Coal feed flexibility in the Exxon Donor Solvent coal liquefaction process

The Exxon Donor Solvent (EDS) Process has been successfully employed to liquefy coals of varying rank. Bituminous, subbituminous and lignitic coals have been processed in the continuous, integrated 40 kg day^?1Recycle Coal Liquefaction Units and 1 t day^?1 Coal Liquefaction Pilot plant located at Baytown, Texas. Recent operations show that significant improvements in total liquid yield, as well as additional flexibility in product distribution, can be achieved with recycle of liquefaction bottoms. The impact of the type of coal and mode of operation on product yield and distribution as well as pilot unit operability are discussed. Specific changes in process configuration have been explored and are desirable for different coals and can be used to produce a variety of products. The implication of these recent results on defining the coal liquefaction reaction paths is discussed.     

Additive effect of cobalt-exchanged coal on the liquefaction of subbituminous coal

In comparison with the liquefaction (420 °C, with tetralin and elemental sulfur) of the proton-exchanged Adaro subbituminous coal with complex (metallocene or metal carbonyl) of cobalt, nickel or iron, the coal liquefaction was enhanced significantly on the liquefaction of cation (Co²⁺, Ni²⁺ or Fe²⁺) exchanged Adaro coal. However, for several kinds of subbituminous coal, metal cations hardly exchange due to the small content of carboxyl group in coal. Therefore, the simultaneous liquefaction of subbituminous coal and cobalt-exchanged coal were carried out in this study. Further, the simultaneous liquefaction of lower rank and subbituminous coals with Co₂(CO)₈ was carried out to compare the catalytic effect of cobalt derived from either the exchanged coal or Co₂(CO)₈. The catalytic effect of cobalt in the exchanged Morwell brown coal was appeared as the decrease of residue yield on the liquefaction with Yilan subbituminous coal. On the other hand, there was little enhancement on the simultaneous liquefaction of cobalt-exchanged Adaro and Yilan subbituminous coals. Accordingly, it is clarified that the hydroliquefaction of subbituminous coal, which was exchanged cation hardly, was enhanced by the addition of cobalt-exchanged brown coal.     

Differences in the behaviour of US bituminous and subbituminous coals during short contact time liquefaction

The short contact time (SCT) liquefaction of Belle Ayr subbituminous coal has been compared with that of Illinois No. 6 and Pittsburgh seam bituminous coals. Each bituminous coal was highly solubilized (90 wt%, daf coal) in 3–4 min at 450 °C and 13–16 MPa hydrogen pressure. More than 80 wt% of each coal was converted to solvent-refined coal (SRC, pyridine-soluble residuum), with only small quantities of distillate oil and C₁–C₄ gas being formed. A longer reaction (up to 30 min) gave only a small increase in total conversion, but gas and distillate yields increased significantly. Iron sulphides did not appear to catalyse coal solubilization. By contrast, only 65 wt% of the Belle Ayr coal dissolved rapidly in SCT liquefaction and pyrite addition catalysed the conversion of the remaining insoluble organic matter (IOM). With an equivalent amount of pyrite present the Belle Ayr coal also gave more C₁–C₄ gas and substantially more distillate in SCT liquefaction than the bituminous coals. These differences in product distributions obtained from bituminous and subbituminous coals in SCT liquefaction can be rationalized on the basis of differences in the structures of the starting coals. However, the origin of high IOM yields with the Belle Ayr coal remains unclear.     

Catalytic liquefaction of various coals using a mixture of carbon monoxide and water

In a batch-autoclave, twenty coals were liquefied using a cobalt-molybdenum oxide catalyst with a mixture of CO and H₂0 at 400 °C with or without vehicle oil. Furthermore, lignite and peat were liquefied on tungsten oxide catalyst at 300 °C in the absence of CO. The reactivity of coal in this liquefaction is found to depend strongly on its rank. The vehicle oil significantly influences the extent of the water-gas shift reaction, especially when bituminous coals are liquefied, by dissolving such coals. Liquefaction of coal by this process is considered to take place via three routes: hydrogenolysis by the nascent hydrogen produced from the shift reaction; dissolution of coal into the vehicle oil which is an initial stage of hydrogenolysis; and a solvolytic reaction with H₂O, such as hydrolysis.     

Coal liquefaction by in-situ hydrogen generation.: 1. Zinc-water-coal reaction

Liquefaction of coal was carried out in a zinc—water—solvent system to give a product with high concentration of pyridine and benzene solubles. In this system the metal reacts with water to produce the corresponding metal oxide and hydrogen. This hydrogen was used for in-situ hydrogenation of coal. The effects of reaction time, temperature, type of solvent, the quantity of metal used and the rank of coal were investigated. The solvent has a very marked effect on the conversion of coal to benzene-soluble materials, especially at short reaction times. A maximum benzene conversion of 96% for Taiheiyo coal was obtained when it was treated at 445 °C for 1 h using wash oil as solvent. With regard to the influence of coal rank it was found that low rank coals were more reactive than high rank coals. The amount of preasphaltene is only slightly influenced by coal rank but depends on the temperature and the type of solvent used.     

Catalytic liquefaction of coal with supercritical water/CO/solvent media

This paper describes studies of the catalytic activity of cobalt molybdenum sulphide, cobalt molybdenum oxide, iron sulphate, iron acetate dibasic and H₂S during the reaction between supercritical CO and water, and during liquefaction of coal using supercritical CO-water-solvent mixtures. The kinetics of the water gas shift reaction was studied first and was found to be first order in all the catalysts studied. The activity of the catalysts decreased in the following order: cobalt molybdenum sulphide > cobalt molybdenum oxide > iron salts. The presence of toluene, tetralin, and THQ decreased the CO conversion on the cobalt catalysts but increased CO conversion in the presence of iron salts catalysts. Moderate coal conversion of toluene soluble products (25%–35%) were obtained in the presence of supercritical water and water/CO mixtures. Addition of organic solvents to a supercritical water/CO medium increased conversion of toluene soluble products to 70–80% for THQ, to 50–60% for tetralin, and to 35–40% for toluene. Addition of H₂S to the solvent/water/CO medium increased conversion to toluene soluble products even further. In the presence of H₂S/solvent/water/CO, the presence of catalysts had only a minor effect on coal conversion and were not required to achieve high coal conversions. The optimum operating conditions for an Illinois No. 6 coal were obtained using a H₂S/THQ/CO/water medium at 3600 psi, and 400 °C. Higher conversions were attained with a subbituminous Wyodak coal. These studies clearly demonstrate that high conversions to soluble products can be attained using a supercritical water/Co/solvent medium.     

Effect of Wyodak coal properties on hydroliquefaction reactivity

The effect of Wyodak coal properties on liquefaction reactivity as measured by distillate yield and cyclohexane conversion has been investigated. Spot samples of four Wyodak subbituminous coals from the Anderson and Canyon coal seams in the Powder River Basin of NE Wyoming were liquefied in microautoclave and batch reactor experiments. Runs were made using two different Wyodak coal-derived solvents. Emphasis in this work was directed toward correlation of C₄-700 K distillate yield and cyclohexane conversion as functions of measurable physical, chemical and petrographic properties of the feed coal. Reactivity rankings were found to be the same using either measure of coal reactivity. However, the data indicated that distillate yields were a function of both solvent quality and feed coal properties. For each solvent studied, selected coal properties, including carbon content, total and organic sulphur content, vitrinite content and total reactive maceral (vitrinite plus exinite) content, were found to give statistically significant correlations with distillate production and cyclohexane conversion. Pyritic and sulphate sulphur contents did not appear to enhance liquefaction yield or conversion at the reaction conditions studied. However, any catalytic effects due to pyrite or sulphate sulphur may have been masked by the use of two high quality liquefaction solvents.     

NOx generation from the combustion of Australian brown and subbituminous coals

The formation of NO_x during the combustion of pulverized brown and subbituminous coals from Victoria and Queensland respectively was investigated in an entrainment reactor. As no NO₂ was detected, all the NO_x was present in the form of NO. The brown coals exhibited a significantly greater potential for NO emission under fuel-lean conditions than did the subbituminous coal, even though the latter coal had a higher nitrogen content. However, under fuel-rich conditions the conversion of coal nitrogen to NO for the subbituminous coal was higher than for the brown coals. The differences in conversion efficiency may have been related in part to the nature and reactivity of the volatile nitrogen species. Reactivity differences between the chars produced from the brown and subbituminous coals may also have accounted for different extents of removal of NO. There was a significant reduction in the amount of NO emitted when brown coal was added to a combustion gas stream containing an appreciable quantity of NO before coal injection.     

Effect of coal liquefaction conditions on the trace element content of the soluble non-volatile product

Analysis of 18 trace elements via inductively coupled plasma atomic emission spectrometry has been performed on ‘in-house’ coal-derived non-volatile products. Analyses were conducted in a pyridine matrix to determine the effect of various conversion parameters on metal content. Four subbituminous coals (Wyodak 1–4) and one bituminous coal (Indiana V) were employed in conjunction with both non-basic (tetralin) and basic (1,2,3,4-tetrahydroquinoline) model process solvents. Trace metal data on solvent-refined coals as a function of feed coal, process solvent, reaction time, reaction temperature and extraction solvent are reported. Few trends in metal concentration are apparent on changing various liquefaction parameters. Metal concentrations are, however, approximately one to two orders of magnitude higher in pyridine soluble SRC relative to toluene soluble SRC. The majority of soluble metals, it is therefore suggested, are in the form of coordination complexes rather than true organometallics in SRC. Information regarding the effective molecular size of metal-containing species has been obtained via size exclusion chromatography with specific metal detection. Subtle changes are observed in the effective molecular size of metal-containing materials using different liquefaction parameters. For example, a greater fraction of each metal appears to be bonded to larger ‘sized’ molecules in pyridine soluble fractions than in toluene soluble fractions.     

Mechanisms leading to process improvements in lignite liquefaction using CO and H2S

Optimum distillate yields from US lignites can be as high on a dry, ash-free basis as those obtained from bituminous coals, but only if the vacuum bottoms are recycled. Lignites are more readily liquefied if the reducing gas contains some carbon monoxide and water, which together with bottoms recycle has proven to yield the highest conversions and the best bench-unit operability. The recycle solvent in the reported tests consisted of unseparated product slurry, including coal mineral constituents. Variability in coal minerals among nine widely representative US low-rank coals did not appear to correlate with liquefaction behaviour. Addition of iron pyrite did, however, improve yields and product quality, as measured by hydrogen-to-carbon ratio. Future improvements in liquefaction processes for lignite must maintain high liquid yields at reduced levels of temperature, pressure, and reaction time whilst using less reductant, preferably in the form of synthesis gas (CO + H₂) and water instead of the more expensive pure hydrogen. Understanding the process chemistry of carbon monoxide and sulphur (including H₂S) during lignite liquefaction is a key factor in accomplishing these improvements. This Paper reviews proposed mechanisms for such reactions from the viewpoint of their relative importance in affecting process improvements. The alkali formate mechanism first proposed to explain the reduction by CO does not adequately explain its role in lignite liquefaction. Other possible mechanisms include an isoformate intermediate, a formic acid intermediate, a carbon monoxide radical anion, direct reaction with lignite, and the activation of CO by alkali and alkaline earth cations and by hydrogen sulphide. Hydrogen sulphide reacts with model compounds which represent key bond types in low-rank coal in the following ways: (1) hydrocracking; (2) hydrogen donor; (3) insertion reactions in aromatic rings; (4) hydrogen abstraction, with elemental sulphur as a reaction intermediate; and (5) catalysis of the water-gas shift reaction. It appears that all of these reaction pathways may be operative when catalytic amounts of H₂S are added during liquefaction of lignite. In bench recycle tests, the addition of H₂S as a homogeneous catalyst reduced reductant consumption as much as three-fold whilst maintaining high yield levels when the reaction temperature was reduced by 60°C. Attainment of the high distillate yield at 400°C was accompanied by a marked decrease in the production of hydrocarbon gases, which normally is a major cause of unproductive hydrogen consumption and solvent degradation via hydrocracking. Processing with synthesis gas and inherent coal moisture using bottoms recycle and H₂S as a catalyst appears to be the most promising alternative combination of conditions for producing liquids from lignite at reduced cost.     

Dissolution of subbituminous coal in tetrahydroquinoline

Two different samples of Wyodak subbituminous coal from the Powder River Basin in Wyoming were liquefied in a two dm³ batch reactor using 1,2,3,4 tetrahydroquinoline, THQ, as a solvent. Sufficient sample was produced to determine product boiling ranges by distillation and to measure THQ distribution in the product. Product distillation showed that even at cyclohexane conversions greater than 50%, net distillate yields produced using THQ as a solvent were negative. In some cases, high boiling, coal-derived resid yields were greater than the dry coal charged to the reactor. These observations have been attributed to THQ losses resulting from dimerization of the THQ and reactions between THQ and coal derived components.     

Short contact time liquefaction of coal using hydrogen donor solvent

Liquefaction of coals to form benzene-soluble materials was studied at 400 °C under autogenous pressure conditions using tetrahydroquinoline (THQ), tetrahydroisoquinoline (THIQ) and tetralin (TL) as the hydrogen donating solvents. THQ was the best solvent with a conversion rate of 90% within 15 min for low rank coals (< 80% C). In contrast, it took 50 min to achieve a conversion of 80% when TL was used as the solvent, although both solvents could achieve almost complete conversion of coals into quinoline-soluble material within 10 min. THQ also showed excellent activity with blended coals. Some binary solvents exhibited activities which varied with the THQ content. A 1:3 by weight mixture of THQ and petroleum pitch produced the highest conversion of 100%.     

Liquefaction of partially dried and oxidized coals: 3. Liquefaction results

Samples of partially dried and oxidized Belle Ayr subbituminous coal were liquefied in a recycle donor solvent (SRC-ll heavy distillate) to observe the effect of coal pretreatment on conversion. Because subbituminous coals have moisture contents typically > 25%, it would appear useful to dry these coals prior to liquefaction; however, the drying of Belle Ayr coal, either in nitrogen or oxygen-containing gases, resulted in a significant decrease in yields of liquefied coal products. The liquefaction residues recovered from these runs were examined by optical microscopy and were found to contain high levels of coke. This coke appeared to have formed by polymerization of coal-derived liquid products.     

Determination of specific heats of coal powders by differential scanning calorimetry

Specific heats (C_p) of bituminous and subbituminous coals were investigated in the temperature range 300–360K by differential scanning calorimetry (DSC). To establish the validity of the procedure, specific heats of glass beads and graphites in powdered and bulk form were determined. Good agreement was obtained with the values for the specific heats of glass and graphites in the literature, and it was established that the specific heats were not dependent on the degree of diminution of these materials. Specific heats of coal samples were found to depend upon mesh size, temperature, rank, moisture content and whether the coal powder was wet- or dry-screened. However, there were only minor differences in C_p between bituminous and subbituminous coals.     

Kinetics of thermal liquefaction of coal

A model is presented for the kinetic study of the thermal liquefaction of Belle Ayr subbituminous and Burning Star bituminous coals with anthracene oil, hydrogenated anthracene oil and hydrogenated phenanthrene. All experiments were performed in a continuous-feed, stirred tank reactor, at a temperature of 450 °C and a space time of approximately 5 to 55 min. A kinetic model which includes a reaction: coal + oil→more reactive coal, correlates the data reasonably well. This reaction explains the net consumption of anthracene oil during the initial stages of liquefaction. Such a reaction may account for a portion of the swelling of coal at low space times and the sizable increase of viscosity of reaction slurry during these initial stages of liquefaction. It is also observed that the yield of oil increases when solvents of increasing hydrogen donor capacity are used.     

Solvolytic liquefaction of coals with a series of solvents

Solvolytic liquefaction of coals of different rank was studied with a variety of solvents at 370–390 °C under nitrogen in order to elucidate the role of solvent in coal liquefaction of this kind and to find a suitable solvent for the highest yields of liquefaction. The yield was found to depend strongly upon the nature of the coal as well as the solvent under these conditions. Pyrene and a SRC-BS pitch were excellent solvents for Miike coal, which was fusible with high fluidity at these temperatures. However, the former was less efficient for Itmann and Taiheiyō coals which were fusible at a higher temperature and non-fusible, respectively. The mechanism of solvolytic liquefaction is discussed, including nature of coal and solvent at reaction temperatures, in order to understand the properties required for high yields with non-fusible coals in solvolytic liquefaction. It is found that for liquefaction with a high yield if the coal is non-fusible, solvolytic reaction should take place between solvent and coal, so giving a liquid phase of low viscosity at the reaction temperature. The solvolytic reaction may be one of hydrogen transfer when SRC-BS is used as the solvent.     

Ignition behaviour of different rank coals in an entrained flow reactor

An experimental study to determine the temperature and mechanism of coal ignition was carried out by using an entrained flow reactor (EFR) at relatively high coal feed rates (0.5 g min⁻¹). Seven coals ranging in rank from subbituminous to semianthracite, were tested and the evolved gases (O₂, CO, CO₂, NO) were measured continuously. The ignition temperature was evaluated from the gas evolution profiles, and it was found to be inversely correlated to the reactivity of the coal, as reflected by the increasing values of the ignition temperature in the sequence: subbituminous, high volatile bituminous, low volatile bituminous and semianthracite coals. The mechanism of ignition varied from a heterogeneous mechanism for subbituminous, low volatile bituminous and semianthracite coals, to a homogeneous mechanism for high volatile bituminous coals. A thermogravimetric analyser (TGA) was also used to evaluate coal ignition behaviour. Both methods, TGA and EFR, were in agreement as regards the mechanism of coal ignition. From the SEM micrographs of the coal particles retrieved from the cyclone, it was possible to observe the external appearance of the particles before, during and after ignition. The micrographs confirmed the mechanism deduced from the gas profiles.     

Separation of Solids from Coal Liquefaction Products via Countercurrent Solvent Deasphalting

Abstract

Counter-current deasphalting is a key unit operation in the Dow Liquefaction Process. The Dow developed deasphalter removes typically 99.8–99.9% of the ash and 94–95% of the toluene insoluble hydrocarbons from the coal liquefaction product.

The deasphalted oil, after separation from solvent, contains less than 0.05% ash. Deasphalted coal liquefaction product, a low sulfur, low residue, premium synthetic oil, constitutes 40% of the net liquefaction products. The deasphalter residue typically contains 40% ash and up to 25% hydrocarbonaceous coal solids. It has a heating value of approximately 9000 BTU/1b and is produced as a viscous pumpable fluid. These properties are important in the Dow Process where deasphalter residue is gasified to generate the hydrogen for the liquefaction step. The solvent used in the deasphalter is a paraffinic 50–75° light oil cut from the liquefaction unit.

The counter-current deasphalting technology is based on the results from model and batch studies and results from the operation of two different 23 kg/hr continuous deasphalters. Operational parameters which have been studied include: extraction temperature and pressure, solvent to oil ratio, feedrate, coal oil composition and variations in column design.

The mass transfer which occurs during the deasphalting of coal liquefaction product is accompanied by interfacial tension gradients which develop naturally during the extraction process. The resulting interfacial convection (referred to as a Marangoni instability) profoundly enhances the rate of the extraction process and greatly simplifies the design of the extraction column.     

Liquefaction mechanism of Wandoan coal using tritium and 14C tracer methods: 1. Liquefaction in 3H and 14C labelled solvent

Liquefaction of Wandoan coal using a ³H labelled tetralin solvent which contains a small amount of ¹⁴C labelled naphthalene has been studied at 400 °C under an initial hydrogen pressure of 5.9 MPa, in the presence or absence of NiMoAl₂O₃ catalyst. The amounts of ³H and ¹⁴C transferred from the solvent to the products were measured as liquefaction progressed. The reaction pathways in the presence and absence of the catalyst were discussed and their reaction rate constants were calculated. According to the mass balances of hydrogen and ³H, in the absence of catalyst, tetralin provided coal with hydrogen atoms, and the degree of hydrogen exchange between coal and solvent was small. The catalyst decreased the hydrogen addition from solvent to coal and increased that from gas to coal.     

Interrelationship of chemical characterization and rheological behaviour of coal liquefaction bottoms

In the development of the Exxon Donor Solvent (EDS) process, bituminous and subbituminous coals have been processed in a one ton-per-day coal liquefaction pilot plant using the feed coal slurried with solvent and with or without bottoms recycle. The liquefaction bottoms from both once-through and bottoms recycle operation exhibit non-Newtonian viscometric behaviour. The recycled bottoms, however, are more viscosity/shear dependent, less viscosity/temperature dependent and more thermally stable than the once-through bottoms. Chemical characterization of these bottoms reveals that alkyl and phenoxy groups are important functional groups responsible for the viscometric behaviour of bottoms. The increase of bottoms viscosity is postulated to involve the elimination/condensation of methylene units are phenolic functional groups to form the crosslinkages of large aromatic clusters.