develop more accurate prediction of flash points

2
PROCESS DEVELOPMENTS Develop more accurate prediction of flash points New calculation improves estimating minimum flash temperature for petroleum products K" ffi . §,&§V§,&K&.§§&*i&ffi, {,*rp. l-lr:i", fr rj*r*1, '::""'"':'"'):':he flash point of a hydrocarbon mixture is a very imporrant ,¡: parameter especially with regards to the safe handiing, storage r,, and transportation of hydrocarbons. By definition, the flash point is the minimum temperature at which vapors from the mixtu¡e would produce a momentary flash when subjected to a standard test flame. The two widely accepted test methods for fiash-point determi- nations are Pens§ Martins and Abels flash apparatus. Flash points are extremely dependent on the lighter ends of any petroleum fraction. Methods. Riazi and Daubert determined that for petroleum fractions, ASTM 10% is more appropriate for calculating flash points. They proposed two equations for predicting flash points: tfr, = -o.ot+568+(2.84e4t lr,)*(,.r0: " ro-')r" r, (r ) where Q and I1 are in'R ( is the ASTM 10% temperature of petroleum fraction. Tf = -t24.72+0.707047,(+ *O { are in'F) Q) Samples Density Calculated flash points of diesel with varying ASTMs for different refineries ASTM distillation IBP 5o/o 'l0o/o 50% 90!o For fractions up to 500"F, Eq. 2 predicts flash poinrs with reasonable accuracy, while Eq. 1 is applicable for higher boiling point fractions. Better prediction method. A slight modification can predict the flash point value even more closer to the experimental value. Instead of taking ASTM 107o, if one replaces the value of ASTM 10% with the average of the initial boiling point (IBP), 5o/o and 10% in Eq. 1, the calculated flash point values obtained are very much closer to experimental values. Táble 1 summa¡izes the obser- vations for diesel flash points ofvarious units from a refinery. rf rr: -o.or+s6s +(z.a4t47lr^"0)* (r.loe, ro-3 )ua"o (3) where f,,,¿ ir {uun . 5% + 10%)/31 in 'R and ! is the fiash in "R. Error analysis. Táble 2 lists the error analysis berween experi- mental and predicted values using Eq. 3. Exp Dl D2 Mod 95o/o FBP Flash pt Flash, "C Flash2, 'C f (0,5,10) Flash, "C Diesel 1 816.5 162 Diesel 2 824.3 194 Diesel 3 853.9 230 Diesel4 sil.¡- - is) Diesel5 8058 152 Exp = s¡ps,¡r..,, flash point, 'C D1 = flash point estrmated from Eq. l D2 = f ash po¡nt estrmated from Eq. 2 185 195 220 233 255 268 260 275 '168 176 256 213 314 317 263 JJ I 330 367 3to 348 352 346 381 385 358 360 392 3g7 372 55 b9 97 103 45 66.96 90.40 10e.1á 112.59 54.10 63.38 g0.24 lti,..lg 11g.34 49.94 180.67 57.32 215.67 Aó.0é zsi.óo roo.jg 257.33 103.74 '165.33 46.56 25 Modfed=Flashpointestim¿tedfrommodifredEq. l,bysubstituting(1BP+5%+'T0%)/3 in place of ASTM 1 0% Error analysis between experimental and predicted values Samples D1 D2 Mod 20 15 t' o Elo 5 0 Diesel 1 Diesel 2 Diesei 3 Diesel 4 Diesel 5 1'l .96 21.40 12.16 9.59 e.io 8.38 21.24 17.99 16.g4 4.94 2.32 1 1.09 i.¡s 0.7 4 1.56 Diesel 2 Diesel 3 Diesel 4 Diesel 5 iesel 't HyDRocARBoNpRocrss\c ruavzooa |, ¡

Upload: nxvn

Post on 10-Feb-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

7/22/2019 Develop More Accurate Prediction of Flash Points

http://slidepdf.com/reader/full/develop-more-accurate-prediction-of-flash-points 1/2

PROCESS DEVELOPMENTS

Develop more accurate predictionof flash pointsNew calculation improves estimating minimum flash temperaturefor petroleum products

K" ffi . §,&§V§,&K&.§§&*i&ffi, {,*rp. l-lr:i", fr rj*r*1,

'::""'"':'"'):':he flash point of a hydrocarbon mixture is a very imporrant

,¡: parameter especially with regards to the safe handiing, storager,, and transportation of hydrocarbons. By definition, the flash

point is the minimum temperature at which vapors from the mixtu¡e

would produce a momentary flash when subjected to a standard testflame. The two widely accepted test methods for fiash-point determi-nations are Pens§ Martins and Abels flash apparatus. Flash points are

extremely dependent on the lighter ends of any petroleum fraction.

Methods. Riazi and Daubert determined that for petroleumfractions, ASTM 10% is more appropriate for calculating flashpoints. They proposed two equations for predicting flash points:

tfr, = -o.ot+568+(2.84e4t lr,)*(,.r0: " ro-')r" r, (r )

where Q and I1 are in'R( is the ASTM 10% temperature of petroleum fraction.

Tf = -t24.72+0.707047,(+ *O { are in'F) Q)

Samples Density

Calculated flash points of diesel with varying ASTMs for different refineries

ASTM distillation

IBP 5o/o 'l0o/o 50% 90!o

For fractions up to 500"F, Eq. 2 predicts flash poinrs withreasonable accuracy, while Eq. 1 is applicable for higher boilingpoint fractions.

Better prediction method. A slight modification can predictthe flash point value even more closer to the experimental value.Instead of taking ASTM 107o, if one replaces the value of ASTM10% with the average of the initial boiling point (IBP), 5o/o and10% in Eq. 1, the calculated flash point values obtained are verymuch closer to experimental values. Táble 1 summa¡izes the obser-

vations for diesel flash points ofvarious units from a refinery.

rf rr: -o.or+s6s +(z.a4t47lr^"0)* (r.loe, ro-3 )ua"o(3)

where f,,,¿ ir {uun . 5% + 10%)/31 in 'R and ! is the fiash point in "R.

Error analysis. Táble 2 lists the error analysis berween experi-

mental and predicted values using Eq. 3.

Exp Dl D2 Mod

95o/o FBP Flash pt Flash, "C Flash2, 'C f (0,5,10) Flash, "C

Diesel 1 816.5 162

Diesel 2 824.3 194

Diesel 3 853.9 230

Diesel4 sil.¡- - is)Diesel5 8058 152

Exp = s¡ps,¡r..,, flash point, 'CD1 = flash point estrmated from Eq. l

D2 = f ash po¡nt estrmated from Eq. 2

185 195

220 233

255 268

260 275'168 176

256

213

314

317

263

JJ I

330

367

3to

348

352

346

381

385

358

360

392

3g7

372

55

b9

97

103

45

66.96

90.40

10e.1á

112.59

54.10

63.38

g0.24

lti,..lg

11g.34

49.94

180.67 57.32

215.67 Aó.0é

zsi.óo roo.jg

257.33 103.74

'165.33 46.56

25

Modfed=Flashpointestim¿tedfrommodifredEq. l,bysubstituting(1BP+5%+'T0%)/3in place of ASTM 1 0%

Error analysis between experimental andpredicted values

Samples D1 D2 Mod

20

r¡ 15

t'oElo

5

0

Diesel 1

Diesel 2

Diesei 3

Diesel 4

Diesel 5

1'l .96

21.40

12.16

9.59

e.io

8.38

21.24

17.99

16.g4

4.94

2.32

1 1.09

i.¡s0.7 4

1.56

Diesel 2 Diesel 3 Diesel 4 Diesel 5iesel 't

HyDRocARBoNpRocrss\c ruavzooa |,¡

7/22/2019 Develop More Accurate Prediction of Flash Points

http://slidepdf.com/reader/full/develop-more-accurate-prediction-of-flash-points 2/2

PROCESS DEVELOPM ENTS

Comparison of predicted flash points withvarious dieselfractions using the f (0,5,10) method

Samples r (0,5,10) Dl D2 Mod Exp

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5

Sample 6

Sample 7

Sample 8

Sample 9

Sample 10

Sample 11

161 .3 52

162.6 49.8

163.6 49.8

164.6 53.4

166.3 51 .2

169 57.5

110 76.5

177.3 88.6

17g 66.29

296.6 124.1

31 3.6 13'r .6

47.8 +31

45.7 44.6

45.6 45.3

4g.2 46.1

41.1 47.2

53.4 49.1

13.e +:e.8

88.1 50.8

62.6 56.1

139 122.5

'1 51 129.5

41

45

46

42

43

48

40

39

49

138

141

160

140

160

140

160

140

120O.^100Eo ót¡E360E

4020

0

As is quite evident fromTable 2 and Fig. 1, using f (0, 5,10)-Eq. 3-in place of only 10% yields a flash point value whichis close to the experimentally determined value. AIso, from a saferyviewpoint, predicting a lower flash value is always safer rvhenhandling and transporting petroleum fractions. To further test its

compatibiliry a comparison using different fractions rvas carriedout. Table 3 and Figs. 2-4list the resulrs of this study.

From Figs. 2-4, the modified Riazi and Daubertt eqi-ration is

accurate for all fractions with minor deviations from the experimenta-l

values in the mid range. Most modern simulation and chemical soft-ware packages take into account Daubert equation for predicting flash

§r#Kfip§§- Iis"{r* #*rx

jj,ii,,t:.i &§úÉ*JJi* f**+¿+*s:**; i;¿* .&¡¿;*,¡.;"f

6Samples

poir.rts of petroleum fractions. By considering the average of IB!,5o/oand 10%, o.ne can predict flash points even more accurately. Since,

at the IBB we get the first drop of the fraction, it is at this point that

the flash is predominant. So, the average of the first three points willgive a more accllrate prediction of the flash point. HP

LITERATURE CITEDr Riazi, M. R. and T. F.. Dauberr, "Predicting flash ancl pour points," Hylrocdrbon

Prorcsing,Scpternber 1 987, pp. 81-83.

K. R. Ramakumar is a product on engineer at Indian Orl

Corp. Ltd., Gujarat, lnd a. He has a BS degree in chemical engr-neering from Sardar Val abhbhai National lnstitute of Technology,

Surat, Gularat, where he was a bachelor of chemica enqlneer nq

120o"-100'e 80o

860r40

¿U

0

6

Samples

t)

Samples120

120o"-100

'6 80ot360iE

40

20

0

I

l;;;pl '-It-aroeatl--v

,r'\ \d/<*r*#l"L---rz_-\.-

I

--l +ozI rr- fxp

á.

lLaroeazI\,r I

,/\l \/*- -*,1

I

_l+.No¿ l_l.c- exp

I

III

J

Select 1 79 at www.HydrocarbonProcessing.com/RS