Optimisation of an old 200 MW coal-fired boiler with urea injection through the use of supervised machine learning algorithms to achieve cleaner power generation

Due to the ever more stringent environmental regulations focused on cleaner power generation, thermal power plants that produce energy by burning fossil fuels are forced to optimise combustion processes or invest in new, more modern combustion plants meeting the environmental regulations. The purpos...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of cleaner production Ročník 290; s. 125200
Hlavní autoři: Strušnik, Dušan, Agrež, Marko, Avsec, Jurij, Golob, Marjan
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Ltd 25.03.2021
Témata:
algorithms
artificial intelligence
Artificial neural network
Boiler
burning
coal
combustion
data collection
energy
environmental law
EU regulations
European Union
evaporation
greenhouse gas emissions
heat
injection
Machine-learning
neural networks
Neuro-fuzzy
nitrogen oxides
operating costs
power generation
power plants
Urea
Boiler
Urea
EU regulations
Artificial neural network
Machine-learning
Neuro-fuzzy
ISSN:0959-6526, 1879-1786
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Due to the ever more stringent environmental regulations focused on cleaner power generation, thermal power plants that produce energy by burning fossil fuels are forced to optimise combustion processes or invest in new, more modern combustion plants meeting the environmental regulations. The purpose of the research is economic-ecological optimisation based on the minimisation of the consumption of reagents and a thermodynamic analysis of the impact of the injection of urea into the combustion chamber, demonstrating the innovative aspect of the study. In order to minimise additional operating costs and ensure the half-hour average nitrogen oxides emission values below 200 mg/m3, we created supervised machine learning algorithms. The supervised machine learning algorithms are applied by using the supervised machine learning methods such as artificial neural network and local linear neuro-fuzzy models. The proposed non-linear models are based on a wide range of real process operational datasets from a combined heat and power system in a thermal power plant. The results show that by controlling urea direct injection the supervised machine learning algorithms significantly minimise the operating costs and ensure, at the same time, that requirements regarding the nitrogen oxides emissions prescribed by the European Union directive are met. Moreover, it is evident from the results of the analysis that 2.297 MW of heat from the firebox is consumed on average over the period analysed due to the evaporation and preheating of the mixture injected into the firebox via urea direct injection. As a result, the coal consumption and greenhouse gas emissions are increased. [Display omitted] •The SMLAs for the prediction of the UDI system consumption using the real process data are applied.•The environmental-economical optimisation and thermal aspect regarding the UDI system operation are analysed.•The UDI system operation has a positive as well as a negative impact on the environment.•The positive impact is reflected in reduced NOx toxic gas emissions.•The negative impacts: higher coal consumption, higher greenhouse gas emissions, increased release of heat into the atmosphere..
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0959-6526
1879-1786
DOI:10.1016/j.jclepro.2020.125200