INDUSTRIAL AND ENGINEERING CHEMISTRY
2182
(7) Reid, J. C., and Hersberger, A. B., A S T M Bull., 145, 7 7 ,
LITERATURE CITED
(1) Burkhardt, C. H., Fueloil d Oil Heat, 7 , 50,1 2 , 58 (1949). (2) cauley, s, p,, and ~ ~E. B.,oil i &s J~, , 45, so, ~ 10, 166 ~
(1946).
(3) Cauley, S. P., and Linden, H. R., Ibid., 45, S o . 12, 80 (1946).
(4)Hunt, R. A., IKD. EKG.CHEM.,45, 602 (1953). (5) Kewleu, J.. and Jackson, J. S..J . Inst. PetroZeum Technol., 13, (6)
Vol. 46, No. 10
364-(1927). Lochlin, D. TV,,and Parmalee, G. V., Heafing. Piping, Air Conditioning, 22, No. 12, 121 (1950).
(1947). ~(8) Terry, , J. B., and Field, E., 1x1,EM. CHEN., 8, 293 (1936). (9) Western Petroleum Refiners ;ISJOC.. Tulsa, Okla., “Burner Oils-Report of Research Project,” 1945. (10) Woodrow, W.A , , World Petroleum Congr., Proc., 1st Congr., London, 1933, Vol. 11, p , 7 3 2 . RECEIVED for review January 27, 1954. ACCEPTEDM R 24, ~ 1954. Presented before the Division of PetroleJm Chemistry a t the 125th Meeting CHEXICALSOCIETY, Kanaas City, Mo. of the AMIEXICAK
Prediction of ASTM End Points of Blended Light Petroleum Products MAURICE E. STASLEY AND GWENDOLYX D. PINGREY The Texas Co.,P o r t rlrthur, Ten.
ETROLEUM refinery designers and operating engineers are frequently called upon to estimate the XSThI distillation curves of mixtures of stocks. I n the past, such predictions have relied heavily upon the experience and judgment of the estimator, and there has been no satisfactory method for predicting the results of blends of widely dissimilar stocks. The increasing consumption of fuels over the past severa.1 years has made it imperative to blend closer and closer to limiting volatility specifications in order to obtain masimum production. At the same time, the development of new processes such as alkylaiion and catalytic cracking has made new stocks available for use which have such imrrow boiling ranges that they considerably complicat,e blending predictions. Therefore, the need for more precise met,hods for estimating distillation points has inweased steadily over the past several years. The effect of natural gasoline on the ASTRI distillation curve of gasoline blends is discussed by Sash and Howes ( 4 )who present a summary of work done by Zublin ( 6 ) and by Alden and Blair ( 1 ) . The correlations of these authors are somewhat limited in their applicability to modern refinery calculations. However, a method presented by Haskell and Beavon ( 3 )appears applicable for estimating the 10% points of most refinery blends containing natural gasoline, butane fractions, and pentane fractions. Kelson (6) states that the boiling range of naphtha-natural gasoline blends can be estimated by averaging the material boiling up t’o a given temperature in the ratio of the quantity of each blending agent used. Although this method is suggested only for naphthanatural gasoline blends, it has found wide usage for many gasoline blending problems and is thought to be generally applicable for the mid-range of gasoline blends provided the components do not differ too widely in distillation characteristics. This paper presents a method for predicting the ASTRI. end points of blends of light petroleum products when only the composition of the blends and the ASTLI distillations of t’he component stocks are known. This method appears to be sufficiently accurate for most refinery calculations and is applicable to stocks varying in distillation characteristics from butane and pentane through full range naphthas and narrow boiling range products to kerosine. EXPERIMENTAL PROCEDURE
The development of a blending correlation requires a large amount of accurate data on a representat’ive portion of the many combinations of stocks that may be encountered in refinery operation. Therefore, R survey was made in a large re-
finery to determine how many tjpes of stoclre were manufactured. On the basis of this survev, the thirteen stocks shown in Table I were chosen t o represent the various fractions that might be encountered in refinery calculations. Butanes and pentanes were considered sepaiately, as discussed later. The list included wide boiling range fractions such as thermal cracked gasoline (G), narroF boiling range fractions such as Stoddard Solvent ( P ) , stocks that have a long “tail” such as light catalytic cracked gaqoline (E),and stocks iyith short “tails” such as aviation straight run gasoline ( A ) . “Tail” is used t o desciibe the final portion of an ASTM distillation Curve where the temperature normally iises much more rapidly than it doe> during the midportion of the curve. The stocks ranged from low boiling aviation gasoline fractions through motor gasoline and special solvent fractions to stocks which might be uwd in kerosine and jet fuel manufacture. From these thirteen stocka, 108 blends were prepared and
420
280
t
‘
I
20
I 40
I 60
I 80
r 100
% HIGH END POINT COMPONEP1T
Figure 1. Typical End Point-Composition Curves
I N D U S-TR I A L A N D E N G I N E E R I N G C H E M I S T R Y
October 1954
2183
TABLE I. STOCKS
Start 10 20 30 40 50 60 70 80 90 95
End point
...
110 146 158 169 178 185 192 200 208 218 227 256
257 266 271 276 281 286 292 299 311 320 341
346 350 355 360 373
F.
5 10 15 20 25 30 35 40 ._ 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145
(Use in Conjunction with Table 111) High End Point Stock in Blend, % _____ 5
10
15
20
0 1 1 2 2 3 3 4 4 5 6 6 7 8 9 10 11 12 14 15 17 19 21 23 26 29 33 38 45
1 2 2 3 4 5 6 8 9 10 12 13 15 17 19 21 23 26 29 32 35 38 42 46 51 56 62 69 77
1 2 3 4 6 7 8 10 12 14 16 18 20 23 26 29 32 36
1 3 4 6 8 9 11 13 16 18 20 23 26 29 33 36 40 44 48 53 58 63 68 73 78 84 90 96 101
40
44 48 53 58 63 68 74 79 85 91
25
30
40
50
60
70
...
420
333 351 358 365 372 379 389 399 413 430 446 481
327 357 373 3afi 400 412 424 437 . .. 451 467 479 494
As nil1 be disrusscd later, the effect of added butanes and pentanes on the ASTM distillation end point was determined from data on several earlier blends. These blends had been tested only once or twice each.
TABLE 11. ASTAMBLENDIXG CORRELATION-A Diff. in End Points,
173 263 283 296 306 315 324 333 343 357 369 407
80
DEVELOPMENT OF END POINT BLENDING CORRELATION
Plots of the average end points versus the composition of the blends, such as Figure 1, showed that the locus of the end points of blends of two stocks is not a straight line but is a curvilinear function whose shape varies depending upon which components are blended. Study of the various plots disclosed that both the difference between the end points of the components and the slope of the tail of the distillation curve of the higher end point component affect the shape of the curves. The end point minus the 90% point was chosen as the correlating factor because the necessary data are normally available in most distillatipn blending problems. The relationships which were finally adopted are presented in Tables I1 and 111. U S E OF END POINT BLENDING CORRELATION
The procedure for use of Tables I1 and I11 can best be shown by an example:
It is desired to know the end point that will be obtained with a tested. No attempt was made to prepare every conceivable blend containing 75% of the light straight run gasoline ( B ) and ?5% of the heavy catalytic cracked gasoline ( H ) shown in Table combination of stocks, but it is believed that enough combinations were tested to cover most types of blending problems that might be encountered in refinery design and operation. End point of H 400" F. At least five ASTM distillation tests were made on each of the End point of B 290' F. stocks and blends to be certain that the basic data used for deDifference 110" F. velopment of the blending correlation would be accurate. An attempt was made to have each sample tested by severd different operators using TABLE 111. ASTM BLENDING CORRELATION-B different sets of apparatus to minimize syste(Use in conjunction with Table 11) mat,ic errors in the testing. In cases where the End Pointfive distillations showed wide discrepancies, addi90,% Point, O.F. High End Point Stock in Blend, % End Point tional tests were made until the true end point had Stock) 5 10 15 20 25 30 40 50 60 70 80 been established with a satisfactory degree of ac5.5 3.18 2 . 3 8 2 . 0 8 1.90 1 . 7 5 1.54 1 . 4 2 1 . 3 3 1.28 1.24 I8 4 . 8 2.70 2.05 1.80 1.65 1.54 1.40 1.29 1.22 1 . 1 7 1 .15 curacy. The several distillations were then 2o 3.3 2.23 1 . 7 4 1.55 1.43 1.34 1 . 2 4 1 . 1 8 1.13 1.10 1.08 25 averaged. In those few cases where the end point 30 2.55 1.92 1 . 6 5 1 . 4 2 1 . 3 2 1.26 1.17 1 . 1 2 1 . 0 8 1.05 1.04 35 2.04 1.65 1.40 1.30 1.24 1.20 1.12 1.08 1.05 1 . 0 2 1.01 of one distillation diffeled by more than 10' F. 40 1.68 1.44 1.28 1.22 1 . 1 7 1.14 1.08 1.05 1 . 0 3 1 . 0 1 1.00 45 1 . 4 0 1 . 2 7 1.18 1.14 1.11 1.09 1.05 1 . 0 3 1 . 0 1 1.00 1.00 (twice the standard error) from the average of all 50 1.18 1.13 1.09 1 . 0 7 1.05 1.04 1 . 0 2 1.01 1.00 1.00 1.00 the distillations on that stock or blend, the errant 55 1.00 1.00 1.00 1.00 1 . 0 0 1.00 1 . 0 0 1.00 1.00 1 . 0 0 1.00 60 0 . 8 8 0.90 0 . 9 2 0 94 0 . 9 5 0.96 0 . 9 7 0 98 0 . 9 9 0 . 9 9 1.00 value was discarded and the remaining values were 65 0 . 7 8 0.82 0 . 8 5 0.88 0.90 0 . 9 1 0 . 9 3 0.95 0.97 0.98 0.99 70 0.70 0.74 0.78 0 . 8 2 0 . 8 5 0.87 0.90 0.93 0.95 0.97 0.98 averaged. Calculations based on t'he 854 distilla75 0 . 6 2 0 . 6 7 0 . 7 2 0.76 0.80 0.82 0 . 8 6 0.90 0 . 9 3 0 . 9 5 0 . 9 7 80 0.55 0.60 0.66 0 . 7 1 0.75 0.78 0.83 0.87 0 . 9 1 0.93 0.96 tions made during the course of this work showed 85 0 . 4 9 0 . 5 5 0 . 6 1 0.66 0 . 7 0 0.74 0.79 0.84 0.88 0 . 9 1 0.94 t'hat the standard error of a single ASTRI end 90 0.44 0.50 0.56 0 . 6 1 0.65 0.69 0 . 7 5 0.80 0 . 8 4 0 . 8 8 0 . 9 1 95 0.39 0.46 0 . 5 1 0.56 0.60 0.64 0 . 7 1 0 . 7 6 0.80 0.84 0 . 8 7 point determination as nieasured in this laboratory 100 0.34 0.42 0.47 0.51 .56 0.60 0.66 0 . 7 2 0.76 0 . 8 0 0.83 I s 14.8' F.
INDUSTRIAL AND ENGINEERING CHEMISTRY
2184
Vol. 46, No. 10
TABLE IV. ACCURACY OF PREDICTION-BIXARY BLENDS Composition,
%
95Af5G 90 A 10 G 85 A 15 G 75 A 25 G 50 A 50 G 25 A 75 G 95B+5G 90 B 10 G 85 B 15 G 75 B 25 G 50 B 50 G 25 B 75 G 95C+5G 9OC 10G 85 C 15 G 75 C 25 G 50 C 50 G 25 C 75 G 75 D 25 G 50 D 50 G 25 D 75 G 75 E 25 G 50 E 50 G 25 E 75 G 95F+5G BO F 10 G 85 F 15 G 75 F 25 G 50 F 60 G 25 F 75 G 75 G 25 H 50 G 50 H 25 G 75 H 75 G 25 I 50 G 50 I 25 G 75 I 75 G 25 J 50 G 50 J 25 G 75 J 95Gf5K 90G+10K 85 G 15 K 75 G 25 K 50 G 50 K 25 G 75 K 95G+5L 90 G 10 L 85 G 15 L 75 G 25 L 50 G 50 L 25 G 75 L
End Point, F. Actual Estd.
Error,
++ ++ +
$3
++ ++ +
+3
++ + ++ ++ + ++ +
++ ++ + ++ + ++ + +++ ++ ++
+ +++ +
+4
+2
+A
0
+4 0
-1 -4
-1 -1 +3 +4
3.3 -2 +I 365 383 394 379 390 396 377 382 3x3 ~. 383 386 393 403 402
403 400 402 402 404 412 419 420 43 1 439 452 469 4i7 437 459 465 476 487 492
3GQ 383 393 379 388 394 375 377 3i9 382 389 395 399 400 400 402 404 405 405 411 41.5
410 423 431 445 465 475 440 460 466 477 49 1 493
+:
-1 0 -2 -2 -2 -5 -4 -1 +3 +2 -4 -2 -3 +2 +2 +3 +I -1 -4 10 -8 -8*
-
--I
-4 -2
Csing this 110" F. A end point and the composition of the blend (25% heavy component), Table I1 gives an uncorrected value of 71' F. to be added to the end point of stock B. End point of H 90% point of H Difference
Composition,
F.
400" F. 373" F. 27' F.
Table I11 gives a correction factor of 1.39 for blends containing 25% of a heavy component which has an end point 90% point slope of 27" F. 1.39 X 71 = 99 = corrected F. to be added to low end point. 290 99 = 389" F. = predicted end point The actual end point of this blend was 388' F.
+
hIulticomponerit blends require stepwise calculation carried out as though the final blend were the result of a series of binary blends. Although it is theoretically possible to combine the various components in any desired order, it is easiest to start with the two lowest end point components and successively add higher end point materials since this procedure eliminates the necessity for estimating the 90% point of the intermediate blends. It may be somewhat surprising to find that the end point estimated by Tables I1 and I11 is sometimes higher than the end point of either component. IIowever, this phenomenon is in accord with the actual data which showed that 11 out of the 108 blends tested had end points higher than any component of the blend. Certain blends containing component F , Stoddard Solvent, had end
%
+ +++ + + ++ + ++
95 G 5 M 90 G -I- 10 A1 8.5 G 15 11 75 G 25 41 50 G 50 h1 25 G 75 11 95 .4 f 5 B 90 A 10 B 85 A -C 15 B 75 A 25 B 50 A 50 B 25 A 75 B 95A 5 E 90 A 10 E 85A+l5E 75 A 25 E 50 A 50 E 25 A 75 E 95B 5C 9OB 10C 85B 15C 75 B 25 C 50 B 50 C 25B 75 C 95B + 5 F 90 B 10 F 85B l5F 75 B 28 F 50 B I50 F 25 B 75 F 95B+5H 90 B 10 H 85B 15H 7 5 B -C 25 H 50 B 1. 50 r-I 25 B 75 €1 75 C 25 E 50 C 4 50 E 25 C 75 E 75E t 25F 60 E 50 F 75 F 25 E 75 E 25 H 50 E 50 H 25 E 75 H
+++ ++ ++ ++ ++ + + ++
+ + + ++ ++ +
E n d Point, F. Actual Estd. 434 432 455 455 470 468 483 482 499 498 504 503 262 262 266 266 264 267 272 272 279 277 282 280 203 267 272 273 281 280 297 300 325 326 341 316 292 287 296 295 298 298 306 299 320 316 330 327 3.50 348 360 360 366 364 378 376 381 385 384 385 342 347 370 370 379 381 3 88 389 399 398 402 401 341 346 315 381 343s 357 374 371 377 372 383 373 387 384 397 392 401 397
Error,
F.
-2 0 -2
+1 +1 -1 0
0
+3
0
-2 $2 -4 -1 +1 -3 -1
-5
+5
+At T +4
+3
+E-2 -2
$4 +1
+; +z
4-1 +1 -1
+5 +G +14
-3
5 -- 10 -3 -5
-'I
a Appears exceptionally far o u t of line on plot of end point versus composition.
points as much as 11" F. higher than the end point of either component. EFFECT OF PRESSURIXG AGEUTS
Butane and pentane fractions are usually added to gasoline blends t o improve volatility characteristics. Seither butanes nor pentanes were included in this work, and their effect on end points cannot be determined by Tables I1 and I11 since ( a ) the end points of the pressuring agents cannot be determined by the standard BSTM distillation test and ( b ) even if the end points could be determined, the difference between these end points and those of the other components of a normal gasoline blend would be greater than the range of end points of the stocks used
T.4BLE
v.
ACCURACY
Composition, 33118 40 B 40 B 33.18 40 B 40 B 331/3 40 B 40 B 331,'s 20 B 40 B
OF
PREDICTIOX-TERNARY BLENDS E n d Point, F. Actual Estd. 37 1 372 377 378 361 330 391 384 371 386 396 389 391 392 393 39 1 396 389 394 394 399 399 392 390
E
~ F. -2 -7
- 12 -7
-2
-2 -,
0
0
+z
~
~
~
,
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1954
TABLE VI.
.N
ACCURACY OF PREDICTION-BUTANE BLENDS End Point, F. Actual Estd.
Composition, %
++ 510n-butane n-butane + 515isobutane-n-butane n-butane + mixture N + 10 isobutane-n-butane mixture N + 15 isobutane-n-butane mixture N + 5 catalytic butylenes N + 10 catalyt,ic butylenes N + 5 thermal butylenes N + 10 thermal butylenes 0 95 0 + 5 n-butane 90 0 + 10 n-butane 85 0 + 15 n-butane 95 0 $- 5 isobutane-n-butane mixture 90 0 + 10 isobutane-n-butane mixture 85 0 + 15 isobutane-n-butane mixture 95 0 + 5 catalytic butylenes 90 0 + 10 catalytic butylenes 95 0 + 5 thermal butylenes 90 0 + 10 thermal butylenes 95 90 85 95 90 85 95 90 95 90
P 95 90 85 95 90 85 95 90 95 90
Q
90 90 85 95 90 95 90
N
N N N
++ 510n-butane n-butane P + 15 n-butane P + 5 isobutane-n-butane mixture P + 10 isobutane-n-butane mixture P + 15 isobutane-n-butane mixture P + 5 catalytic butylenes P + 10 catalytic butylenes P + 5 thermal butylenes P + 10 thermal butylenes P
P
+ Q + 10 n-butane-isobutane Q + 15 n-butane-isobutane Q + 5 catalytic butylenes Q + 10 catalytic butylenes Q + 5 thermal butylenes Q + 10 thermal butylenes Q
..
mixture mixture
379 380 375 376 380 378 372 379 377 379 380 382 384 381 381 380 381 379 379 376 382 384 381 378 377 382 376 376 372 378 376 380 374 394 39 1 393 390 395 394 393 392
...
378 376 375 378 376 375 378 376 378 376
...
381 380 378 381 380 378 381 380 381 380
...
-2
2: -2 -2 +3
-1 -1 -1
-4
... -3 -1 -3
+1 -1 -1 +2
2;: -4
380 378 377 380 378 377 380 378 380 378
72
...
*..
392 392 390 393 392 393 392
$1 -5 $4 +2 +5 +2
+; +4
+: 0 -2 -2
0 0
in developing Table 11. However, a review of available data on the 37 butane blends and 37 pentane blends shown in Tables VI and VI1 indicated that from 0 to 15%, addition of butanes decreases the end point of a blend 0.25” F. for each per cent used, and from 0 to 50%, addition of pentanes decreases the end point of the blend 0.28’ F. for each per cent used. The accuracy of the data from which these figures were derived was not sufficient to determine whether the type of base stock or the composition of the pressuring agent has an effect on the rate of end point reduction with increasing concentration of pressuring agent; therefore, it is recommended that the use of these figures be limited to normal gasoline blends. ACCURACY
*’
TABLEVII. ACCURACYOB PREDICTION-PENTANE BLEND^
Error, O F.
...
Tables IV, V, VI, and VI1 show the accuracy of the proposed method for predicting the end points of the various blends tested. The largest error noted for any of these blends was 14’ F. for the mixture of 25% C plus 75% E shown in Table IV. However, the plot oi end point versus composition for this series of blends showed that this blend was definitely “out of line” with the others in the series so that at least part of this error may be attributed either to blending or testing errors. Likewise, some of the other errors may be partially attributed either to blending or testing. Even when these errors are included, however, the standard error of prediction
of the 108 binary and ternary blends shown in Tables I1 and I11 is only 13.8’ F. This figure is, of course, based on the averages of multiple distillations on each base stock and blend. In order to estimate the accuracy of prediction when only single distillation determinations are available, calculations were made on 25 blends chosen a t random from the 108 blends shown in Tables IV and V using tests chosen at random from the several distilla-
2185
*End Point, Actual Estd.
Composition, %
R
...
Error, F.
...
+ 5.05 isopentane 9.75 isopentane +++49.7isopentane 14.6 isopentane
285 286 276 279 274
284 282 281 271
++5.210.25 isopentane isopentane + 17.25 isopentane T 64.25 T + 5.75isopentane 90.25 T + 9.75 isopentane 84.6 T + 15.4 isopentane 78.5 T + 21.5 isopentane 49.8 T C 50.2 isopentane
286 280 278 278
285 283 281
+i
304 302 299 294 292 284
302 301 300 298 290
...
...
...
...
94.95 R 90.25 R 85.4 R 50.3 R
s
94.8 9 89.75 S 82.75 S
”
TT
++ ++
95 U 5 isopentane 89 U 11 isopentane 84.75 U 15.25 isopentane 49.9 U 50.1 isopentane
W
...
290 284 288 286 278
289 287 286 276
292 296 292 285 281
291 289 288 278
... ...
5 isopentane +++ 50 15 isopentane isopentane X 90 X + 10 isopentane 85 X + 15 isopentane
296 295 295 278
295 292 282
288 287 285
285 284
Y 94.1 Y 89.5 Y 82.7 Y
372 369 370 369
370 369 367 401 400 398
95 W 85 W 50 W
+ 6.9 Ca fraction + 17.3 Ca fraction z 93.8 Z + 6.2 Ca fraction 88.3 2 + 11.7 Ca fraction
3- 10.5 Cs fraction
... ...
...
82.5 Z
4- 17.5 Cs fraction
403 400 399 398
AB 90 AA 70 AA
++ 3010 catalytic Ca’s catalytic Cs’s
380 376 371
377 372
BB 90 BB 70 BB
++ 3010 catalytic Ca’s catalytic Ca’s
380 380 373
377 372
++ 3010 catalytio Ca’s catalytic Ca’s
376 374 371
cc
90 CC 70 CC
...
-2 +6 +2 -3
+5 $3
0 $2 +6 +6
+6
i: 0
-2
... -5 -3
2; . .0.
-3 +4
... -2
-1
... +1 -1 -2
+i
+; ...
+1
+1
...
... -3
...
... -1
3 73 368
-1
-3
tions on each blend and component. Based on these 25 calculations, the standard error of prediction is 1 5 . 6 ’ F. The standard errors of prediction of the butane and pentane blends shown in Tables V I and VI1 are h2.4’ F. and 13.6’ F., respectively. ACKNOWLEDGMENT
The authors wish to express their thanks to N. B. Haskell for aid and encouragement in the development of the method described and to The Texas Co. for permission to publish this information. REFERENCES
(1) Alden, R. C., and Blair, 31.G., Natl. Petroleum News, 22, No. 46,107 (1930).
(2) American Society for Testing Materials, “Standard Method of Test for Distillation of Gasoline, Kerosine and Similar Petroleum Products,” D 86-46,p. 7, Philadelphia, 1951. (3) Haskell, N. B., and Beavon, D. K., IND.ENO.CHEM.,34, 167-70 (1942). (4) Nash, A. M., and Howes, D. A., “Principals of Motor Fuel Preparation and Application,” Vol. I, pp. 134-46, Wiley, New York, 1935. (5) Nelson, W. L., “Petroleum Refinery Engineering,” p. 724, McGraw-Hill, New York, 1949. (6) Zublin,E.W., Oiland GasJ., 29,No. 1,182-3 (1930). RECEIVED for review June 13, 1953.
ACCIOPTED April 22, 1954.