Upgrading of flash pyrolysis oil and utilization in refineries

Upgrading of flash pyrolysis oil and utilization in refineries

&maw and Biocnerg~ Vol. 7. Nos. l-6, pp. 237-244, 1994 Copyright c 1995 Elsevier Science Lid Printed in Great Britain. All rights reserved 0961-9534(9...

164KB Sizes 0 Downloads 9 Views

&maw and Biocnerg~ Vol. 7. Nos. l-6, pp. 237-244, 1994 Copyright c 1995 Elsevier Science Lid Printed in Great Britain. All rights reserved 0961-9534(94)000654 0961-9534194 $7.00 + 0.00


UPGRADING OF FLASH PYROLYSIS OIL AND UTILIZATION IN REFINERIES W. BALDAUF , U. BALFANZ and M. RUPP VEBA OEL AG, Pawiker Str. 30, D-45876 Gelsenkirchen, Germany Abstract-Flash pyrolysis oil from an ENSYN RTP pilot plant was upgraded in a continuous bench scale unit with commercial CoMo and NiMo catalysts in anticipation of scaling up the process. Large amounts of product were produced in a pilot plant for use in an extended analytical characterisation programme. In bench-scale experiments, high deoxygenation rates of 88-99.9% were achieved. Low liquid and high water yields were obtained. The fractionated products of the production run did not fulfil the requirements for direct use as gasoline and diesel. The process is restricted by several operational problems such as rapid catalyst deactivation, coking and plugging. Due to high feedstock and hydrogen addition costs, pyrolysis upgraded oil by the process tested is significantly more expensive than petroleum-derived oil at present oil prices.

from the hydrotreating of conventional coker distillates are the background to this study.


Pyrolysis oils from biomass are characterized by their high oxygen content. They are also very unstable, and properties like viscosity and water content change during storage and heating. The evidence for this is water formation and polymerisation initiated by side reactions.’ The properties and composition of pyrolysis oils mainly depend on the pyrolysis conditions. Rapid or fast pyrolysis processe? are characterised by a temperature range of 450”-900°C and short residence times (< 1 s). Under these conditions very low amounts of char are produced. In the region of 450”-6Oo”C, most of the products are liquids. These liquids contain high amounts of water (approx. 30%) and acids and have low viscosities, especially when compared with slow pyrolysis oils which have high viscosities. Because of their poor quality, the utilisation of raw pyrolysis oils is difficult. Possible applications are their use as fuel oil for electricity production in boilers and gas turbines5 or as feedstock in petroleum refineries. For both routes it is necessary to upgrade the oil, e.g. by hydrotreatment. Recent investigations cover laboratory experiments on a small scale>’while the target of the reported project is to investigate possible routes for pyrolysis oils from biomass to be processed in standard petroleum refineries. Applied technology and experience


A petroleum refinery is a very complex sequence of processes, which convert petroleum crude oils into lighter marketable fuels and chemical feedstocks. A simplified scheme is shown in Fig. 1, which shows only the major processes and major product streams. The crude oil enters the refinery via the desalting unit. After desalting the crude oil is atmospherically distilled and fractionated into naphtha and middle distillate. The naphtha cut is either routed to the naphtha pool, reformed to produce high-octane naphtha components, or treated in olefin plants to produce ethene and propene. The middle distillate can be directly routed to the middle distillate pool, or disulphurised prior to this. These pools supply gasoline, jet fuel, diesel and fuel oil to the market. The high boiling residue from the atmospheric distillation tower is further distilled in a vacuum tower. The resulting distillates are catalytically cracked in hydrocrackers or fluid cat crackers, respectively. The residues are thermally cracked in visbreaking or coking units, or in thermal hydrocracking processes. The lighter products are treated in the same way as the straight-run distillates. 237




et al.

Atm. Disttllatton

v&z. Diiillatton

Fig. 1. Refinery schematic.

Raw pyrolysis oils are not miscible with conventional petroleum crudes or any fractions and this leads to the need for a pretreatment step prior to processing pyrolysis oil in the refinery. Depending on the quality of the pyrolysis oil after pretreatment, it can be routed to either atmospheric or vacuum distillation, from where it ends up very diluted in the three distillate streams. Another possibility is to feed it directly into the conversion units. If pretreating product qualities allow, it could also be possible to distil the hydrotreated oil separately and to feed the distillates to the corresponding conversion units or to the corresponding product pools. The possible feed points for pyrolysis oils are indicated in Fig. 1.

Table 1. Characterisation of raw pyrolysis oil Water content % 29.85 Density (30°C) g cm-” 1.192 HHV MJ kg-’ 16.23 LHV MJ kg-’ 14.47 -30 “C Pourpoint 51 “C Flashpoint Elemental analysis (solids free) 39.17 wt% C 8.04 wt% H 5 S ppm wt% 0.05 N 52.74 wt% 0 Solids Acetone insolubles wt% 2.2 Pyridine insolubles wt% 2.1 Ash content wt% 0.15


Flash pyrolysis oil was provided by ENSYN (Canada). The oil was produced from mixed hardwood sawdust (maple/oak). The pyrolysis plant operated at a temperature of 525°C and a residence time of 0.35 s. The ENSYN oil was analysed with standard analytical methods. Results are listed in Table 1. Because the crude pyrolysis oil contained 2% of char the oil had to be filtered prior to hydrotreating. EXPERIMENTAL

The upgrading test runs were performed in a continuously operating bench-scale catalyst testing unit. The unit is designed for pressures of up to 30 mPa, temperatures of up to 500°C and flow rates up to 300 g h-‘. Especially for processing pyrolysis oil the unit was modified for an optional down- or upflow mode operation. A simplified flow scheme of this continuous-flow fixed-bed unit is shown in Fig. 2. The make-up hydrogen was supplied from a high-pressure hydrogen supply and the liquid feedstock was fed by means of a high-pressure pump. Both feed rates were continuously measured by flowmeters. After contacting gas and liquid the reactor was charged in the selected mode. The reactor (30mm internal diameter and 1130 mm long) had a volume of 7 16 cm3, minus the volume of a central thermowell which con-


Upgrading of flash pyrolysis oil


Stab. Gas

Fig. 2. Bench-scale unit for catalyst testing.

tained thermocouples for measuring the temperatures along the reaction section. The reactor was externally heated by five heating zones which were independently controlled and monitored by means of the thermocouples. Preheating was performed in the first part of the reactor, which was filled with inert SIC. The

centre part of the reactor was filled with catalyst. The feedstock was further heated up to the reaction temperature in the first third of the reactor which was isothermal in the remaining catalyst zone. The catalysts were standard refinery catalysts in their sulphided form. To prevent desulphiding and thus deactivation of the catalyst small amounts of DMDS (dimethyldisul-



B A L D A U F et a l .

WHSV:0.40 *








Temperature I [“Cl

Fig. 3. Influence of temperature on deoxygenation.

RESULTS phide) were added to the feed to ensure a minimum H, S partial pressure. During the experiments the effects of temThe product leaving the reactor was cooled down at pressure. The offgas was separated p e r a t u r e (350”-37O”C), s p a c e v e l o c i t y from the liquid phase in the high-pressure sepa- (0.154.80 kg (kg h))‘), up- and downflow rator and depressurised by a control valve which mode, and different catalysts (CoMo, NiMo) established the operating pressure of the unit. were investigated. In order to reduce the number of independent The liquid phase was depressurised and separated into a liquid product and stabilisation gas variables (temperature, space velocity) the severity parameter “deoxygenation rate” was in the stabiliser. The production run was performed in a hy- introduced for the following interpretation of drotreating pilot plant which was designed for the test results. Deoxygenation rate represents oil flow rates of up to 10 kg h-‘. The configur- the reduction of concentration of organic oxyation of the plant is similar to that of the gen in product oil related to organic oxygen bench-scale unit described above. A detailed concentration in the feed. Depending on the operating conditions, deoxygenation rate indescription of the plant is given in ref. [8]. Material balances were performed by creases with increasing severity which is illussampling the liquid and gaseous products. The trated in Fig. 3. The results of analytical inspection of product observation of mass balance was 48 h after establishment of stable conditions. oils are summarised in Table 2. The correspondRaw mass balances were corrected by a calcu- ing balances and product distributions are listed lation procedure which closed elemental bal- in Table 3. ances of carbon, hydrogen, oxygen, sulphur and Oil yields decrease from 36 to 32% with nitrogen. Hydrogen consumption was calcu- increasing deoxygenation. Water shows the lated as the difference in hydrogen content of highest yields. In addition to the water in the the product streams minus hydrogen content in feed, 21-25.5 wt% of water was produced. Gas the liquid feed. H,S, NH3 and additional H,O and water yields increase at higher severities. In were calculated by balancing S, N and 0 in addition to water formation, significant liquid product and feed (water in feed oil and amounts of oxygen were removed as CO and DMDS were taken into account). Finally c o , . With increasing deoxygenation the flow rates were corrected by closing the carbon amounts of CO and CO* decrease. This means balance. C,-C,, CO, CO*, and C,, were calcu- that, at higher severity, oxygen removal shifts lated from corrected flow rates and the compo- more and more to water production instead of sitions of liquid products, off gases and stripper CO, production. Oxygen rejection via water gases. increases hydrogen consumption. As deoxy-

Upgrading of flash pyrolysis oil


Table 2. Characterisation of product oils 540 Run no. downflow Flow direction CoMo Catalyst type 370 Temperature (“C) 0.25 Space velocity (kg(kg h))‘) 99.92 Deoxygenation rate (X) Elemenial analysis (water .free) 86.79 c (wt%)

H (wt%) s (wt%) N (wt%) 0 (wt%) Density (g cn-‘) H,O content (ppm) CCR (wt%) arom. C (wt%) HHV (MJ kg-‘) LHV (MJ kg-‘) Boiling range 0-200”c (wt%) 200-350°C (wt%)

350-500°C (wt%) > 5OOC (wt%)

13.17 0.070 0.003 0.021 0.83 50 0 10.0 45.96 43.09 40.8

38.5 20.7

541 downflow CoMo 370 0.40 98.53

542 downflow CoMo 350 0.25 95.17

543 downflow CoMo 350 0.40 91.78

544 downflow CoMo 370 0.80 88.32

545 downflow CoMo 370 0.40 95.88

87.22 12.35 0.004 0.033 0.39 0.86 210 0.09 17.0 45.26 42.60

86.84 11.82 0.004 0.082 1.25 0.896 510 1.40 18.8 44.46 41.87

86.50 11.23 0.008 0.021 2.14 0.928 1900 3.39 25.9 43.43 40.99

85.94 10.79 0.055 0.152 3.06 0.94 4200 5.52 27.0 41.20 38.87

86.93 11.87 0.055 0.129

40.0 36.2 21.2 2.6

30.6 31.4 23.1 14.9

27.6 29.6 23.0 19.8

25.3 29.2 21.6 23.9

genation increases from 88 to lOO%, hydrogen consumption increases by about 50%. Product properties show the following dependencies with decreasing severity: specific gravity increases from 0.83 to 0.94 g ml-‘; ?? water content in product oil increases from 50 to 4200 ppm; ?? higher and lower heating value decrease from 46.0 (43.1) to 41.2 (38.9) MJ kg-‘; ?? Conradson Carbon Residue (CCR) content increases from 0 to 5.5%; ??

CoMo 370 0.40 95.25

557 downflow NiMo 370 0.15 97.20

0.90 350 1.79 18.8 44.38 41.68

86.82 11.52 0.007 0.23 1.20 0.93 I 351 1.60 22.7 43.44 40.91

86.47 12.90 0.021 0.009 0.60 0.866 102 0.28 9.8 44.83 42.00

29.8 30.2 23.8 16.2

21.4 32.1 27.6 18.9

38.5 36.3 22.3 3.0




?? aromatic

carbon content increases from approx. 10 to 27%; ?? naphtha and middle distillate concentrations in the liquid product decrease from 41 and 39% to 21 and 29%, while vacuum gas oil remains relatively constant at approx. 20-22% and residue > 500°C increases from 0 to 24%. In order to get a measure of catalyst stability, run 541 was repeated at similar operating conditions after approx. 150 h (run 545). A significant drop of deoxygenation rate from 98.5 to 95.9% indicates the rapid catalyst deactivation

Table 3. Mass balances and yields Balance no. Flow direction Catalyst type Temperature (“C) Space velocity (kg(kg h))‘) Deoxygenation rate (X) Gases (total) (wt%) c, (wt%) c, (wt%) c, (wt%) c.$ (wt%) co (wt%) co, (wt%) Water (wt%) Oil (wt%) Total products (wt%) H, consumption (wt%)

540 downflow CoMo 370 0.25 99.92

541 downflow CoMo 370 0.40 98.53

542 downflow CoMo 350 0.25 95.17

543 downflow CoMo 350 0.40 91.78

544 downflow CoMo 370 0.80 88.32

545 downflow CoMo 370 0.40 95.88

551 upflow CoMo 370 0.40 95.25

557 downflow NiMo 370 0.15 97.20

16.69 3.22 4.06 2.70 2.04 0.13 4.46 55.51 32.81 105.01

18.47 3.32 4.04 2.49 1.88 0.16 6.53 53.61 32.28 104.42

18.53 3.12 4.27 2.20 1.66 0.11 7.11 52.89 32.68 104.10

16.72 2.67 3.69 1.76 1.29 0.15 7.15 52.45 34.51 103.68

15.85 2.57 2.62 1.25 0.90 0.32 8.20 50.97 36.34 103.16

16.76 3.35 3.81 1.81 1.31 0.15 6.33 53.58 33.78 104.12

17.12 3.53 3.61 1.85 1.34 0.15 6.51 53.14 33.75 104.01

16.90 3.68 3.63 2.32 0.57 0.55 4.70 54.53 33.58 105.01











et al.

Table 4. Characterisation of product oil from the oilot plant production run Total product

Naphtha >soo”c

Diesel <18o”C

86.45 13.22 851 83 0.2

85.45 14.72 130 2 0.05

87.08 12.88 109 51 0.07

Product Elemental analysis c (wt %) H (wt%) S (ppm) N (ppm) 0 (wt%) Ca @pm) Ni (ppm) V (ppm) Mo (ppm) Fe @pm) Na (ppm) Density (g cm-‘) n-Paraffins (wt%) iso-Paraffins (wt%) Naphthenes (wt%) Polynaphthenes (wt%) Aromatics (wt%) Mono aromatics (wt%) Diaromatics (wt%) Polycyclic aromatics (wt%) MOZ/ROZ (-) Cetane number (-) Cetane index (-) Cloudpoint (“C) CCR (wt%) Stability test (ASTM D 2274)

VGO Residue I go”-350°C 350”-500°C 87.54 12.26 426 162 0.4

0.7517 20.56 12.52 61.41 I .47 4.03




29.3 0.6 0.7 62162 45 39.7 -34 0.02 neg.

which is not acceptable for technical fixed-bed reactor systems. The result of rapid deactivation was proved by the fact that the test run was terminated due to pressure build-up in the reactor caused by plugging. While the test runs 540 to 545 were performed with downflow mode and with a CoMo catalyst, in additional test runs the influences of a different catalyst (NiMo, run 557) and of upflow mode (run 551) were tested. The influences on liquid product qualities can be interpreted as follows:

?? the

?? at

Despite operational problems during benchscale test runs, the potential of upgraded pyrolysis oil as feedstock for petroleum refineries should be demonstrated. To provide large amounts of upgraded oil, a production run was performed in a pilot plant. The bench-scale experience led to operation in downflow mode with NiMo catalyst. Operational parameters were adjusted for a medium deoxygenation rate of 97.3%. The product from the pilot plant test run was used for characterisation of total crude and its related fractions with respect to refinery requirements. Results are shown in Table 4. Total liquid product and fractions can be defined as follows.

similar operating conditions, deoxygenation rate in the upflow mode is less than in the downflow mode; odeoxygenation with the NiMo catalyst is less than with CoMo catalyst; @in upflow mode at the same deoxygenation rate, lower amounts of naphtha and higher amounts of vacuum gas oil are formed; consequently the liquid product density in the upflow mode is higher; ?? nitrogen removal in the upflow mode is significantly less than in the downflow mode; sulphur content in the liquid product is influenced by sulphur injection to the feed and does not represent the behaviour of the hydrotreating process;

content of aromatic carbon is lower with NiMo catalyst; ?? test runs with CoMo catalyst were terminated by coking and plugging in the low-temperature catalyst zone, whereas the test run with NiMo catalyst ended due to plugging of product lines with a gum-like substance; ?? all other properties are only determined by deoxygenation rate or severity, but are independent of flow mode and catalyst.

Upgrading of flash pyrolysis oil



01 0

I 50



Pyrolysis Oil I ECU/t

Fig. 4. Economics of upgrading.

Total liquid product

The product oil is a light brown low viscosity liquid with a boiling range from 35”-535°C. Naphtha (boiling range < 180°C)

High naphthene content and low octane number reduce its suitability as a gasoline pool component. Only small amounts can be blended as gasoline. Further upgrading in the reformer plant and conversion to olefins in steam crackers are more promising routes. Diesel (boiling range 180 ‘-350°C)

The diesel fraction is characterised by high density, low cetane number and high aromatics content. This fraction does not meet the specifications of a marketable diesel product. The oxygen stability test was negative. The need for further upgrading of this fraction for stabilisation and aromatics reduction prior to use as a diesel pool component is evident. Vacuum gasoil{ VGO (boiling range 350 ‘-500°C)

CCR and metals content of VGO are at the limits recommended for feedstocks for hydrocracking (HC) and fluid cat cracking (FCC) plants. Alternatively, the fraction might be used without further upgrading as heavy fuel oil. Residue ( > 500” C)

The residue fraction represents only a very small amount with a final boiling point of 535°C. Commercial vacuum distillation units will not separate this fraction because VGO cut points usually are > 535°C. That means that the residue will be part of the VGO fraction and is treated as described above. Restrictions might occur if CCR and metal contents of vacuum gas oil are significantly changed by the shift of the cut point.

Results of yield structure and fraction qualities indicate that the total liquid product cannot be fed to a single further upgrading process. Separate fractionation or feeding to the petroleum crude oil distillation unit (which is more economic in an existing refinery) is necessary to split the oil into the main fractions. The reported possible suitability routes are to be evaluated on the basis of the major analytical properties of the fractions. The final decision for suitability has to be confirmed by experimental test programmes. ECONOMICS OF UPGRADING

Low yields and high hydrogen costs are responsible for the high cost of upgraded pyrolysis oil. In Fig. 4 the costs for highly upgraded oil are shown as a function of raw pyrolysis oil price. Product costs are calculated as the sum of the feedstock cost, upgrading cost and hydrogen cost, minus a credit for the hydrocarbon gases on the following basis: ?? raw

pyrolysis price is varied between 50 and 15OECU t-‘; ?? upgrading costs are 15 ECU t- ’ feedstock at a hydrotreating plant capacity of 0.8 x lo6 t year-‘; ?? hydrogen costs are 0.09 ECU Nmd3 for steam reforming of natural gas; ?? hydrocarbon product gases are credited at a price proportional to their heating value which leads to approx. 30-50 ECU t-’ of gas. The result is that upgraded pyrolysis oil is more expensive by a factor of 1.5-3 than equivalent petroleum oil products. At typical feedstock costs of 100 ECU t-’for raw pyrolysis oil, 1 t of upgraded oil costs about 450 ECU t-l, compared with 18&220 ECU t- ’ for petroleum-



B ALDAUF et al.

derived naphtha and middle distillate and SO-100 ECU t-’ for fuel oil. The major contribution to this cost is the feedstock price (3 t of feedstock are necessary to produce 1 t of upgraded pyrolysis oil); about 150 ECU t-’ of product has to be expended for hydrogen. Credits for the hydrocarbon product gases and the upgrading costs offset each other. The situation is similar for partially upgraded oil, compared to prices of conventional fuel oils. The value of extenuities such as CO, tax or a credit for zero sulphur content can improve the price relations for biomass resources. These benefits are reduced if the hydrogen and process energy production by fossil resources are taken into account.

Low cetane number (45) high density (0.873 g ml-‘), and oxygen instability illustrate the poor potential of the diesel fraction for direct use as a pool component. Further upgrading is necessary. Total liquid product from the pretreatment step is recommended to be routed to the crude distillation tower, where the fractions end up diluted in the naphtha, diesel and VGO cut and are further upgraded in subsequent conversion units together with petroleum-derived products. A fixed-bed reactor system is not recommended, since the rapid catalyst deactivation must be compensated by continuous regeneration. An ebullated bed or liquid-phase reactor system with homogenous or without catalysts might be more promising.


Acknowledgements-The work was performed with the financial support of the European Union, DG XII in the framework of the JOULE programme. We thank ENSYN, Canada and Energy and Mines Resources, Canada for providing the pyrolysis oil.

Catalytic upgrading of flash pyrolysis oil in a fixed-bed reactor system was realised for short run times. High deoxygenation rates were obtained. Water is the major product and the oil yield is only 30-35%. The products are light liquids with low oxygen and water content and high heating values. At lower severity of lower deoxygenation rate the solubility of product oil in water is increased and the separation of aqueous and organic phases is increasingly difficult. The catalyst deactivated very rapidly and steady-state conditions could only be obtained for a short period. Several operational problems caused by the instability and coking potential of the pyrolysis oil made processing difficult. Tubes and valves were plugged or covered with gum-like deposits. Operating and control systems such as flow meters and level controllers were disturbed by adhesive deposits. With CoMo catalysts the catalyst bed plugged and coked in the preheater, whereas with a NiMo catalyst the reactor outlet was plugged by gumlike substances. Although the products are already hydrogenated, some fractions did not fulfil the required product specifications. The naphtha fraction is unsuitable for direct use as gasoline. Only very small amounts could be routed to the gasoline pool. An additional upgrading (e.g. in a reformer) or the use of the naphtha fraction as a feedstock for olefin production should be more promising.


1. M. Polk and M. Phingbodhipakkiya, U.S. Environmental Protection Agency, Off. Res. Dev. [Rep.] EPA, EPA-600/9-79-023b, Munic, Solid Waste: Resour. Recovery, pp. 118-125 (1979). 2. R. G. Graham, B. A. Free1 and M. A. Bergougnou, The production of pyrolytic liquids, gas and char from wood and cellulose by fast pyrolysis, In Research in Thermochemical Biomass Conversion (eds A. Bridgwater and J. Kiister), p. 629. Elsevier Applied Science, London and New York (1988). 3. R. G. Graham, B. A. Freel, D. R. Huffman and M. A. Bergougnou, The production of liquid fuels and chemicals from biomass by rapid thermal processing (RTP), 1st European Forum on Electricity Production from Biomass and Solid Wasres by Advanced Technologies,

Florence, Italy (27-29 November 1991). 4. D. Scott and J. Piskorz, The flash pyrolysis of aspen-poplar wood. Can. J. Chem. Engng 60,666 (1982). 5. R. G. Graham, 8. A. Freel, D. R. Huffman and A. J. Vogiatzis, The characterization and combustion of fast pyrolysis bio-oils, 1st European Forum on Electricity Production from Biomass and Solid Wastes by Advanced Technologies, Florence, Italy (27-29 November 1991).

6. E. G. Baker and D. L. Elliott, Catalytic upgrading of biomass pyrolysis oils, In Research in Thermochemical Biomass Conversion (eds A. Bridgwater and J. Kiister), p. 883. Elsevier Applied Science, London and New York (1988). __ 7. E. Churin, R. Maggi, P. Grange and B. Delmon, Characterization and upgrading of a bio-oil produced by pyrolysis of biomass, In Research in Thermochemical Biomass Conversion (eds A. Bridawater and J. Kiister). p. 896. Elsevier Applied Science, London and New York (1988). 8. W. Baldauf and U. Balfanz, Upgrading of pyrolysis oils in existing refinery structures, Proc. Energy from Biomass Conrractors Meeting (eds G. Grassi and A. Bridgwater), Gent, Belgium (1991).