| OCR Text |
Show demonstrate how the simulations can be used to supersede lab testing or to expedite and support pilot- and full-scale testing. RESULTS Only simulation results are presented here because extensive formulations [1-3] and performance evaluations [6,7] are already available. In the forthcoming FLASHCHAIN predictions, the operating conditions of temperature, heating rate, time, and/or pressure were varied to match those in the experiments, while the samples' ultimate analyses was used to specify all structural and reactivity parameters according to previously-defined regressions. A simulation of each thermal history requires less than 20 seconds on a 1 OO-~, Pentium-based personal microcomputer. The greatest impediment to widespread applications of coal devolatilization models is an attitude, termed "the Coal Problem," that the devolatilization behavior of various coals is far too random to ever be predicted by a tractable mathematical model. As seen, for example, in Fig. 1, there is finn empirical basis for this attitude. The filled data points indicate the measured total (top) and tar (bottom) yields for atmospheric devolatilization at heating rates from several hundred to several thousand degrees per second at temperatures hot enough. to achieve ultimate yields in the available residence time. As explained more fully in ref. 1, these coal samples have very similar compositions and the same nominal high volatile bituminous rank. Yet the weight loss ranges from 40 to 60 wt. % and tar yields range from 20 to 40 wt. %. Figure 1 also demonstrates the model's performance in predicting the sample-tosample variability among these very similar coal samples. The FLASHCHAIN predictions for total and tar yields appear as the open points connected by the solid and dashed lines, respectively. The predictions display an uncanny ability to mimic the sample-to-sample variations in this dataset, and also accurately match the extreme values of both total and tar yields. Predicted weight loss is within 4 wt. % of the observed values in 11 of 15 cases. the same tolerance is satisfied in 10 of 13 tar yields. The sample-to-sample variations are correctly predicted for all of the weight loss values except for the 3 samples with the largest carbon contents. The predicted sample-tosample variations in the tar yields are correct in all but 2 cases. 2 |
