文献类型：期刊文献

英文题名：New methodology for creating probit models to assess the domino effect caused by overpressure

作者：Sun, Dong-Liang[1]; Huang, Guang-Tuan[1]; Jiang, Jun-Cheng[2]; Zhang, Ming-Guang[2]; Wang, Zhi-Rong[2]

机构：[1] State Environ Protection Key Lab of Environmental Risk Assessment and Control on Chemical Process, School of Resources and Environmental Engineering, East China University of Science and Technology, Shanghai 200237, China; [2] Jiangsu Key Laboratory of Urban and Industrial Safety, School of Urban Construction and Safety Engineering, Nanjing University of Technology, Nanjing 210009, China

年份：2013

卷号：34

期号：SUPPL.1

起止页码：265

外文期刊名：Dongbei Daxue Xuebao/Journal of Northeastern University

收录：EI(收录号：20132916509033)

语种：英文

外文关键词：Random errors - Probability distributions - Chemical equipment - Separation

摘要：Chemical process equipment and the damage phenomenon were categorized. Data were collected and unnecessary overpressure values together with the corresponding damage phenomenon in which the domino effect did not appear were deleted. The entire range of damage probability was divided into three categories according to degree of damage, by two segmentation points of random variables, and the model of damage probability following a uniform distribution in each degree of damage was proposed, in order to establish more detailed and realistic probit models. The influence of the two random separation points on the mean square errors was investigated and the average of the errors was regarded as the criterion for integral accuracy of the created models. The results showed that the mean square errors of different equipment and their average decreased according to the smaller separation point, and decreased initially and then increased with the larger separation point. The prediction model with high reliability was obtained by data fitting to calculate the average of the mean square errors, which reached a minimum when the values of both break points were 0.3363. These probit models of the four equipment categories with lower mean square errors and better consistency between predicted and observed data were proposed after correcting the few and unnecessary observed data used in curve regression and oversimplification in the probability arrangement in a previous study. The threshold value for damage of 33.63% was 19, 37, 26 and 31 kPa for atmospheric, pressurized, elongated and small equipment, respectively.