Spark ignition of annular non-premixed combustors

• The ignition behaviour of a laboratory-scale multiple-burner annular combustion chamber is investigated experimentally focusing on the lightround mechanism ensuring burner-to-burner flame propagation.
• We study the influence of the bulk velocity, the inter-burner spacing and the overall equivalence ratio on the burner-to-burner propagation.
• The measurements show that the extinction stability limits are much wider than the ignitability limits, but when the inter-burner spacing is reduced, they become closer.
• Side visualisation shows that flame propagation consists in a succession of successful and failed events whereas top visualisation shows that the time taken from one burner to ignite the next one varies.
• With an increase in velocity, the time taken for the overall combustor to ignite did not change significantly. However, decreasing the spacing between burners resulted in an increase in the speed of lightround and decreasing the overall equivalence ratio resulted in a slower burner-to-burner propagation.

The ignition behaviour of a laboratory-scale multiple-burner annular combustion chamber is investigated experimentally in this paper. The work specifically focuses on the lightround mechanism ensuring burner-to-burner flame propagation. The system comprises 12, 15 or 18 bluff-body non-premixed burners, each fitted with a swirler, and an annular combustion chamber. A spark located in the neighbourhood of one injector initiates the combustion. The measurements show that the extinction stability limits are much wider than the ignitability limits, but when the inter-burner spacing is reduced, they become closer. Side visualisation shows that successful flame propagation from one ignited burner to its adjacent non-ignited one is associated with the latter being ignited by its own recirculation zone capturing a flame fragment from downstream. The sequential progression of the ignition front from burner to burner was determined by fast OH* imaging from the top of the chamber. The time taken from one burner to ignite the next one varies during the full propagation sequence. The speed of lightround was, in every case, faster in the direction of the swirl of each injector due to the differential tangential velocity induced between the inner and outer combustion chamber walls. With an increase in velocity, the time taken for the overall combustor to ignite did not change significantly. However, decreasing the spacing between burners resulted in an increase in the speed of lightround and decreasing the overall equivalence ratio resulted in a slower burner-to-burner propagation. The results presented in this paper can be used for validation of numerical models of transient combustion processes.