OCR Text |
Show 5.1. Transmission of the mechanical or electric valve square signal to the gas. In this section, the performances of the solenoid electrovalve and the rotary-plug valve are compared. Downstream the valve a pressure sensor was installed to visualize on an oscilloscope the gas pressure versus time. A comparison between the 3 signals (idealized square signal, solenoid and rotary-plug valves) is shown on figure 6. Both the solenoid and the rotary-plug valves gave the expected ratio OfF. The rotary valve signal is less "square" than the electrovalve signal. The performance of the rotary valve could be improved on that point by modifying the shape of the rotary plug. Further experiments will be run to evaluate its potential effect on NOx emissions. 5.2. Solenoid electrovalve pulsation 5.2.1 Natural gas or oxygen pulsation All the NOx results have been expressed on a relative basis as NOx/NOx(ref). The reference NOx(ref) has been defined as the NOx emissions of the equivalent non-pulsated flame. This reference point was systematically checked after each run in order to take into account effects such as variations of air entrainment, N2% in natural gas or other experimental fluctuations. Vhen pulsating only natural gas, the effect of pulsation frequency on NOx emissions was first investigated. It was observed that when the frequency f decreases, NOx concentration decreases. Reductions in NOx emissions of up to 60% of the non-pulsated equivalent were achieved. The NOx versus frequency curve follows an S-type function . At frequencies belov the inflection point, CO begins to form, because of intermitent incomplete combustion between rela tively long-intervaled pulsations . The transient occurrence of CO does not represent a significant drawback of the pulsating combustion technique: in the experimental furnace the flue gas sample has been collected from the bottom of the combus tion chamber in a high temperature zone (1300 0 C) quite close to the end of the flame. This sampling procedure does not provide enough residence time for the gases to allow the eventual mlxlng of the rich and lean portions of the product gases. The CO formed by the very rich sequence of the pulsated flame reaches the sampling point before convertion to CO2 by the excess O2 available from the following lean sequence . In an industrial furnace, this limitation will not occur . Complete combustion (or postcombustion in thi s pulsated case ) will take place withi n · the combustion chamber itself, resulting perhaps in a relatively longer flame . Should change in flam geometry affect the thermal function of the flame , this low frequency value coul d be considered as a practical limit of th pulsating combustion technlque. Because of the larger furnace-to - flame siz ratio in the case f an i ndustrial furnace , it is believed that the impac t of the transien t ccur enc of CO on the heat transfer prof ile wi ll be negligible. Figure 7 shows NOx and CO profiles versus pulsation freq uency. Similar experiments have been run when pulsating oxygen only. The same 12 |