Effect of pressure on the kinetics of kerogen pyrolysis - American

May 19, 1993 - Effect of Pressure on the Kinetics of Kerogen Pyrolysis. Howard Freund,* Jamie A. Clouse, and Glenn A. Otten. Exxon Production Research...
0 downloads 0 Views 1MB Size
1088

Energy & Fuels 1993,7, 1088-1094

Effect of Pressure on the Kinetics of Kerogen Pyrolysis Howard Freund; Jamie A. Clouse, and Glenn A. Otten Exxon Production Research Co., P.O.Box 2189, Houston, Texas 77252 Received May 19,1993. Revised Manuscript Received August 6,199P

Laboratory experiments on three different source rock samples indicate that pressure effects on generation kinetics are measurable but minor. Pressure effects are quantified in terms of an activated volume which is analogous to the use of activation energy to quantify temperature effects. Although the three shales had slightly different activated volumes, a value of 27 cm3/mol is recommended as typical for the activated volume of the transformation of kerogen to products. With this activated volume, a 1380 bar pressure corresponds to about a 7 "C increase in generation temperature. Since generation commonly occurs a t pressures less than 1380 bar, the effects of pressure on generation timing appear minor and well within the range of uncertainty from other causes.

Introduction In hydrocarbon exploration, one needs to know when hydrocarbons were generated and the extent of cracking to gas of these hydrocarbons. Often, estimates must be made where direct measurements on rock samples and liquids are not possible. Usually, kinetic models are then used to help make these estimates and are derived from laboratory kinetics. The laboratory determination of these kinetics is done using the immature shale of interest and subjecting the material to high temperatures to accelerate the conversion. The kinetics are then extrapolated back to the geological conditions believed to exist a t the time of generation. The validity of the extrapolation of these rate constants over a wide range of temperatures (Freundl and Domin@) and pressures (DominCb and Mallinson2b)has recently been studied by looking a t hydrocarbon cracking. Other authors have addressed the issue of pressure. Teichmuller and Teichmullel.3 have suggested that high pressure retards the coalification process based on field measurements of vitrinite reflectance. Also using vitrinite reflectance measurements from the offshore of northwestern Europe, McTavish developed a correlation quantitatively relating pressure to vitrinite refle~tance.~ More recently, Sajgo et al.5 compressed alginite and lignite into disks and determined the effect of pressure on their thermal alteration. They observed significant retardation at 1 kbar equivalent to about 50 "C (i.e., the high-pressure case had to be raised 50 "C to obtain the same extent of conversion). Price and Wenger have performed aqueous pyrolysis experiments on Phosphoria shale and report a similar retardation6 Present addrese: Euon Research& Eng. Co., Rt 22E,Annandale,

NJ 08801.

Abstract published in Advance ACS Abstracts, October 1, 1993. (1) Freund, H. Energy Fuels 1992,6,318-326. (2)(a) Domine, F. Org. Ceochem. 1991,17,619-634. (b) Mallinson, R G.; Braun, R. L.; Westbrook, C. K.; Bumham, A. K. I d . Eng. Chem. Res. 1992,31,37-45. (3)Teichmuller,M.;Teichmuller,R. In Coal and Coal-BearingStrata; Murchison, D., Weatoll, T. S., Eds.; American Elsevier: New York,1968. (4)McTavish, R. A. Nature 1978,271,594-596. ( 5 ) Sajgo, C.; McEvoy, J.; Wolff, G. A.; Horvath, 2. A. Adu. Org. Geochem. 1985,10,331-337. (6)Price, L.C.;Wenger, L. M. Org. Ceochem. 1992,29,141-159.

0887-0624/93/25@7-1088$04.00/0

t

........................................................................................

1

Figure 1. A schematic of the experimental system and sample preparation.

The role of pressure in the generation of hydrocarbons has also been addressed theoretically by Perlovsky and Vinkovetsky' who presented a treatment involving an activated volume to suggest that increased pressure will retard the hydrocarbon generation reactions. In another theoretical paper, Neto8 hypothesized that reactions only occur when reactants are within a critical distance between each other. He predicted a pressure dependence on the kinetics but did not quantify it. Monthioux et aL9 studied the effect of reactor configuration, water, and pressure in the artificial maturation of coal. He found the effect of pressure (aswell as water) to be minimal if the experiments were done in a confined system, with the pressure provided externally. In an effort to elucidate the role of pressure in oil generation, we have done experiments to quantify the effect of pressure on the kinetics of maturation. Our approach was to determine an activated volume from kinetic analysis of residual Rock Eval potential. We have examined three kerogen types: Bakken, a type I1 marine shale, Monterey, a type 11smarine shale, and Green River, a type I lacustrine kerogen. We have also investigated the effect of the amount of water added and the degree of openness of the system. (7)Perlovsky, L.I.; Vinkoveteky,Y. A. Boll. Ceofis. Teor. Appl. 1989, 31.87-89. (8)Neb, C. C. Org. Gemhem. 1991,27,57W584. (9)Monthioux,M.;Landais,P.; Monin,J. Org. Ceochem. 1986,8,!2%292.

0 1993 American Chemical Society

Energy & Fuels, Vol. 7,No. 6,1993 1089

Kinetics of Kerogen Pyrolysis 35 S,(initial)

\

T

\

1

--

75 mg/g 324% 24hS

-0.75

-0.7

'' '

\

-0.65

lob0 1sbo Pressure (bar)

560

9

f

tz

f

'

/ -

'.

2000

-0.6

2500 - -

Figure 2. Sz and conversion as a function of pressure for the Bakken shale.

0.7-

317% ,A

A,'

o'6-

,'

6

.... ............ .......'

,a..'"'308

50

"C

100

P A E s

290 "C

-----1

'a

0

P -1 720 bar A =9.5 E 16 sed' (iixed) E = 60300cal/mole (vary) s 3.73 (iixed)

-0.0

I

I

40

60

I

I

120 Time (hrs)

160

0

Figure 4. High-pressure example of the fit obtained from the KINETICS program for the Bakken shale. Curves are fit to data varying only the activation energy. A and s are determined from the 276 bar run. A calculated curve at slightly higher activation energy has been included for the T = 324 "C case to illustrate the sensitivity to model parameters.

---

276 bar 9.5 E 16 sed' 59156 cal/mole 3.73%

150

or0

I

T i m (hrs)

Figure 3. Determination of the kinetic parameters using the KINETICS program for the Bakken shale. Curves are best fit to the experimental points varying A, E, and s.

Experimental/Analytical Section Several experimental techniques were used in this investigation. Reactor configurations were used which had essentially no vapor space as well as ones which did to probe the effect of openness on the results. Reacted samples were typically extracted with solvent but experiments were done in which the samples were not extracted afterward. Effects of amount of water, degree of openness, and effectiveness of solvent extraction were also examined. The samples to be heated with minimal vapor space were prepared in small gold tubes (-25 mm long x 4 mm 0.d.) welded a t one end before adding sample. Typically, 200 mg of ground shale was loaded into the tubes along with 25 mg of seawater. The tubes were then welded shut under an argon atmosphere. The tubes were inserted into a pressure vessel and placed in a furnace. A thermocouple inserted in the end of the pressure vessel measured the temperature of the sample. It had been previously calibrated against a thermocouple inserted down the inside of the vessel. The vessel was then pressurized with argon causing the gold tube to collapse tightly around ita contents. Figure 1 shows a schematic of the gold tube pressure vessel. For the experimental configuration with a vapor space, a pressure vessel 30 cm X 2.5 cm wide with a 4.8 mm i.d. was used. Ground shale sample was mixed with water to a consistencywhich allowed thin "rolls" of the mixture to be made which could then be inserted into the reactor (a shale/HzO ratio of 3.5). The vessel was pressurized to a value roughly half the final pressure at

temperature with inert gas, either helium or argon. The system was then sealed and placed in a furnace for heating. These experiments were generally run paired, Le., a low and a high pressure bomb were run in the same furnace assuring a similar time-temperature history. Experiments for a given shale were done varying the time, temperature, and pressure to which the sample was subjected. After heating, the samples were removed; in the case of the gold tubes, the tubes were punctured. The shale was removed and typically allowed to air-dry, removing water and light hydrocarbons. The shale was then extracted using room temperature dichloromethane until the extract was visually colorless. Some extractions were done a t the boiling point of dichloromethane to examine the effectivenessof room temperature extraction. After extraction, the shale was then analyzed by Rock-Eval pyrolysis. Conversion was defined based on Rock-Eval Spdata: conversion = 1- (S,(sample)/S,(raw)) The Rock-Eval parameters for the raw (initial, unheated) sample were also done on rock extracted by the above procedure. For some experiments, the samples were not extracted and Rock Eval was done on the unextracted samples. Conversion was still defined as above. In other experiments, the extract itself was analyzed.

Results Data Analysis. The results of a series of runs using Bakken shale with t h e gold tubes at different pressures are shown in Figure 2, plotting SZand conversion versus pressure. The lines are a quadratic best fit to t h e data. In order to quantify the effect of pressure, we needed to put the data on a consistent kinetic basis. A single firstorder kinetic expression does n o t describe t h e complex behavior of most kerogens. However, kerogen can still be represented b y aset of parallel first-order reactions, usually with one A factor and multiple activation energies. I n t h i s work, we chose to use a Gaussian distribution of activation energies to fit the data. This function gives t h e relative proportions of each fraction as well as t h e activation energy. With this functional form, we need to determine three kinetic parameters: A, the Arrhenius preexponential factor; E, t h e center of the distribution; s,

1090 Energy & Fuels, Vol. 7, No. 6,1993

Freund et al.

Table I. Experimental Conditions and Rerults for All Runs run

1 2

3 4 5 6 7 8 9 10 11

12 13 14 15 16 17 18 19 20 21 22

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

t (h)

T ("0

0

25 250 260 260 275 290 290 290 290 290 290 290 290 290 290 290 290 290 290 290 290 290 308 308 308 308 308 308 308 308 308 308 308 308 317 317 324 324 324 324 324 340 340 340 340 340 340 350 360

276 276 276 276 276 276 276 276 276 551 551 551 551 1030 1030 1030 1030 1720 1720 1720 1720 276 276 276 551 551 551 1030 1030 1030 1720 1720 1720 276 276 138 276 551 1030 2070 276 276 276 551 1030 1720 276 276

290 290 290 300 317 320 320 340 365 300 320 320 320 340 365 365

276 276 276 276 276 276 276 276 276 1720 1720 1720 1720 1720 1720 1720

270 270 285 300 300 270 270

138 138 138 138 138 1030 1030

504 96 504 192 24 48 48 96 192 24 96 192 48 24 48 96 192 24 48 96 192 24 48 96 24 96 48 24 48 96 24 48 96 24 48 24 24 24 24 24 6 24 24 24 24 24 6 6 not extracted 48 48 96 192 24 48 96 24 24 192 48 192 96 24 6 24

P (bar)

0

65 66 67 68 69 70 71

48 192 96 6 48 48 192

SZ(mg/g) Bakken cold seal samples 75.2 62.5 66.9 60.8 52.9 60.9 53.9 53.8 44.7 33.6 72.9 49.9 37.8 52.6 74.9 56.6 48.3 39.6 77.1 60.5 54.2 48.8 42.5 32.2 25.4 61.4 27.3 34 39.2 38.7 29 58.1 46.5 33.6 26.5 21.7 14.8 16.3 17.8 25.3 32.4 25 11.55 11.8 8.3 10.9 12.6 14.8 8.8 81.4 73.2 74.4 69.7 52.4 63.3 53.6 43.2 38 18.1 59.8 58.9 43.3 44.5 45.5 40.3 21.4 Monterey Samples 64.8 46.8 29.9 23.5 40.3 15.7 54.3 38.1

conv

HI

comments

0

580 458 480 463 431 451 442 433 391 325 539 419 346 430 554 450 416 357 570 457 436 403 364 314 263 502 284 325 342 349 293 481 398 317 280 234 201 223 240 284 334 264 128 132 107 129 136 175

raw Bakken HgO/shale = 0.125 for samples 1-84

0.169 0.110 0.191 0.297 0.191 0.283 0.285 0.406 0.553 0.031 0.336 0.497 0.301 0.004 0.247 0.358 0.473 -0.025 0.195 0.279 0.351 0.435 0.572 0.662 0.184 0.637 0.548 0.479 0.485 0.614 0.227 0.382 0.553 0.648 0.711 0.772 0.738 0.725 0.610 0.500 0.668 0.846 0.843 0.890 0.855 0.832 0.803 0.883

112

0.101 0.086 0.144 0.356 0.222 0.342 0.469 0.533 0.778 0.265 0.276 0.468 0.453 0.441 0.505 0.737

486 507 473 378 426 390 329 300 151 423 414 326 324 311 306 162

0.278 0.539 0.637 0.378 0.758 0.162 0.412

573 524 431 386 513 329 546 469

raw Bakken samples 49-64 are unextractad

raw Monterey

Kinetics of Kerogen Pyrolysis Table I (Continued) run t (h) 72 73 74 75 76 77 78 79 80 81 82 83 84

85 86 87 88 89

90 91 92 93 94 95 96 97 98 99 100 101 102 103 104

T ("C)

Energy & Fuels, Vol. 7, No. 6,1993 1091 P (bar)

S2 (mg/g) conv HI comments Monterey Samples 96 285 1030 25.9 0.600 406 514 6 300 1030 44.3 0.316 72 300 1030 14.4 0.778 281 Green River (extracted) 74.2 raw Green River 747 24 330 276 14.3 0.807 647 24 330 1380 31.7 0.573 740 112 24 350 276 0.37 0.995 24 350 1380 0.84 0.989 323 Green River (unextracted) 87.8 raw Green River 771 24 330 276 65.2 0.257 670 24 330 1380 70.3 0.199 675 24 350 276 44 0.499 521 628 24 350 1380 57.2 0.349 403 51 350 276 28.2 0.679 51 350 1380 43.5 0.505 519 Miscellaneous Large Capacity Cold Seal, Room Temperature Extraction Bakken 24 317 276 38.2 0.492 362 24 317 1720 44.2 0.412 379 Large Capacity Cold Seal, Reflux Extraction Bakken 0 70 O.OO0 480 raw Bakken 24 317 276 28.9 0.587 309 24 317 1720 36.4 0.480 349 regular cold seal Bakken samples 276 45.1 0.400 384 HzO/shale = 1 24 308 276 44.6 0.407 380 0.125 24 308 0.402 378 0 276 45 24 308 276 40.9 0.456 350 10 24 317 1720 54.9 0.270 428 10 24 317 276 45.6 24 317 0.394 370 open tube 689 44.6 0.407 357 open tube 24 317 open tube 1720 47.4 0.370 391 24 317 40.6 raw Monterey (extracted) bomb expta 275 0.771 319 dead vol = 0.9 cm3 138 9.3 720 0.677 391 1380 13.1 275 720 0.754 307 138 10 dead vol = 3.6 cm3 300 48 1380 14.6 0.640 371 300 48 bomb expta Bakken 0.952 41 extracted 350 138 3.6 48 0.949 39 1380 3.8 extracted 48 350 0.806 138 not extracted 138 15.8 48 350 0.749 168 not extracted 1380 20.4 350 48

a parameter that determines the breadth of the distribution. To do this, data at a pressure of 276 bar (this is our low pressure run) were input to the Lawrence Livermore KINETICS0 program. This program calculates via a least-squares procedure the parameters which best fit the data. Figure 3 illustrates the fit obtained for the Bakken shale. As pressure is increased, there is a retardation in the overall rate. Because we wanted to compare kinetic parameters as a function of pressure, the values for the A factor and the breadth of the distribution, s, were then fixed and the activation energy parameter was determined at other pressures by fitting the data with the KINETICS program. Hence, we obtained different activation energies as a function of pressure. Figure 4 is a plot of the Bakken shale at high pressure showing an example of the fit obtained. Also included in Figure 4 is the sensitivity to a change in activation energy of 200 cal/mol a t a temperature of 324 "C. Because we are constraining the fit by fixing A and s,the overall fit is not quite as good. A total of 48 runs were made with the Bakken shale. The data are tabulated in Table I. To quantify the effect of pressure for a reaction with rate constant k, one typically plots In k versus P , where (10) Braun, R. L.; Burnham, A. K. Kinetics: A Computer Program to Analyze Chemical Reaction Data, Report UCID-21588, Lawrence Livermore National Laboratory: Livermore: CA, Nov 1988.

P i s the system pressure. The slope is defined as VJRT, where R is the universal gas constant, T is the absolute temperature and the parameter V , is the activated volume.ll It is the pressure analog to the Arrhenius activation energy and, for simple or elementary reactions, represents the change in volume of the activated complex relative to the reactants. In more complex systems, as we have, the methodology is still valid but the activated volume is best thought of as a parameter which quantifies the pressure sensitivity of the system. As we mentioned, we have reduced our kinetic data in the form of a distributed Gaussian activation energy. As the pressure is varied, we expect this distribution to shift; Le., the center of the activation energy distribution, E, will vary with pressure. Because this activation energy tends to track In k,we can relate E to the activated volume,

V, E = PV, + E, where E is the center of the activation energy distribution determined from the kinetics program at a given pressure and EO is the center of the distribution at very low pressures. Hence, we plot in Figure 5 the determined activation energy as a function of pressure. The solid line (11) Isaacs, N. S. Liquid Phase High Pressure Chemistry;John Wiley

& Sons: New York, 1981.

Freund et al.

1092 Energy & Fuels, Vol. 7, No. 6, 1993

60200

-

3 59800

j

Ly

59400

59000 Pressure (bar)

Figure 5. A plot of the center of the activation energy distribution,&,as a functionof pressure. The temperatureranged from 250 to 360 OC for the Bakken shale. Table 11. Experimentally Determined Activation Volume (cm*/mol) for Several Kerogens Determined from Extracted as Well as unextracted Samdee ~

~~~

sample V, sample V, Bakken (extracted) 33 f 3 Green River (extracted) 30 f 6 Bakken (unextracted) 13 f 6 Green River (unextracted) 27 f 6 Monterey (extracted) 23 h 6

is a least-squares fit through the Bakken data-the slope yields a value of 33 cm3/mol for the activated volume of the Bakken. With an activated volume determined, it is straightforward to incorporate it into kinetic models. A firstorder rate constant would be expressed as k = A exp[-(Ea + PVa)/RT]. Here, E,, is the laboratory derived activation energy at low pressure and PVa is the correction for the pressure retardation. NonextractedData. Conversion in these experiments is based on the SZ peak in a Rock-Eval analysis after extraction. Hence, material which is very heavy yet soluble in dichloromethane is considered product and removed. This material, during the Rock-Eval analysis would tend to decompose and increase SZif it were not extracted. We did a series of experiments in which no extraction was done on the samples after reaction. In these experiments, we still define conversion as before, based on SZ,but now the reacted material submitted for Rock-Eval contains, in addition to the smaller product molecules,those molecules (or pieces of kerogen) which have been released from the starting material but are very large, soluble molecules. During Rock-Eval, this heavy material will decompose leading to a higher SZand hence lower calculated conversion. In this mode of data reduction, the products are considered the light material released as SI(aswell as any gas already lost during handling). We constrain the optimization to have the same activation energy as the extracted case and determine a new A factor. The A factor for the unextracted runs at low pressure for the Bakken is 7.2 X 10l6s-1 compared to the extracted case of 9.5 X lo1*s-l. Note that the extracted constant is over an order of magnitude larger than the unextracted case. These two methods of defining product/conversion allow a deeper cut into the kerogen since when the conversion based on extracted data is 90-99%, the conversion based on unextracted data will only be in the range of 50-85 % .The kinetics based on extraction data probe the early part of

the reaction of the kerogen, while the unextractad kinetics probe the formation of more oil-like products. The effect of pressure for the unextracted case was examined at one high pressure (P = 1720bar). Using the aforementioned methodology, we obtained an activated volume of 13 cm3/mol for the Bakken shale. Monterey and Green River Data. To examine the dependence on kerogen type, we measured the pressure dependence of a high-sulfur type 11skerogen, a Monterey formation shale. The experiments were run as described earlier although only 14 runs were done and two pressures were used. The KINETICS program determined the following parameters: For P = 138 bar, A = 7.5 X 1016 8-1, E = 54376 cal/mol, and s = 3.46%. At P = 1030 bar, E = 54866 cal/mol. This led to an activated volume of 23 cm3/ mol. We also investigated several runs using the Green River kerogen. For this system, we mixed the kerogen with fine quartz powder to form a 15% mixture of kerogenlquartz. The kinetics for this material (kerogen) had been examined' and found to be very well represented by a single activation energy process, k = 1.0 X 10'3 exp(-51300/RT) s-l. In our analysis, we used this activation energy and determined the A factor by best fitting to the data. For the extracted case, A = 7.9 X 1013 s-l. From the highpressure runs, we calculate a Va of 30 cm3/mol. Using conversions based on unextracted data, we found A to be 7.4 X 1Ol2 s-1 for the low pressure case. Then the high pressure data yield an activation volume of 27 cm3lmol. The activation volume data are summarized in Table I1 including estimated uncertainties in the values. Miscellaneous Runs. Several experiments were run to determine the effect of openness of the system and the effect of water on the results. These data are included in Table I, sample runs 89-96. To examine the effect of openness, we did not weld shut the gold tubes. In addition, no water was added. The shale loadings were the same as in the cold seal experiments but the free volume available for the product gases was about 130 cm3. We did experiments a t 276, 689, and 1720 bar in this open configuration. Although the conversions and hence the rates were different than our gold tube results, little effect of pressure was observed. Similarly, experiments were done to examine the effect of HzO. Except for the runs in which the HzO/shale ratio was quite large (lO:l), little effect of HzO was observed. We also examined the effectiveness of the extraction. Most of our extractions were done at room temperature; compared to extraction under reflux conditions, this does not remove all of the extractable organic material. Runs 85-88 show the effect of a higher temperature extraction (done under reflux conditions). Although a "better" extraction is indeed made, so is the extraction of the raw or initial material. Although slightly different conversions were obtained, similar effects with pressure were observed. The bulk of our experiments has been done in a gold tube, confining pressure system. We did experiments in a different configuration to examine whether the results were configuration dependent. The reactor was described earlier. Pairsof high and low pressure reactors were loaded into a furnace. Results using Monterey shale are indicated in Table I as runs 97-100 and those with Bakken shale listed as runs 101-104. The conversions determined from these experiments are compared with the data from the gold tube runs and are shown in Table 111. The closed

Kinetics of Kerogen Pyrolysis

Energy & Fuels, Vol. 7, No. 6,1993 1093 Table IV. Extract Data of Several Oils

,

A = 9.5 E 16 SBC.l E l =58126cal/mole E2 =59126 "

0.84

s

Monterey

~~

= 3.73%

0.7El

0.60

.

/

/E2

A E = 1000cal/mole

0.5-

c

6 0.4-. 0.30.20.1

0

I

'

1

'

1

'

1

Green River

'

1

'

1

-

Temperature ('C)

Figure 6. Effect of a 1kcal/mol change in activation energy on the kinetics of oil generation from kerogen. Table 111. Results from a Closed System Reactor Compared to Results from the Gold Tube Configuration.

sample and confia conditions 138bar 1380bar Monterey closed sys T = 275 "C,t = 30 days 77 f 5 68 f 5 Monterey gold tube 82 f 3 (calc) 74 f 3 (calc) 75 f 5 Monterey cloeed sys T = 300 O C , t = 48 h 64f5 68 f 3 (calc) Monterey gold tube 75 f 5 Baken closed sys T = 350 O C , t = 48 h 95 f 3 95 f 3 Bakken gold tube 97 f 5 (calc) 92 f 3 (calc) Some of the values (calc) have been determined using the kinetic parametera previouslydetermined for the Monterey or Bakken shale. Alldataareanalyzed baeedonextractedshaleand aregiven inpercent conversion aa defined by Sz. a

system conversions are to be compared with the gold tube results directly below them. Within the accuracy of the data, there is good quantitative agreement in the kinetics between the two configurations at both low and high pressures. Extract Analysis. We have also analyzed the extracts from several experiments. Table IV shows data compiled from the extracts of a Monterey shale run in a closed vessel bomb at low and high pressure compared to an extract from a low pressure (confining) run from a gold tube. Also shown in the table are the data from gold tube extracts from the Green River kerogen. The table includes gross compositional data as well as particular biomarker ratios. Biomarker analyses of the extracts (GC-MS) showed similar results for the two pressure runs. Although all these extracts are rather immature, the kerogen has been significantly altered as evidenced by the conversions in the 7 6 8 0 % range for the Monterey and 98-99% for the Green River (based on extracted data). As far as the composition of the extracts is concerned, the effect of pressure (138 vs 1380 bar) appears to be small.

Discussion The activated volumes obtained in this work range from 13 to 33 cms/mol. We have no explanation for the significant difference in activation volumes between the extracted and nonextracted data for the Bakken shale. Activated volumes are known for simple elementary free radical reactions, e.g., V , for H-atom abstraction reactions range 10-20 cm3/mol. For material as complex as kerogen,

run no. time (days) temperature(OC) pressure (bar at )'2 % conversion % saturates % aromatics % NSO % aaphaltenes CPP Pr/Ph Pr/nC17 Ph/nC18 % C29 20Sb 5% C27C % C2Sd % c29c Dia/Reg. Steraned % C32 2 2 9 % Ts/(Ts+Tm)h % Tricyclicsi MPIj

99 2 300 -138

75.4 2 8 13 78 1.00 1.82 1.78 0.94 44.0 40.9 33.1 26.0 0.02 55.6 11.2 15.7 0.83

100 2 300 -1380 64.0 2 7 14 77 1.01 1.84 1.78 0.94 43.1 39.2 32.7 28.1 0.03 54.9 13.4 15.7 0.79

69 2 300 138 75.8 2

5 11

82 0.97 1.60 1.72 0.94 43.4 42.7 27.7 28.3 0.01 56.0 9.6 18.7 0.72

77 1 350 276 99.5 12 11 19 58 1.29 1.27 1.32 0.89 51.4 15.6 30.9 51.5 0.02 58.0 9.2 33.4 0.84

78 1 350 1380 98.9 11 9 15 65 1.30 1.55 1.78 1.07 50.0 16.8 28.9 52.3 0.02 56.8 8.9 28.1 0.84

a CPI = carbon preference index in range nC24 to nC34. % C29 205 = [C29 20S/(20S + 20R)I steranes. %C27 = C27 steranes aa % of C27 + C28 + C29 regular steranes. % C28 = C28 steranes aa 5'% of C27 + C28 + C29 regular steranes. e % C29 = C29 steranes aa % ofC27 t C28 + C29regular steranes. f Diaateranes/regularsteranes. g %C32 22s = [22S/(22S + 22R)I triterpanes. Ts = C27 Trisnorneohopane; Tm = C27 Trisnorhopane. % tricyclics relative to pentacyclic triterpanes from m / t 191. j MPI = 1.5(3-MP + 2-MP)l (P + 9-MP + 1-MP), MPI = methylphenanthrene index.

the authors are unaware of any measurements of activated volume for decomposition. Typical values for the activation volume in polymerization reactions are in the range -(15-25) cm3/mol.11 Because bonds are forming, the activation volumes are negative, meaning the rates accelerate with increasing pressure. The polymerization reactions, however, are similar mechanistically in that they are free radical reactions involving initiation, propagation, and termination. It is reasonable to expect that the pyrolysis of kerogen (essentially a depolymerization reaction) would have a similar, albeit positive, activation volume because of the similar kinds of reactions occurring. Using an activated volume of 27 cm3/mol,a pressure of 1380 bar is equivalent to an activation energy increase of about 1kcal/mol. If we assume that temperature would not effect the measured activated volume (our measurements were determined around 300 "C), then it is valid to apply the activated volume to the maturation of kerogen a t much lower temperatures. Figure 6 shows the effect a 1 kcal/mol difference between two kerogens would have under maturation conditions. We have taken the heating rate to be 1OC/million years and have used the parameters obtained for the Bakken. A t 50% conversion there is an offset of 7O between the two generation curves. This temperature offset is about the same as the extrapolation uncertainty from the determination of the high temperature rate constants. Hence the role of pressure is small. Note that Figure 2 indicates a significant change in S2 as the pressure increases from 138 to 2070 bar. Although Sa, the amount of pyrolyzed volatile5 expelled in RockEval, increases over a factor of 2, the conversion only drops from around 80% to 58%. From a kinetic perspective, it is the conversion,not S2, which is the important parameter. SZis a measure of what is still left behind unreacted and must be related to the initial or raw value to obtain the conversion.

1094 Energy & Fuels, Vol. 7, No. 6,1993

Table IV indicates there are relatively small differences between the extracts from the low and high pressure experiments. In addition, the extracts from the confining pressure configuration (gold tubes) look similar to those generated in the closed system (hydrous pyrolysis). The A factor of the first-order rate constant for the unextracted low pressure runs of the Green River kerogen is very close to the value determined under nonisothermal conditions in a TGA (A = 7.4 X 1012vs 1 X l O I 3 s-l).l Similar agreement was obtained with the unextracted Bakken data compared to nonisothermal open system pyrolysis data. Although the two techniques measure slightly different products, we believe it lends further support to the validity of the kinetics obtained from the confining pressure configuration as compared to other systems. The results of Sajgo et al.5and Price and Wenger6would suggest a considerable effect of pressure on the kinetics of pyrolysis. The discrepancy between their results and ours is disconcerting. We see a relatively minor effect with pressure whereas they see differences amounting to an increase of more than 5000 cal/mol in the activation energy at 1030 bar compared to 138 bar. Sajgo et al. examined lignite, a type I11 kerogen, and alginite, a type I kerogen, whereas we have investigated types I, 11, and 11s. Price and Wenger examined the Phosphoria, a type IIS shale. We do not believe kerogen type is the cause of the differences. The reactor configuration used by Sajgo was quite different from our experimental apparatus. However, several of our experiments were quite similar in

Freund et al. configuration to the closed system bomb configuration of Price and Wenger. We cannot explain the different results.

Summary We have measured the activated volume for the pyrolysis of three shales based on a large number of experiments (- 100 runs) in different configurations under a range of experimental conditions. We recommend an activated volume of 27 cm3/mol for the pyrolysis/maturation of a typical kerogen. At geological pressures associated with generation, this would lead to an uncertainty of less than 10 "C for the timing of generation. In addition, we have analyzed the extracts of high and low pressure experiments and found only minor differences. Our results support our belief that pressure plays a minor role in the generation of oil from kerogen. We are currently addressing the issues involved in the use of laboratory data when applied to the further cracking of oil to lighter products and gas. Acknowledgment. The authors wish to thank the Saskatchewan Energy and Mines Subsurface Geological Laboratory in Regina, Canada, for supplying us with the Bakken core used in this work. We also gratefully acknowledge the careful work of John Okafor who performed the Rock-Eval measurements. Discussions with Dr. Erik Sandvik were particularly helpful. Analytical contributions from Sherlone Robertson, Hettye Hunter, Phuc Nguyen, and Marlene Schaps were also much appreciated.