BioMatNet Item: JOULE JOR3-CT97-0170 - Advanced neural network based devolatilisation modelling for higher energy efficiency and lower NOx emissions from biomass fuel systems (BIONET)

[BioMatNet Database - European Commission]

JOULE JOR3-CT97-0170
Advanced neural network based devolatilisation modelling for higher energy efficiency and lower NOx emissions from biomass fuel systems (BIONET)

Contacts

To find similar Items, click on a keyword below:
JOULE/THERMIE Programme : Solid Biofuels : Thermochemical Conversion

	Type of Project	Shared Cost
	Contract No	JOR3-CT97-0170
	EC Contribution	600,000 ECU
	Start Date	01/03/98
	Duration	36 Months

Advanced neural network based devolatilisation modelling for higher energy efficiency and lower NO_x emissions from biomass fuel systems (BIONET)

Objectives

This research work will improve and extend the theoretical models used for the prediction of volatile release from solid fuels in both high and low temperature combustion systems. The improvements, and other database information on fuels, will be used to train a neural network This will be linked to an existing combustion predictive technique to enable the combustion performance of biomass fuels and fuel blends to be calculated more effectively.

Technical Approach

The total gaseous and solid emissions from the combustion of fuels are largely determined in the early stages of combustion when the volatile compounds are released during particle heating. The rate of release is set by a number of complex internal and external interacting factors, such as physical and chemical properties of the fuels and the physical environment. New information on volatile release will be produced in this programme by measurements on particles at laboratory scale using laser-heating techniques to. These give the higher heating rates that are more representative of the physical conditions in either a fluid bed or suspension fired combustor. Measurements will be taken on datum fuels including wood chips, sewage sludge and straw in blends with high and low volatile coals.

A neural network will be trained with inputs from the main chemical and physical data to provide output data on volatile release and kinetic data. This output will be used in combustion calculations and the selected fuels will be also fired in both high and low temperature combustion environments, at an industrial scale using suspension firing and fluid bed combustors. Detailed measurements of gaseous species and temperatures will be made in the flame region and post flame and these will be used to confirm the integrity of the volatile release model.

Expected results and exploitation

The new volatile release model will improve the detailed knowledge of how emissions from biomass fuels and blends with coal behave under the different heat release conditions of suspension firing and fluid bed firing. The neural network will form the basis for a prediction tool to make assessments of change in emissions when fuel quality changes occur. The software will be suitable for use in both combustor design and control schemes.

Contacts

Coordinator

Clifford STEVENSON / ABB Combustion Services Ltd Research and Development Dept. / UNITED KINGDOM

Participant