Numerical Simulations of SCR DeNOx System for a 660MW coal-fired power station

.Aimed at the selective catalytic reduction (SCR) DeNOx system of a 660 MW coal-fired power station, which is limited by low denitrification efficiency, large ammonia consumption and over-high ammonia escape rate, numerical simulations were conducted by employing STAR-CCM+ (CFD tool). The simulations results revealed the problems existed in the SCR DeNOx system. Aimed at limitations of the target SCR DeNOx system, factors affecting the denitrification performance of SCR, including the structural parameters and ammonia injected by the ammonia nozzles, were optimized. Under the optimized operational conditions, the denitrification efficiency of the SCR system was enhanced, while the ammonia escape rate was reduced below 3ppm. This study serves as references for optimization and modification of SCR systems.


Introduction
Owing to the "poor in oil and gas, rich in coal" characteristic of China, the coal-fired power generation dominates the domestic power industry [1] . In 2012, the domestic power generation was 4.97 trillion kW•h and the coal-fired power generation was 3.91 trillion kW•h (78.6%) [2] . The dominant role of coal-fired power generation makes significant reductions of NOx by coal-fired furnaces extremely challenging. Nevertheless, optimization methods of coal-fired furnaces reported previously do not meet requirements by current standards [3] . Among various NOx treatment methods, the SCR denitrification technology is one of the most widely applied and most effective methods [4] . The mechanism of SCR denitrification is that reducers (e.g., NH 3 and urea) selectively react with NOx to generate N 2 and H 2 O under prescribed catalysts.
In denitrification processes by the selective catalytic red uction (SCR), the concentration distribution of reducers in S CR reactor and airflow distribution has a direct effect on its denitrification efficiency. Currently, the flow fields of the sy stem were analyzed using computational fluid dynamic (CF D) methods [5][6] .

Equipment
The denitrification of the target coal-fired station was achieved by SCR with double flue, double reactor, and 2+1 catalyst layers (one spare layer). Liquid ammonia was used as reducer and the denitrification rate was no less than 80% (NOx at standard status, 6% oxygen content, dry basis). Each reactor in the denitrification system has a size of 13950mm (L) ×11200 mm (W) ×19080mm (H) (requirements by flue gas under 50%BMCR working conditions). Table 1. summarizes key design parameters of the proposed SCR denitrification system. 3 Numerical simulations of SCR DeNOx system

Establishment of numerical simulations model
A 1:1 model from the economizer outlet to the air preheater inlet was established according to the structure of the SCR reactor. The SCR reactor has a length of 13950mm, width of 11200mm, and height of 19080mm, and it contains two catalyst layers. Deflectors were designed according to the actual arrangement, ammonia injection mixer and AIG were designed at the flue inlet, and rectifier grills were designed on top of the catalyst layer.
Owing to the complicated structure of SCR, which contains various deflectors and rectifier grills, the model was divided into seven areas and the 3D model was divided by polyhedral grid [7] . Meshes at certain locations (e.g., small AIG nozzles) were encrypted to observe flow fields and ammonia injections around nozzles. Boundaries at deflectors and side walls are divided into two layers to enhance the accuracy of numerical simulations.
For numerical simulations of SCR, several assumptions should be proposed. First, the hot flue gas is assumed to be Newtonian fluid and uniformly distributed at the model inlet. Second, the catalyst layer of the reactor is assumed to be porous medium. Third, chemical reactions (ammonia and NOx) are assumed to be negligible as the temperature is below the critical value. The standard k-Epsilon model [8] , which has been widely used for calculations of Reynolds number of high turbulence and fluid flows far from side walls [9] , was selected. The boundary conditions at the flue gas inlet were aligned with those at the velocity inlet and the compositions of flue gas at inlet were designed accordingly. The boundary conditions at AIG were aligned with those at the velocity inlet and the compositions of flue gas at inlet were designed according to actual ammonia/air ratio (see Table 2.).

3.2Numerical simulation results and discussion
As the distributions of ammonia concentration and flue gas velocity at the first catalyst layer inlet have significant effec ts on denitrification by SCR, the first catalyst layer is the fo cused area for study of flow field. Moreover, the set-up cont ains two catalyst layers and distributions of flue gas velocity and ammonia concentration at the second catalyst layer outl et should also be paid attention. Herein, the standard deviati on coefficient (CV) was used to evaluate performance of the SCR system. C V is defined as the percentage of standard de viation over average for velocity/concentration at different c ross sections in the SCR reactor. It can be calculated by Eq (1) and (2): where 错误!未找到引用源。 refers to the standard deviation, 错误!未找到引用源。(2) Fig. 1 shows distributions of flue gas velocity and concentration at the first catalyst layer inlet under BMCR working conditions obtained by numerical simulations. As observed, C V of flue gas velocity at the first catalyst layer inlet was 8.30%, which is lower than the critical value(15%). As shown in Fig. 1 (b), the ammonia concentration field was high at the region close to X-axis. Indeed, C V of ammonia concentration at this location was 14.9%, while the critical C V of ammonia concentration is lower than 10.0%.
(a) flue gas velocity distribution at the first catalyst layer inlet.
(b) ammonia concentration distribution at the first catalyst layer inlet (gas volumetric ratio) .  Fig 1. Flow field at the first catalyst layer inlet in initial design. Fig 1. (a) shows the distribution of flue gas velocity at the first catalyst layer inlet. As observed, the flue gas velocity at the nozzle was up to 4.5m/s, which can be attributed to the angle of the static mixer. Fig 1. (b) shows the distribution of ammonia concentration field at the first catalyst layer inlet. As observed, the ammonia concentration was high at bottom nozzle which can be attributed to the angle of the static mixer. More specifically, the flue gas flow in this area was relatively high, resulting in high ammonia injections and ammonia concentrations in this area as the gas involved was a mixture of ammonia and flue gas.
(a)flue gas velocity distribution at the second catalyst layer outlet.
(b)ammonia concentration distribution at the second catalyst layer outlet. Fig. 2 Flow field at exit of the second catalyst layer in initial design. Fig 2. shows numerical simulations of flue gas velocity distribution and ammonia concentration distribution at the second catalyst layer outlet. As observed, flue gas velocity was uniformly distributed, while slightly high at the top. Meanwhile, ammonia concentration uniformly distributed, although it was relatively high at top left corner and relatively low at bottom left corner.
The distributions of velocity and concentration obtained by the actual operational data of the SCR DeNOx system were consistent with those obtained by numerical simulations, despite slight errors induced by equipment. Owing to this phenomenon, potential problems in the design of SCR system can be predicted by numerical simulations. In other words, non-uniform distributions of flue gas in the catalyst layer can be relieved in design, thus optimizing the performance of the SCR DeNOx system. Additionally, vector diagrams of numerical simulations velocity illustrate flow field distributions, thus avoiding non-uniform flue gas distribution and unexpected wear of the catalyst layer.
It is, therefore, can be concluded that the distributions of concentration and velocity in SCR are affected by factors such as structure and angle of the mixer, location and angles of deflectors, and ammonia injections by each nozzle.Hence, these factors are optimized by numerical simulations to obtain optimized design parameters. In this study, ammonia injections by each nozzle were optimized with the overall ammonia injection staying constant.

Ammonia injection optimization
Before entering AIG, the flue gas distribution was non-unif orm and the flow of ammonia was determined by the flue ga s velocity and the flue gas/ammonia ratio. Therefore, ammo nia injection by each nozzle in AIG was adjusted to achieve uniform mixing of flue gas and ammonia in AIG. With cons tant overall ammonia injection, define the upper and lower n ozzles before AIG as nozzles in Row A and B, respectively; adjust ammonia injections by nozzles in Row A and B succ essively. In order to enhance denitrification efficiency and r educe ammonia consumption and ammonia escape rate, CV of ammonia concentration field at the first catalyst layer inle t of SCR should be minimized and C V of velocity field shou ld be no larger than15%.  Under BMCR working conditions, the ammonia injections by nozzles in Row A and B were adjusted. Fig 3. summarizes the results of numerical simulations. As observed, denitrification by SCR is optimized by adjusting ammonia injections by nozzles in Row A and Row B to 55% and 145% of the initial values, respectively.
Under BMCR working conditions of ammonia injection optimization, the ammonia concentration distribution at the first catalyst layer inlet was uniform (see Fig 4.). Indeed, CV of ammonia concentration distribution was reduced from 14.9% to 8.7% and CV of flue gas velocity was 8.2%, which was slightly enhanced.

Hot tests
With constant overall ammonia injection, optimizations of a mmonia injections by nozzles in Row A and B were adjuste d successively. Table 3. summarizes results of hot tests in th is thermal power set under conventional working conditions. To verify the ammonia injection optimization, under typical working conditions, NOx concentration distributions at Outlet A before (see Table 4) and after ammonia injection optimization (see Table 5) were obtained. Under working conditions 1 (before ammonia injection optimization), the average NOX at Outlet A of the SCR system was 87.7mg/Nm3 (deviation = 16.86%), resulting in large ammonia consumption, high ammonia escape rate, and low denitrification efficiency. Under working conditions 2 (after ammonia injection optimization), the average NOx at Outlet A of the SCR system was 83.95mg/Nm3 (deviation = 9.89%) and concentration distributions (<10%) meet requirements. Indeed, the ammonia consumption was reduced from 137.5kg/h to 135.9kg/h, the ammonia escape rate was reduced from 3.92ppm to 2.93ppm, and the denitrification efficiency was enhanced from 80.4% to 81.58%.

Conclusions
In this study, flue gas velocity field and concentration field distributions of SCR DeNOx system for a 660MW coal-fire d power station were simulated ,the results of the proposed SCR system before and after optimization were compared w ith each other. The following conclusions can be obtained: (1) Field distributions obtained by numerical simulations are consistent with the actual operational data, indicating th at numerical simulations can predict distributions of flue gas in the catalyst layer for design of SCR systems, thus facilita ting optimization of denitrification by SCR systems by elim inating non-uniform flue gas distribution and unexpected w ear.
(2) With a constant overall ammonia injection, denitrific ation by SCR is optimized by adjusting ammonia injections by nozzles in Row A and Row B to 55% and 145% of the in itial values, respectively. As a result, Cv of concentration di stribution was reduced from 14.9% to 8.7% and that of flue gas velocity was 8.2%.
(3) Hot tests indicated that the mixing of ammonia and flue gas is affected by the ammonia injection ratio. With a constant overall ammonia injection, optimization can be further enhanced by adjusting ammonia injection by each nozzle and the denitrification efficiency of the SCR system is improved. Indeed, the ammonia escape rate was reduced below 3ppm and the denitrification ammonia consumption was reduced. This study provides references for optimization of SCR systems.