EXPERIMENTAL STUDY OF FLAME COMBUSTION PARAMETERS FOR TRAINING A NEURAL NETWORK MODEL FOR THERMAL POWER EQUIPMENT CONTROL
E.P. Kop’ev, A.V. Kuznetsov, M.A. Tarulin, E.Yu. Shadrin
Kutateladze Institute of Thermophysics, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia
Keywords: burner, liquid hydrocarbon fuel, gas analysis, flame visualization
Abstract
A comprehensive experimental study aimed at generating a training dataset for the development of a neural network model to optimize the control of environmental and energy performance parameters of hydrocarbon fuel flame combustion is presented. Flame characteristics are investigated for liquid fuel combustion using a flare burner with a thermal capacity of 300-1500 kW and mechanical atomization. The experiments examine operating modes at various power levels (400, 700, and 1000 kW) and equivalence ratios (1.06, 1.17, 1.28, and 1.40). Simultaneous measurements of flame temperature, combustion product gas composition analysis, and visual flame monitoring are performed. The dependence of combustion parameters on operating conditions is established: an increase in power and a decrease in the equivalence ratio elevate the flame temperature. Alternatively, dropping airflow reduces CO concentration to <10 mg/m3 but increases NOx emissions to >125 mg/m³. Correlations between the visual flame characteristics (size and luminous intensity) and the equipment operating parameters are identified, which opens prospects for the development of automated machine-vision-based control systems.
|
