K A = adsorption equilibrium constant, ml/g of foam material KA' = adsorption equilibrium constant of pure carbon particles, l./mg of particles kf = mass transfer coefficient, cm/sec M o = zeroth moment o f t vs. (1 - C/Co) curve, min M I = first moment, min2 Mz = second moment, min3 M3 = third moment, min4 No = capacity of adsorbent Q = flow through sample, l./min R = radius of spherical particle of adsorbent, cm r = length coordinate in the spherical particle of adsorbent, measured from the center of the particle, cm T = temperature of sample and gas, OC t = time,min 1/UA = overall mass transfer coefficient, cc of CC14/(g of foam min) V = interstitial velocity, cm/min Wt = sample weight,g z = length coordinate of bed of adsorbent, cm
Greek Letters a = cloth porosity (void volume/total volume) = intraparticle void fraction of carbon particle 6 = weight of particles per unit weight of foam material t = void fraction representing the bed volume not occupied by carbon particles divided by the total volume pm = apparent density of foam material, g of solids/cc of total volume pp = apparent particle density, g of carbon/cc carbon
Literature Cited Aris, R., "On the Dispersion of Linear Kinematic Waves", Roc. Roy. SOC. London, Ser. A, 245, 268 (1958). Marshall, W. R., Jr., Pigford, R. L., "The Appleation of Differential Equations to Chemical Engineering Problems". University of Delaware, Newark, Del., 1947. Masamune. S., Smith, J. M., AIChE J., 11, 34 (1965). Rosen, J. E., J. Chem. Phys., 20, 387 (1952). Schneider. P.. Smith, J. M., AlChEJ., 14, 762 (1968). Vogel, R. F., Mitchell, 6.R., Massoth. F. E.,Environ. Sci. Techno/., 8, 432
(1974).
Receiued for review M a r c h 10, 1975 Accepted September 29,1975
Startup Solvent Selection for the Liquefaction of Lignite Warren P. Scarrahl and Richard R. Dlllon Project Lignite. Department of Chemical Engineering, University of North Dakota, Grand Forks, North Dakota 5820 1
The liquefaction of lignite involves the reaction of lignite and hydrogen in the presence of an organic solvent which acts as a hydrogen donor. During continuous operation the solvent will be produced from the lignite: however, during the initial startup-a solvent from an outside source must be provided. Batch autoclave liquefactions were run with 15 coal-derived and 10 petroleum-derived solvents to screen potential startup solvents. The criteria for comparing the solvents were divided into (1) pre-liquefaction characteristics and (2) liquefaction performance. Pre-liquefaction characteristics included (a) the percentage of liquefaction solvent contained in the raw solvents and (b) the fluidity of the raw solvents. Liquefaction performance included (a) lignite conversion and product yields, (b) flow and filtration characteristics, (c) product hydrogen:carbon atom ratios and sulfur contents, and (d) recycle solvent properties (aromaticities, sulfur contents, and recoveries). Several coal and petroleum-derived solvents were found to be suitable for startup purposes.
Introduction Solution hydrogenation is a process being extensively investigated for converting coal into liquid, gaseous, and upgraded solid fuel products. Coal and hydrogen gas are reacted in the presence of an organic solvent which acts as a hydrogen donor; the hydrogenation products are then separated from the unconverted coal and mineral matter (Kloepper et al., 1965; Severson et al., 1970, 1973). At the University of North Dakota one goal of Project Lignite, established under the sponsorship of the Office of Coal Research, is to study the solution hydrogenation (liquefaction) of lignite. A continuous process development unit (PDU) is currently under construction; the principal product will be an upgraded solid fuel, solvent refined lignite (SRL). A requirement for a viable process is that it generate its own solvent. However, during the initial startup, there will have been no previous operation of the process and solvent will have to be supplied from an outside source. The experiments to be described were run to screen potential solvents and select one or more suitable for the PDU startup. 122
Ind. Eng. Chem., Process Des. Dev., Vol. 15. No. 1, 1976
Previous tests had established the fraction of the reactor discharge slurry (RDS) which would be separated and recycled as solvent. Light material had little tendency to dissolve the lignite while vacuum distillation of the filtrate fixed the upper temperature a t which liquid could be readily separated from the dissolved solid product. Although some of the solvents investigated were used "as received" from the supplier, the "standard" solvent that will normally be recycled throtigh the process will consist of material boiling between 100 and 2 3 O O C a t an absolute pressure of 1.6 mmHg. While it appears logical that a coal-derived solvent should be used in the initial startup of the PDU, the current demand for coal-derived solvents is so great that their availability for large-scale operation is questionable. Therefore it was decided that potential petroleum-derived solvents should also be screened. Coal-Derived Solvents. The solvents evaluated were creosote oil, anthracene oil, Light Creosote, and MiddleHeavy Creosote. The first two solvents were obtained from a manufacturer of coal tar products while the latter two
Commercial Laboratory -
Sol ven t
Solvent
Cut
R u n No.
Treat ment Troatment -
Heavv Ends
463
Raw
Anthrac&,..
Solvent Cut
HAN
As Received
No 6 Fuel Oil
Standard Solvent
R u n No
519
473.478
513
Creosote Oil
.--
Solvent -
n.,
IL d
Raw
~
1-
Heavv Ends Standard Solvent
462 475
Hvdroaenated
S t a n u n t
474
Heavv E n d s Standard Solvent
51 I 455
I IHvdroaenated
Y H y d r w e n a t e d
1
S t a n d a r d Solvent 468 Heovv Ends SI0 Standard Solvent 4 6 5 . 4 8 3 . 4 8 4 S t a n d a r d Solvent
471,480
Light Creosote
As R e c e i v e d
503
Middle-Heavv C r e o s o t e
As R e c e i v e d
507.508
Figure 1. Hierarchy of experimental liquefaction runs using coalderived solvents.
solvents were by-products from the coke production facilities of a steel manufacturer. Creosote and anthracene oils differ in that the former is a lighter fraction of the coal tar. Two types each of creosote and anthracene oils were acquired: (1)“raw” oils that were separated only by an industrial distillation process and (2) “chilled” oils that had been slowly chilled and decanted after distillation to remove crystalline material, e.g., carbazole. Some of these oils were subjected to hydrogenation to determine the effect it would have on their hydrogen-donor characteristics. Finally, these oils were further refined by one of two methods: (1) the “heavy ends” were recovered by topping, i.e., stripping off the components boiling a t temperatures less than 350’F during atmospheric distillation, or (2) the “standard” solvent was obtained by retaining the fraction boiling between 100 and 230’C during vacuum distillation a t 1.6 mmHg of the “as received” solvent. This “standard” solvent corresponded to a cut with an atmospheric pressure boiling range of approximately 537 to 830’F. The Light Creosote and Middle-Heavy Creosote were used “as received” from the manufacturer. The hierarchy of coal-derived solvents is shown in Figure 1 along with the corresponding experimental run in which each was used. Petroleum-Derived Solvents. Listed in order of progressively heavier fractions, the commercial petroleum solvents tested were: HAN (highly aromatic solvent), No. 6 Fuel Oil, No. 5 Fuel Oil, Aromatic Tar S-2 (the bottoms from a steam cracking unit), FS-120 (carbon black feedstock), and Aromatic Concentrate (heavy catalytic cracking recycle stock). Sulfur contents were higher for all the petroleum-derived solvents than for those derived from coal. These solvents were used (1)“as received” from the manufacturer or (2) after separating a “standard” solvent cut by vacuum distillation of the “as received” solvent. In Figure 2 the petroleum-derived solvents and corresponding experimental runs are summarized.
Criteria for Evaluation I t was necessary to both (1)determine solvent characteristics prior to liquefaction and (2) evaluate the effects of the solvent on the liquefaction process. The former information would indicate the type and extent of facilities required for startup solvent preparation; the latter information would determine the effects of the startup solvent on the liquefaction process itself. The criteria selected to compare the solvents were the following. 1. Pre-Liquefaction Characteristics. (a) “Standard” Solvent Yield. The percentage of the “as received” solvent that boiled in the “standard” solvent range was determined
486
496.509 495.499 Aa Received
514
Stondard Solvent
485.497 500.506
A t m u t i c Concentrate
vent
494.501
Figure 2. Hierarchy of experiment,alliquefaction runs using petroleum-derived solvents.
by vacuum distillation at 1.6 mmHg. The remaining solvent could be removed prior to liquefaction or separated from the solvent prepared for recycle after liquefaction. (b) Handling Characteristics. Solvents that were liquids a t room temperature would be much easier to handle during slurry preparation and transfer than those that were solid; room-temperature viscosity measurements provide a quant,itative comparison among the liquid solvents. 2. Liquefaction Performance. (a) Conversions and Yields. Yields were based on the moisture and ash-free (MAF) lignite to allow for the comparison of runs having variations in the quantities and/or compositions of the input solvent, gases, and water. The weights of the experimental outputs were first adjusted to give 100% recovery (actual recoveries ranged from 94.3 to 99.6 wt %). Macroscopic material balances were then made on the (1) gases, (2) water (including lignite moisture) and ash, and (3) the MAF liquids. Because of the adjustment to 100% recovery, the conversion of the MAF lignite was equal to the sum of the weight changes obtained from the above three material balances. The net product distribution based on MAF lignite was then reported as: gas, liquid, unconverted, and water. The liquid was further distributed as light oil (boiling below 100°C a t 1.6 mmHg) and net solvent refined lignite (SRL). Net “gas” and “light oil” yields indirectly indicated the durability of the solvent as they were partially formed from the cracking of the solvent. “Net SRL” indicated the usefulness of the solvent for converting lignite into the desired product. “Unconverted” represented the degree of lignite liquefaction attained and demonstrated the suitability of the solvent for solution hydrogenation. The negative net water yield primarily represented the generation of hydrogen via the water gas shift reaction and could indicate any solvent effect on that reaction. (b) Flow Characteristics. Areas of particular concern were the tendency of process streams to plug lines and the filtration of the reactor discharge slurry. To compare filtration characteristics an empirical filtration index was developed. The filtration index is the percentage of the total recycle solvent that was retained in the filter cake, residue, and pot residue; therefore, it was a measure of (1)the ease of separation of the filter cake and filtrate and (2) the tendency of the reactor discharge slurry to adhere to equipment surfaces. (c) Product Characteristics. Particularly important characteristics of the SRL related to the solvent were its hydrogen:carbon atom ratio and sulfur content. The hydrogen:carbon atom ratio indicated how well the solvent acted as a donor of hydrogen to the lignite. It was also important to determine the relationship between the sulfur content of the solvent and that of the product. Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976
123
Table I. Test Conditions for Batch Autoclave Liquefaction of Lignite
INPUTS
EXPERIMENTAL OUTPUTS
CALCULATED PRODUCTS
-
Reactants: Lignite (-100 mesh), 200 g (MAF) Solvent, 400 g Syngas, 1:l molar ratio of CO to H, Water (including lignite moisture), 100 g Initial pressure, 1000 psig Maximum pressure, 2850-3300 psig Test temperature, 752°F Run time (at temperature), 30 min
(d) Recycle Solvent Characteristics. Infrared ratios (IRR), sulfur contents, and solvent recoveries were measured. The IRR indicated the relative unsaturatedness or aromaticity of the solvent-the higher the value the greater the proportion of aromatics present (Wright and Severson, 1972). Comparison of the IRR of the input and recycle solvents provided a measure of change due to hydrogenation of the solvent. A comparison of the sulfur contents of the input and recycle solvents indicated whether sulfur had been removed from or added to the solvent. The solvent recoveries were calculated assuming that there were no losses in the process; recoveries less than 100% implied that the remainder of the recycle solvent would be supplied by using part of the solvent refined lignite. Recoveries of “heavy ends” or “as received” solvents were determined by comparing the quantities of recycle solvent with the “standard” solvent fraction of the input solvent; the latter had been determined by vacuum distillation of samples of “heavy ends” and “as received” solvents. Experimental Section Materials Used. Lignite was obtained from North American Coal Corp., Zap-Indianhead Mine, Zap, North Dakota. Solvents used were the following. Raw and chilled creosote oil was obtained from Reilly Tar and Chemical Corp., Indianapolis, Ind. Raw and chilled anthracene oil was obtained from Reilly Tar and Chemical Corp., Indianapolis, Ind. Light Coal Tar Fraction (Light Creosote), distillation range 230 to 315OC, was obtained from United States Steel Chemicals, Pittsburgh, Pa. Middle-Heavy Coal Tar Fraction (Middle-Heavy Creosote), distillation range 315 to 355OC, was obtained from United States Steel Chemicals, Pittsburgh, Pa. HAN (highly aromatic solvent) was obtained from Exxon Chemical Company, Houston, Texas. No. 6 Fuel Oil was obtained from Exxon Company, U S A . , Billings, Mont. No. 5 Fuel Oil was obtained from Standard Oil Company, Mandan, N.D. Aromatic Tar S-2 (bottoms from steam cracking) was obtained from Exxon Chemical Co., Houston, Texas. FS-120 (Gulf carbon black feedstock) was obtained from Gulf Oil Corp., Port Arthur, Texas. Aromatic Concentrate (heavy catalytic cracking recycle stock) was obtained from Exxon Chemical Co., Houston, Texas. Procedure Equipment. All experiments were conducted in a 1-gal MagneDrive 316 stainless steel autoclave manufactured by Autoclave Engineers, Inc., Erie, Pa. The autoclave unit was equipped with a magnetic stirrer assembly, thermocouple well for monitoring reaction mix temperature, reactor discharge slurry withdrawal tube, gas ports, and a four-element 9-kW jacket heater. Compressed gases of nitrogen (N2), carbon monoxide (CO), and hydrogen (H2) used in the experiments were manifolded to the autoclave with 78in. 0.d. high-pressure stainless steel tubing. Temperature control was achieved by a Honeywell Electr-0-Volt Con124
Ind. Eng. Chem.. Process Des. Dev., Vol. 15, No. l , 1976
WATER
J
t
NPE@’
4
,MINERAL RESIDUES
“‘PYRIDINE EXTRACTABLE @’NON-PYRIDINE MTRACTABLE
Figure 3. Schematic of material balance calculations. troller equipped with two Electronik 15 strip chart recorders. The temperature controller permitted either manual or automatic control of heating rate; the recorders enabled the operator to monitor both external and internal temperature. The autoclave internal pressure was also monitored on one of the above recorders via a pressure transducer unit. 2. Test Operation. Test conditions are summarized in Table I. The actual lignite charge was about 300 g due to moisture and ash; however, the solvent to MAF lignite weight ratio was maintained a t 2:l. The lignite, solvent, and water were charged and the reactor assembled and leak-tested with N2 at 1000 psig. The nitrogen was bled off and the reactor purged with CO. The reactant gases were then added to a pressure of 1000 psig and a gas sample analyzed to ensure there was a 1:l molar ratio of CO to Hz. The reactor was heated at a rate of 5OF/ min until the test temperature of 752’F was attained; the test temperature was maintained for 30 min. Upon completion of a test, the autoclave was cooled to about 400°F, and the product gases were exhausted a t a rate of about 0.3 cfm through a series of cold traps into a gas bag. Water and light oil fractions were collected in the cold traps. After discharge of product gas, the autoclave was repressurized with CO and the slurry, still at 400’F, discharged through a dip-tube onto a heated Buchner funnel where the solvent and coal-derived liquids were separated from the mineral constituents and unconverted coal via suction filtration. The products collected were filtrate, filter cake, and residues. Product Workup. As shown in Figure 3 the converted input materials were handled as either experimentally measurable outputs or as calculated products. The calculated products would have been recovered from a facility that included separation equipment as well as a reactor. Inputs included weighed quantities of added water, solvent, and lignite. The gas input was calculated based on calibrations of the reactor made at different gas pressures and compositions. Experimental outputs included gas, water, condensed light oils, filtrate, cake, pot residue, and residue. The quantity of gas was determined from its volume and density. The water and condensed light oils were the weighed quantities collected in the cold traps. The filtrate and cake were weighed after the RDS material was filtered. The pot residue was the weighed quantity of material remaining in the reactor after emptying while the residue was the weight of the material adhering to the filtration assembly. The calculated products include light oils, reclaimed solvent, SRL, and mineral residues. The following will de-
scribe how the distribution of these calculated products was obtained. The oils were assigned to the filtrate. The pot residue and residue were extracted with pyridine and the soluble material assigned t o filtrate with the insoluble material assigned to mineral residues. The pyridine soluble portion of the cake was then assigned to the filtrate and the remainder of the cake to mineral residues. A vacuum distillation of a portion of the filtrate was the basis for distributing all the filtrate into light oils, reclaimed solvent, and SRL. Analytical Methods. Standard (ASTM) procedures were used to characterize the raw lignite and SRL, liquefaction solvents, and intermediate coal liquid products obtained from the batch autoclave liquefaction experiments. Moisture of pulverized lignite was analyzed per ASTM D3173. Ash content of the lignite was determined per ASTM D3174. Volatile matter was determined on raw lignite per ASTM D-3175. Carbon and hydrogen were determined on all starting materials and coal products via the quartz tube combustion technique per ASTM D-271. Total sulfur was determined on both liquid and solid samples using an induction furnace technique per ASTM D-1552, modified for use with an automatic titrator. Total nitrogen in both liquid and solid samples was determined using the Kjeldahl-Gunning method per ASTM D-271. Specific gravity of input solvent, recycle solvent, and filtrate was measured per ASTM D-1298. Pyridine extractable hydrocarbon was determined in filter cake and coal residue products per ASTM D-473-69. Calorific value of the SRL from the batch autoclave test was determined per ASTM D 2015-66. Infrared ratio (ratio of aromatic t n aliphatic hydrogen) was determined on both input and recycle solvents; the method used an ir spectrophotometer which scanned the ir spectrum of the solvent sample over the wavelength range of 2.5 to 16.0 w. Relative intensity of the absorbance of aromatic vs. aliphatic hydrogen was measured a t 3.28 and 3.41 p, respectively, to determine the degree of aromaticity in the solvent. Brookfield viscosity was determined a t room temperature on liquid samples using a Brookfield Model LVT Synchro-Electric viscometer with a Model C Helipath stand; spindle sizes and configurations varied depending upon the nature of the sample. Hydrogen sulfide in product gas was determined by bubbling the gas sample through ammoniacal zinc sulfate solution and then using an iodometric titration technique to determine the amount of hydrogen sulfide absorbed. Specific gravity of the product gas was determined by the evacuated glass bulb technique a t room temperature. Composition of the product gas was determined by gassolid chromatography using both a Porasil A and a Linde 5A molecular sieve column a t 100OC. The Porasil column was used to resolve ethane, carbon dioxide, and traces of C:< and Cd hydrocarbons in the gas sample; the Linde column was used to resolve hydrogen, nitrogen, methane, and carbon monoxide. Distillation Methqds. 1. Atmospheric Distillation. Atmospheric distillation was performed per ASTM D-24659 on all incoming solvents used in the coal liquefaction tests as a means of characterizing basic properties, e.g., initial boiling point temperatures and estimated true boiling point curves at atmospheric pressure. 2. Vacuum Distillation. ”Standard” liquefaction solvent and all recycle solvents were prepared by vacuum distillation at 1.6 mmHg. The apparatus included a Vigreux glass fractionating column with an effective length of 15 cm and a straight-tube water-cooled condenser. Three boiling range fractions were taken during the distillation; IBP to
100°C (light oil), 100 to 230°C (solvent fraction), and 230 to 255°C (heavy oil fraction).
Results and Discussion Coal-Derived Solvents. Data collected during the liquefaction of lignite using coal-derived solvents is presented in Tables 11, 111, and IV. The creosote oils, anthracene oils, and by-product creosotes will be compared relative to the previously outlined criteria. Vacuum distillation yields of “‘standard” solvent from the “as received” solvents were about 70 wt % for creosote oils and 85 wt % for anthracene oils; thus anthracene oils would require less makeup solvent during the startup period. The by-product Light Creosote contained less than 30% “standard” solvent; no distillations were made on the Middle-Heavy Creosote. Regarding handling characteristics, the ratings of the solvents from best to worst were: 1, Anthracene oils; 2, Light Creosote; 3, Creosote oils; 4, Middle-Heavy Creosote. The yields of unconverted lignite ranged from 3.1 to 15.4 wt % with most in the range of 4.6 to 10.6 wt %. Gas yields and water consumptions (water gas shift reaction) showed little variation; the range of the former was 34.9 to 46.5 wt % and of the latter 8.1 to 17.3 w t %with one exception (Run 479). Light oil and net SRL yields were related; a high light oil yield would result in a low net SRL yield and vice versa. The primary source of light oil appeared to be the solvent rather than the lignite; the runs with the highest light oil yields (36.2, 48.8, and 51.4 wt %) also had correspondingly low solvent recoveries for recycle (85.8, 76.1, and 79.9 wt %). Therefore, a high light oil yield indicated that more of the lighter portion of SRL would be required as makeup for the recycle solvent. Creosote oils had much higher light oil yields than the anthracene oils and by-product creosotes. The “standard” solvent cut of the creosote oils had light oil yields of 20.2-51.4 wt % compared to net light oil yields of 0-6.5 wt % for the “heavy ends” cut. Except for Runs 511 and 455 no significant difference in light oil yields from the two anthracene oil cuts were apparent. In addition, when “standard” solvents from untreated and hydrogenated creosote oils are compared it is apparent that the light oil yields are much higher for the hydrogenated oils (20.2-36.2 wt % vs. 48.8-51.4 wt 96). This indicates that hydrogenation affects the stability of the solvent as well as its ability to act as a hydrogen donor for coal. Hydrogenation of anthracene oil resulted in no significant difference in light oil yields. The RDS withdrawal tube leading to the filter became plugged following liquefaction for four solvents: 1, “standard” raw creosote oil; 2, “standard” hydrogenated raw anthracene oil; 3, “standard” chilled anthracene oil; 4, “as received” Light Creosote. However, once slurry flow was initiated no further plugging was experienced. The ratings of the solvents from best to worst relative to the filtration characteristics of the RDS were: 1, Creosote oils; 2, Light Creosote; 3, Anthracene oils; 4, Middle-Heavy Creosote. From an operating viewpoint, the relative ease of filtration of RDS when using the creosote oils is attractive due to the anticipated difficulty in finding satisfactory filtration equipment. The only SRL hydrogen:carbon atom ratios as large as 0.79:l were obtained using hydrogenated creosote oils (0.8O:l-0.84:l); therefore hydrogenation of the solvent did result in increased hydrogen transfer to the SRL. The only other noticeable difference in SRL hydrogen:carbon atom ratios was the low values (