Publisher DOI: 10.1016/j.fuproc.2018.03.018
Title: Prediction of ash fusion behavior from coal ash composition for entrained-flow gasification
Language: English
Authors: Sasi, Thulasi 
Mighani, Moein 
Örs, Evrim 
Tawani, Ruchika 
Graebner, Martin 
Keywords: Artificial neural network; Ash fusion behavior; ChemApp; Coal ash compositions; Entrained-flow gasification; FactSage; Thermodynamic calculation
Issue Date: Jul-2018
Source: Fuel Processing Technology 176 : 64-75 (2018-07)
Abstract (english): 
In entrained-flow gasification, solid fuel is brought in contact with oxygen and steam, yielding slagging conditions at temperatures of 1250–1800 °C. The process temperature cannot be chosen freely but is determined by the melting behavior of the coal ash. By blending different coals and fluxing agents, the ash fusion temperature can be lowered allowing operation at a lower reactor temperature and savings in oxygen. Since ash fusion behavior is not measurable online, it can be beneficial to use a quickly measured coal ash composition and estimate the ash fusion behavior instantly. In this work, >300 different coal samples from all over the world were investigated. This includes ash compositions determined from X-ray fluorescence (XRF) analysis and standard ash fusion behavior under reducing and oxidizing conditions. In a systematic approach, the ash components were limited to the most significant ones to optimize calculation time. The software ChemApp was used to calculate thermodynamic equilibrium based on FToxid and FactPS databases. The obtained results involve the temperatures at which 10 to 100% of the ash melt are liquid slag calculated in 10%-pts steps. According to the applied atmosphere, the obtained results have been statistically correlated to the experimentally determined fusion temperatures. In parallel, a neural network approach was tested to accomplish the same task. It was found that the hemispherical temperature correlates best to a liquid slag fraction of 85.0 wt% under reducing and 80.1 wt% under oxidizing conditions. The thermodynamic model is able to predict the hemispherical temperature under reducing conditions for 32% of the data while exclusion criteria defining the validity range have been formulated. The neural network model shows in average a higher accuracy of predicting ash fusion behavior from ash composition covering also temperatures of initial deformation and fluidity and appears as a suitable alternative to the thermodynamic calculation if sufficient data are available (i.e. covering the coal ash composition range of interest).
URI: http://hdl.handle.net/11420/13640
ISSN: 0378-3820
Journal: Fuel processing technology 
Document Type: Article
Appears in Collections:Publications without fulltext

Show full item record

Page view(s)

22
checked on Oct 3, 2022

Google ScholarTM

Check

Add Files to Item

Note about this record

Cite this record

Export

Items in TORE are protected by copyright, with all rights reserved, unless otherwise indicated.