Clustering is the term used in several research communities to describe methods for grouping of unlabeled data. The objective in clustering is to divide the data set in such a way that the intra-class similarity is high and inter-class similarity is low. It is very helpful to find a system's states distributions, but when the system's scale is large, e.g., considering a distillation column that has crude oil as its feed, if one of desired is finding concentration of most popular hydrocarbons (ethane, propane, octane, etc.) on each tray, it should be solve species continuity for each component. A mixture of 100 hydrocarbons was classified into 10 clusters by different clustering methods, e.g., Fuzzy C-Means, K-medoids, K-Mean, Gustafson-Kessel and the accuracy of this grouping was tested using "Flash Calculation". The thermodynamic and transport properties of the components and the mixture were estimated using the thermodynamic and transport properties of the clusters centers and fuzzy memberships of the clusters. The results were satisfactory and accurate. This is an abstract of a paper presented at the 19th International Congress of Chemical and Process Engineering and 7th European Congress of Chemical Engineering (Prague, Czech Republic 8/28/2010-9/1/2010).
|Original language||English (US)|
|State||Published - Dec 1 2010|
|Event||19th International Congress of Chemical and Process Engineering, CHISA 2010 and 7th European Congress of Chemical Engineering, ECCE-7 - Prague, Czech Republic|
Duration: Aug 28 2010 → Sep 1 2010
|Other||19th International Congress of Chemical and Process Engineering, CHISA 2010 and 7th European Congress of Chemical Engineering, ECCE-7|
|Period||8/28/10 → 9/1/10|