在我們先前的研究中，將二十種可燃氣體氮氣惰化時之燃燒上限資料，以回歸的方式求得預測模式參數，發現這二十種物質所得的燃燒上限預測模式之值為0.9954，平均誤差除了乙烯(10.9020%)以及二甲醚(16.6708%)外，其餘均在4.2263%以下，然而，燃燒上限為可燃物質在進行程序設計的一個重要參數，因此我們預計在本計畫中考慮參數信賴區間的概念用來估計預測燃燒上限的誤差。另外，1,1-二氟乙烷並未在模式建置中的二十種可燃氣體範圍內，但我們利用預測模式進行估算，發現其在氮氣惰化情況下最大誤差為4.6381%，其預測結果甚至比Kondo等人考慮1,1-二氟乙烷所建置的預測模式更為準確，因此在本計畫中另一個重要的工作項目便是分析可燃氣體以及惰性氣體在何種性質(絕熱火燄溫度、比熱容積、官能基等)相近的情況下可以用同一種預測模式進行預測。
In our earlier work, experimental data for twenty flammable gases are collected and used to determine the unspecified parameter in our theoretical model. All the compounds are unified in a same model, it is found the R2 value is 0.9954 for inert nitrogen and average errors except those for ethylene (10.9020%) and dimethyl ether (16.6708%) smaller than 4.2263%. However, the upper flammability limit (UFL) is an important parameter in the design of processes involving flammable compound, it is valuable to be able to estimate the error in the predicted values. Therefore, we have also presented a method to calculate the confidence limits of the estimated UFL. Besides, experimental data of 1,1- difluoroethane, which is not included in the original semi-empirical model, is used to validate the predictive ability for the proposed unified model, and it is found that the maximum predictive error is 4.6381% for inert nitrogen. As compared to the models suggested by Kondo et al., the proposed models result in much better predictive performance in the case of inert nitrogen. In this study, we want to analysis which kind of flammable gases’ and inert gases’ properties can use the same semi-empirical model for predicating (UFL) for flammable gases diluted with inert nitrogen and with carbon dioxide are developed, respectively.
