Energy & Fuels 1993, 7,610-619
610
Pyrolysis Kinetics and Maturation of Coals from the San Juan Basin John G. Reynolds and Alan K. Burnham* Lawrence Livermore National Laboratory, University of California, Livermore, California 94551 Received December 21, 1992. Revised Manuscript Received May 3, 1993
Temperature-programmed pyrolysis was used to analyze 18 coals and coaly shales from the San Juan Basin of the southwestern US.,giving temperatures of maximum evolution rate (2'-) and pyrolysis yields. T m , values increased with maturity (as measured by vitrinite reflectance % Rm]). The pyrolysis yields remained constant or increased slightly with increasing maturity until a value for 9%Rm of approximately 1.0 was reached, after which the yields declined rapidly. Pyrolysis-gas chromatography and pyrolysis-FTIR showed that alkane and alkene yields followed similar trends. A subgroup of coals from the Fruitland seam of the San Juan Basin was also analyzed by micropyrolysis at several constant heating rates to determine laboratory pyrolysis kinetics. The kinetic calculations yielded the energy of activation by the approximate method (Eappr03 and the principal energy of in the range 55-57 kcal/mol for the coals in the activation by the discrete method (principal E&erets) 9% Rm range 0.4-0.9. However, the coal with the highest 9% Rm (1.30) had Eapprox and principal E h e b values around 63 kcal/mol. These Fruitland seam coals were also extracted with organic solvents, and the kinetic parameters were found to be within experimental error of their unextracted counterparts. Application of the multiple parallel reaction model to the kinetic parameters determined for the four Fruitland seam coals showed only fair agreement between calculated and measured values for Tm, and yield relationships with maturity. This inability of the model to adequately describe the measured data is attributed to the pyrolysismechanism changing with increasing maturity.
Introduction The San Juan Basin in the southwestern U.S.(Figure 1)is a major producer of hydrocarbon gas. As of December 1987, the resource was evaluated to contain 16.2 tcf of nonassociated gas, 425 bcf of associated gas, 60 million bbls of condensate, and 23 tcf of nonassociated gas potential. Of this potential, coal is considered the source of gas in the Four corners platform and Chaco slope.1.2 Recently, there has been activity in the exploration for gas in the upper Cretaceous Fruitland seam, which is the major deposit of coal in the basin. The Fruitland seam is considered to have the potential of 50-56 tcf of gas production, as well as the largest reserves of coal in the basin-approximately 200 billion tons. The seam is also thickest and most continuous in the low part of the basin, making the basal coals of the Fruitland formation the primary target of gas exploration. Below the Fruitland seam is the Pictured Cliffs sandstone which also produces mainly nonassociated gas. Most of the gas generation is probably due to thermal maturation over geological time. Although the gases produced from the Fruitland seam and the Pictured Cliffs sandstone are similar in isotopic composition, they are quite different chemically-Fruitland gas is almost all CHd with only 6% CO2; Pictured Cliffs gas has an additional 15% C2+ componenb and less than 2% CO2. Several features of the Fruitland formation are important to note with respect to the position in the basin: (1) (1)Rice, D. D.; Threlkeld, C. N.; Vuletich, A. V.; Pawlewicz, M. J. Oil Gas J . 1990, Aug 13,60431. (2) Faesett, J. E. In Oil and Gas Section of: Economic Geology of the
UnitedStates;Rice,D.D.,Ed.;DecadeofNorth AmericanGeology Series; Geol. Soc. Am.: Boulder, CO, 1992; Vol. P-2.
Coal rank increases to the northeast from subbituminous to medium-volatile bituminous [probably due to (i) a deeply buried heat source located in the northern part of the basin or (ii) the circulation of relatively hot fluids into the basin from a heat source located in the vicinity of the volcanic San Juan Mountains31. (2) Coal gas is isotopically heavier and chemically drier with increased burial depth and increasing source-coalmaturity.' (3) Coal gas is mostly methane with decreasing amounts of C02 accompanying increasing levels of source-coal m a t ~ r i t y . ~ Laboratory pyrolysis of foseil fuel samples has been shown to have relevance in the understanding of hydrocarbon formation from source rocks (for a recent review, see Ungerer6). One such method for this analysis is the Pyromat I1 micropyrolyzer, which has been utilized to measure the laboratory pyrolysis kinetics of several oil shales and source rocks.6 Through selected reaction models, kinetic parameters have been deduced and extrapolated to estimate relevant maturation parameters. We have developed both type I and I1 kerogen reaction models. However, work continues on a type I11 kerogen model, which is relevant to coal-bed methane and relatively deep gas formation from hydrogen-poor kerogens. Although much has been done on source rocks and oil shales with the Pyromat I1 pyrolysis technique, little has been done on coals. This report examines the evolution behavior at selected heating rates and the kinetic parameters derived from analysis of evolution behavior at (3) Law, B. E. Geol. SOC. Am. Bull. 1992,104,192-207. (4) Rice, D. D.; Threlkeld, C. N.; Vuletich, A. K.; Pawlewicz, M. J. Rocky Mt. Assoc. Geol. 1988, 51-59. (5) Ungerer, P. Org. Geochem. 1990,16, 1-25. (6) Braun, R. L.; Burnham, A. K.; Reynolds, J. G.; Clarkson, J. E. Energy Fuels 1991,5, 192-204.
0887-0624/93/2507-0610$04.00/00 1993 American Chemical Society
Energy & Fuels, Vot. 7, No.5,1993 611
Coals from the Son Juan Basin 109"
108"
I
I
I
I
107"
I
La Plata
' COLORADO.,----.
Alamos
Figure 1. Map of the San Juan Basin in the four corners area of the southwestern U.S.showing the locations of the samples studied (see Table I for well names): (1)HAFA, HAFB, and HAFC (23N, 7W, sec 29);(2) CHSA, CH5B, CH5C, CH5D, and CH5E (23N,low, sec 5); (3) M O M and M O B (30N, 8W, sec 5); (4) KB5A and KB5B (31N, 8W, sec 28); ( 5 ) GR3A and GR3B (31N, 11W, sec 20); (6) CUBA (32N, 5W, sec 23); (7) SJ8A and SJ8B (32N, 6W, aec 27); (8) CU5A {32N,4W, sec 17). Table I. Selected Information on Coals from the San Juan Basin
sample ID HAFA0 HAFBO HAFCO CH5Ab CH5Bb CH5Cb CH5Db CH5Eb MOAAe MOABe KE35Ad KB5Bd GR3Ae GFt3Be CU2Af sJ8A8 SJ8W CU5Ah
formation Fruitland Menefee Menefee Fruitland HogMountain Hog Mountain Menefee Menefee Fruitland Fruitland Fruitland Fruitland Fruitland Fruitland Fruitland Fruitland Fruitland Fruitland
depth, f t vitrnite reflectance, %R, 0.46 1528-1534 0.57 3049-3054 3-3810 0.58 950-960 0.44 1880-1890 0.52 0.49 1910-1920 3170-3180 0.54 0.54 321+3220 2927-2937 0.76 0.82 3085-3106 0.82 3046-3056 3169-3184 0.93 0.62 2520-2540 254+2550 0.66 1.08 3-3990 1.22 3150-3160 3148-3158 1.24 1.30 4180-4200
Wells:a Henry AGC Fed #1(Yam Petroleum Co.). b Champ #5 (Dugan Petroleum (20.1. Moore A #8 (Amoco Production Co.). d Kemaghan B #5 (AmocoProductionCo.). Grenier #lo3 (Meridian Oil). f Carram Unit 23A #2 (Nassau Resources). 8 San Juan 32-5 #lo8 (Meridian Oil). Carram Unit 17B #15 (Nassau Reeourcea).
multiple heating rates measured by the Pyromat I1 micropyrolyzer on selected coals from the San Juan Basin. In addition, the chemical behavior of the coals during pyrolysis is also examined in efforts to understand changes in product evolution mechanismsas a function of maturity.
Experimental Section Samples. Vito Nuccio (USGS) supplied 18 samples from various parts of the San Juan Basin-Fruitland formation, 12; Menefee formation, 4; Hog Mountain tongue, 2. Table I shows
the formation, the aample identification symbols,the well and depth (feet),and the vitrinite reflectance values (mean random
reflectance values), and Figure 1 locates the wells in the basin. The samples range from essentially subbituminous rank to medium-volatilebituminous (%R, of 0.44-1.30). The depths of the Fruitland samples vary considerably and are not directly correlatedwithrank probablybecauseofeither localized intrusive heating or nonuniform uplift and e r o s i ~ n .For ~ Henry AGC Fed #1, the shallowersamples are from the Fruitland formation, and the deeper samplesare from the Menefee formation (correlation charts show that the well probably passes through the Pictured Cliffs and Cliff House sandstone formational). For Champ #5, the shallower sample is from the Fruitland formation, more to the North than the Henry AGC Fed Iy1 well, the intermediate samples are from the Hog Mountain tongue, and the deepest samples are from the Menefee formation. The maturity of the Menefee formation samples from these two wells is only slightly higher than that of the shallower Fruitland samples. All samples were received in sealed plastic bags. All were visually dry. The whole allotment (5-10 g) was ground with a mortar and pestle in a nitrogen-purged glove bag to inhibit oxidation. Homogeneityproblems were encountered with some samples, and these were reground. Extraction of 0.2-0.5 g of each sample was performed in a micro-Soxhlet extractor for 36 h using either the 92% CH2Cld 8% MeOH azeotrope or 100% tetrahydrofuran (THF) as the extraction media. After the extraction was complete, the extracted coal was dried in oacuo and the solvent containing the extracted bitumen was evaporated under a stream of N2 gas. Mass balancesexhibited over 98% recovery forthe CHgldMeOH extractions. The THF extraction exhibited over 100%recovery after extensivedrying,indicatingsome permanent incorporation of the solvent into the sample. Elemental Analyses. The samples were analyzed for total C, H, and N by either a LECO (St. Joseph's, MI) 800 C, H,and N analyzeror a Heraeus (Germany)C,H, N, and 0 Rapid Analyzer and for mineral C02 by collection of HC1-liberatedCOPon ascarite. Total organic carbon (TOC) was determined by the difference of total and mineral carbon. Table I1 lists the results of these analyses (there was not enough CH5D sample to analyze). Most of the samples have TOC levels which fall in the range of 30 to
612 Energy &Fuels, Vol. 7, No. 5, 1993
Table 11. Selected Elemental Analyses for Coal and Coaly Shale Samples from the San Juan Basin. totalc, totalH, totalN, mineral TOC, sampleb % % % coz, % % 3.8(0.1) 1.3 (0.1) 1.0 41.6 HAFA 41.8(0.4) 1.3(0.2) HAFB 40.2 (0.1) 3.6 (0.0) 0.5 40.1 1.4 (0.2) 0.5 59.6 HAFC 59.7 (1.3) 4.7 (0.1) 1.1(0.1) 2.7 51.2 CH5A 51.9 (0.2) 4.0 (0.1) 1.5(0.2) 0.6 61.7 CH5B 61.9(0.1) 4.8 (0.2) 1.1(0.0) 0.7 45.9 CH5C 46.1 (1.2) 3.7 (0.1) na na na na CH5D nac 0.6 25.9 CH5E 26.1 (3.2) 2.3 (0.1) 2.2 (0.1) 2.3 37.6 MOAA 38.2 (3.8) 3.1 (0.1) 1.1(0.2) 0.9 63.0 MOAB 63.2 (1.9) 4.1 (0.1) 1.5(0.0) 1.4 (0.2) 1.4 59.9 KB5A 60.2 (1.7) 4.4 (0.3) 1.9 (0.3) 1.4 63.7 KB5B 64.1 (1.1) 4.5 (0.1) 18.4 (0.3) 2.0(0.0) 0.6 (0.0) 1.8 17.9 GR3A 1.6(0.1) 1.7 55.2 GR3B 55.7 (2.3) 5.0 (0.9) CU2A 1.9 (0.1) 0.5 (0.1) ndd 2.9 1.1 1.2(0.1) 1.0 36.1 SJ8A 36.4(0.6) 2.7 (0.1) 33.4 (1.3) 2.4(0.1) O.g(O.0) 1.3 33.1 SJ8B 1.2(0.1) 0.1 55.8 CU5A 56.3 (0.6) 3.3 (0.1) "Values in parentheses are the standard deviation for three determinations. Samples were analyzed on an as-received basis. Not enough sample to analyze, na. Analyzer problem with N determination,nd. above 60 wt % ,with KB5B being the richest. Two are less than 30 wt % , and one, CUZA, is extremely lean in organic matter. This sample exhibits anomalous evolution behavior (see below). The samples are also very low in carbonate content (%COz), indicating that the mineral matter is primarily silica based. Ash, sulfur, and oxygen content were not determined because they were outside the scope of this study. Kinetic Analysis. The method of kinetic analysis using the Pyromat I1 has been described in detail elsewhere.6 Briefly, the Pyromat I1pyrolyzessmallsamples (2-15 mg) at constant heating rates (0.5-75 OC/min) under He flow and detects evolvingvolatile material with an FID detector. Temperature is measured by direct contact of a thermocouple (type K with a 0.040-in. 304 stainless steel sheath) with the sample. The kinetics were determined from multiple runs at constant heating rates on approximately 4-10-mgsamples. Generally, three 50 "C/min, one 7 "C/min, and two 1"C/min runs were performed. If T,, values and profile shapes were not in agreement, more runs at these heating rates were performed. Kinetic calculations were determined by both the T--shift method (yieldingapproximate parameters, E,,, and A,,,J and the discrete distribution method (yielding discrete parameters, principal Ebb and A w ) . Rate data were analyzed by using the regressionanalysis program KINETICS? which was applied previously to the Argonne premium coals.8 Yield Analysis. Pyrolysis yields were determined for all samples by comparison to the yield from AP22 oil shale. This yield has been determined from Fischer Assay and Rock-Eval analysis to be 88 mg pyrolysate/gram oil shale. The furnace in the Pyromat I1 was replaced during the course of this work. For most samples, the yield was determined from two or three runs at the nominal heating rate of 25 "C/min, from 250 to 700 "C using the first furnace. The AP22 standard was run twice daily, and a single calibration factor was determined for the entire set of runs (over a period of 4 days). For the four samples used in the kinetic analyses,severaldeterminations were performed using the second furnace. To assure more accurate yields,the standard was run immediately before each sample. The standard values were then averaged each day, providing a daily calibration factor. Multiple determinations were made for most of the properties measured in this study. Full listing of the T-, yield values, calibrations, and kinetic calculations are reported elsewhere? (7) Burnham,A. K.; Braun, R. L.; Gregg, H. R.; Samoun,A. M. Energy Fuels 1987, 1, 452-458. (8) Burnham,A. K.; Oh, M. S.; Crawford,R.W.; Samoun, A. M. Energy Fuels 1989, 3, 42-55.
Reynolds and Burnham Fruiland H~gMOunlain
- Calculated
0.4
I
I
I
I
I
I
0.8
0.8
1 .o
1.2
1.4
Vitrinite Reflectance, %R, Figure 2. Relationship between maximum rate of evolution (T-) and vitrinite reflectance (%It,) at the nominal heating rate of 25 OC/min for selected samples from the San Juan Basin. Data points were measured directly by the Pyromat I1 micropyrolyzer, and the solid line represents the calculated values from activation energy parameters measured on CH5A, MOAB, KBSB, and CU5A using the multiple-parallel-reaction model. Pyrolysis-FTIR. CHBA, KB5B, and CUBA were analyzed by pyrolysis-FTIR (Advanced Fuel Research, East Hartford, CT; see Solomon et al.lo for description of the technique applied to coals). Pyrolysis-Gas Chromatography. CH5A, MOAA, KB5A, and CUBA were examined by pyrolysis-GC at Norsk Hydro Research Center, Bergen, Norway. The samples were first extracted for 4 h with 92% CHzC12/8% MeOH in a Soxtec apparatus (Tecator, Herndon, VA) to remove native bitumen. Soxtec extraction differs from Soxhlet extraction in that the sample is boiled in the solvent for the first hour and then rinsed for 3 h to achieve equivalent extraction. The samples were pyrolyzed at an average heating rate of 25 OC/min, although the heating rate was not very constant. CZ+products were captured at the front of the capillary column by cooling the fiist 30 cm of the column with liquid nitrogen.
T- and Yield Data Relationship between T- and Vitrinite Reflectance. Figure 2 shows the San Juan Basin coal and coaly shale Tmaxvalues measured a t the nominal heating rate of 25 OC/min as a function of maturity (measured by vitrinite reflectance). These are true temperatures; 35 OC must be subtracted to compare to Rock-Eval Tmax values.s Values shown are for samples from the Fruitland and other seams. Generally, two determinations were made for each sample, and both determinations were in good agreement, except for CU2A. The evolution behavior of this sample was particularly susceptible to sample size.ll To obtain reasonably reproducible data, 10 determinations were made. This sample has a very low TOC (see Table 11), and mineral matter contributions, inhomogeneity, or migrated bitumen could be causes for variations in the pyrolysis behavior. l2 The relationship between T,, and %R, is roughly linear, with increasing T,, with increasing maturity. The figure shows deviation from the linearity for some of the (9) Reynolds, J. G. Lawrence Livermore National Laboratory Report UCRL-ID-109214:Dec. 9.1991. (10) Solomon, P. R.; Serio, M. A.; Carangelo, R. M.; B~ilakia,R.; Gravel, D.; Baillargeon, M.;Baudie, F.; Vail, G. Energy Fuels, 1990,4,
319-333.
(11)Reynolds, J. G.; Murray, A. M. Lawrence Livermore National Laboratory Report UCRL-ID-106605;January 1991. (12) Dembicki, H., Jr.; Horsfield, B.; Ho, T. T. Y. AAPG Bull. 1983,
67,1094-1103.
Energy & Fuels, Vol. 7, No. 5, 1993 613
Coals from the San Juan Basin 5004
Table 111. Summary of Yield Data by Pyromat I1 Micropyrolysis for Selected Coals and Bitumen-Extracted Coals from the San Juan Basin at the Nominal Heating Rate of 25 OC/min
0 0
400
coal
extraction solvent
yield, mg pyrolysate/ g coal
CH5A CH5A
none 92% CHzC12/8% MeOH 100% THF none 92% CHzC12/8% MeOH none 92% CHzC12/8% MeOH none 92% CHzC12/8% MeOH
131 104 108 191 175 192 182 91 59
CH5A
L
0
1 0 0 4
0 8
1 2
1.4
0 8
Vitrinite Reflectance, %R, Figure 3. Rektionship between pyrolysate yield (mg/g TOC) and vitrinite reflectance (%I?,) at the nominal heating rate of 25 OC/min for selected samples from the San Juan Basin. Data pointa were measured directly by the Pyromat I1 micropyrolyzer using AP22 oil shale as the yield standard (88 mg pyrolysate/g shale), and the solid line represents the calculated values from activation energyparametersmeasured on CHSA, MOAB,KBSB, and CU5A using the multiple-parallel-reactionmodel.
lower maturity samples, but except for GR3A, if the Menefee and Hog Mountain samples are not included, the relationship becomes very linear. Because of its apparent outlying behavior, extra determinations were made for GR3A, and the results were found to be comparable to the results in Figure 2. Michael et al.13 have also found a linear relationship between 3'% R, and T- for Rock-Eval measurements on coals from the San Juan Basin. Relationship between Yield and Vitrinite Reflectance. The yields were also measured at the nominal heating rate of 25 OC/min for all samples, and these results are shown in Figure 3. Because the Pyromat I1 has no direct method of measuring yield, these values were measured by comparing the pyrolysis yield to that of an AP22 standard (see Experimental Section). For most of the data shown, the calibration factor from the AP22 standard was an overall average. For selected samples, the yields were checked more times, particularly if the yield did not follow the trend in Figure 3 (HAFA, for example). These replicate yields were measured by the modified method discussed below. The yield values shown exhibit some scatter, particularly for the low-% Rmsamples,althoughreproducibility is good for a given sample. Excluding GR3B, the yields appear to be fairly constant until a % R, of 0.9, where they start to decrease. GR3B does not really follow this trend, nor does CUZA. We had significant problems with every aspect of CUBA and have little confidence in the reproducibility (see above). The trend of decreasing yield above a %Rm of 0.9 has been seen for the Fruitland seam previ0us1y.l~ GC fingerprints of bitumen extracts of selected coals from the San Juan Basin13J4and pyrolysis-GC profiles of these coals (see below) show a definite change in hydrocarbon constitution also occurring around a % R, of 0.9. Effect of Bitumen Extraction on Pyrolysis Yields. Four Fruitland coals CH5A, MOAB,KB5B, and CUBA were extracted with organic solvents to remove native bitumen. Table I11 shows the pyrolysis yields of the (13) Michael, G. E.; Anders, D. E.; Law, B. E. Org. Geochem. 1993,20, 476-498. (14) Clayton, J. L.; Rice, D. D.; Michael, G. E.; Org. Geochem. 1991, 17,735-742.
MOAB MOAB KB5B KB5B CU5A CU5A
% of unextracted pyrolysis yield 100 79 82
100 92 100 95 100 65
extracted coals and compares them with the yields determined for the corresponding unextracted coal. Consistent with Figure 3 and with the yields determined for corresponding unextracted coal, the pyrolysis yields of the extracted coals increase with increasing %R, for samples CH5A, MOAB, and KB5B, and decrease for % Rm > 0.9 as shown by sample CU5A. In all cases, the extraction reduces the pyrolysis yield, and the magnitude depends upon the coal and the extraction solvent. Because we had no previous experience with these samples, we selected 92% CHzC12/8% MeOH as the extraction solvent because of (1) its effectiveness in extracting shales, (2) the reduced likelihood of damaging the coal structure by swelling, and (3) the minimal irreversible binding. The extraction yields using this solvent calculated from the weight of extracted bitumen were as follows: CH5A, 4.4 5% ;MOAB, 4.1 5% ;KBBB, 4.8 7% ; and CU5A, 4.2%. However, these yields seem relatively low compared to extracted bitumen yields for other coals, so THF was selected as an alternate solvent. This solvent has a higher solubility parameteP and could possibly extract more bitumen. The yield for CH5A using 100% T H F as the extraction solvent was 6.9%, which was not enough higher to motivate reextraction of the samples. Solvents such as pyridine were ruled out because of their destructive interactions with the coal structure and the noted irreversible binding.16 The native bitumen extraction yields are essentially constant for all four Fruitland samples, while the reduction in pyrolysis yield due to extraction decreases with increasing maturity for CHBA, MOAB, and KB5B. In contrast, CU5A stands out having the largest reduction in pyrolysis yield. In addition, all the extractions of CU5A (four) showed a mass balance over loo%, unlike the other samples, indicating solvent was incorporated into the coal structure. Again, this behavior is consistent with changes in other properties occurring around a % Rm of 0.9. Change in Pyrolysis Products with Maturity. The pyrolysate chromatograms in Figure 4 show that the Fruitland coals are rich in branched and normal hydrocarbon potential, as might be expected from their relatively high pyrolysis yields. The normal hydrocarbon potential remains mostly intact up through % Rm = 0.82. Even so, there is a marked decrease between ?6R, = 0.44 and 0.76 in branched and cyclic biomarker hydrocarbons (e.g., pristenes between C1, and C ~and B polycyclics around C30) and possibly in some aromatic hydrocarbons. At the (15) Herbandson, H. F.; Neufeld, F. R. J. Org. Chem. 1966,31,1140-
1143.
(16) Whitehurst,D. D.; Mitchell,T. O.;Farcasiu,M. Coal Liquefaction; Academic Press: New York, 1980.
Reynolds and Burnham
614 Energy &Fuels, Vol. 7, No. 5, 1993 CH5A %Rm I 0.44
water, CO, and COz (not shown) potential is eliminated between 7% Rm 0.44 and 0.93. AS a result, 7% Rm = 0.9 might be considered as the transition from primarily eliminating oxygen from the structure to generation of oil-like products. For 7% R m = 1.3,CHr is the dominant species evolved, opposite to the lower maturity samples where tar and oxygen components are the dominant species evolved. Therefore, gas represents a much larger portion of the remaining potential at 7%Rm = 1.3. Kinetics of Evolution of Organic Materials by Pyrolysis
-
0.76
-
0.85
MOAA %Rm
KB5A %Rm
0 U
20
30
40
50
60
CU5A %Rm = 1.30
L--L--L--&.--4_iLL---i+o 20 30
50
60
Retention time (min)
Figure 4. Pyrolysis-gas chromatograms for selected extracted coals from the San Juan Basin of varying maturity. Although most biomarkera are generated at lower maturity, most normal oil potential is retained up to about %R, = 0.9.
highest maturity, there is still some normal hydrocarbon potential, but the major pyrolysis products are benzenes and naphthalenes. This dramatic change in evolved hydrocarbon composition at the highest maturation suggests a radical change in coal structure between 7%Rm = 0.82 and 1.30,which is consistent with the pyrolysis-FTIR results. FTIR evolution profiles for tar and three important gases are summarized in Figure 5. As in pyrolysisGC, the highest maturity sample is notably lower in tar yield than the other samples. In contrast, most of the
Fruitland Coals. Four Fruitland coals, CH5A, MOAB, KBBB, and CU5A, were examined to determine their kinetic parameters for hydrocarbon evolution. These coals were picked because their range in 7% R m covers from the least to the most mature of the samples (see Table I). Figure 6 shows the kinetic parameters determined by the discrete and approximate analyses. The EappmxandAn,,, parameters are given in the insets of the distributions on the left side of the figure. The bar graphs show the Ewe@ distribution, and the A w e b is listed above the residuals of the least squares analyses of the discrete fit to the measured data, En2. The right side of the figure shows the measured rates (circles) from three different heating rates (nominally 1,7,and 50 OC/min) and the calculated rates (solid line) generated by the discrete kinetic analysis of the measured data. The best kinetic parameter sets were chosen from over 50 independent determinations. It was necessary to trim the low-temperature part of the pyrolysis curve, which most likely has contributions due to volatile bitumen, to obtain satisfactory fits. Complete listings of the data sets are reported el~ewhere.~ The best kinetic parameters show some interesting trends. The parameters for the lower rank coals are all very similar, with Eapprox values of 55-57 kcal/mol and and principal Ed-b values of 56-57 kcal/mol. EapPmx principal Ekrebvalues determined for CH5A are slightly higher than those determined for MOAB and KB5B but are within the experimental error of the technique.6 CU5A has the highest 7% Rm of the samples studied and stands out as having distinctly different kinetic parameters, with an Eapprox value of 62.6 kcal/mol and a principal E h h value of 62 kcal/mol. This relationship of the kinetic parameters to 7% Rm appears to qualitatively follow the behavior seen before in the kinetic studies of Argonne premium coals? which show a decrease in activation energy with increasing rank for the lower rank coals and an increase in activation energy with increasing rank for the higher rank coals. The right side of Figure6 shows good agreement between the measured and calculated data for samples CH5A, MOAB, and KB5B, which is verified by the very low residuals. In contrast, sample CU5A showspoor agreement between the measured and the calculated rates, which is also verified by the residuals. This is further discussed below with respect to selection of final kinetic parameters. Extracted Fruitland Coals. To understand the effects of native bitumen on the kinetic parameters of these coals, pyrolysis kinetics were also determined from multipleheating rate experiments for extracted CHBA, KB5B, MOAB, and CU5A. The best kinetic parameters are shown in Figure 7 (which is arranged the same as Figure 6). As in the case for the unextracted coals, the lower rank coals all have very similar kinetic parameters, with Eapprox
Coals from the San Juan Basin
Energy & Fuels, Vol. 7, No. 5, 1993 615
90fn
70-
@. cn
50-
F
30-
d, I
10-104
100
300
500
700
900
Temperature, C
7
7.
u)
@
"I
m 5 -
$? 3-.
. F 0,
3-
F .
a i .
ij
8
1......_
co
.................................
-1 7
and principal Ediscrete values of 55-56 kcal/mol. Eapprox and principal Ediacrete values determined for CH5X are slightly higher than those determined for both MOAX and KB5X, but the values are considered to be within experimental error as determined by the technique. CU5X stands out having an Eappmx of 63.2 kcaVmo1and a principal E k t e of 62 kcal/mol. This behavior parallelsthe behavior of the unextracted coals. The right side of Figure 7 shows good agreement between the measured and calculated rates for the extracted samples CH5X, MOAX, and KB5X. This is verified by the low residuals. Contrary to the unextracted CU5A sample, the CU5X sample also shows good agreement between the rates, also supported by the low residuals. Because the kinetic parameters for sample CU5X are the same as those for the unextracted sample CU5A and because the agreement between the calculated and measured rates is good for sample CU5X, we have confidence in the kinetic parameters for either sample. Comparison of Figures 6 and 7 shows that the kinetic parameters for the unextracted and corresponding extracted coals are within experimental error. CH5X and CH5A values show the most difference. Noting the possibility of compensating A values for the lower activation energy values for CH5X, the discrete kinetic parameters were recalculated for CH5X holding the A value fixed a t 1.74 X 1015. Table IV shows these results. Although the residual sum of squares was not quite as good as for the original CH5X kinetic set (0.130 vs 0.112), the resulting discrete parameters were almost identical to those of the unextracted CH5A coal. The similarity of the parameters for the unextracted coals and the corresponding extracted coals indicates that extraction does not affect the coal structure in a way that significantly influences the kinetic parameters. The biggest differences are seen in the calculated Tmru values at the heating rate of 25 OC/min (samples CH5A, MOAB,
1.......
-1.
'
KB5B, and CU5A-464.5, 489.6, 494.4, and 515.5 "C, respectively;samples CH5X, MOAX,KBBX, and CU5X468.1, 491.8, 494.9, and 514.5 "C, respectively). The extracted coal values are slightly higher than the corresponding unextracted coal values for the lower rank coals and are essentially identical for the higher rank coals. This differenceis probably not significant enough to confidently say that the extraction affects laboratory pyrolysis evolution kinetics or that extraction is necessary in these cases to obtain valid kinetic parameters. For CU5A, however the kinetic determinations were not as easy to interpret. Multiple determinations were performed on the unextracted coal without producing satisfactory kinetic parameters, primarily because the agreement between the measured and calculated data from the discrete method was so poor. The satisfactory kinetic parameters were produced only after examining the extracted CU5A coal, suggestingthe necessity of extraction for at least this coal. The cause of this can be seen in Figure 8 which compares evolution profiles for extracted and unextracted CUBA a t the nominal heating rate of 25 OC/min. The differencebetween the two profiles primarily occurs in the low-temperature range, indicating that the extraction removes soluble material which is volatile in the low-temperature range. This had a significant effect on the agreement between the measured and calculated rates from the discrete kinetic analysis. Also, the values of u in Figures 6 and 7 show that this extraction affects the peak width of evolving materials in the kerogen pyrolysis range. As stated in the yield section, CH5A was also extracted with THF (sample CH5T). The best kinetic parameter set is shown in Table IV, and this is compared with the best kinetic parameter sets for CH5A and CH5X. The THF extraction had little effect on the values of the kinetic parameters. The Eapprox and the principal E b b values are slightly decreasing from 57 to 55 kcal/mol with the
Reynolds and Burnham
616 Energy & Fuels, Vol. 7, No. 5, 1993
0 8
75
0 6 10
0 4
5
0 2
0 0
0
1 0
20 0 8 15
0 6 10
0 4
s
a
5
5 0 2
3 3
,
I
3
0
--
L 3
c:
I!
U
-
I
0
;a
20
A = 160 x 10" X-?= 0 106
Y O 0
T-
KBSR 0 %
I
, ~ ,= ~ 55 ,6 10 1 1 ) kcal'mol 10" 1 sec
A ~ , ,= 1 19x
1
wc
(I
?I
0
0
z 0 8
= 2 29'6 Of , , , ,E
:5
0 6
10
0 4
5
0 2
0
-
_--
E
II
0 0
I4
72
A = 264 x 10'~ E,?= 1 577
1 .o
i'sec
0.8
70
8
0.6
6
0.4
4
0.2
2 0.0
0 45
50
55
60
65
Energy 0' Actlvat,on. kcal mol
70
300
400 500 Temperature. "C
600
Figure 6. Approximate and discrete kinetic parameters from the best kinetic parameter seta selected for CHBA, MOAB,KB5B, and CUSA coals. The circles are the data pointa and the solid lines the calculated reaction rates at (nominally) 1,7,and 50 OC/min.
extent of bitumen extraction (CHSA, CHSX, CHST). These values, however, are within experimental error of the values derived for the CHZC12/MeOH extraction.
Model Calculations The results from the discrete kinetics analyses of samples CH5A, M O B , KBSB, and CUSA (Figure 6) were used to calculate coal properties as a function of maturity. The solid line in Figure 2 shows the calculated relationship ' " and %Rm. 7 ' " was calculated using a between 7 spreadsheet that mathematically reacted the sample at a heating rate of 3 OC/million years to various temperatures then further reacted the sample at 25 "C/min. %R m was calculated for the same geologicalheating conditions using the EASY %Ro ~preadsheet.1~ When compared to the measurements, the calculated 7'- vs %IRmcurve is similar
at high and low maturity but has notably different curvature and value at mid maturity. The solid line in Figure 3 shows the calculated relationship of the pyrolysis yield to %R,. When compared to the measurements, the model agrees well at lower vitrinite reflectance. The model begins to show a decrease in yield at % R m = 0.8, but the measured decline occurs a t % R m > 0.9. Comparingthe calculated and measured maturity trends in Figures 2 and 3indicates that the pardel-reaction model is not adequate, particularly for predicting -'2 a t lower % R , values. Essentially, the model requires change to occur simultaneously in both 7 '- and yield. However, '- to increasewhile yield remains constant nature allows 7 a t low maturity. ~
(17) Sweeney, J. J. Burnham, A.
K.AAPC Bull. 1990,74,1559-1570.
Energy & Fuels, Vol. 7, No. 5,1993 617
Coals from the San Juan Basin 20
1, , , E
CH5A Extracted A = 6 4 7 X 10'' 1
6
=
55 6 (0 041 kcal'mol
~
15
10
5
0 25 20
,~ ,
MoAB Extracted A = 1 65 x 10" Z,l = 0 093
,,A,
im
(I
7 (OM,kcal'mol = 1 65 X 10'' l
~
c
= 2 34 '10 Of ,,,E ,
15 10 - 5 P CI
w-m o c
c 0
KB5B Extracted
;2 a c
EmOr = 55 2 10 09) kcal'mol = 8 46 x 1 0 ' ~ ilwc O = 2 40 "a Of E-,.
, ,A A = 1 13 X 10'' t W c
15
IC 5
0 u i
14
1;
IC 8
6
10 A = 2 40 x 1 0 ' ~I/WC 1")= 0 344
0 8
--,
0 6
= 63 2 (020) kCallml I ~ , =, 5 52 x 1 0 ' ~ QI C S = 3 16 'lo Of EmOI
04
4
02 2 0
T i l 45 50 55 60 65 Energy of Activation. kcallmoi
0 0 l
-
70
300
400
500
Temperature.
600
"C
Figure 7. Approximate and discrete kinetic parameters from the best kinetic parameter sets selected for extracted CH5A (CH5X), MOAB (MOAX), KB5B (KB5X), and CU5A (CU5X) coals. The circles are the data points and the solid lines the calculated reaction rates at (nominally) 1, 7, and 50 OC/min.
Previous work18 has shown that open-system pyrolysis does not follow the same maturation trend as natural maturation, although high-pressure pyrolysis does. The open-system pyrolysis residues tend to have a higher oxygen content for a given extent of hydrogen elimination. A parallel-reaction model can calibrated to model many aspects of natural and high-pressure pyrolysis, as done by Burnham and Sweeney.lg However, such a model would not be expected to predict product evolution during open pyrolysis, including the substantial increase in open pyrolysis T- during the early stage of hydrocarbon elimination.20 In fact, the difference between open-system (18) Monthioux, M.; Landaie, P.; Monin, J.-C. Org. Ceochem. 1985.8, 275-292.
(19) Burnham, A. K.; Sweeney, J. J. Ceochim. Cosmochim.Acto 1989, 53,2644-2657.
and high-pressure pyrolysis is itself a reflection of the failure of the parallel-reaction model. Insight into the failure of the parallel-reaction model can be found in the pyrolysis-FTIR data in Figure 5. This data and derived conclusionsare similar to those presented earlier for the Argonne premium coals.8 The tar potential remains roughly constant between %R, = 0.44 and 0.93 while -'2 increases significantly, consistent with the trends in Figures 2 and 3. A t the same time, most of the oxygen has been eliminated from the coal structure, as reflected in the water and CO evolution profiles. This elimination of oxygen from the coal structure has caused the remaining hydrocarbon moieties to become more refractory even though most of them are retained in the (20) Burnham,A. K.; Sweeney, J. J. Ceochim.Cosmochim.Acta 1991,
55,643-644.
Reynolds and Burnham
618 Energy & Fuels, Vol. 7, No. 5, 1993
Table IV. Approximate and Discrete Kinetic Parameters from the Best Kinetic Sets Selected for CHSA, CHSX (92% CH&l&W MeOH Extracted CHSA), CHBX with Fixed A (from CHSA Determination), and CHST (THF Extracted CHSA)
approximate E,O kcal/mol approximate A,
CH5A
CH5X
CH5X fixedA
CH5T
56.7 (0.07)
55.9 (0.05)
55.9 (0.05)
55.7 (0.03)
9.99 X 1014 4.83 X 10"
4.83 X 1014 4.06 X 10"
3.58
3.14
11s
approximate u,
3.14
3.05
% ofE % oftotal
discrete E ,
kcal/mol 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
7.72 0.40 3.52 0.04 2.39
&
0.99 1.65 0.13 3.32 3.34 10.56 23.76 15.41 16.20 4.57 4.90 3.51 0.98 2.62 0.87
3.88 0.03
2.38 4.71 0.82 7.97
2.84
1.48 7.27
OC, at 25 OC/min
0
0.24 0.23 0.39 0.15 0.68 0.79 1.02 3.48 4.28 11.59 20.41 17.40 13.97 7.40 3.97 2.79 3.49
0.89 1.73
1.85 0.21
1.14 0.85 0.18 1.95 0.68
0.28 0.66
0.46 0.29 0.42 0.05 1.03 0.38 1.74 3.65 5.26 16.70 21.12 16.08 11.98 4.91 3.41 3.17 2.03 1.36 2.55
2.41 1.42 2.76 1.53 5.92 5.87 12.80 19.60 14.07 16.02
discrete A , l/s T-9
0.49 0.33 0.44
1.74 X 10l6 6.47 x 1014 1.74 x 1016 464.48 468.06 468.06
4.06 x 1014 468.87
error in kcal/mol in parentheses.
_...._-.
-.^e.
I
300
-I
I
500 Temperature, "C 400
I
600
Figure 8. Evolution profiles for CU5A and extracted CU5A coals at the nominal heating rate of 25 Wmin.
structure, which is incompatible with a parallel-reaction model. More important, if open-system laboratory pyrolysis were following the same parallel-reaction pathway as natural maturation, it would not be possible to reduce the amount of CO evolved at high temperature at this level of maturity, because the maturation severity has not been great enough to reduce the amount of methane evolved at a lower temperature. Therefore, it is evident
that high-temperature, open-system pyrolysis results in formation of refractory CO bonds that are not formed in natural maturation. (This is consistent with the Van Krevelan diagram of Monthioux et al.18) Clearly, elimination of oxygen from the coal structure occurs differently in nature than in open pyrolysis, and the difference affects the timing of hydrocarbon generation. As a result, it appears that an approach different from multiple-parallelreaction modeling must be used for calibrating geologic maturation models of high-oxygen kerogens. One possibility is the cross-linking model of Solomon et aLZ1 The discrepancy between our simple model calculation and maturity observations cannot be explained by the timing of hydrocarbon generation proposed by Michael et al.l2 They suggest that the peak in their Rock-Eval production index [Sd(S1+SZ),where S1is the free volatile hydrocarbon yield upon heating to 300 "C and SZis the pyrolysis yield] data at % Rm = 0.85 corresponds to peak generation of hydrocarbons. They also draw a straight line through their hydrogen index (pyrolysis yield, mg/g TOC) vs reflectance data, indicating a continuous decline for %Rm > 0.6, although the scatter is so great at low maturity that a reverse trend would be just as plausible for % Rm< 1.0. An explanation more consistent with both our and their data is that some hydrocarbons, particularly biomarker compounds, are generated between %R , = 0.5 and 0.9 and that the main oil generation reaction commencing at % Rm = 0.9 is not observed in the production index because oil expulsion becomes very efficient at that same maturity.
Conclusions For the San Juan Basin coals in this study: (1) T ,, increases systematically with increasing maturity (as measured by vitrinite reflectance). (2) Total pyrolysate yield increases slightly with increasing maturity until a % Rm of approximately 0.9 is reached. After this, the yield decreases. Most of the normal hydrocarbon potential is released between % Ro = 0.9 and 1.3. For the Fruitland seam coals in this study: (1) Extraction with 92% CHzC12/8% MeOH removes approximately 4% by weight of total sample. This qualitatively agreeswith the reduction of pyrolysis yield upon extraction, except for the highest rank coal, CU5A. (2) Extraction of CH5A with THF gives a higher bitumen yield than extraction with 92% CHzC12/8% MeOH. However, the pyrolysis yield of the CH5A extracted with THF was also higher than the pyrolysis yield of CH5A extracted with 92% CH&12/8% MeOH. This suggests that THF is incorporated into the coal. (3)Kinetic parameters derived for CH5A, MOAB, and KB5B (%Rm = 0.44, 0.82, and 0.93, respectively) were similar. However, those derived for CUBA exhibited much higher activation energies. (4) Kinetic parameters derived for extracted CH5A, MOAB, and KB5B were similar. However, those derived for extracted CU5A exhibited much higher activation energies. (5) Kinetic parameters derived for CHBA, MOAB, and KB5B coals were almost identical to the parameters for the corresponding unextracted coals indicating extraction does little to effect the coal structure and is probably not necessary for determining accurate kinetics parameters for these coals. (6) Rates calculated from the discrete distribution kinetic parameters for CU5A did not agree (21) Solomon,P.R.;Hamblen,D.G.;Serio,M.A.;Yu,Z.-Z.;Charpenay, S. Prepr. Pap-Am. Chem. SOC.,Diu. Fuel Chem. 1 9 9 1 , s (1) 267-300.
Coals from the San Juan Basin well with measured rates. Kinetic calculations using this same method for extracted CUBA produced calculated rates that were in good agreement with the measured rates, which indicates extraction is necessary for this coal for determining reliable kinetic parameters. (7) A parallelreaction kinetic model predicts some features of geologic maturation but misses other important aspects.
Acknowledgment. We thank Edward L. Jones and Ann M. Murray for experimental assistance and Vito
Energy & Fuels, Vol. 7, No. 5, 1993 619 Nuccio of the U.S.Geological survey for the samples and the correspondingvitrinite reflectancedata. We also thank Norsk Hydro personnel for their help with the pyrolysisGC experiments. We also acknowledge Rosemary Bassilakis for conductingthe pyrolysis-FTIR experiments.This work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under contract number W-7405-ENG48. Funding was provided through the Morgantown Energy Technology Center of DOE.