Liquefaction co-processing of coal shale oil at - Argonne National ...
Liquefaction co-processing of coal shale oil at - Argonne National ...
Liquefaction co-processing of coal shale oil at - Argonne National ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
eaction, the pl<strong>at</strong>form was lowered and the microreactor was <strong>co</strong>oled with a stream<br />
<strong>of</strong> room-temper<strong>at</strong>ure air. The outside <strong>of</strong> the <strong>co</strong>oled microreactor assembly was<br />
cleaned (to remove sand) with <strong>co</strong>mpressed air. Gases were vented, and the micro-<br />
reactor was disassembled.<br />
The reacted suspension was pipetted into a 50-mL volumetric flask <strong>co</strong>ntaining<br />
0.100 gm durene and 0.100 gm fluorenone, which were the internal standards used<br />
for quantit<strong>at</strong>ive capillary gas chrom<strong>at</strong>ography, The remaining residue and<br />
synthetic solvent in the microreactor were removed with tetrahydr<strong>of</strong>uran (THF) and<br />
added to the flask. Aliquots were then analyzed using a 50-meter highly cross-<br />
linked phenylmethylsili<strong>co</strong>ne capillary <strong>co</strong>lumn. Flame ioniz<strong>at</strong>ion was used to detect<br />
individual <strong>co</strong>mponents <strong>of</strong> the tre<strong>at</strong>ed synthetic solvent, although in certain<br />
instances a mass-selective detector was also used to assist in the identific<strong>at</strong>ion<br />
<strong>of</strong> products.<br />
Solvent loss <strong>of</strong> each <strong>co</strong>mponent was determined by calcul<strong>at</strong>ing the difference<br />
between the original amount <strong>of</strong> the <strong>co</strong>mponent and the re<strong>co</strong>vered amount <strong>of</strong> the <strong>co</strong>m-<br />
ponent and its reaction by-products. Solvent th<strong>at</strong> was lost and unac<strong>co</strong>unted for<br />
represented adducted or polymerized m<strong>at</strong>erial. A l l percentages will be discussed<br />
on an absolute basis; therefore, the loss <strong>of</strong> 10% tetralin would represent 10% <strong>of</strong><br />
the synthetic solvent and not 10% <strong>of</strong> the 20% tetralin present in the synthetic<br />
solvent. The solvent losses would be five times larger if they were based on the<br />
weight <strong>of</strong> the <strong>co</strong>al sample, since the so1vent:<strong>co</strong>al r<strong>at</strong>io was 5:1, i.e., loss <strong>of</strong> 3%<br />
<strong>of</strong> the solvent by adduction would represent a 15% addition to the weight <strong>of</strong> the<br />
<strong>co</strong>al.<br />
RESULTS AND DISCUSSION<br />
The effect <strong>of</strong> reaction <strong>co</strong>nditions on solvent loss, and specifically on pref-<br />
erential loss <strong>of</strong> <strong>co</strong>mponents <strong>of</strong> the solvent, was examined <strong>at</strong> three sets <strong>of</strong> reaction<br />
<strong>co</strong>nditions. Experiments were <strong>co</strong>nducted under SCTL <strong>co</strong>nditions (rapid he<strong>at</strong>-up, 3<br />
min <strong>at</strong> 425OC). traditional <strong>co</strong>nditions (gradual he<strong>at</strong>-up, 30 min <strong>at</strong> 425OC), and<br />
severe <strong>co</strong>nditions (gradual he<strong>at</strong>-up, 6 hours <strong>at</strong> 425OC). The advantages in using<br />
this synthetic solvent are th<strong>at</strong> all <strong>co</strong>mponents <strong>of</strong> the solvent are known, its<br />
elemental <strong>co</strong>mposition is similar to a <strong>co</strong>al-derived recycle solvent, and the multi-<br />
<strong>co</strong>mponent solvent simul<strong>at</strong>es a recycle solvent better than a one-<strong>co</strong>mponent solvent<br />
does. The use <strong>of</strong> a synthetic solvent also allows the study <strong>of</strong> individual <strong>co</strong>m-<br />
ponents in a more realistic environment.<br />
The objective was to examine solvent re<strong>co</strong>very with increasing severity <strong>of</strong><br />
<strong>co</strong>al-liquefaction reaction <strong>co</strong>nditions, The SCTL stage <strong>of</strong> an ITSL process involves<br />
rapid he<strong>at</strong>-up, followed by a short residence time <strong>at</strong> reaction temper<strong>at</strong>ure. During<br />
rapid he<strong>at</strong>-up, the r<strong>at</strong>e <strong>of</strong> free-radical production should increase significantly<br />
in a SCTL stage. An increase in free-radical <strong>co</strong>ncentr<strong>at</strong>ion was hypothesized,<br />
since the demand for hydrogen, with increased free-radical production, would<br />
increase and would be less likely s<strong>at</strong>isfied by hydrogen donors and gaseous<br />
hydrogen. With an increase in free-radical <strong>co</strong>ncentr<strong>at</strong>ion, a <strong>co</strong>n<strong>co</strong>mitant increase<br />
in solvent adduction was hypothesized, since free-radical addition, arom<strong>at</strong>ic sub-<br />
stitution, and polymeriz<strong>at</strong>ion reactions involving solvent and <strong>co</strong>al free-radicals<br />
xould be likely.<br />
In <strong>co</strong>mparing results obtained under SCTL <strong>co</strong>nditions (Table 1) with results<br />
obtained from experiments <strong>co</strong>nducted under traditional liquefaction <strong>co</strong>nditions<br />
(gradual he<strong>at</strong>-up to 425OC, 30 min <strong>at</strong> 425OC), solvent loss due to adduction should<br />
also be occurring under traditional <strong>co</strong>nditions. As the severity <strong>of</strong> the lique-<br />
faction <strong>co</strong>nditions is increased, adducted solvent should undergo cleavage (crack-<br />
ing, hydrogenolysis, etc. ) and re-form solvent-like products, resulting in<br />
improved solvent balance. While the original <strong>co</strong>mponents may not be re<strong>co</strong>vered,<br />
products having similar structures and chemical properties should be. For<br />
example, adduction <strong>of</strong> phenanthrene with a <strong>co</strong>al-derived benzylic radical may occur<br />
332