New Opportunities for Co-Processing Renewable Feeds in ... · PDF fileCo-Processing Renewable...

Catalagram 103 Spring 2008 1

Brian WatkinsSupervisor,Laboratory Technology,Advanced Refining Technologies

Charles OlsenWorldwide Technical ServicesManager, Advanced RefiningTechnologies

Kevin SutovichSenior R&D Chemist,Grace Davison RefiningTechnologies

Natalie PettiVandelay Management

he use of renewable or bio-based sources of feed to pro-duce fuels is becoming more

widely employed as a means ofdecreasing dependence on non-renewable fossil fuel sources. Thereare typically three common produc-tion routes for biodiesel. Fuel which isproduced by the FAME (Fatty AcidMethyl Ester) process to meet a fuelspecification of ASTM D6751 is con-sidered biodiesel. Fuels producedfrom biological material using thermaldepolymerization to meet ASTM D975or ASTM D396 are considered renew-able diesel. Fuels that are producedwhen vegetable oils or animal fats areprocessed in traditional refiningprocesses are considered co-

processed renewable diesel. Somecommon sources of renewablefeeds are those produced for foodgrade oils such as soybean, rape-seed and other vegetable oils. Thetraditional process for introducingthese sources into the diesel pool isto use the transesterification reac-tion for breaking the glycerol fromthe fatty acid chains. This reactionrequires the use of an alcohol (suchas methanol) and a catalyst (suchas sodium or potassium hydroxide,NaOH or KOH) in order to break thelong chained fatty acids apart fromthe glycerin molecule. (Figure 1)

New Opportunities forCo-Processing RenewableFeeds in RefineryProcesses

T

www.e-catalysts.com2

These long chained fatty acids(Fatty Acid Methyl Esters or FAME)are most commonly between 16-20carbons in length with a few excep-tions. The long carbon chains aresimilar in structure to the 16-20 car-bon chains found in typical dieselexcept that FAME compounds con-tain almost no sulfur or nitrogen andno aromatics, which make themexcellent blending components inthe diesel pool. FAME products,however, do have a high percentageof oxygen that enables them to betracked directly at the pump. FAMEproduction occurs in separate facil-ities and requires a distributioninfrastructure to transport the biofu-els to a location where they can be

blended with conventional diesel.Discrete storage and supportinginventory of the biofuels would berequired to guarantee continuous sup-ply. Using one of ART’s high perform-ance catalysts, we have been able tocapitalize on renewable sources offuel by bypassing the purchase ofFAME products and instead process-ing the raw materials through conven-tional hydrotreating equipment to pro-duce a higher quality ULSD product.Refiners who use the co-processingmethod would have exact knowledgeof the bio-based fuels that are incor-porated into the diesel pool whichhelps to ensure the finished blendquality.

Background

Taking a detailed look at the com-pounds found in typical renewableoils shows that these oils can betreated as classic petroleum basedcompounds. ART analyzed severaldifferent renewable sources of fuelsin order to better understand thepossible chemistry that would occurif they were processed in a conven-tional hydrotreater. Soybean, palmand rapeseed oils were studied, asthese materials are readily avail-able. A table listing various oils andtheir structural makeup is shown inTable I. Bio-based sources of oilscan be of significant value whenincorporated into the ULS dieselpool due to the low contaminantconcentrations and high cetanenumber of the resulting products.

Components and Description

The major saturated and unsaturat-ed fatty acids found in these oilsconsist of palmitic acid (C16:0),linolenic acid (C18:3), linoleic acid(C18:2), oleic acid (C18:1),eicosenoic (C20:1) and erucicacids (C22:1) in varying percent-ages. If each of these fatty acidchains were to be separated fromthe glycerol molecule they can beincluded easily into the diesel pool

CH3OH

R1

R3

R2

CH2 O

O

C

CH2 O

O

C

CH2 O

O

C

Catalyst

R1

R3

R2

CH2 O

O

C

CH2 O

O

C

CH2 O

O

C

CH2 OH

CH OH

CH2 OH

Figure 1Transesterification reaction

Fats and OilsMolecular wt.Tallow, wt.%Lard, wt.%Butter, wt.%Coconut, wt.%Palmkernel, wt.%Palm, wt.%Safflower, wt.%Peanut, wt.%Cottonseed, wt.%Maize, wt.%Olive, wt.%Sunfower, wt.%Soy, wt.%Rapeseed, wt.%Mustard, wt.%Codliver Oil, wt.%Linseed, wt.%Tung, wt.%

C4:0

Butyric

C6:0

Caproic

C8:0

Caprylic

C10:0

Capric

C12:0

Lauric

C14:0

Myristic

C16:0

Palmitic

C16:1

Palmitoleic

C18:0

Stearic

C18:1

Oleic

C18:2

Linoleic

C18:3

Linolenic

C20:0

C22:0

Arachydic-

Behenic &

Mono-

unsaturated

acids

<C16:1

C20:1 C22:1

Arachidonic-

Erucic &

others

3.5 1.5

8

3

25

85

3

4848.5

3

1

11

16

17

3.5

0.5

1.5

1

1

0.3

4

0.2

27

26

30

8.5

7.5

395

52

7

19

9

13

6

7.8

3.5

3

10

6

2

2

3.5

0.5

1.5

1.5

2

0.4

0.2

14.5

24.1

13

12

2.5

2

3.5

2.2

4.5

2

2.5

2

4.2

2.5

2

1.5

0.5

5

40.7

45.2

26

6.5

14

46

76.4

52

31

40

68

18.7

26

13.5

39.5

28

17.3

8

2

10.3

3

2

1

7.5

16.2

27

44

45

12

69.4

51

17

12

16

12

0.3

2.5

0.85

0.5

2.5

1

1

56.3

36

421

1.5

0.7

1.65

1.5

7.5

0.5

1.4

7

0.9

0.5

0.3

5

7.5

8

55

80

88 116 144 172 200 228 256 254 284 282 280 278 326 226 324

Table IComposition of Various Oils and Fats1,2


as normal paraffin components inthe 500-650˚F boiling range. Thesen-paraffins can be of significantvalue for ULSD as they have typi-cal cetane numbers ranging from95 to 110, which can provide a sig-nificant boost for those refinersprocessing feeds with lowercetane (i.e. FCC LCO’s). The typi-cal diesel hydrotreater has only asmall effect on cetane with cetaneupgrade of about 2-4 numbers.

In the unbroken, unprocessed form,the triglyceride molecules are signif-icantly outside the diesel pool rangeas they have molecular weights of700 or greater, while the typicaldiesel pool has a molecular weightof less than 400. The simulated GCanalysis of soybean oil is shown inFigure 2 and indicates that thesematerials have a fairly narrow distil-lation showing up in the C50-C60range. Note that simulated distilla-tion of these compounds is basedon the carbon content and molecu-lar weight of the materials and thiscan sometimes skew the estimatedboiling points. Biofeed sources typ-ically have a true boiling point that ismuch lower than that reported bysimulated distillation equipment dueto molecular weight interference. Inthe unconverted state these triglyc-eride molecules cannot be blendedinto the diesel pool at the levelsrequired to meet renewable fuelstandards.

Another concern is that theserenewable feed sources caninclude various contaminants. Ananalysis of several different bio-feed sources has indicated thepresence of contaminants such assodium, calcium and phosphorus.Table II shows the measured con-taminant levels of the soybean,rapeseed oil and palm oils used inthis work. The palm oil shows notrace impurities, which indicatesthat it has been previouslyprocessed while the soybean andrapeseed oils have not. In the fore-seeable future it is unlikely that theuse of these renewable sources

Chromatogram:

0 5 10 15 20 25 30 35 40

Retention time (min)

0.000e+000

2.000e+005

4.000e+005

6.000e+005

8.000e+005

1.000e+006

1.200e+006

1.400e+006

1.600e+006

1.800e+006

2.000e+006

2.200e+006

Sign

al

100200300 400 500 600 700 800 900 1000 1100

Boiling point (°F)

Star

tTim

e

End

Tim

e

IBP

FBP

Figure 2Simulated Distillation (D2887) of Soybean oil

SoybeanOil

Rapeseed Oil

*Palm Oil

API (°) 21.58 21.98 22.98Specific Gravity (g/cc) 21.6 22.0 23.0Sulfur, ppm 0 3 1Oxygen, wt.% 10.5 10.62 11.33Nitrogen, ppm 3.9 16 1.6 D2887 Distillation, °F

IBP 702 710 6255% 1059 1065 941

10% 1069 1077 102630% 1090 1095 106250% 1102 1106 107970% 1111 1115 109090% 1183 1188 114695% 1232 1238 1197FBP 1301 1311 1302

Metals Contamination, ppm

Na 2.0 4.7 0.0Ca 3.0 13.8 Mg 0.9 0.3 P 6.5 4.0

Zn 0.1 0.6 Al 0.1 0.2 Mn 0.0 0.1

* Oil was pre-processed to remove impurities

0.00.00.00.00.00.0

Table IIAnalysis of Different Biofeed Sources


would exceed 20% in conventionalhydrotreating applications, whichwould bring the level of all of thesecontaminants down to 2.5 ppm orless. At these levels ART’s highcapacity guard materials and GraceDavison specialty catalysts arecapable of protecting the down-stream high activity catalysts fromthese damaging poisons.

Since these renewable feeds arederived from a biological source,they also contain a high concentra-tion of oxygen. For the materialslisted in Table I the oxygen contentranges from 10 to 15%, and isentirely dependent on the lengthand degree of saturation of the fattyacid chains.

This quantity of oxygen is important,as under normal hydrotreating con-ditions the oxygen will react with thehydrogen to form water. This water,if generated in a significant enoughquantity, may cause problems suchas weakening the catalyst supportor redistribution of the active metalsand loss of surface area. At theexpected blending ratios of 10%,the oxygen content is around 1 to1.5 wt.%, and even if all the oxygenis converted, this is unlikely to gen-erate enough water to be a signifi-cant problem.

General Co-Processing Ideas

Looking at these compounds from ahydrotreating perspective, the bio-feeds can be classified as monoand di-olefins, since a majority ofthese compounds have one or twodouble bonds per fatty acid chainwith a few having three. Using thenumbers listed in Table I, palm oilhas an average of 1.1 double bondsper chain, rapeseed oil has 1.3 andsoybean oil has 1.7. It has beenwidely established that the olefinsaturation reactions occur quiterapidly and tend to happen near thetop of the catalyst bed in ahydrotreater. The reactions go tonear completion at typicalhydrotreating conditions, and willgenerate between 130-150 BTU’s

per standard cubic foot of hydrogenconsumed. The reaction pathways tohydrotreat the bio-oils is shown inFigure 3.

In the first step of the reaction, theunsaturated fatty acid chains arequickly converted into fully saturatedn-paraffins. The second reaction thatmust occur in order to ensure that thecompounds will be of the appropriatesize for the diesel pool is the breakingof the fatty acid chains away from theglycerin molecule which requirescleavage of a carbon – oxygen bond.

CO

O- =-

CO

O- =-

CO

O- =-CH

C

C

H

HH

H

CO

O- =-

CO

O- =-

CO

O- =-CH

C

C

H

HH

HH

H

H

H H

H

HH H

HH H

HH

HHH

H-

-

-CHCC

HH

H

HH

H

H

+ 6H2

HEAT

HEAT

+4.5H2

+ 3 O2

+ 6H2

+ 3 H2O

Figure 3Pathways to Hydrotreating of Renewable Oils

Figure 4Chromatogram of Sulfur, Carbon & Oxygen

in a Diesel Feedstock

This reaction is expected to takeplace at a rate similar to that of sul-fur compounds such as sulfidesand disulfides. In order to verify thatthis is indeed the reaction that istaking place, several different feedsources were analyzed for oxygenin order better understand wherethey are likely to be distributed inthe feeds. This analysis has adetection limit of approximately 2.6ppm oxygen. Figure 4 is a Carbon-Sulfur-Oxygen chromatogram for adiesel feedstock that contains 263ppm oxygen.

Figure 5Chromatogram of Sulfur, Carbon andOxygen in a Co-Processed Product

Analytical techniques using GC-AED have shown that in normalULSD operation, no oxygen isdetected in the products at levelsbelow 500 ppm sulfur. This can beseen in Figure 5 which shows theanalysis of one of the co-processedproducts which has a total sulfur of31 ppm and less than 1 ppm nitro-gen.

Pilot Plant Testing of RenewableOils

In order to understand the processfor co-treating renewable fuel com-ponents in a hydrotreater, AdvancedRefining Technologies completed anumber of pilot plant studies. Awide range of ULSD operating con-ditions were investigated to deter-mine if there is an optimal operatingwindow for processing these typesof feeds. The conditions includedhydrogen pressures from 450 to1100 psia and hydrogen to oil ratiosof 1000 to 3000 SCFB.

The three different renewablesources of oil were blended in sepa-rately with a typical straight run (SR)diesel feedstock. The renewablecomponent level was varied from10% to 80% and hydrotreated overthe range of processing conditionslisted above. The SR componentproperties are listed in Table III, alongwith 5 different blends of the biocomponents. As can be seen in thetable, the effects of blending in therenewable source are to dilute the

sulfur, nitrogen and aromatic contentsand decrease the API gravity.

ART then conducted testing on blendscontaining the various renewable

sources of feedstock at the target-ed 10% level. Figure 6 summarizessome of the results of the testing.The testing showed that soybeanand rapeseed oils behave similarlywhen co-processed in a SR diesel.The feed blends required essential-ly the same temperature for 10 ppmproduct sulfur, and the apparentactivation energy (temperatureresponse) for the two feed blends issimilar to that of the SR feedstockalone. The palm based oil, whichhad been previously processed,was apparently easier to treat to lowsulfur diesel levels, but for 10 ppmproduct sulfur the temperature wasonly slightly lower than that for theSR feed. The apparent activationenergy for this feed blend was lowerthan the SR component indicatingthe temperature response in the unitwas lower. Comparing the feeds at

SR Oil

10%Soybean

10%Palm

10%Rapeseed

40%Soybean

80%Soybean

API 34.44 33.03 33.50 33.29 29.24 24.38 Specific Gravity, g/cc

0.852 0.859 0.857 0.858 0.879 0.907

Sulfur, wt.% 1.123 1.083 1.092 1.042 0.670 0.210

ppmNitrogen, 130 82 75 67 47 16

wt.%Oxygen, 0.0 1.2 1.2 1.1 4.7 8.9

Aromatics,wt.%

Mono 17.76 15.85 15.87 15.86 10.32 3.34

Di 7.39 6.60 6.60 6.60 4.29 1.39

Poly 2.1 1.87 1.88 1.87 1.22 0.39

Total 27.25 24.32 24.34 24.33 15.84 5.12 D2887Distillation,˚F

IBP 222 209 209 210 239 329

10% 477 465 459 465 498 571

30% 579 559 557 559 579 1009

50% 613 595 592 594 631 1119

70% 643 632 628 630 1108 1130

90% 681 720 688 715 1130 1135

FBP 740 1127 1121 1127 1134 1139 CloudPoint, ˚F

19.9 21.7 24.1 22.8 22.0 19.1

Cetane Index 53.8 50.6 51.2 50.9 46.4 NA

Table IIIStraight-Run (SR) and Bio-Blend Analyses



ultra low sulfur levels suggests theco-processing of the renewable oilshas only a small effect on the per-formance of the hydrotreater. Thedata indicates that the least reactiveblend contains rapeseed oil, fol-lowed by the soybean oil blend andfinally the palm oil blend. The differ-ence, however, is only about 10˚F for<10 ppm product sulfur.

Looking at other diesel productproperties that are important torefiners reveals that there is a boostin the product cetane index byalmost 2 numbers. This increase incetane was seen over the widerange of conditions tested and is areflection of the normal paraffinsfrom the renewable oil discussedabove. Figure 7 summarizes someof the cetane index results achievedat higher pressure for each of thebio-feed blends.

The impact of low pressure opera-tion on cetane improvement isshown in Figure 8 for a 10% renew-able feed blend. Not surprisingly,lower pressure operation results in alower cetane index for the SR feed,but the addition of the renewable oilagain provides a consistent twonumber increase in cetane index.

This is a good indicator that thelarge fatty acid molecules are beingbroken down into the three individ-ual fatty acid chains via the break-ing of the C-O bonds. Figure 9compares the D-2887 distillationchromatograms of the SR productsat 10 ppm sulfur to that of the co-processed products, and it is evi-dent that there is an increase in theconcentration of the n-paraffinsbetween 500°F and 600°F boilingpoints. It is this increase that yieldsthe significant boost in cetane.

The hydrotreating of the bio-blend-ed oil results in a product that nolonger contains material in the C50to C60 range. This is consistent withthe theory that the individual fattyacid chains are being broken apartto hydrocarbons of similar size tothose in the SR diesel.

570

580

590

600

610

620

630

0 50 100 150

10% Levels

Product Sulfur, ppm

WA

BT,

˚F

SR 10% Soy 10% palm 10% rapeseed

Figure 6Results of Various Renewable Components

55

56

57

58

59

60

61

62

63

570 580 590 600 610 620 630

Prod

uctC

etan

eIn

dex

WABT, ˚F


Figure 7Cetane Boost when Co-Processing Bio-Feeds

at High Pressure

Prod

uctC

etan

eIn

dex

WABT, ˚F

55

56

57

58

59

60

590 600 610 620 630 640 650

SR Renewable

Figure 8Cetane Boost when Co-Processing Bio-Feeds

at Low Pressure

Due to the addition of unsaturatedchains from the bio component,there is expected to be an increasein hydrogen consumption to satu-rate these C=C bonds. With thisadditional hydrogen usage, it isimportant to also be aware of anychanges in product aromatics, asreactions to saturate aromatics arehigh consumers of hydrogen andwould compete with the saturationreactions under hydrogen limitedconditions. Figure 10 summarizesthe product aromatics for one of thebio-blended feeds. The total aro-matics are consistently two num-bers lower than the SR feed, whichis the same as the actual differencein the total aromatic content of thetwo feeds. The lower aromatic con-tent of bio-blended feeds allows therefiner to achieve lower product aro-matic content, which may be valu-able as future regulations mayrequire a lower total aromatic limiton diesel fuel.

At lower pressure and H2/oil ratios,the total aromatic content shows asimilar response, with two numberslower total aromatics when co-pro-cessing bio-based feedstock. Withreduced operating pressure, thereis a decrease in the aromatic satu-ration ability of the catalyst, and thepossible use of renewable oils mayhelp to offset this.

Cloud point specifications varybased on the location of the refineryand the end user of the fuel.Although the blending of the biocomponents yields improvements inthe diesel cetane, there is the con-cern about the biofeed baseddiesel cloud point. It is widelyknown that n-paraffins have a signif-icantly higher cloud point than othersame carbon number hydrocar-bons. Since hydrotreating convertsthe fatty acid chains into longchained n-paraffins, the cloud pointof the mixture will increase. Figure11 summarizes the product cloudpoints after hydrotreating the threeblended feeds. The SR feed isincluded for comparison.

Chromatogram:

0 5 10 15 20 25 30 35 40


0.000e+000

1.000e+005

2.000e+005

3.000e+005

4.000e+005

5.000e+005

6.000e+005

7.000e+005

8.000e+005

9.000e+005

1.000e+006

Signal

100200300 400 500 600 700 800 900 1000 1100

Boiling point (°F)

Star

tTim

e

End

Tim

e

IBP

FBP

Chromatogram:

0 5 10 15 20 25 30 35 40


0.000e+000

2.000e+005

4.000e+005

6.000e+005

8.000e+005

1.000e+006

1.200e+006

1.400e+006

1.600e+006

1.800e+006

Signal

100200300 400 500 600 700 800 900 1000 1100

Boiling point (°F)

Star

tTim

e

End

Tim

e

IBP

FBP

Figure 9Boiling Point Comparison Between SR and 10% Bio Blends

WABT, ˚F

Tota

lPro

du

ctA

rom

atic

s,w

t.%

5

10

15

20

25

30

570 580 590 600 610 620 630

SR 10% palm

Figure 10Comparison of Total Aromatics of SR Oil with

10% Renewable Oil at High Pressure

WABT, ˚F

Pro

du

ctC

lou

dP

oin

t,˚F


16

18

20

22

24

26

28

30

32

570 580 590 600 610 620 630

Figure 11Cloud Points of Bio-Blended Feeds Compared to SR Feed


Straight-run product@ 10 ppm sulfur

Co-processed product@ 10 ppm sulfur


Processing the SR feed has essen-tially no impact on the productcloud point and the products are allwithin a few degrees of the feedcloud point shown in Table III. Therenewable containing feeds all haveslightly higher cloud points com-pared to the SR feed (see Table III),and after hydrotreating the cloudpoint increases by 6 to 10°F. Thisincrease in cloud point can be sig-nificant especially in cold weatherclimates. By using other technologyprovided by ART, the problem ofincreased cloud point can bereduced or eliminated.

Co-Processing in FCC Units

Another option for refiners could beto co-process bio components influid catalytic cracking units(FCCUs). Unlike hydrotreater unitswhere catalyst cannot be changedwithout taking the unit out of serviceto reload the reactor, continuousreplacement of catalyst in the FCCunit enables the refiner to adjust thecatalyst formulation to optimize

yields. Characterization of the bio-feedstock and an understanding ofthe refiner’s objectives can allow thecatalyst supplier to develop a formula-tion to maximize profits.

Pilot Plant Testing of RenewableOils

To illustrate the impact on FCCyields with the incorporation of veg-etable oil feedstocks into FCC feed,a pilot plant study was conductedby starting with a composite feed ofVGO and resid with propertiesshown in Table IV, then the soybean,palm and rapeseed oils used in thehydrotreating testing were blendedto 0%, 7.5% and 15% concentra-tions. The hydrotreated VGO in thistable is used in a later study to com-pare the effects of different basefeedstocks.

The blended FCC feeds werecracked over an FCC catalyst,which was deactivated using aCPS-3 type protocol3,4 to 1000 ppmnickel and 2000 ppm vanadium.The catalyst was formulated to pro-vide maximum bottoms upgrading.Properties of the deactivated cata-lyst are shown in Table V.

The pilot unit was run at a constantreactor temperature of 1000˚F, andthe deactivated catalyst was testedat three catalyst-to-oil ratios (4,6,8)for each of the blended feeds in thepilot unit.

Figure 12 shows that all three of theblended bio-feeds are easier tocrack than the base feed. The addi-tion of the bio component increasesthe cracking activity (lower cat to oil

delta

Conv

ersi

on,w

t.%

Cat to Oil

94.0

97.0

100.0

103.0

106.0

109.0

112.0

4 6 8

7.5% Bio

Base Feed

15% Bio

Figure 12Conversion vs. Cat to Oil

HydrotreatedVGO

APISulfur, wt.%Nitrogen Total/Basic, ppmConcarbon, wt.%

K-FactorAromatic Ring, wt.%Naphthenic Ring, wt.%Paraffinic Carbons, wt.%

Vanadium, ppmNickel, ppm

D6352 Distillation, ºFIBP103050709095

FBP

CompositeResid

27.30.20

800/2800.2

12.2318.911.569.5

1.80.7

460703786849923

103410881226

24.40.53

813/2871.12

11.9622.713.663.8

2.51.1

494689775834899

101811101279

Table IVFeedstock Analysis for FCC Study

for a given conversion), andincreasing the concentration of agiven bio component yields anincrease in conversion at constantcat to oil ratio.

Figure 13 shows that the bio-feedsproduce significantly less coke perunit conversion than the base feed.Increasing the concentration of anyof the three bio components furtherreduces the amount of coke pro-duced for a given conversion. Therapeseed oil is the easiest to crack,followed by the soybean oil with thepalm oil being the most difficult ofthe components to crack.

The trends observed in Figures 12and 13 can be explained by theindividual fatty acid compositionsfor each of the bio componentsreported previously in Table I. Thereaction pathways for triglycerideshave been studied extensively byDupain et. al.5, and under FCC con-ditions, were found to occur in twodistinct steps. First, thermal crack-ing of the triglyceride occurs,releasing the fatty acid chains fromthe glycerin backbone. Conversionof free fatty acids via thermal crack-ing is very low, requiring catalyticcracking in the subsequent step tobreak the high molecular weightfatty acid molecules into smaller,more valuable products. Free fattyacids with more saturation will bemore difficult to crack than thosethat contain greater amounts ofdouble bonds6. With the lowest aver-

age of 1.1 double bonds per fatty acidchain, this explains the performanceof the palm oil versus the soy or rape-seed oils.

The bio-feeds also have significantlyless contaminants (sulfur, nitrogen,and potentially metals) than the basefeed, but additional characterization ofthese materials is very difficult.Traditional analysis methods devel-oped for fossil based hydrocarbonswill not apply to the bio-based materi-als. Measurements such as concar-bon, n-d-M (which is an estimate ofthe chemical composition of the feed-stock using refractive index (n), thedensity (d) and the molecular weight(M) of the feed to calculate theamount of paraffinic (Cp), napthenic(Cn), and aromatic (Ca) carbonspecies in the feed) and even GCbased distillations can be flawed dueto the thermal cracking response ofthe bio-feeds. With virtually no sulfurin each of the bio-feeds, incorporatingthem into FCC feed would dilute keyFCC product stream sulfur levels,

potentially allowing for downstreamhydrotreating benefits (less gaso-line octane loss, extended catalystrun length) or for a lower cost baseFCC feed at constant product sulfur.

Since the increase in the concentra-tion of the biofeeds in the feeddirectionally increases the magni-tude of the response in the FCCyields, to simplify the trends theremaining results will be presentedfor only 15% blends. Additionally,while commercial FCC units operateto constant coke, if there is a signif-icant difference in coke yield whencomparing individual productyields, the constant coke compari-son will accentuate the differencesin yields. Comparing data on a con-stant conversion basis will smoothout the data and allow for a morerepresentative comparison. Figure14 shows the yields for the bio-feeds at 15% concentration for con-stant coke and conversion com-pared to the base feed.

Beginning with hydrogen, Figure 15confirms that the reduction in hydro-gen yield with the biofeeds is signif-icant. Hydrogen can come fromthree sources: a by-product ofdehydrogenation with metals, aproduct of thermal cracking, or aproduct of catalytic cracking.

One potential method to determinethe reason for the reduction inhydrogen with the bio-based materi-als involves a comparison of the C5olefins yields in gasoline for thefeeds tested. During dehydrogena-tion reactions, gasoline rangeolefins react with the metals on thecatalyst to produce hydrogen7. If

Coke, wt.%

Biofeed Pilot Plant Study

71.0

72.0

73.0

74.0

75.0

76.0

77.0

78.0

79.0

80.0

81.0

3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5

Con

vers

ion,

wt.%

7.5% Palm 15% Palm7.5% Soy 15% Soy7.5% Rapeseed 15% RapeseedBase Feed

Figure 13Conversion vs. Coke

MAT, wt.%1759976

Unit Cell Size, Å 24.29Rare Earth, wt.% 2.04Alumina, wt.% 50.23

Total Surface, m2/gZeolite Surface, m2/gMatrix Surface, m2/g

Table VDeactivated Catalyst Properties


the biofeed materials produce lessgasoline range olefins than the basefeed, this could be the cause for thereduction in hydrogen.

From our analysis, the addition ofbiofeeds to the base feed does notreduce the amount of gasolineolefins produced, and therefore thereduction in hydrogen observedwith the biofeeds is not likely due toreduced dehydrogenation reac-tions.

Alternatively, the potential for thebiofeeds to produce hydrogen dur-ing the catalytic cracking processcan be evaluated by comparing thepropane yield for the feeds tested.In cracking a biofeed, the fatty acidmolecules that were liberated by theinitial thermal cracking step will sub-sequently crack along the pathwaysdefined for either paraffin or olefinmolecules. If the fatty acid mixtureis more olefinic, it will be very reac-tive and will easily crack to producesmaller gasoline range olefins6. Amore paraffinic fatty acid mixturecan react along multiple potentialpathways to produce a variety ofproducts (Figure 16).

The initiation step occurs on eitherBronsted or Lewis acid sites on thecatalyst, and on the Bronsted sitethere are two additional potentialpathways for the paraffin to react toproduce a carbenium ion. Protolyticcracking can occur with the elimina-tion of the paraffin, or the carbeni-um ion can be formed with the elim-ination of a hydrogen molecule(Figure 17.)

If protolytic cracking is prevalent,the chain terminating product ispropane. Figure 18 shows that thepropane yield for all three of the bio-feeds is lower than the base feed.Thus there is an indication that thereduction in propane and hydrogenis potentially due to reduced pro-tolytic cracking of the fatty acidscompared to a typical FCC feed.This also indicates that the doublebonds on each of the free fatty acidmolecules, which are more reactive,are the initiation sites for the crack-ing reactions.

Constant Coke, 5%Constant Conversion, 76%

Hydrogen

TotalDryGas

Propylene

TotalC3's

TotalC4='s

TotalC4's

C5+Gasoline

RON

MON

Gaso.Isoparaffins

Gaso.Aromatics

Gaso.Napthenes

Gaso.Olefins

LCO

Bottoms

Coke

Conversion

Del

taC

hang

e,w

t.%ex

-oct

ane

Figure 14Comparison of 15% Blended Feeds


0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.55

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0

Conversion, wt.%

Hyd

roge

n,w

t.%

15% Palm 15% Soy 15% RapeseedBase Feed

Figure 15Hydrogen Yield vs. Conversion

PROTOLYTICCRACKING

ALKANE

+ H+STRONG

BRONSTED

STRONG

LEWIS

CARBONIUM ION

CARBENIUM ION

CARBENIUM ION

- H+

ßSCISSION

ALKANE (Product)

OLEFIN(Product)

ALKANE (Product)

ALKANE (Product)

OLEFIN(Product)

+ ALKANE

- H+

+ H+REFORMLEWIS

REFORM

BRONSTED

INITIALREACTIONS

CHAINREACTIONS

INITIATION

TERMINATION

PROPAGATION

Figure 16Reaction Pathway Network for Alkane Cracking


Catalytic Cracking Catalysts, Chemistry and Kinetics, ChemicalIndustries 25, reprinted by permission

The fatty acid molecules are crack-ing into gasoline and propylene, ascan be seen in Figures 19 and 20.

The gasoline trend indicates over-cracking for all three of the bio-feeds, and also for the base feed.Over cracking occurs when thegasoline molecules produced aresubsequently cracked into lightermolecules. Catalyst activity, cat tooil ratio, and temperature can alldrive a feed to over-cracking, andthe easier the feed is to crack, thehigher the tendency to over-crack.Table IV shows that the base feeditself is fairly paraffinic, with aWatson K-Factor that indicates it willproduce high conversions. Theincorporation of the highly paraf-finic fatty acids from the biofeedsmakes the blended feeds even eas-ier to crack. Catalyst activity will bedetermined based on specific unitconstraints and objectives whencommercial processing of theblended bio-containing feeds. Theovercracking trends observed in thepilot plant testing can be correctedand is not expected in commercialoperations.

The influence of the bio-basedmaterials on the gasoline propertiesis important, as the oxygen speciesin the triglycerides could breakdown to water, or they could poten-tially be converted into aldehydeand furan species which are unde-sirable from an environmental,health and safety perspective. GasChromatography-Atomic EmissionDetector (GC-AED) was performedon the liquid product, which wasrecovered after each test runthrough the pilot unit. The GC-AEDwas run in oxygen mode in order todetect oxygen species using thesame technique as in Figure 4 and5. The only peaks that showed upwere initial peaks between two andfour minutes from the syringe washsolvents. No discernible oxygenpeaks were present in the liquidproduct thus, no undesirable oxy-gen species were present.

R CH2 R’CH2 HZ R’R CH2 C

H H

H Z-

+

-RH

-H2

R’R CH2 C

H

Z -+

R CH2 C

H

H

Z -+

Figure 17Paraffin Reaction Pathway


0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0

Conversion, wt.%

Prop

ane,

wt.%


Figure 18Propane Yield vs. Conversion


48.0

48.5

49.0

49.5

50.0

50.5

51.0

51.5

52.0

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0

Conversion, wt.%

Gas

olin

e,w

t.%


Figure 19Gasoline Yield vs. Conversion


The (R+M)/2 octane of the gasolineproduced by the biofeeds is lowerthan the base feed, with contribu-tions from both RON and MON con-tributing to the trend. The magni-tude of the response in gasolineRON may depend on the source ofthe biofeed, but in general the lossof RON is fairly small. (Figure 21)MON is consistently lower with thebiofeeds, and this is due to thelower aromatics content of thegasoline produced with the bio-based materials in the feed.

The response in C4 yields can beexplained by again reviewing theindividual fatty acid compositionsfor the three biofeeds as shown ear-lier. Palm oil contains significant per-centages of both C16 (palmitic) andC18 (oleic) acids. The highest %fatty acid in soybean oil is C18(linoleic), and C20 and C22 (arachi-donic and erucic) acids are presentin the largest concentrations inrapeseed oil. Once these mole-cules are liberated from the glycerinbackbone via thermal cracking,they follow a typical FCC rule ofthumb, which is that the longer thechain (ie. the higher the carbonnumber in the molecule), the morebroad the distribution of productolefins that will result from catalyticcracking of the molecule. Thus, thepalm oil produces the largestamount of C4 olefins, as seen inFigure 22.

Influence of Base Properties

Ultimately the yield response for theaddition of a bio-feed material intoFCC feed will depend on the prop-erties of the base FCC feed.Dupain et al.5 observed that theaddition of rapeseed oil to ahydrowax feed would yield lessgasoline and C4 minus products,and higher amounts of LCO, slurryand coke. (Figure 23)

While the properties of the specifichydrowax feed are not known, ingeneral these feeds are highlyparaffinic and the addition of veg-etable oils to that type of feed wouldin fact degrade the overall proper-


4.0

4.5

5.0

5.5

6.0

6.5

7.0

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0

Conversion, wt.%Pr

opyl

ene,

wt.%


Figure 20Propylene Yield vs. Conversion

90

90.5

91

91.5

92

92.5

93

93.5

94

71 73 75 77 79 81

Gas

olin

eR

ON

79

80

81

82

83

84

85

Gas

olin

eM

ON

Base Feed Palm Soy Rapeseed

Figure 21Gasoline RON & MON for Bio-blends


6.8

7.0

7.2

7.4

7.6

7.8

8.0

8.2

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0

Conversion, wt.%

Tota

lC4=

's,w

t.%


Figure 22Butylene Yield vs. Conversion


ties of the feed. This is confirmedby comparing the response inyields for the addition of rapeseedoil to the two base FCC feeds fromTable IV. The general trends inyields for constant conversion areshown in Figure 24. The addition ofrapeseed oil to the hydrotreatedVGO feed produced more coke andhydrogen, at constant conversion,but at constant coke the conversionwas actually lower with the rape-seed oil in the feed. By comparison,the composite feed shows strongimprovements in yields with theaddition of rapeseed.

These trends suggest that if a baseFCC feed is paraffinic, yields may infact worsen with the addition of abio based material into the feed,whereas for VGO or resid types offeed, yields may instead improve.Future work will include examiningthe role of the base feed on yieldswith the incorporation of biofeedmaterials into the FCC feed.

Conclusion

Based on these results, the use ofART’s high activity hydroprocessingcatalysts or Grace’s high perform-ance FCC catalysts can enablerefiners to co-process renewableoils through conventional refiningequipment. Co-processing can beincorporated into a refiner’s operat-ing strategy with minimal detriment

to catalyst stability or yields, but theeffect on an individual operation willdepend on the base feed and condi-tions. Grace Davison’s BiofuelTechnologies Group utilizes resourcesacross Grace and its affiliates to evalu-ate options for refiners who wish toconsider incorporating co-processingbiofeeds into their operation, but wantto understand the optimum configura-tion to maximize their profitability.

The authors of this paper would like tothank the following people for theircontributions to this project:

Susan Ehrlich, Business DirectorBiofuel Technologies Group; RickWormsbecher, Research andDevelopment Fellow, RefiningTechnologies

References

1. Data derived from OrganicChemistry, W.W. Linstromberg, D.C. Healthand Co., Lexington, MA, 1970

2. Data derived from OrganicChemistry, Morrison and Boyd, 6th Edition,1992

3. Wallenstein, D., Roberie, T., andBruhin, T., Catalysis Today 127, 2007, pp. 54-69.

4. Wallenstein, D., Harding, R.H., Nee,J.R.D., and Boock, L.T., Applied Catalysis A:General 204, 2000, pp. 89-106.

5. Dupain, X., Costa, D.J., Schaverien,C.J., Makkee, M., Moulijn, J.A., AppliedCatalysis B: Environmental 72, 2007, pp. 44-61.

6. W.-C. Cheng , E.T. Habib, Jr., K.Rajagopalan, T.G. Roberie, R.F.Wormsbecher, M.S. Ziebarth, Handbook ofHeterogeneous Catalysis, 2nd edition, G.Ertl, H. Knoezinger, F. Schueth, J. Weitkamp(Editors), Wiley-VCH, Weinheim, 2008,Chapter 13.5.

7. Zhao, X., J. A. Rudesill, W-C. Cheng,Preprints of Symposia - American ChemicalSociety, Division of Fuel Chemistry, 46(1),2001, pp. 240-244.

8. B. W. Wojciechowski, and A. Corma,Cataytic Cracking. Catalysts, Chemistry,and Kinetics. Chemical Industries 25, 1986,New York: Marcel Kekker.

0

10

20

30

40

50

60

70

0.0 0.2 0.4 0.6 0.8 1.0

RSO fraction in feed blend (-)

Prod

uctf

ract

ion,

wt.%

Gasoline

LCO

HCO

Gas

Coke

525 °C/CTO 4/21.2m

(a) HWX/RSO product fractions

Figure 23Quantitative Yields for DifferentHydrowax/Rapeseed Oil Blends

Bottoms

Coke

LCO

Gasoline

Wet Gas

Hydrogen

Composite FeedHydrotreated VGO

Bottoms

Coke

LCO

Gasoline

Wet Gas

Hydrogen

Composite FeedHydrotreated VGO

Red: Undesirable Green: Undesirable

Grey: Depends on the refiners objectives and constraints

Figure 24Yield Trend Comparison for Feeds at Constant Conversion


Applied Catalysis B, Environmental 72, reprinted by permission

New Opportunities for Co-Processing Renewable Feeds in ... · PDF fileCo-Processing Renewable...

Documents

Transcript of New Opportunities for Co-Processing Renewable Feeds in ... · PDF fileCo-Processing Renewable...