US EPA

9418182 

Journal Article 

Diagnosis analysis of a small-scale incinerator by neural networks model 

Chen, JC; Chen, WeiH; Chang, NiBin 

2008 

Yes 

Civil Engineering and Environmental Systems
ISSN: 1028-6608
EISSN: 1029-0249 

25 

3 

201-213 

10.1080/10286600701838683 

WOS:000258362000002 

The formation of PCDD/Fs (dioxins/furans) due to incomplete combustion in waste combustion process caused tremendous public concern in the last two decades. Since then, more stringent standards for combustion and emission control have been implemented for mitigating such impacts. This change resulted in shutting down many small-scale incinerators because of costly expense in retrofit for meeting higher emission standards. Yet there is still an acute need for building small-scale incinerators for the purpose of disease control, environmental sanitation, and budget saving in rural areas and remote communities. For this reason, improving the diagnosis skill that leads to identifying the optimal management strategy of those small-scale incinerators is still worthwhile to pursue. Using a neural networks model in conjunction with Garson index in this paper, the generation of a series of element contribution analyses helps to diagnose operating discrepancies and differentiate the most influential parameter tied in with the combustion process. Research findings clearly indicate that injecting the auxiliary fuel via an on/off control unit is not an ideal way, as evidenced by its relatively lower Garson index. The process control of auxiliary fuel injection system should be properly upgraded, in order to improve its handling. The results also show that the amount of waste in batch-charging mode and the lowest temperature in the primary combustion chamber during the last feeding are two critical operating factors. Controlling the charging amount per each feed to be 30 kg is a good choice to mitigate the discrepancies of the combustion process in this type of small-scale incinerator. 

solid waste; incineration; neural networks model; element contribution analysis; artificial intelligence