Sensitivity analysis of several operational parameters on gas temperature deviation in a tangential firing boiler

. In this work, an integrated numerical model combining combustion and fluid heating was established for a 660 MW coal-fired boiler. To validate the accuracy of the developed model, related experiment was also conducted, and shows that the relative errors (REs) of several predicted parameters on both flue-gas side and water-steam side of the boiler are less than 7 %, which meets the requirement of analysis. The results indicate that the effect of SOFA tilt angle on gas temperature deviation (GTD) is greatest, and that of SOFA’s proportion follows; while GTD presents a smallest sensitivity to the concentration of O 2 . This is because O 2 concentration mainly affects the average value of the gas temperature, but barely changes its distribution. By contrast, altering SOFA’s tilt angle and its proportion directly varies the shape, location, and area of the high-temperature and low-temperature regions at the entrance profile of the horizontal crossover pass, so these two factors play a more important role in the variation of GTD.


Introduction
During the past decades, to improve the thermal efficiency of the coal-fired power plants, utility boilers have been developed towards a larger capacity and a higher steam parameters. In this context, the fluid inside the tubes can be heated to a supercritical or even ultrasupercritical pressure, and the steam temperature of some boilers can reach as high as 700 ºC [1]. However, once overheating failure occurs among these tubes, not only would it waste the fuel and energy, but also the leaked steam with high momentum and temperature could impact the adjacent tubes, causing a secondary damage. Therefore, it is necessary to guarantee the security of the steam tube system during the process of the electricity generation.
Among all heat exchangers, final superheaters and reheaters receive the greatest attention, because their tube walls attain the highest temperatures in the boiler [2]. It has been suggested that the tube overheating problem of these high-temperature heating surfaces results from an unexpected gas temperature deviation (GTD) in the horizontal crossover pass of the boiler [3][4]. In recent years, several researchers [5][6][7][8] have paid attention to revealing the causes and characteristics of this deviation. Liu [5] found that the residual swirl at the furnace exit plays an important part in generating GTD. By retrofitting the boiler's structure, Tan et al [6] optimized the in-furnace flow field and successfully alleviated such a temperature deviation. In addition, the effects of several boiler's settings such as the opposing tangential angle of the air jets [7] and the vertical burner tilt angle [8] on GTD have also been studied. However, few studies have conducted a sensitivity analysis to compare the degrees and mechanisms that these factors act on GTD.
In this work, a simulation which combines coal combustion and fluid heating was carried out for a 660 MW tangential firing boiler, and related experiment were performed for the model validation. Based on the model, the effects of O2 concentration, separated over-fire air (SOFA) tilt angle, and SOFA's proportion on GTD were further discussed.

Boiler description
As shown in Fig. 1, the boiler considered in the present work is a 660 MW supercritical pulverized coal-fired boiler, which has a furnace equipped with a tangentially fired system. A set of low-NOx concentric firing burners were installed in the furnace, composed of six groups of primary air nozzles, six groups of secondary air nozzles, two groups of CCOFA nozzles and five groups of SOFA nozzles.

Operational condition and measurements
To validate the numerical model of the boiler, an experiment at the full load condition was conducted, and several key parameters were measured, including the concentrations of O 2 and CO after the economizer, the gas temperature at the exit of the final reheater (RH), and the pressure, temperature and mass flow rate at the inlet and outlet of four high-temperature heat exchangers, namely the division superheater (SH), the platen SH, the final RH, and the final SH. SOFA Table. 1 provides the coal properties during the experiment. Apart from that, the value of pulverized-coal finenesses R90 and R200 were also measured as 33.6% and 3.3%, respectively. Thus, the distribution of the coal particles can be estimated, which follows the Rosin-Rammler ruler with a mean diameter of 61 m and a spread parameter of 1.15. The experiment was carried out at the full load of 660 MW. In this case, the total mass flow rate of the pulverized coal was 282 t/h, which was equally distributed to the mills A to F. The air had a total mass flow rate of 2360 t/h, of which 27.4% was the primary air with a temperature of 351 K and 72.6 % was the secondary air with a temperature of 581 K. Five layers of SOFA took up 26 % of the total air and had a same vertical tilt angle of −12°. For the water-steam cycle, the mass flow rates of the SH and the RH were 1955 t/h and 1632 t/h, respectively. At the inlet of the water wall heaters, the division SH, and the final RH, the temperature and pressure of the fluid were separately 596 K/26.9 MPa, 687 K/25.9 MPa, and 733 K/4.4 MPa.

Mathematical models
The mathematical models are presented in Table. 2. These models were applied to describe the complex physical and chemical processes within the furnace, and considered the conjugate heat transfer in the tubes of the water wall heaters, the division SH, the platen SH, the final RH, and the final SH. In the present work, the models related to coal combustion was solved using Fluent 14.5, while the thermal hydraulic model inside the tubes was numerical simulated via MATLAB coding. More detailed descriptions of these models, the modelling parameters in them, and the coupled methods have been provided in our previous work [4].  Table. 3 shows some key operational parameters on both flue-gas side and water-steam side of the boiler. It can be seen that the predicted data from the CFD modelling and the experimental results are close. The relative errors (REs) of all the parameters are less than 7 %, which means the accuracy of the developed model is acceptable and this model can be used for the following analyses.

Causes of gas temperature deviation
As shown in the Fig. 2, the pulverized coal and air were injected into the furnace, combusting and forming a counter-clockwise fire ball, which results in a residual swirling flow at the furnace exit. Thus, on the Crosssection P1, the velocity on the left side is higher than that on the right side. This further affects the distribution of gas temperature. A higher local velocity means the hot flue gas releases less heat in this place, namely at the left part of the upper furnace, the division SH and platen SH absorb less heat. Therefore, on the P1 plane, the gas temperature on the left side is also higher than that on the right side, generating a GTD at the entrance of the horizonal crossover pass.  Fig. 3 shows the gas temperature distributions on plane 1, which is the entrance profile of the final RH and SH. To perform a further quantitative analysis of GTD on this plane, Eq. 1 was defined as following: dev ave,left ave,right

Effects of O2 and SOFA on gas temperature deviation
where T ave,left and T ave,right are the gas average temperature in the left and right parts on the P1 plane, respectively. As shown in Fig. 4, when the concentration of O 2 , the SOFA tilt angle, and the proportion of SOFA increase within their regular operational intervals, the maximal differences of T dev are separately 14 K, 29 K, and 53 K, which indicates that the proportion of SOFA has the greatest influence on GTD, and the effect of SOFA tilt angle follows; while this temperature deviation presents a smallest sensitivity to the O 2 concentration.
It is found from Fig. 3(a) to Fig. 3(c) that, as the boiler operated at different O 2 contents, the temperature distributions on the P1 plane are quite similar, which means that the shape, location, and area of the hightemperature and low-temperature regions show few differences. When adjusting the O 2 concentration from 2 % to 3 % and 3.5 %, the gas average temperature on the left side of P1 separately dropped from 1059 K to 1043 K and 1038 K, and that on the right side separately declined from 1017 K to 1000 K and 981 K. This unbalanced decreases in gas temperature of the two parts cause the variations of GTD. By contrast, as shown from Fig. 3(d) to Fig. 3(i), altering SOFA, namely its tilt angle and proportion, directly changes the distribution of the gas temperature on the P1 plane. This is because, by determining the ratio of the air in the main stream to the air with an opposing tangential angle, SOFA settings can induce a different distribution of momentum within the furnace, so have a greater effect on GTD.

Conclusion
In this work, an integrated numerical model was established for a 660 MW tangential firing boiler, and related experiment was also conducted for the model validation. Based on this model, the effects of the O 2 concentration, the SOFA tilt angle, and the proportion of SOFA on GTD were compared. The main conclusions can be drawn as follows: (1) The effects of SOFA's tilt angle on GTD is the greatest, and that of its proportion follows; while GTD presents a smallest sensitivity to the O 2 concentration.
(2) The way that O 2 concentration affects GTD is by producing an unbalanced temperature variations on both sides of the P1 plane, but rarely changing the distribution of temperature.
(3) Altering SOFA's tilt angle or its proportion directly change the shape, location, and area of the hightemperature and low-temperature regions on the P1 plane, so these two factors show a greater effect on the variations of GTD.
(4) SOFA is an important factor both in changing the value of GTD and reducing NOx emission, so it needs to take a comprehensive account for this when doing combustion optimization.