Upgrading of flash pyrolysis tars to synthetic crude oil

Upgrading of flash pyrolysis tars to synthetic crude oil

Upgrading crude oil of flash pyrolysis tars to synthetic 4. Hydrotreatment reactor James H. Edwards, with iron catalyst Kym Schluter in a slurr...

639KB Sizes 5 Downloads 101 Views

Upgrading crude oil

of flash pyrolysis tars to synthetic

4. Hydrotreatment reactor James

H. Edwards,


iron catalyst

Kym Schluter

in a slurry-phase

and Ralph J. Tyler

CSIRO Division of Fossil Fuels, PO Box 136, North Ryde, NS W 2 113, Australia (Received 19 March 7986; revised 22 August 1986)

Flash pyrolysis tars, produced from Millmerran (Queensland subbituminous) and Piercetield (New South Wales bituminous) coals, have been hydrogenated in a slurry-phase bubblecolumn reactor using red mud and sulphur as catalyst. Oils of low coking propensity and high volatility were produced from both coal tars. The slurry-phase reactor successfully overcame the severe coking problems previously encountered when hydrogenating flash pyrolysis tars in fixed-bed reactors, completely eliminating coke formation. For Millmerran coal tar the effect of reaction temperature on reactor performance was investigated in the range 42&46o”C. The major effects of increasing temperature were to reduce the coking propensity of the product oil and to increase its volatility. Product oils from both tars were high in aromatics and heteroatoms, and the distillate fractions derived from these oils would require further refining to produce finished products (i.e., gasoline and diesel). It should be possible to achieve this using conventional relining technology. (Keywords:

flash pyrolysis tar; hydrotreatment;

iron catalyst; slurry-phase reactor)

The upgrading of tar to synthetic crude oil by catalytic hydrotreatment is an important part of the CSIRO flash pyrolysis process for producing liquid fuels from coal’-3. The process involves the rapid pyrolysis of pulverized coal in a fluidized bed to produce tar, which is then converted to oil by hydrogenation. Development of a suitable hydrogenation reactor has posed problems because the flash pyrolysis tars have a high propensity to deposit coke on hydrogenation catalysts and contain small but significant quantities of fine solids carried over from the pyrolyser. One system investigated for this purpose involved treating the tar with hydrogen over a packed bed of sulphided steelwool to reduce its coking propensity by mild hydrotreatment, followed by further upgrading in a fixed bed of conventional Ni/Mo hydrotreating catalyst. The performance of this two-stage reactor has been reported previously4-6. Results showed that the steelwool reactor was a useful laboratoryand PDU-scale (PDU = process development unit) method for treating whole tar. However, it was not a practical system for fullscale application since its operating life was limited by the rapid build-up of carbonaceous deposits (char and chemically formed coke). In addition, coking problems were experienced under certain circumstances in the second-stage (Ni/Mo) reactor, indicating that more effective primary upgrading of the tar was required. Subsequent efforts concentrated on finding an alternative to the steelwool reactor. The principal objective was to produce an oil with a high volatility and low coking propensity, without coke formation, in a reaction system that could be scaled up for practical use. A slurry-phase reactor using red mud (ironrich byproduct of bauxite refining) and sulphur as catalyst has been successfully developed to achieve this objective. 0016-2361/87/050637-06.$3.00 0 1987 Butterworth & Co. (Publishers) Ltd.

Results from the upgrading of tars from two Australian coals in the slurry-phase reactor are presented in this paper. EXPERIMENTAL Reaction system and operating conditions The reactor consisted of an electrically-heated type 316 stainless steel tube (13.8 mm i.d. x 1320 mm long), which was mounted vertically and operated as a slurry-phase bubblecolumn reactor. The liquid and hydrogen feed, product letdown and gas analysis systems were the same as those described previously for the two-stage reactor experiments4+. All runs were conducted at 13.8 MPa system pressure, tar space velocity of 0.45 h- 1 (vol basis) and hydrogen/tar ratio of 3.2m3 (NTP) kg-‘. Red mud and elemental sulphur were added directly to the tar in powdered form at constant levels of 5 and 2 wt ‘A tar, respectively. Runs lasted for 13 to 22 h and were terminated voluntarily. During each run product oil was recovered from the high pressure separator at regular intervals (usually every 34 h). Tar feedstocks (Table I) were prepared by the flash pyrolysis of Millmerran (Queensland subbituminous) and Piercefield (New South Wales bituminous) coals at 6OO“C in a 20 kg h-i PDU-scale pyrolyser. Tars were recovered from the pyrolyser exit gas stream in an electrostatic precipitator and were used in subsequent experiments without dilution with solvents. Four runs were carried out with Millmerran tar and one with Piercefield tar. Run 1 (419°C) was carried out under conditions similar to those used for the steelwool reactor to enable the performances of the two reactors to be compared. The effect on reactor performance and on the

FUEL, 1987, Vol 66, May


Hydrotreatment Table 1

of flash pyrolysis

tars in a slurry-phase

Properties of tar feedstocks Milhnerran”

Elemental analysis (wt % as fed) Carbon Hydrogen Nitrogen Sulphur Oxygen (by diff) Atomic H/C ratio Phenolic OH (wt %) Ash (wt %) Carbon aromaticity (“C n.m.r.) Quinohne-insolubles (wt%) Conradson carbon (wt % dcf) Specific gravity


Batch 1

Batch 2

80.6 1.7 1.4 0.6 9.7 1.15 n.d. 2.0

79.9 7.8 1.1 0.5 10.7 1.17 5.0 1.1

80.9 6.1 1.9 0.4 10.1 0.99 5.8 1.9




4.0 30.0 1.14

nd. 30.7 n.d.

4.5 38.6 n.d.

“Batch 1 used in Runs 1 and 2; batch 2 used in Runs 3 and 4; n.d. not determined

nature of the product oil of increasing reaction temperature from 443 (Phase 1) to 462°C (Phase 2) was investigated in Run 2. Runs 3 and 4 were carried out at ~4460°C to obtain complete product yields at this temperature. A single run (Run 5) was conducted with Piercelield tar using the same conditions as Runs 3 and 4. Catalyst activation

Red mud from an Australian bauxite refinery was treated according to the procedure recommended by Pratt and Christoverson7. Briefly this involved digestion of the red mud in hydrochloric acid followed by precipitation of the iron with ammonia at pH 8.0. The resultant slurry was filtered and the solids washed with distilled water, dried at 110°C and then calcined in air for 2 h at 500°C. Analyses of the raw and treated red mud are given in Table 2. Although pre-treatment increased the concentrations of iron, aluminium, silicon and titanium significantly, the largest change was in the BET surface area which increased from 27 to 97 m2 g-‘. Treated catalyst was used in all runs except Run 4, in which untreated red mud was used to determine whether activation of the catalyst produced any improvements in oil yield or quality. Characterization of products

All product oil samples from Runs 2 to 5 were treated in the following manner, not only to remove moisture and solids but also to determine their contents. The raw oil was distilled at 100°C and material (light oil and water) evaporated under these conditions was condensed and collected. Distillation was considered complete when no further water was recovered. The quantity of water was measured directly and the light oil separated and retained. The dry oil was then filtered and the filter cake extracted with tetrahydrofuran (THF) to measure the amount of THF-insoluble material in the raw oil. The THF was evaporated from the extract and the extract plus light oil recombined with the bulk oil to produce a dry, solids-free oil. Overall recovery of products (oil, water and THF-insoluble material) by this procedure amounted to more than 98 wt % of the raw oil.


FUEL, 1987, Vol 66, May

reactor: J. H. Edwards et al.

The product oils from Run 1 were too viscous to be processed by this procedure. Moisture contents were determined on sub-samples of this material by the Dean and Stark method. Further samples were dried by passing gaseous nitrogen through them at 60°C and these were then extracted with THF and filtered to remove and measure the solids content. THF was evaporated from the filtrate to produce dry, solids-free oil. Oil analyses reported here are based on dry, solids-free oils. Distillation of the raw tar was conducted according to the ASTM D-1160 method, while product oils were distilled in a spinning band microstill (100 theoretical plates, 50: 1 reflux ratio). The propensity of tar and oils to form coke on heating was measured by the Conradson Method (ASTM 189-76). Phenolic OH contents were determined by enthalpimetric titration and carbon aromaticity by 13C n.m.r. Coke formation was checked at the end of each run by comparing the amount of THF-insoluble material in the reactor contents and product oil with that measured in the feed tar. The forms of iron present in the fresh and used catalyst were determined by Mossbauer spectroscopy of selected samples. RESULTS


Reactor performance

Performance data for Runs 1, 3, 4 and 5 are given in was varied throughout the run, it was only possible to measure the C,C, hydrocarbon gas yields, these being 5.0 and 8.5 wt % of the dry, char-free (dcf) tar input at 443 and 462°C respectively. Total products accounted for ranged from 96.0 to 98.7 wt % dcf tar. Recovered oil yields from Millmerran tar declined with increasing temperature, being 84.6 wt % dcf tar at 419°C and 77.8 wt % at 463°C. The 463°C oil yield from Piercefield tar was slightly lower at 75.3 wt %. Tar yields from the pyrolysis of Millmerran and Piercefield coals were 35 and 25 wt % of the dry, ash-free (daf) coal and thus the net oil yields at 463°C were 27.2 and 18.9 wt %, respectively, of the daf coal fed to the pyrolyser. For Millmerran tar, the C,C, hydrocarbon yield increased substantially with temperature, rising from 3.2 wt % dcf tar at 419°C to 10.0 wt % at 463°C. The increased gas yield is largely responsible for the doubling of the hydrogen consumption over this temperature range. The gas yield from Piercelield tar was even higher, being 13.4wt % at 463°C. In a commercial pyrolysis Table 3. In Run 2, where the reactor temperature

Table 2

Analyses of raw and treated red mud

Component (wt % dry basis) Fe,& AI,& SiO, TiO, MnO MgQ Na,O R,Q P,Q, CaO BET surface area (m* g-l)



31.5 19.5 11.2 7.0 0.01 0.07 2.7 0.06 0.16 3.2 27

41.2 25.5 12.7 9.5 0.01 0.12 0.12 0.09 0.23 0.56 97


of flash pyrolysis

reactor: J. H. Edwards et al.

tars in a slurry-phase

Table 3 Slurry-phase reactor performance data Piercefield

Millmerran Run number Temperature (“C) Product yields (wt % dcf tar) Daf oil CIC, hydrocarbons Formed water Formed coke NH, H,S co CO, Total products Hydrogen consumption (wt % dcf tar) Carbon balance (%) Heteroatom removal efficiency (%) Oxygen Nitrogen Sulphur Overall oil yield

1 419

3 460

4 463

5 463

84.6 3.2 6.6 nil 0.3 0.3 0.5 0.5 96.0 1.4

18.5 9.7 8.2 nil 0.4 0.3 0.7 0.6 98.4 2.8

17.8 10.0 6.8 nil 0.3 0.4 0.8 1.3 91.4 2.1

75.3 13.4 7.9 nil 0.7 0.3 0.3 0.8 98.1 3.6





66 16 51 29.6

75 21 67 27.5

70 21 69 27.2

76 29 81 18.9

(wt % daf coal fed to pyrolyser)

process the gas could be effectively utilized by steam reforming to produce the hydrogen required for tar upgrading. However, it is considered that operating at temperatures much above 460°C would result in unacceptably high losses of product oil through the formation of hydrocarbon gases. Substantial quantities of oxygen and sulphur were removed from both tars but there was little reduction in nitrogen content. For Millmerran tar, the heteroatom removal efficiencies were only marginally improved by increasing the temperature from 419 to 463°C. A notable achievement of the slurry-phase reactor was that no coke formation was encountered in any of the runs with Millmerran and Piercelield tars. In all cases the amount of THF-insoluble material contained in the product oil and reactor contents at the end of each run was equal to or less than that present in the feed slurry (i.e. added catalyst and char carried over from the pyrolyser char recovery system). Microscopic examination of the THF-insoluble material in both oil and reactor contents showed no evidence of mesophase formation (M. Shibaoka, CSIRO Division of Fossil Fuels, personal communication). In addition reactor operation was not affected by the solids in the feed tar. These passed completely through the reactor and were continuously removed from the system with the product oil. In a largescale plant the solids could then be separated from the oil by either filtration or vacuum distillation. The ability of the slurry-phase reactor to process tar containing fine solids and to eliminate coke formation has solved the major operational problems encountered with the steelwool reactor. Therefore the slurry-phase reactor should be suitable, in these respects, for scaling up to a commercial size. Properties of product oils

Properties of the oils produced at similar operating conditions in the slurry-phase and stcelwool reactors are given in Table 4. Oil from the steelwool reactor is lower in nitrogen than the slurry-phase product but this reflects

Table 4

Comparison of oils from slurry-phase and steelwool reactors

Temperature (“C) Elemental analysis (wt % daf) Carbon Hydrogen Nitrogen Sulphur Oxygen (by diff.) Atomic H/C ratio Phenolic OH (wt % daf) Ash (wt %) Conradson carbon (wt % daf) Carbon aromaticity (‘% n.m.r.) Distillation (wt %) IBP-230°C gasoline/kerosine 230-350°C diesel > 350°C residual

Slurry-phase reactor 419

Steelwool reactor 419

85.4 9.1 1.4 0.3 3.8 1.28 2.1 0.07 11.0 0.54

85.5 8.7 0.8 0.2 4.8 1.22 3.3 0.08 16.9 0.58

6 30 64

5 28 61

the lower nitrogen content of the tar used in the steelwool reactor experiments rather than an improved efficiency for this heteroatom. In other respects the analyses of the tar feeds were very similar. Overall, the data in Table 4 show that the slurry-phase product is of slightly better quality than oil from the steelwool reactor, with respect to heteroatom content and coking propensity, but the difference is not significant. However, it will be remembered that the slurry-phase product was generated without coke formation, whereas in the steelwool reactor 4 wt % of the tar was converted to coke. Properties of oils produced in all the slurry-phase reactor runs are given in Table 5. The data are composite analyses for each run since in all cases the variation in oil properties during the run was small. Results for the Millmerran tar show that in some respects the red mud was a poor hydrogenation catalyst: the aromatics and heteroatom contents of the oil were high and there was little reduction in these components with increasing temperature. The hydrogen content increased only



Vol 66, May


Hydrotreatment Table 5

of flash pyrolysis

tars in a slurry-phase

reactor: J. H. Edwards et al.

Properties of product oils Millmerran

Run number


Temperature (“C)


Phase 1 443

Elemental analysis (wt % daf) Carbon Hydrogen Nitrogen Sulphur Oxygen (by diff) Atomic H/C ratio Phenolic OH (wt % daf) Ash (wt %) Conradson carbon (wt % daf) Carbon aromaticity (r3C n.m.r.) Specific gravity

85.4 9.1 1.4 0.3 3.8 1.28 2.7 0.07 11.0 0.54 n.d.

Distillation (wt %) IBP-230°C gasolinefierosine 23&35o”C diesel > 350°C residual

6 30 64







Phase 2 462




85.3 9.4 1.2 0.3 3.8 1.32 3.0 0.08 7.5 0.57 n.d.

85.8 9.8 I.0 0.3 3.1 1.37 2.7 0.04 2.9 n.d. 0.98

85.6 9.7 1.1 0.2 3.4 1.36 2.8 0.01 2.8 0.48 0.96

84.8 9.7 1.1 0.2 4.2 1.37 3.0 0.01 3.4 0.52 n.d.

86.3 8.6 1.8 0.1 3.2 3.0 0.03 8.6 0.63 n.d.

14 35 51

26 41 33

24 48 28

24 46 30

17 42 41



Figure 1 Distillation curves for feed and product oils, slurry-phase reactor, Millmerran coal tar. 0, Feed tar; l ,419”C product; 0,443”C product ; A, 463°C product

marginally from 9.1 wt % at 419°C to 9.7 wt % at 463°C. However, the catalyst was very effective in improving two important properties, namely volatility and coking propensity. The dramatic increase in volatility is shown by the distillation curves for the feed and product oils in Figure 1. The effect of reactor temperature on these properties is shown in Figure 2. The yield of diesel and lighter transport fuel fractions increased from 36 wt % of the oil at 419°C to >70wt % at 463°C whilst the Conradson carbon level in the oil decreased from 11 to 3 wt %. It is notable that such large improvements in volatility and coking propensity were achieved over a relatively small temperature range and with only a modest increase in the atomic H/C ratio of the oil from 1.28 at 419°C to 1.37 at 463°C. The Piercefield tar was more difficult to upgrade than




Vol 66, May



430 440 450 460 Reaction temperature (“C 1


Figure 2 Effect of temperature on the properties of the product oil, slurry-phase reactor, Millmerran coal tar

the Millmerran tar since it was more aromatic and had higher Conradson carbon and nitrogen levels. Consequently the Piercefield oil is of poorer quality than the Millmerran 463°C product. Nonetheless, the slurryphase reactor successfully processed the bituminous coal tar to give an oil that contained almost 60 wt % material in the diesel and lighter boiling range.


of flash pyrolysis

Properties of distillate j-actions

The high volatility of the oil produced at 463°C means that it can be distilled directly into gasoline, kerosine and diesel fractions without further treatment. This has the advantage that the subsequent refining of each fraction can then be optimized separately to meet the required specifications for the finished products. The properties of each distillate fraction from both tars are given in Table 6. All fractions have unacceptably high levels of heteroatoms and would require severe hydrotreating for their removal. Refining of the gasoline fractions would need to optimize hydrotreating conditions for heteroatom removal whilst, if possible, minimizing aromatics saturation. On the other hand both the kerosine and diesel components would require severe hydrocracking to reduce aromatics to the maximum l&20% levels tolerable in the relined products. The Millmerran and Piercefield distillate fractions have similar properties to the corresponding materials produced from Wandoan (Queensland, subbituminous) and Piercefield coals in the coal hydrogenation PDU at the Australian Coal Industry Research Laboratories (ACIRL)8. A study by Ampol Research and Development Laboratory has demonstrated that severe refining processes are capable of producing satisfactory transport fuels from the ACIRL coal-derived liquids’. On this basis it should be technically feasible to produce specificationgrade fuels from the flash pyrolysis distillate fractions. Role of the catalyst

The use of red mud plus sulphur as an effective catalyst for the liquefaction of coal and the upgrading of heavy oils has been extensively reported’*‘*. Results from the current study on tar upgrading have indicated that, although it is a poor hydrogenation catalyst, the red mud is very effective in suppressing coke formation. This suggests that the red mud efficiently catalyses the reactions between hydrogen and the lower molecular weight radicals, produced by cracking of the tar, to stabilize these fragments before they can polymerize to produce coke. The cracking reactions may have also been catalysed by components in the red mud (e.g. silica and alumina). However, the fact that the 419°C slurry-phase reactor product is not significantly more volatile than the oil from the steelwool reactor (see Table 4) suggests that the cracking is predominantly thermal rather than catalytic. The complete absence of coke formation suggests that the slurry-phase reactor promotes efficient contact of the three phases of tar, hydrogen and catalyst.

Table 6

tars in a slurry-phase

reactor: J. H. Edwards et al.

The Miissbauer spectra of fresh and used catalysts confirm the observations of other workers” that, in the presence of H,S under hydrogenation conditions, the iron is converted from haematite (Fe,O,) to the nonstoichiometric pyrrhotite form (Fe, _,S). Comparison of the results obtained for Runs 3 and 4 shows that activating the red mud has had no significant effect on either the product yields or oil properties. However, no attempt was made in this work to find the optimum catalyst concentration. In a commercial plant it would be important to minimize the amount of catalyst required to reduce the quantity of solids that must subsequently be separated from the oil. Under these circumstances the activation procedure may prove beneficial at catalyst levels lower than the 5 wt% concentration employed here.

CONCLUSIONS A slurry-phase reactor incorporating red mud and sulphur as disposable catalyst is well suited as the primary upgrading step for the conversion of flash pyrolysis tars to liquid fuels. In this system the major objective of producing a highly volatile oil with low coking propensity has been achieved and the coking problem has been overcome by completely eliminating coke formation. However, because of the poor hydrogenation activity of the red mud, the product oils were high in both aromatics and heteroatoms and the distillate fractions derived from these oils would require severe refining to reduce these components to the maximum levels tolerable in the finished products. It should be possible to achieve this using conventional relining technology.

ACKNOWLEDGEMENTS The authors wish to thank their CSIRO colleagues who have contributed to the information in this paper. Support was provided under the National Energy Research, Development and Demonstration Program administered by the Department of Resources and Energy.


Smith, I. W. ‘EPRI Conference on Coal Pyrolysis’, Palo Alto, Paper No. 9, 1981

Properties of 463°C product oil distillate fractions


Gasoline (IBP-200°C)

Kerosine (20&23O”C)

Diesel (23%350°C)








Elemental analysis (wt % daf) Carbon Hydrogen Nitrogen Sulphur Oxygen (by diff) Atomic H/C ratio Carbon aromaticity (‘% n.m.r.) Phenolic OH (wt % daf)

82.2 12.2 0.6 0.2 4.8 1.77 0.33 2.2

81.3 11.1 0.7 0.1 6.8 1.64 0.46 3.9

83.6 9.7 0.5 0.1 6.1 1.39 0.56 5.3

83.4 9.3 0.9 0.1 6.3 1.34 0.66 6.3

85.9 9.9 1.0 0.2 3.0 1.38 0.50 2.5

86.2 9.4 1.7 0.1 2.6 1.30 0.59 2.3

M = Millmerran; P = Piercefield


1987, Vol 66, May


Hydrotreatment 2


4 5 6 7


of flash pyrolysis

tars in a slurry-phase

Wailes, P. C. and Kershaw, J. R. ‘The Characterisation and Upgrading of Flash Pyrolysis Coal Tars’, Project Report No. 3, CSIRO Division of Applied Organic Chemistry, 1982 Edwards, J. H., Smith, I. W. and Tyler, R. J. ‘The CSIRO Flash Pyrolysis Project: Compendium of Data’, Investigation Report 140, CSIRO Institute of Energy and Earth Resources, 1983 Edwards, J. H., Schluter, K. and Tyler, R. J. Fuel 198564, 594 Edwards, J. H., Schluter, K. and Tyler, R. J. Fuel 198665, 202 Edwards, J. H., Schluter, K. and Tyler, R. J. Fuel 1986,65,208 Pratt, K. C. and Christoverson, V. Fuel 1982,61, 460

FUEL, 1987, Vol 66, May

reactor: J. H. Edwards et al. 8

9 10 11

Armstrong, L., Atkins, A. and Copson, B. ‘Relining of ACIRL Syncrudes to Transport Fuels’, NERDDP/EG/83/190, July 1983 Garg, D. and Givens, E. N. Ind. Eng. Chem. Process Des. Dev. 198524, 66 Strobel, B. 0. and Friedrich, F. ‘Proc. 1985 Int. Conf. Coal Sci.‘, Pergamon, Sydney, p. 7 Montano, P. A., Lee, Y. C., Yeye-Odu, A. and Chien, C. H. Am. Chem. Sot. Dio. Fuel. Chem., Prep.

1984, 29(4), 279