mehdi sedighi - chemical engineering department faculty of engineering tarbiat modaress university
jafar towfighi

products distribution of thermal cracking of heavy liquid hydrocarbon feedstock were investigated in a tubular reactor at atmospheric pressure. Central composite design CCD was used as an experimental design. The design variables were coli outlet temperature COT steam ratio and feed flow rate. maximum yields of ethylene were 30.37 wt.% at COT,steam ratio and residence time of 869 C , 1.22 gr/gr and 0.208 sec respectively. maximum yields of propylene was 15.37 wt.% at COT , steam ratio and residence time of 825 C , 0.95gr/gr and 0.147 sec respectively. maximum yield of olefin ethylene +propylene was obtained at 842.5 C, steam ratio of 1.24 gr/gr and residence time of 0.17 sec. in these conditions , the yields of ethylene and propylene are 27.2 and 13.6 weight percent respectively. finally , a three layer perceptron neural network , with back propagation BP training algorithm , was developed for modeling of thermal cracking of heay feedstock. the optimum structure of neural network was determined by a trial and error method and different structures were tried.

thermal cracking, heavy liquid hydrocarbon, modeling , neural networks , performance function.