OCR Text |
Show The starting point for the calculations was the calculation of the exhaust gas enthalpies, starting with the determination of the excess air. The Ostwald plot of Fig. 2 shows that the data lie on an Ostwald line passing through 21 % oxygen at zero CO2, and 14% CO2 at zero oxygen. The fuel analysis supplied for the natural gas was: CO2, 0.9%; N2, 0.4%; CR., 93.4%; C2IL, 3.93%; C3Hs, 0.96%; CJIlO, 0.28%; CSH12, 0.06%; CJf14, 0.07%. The calculated maximum CO2 from this analysis was 12.03% compared with the 14% from the Ostwald plot. The discrepancy was enquired into at the time (1966) and was identified as most probably fluctuations in the longer chain hydrocarbons in the gas as delivered; nevertheless, we note that the value of 14% for the maximum CO2 approximates that for a refinery oil gas from vapor phase cracking. The equivalent hydrocarbon analysis in the form CHn yielded: n = 2.532; and the stoichiometric molar ratio of oxygen to fuel was X. = 1.633 [using XI = 1 + nl4]. These values provided the stoichiometry values used in the subsequent computations. The specific enthalpy of each exhaust mixture was calculated from the composition and the exhaust temperatures (Table 1), and using standard data tables for the specific heats of the components. The exhaust enthalpies, Hg were then calculated from Eq. 4 using the thermal input value, Hr, in Eq. 4 and the calculated specific exhaust enthalpy ratios (hglhr). Wall losses (Hw) were determined by difference, as noted, and compared with the values obtained from the surface temperatures and the related calculations, as indicated- above: the results were found to be in good agreement. This completed creation of the primary data base. 5. Results. The primary data base listed as Table 1 was transformed by calculation of the thermal inputs (firing rates), Hr, and exhaust enthalpies, Hg, to a common stoichiometry using Eq. 12, with 25% excess air selected in this instance. This reduction to a common stoichiometry then permitted test of applicability of the performance equations, notably Eqs. 1, 2 and 9, as we now show. In this tranfonnation, the primary assumption was, as noted earlier, that the wall loss was a function only of output, HI, so that in the recalculations for each run, the originally determined wall loss and output values remained the same. Notably, as will be seen in the presentation of the results following, the 10 |