (50% evaporated point) determined with A.S.T.M. distillation equipment (Procedure D86-40). By applying a n emergent stem correction t o the boiling point, a satisfactory approximation of the true boiling point is obtained where the barometric pressure is in the range 750-770 mm. , (Pressure correction should be applied if pressure is above 770 or below 750.) These correctiom for A.S.T.M. distillations may be obtained from Figure 4. Both curves are based on a room temperature between 77" and 90' F. and were calculated by the following equation: C = K n (T - t ) where C = correction, O F. K 0.00009 n = number of degrfes of exposed stem T = temp. of bulb, F. t = temp,at mid-point of exposed stem
-
..
MOLECULAR
Figure 2.
WEIGHT
Reciprocal Boiling Pbint us. Molecular Weight
Experimental data for temperature of this mid-point of the exposed stem were found t o be in essential agreement with data obtained in another laboratory. The accuracy of Figure 3 depends somewhat on the law of averages since it is based on the average physical properties for mixtures of isomers, If it is used on a very narrow cut in which there may be a high concentration of one isomer, more variation may be encountered than with mixtures of a considerable number of isomers. This point is illustrated in Table I which gives data for the C, paraffins and naphthenes. Table I1 shows the agreement of Figure 3 for various classes of pure hydrocarbons. The normal paraffins, aa previously stated, do not agree particularly well with this correlation, since it was deliberately set to agree with average paraffin data. For the whole group of 134 pure hydrocarbons, the average deviation of the calculated molecular
was used. The relation presented in Figure 2 shows that, if reciprocal absolute boiling point is plotted against molecular weight for the normal paraffins, a smooth curve is obtained (line A ) . If the data for the noncondensed CSring naphthenes without side chain are plotted, a curve equidistant from 'that for the n-paraffins is obtained (line B). If the intersections between the constant molecular weight lines and the normal paraffin line (from Figure 1) up to 150 molecular weight - are plotted on this graph the resulting curve I S also equidistant from the actual n-paraffin line (curve C ) . It DENSITY OC. was assumed that the same equidistance would hold above 150 molec0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 , 1.05 I I I I I I I I 7 ular weight, and line C was extended 4001 I on this basis. Using B and C , the intercepts of - 700 the lines of constant molecular weight with curves 1 and 7 of Figure 1 were established above 150 molecular - 600 weight. By asing curve C (Figure 2) rather than A in establishing the lines of constant moleoular :weight - r ; the same deviation is assigned to the n-paraffins above and below 150 mo500 c lecular weight. 5 0 The two series of compounds were - a . selected because they are far enough -400 apart in Figure 1to establish the slope of the constant molecular weight d 0 lines, and because their data were m considered most reliable in this mo300 ;C (ecular weight range. The data €or 0 u) the naphthalene series (curve 6, Figure 1) were used to establish the molecular weight lines in the high - 200 density range. These data indicated t,hat some curvature had to be applied to the molecular weight lines to fit this series, and they were drawn SO that - 100 - 80 the curvature occurs for the most I I I I I 0.75 0.85 0.95 1.0 part beyond curve 7. SPECIFIC GRAVITY 60160 OF. Since reciprocal absolute temFigure 3. True 50% Boiling Point, Specific Gravity, and Molechlar Weight perature is an awkward term to
9
. -
-
INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY
444
TABLEI. ACCURACY OF BOILINGPOINTGRAVITY-MOLECULAR L f T CORRELATIOK ~ ~ ~ ~ FOR ~ C? PARAFFIN ASD NAPHTHENE ISOMERS Calod.
Compound (Mol. Kt.)
B.P., SP. Gr., O
Paraffins (100) 2 ,!&Dimethylpentane 2 4-Dimethylpentane 2:2,3-Trimethylbutane 3,3-Dimethylpentane 2,3-Dimethylpentane 2-1RIethylhexrane 3-hIethylhexane 3-Ethylpentane n-Heptane
C.
79 81 81 86 90 90 92 93 98
d:'
0.6736 0.6729 0.6900 0.6932 0.6951 0.6787 0.6870 0.6984 0.6839
81 87 91 ng) 92 ing) 99 100 101
%'t. 96 97 95 97 100 102 102 101 107
Methylcgclohexane Ethylcyclopentane
0.7101 0.7547 0.7479 0.7495 0.7718 0.7707 0.7610
ING TO A.S.T.M. PROCEDURE
Deviation, %
MOL
-4 -3 -.5
-3 0 4-2 4-2
$;
Average deviation
Naphthene (98)
3
92 92
-6 -6 -3 -2 -1 0
95
96 97 98 99
i-1
Average deviation
2.7
TABLE 11. ACCURACY OF ROILIKGPOINT-GR-~VITY CORREI,.+ TION FOR VARXOGS HYDROCARBOX TYPES No. of ComType Compound pounds n-Para ffin s 14 20 Isoparaffins Monocyclic naphthenes Low mol. w t . (Ca, cs,Ce rings) 14 High mol. art. (Cs rings) 8 Alkylbenzenes 12 Noncondensed dicyclic naphthenes ( C S ,CGrings) 9 Naphthalenes Tetralins Decalins 8 Noncyclio mono-olefins 15 14 Cyclic mono-olefins Total 134
-
hlol. Wt. Range 100-282 100-184
Av. Deviation, % 5.7 2.7
Deviation of Av., 74 +5.7
98-154 140-252 92-190 138-236
3.3 2.2 4.0 4.4
-2.6 +l.9 -3.9 -1.9
138-278 98-224 96-166
3.1 2 .6 2.4
-2
3
-1-1.8 + A 0
TABLE111. COMPARISON
O F BOILING POINT--GRAT'I'TY AND EXPERINENTAL MOLECULAR WEIGHTDATAOF FITZSIMONS AKD
THIELE('7)
228
231 232 235 238 242 246 249 250 300 305 306 307 309 810 315 330 350 a55
Exptl. hlol. Wt. 78 89 101 106 116 131 143 153 173 9B 127 165
179 188 203 232 257 266 141 118 187 266 241 239 222 205 193
Deviation, To Calcd. hIol. W t . +6.4 83 91 f2.2 104 f3.0 110 +3.8 119 +3.5 133 4-1.5 146 +1.3 157 +2.6 179 +3.4 105 +9.4 130 i3.4 173 +4.8 187 4-4.5 196 +4.2 219 +7.8 i-2.2 237 262 f1.9 267 -3.4 140 -0.7 120 f- 11 ..17 185 240 -9.7 250 -0.4 227 -5,o 223 +0.5 213 1-3.9 +2.1 197 Average deviation 3,s
200 300 TEMPERATURE
400 'F.
500
600
Figure 4. Stem Corrections for A.S.T.RI. Low- and High-Distillat i o n Thermometers
-2.1
\wights is 2.4%. Table I11 shows the agreement of Figure 3 with the experimental molecular weights of Fit'zSimons and Thiele ( 7 ) for twenty-seven samples ranging in m o l e c u h weight from 89 t o 266. The average deviation for this group is 3.570. R~OLECULAR ITEIGHTS FROJI P O D B I E L N I A K DISTILLATION.I n connection with design of stabilization units and similar work, it is frequently necessary for engineering purposes to have the moiecular TTeight of a mixture which may be about 50% c, and
Sample No. 212 2 13 215 216 218 221 223 225
100
lighter and about SOYc C, and heavier. Materials of this class are too volatile for determination of molecular weight by the methods usually available. W t ' h samples of this type a low temperature Podbielniak distillation is ah-ays run to obtain th(3 composition of the lower boiling fractions. Lsi~ally20-30 ml. of residue from the Podbielniak distillation can he obtained if a large distilling bulb is used. The molecular weight of the residue can be det'ermined from its boiling point arid gravit,y. Gravity is obtained v i t h a small hydrometer or a Sun Oil Company precision pycnometer (I:). The boiling point can be obtained with the setup sho\vn in Figure 6. This is similar to tkle con\,entiona] A.S.T.31.distillation apparatus (D-86-40) except that a, 50-ml. flask and a smaller condenser are used. The ice bat,h car1 be made from a 1-gallon can havilig a 4 X 6 inch cross section. Table I V presents the agreemellt bet,Tveen standard 100-ml. A,S.T.&f. distillations and distillations carried out in the in&cat,ed apparatus \Tit11 25-1nl. and 15-mi. charges. These data shox that the 5070 evaporated points of the small distillations and the standard A.S.T.M. distillations agree fairly viell. 'VVhile the use of the 50% recovered point instead of the recommended 50y0evaporated point does not, cause much error in the case of the st'andard -4.S.T.X. distillation, for the small distillations the latter should ala-ays be employed. Using the 50yo evaporated point and correcting as usual for emergent stem, niolecular weight's 17-ere found t o agree fairly well with experimentally determined molecular x-eights as indicated in Table 1'. MOLECULAR i y E I G H T O F ' r R A M F O R I \ I E H OILS. The boiling pointdensity correlat,ion can also be applied t o transformer oils and similar light oils since Engkr distillations are normally r u ~ i on such oils. These oils are of particular interest, since their viscosities are high enough for viscosity-density correlations to be used also. Table T-I gives data for sixteen light oils showing an average deviation of 4.1 yo for the boiling point-molecular weight correlation. Five of these n-ere reported b y E'enske (1;) and four by FitzSimons and Thiele ( 7 ) ; seven n-ere determined in our laboratories. The molecular weights reported from this laboratory were determined cryoscopically in benzene, i n the absence of a drying agent. Subsequently, on the basis of a limited amount of data, i t was found that molecular weights determined in the absence of a drying agent averaged 5 r o loner than those in the presence of a drying agent. Consequently, the data have been corrected for the effect of moisture in the oldw
~ ? y& ; : ~ 1: ~ ~ 2.9
voi. 38, N ~ 4.
INDUSTRIAL AND ENGINEERING CHEMISTRY
April, 1946 ~~~
~~
~
445
~~~
~~
OF 15- AND 25-Cc. DISTILLATIONS WITH A.S.T.M. 100-Cc. DISTILLATION (D86-40) TABLE IV. COMPARISON
East Texas Light Gasoline 100-cc. 25-cc. Distn., O F. Initial 5% over
80%
94 120 133 155 170 184 196 207 218 229 3
92 114 129 153 171 186 197 210 221 238 8
Podbielniak Reaidue 25-0~. 15-cc. 15-cc. (A) (B) 122 148 160 176 189 203 214 229 248 275 2
122 141 156 176 189 201 216 231 250 280 2
121 140 157 174 189 202 215 229 248 280 3
Composite of Podbielniak Residues .lOO-cc. 25-cc. 15-cc. (A) 130 164 176 191 206 220 236 253 276 300 2
128 152 166 182 197 214 233 251 274 299 2
Loss, % SO%,evapd. pointa O F . 192 188 212 213 211 233 229 Dev. fro& A.S.T.M., F. -4 +le -1) -4 5 Assuming distillation curve to be straight line between 40 and 50%. b Deviation from 25-cc. distillation.
.. .
...
...
cryoscopic procedure by adding 5% to the experimentally determined values. While occasional deviations of 8-9% are found; in general, they are 5% or less. In summary, the boiling point-gravity correlation is believed to be satisfactory for determining molecular weight for any petroleum fraction of 70-300 molecular weight for which a good 50% boiling point can be obtained. Obviously, it should not be applied t o unusual blends, such as a solution of asphalt in spirits, to cite an extreme example. The crowding of the lines of constant molecular weight as molecular weight is increased, as well as increased difficulty in measuring the boiling point, make other types of correlation more suitable for the fractions of higher molecular weight. Table VI1 gives sufficient data for the construction of a graph for the boiling point-gravity correlation.
15-00; (B)
Gasoline 100-00. 25-00. 15-cc. (A)
15-cc. (B)
130 138 164 181 197 213 234 254 277 303 2
130 154 163 181 200 216 232 248 277 292 2
90 117 132 160 188 211 240 263 288 316 3
92 110 131 160 188 214 240 267 302 332 4
94 109 125 156 191 221 247 277 307 344 5
94 108 127 164 197 221 246 280 310
230
229
231
230
234
-3
-4
-1
+3
...
Heavy Gasoline Fraction 100-cc. 25-cc.
Mineral Spirits 100-cc. 25-oc.
6
252 285 294 304 314 322 327 336 346 358 2
258 282 291 303 312 321 330 339 350 364 2
302 319 324 331 336 343 348 354 361 371 2
295 318 324 330 336 341 349 356 364 375 3
230
326
328
347
347
-1
...
+2
...
0
.. .
The greatest difficulty in attempting to correlate molecular weight in the lubricating oil range is the obvious discrepancy in data obtained in laboratories which should be expected t o be equally reliable. I n this connection, it is well to consider the cooperative molecular weight work carried out by Rall and Smith (26). These data are summarized in Table VIII, which also includes later date on the same oils by Hanson and Bowman (9). It appears that molecular weights, even when obtained by experienced operators, are subject to considerable error. Thus it is conceivable that any one worker may obtain molecular weight data which are far better in reproducibility than in absolute accuracy. I n view of these data, it is also clear that a
TABLEV.
ACCURACY OF BOILINGPOINTGRAVITY CORRELATION FOR PODBIELNIAK RESIDUES
VISCOSITY-DENSITY AND VISCOSITY SLOPE Sample No.
1
3
2
4
5
6
Several correlations of the molecular weights of lubricating oil 139 149 50'%0bf..tp.p~I~C. 100 100 140 100 fractions with viscosity slope (change of viscosity with temperaG~., 59.8 56 8 48 8 59 0 49 6 45 1 101 114 99 116 119 ture) and viscosity-density data have been proposed. This pregir;:: ;:; 100 120 101 119 123 vious literature has been discussed by Hirschler ( l a ) . He also Deviation, % +3 -1 +5 +2 t 3 c3 proposed ( l a ) correlations of kinematic viscosity a t 100' and 210' F., density a t 25' C., and molecular weight, whereby AND BOILINGPOINTGRAVITY TABLE VI. AGREEMENTBETWEEN VISCOSITY-GRAVITY any two of these four constants could be CORRELATIONS IN OVERLAPRANGE used to calculate the other two. Molecular weights could thus be calculated from Saybolt B.P.-Gravity Vision-Gr. VIsn-Gr. sp'60/Gr" ViaCosity Mol. DeviaMol. DeviaMol. Deviathe kinematic viscosities at 100' and 60° F. IOO'F. 210' F. "C:' Wt. wt.a tion, % wt. tion, wt. tion, % 210" F. or from the density and either D a t a of Fenske, et a2.6 (6) viscosity. 310 37 3 370 4-3 2 77.6 +1 6 0 867 320 315 322 4-3 9 For the viscosity-density correlations 0 867 113 4 338 40 1 395 +5 0 363 +8.1 355 360 +6 5 280 -8.2 282 37 9 360 305 -7 5 283 -7 2 932 102 3 presented in the present paper, Saybolt 0 370 328 295 -10 0 317 188 5 41 8 -3 4 315 0 933 -4 0 365 304 87 0 37 5 $2 0 4-1 0 310 307 310 +2 0 viscosities and specific gravity (60/60' F.) 0 888 D a t a of FitsSimons and Thiele (7) were employed since inspection tests are - still ordinarily given in these latter units. 0 8868 113 3 40 2 398 337 345 4-2 4 335 -0 6 340 fO 9 0 9191 55 7 34 1 340 257 264 +2 7 252 +1 9 255 -0 8 The present viscosity-density correlations 0 8510 44 4 . 316 266 257 -3 4 258 -3 0 0 8516 44 2 303 266 242 -7 3 257 -3 4 . , are not derived directly from those of Present D a t a C Hirschler (12) by a simple change in units; +1.5 +2.7 57 332 261 265 268 0,8877 the entire correlation was revised, 321 54 260 256 -1.5 +3.1 268 0.8751 336 288 -1.1 c2.6 281 278 0,8529 56 smoothed somewhat, and extended further 340 272 -2.2 0 272 266 73 0.9147 into the low viscosity region. These 349 302 -4.7 +1.3 84 298 284 0.8883 327 272 $5.6 270 255 $0.7 0.9018 65 modifications have not resulted in any 0.8649 57 ... 338 293 . 278 - 6 . 1 280 - 4 . 4 considerable differences from the earlier Average deviation from exptl. data 4.1 2.8 3.8 0 +0.2 Deviation of average from exptl. data +1.3 correlation but are sufficient t o require an 2.1 3.7 Average deviation from b.p.-gr. value ... f2.4 independent evaluation. Both the present Deviation of average from b.p.-gr. value 4-0.7 0 Molecular weights above 300 obtained b y extrapolation. correlations and the earlier ones ( l a ) were b Boiling points corrected from IO-mm. boiling points. based primarily on the data of Mair and c A 5% correction was applied to experimental molecular weights for reasons given in text. co-workers (21, 23, 23).
gi: :El
"9
4
.
...
0
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
446
Vol. 38. No. 4
.
A. S.T. M LOW O i S T I L L A T I O N THERMOMETER
50-cc. ENGLER FLASK
y;! 70 80 I10
loll
110 120 130 140 1.50 160 170 180 190 200 210 220 230 240 250 260 270
99"
280 290 300
~,~-~--.---..-----Uoilinr l'oint i n ' C. 1Vher1 Density 28 47 tj4
.. .. ..
0.70 33 55 74 00 104 121
... ,
,
(
I
.
...
...
...
..
0.75 42 07
82 100 116 132 147 162
176 190 203 216 228
.. . , .
. . ... ..
0 80
0.85
0 00
50 72 92 110 126 143 158 174 187 200 213 225 238 249 262 272 282 292 301 310 320 328 337 343
58' 80 101 120 138 154 170 184 198 2 10 222 236 248 289 271 281 291 301 310 319 328 :37
..
.>.I5 3 d1
1l(1 180
148 165 182 196 209 222 234 247 260 272 283 294 303 3 14 323 333 340 349
357 384
(20" C.) -Is: 0.95 I 00
..
~
1.02
1
... ... . .
iici
195 209 222 236 250 262 275 286 300 310 320 330 340 349 3.56 368 873 380
2 10
224 239
233
268 282 294 306 320 332 340 352 361 270 378 386 394 401
2111 232 247 263 278 29 1 303 315 830
241 380 3 62 372 380 388 397 404 411
STANDARD blSTl L L AT ION SHIELD
Figure 5 .
Cryoscopic (Benzene) Rall & Smith (26) Haiison 8z R o a m a n 11 lab., mean Ebullioscopic (0) Benzene
DiatillaLioii Equipment for Podbielniah Residues
correlation oi meilccu1:ir weight n.it,li physical propcrtics \Till not necessarily fit, equally \veil the data from differelit lnboratories. It is therefore necessary t o decide which dat,a nrc most fur correlation. The data of ) i ~ o r rgiven greatest sriglit in deriving the sorrclations presented in this paper. The d a t a of Mair ivvre obt,ained by prcciac clullioscopic technique. Our expericnw \\.it11 the ebullioscopic method. lio~\-ever, does not l e d us to recommend it for general m e since it i,s more time consuming than the cryoscopic procedure sild h a as many, if not more, possibilitics for crpcrimental error. Howcver, in the hands of a csseful laboratory technician it yields good results.
Cyrlohexa~ic Mean C:ilcd. mol. w t . (visioo-pr.
491 442 447
3x3 358 317
G7 I (552
450 445 447 482
326
861 BY0
3 1% 319 334
I11 rrgnrd ttJ tlic cryoscopic technicjue, tlic previous wcornmendstion? (fd, 26') t l t d tlio freezing point depremion be dotermiiicd in t,lie prcseiice of a drying agrrit, are heartily ctidurscd. At present tlic 5yodifference u-hich I\ c liavc observcd in moiccular xeighi s, det,errnined Kith a n d n.ifliout :I drying agent, appears to be due to a small t i w e of moi:;tnre picked up at the time l,he sample is introduced. Thin 57(>differcnctx, if due to this cRuse, ~ o u l not d be duplicated exactly in otlior I>ibomt,oriessiricc details of technique during introduction of sxnplct and the h~irnidit~y of this atmosphere would be controlliiig influeric,es.
SPECIFIC GRAVITY Figure 5.
iO9
7G4 893
60/60 O F
Saibolt Viscosity a t 210" F., Specific Gravity, and JIoleciilar Weight
IN D U S T R I A L A N D E N G IN E E R I N G C H E M I S.T R Y
April, 1946
447
TABLEIX.' DATAFOR CONSTRUCTION OF CHART CORRELATINQ krOLECULAR WEIQHTWITH SPECIFIC GRAVITYAT 60/60° F. AND SAYBOLT VISCOSITY AT 210' F. Mol. Visno W h e n Grso/ao Is: Grao! EO at Wt. 0 84 0.87 0.90 0.93 0.96 0.99 1.02 1.05 1.07 Visuo
44.5 50.0 65.5 101 162
.. .*
..
400 40.8 43.8 48.7 57.5 76.2 124 420 42.0 46.0 52.0 63.0 87.2 145
440 460 480 500 520 540 560 ,580
600 620 640 660 680
.
43.9 45.8 47.5. 49.4 51.6 53.9 56.5 59.0 61.8 65.0 68.5 72.2 75.5
48.2 55.7 51.0 59.5 53.0 63.5 56.0 68.5 59.0 73.5 63.0 79.5 66.5 86.0 71.0 93.0 75.5 102 81.0,112 87.0 123 93.0 135 101 150
70.0 . 101 72.0 115 85.5 134 94 153 106 180 117 204 132 236 148 268 164 186 210 232 255
..
174 201 '239
.. ..
..
..
.. .. ..
.. .. ..
222 270
.. ... . .... ..
.. .. ..
... . .. .. .. .. .. .. .. ..
....
.... .... ....
.... ....
0 MAL-CRYSTAL
1.0702 1.0488 1.0358 1.0248 1.0162 1.0070 0.9998 0,9920 0.9852 0.9767 0.9700 0.9628 0.9568 0.9508 0.9446 0.9388 * 0 I9337 0.9282
49'
X NEEDLE
'The difficulties in cryoscopic procedure due to mixed crystal forniation and solute association have been mentioned in the literature (10, $0). Although we d o not claim t o have conclusive evidence on this point, a considerable amount of experience with both the boiling point and freezing point technique as applied to naphthenic acids and to lubricating oil fractions lead us to believe 1.41 1.4 2 1.43 1.44 1.45 that the difficulties due to mixed crystal formation and solute REF. INDEX AT 80°C. association must be considered; however, details such as presFigure 7 . Me1 ting Point vs. Refractive Index at 80" C. ~ I L C Cof moisture, experimental technique, etc., deserve BS much for Waxes consideration. I n short, we are inclined t o believe t h a t errors due t o details of experimental technique are a major source of difficulty and may explain most of the systematic errors which lead However, this correlation has a lower viscosity limit of 2.6 centito reproducibility, in general, being better than absolute accuracy. stokes (35 seconds Saybolt) a t 210" F. which does not permit its Correlations of molecular weight with ~ i s 2 ~ 0 - g rana ~ ~ visloouse on certain fractions of low molecular weight; in addition, for greO/ware presented here. D a t a for the construction of both of many low viscosity cuts, only the 100"F. viscosity is customarily these charts are given in Tables IX and X, and Figure 6 illusdetermined. The visloo-gravity correlation may be applied to trates the correlation. Since the correlation of molecular weight such fractions down t o a molecular weight of 240. with visloo-vis210was presented by Hirschler ( l a ) , i t is not reThe boiling point-gravity correlation can also be used for oils I up t o 300 molecular weight; therefore, in this intermediate rang(. peated here. It seems worth while t o have available more than one correlait is possible t o cross check boiling point-gravity molecular tion, so that in case of doubt or in the case of an unusual sample, weights against viscosity-gravity values (Table VI). The averthe molecular weight obtained on one graph can be cross-checked age deviation between the molecular weights calculated by thew with that obtained on the other graphs. two correlations is 3.7% for this group of oils. Hirschler ( l a ) showed that the vis100-vis210correlation will, in The viscosity-gravity correlations may be presented in t h t. general, fit the various types of pure compounds, particularly the extreme naphthem and aromatic types, with better TABLEX. DATAFOR CONSTRUCTION OF CHARTCORRELATINQ MOLECULAR WEIOH'I mxuracy than a viscosity-gravity correlaWITH SPECIFIC GRAVITY AT 60/60' F. AND SAYBOLT VISCOSITY AT 100' F. tion. His d a t a indicate, however, that Visioo When Grso/u Is: Grewaoet with normal petroleum fractions the disMol. Wt. 0.84 0.87 0.90 0.98 0.96 0.99 1.02 ' 1.05 1.08 Visio,oou tribution of hydrocarbon types is' such 240 72 104 .... ... 49.0 56 184 392 260 582 1440 145' ... 49:O .... 273 56:5 69.5 94 that the viscosity-gravity correlation will 1870 6800 287 280 58.3 72.0 97.0 150 49.8 1.0880 670 242 300 1,0568 6600 545 1570 68.6 56.5 . . 136 90.0 give a reasonably reliable molecular 378 970 3660 320 1.0325 82.8 114 64.9 .. 190 weight. I n the case of the fractions of 570 1700 7500 340 98.0 143 .. 1 ,0250 256 73.5 810 360 2750 116 83.6 1.0135 . . 344 181 higher molecular weight derived from Gulf 1170 4350 380 138 94.0 1.0044 450 225 Coast or California crudes, which are 1650 6900 .. ". 106.4 162 279 605 2220 10,000 189' 340 .. 755 120 highly naphthenic, the molecular weight .. 220 2900 .... .. .. 960 136 408. calculated from our viscosity-gravity cor.. .... .. 153 252 497 .. 1190 3950 . . . . .. 293 597 5250 . . .. 1500 172 relations are too high. Special viscosity.... . . .. 335 .. 1830 6700 193 695 8600 . . . . . . .. . . 2220 385 810 218 gravity correlations for Gulf Coast oils .. .... .. 244 435 .. 2670 950 have been prepared which give good re560 272 495 1090 3220 .. .... .. .. .. 0.9542 sults for this type of oil. They are de580 304 555 1270 3840 ,. .... .. .. .. 0.9500 600 333 627 1460 4600 .. . . . . . . . . . . 0.9460 wribed in the next section. 610 370 706 1700 6560 .. ..,, ,. .. ., 0.9420 640 402 780 1960 6800 .. .... .. .... The correlation is therefore 660 440 870 2300 8100 .. .. .. ... considered most reliable and is recom680 479 977 2670 9500 .. .... .. .. .. .... m e n d 4 when two viscmities arc available.
-
I
.
I
I .
I
. I . .
.
.
I
.
*
.
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
448
Vol. 38, No 4
Since, as stated previously, the. corrclation was b a e d Driniiirilv 240-370 Mol. Kt. 370-4.50 Mol. Wt. >430 1\Iol. wt. on the data of Mair, the agrreS o : of DeviaDev. of No. of DeviaDev. of KO.of Dpvia- Dev.-
