. .

Computation of Petroleum
Fractionation
Estimation of A. S. T. M. Distillation Curves
from True-BoilingPoipt Distillation Analvses J

R. L. GEDDES
Stone & Webster Engineering Corporation, Boston, Mass.

OR the cases of fractiona- The estimation of A. S. T. M. distillation efficient column fractionation of

F tion of relatively simple
mixtures of the lower boil-
ing hydrocarbons into specified
curves from true-boiling-point distillation
curves is a practical problem in the inter-
a sample.
A few general characteristics
of the A. S. T. M. distillation
products, the methods of compu- pretation of theoretical computations of procedure and curve may be
tation of the number of bubble petroleum fractionation. The present noted. This batch distillation
trays and the reflux ratios re- paper offers an improved method of esti- test is empirical, and the exact
quired have been developed until mation of the A. S. T. M. curve from the nature of the curve obtained
they constitute a practical tool depends upon the experimental
for engineering design. The use true-boiling-point curve, based on a survey technique followed. A stand-
of similar methods in fractiona- of data from a large number of distillation ardized procedure is specified
tion problems involving the analyses. It is indicated that the same ( I ) , but within the range of
higher boiling, exceedingly com- correlation may be used conversely for permissible operation several
plex petroleum mixtures has been approximate estimation of the true- slightly different but acceptable
hampered by the lack of a precise distillation curves may be ob-
method of interpreting the calcu- boiling-point curve corresponding to a tained for a given sample of oil.
lated true-boiling-point analyses given A. S. T. M. distillation curve. Under certain conditions super-
of productsin terms of theempiri- heating leads to incorrect vmor
cal A. S. T. M. distillation analy- temperature readings. The dis-
sis (1). The latter convenient-inspection assay, by long usage tillation curve approaches but is not the same as that obtained
in the a r t of commercial fractionation, forms the practical in true differential vaporization, largely because of partial con-
criterion for definition of quality of naphthas, gasolines, and densation in the neck of the distillation flask, lack of true equi-
kerosenes, and of degree of separation between fractions. librium between vapor and liquid, and the liquid holding ca-
Rigorous means are available for computation of the frac- pacity of the condensing system. The degree of fractionation
tionation requirements for complex petroleum mixtures is intermediate between that of an equilibrium flash distillation
(5, 9),the calculated product fractions being defined by true- and a T. B. P. distillation, and the slope of a given portion of the
boiling-point curves. The present paper offers an improved A. S. T. M. curve depends not only upon the slope of the corre-
method of estimating the corresponding A. S. T. M. distillation sponding portion of the T. B. P. curve, but also on the slope
curves. Nelson and Hamburg (6) published a method of and character of the other parts of the latter. These considera-
correlation of A. S. T. M. and true-boiling-point distillation tions lead to the expectation that any statistical correlation
curves, but their correlation does not appear to give suf- of A. S. T. M. and T. B. P. distillation curves will involve con-
ficiently precise results (see also Lewis and Robinson, 4). siderable scattering of the experimental points. However,
During the course of many laboratory research projects it also follows that a rational correlation, once effected, might
and commercial plant performance tests over a period of permit a more reliable estimation of the best A. S. T. M.
several years, there became available a large number of true- curve than would be obtained experimentally from a single
boiling-point analyses of petroleum fractions for which the actual distillation test.
corresponding A. S. T. M. distillation data (1) had been ob- For a given sample of oil the exact character of the T. B. P.
tained. The true-boiling-point distillations were made with curve depends upon the effectiveness of the column fractiona-
a Podbielniak standard high-temperature fractionation tion used. Just what minimum degree of fractionation con-
column (7). The present correlation work is based upon a stitutes a T. B. P. fractionation is not precisely d e h e d in the
survey of the data from approximately two hundred samples; literature, but this is usually determined by the use to be
Table I summarizes part of the basic data used. made of such an analysis. By way of illustration, if we carry
Both the true-boiling-point and A. S. T. M. distillation out a series of successive batch column fractionations on a
curves are the results of plotting as the abscissa the volume given sample, starting with a degree of fractionation less than
percentage of the sample distilled against the still-head tem- T. B. P. requirements and increasing the power of the frac-
perature as the ordinate. The A. S. T. M. procedure involves tionation, the first distillation curves of the series will be
simple batch distillation with no added fractionation except smooth and will be characterized by increasing slope until a
that due to a slight amount of partial condensation in the limiting value of the latter is attained. With subsequent in-
upper parts of the distillation flask. The true-boiling-point creased fractionation, stairsteps in the curve will develop, but
(T. B. P.) distillation, on the other hand, involves a fairly even with the highest attainable fractionation the position
795
796 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 33, No. 6

TABLEI. DISTILLATION
DATA
Sample A . P. I. A . 9. T. M. Distillation (Corrected for Loss), F.
NO. Name Gravity 1 . B . P : 10%. 20% 30% 40% 50% 60% 70% 80% 90% E.P. Residue Lass
1 Rubber solvent 77.2 117 143 153 162 171 180 188 198 210 227 279 1.0 1.o
2 Na htha 66.6 158 181 186 192 198 203 210 217 226 242 283 0.5 0.5
3
4
V. 8. & P. naphtha
Stoddard solvent
61.2
54.1
220
324
240
334
244
337
250
342
254
346
260
351
265
356
270
362
278
370
288
382
316
411
1.0
1.0
0
0
5 Naphtha 61.6 100 162 189 208 222 233 246 257 270 287 322 0.8 2.2
6 Naphtha 47.4 318 328 331 334 337 340 344 348 354 364 392 1.o 0
7 Kerosene 41.0 381 397 405 410 416 422 429 437 449 465 497 1.o 0
8 Gas oil 34.7 456 484 491 498 505 512 52 1 534 550 577 648 1.0 0
9 Naphtha 57.6 113 183 210 230 246 259 275 290 306 327 374 1.2 1.8
10 Blend 50.6 126 214 250 278 302 329 353 379 404 431 471 1.5 0.5
11 Blend 46.3 139 252 296 335 370 400 433 461 491 523 597 1.5 0
12 Gas oil 37.6 387 421 431 442 455 468 481 498 518 552 622 1.o 0
13 Blend 38.9 340 387 401 416 432 451 468 486 509 54 1 616 1.0 0
14 Naphtha 42.6 332 354 362 370 379 389 400 412 425 444 480 1.5 0
15 Gasoline 62.1 128 179 202 223 244 265 286 308 329 352 385 1.0 1.0
16 Naphtha 67.7 117 160 173 185 196 208 219 229 242 259 292 0.9 1.0
17 Xaphtha 56.3 238 269 280 291 303 313 324 334 346 362 390 1.2 0.5
18 Unstripped naphtha 47.9 287 308 313 317 321 325 328 333 339 350 382 1.0 0
19 Unstripped kerosene 41.5 339 375 385 392 398 405 412 420 431 448 481 1.0 0
20 Unstripped gas oil 35.5 349 456 474 484 494 503 512 524 541 567 636 1.0 0
21 Gas oil 30.0 465 536 555 572 586 601 616 635 656 683 736 0.5 0
22 Unstripped naphtha 48.4 296 316 321 324 328 331 334 338 344 351 379 1.0 0
23 Unstripped kerosene 41.4 345 387 398 404 412 419 425 433 444 459 483 1.0 0
24 Unstripped gas oil 35.5 352 470 485 495 602 508 515 522 531 545 586 1.0 0
26 Gas oil 31.0 469 527 542 552 564 573 583 595 609 633 681 0.7 0

26 Unstri ped kerosene 43.6 331 382 396 408 418 430 441 455 472 496 538 1.5 0
27 NaphtRa 51.4 231 286 300 309 319 328 335 346 356 372 415 1.5 0.5
28 Gas oil 28.5 405 472 493 514 540 568 597 625 655 701 750 1.0 0
29 Blend 52.6 102 180 230 279 324 363 396 430 459 495 538 1.0 2.0
30 Cracked yaso:ir.e 39.1 100 157 178 194 206 222 238 261 292 335 404 2.0 1.0
31 Crackcd gasoline 34.5 124 189 202 212 223 237 255 277 307 345 399 1.2 0.3
32 Crscked nnp!itha 34.8 183 197 201 205 209 215 222 233 247 269 305 0.5 0
33 Cracked nnplithn 34.4 181 195 198 203 206 213 220 229 242 260 298 0.5 0
34 Cr.ickc-d nnpLrha 31.5 184 195 197 201 205 210 216 225 238 256 300 0.5 0
35 Polymer gasoline 58.7 98 119 128 137 149 162 177 197 229 295 395 0.8 1.2
36 Polymer gasoline 57.5 98 119 129 141 152 167 183 205 235 292 405 1.o 1.o
37 Polymer gasoline 61.7 96 108 120 132 143 157 175 196 224 280 409 0.8 2.7
38 Polymer gasoline 59.4 96 114 125 137 148 162 178 198 226 280 399 0.8 z.2
39 Polymer gasoline 61.4 94 108 123 137 150 167 186 207 235 286 399 0.9 4.1
40 Light naphtha 82 111 120 129 136 143 150 156 163 174 218 0.5 0.6
41 Light naphtha 74:5 94 121 128 134 140 144 149 152 157 162 205 0.5 0.5
42 Light naphtha 74.2 95 120 131 142 149 156 162 170 176 188 230 0.5 0
43 Light naphtha 77.8 87 111 118 127 134 142 151 162 168 179 222 0.5 2.0
44 Light naphtha 80.8 73 97 110 119 127 138 144 151 159 166 210 0.6 4.5
45 Heavy naphtha 52.1 266 280 283 286 288 292 296 302 310 322 370 1.0 0
46 Heavy naphtha 48.6 265 281 284 286 290 294 297 302 309 320 365 0.5 0
47 Kerosene 39.7 347 380 390 403 416 427 443 456 469 490 522 1.0 0
48 Gasbil 34.5 483 52 1 530 536 54 1 546 551 557 565 576 605 1.o 0
49 Aromatic extract 14.8 431 522 528 533 538 543 549 558 566 582 62 1 1.0 0

50 Gas oil 32.9 555 583 590 596 602 608 614 62 1 631 647 678 2.0 0
51 Gasoline 63.5 97 159 188 210 228 245 262 278 303 327 362 0.5 1.5
52 Naphtha 55.1 199 266 281 294 303 315 326 334 344 360 388 1.5 0.7
53 Naphtha 51.9 178 320 330 338 344 350 357 363 370 384 428 1.o 0.6
54 Kaphtha 50.8 308 343 351 358 364 369 375 382 392 407 451 1.o 0
55 Naphtha 48.8 324 367 376 382 388 396 402 408 418 434 473 1.0 0
56 Kerosene 47.0 344 393 403 409 416 424 432 442 453 471 507 1.0 0
57 Kerosene 44.9 366 422 435 444 452 460 468 476 485 503 546 1.0 0
58 Gas oil 42.8 382 471 485 492 500 508 516 524 538 565 613 1.5 0
Volume per cent

and slope of the equivalent smooth distillation curve

through the stairstep plateaus will change but little. This
latter statement may need qualification in cases where mix-
tures consist of only two or three pure compounds.
It appears that the Podbielniak high-temperature column
(7) is more powerful than the earlier packed T. 13. P. columns
of the Peters type (6). This paper will not attempt a thor-
ough discussion of the factors and their relative importance,
which determine the precision and effectiveness of the T. B. P.
analysis. The most important factors are column length and
efficiency per unit length, reflux ratio, size of batch charged,
and liquid holdup capacity of the column under operating
conditions. The Podbielniak column is equivalent to only
seven to fourteen equilibrium bubble trays ( 2 , 8 ) , depending
upon the diameter of the standard column; however, the
liquid holdup capacity of the column is very small and
enables a low ratio of holdup to charge capacity to be ob-
tained. As a result, at high reflux ratios and slow distillation
FIGURE1. DERIVATION SMOOTH T. B. P.
OF EQUIVALENT rates relatively sharp T. B. P. fractionation may be realized.
CURVE Particularly in analyses of straight-run (uncracked) naphthas
lune, 1941 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 797

TABLE
I. DISTILLATION
DATA(Continued)
Sample A. P., I. True Boiling Point Distillation, F.
NO. Name Gravity I.B.P. l%a 10% 20% 30% 40% 50% 60% 70% 80% 90% 95% 98%
1 Rubber solvent 77.2 30 34 100 131 151 167 180 197 214 231 252 268 282
2 66.6 95 103 137 160 176 190 203 214 227 242 258 268 282
3 ?&@h&BP. naphtha 61.2 150 166 214 222 240 252 260 266 271 284 306 316 326
4 Stoddard solvent 54.1 250 280 309 317 327 341 351 358 368 385 397 404 414
- -
: Naphtha
Naphtha
61.6
47.4
40
268
27
280
109
307
161
317
190
326
212
333
230
340
249
347
262
356
281
365
301
379
320
394
347
413
7 Kerosene 41.0 317 340 367 388 400 412 422 434 449 460 478 495 512
8 Gas oil 34.7 390 465
9 Naphtha 57.6 -
--40 -410
22 132
482
177
496
212
506
235
516
257
532
278
546
301
668
323
600
345
635
364
672
383
10
11
Blend
Blend
50.6
45.3
40
40
- 100 162
197
217 255
305
293 330 264 395 420 455
544
472 496
252 357 401 438 473 503 580 617
12 Gas oil 37.6 310 340 391 415 434 467 476 493 511 536 571 603 645
13 Blend 38.9 240 300 351 380 409 435 460 481 502 522 662 597 638
14 Naphtha 42.6 240 285 330 353 371 387 405 417 430 452 470 492 502
15 Gasoline 62.1 32 52 142 179 211 238 267 290 316 342 374 391 403
16 Naphtha 67.7 32 43 121 152 173 193 211 240 247 262 280 291 310
17 Naphtha 56.3 140 178 238 264 291 307 316 339 353 361 382 400 408
18 Unstripped naphtha, 47.9 200 235 286 300 310 320 329 339 349 358 374 392 412
19 Unstripped kerosene , 41.5 230 275 361 377 396 407 415 430 441 455 474 489 512
20 Unstripped gas oil 35.5 260 300 429 462 478 492 504 517 534 550 589 622 662
21 Gas oil 30.0 360 395 488 514 538 563 591 619 652 685 711 729 742
22 Unstripped naphtha 48.4 170 235 297 312 322 327 331 338 346 353 364 373 383
23 Unstripped kerosene 41.4 182 270 366 383 395 406 422 434 446 460 475 487 500
24 Unstri ped gas oil 35.5 300 305 450 474 487 496 505 518 534 551 575 588 598
25 Gas oiy 31.0 360 400 491 529 550 570 585 599 614 630 653 682 ...
26 Unstri ped kerosene
Naphtfa
43.6
51.4
... 207 364 389 405 422 439 456 475 495 527 553 568
27 124 170 251 279 304 317 330 339 350 364 383 393 403
28 Gas oil 28.6 ... 352 426 464 500 568 738
29 Blend 52.6 68 86 155 216 260
533
298 350
608
393
643
430
682
462 497
794 832
... ...
30 Cracked gasoline 39.1 0 27 97 140 163 183 203 225 249 277 335 390 435
31 Cracked gasoline 34.5 0 25 150 170 186 203 249 238 268 312 353 393 430
32 Cracked naphtha 34.8 * 130 150 160 161 169 181 195 210 227 245 269 277 290
33 Cracked naphtha 34.4 140 160 174 175 187 201 214 227 240 256 280 297 310
34 Cracked naphtha 31.5 150 174 176 176 183 194 210 220 234 249 277 296 310
35 Polymer gasoline 58.7 15 30 86 103 118 136 154 172 198 230 295 348 410
36 Polymer gasoline 57.5 20 40 92 104 130 147 168 189 212 240 295 352 412
37 Polymer gasoline 61.7 -
-20
20
20 69 94 114 136 159 179 207 233 291 355 420
38 Polymer gasoline 59.4 5 61 94 117 140 160 178 200 231 289 350 400
39 Polymer gasoline 61.4 25 35 80 100 120 143 162 183 210 241 296 335 380
40 Light naphtha 15 25 65 89 112 132 145 151 157 170 189 200 203
41 Light naphtha 74:5 15 25 72 102 126 140 146 151 154 160 168 190 203
42 Light naphtha 74.2 15 25 70 99 127 144 153 162 175 189 202 208 212
43 Light naphtha 77.8 0 15 62 84 102 123 139 148 156 169 189 202 212
44 Light naphtha 80.8 -5 5 51 80 97 118 135 144 153 163 181 200 206
45 Heavy naphtha, 52.1 130 210 263 271 280 287 293 298 306 314 335 347 356
46 Heavy naphtha 48.6 158 216 265 275 281 287 294 300 308 319 333 342 350
47 Kerosene 39.7 238 288 346 360 380 400 418 439 454 473 493 510 528
48 Gas oil 34.5 340 370 486 501 518 538 548 552 564 580 603 617 632
49 Aromatic extrict 14.8 346 398 484 508 518 528 537 547 558 573 602 624 645
50 Gas oil 32 9 412 482 565 582 590 598 608 620 626 635 662 678 690
51 Gasoline 63.5 0 16 99 150 184 209 231 251 279 304 332 349 370
52 Naphtha 55.1 104 150 228 262 288 310 324 338 350 363 382 394 402
53 N,apphtha 5P .9 44 204 2%8 316 330 342 356 364 372 385 402 413 422
54 Naphtha 50.8 158 234 326 345 357 365 374 382 392 405 424 442 460
55 Naphtha 48.8 160 255 364 376 388 396 407 419 432 445 465 480 504
56 Kerosene 47.0 2 10 285 378 394 408 426 438 448 462 874 498 515 526
57 Kerosene 44.9 212 262 400 420 432 447 456 472 484 500 520 536 556
58 Gas oil 42.8 145 300 436 452 462 484 502 513 527 548 576 606 640
a Volume per cent.

and gasolines, the Podbielniak column may give pronounced of the many distillations showed that the most common a p
plateaus throughout the entire distillation range if the reflux parent inaccuracy occurred in errors in temperature levels
ratio is maintained relatively high. Hence in the present cor- due to shifts, between analyses, in the calibrations. of' 'the
relation work, the T. B. P. curves of the various samples are thermocouples and thermometers used in the T. B. P. and
arbitrarily defined in terms of the equivalent smooth curves A. S. T. M. distillations, respectively. This is evident from
drawn through the plateaus, since only in this way can a comparisons of the 50 per cent boiling point temperatures
practical simple correlation basis be obtained. The method
of arriving at these equivalent smooth curves is illustrated in
Figure 1; the derived curve is drawn, leaving on both sides of
the equivalent curve approximately equal areas between the
experimental and derived curves. Although this treatment
was necessary in the case of a large number of T. B. P. distilla-
tions, the majority were carried out with less efficient frac-
tionation where the plateaus were less pronounced though
present to an extent indicating effective fractionation.
Basis of Correlation
Both the A. S. T. M. and T. B. P. distillation data used were
taken from routine analyses and are subject to the normal FIGURE
2. T. B. P. 50 PER CENT TEMPERATURE
us.
experimental errors inherent in such work. A critical survey A. S. T. M. 50 PERCENTTEMPERATURE
798 INDUSTRIAL AND ENGINEERING CHEMISTRY
FIGURE3. T. B. P. (50-20 PER CENT)TEMPERATURE DIF-
FERENCEus. A. S. T. M. (50-20 PERCENT)TEMPERATCRE
DIF-
FERENCE
in Table I. Errors in the absolute temperature level read,
however, will simply displace a distillation curve vertically
and not change its shape. For this reason the relation between
the two types of distillation curves is here developed in terms
of comparative temperature differences for similar portions
of both curves, after establishment of one absolute tempera-
ture point on each curve from a correlation between the 50
per cent temperatures. Selection of the most probable values
for the latter relation is made possible by the large number of
analyses considered.
The estimation of the A. S. T. M. distillation curve by the
method here offered involves seven steps, using the graphical
correlations indicated:
T.B. P. Temp A. S. T. M. Temp Figure No.
50% 50% 2
50-20%0 50-20% 3
lO-l%
E10-initial
E R B P. 4
5
G
98-5070 End pojnt-50% 7
98-90% E n d point-90% 8
Thus after the 50 per cent vaporization tem-
perature for the A. S. T. M. curve has been
estimated, the temperatures corresponding to
other percentages are determined as differ-
ences from this fixed base point. The proce-
dure &ems laborious, but in case it is not
necessary that all points on the A. S. T. M.
curve be established (for example, when
only the initial boiling point and end point
are required) the procedure may be
shortened.
As stated previously, the T. B. P. data
used were taken from routine analyses carried
out over an extended period on various ex-
perimental projects. Inmost of those analy-
ses it was not practical to carry the distilla-
tions t o an unusually small amount of still
bottoms. Most of the distillations, how-
ever, were carried sufficiently far so that
the 98 per cent distilled points were attained ,
1 More usable oopies of the oorrelation graphs may be
obtained b y writing to the Stone & Webster Engineer-
ing Corporation.
Vol. 33, No. 6
or could be closely estimated. This seems satisfactory for
the present purpose.
Derivation of Relations
A. S. T. M. 50 PERCENTTENPERATURE. By the method
of correlation adopted, it was necessary to establish only one
absolute temperature point on the A. S. T. M. curve. The
required relation was obtained from consideration of the
mid-boiling points on the two curves, and is shown on
Figure 2 where the 50 per cent A. S. T. M. temperature (cor-
rected for loss) is expressed as a function of the T. B. P. 50 per
cent temperature. For temperatures higher than approxi-
mately 550 F. the curve is of questionable significance, since
the A. S. T. M. 50 per cent points were obtained from distilla-
tions under vacuum, and extensive scattering of the points
around the correlation line was obtained. For lower tempera-
tures the position and slope of the line were firmly established.
In the case of several samples (Table I) the differences be-
tween the 50 per cent temperatures were excessive; the evi-
dence indicates that these were due to errors in the measured
temperature levels resulting from unnoticed variations in the
base calibrations of thermometers or thermocouples. No con-
sistent effect of slope of T. B. P. curve upon the above relation
could be determined.
A. S. T. M. 50-20 AND 80-50 PERCENTTEMPERATURES.
In the investigation of the relative slopes of the T. B. P. and
A. S. T. M. curves between the 20 and 80 per cent points, it
became evident that separate treatment was necessary for
the 20-50 per cent portions and the 5G80 per cent portions.
The relative temperatuqe differences over the 20-50 per cent
sections were found t o depend upon the slopes of the front-
end tail of the T. B. P. curves. The temperature differ-
ence for the 10-20 per cent portion proved to be a satisfactory
correlating variable. Likewise, for the 50-80 per cent por-
tions, the relative temperature differences are dependent upon
the slope of the higher boiling (80-98 per cent) part of the
T. B. P. curve. Figures 3 and 4 show the average correlations
100
5
sz
s
2
-
m
I
50
c
v1
U
0 0
TRUE BOILING POINT (80%-50%), F
4. T. B. P. (80-50 PER CENT) TEMPERATURE
FIGURE DIFFERENCE
vs.
A. s. T. M. (80-50 PERCENT)TEMPERATURE
DIFF~RENCE
June, 1941 INDUSTRIAL AND ENGINEERING CHEMISTRY 799
obtained. In both the specific effect of the temperature level deviation of the experimental points from the correlation is
(as judged by the 50 per cent boiling point of a sample) is approximately 5 O E. ; this is considered very satisfactory
either absent or constitutes a minor variable. since errors of this magnitude may be obtained in both ex-
A. S. T. M. 20-10 AND 10-0 PERCENTTEMPERATURES. perimental distillation curves. The true error of the relation
For a given 20-10 per cent temperature increment on a is probably less. Insufficient data were available for deter-
T. B. P. curve it was found that the temperature difference for mining the trends of the lines beyond the ranges indicated.
the corresponding part of the A. S. T. M. curve is a function One would suspect that the relation would depend, in addi-
of the slope of the main portion (20-80 per cent) of the T. B. P. tion, upon the proportion of the total T. B. P. 98-80 per cent
curve. Likewise, for a given 10-1 per cent increment on the temperature difference obtained in the 90-80 per cent inter-
T. B. P. curve, the corresponding 10 per cent initial-boiling- val. Hence the present correlation should be applied with
point (T. B. P.) portion of the A. S. T. M. curve was found to be caution to abnormal blends whose T. B. P. distillation curves
dependent upon the slope of the 20-80 per cent portion of the are relatively flat in the 80-90 per cent portion, with very long
T. B. P. curve. The use of the 1per cent point on the T. B. P. tails in the 95-98 per cent part; the correlation presented is
based upon products obtained by normal fractionation
processes. As Figure 7 indicates, the increasing effect of
superheating at the A. S. T. M. end point, with decreasing end
point of samples, in giving fictitiously high A. S. T. M. end
points was clearly established statistically. This was par-
ticularly apparent for the available samples having end
points in the range 200-300 F., where the A. S. T. M. end
points were higher than the corresponding points on the
T. B. P. curves. This is consistent with the experience of the
coal tar industry which establishes that the dry point (vapor
temperature at which the last bit of liquid on the bottom of
the flask is just vaporized) is more significant than the end
point (maximum thermometer temperature with continued
heating of the flask) as the criterion of the true final boiling
I I / l / / / l I I / I

100 /
FIGURE
5. T. B. P. (20-10 PERCENT)TEMPERATURE
us. A. S. T. M. (20-10 PERCENT)TEM-
DIFFERENCE y/
PERATURE DIFFERENCE
80
curve, instead of the initial boiling point, was selected because
the former i s more precisely determinable in the distillation.
-
n 70
-
The relations derived are indicated on Figures 5 and 6,and
-6z
Y
60
are limited to so-called stripped petroleum fractions; they
apply with less accuracy to fractions with long front-end + 50
tails, such as gasolines separated in equilibrium with large yi
4
amounts of propanes and butanes. Since the early stages of 40
the A. S. T. M. distillation are somewhat similar to an equilib-
rium flash separation, these correlations are consistent with 30
theoretical expectation that the holding effect for the lowest
boiling constituents would be a function of the slope (and 20
hence average molecular weight) of the main portion of the
10
T.B. P. curve. For both relations i t was again found that
the specific effect of temperature level is a t most but a minor
factor; this is surprising since it would seem that a t the higher 0: ib 2; 3b 4; 50 I50 o,
T BP(iO%-I%),F
8b 9; ,A0 00
temperatures the increased condensate reflux in the A. S. T. M. FIGURE6. T. B. P. (10-1 PERCENT)TEMPERATURE DIF-
distillation might prove a definite factor. Furthermore, i t FERENCE US. A. 8. T. M. (10 PER CENT-INITIAL-BOILING-
would seem that the relation between the 20-10 per cent por- POINT) TEMPERATURE DIFFERENCE
tions would depend to some extent upon the temperature dif-
ference over the 10-1 per cent section in addition to the slope point for low-boiling aromatic products. This superheating
of the 20-80 per cent main portion of the T. B. P. curve. Like- effect is probably most serious in the case of products pro-
wise the relation between the 0-10 per cent portions should be duced by sharp fractionation and having T. B. P. curves rela-
dependent additionally upon the slope of the T. B. P. 20-10 tively flat in the final boiling range. The commercial specifi-
per cent section. However, the experimental data a t hand did cation of dry points instead of end points for low-boiling pe-
not permit establishing these refinements. troleum fractions seems highly desirable.
A. S. T. M. ENDPOINT. The estimation of the end point A. S. T. M. 90 PERCENTTEMPERATURE. Figure 8 gives
(E. P.) of the A. S. T. M. curve proved to be the most a method of estimating the A. S. T. M. 90 per cent point,
difficult problem. After many attempts the correlation after the end point has been determined according to Figure 7.
shown in Figure 7 was obtained. This relates the entire The hooks on the curves shown are due to the effect of super-
latter half of the A. S. T. M. curve with the corresponding heating discussed above. This relationship is not so satisfac-
part of the T. B. P. curve, the 98 per cent point on the latter tory as those previously discussed.
proving the most satisfactory correlating variable. This
procedure enabled more accurate estimation of the A. S. T. M. Application of Method
end point than others attempted where the range was The use of the correlation will be illustrated by a typical
broken up into two or more portions. The average example. Taking the T. B. P. curve data for sample 15 (a low-
800 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 33, No. 6
vapor-pressure gasoline) of Table I, the estimated A. S. T. M. 100 I
/
curve is derived as follows: /
90 I /

T. B. P. Distn. Curve
Estd. -4.S. T. M. Distn.
Curve
Figure
No. I
80
50% temp. 267O F. 5 0 % temp. 266O F. 2
Slope, (80-20%)/60
50-20%
2.72
88' F. 50-20%
...
56O F.
*.
3 I ,
73
80-50 75' F. 80-50% 58" F. 4
37' F. 20-10% 23O F. 5
$I: 90' F. 10%-I. B. P. 57' F. 6
136' F. E. P.-50% 123" F. 7
29' F. E. P.-9070 31' F. 8

The resulting estimated A. S. T. M. curve compares with the

experimental curve :
Initial End
B. P. 10% 20% 50% 80% 90% Point
Estd O F 130 185 208 266 324 358 389
E x d 9 6. 128 179 202 265 329 352 385

l l l l l i l

FIGURE 8. T. B. P. (98-90 PER CENT)TEXPERA-

TURE DIFFERENCE os. A. S. T. M. (ENDPOINT-90
PER CENT) TEhlPERATURE DIFFERENCE

where the plateaus throughout the curve are accentu-

ated. Two points of possible exception may be noted-
in case added fractionation significantly affects the 1 or
98 per cent temperatures over those obtained with the
Podbielniak column. Experimental data on this ques-
tion are needed.
The correlation should prove particularly helpful in
the evaluation of the results of theoretical computa-
tion studies of the effect of changes of reflux ratio, num-
ber of bubble trays, and position of feed introduction,
in the fractionation of complex petroleum fractions.
Such studies permit precise determination of the effect
of changes in only one variable a t a time, a perfection
which is difficult t o realize experimentally. Changes
in the estimated A. S. T. M. distillations of the prod-
ucts should be significant in spite of any question as to
the absolute accuracy. .
FIGURE
7. T. B. P. (98-50 PER C E N T ) TEMPERATURE DIFFEREXCE
T. M. (ENDPOIXT-~O
os. A. S. PERCEST) TEMPERATURE DIFFER- Estimation of T. B. P. Curve from A. S. T. M.
ENCE Curve
By reversal of the above methods it is possible to esti-
Discussion mate the T. B. P. curve corresponding to a given A. S. T. M.
Table I1 compares the estimated and experimental
A. S. T. M. distillation curves for several of the samples of
Table I. The estimated I. B. P. is consistently low for
gasoline and naphtha samples containing appreciable per- TABLE11. COMPARISON OF ESTIMATEDAND EXPERIMENTAL
A. S. T. M. DISTILLATION
ANALYSES
centages of propanes and butanes. It is difficult to establish
Sample A. S. T. &I.Distillation, F.
precisely the absolute accuracy of the correlation. The basic No. Data
7
--
I. B. P. 1 O % Q 20% 50% 80% 90% E . P.'
data were taken from routine work in which only one dis- 1 105 138 153 243 277
tillation was made by each of the two methods of analysis. 117 143 153 227 279
2 153 170 182 251 283
Thus for each specific sample i t is not possible to determine 158 181 186 242 283
3 223 242 245 296 325
the magnitude or sign of the deviations from the "best" dis- 220 240 244 288 316
tillation curves which might have been established by suffi- 4 320 333 336 381 406
324 334 337 382 411
cient check analyses, and hence not possible to determine 6 87 186 182 293 338
100 162 189 287 322
which of the two A. S. T. M. curves, estimated or experi- 6 381 394 402 457 491
mental, is the more accurate in a given case. Recourse 381 397 405 465 497
7 316 327 330 370 403
must be taken t o statistical comparison and to future experi- 318 328 331 364 392
8 462 487 495 574 643
mental investigation. 456 484 49 1 577 648
The correlation is based on T. B. P. distillations made in the 9 ... 177 203 333 371
113 183 210 327 374
Podbielniak column. In general, we believe that the rela- 10 ...
126
208
214
243
250
430
431
470
471
tions found would be essentially unchanged if a more efficient 15 130 185 208 358 389
column had been used, since when properly used, the Pod- 128 179 202 352 385
bielniak column seems to give equivalent-smooth distillation a Volume per cent
curves which are not changed by more powerful fractionation
June, 1941 INDUSTRIAL A N D ENGINEERING CHEMISTRY
curve. The T. B. P. curve thus estimated will be only an ap-
proximation, as consideration of the fundamental relations be-
tween the two types of distillations will disclose: For each
T. B. P. curve a singular A. S.T. M. curve is defined, but a given
A. S. T. M. curve may correspond to several slightly different
T. B. P. analyses. Obviously such an estimation would not
be used to establish the charge-stock T. B. P. curve to be used
in a detailed column computation. However, cases arise in
refinery engineering design where only the A. S. T. M. curve is
available to characterize the charge to a proposed unit. I n
these instances it is desirable that the T. B. P. curve be ap-
proximately defined in order to permit more intelligent esti-
mation of the percentages of products and the temperatures
of operation, the number of plates and reflux ratios being
established from experience. I n the estimation of the T. B. P.
curve from the A. S. T. M., two methods of procedure are
possible: The various relations here presented may be ap-
plied and the A. S. T. M. curve established by successive ap-
proximation, or derived correlation graphs may be drawn
which enable direct estimation of the A. S. T. M. curve from
the T. B. P. analysis.
801
Acknowledgment
The author wishes to acknowledge the valued assistance of
M. W. Riley in compiling much of the data and for aid in the
early stages of the project. Also appreciation is expressed
to the Alco Products division, American Locomotive Com-
pany, in whose laboratory the experimental work was car-
ried out.
Literature Cited
(1) Am. Soc. for Testing Materials Standards on Petroleum Products
and Lubricants, Method D-86-35 (1937).
(2) Baker, R. H., Barkenbus, C., and Roswell, C. A., IND. ENG.
CHEM.,Anal. Ed., 12, 469 (1940).
(3) Docksey, P., J. last. Petroleum Tech., 23, 316 (1937).
(4) Lewis, W. K., and Robinson, C. S., J. IND.ENG.CHEW,14,481
(1922).
(5) Nelson, W.Lo,and Hansburg, M., Oil Gas J., 38, No. 12, 45
(1939).
(6) Peters, W. A., Jr., and Baker, T., IND.ENG.CHEM.,18, 69 (1926).
(7) Podbielniak, W. J., IND.ENO.CHEM.,Anal. Ed., 5, 135 (1933).
(8) Podbielniak, W.J., private communication.
(9) Thiele, E. W.,and Geddes, R. L., IND. ENG. CHEM.,25, 289
(1933).

Treatment of Tannery Wastes

with Flue Gas and Lime
H. B. RIFFENBURG AND W. W. ALLISON
Virginia Polytechnic Institute, Blacksburg, Va.

EATHER manufacture progresses in stages which are
L in general the same for all skins, though process detail
may vary with the character of leather to be produced.
The steps in leather manufacture take place in the following
order: (1) curing of the skin, (2) soaking, (3) liming, (4)
unhairing, (5) deliming or bating, (6) tanning, (7) finishing.
A waste product is produced in each step. The wastes from
steps 1, 2, and 4 contain small amounts of material and are
easily disposed of. The wastes from the other steps contain
large quantities of colloidal and suspended matter, and some
of them are highly colored. This condition makes their dis-
posal difficult because of their tendency to clog sewerage sys-
tems or to pollute streams.
Properties of the Wastes
LIMES. These solutions contain certain decomposed skin
constituents, various forms of sulfur and sulfur
compounds from the decomposed proteins, emul-
sified fatty bodies of the flesh, carbonates, salts,
and other substances. The pH of these solutions
SPENTTAN LIQUORS(OF VEGETABLE ORIGIN). These
solutions are polydispersed systems in water solutions con-
taining particles ranging in size from molecular to coarse
suspensoids, and are composed of some tannin, nontans, and
hydrolyzed substances derived from the wood during extrac-
tion of the tanning extract and from the skins with which the
solutions came in contact. These solutions contain about
33,500 p. p. m. of solid matter, which has an average oxygen
demand of 13,600 p. p. m. and amounts to about half of the
total demand of tannery wastes. They are thought to possess
a negative charge and have a pH of about 5.4 (Table I).
Wilson (6) pointed out that tan liquors vary extensively in
color with pH. He suggested that tan solutions derived from
the regular vegetable sources might be standardized for use
as pH indicators. This property of the waste tan liquors
makes the color appear much more intense when the wastes
TABLEI. CHEMICAL ANALYSISOF TANNERY WASTESAMPLDS
Uirlr.
--_x1

averages 11.5, and their biochemical oxygen de-

mand, 1900 p. p. m. The amount of suspended
solids in the lime solution is about 17,000 p. p. m.,
Waste

Soaka
Specific
Gravity

1.004
1 002
7.45
11.5
Total
PH Waste
Qal./day
9,000
-Total
Solids

8,352
3,300 36,488
Ni-

88 6
28.6
Sub-

496
160
600
1,700
Oxygen
trogen stance B. 0. D. Consumed
Parts p e r mzllion
900
1,750
dissolved solids 20,000 p. p. m., and hide sub- !%Beswash 1.004 8.1 4,000 8,944 225 1257 535 goo
stances 160 p. p. m. Unhairjng and
fleshing 1.000 10 5 3,860 3,860 166 929 3,000 1,050
DELIMINGOR BATING.~ O L U T I O N S . These Deliming and
1.001 6.7 600 5,644 325 1818 1,650 900
wastes are of relatively small quantity but highly Tabn&:t
concentrated. They contain over 6000 p. p. m.
of solids and have an oxygen demand of around
E%
Add POOI
zpe ;A!:
1.013

1.015
5 4

1.8
18,000 33 496
2..':800
1,800
;$$;!
13,460
43.
25.7
204
ib5
13,600 22,000
3,750
2,750
2,000
15,800
9,125
5,425
1600 p. p. m. J