OCR Text |
Show Fuel N0X domain: Fuel Nitrogen -> HCN (18) HCN ^oxidatlon NO HCN ^>reduc"on N 2 After defining the mechanisms, the following transport equations are solved for predicting NOx dYso^„ dyN0_ d (^ndYso) p____i + pWi_________ pD ^ dXi dxX dxi - T - + P « , - T- C/ 0 x, o x, dY, PD^\ + S„c„ ox, + SN0 (19) 3.0 RESULTS A N D DISCUSSION The transport equations arising from the submodels are discretized by the finite volume method using a hybrid scheme. Model predictions are compared to the reported experimental data by Costa et al. (1996) on emissions from an oil fired industrial furnace. The furnace is about 10.74 m long, 6.7m wide, x 1.9 m high, which is represented by 49,000 hexahedral grid cells in the present simulation as shown in Figure 1. The heavy oil and air feed rates used in this simulation are 827 Kg/hr (at 113 °C) and 10,592 Kg/hr (preheated to 1400 °C), respectively. The composition of heavy oil #6 is 88.4wt. % C, 9.3 wt. % H, 2.0 wt. % S , 0.2 wt. % N, and 0.04wt. % ash with the heating value of 43.96 MJ/Kg. The mean oil droplet size used m this study was 200 jim (range 100-300 fim). The glass melting surface was modeled as a fixed temperature of 973 °C during the first 3.25 m along the length, and 1500 °C on the remaining length of the furnace. The furnace is fired from each port alternatively in 20 minute cycles, but in the present simulation firing from the right port is considered. The predicted temperature and NOx concentrations in glass melting (Ports A and B) and refining zones C) are compared with the measured data of Costa et al. (1996) in Figure 2. It may be seen from the Port-A temperature distribution that the visible flame region is between 0.6 to 1.8 m from the wall, with the flame peak at approximately 1.2m from the wall. The predicted flame temperature is slightly low compared to the measured values, which could be attributed to the constant temperature boundary condition of 973°C applied to the first 3.2 m section of the glass surface in furnace. In reality, it could be varying non-linearly with the distance from 50 °C to 1200°C. Apart from the visible flame zone, the predicted temperatures match very well with the experiments. It may be seen from Figure 2(b) that the predicted N O x in on the firing side wall is generally smaller than the measured N O x concentration, however on the exhaust side port the predicted values are in good agreement with the measured values. The peak flame temperature along with the other exhaust species is compared with the measured values of Costa et al. (1996) in Table 1. It can be seen that the peak flame temperature predicted inside the furnace with the soot model tumed-on and without the soot model differ by 368 °C. It is well known that the N O x concentration approximately doubles for every 70 °C increase in |