Experimental and numerical simulation of catalyst deactivation caused by transient turbulence in gradient �ow �eld

In SCR system, the velocity of flue gas field varies with the load, forming a gradient flow 7 field. The characteristics of gradient flow field have important influence on the physical deactivation 8 of catalyst. Through CFD simulation in this paper, it was found that the relative standard coefficients 9 of flow field with characteristic flow velocity were 10.03%, 12.48% and 14.37% respectively. The 10 uniformity of flow field deteriorated with the increase of flow velocity, and the alternating flow field 11 was more likely to scour, wear and block the catalyst channel, leading to its inactivation. This 12 conclusion is also confirmed by the test data obtained from the measuring points installed in various 13 parts of the system. Through LES simulation, it is found that alternating flow field will generate 14 transient turbulent vortices in the system, and with the increase of velocity, the number and 15 distribution range of transient turbulent vortices increase rapidly. In the low-speed flow field, the 16 flow field at the inlet Angle of the flue is disordered, and the velocity varies from 2.42m /s to 8.14m 17 /s. At the corner of flue gas outlet, the flow velocity also varies between 4.86 m/s and 9.03 m/s, but 18 there is laminar flow near the wall. The transient vortices are triggered by the laminar stripping 19 mechanism near the wall. In high velocity flow field, the number of turbulent vortices increases 20 sharply, especially on the surface of the first layer catalyst, which has a great influence on its activity.


Introduction
Nitrogen oxides (NOx) are one of the main atmospheric pollutants.They are composed of various forms such as NO, NO2, N2O, N2O3, N2O4 and N2O5, NO and NO2 account for more than 90% of NOx [1][2][3][4].More than 90% of NOx comes from the burning of coal, oil and gas, of which 70% comes from the burning of coal.NOx in the atmosphere is one of the main causes of the formation of acid rain, PM10 and PM2.5 fine particulate matter, and photochemical smog.It can be seen that the existence of NOx in the atmosphere not only causes environmental pollution, but also affects human health [5][6][7][8][9].At present, the commonly used denitrification technologies of flue gas mainly include: Selective Non-catalytic reduction (SNCR) [10,11], selective Catalytic Reduction (SCR) [12,13] and SNCR/SCR [14] hybrid technologies.SCR flue gas denitrification technology is a low-temperature (300-400℃) denitrification technology.The main reactor is arranged after the economizer, and the flue gas reacts with the reducing agent when flowing through the reactor with catalyst, NOx is reduced to N2 and H2O, SCR technology has high initial investment, but has high denitration efficiency (up to 90%) [15][16][17][18], so it has become the first choice for flue gas denitration.
However, due to load changes, uneven flow field, catalyst "poisoning", coal type changes and other reasons in operation, the uneven mixing of ammonia gas and flue gas, unreasonable molar ratio of ammonia nitrogen, catalytic efficiency decrease, resulting in SCR denitration efficiency decrease or ammonia escape increase, which not only brings economic losses, but also causes secondary pollution, and even affects the safe operation of associated equipment [19][20][21].In the field of preventing catalyst deactivation many researchers have given considerable attention.Shi Jie [22] used Dymola software to establish the kinetic model of SCR reaction based on Modelica language, and predicted the denitration efficiency, ammonia escape rate and other parameters of SCR system.The model is verified by WD10 diesel SCR system test bench.The effects of temperature, flow rate, NH3/NOx stoichiometric ratio (NSR) and cell density on the denitrification performance of SCR system were analyzed by using the established model.The turbulent flow, multi-component diffusion and chemical reaction of flue gas in SCR system are calculated by CFD numerical simulation [23].
Chen Lin [24] modified the catalyst with Samarium (Sm) on the basis of manganese oxide biochar (BC), and determined the catalytic reduction activity of NH3 for NO (LT NH3-SCR), which proved that chemical modification could increase the activity of catalysts at low temperatures.Commercial V2O5-TiO2 catalysts are prone to alkali metal poisoning in flue gas, Huang Li [25] proposed a simple method to improve the resistance of V2O5-MoO3/TiO2 (V-Mo/Ti) catalyst to Na component by loading P component into the original catalyst.The results show that on the V-Mo-P /Ti catalyst, the Na component reacts preferentially with the P component, and the V2O5 component retains most of the REDOX sites (V=O structure) and acid sites (V-OH structure).Yao Xiaojiang [26] selected MnOx/CeO2-ZrO2 nanorods as the reference catalyst and TiO2 as the modifier to modify the original catalyst to inhibit the formation of N2O, improving the catalytic activity of Mn/CZ-NR catalyst and the wear resistance of H2O+SO2.Wang Mingming[27] studied and compared the deactivation effect of PbCl2, Pb(NO3)2 and PbSO4 on NO low-temperature NH3-SCR catalyst supported by Mn-Ce activated carbon.Finally, the mechanism model of poisoning of different lead salts on Mn-Ce/AC catalyst was proposed.Liu Xuan [28] studied the effect of alkali metal deposition on the denitrification activity of selective catalytic reduction (SCR) catalyst by entrained flow combustion method.In addition, the mechanism of alkali metal deactivation was analyzed through various physicochemical characterization.The results showed that the deactivation mechanism of K and Na was similar.The effect of K on catalyst deactivation is stronger than that of Na.Based on real driving emissions (RDE) tests conducted in urban, suburban, and highway sections in Naples, Italy, Alberto [29] derived a control-oriented model of the SCR system and then used this model to design an alternative urea injection logic that minimizes ammonia skid in the exhaust pipe while minimizing efficiency.Um Hyung Sik [30] conducted numerical simulation of urea decomposition chamber (UDC) in low pressure SCR system.The effects of chamber diameter, shape of entrance and exit chambers on denitrification were simulated.The kinetics of urea decomposition was calculated and verified with the experimental data in the range of 300-450℃.The influence of design parameters on the performance of UDC was comprehensively evaluated.
Based on previous studies, the research on catalyst deactivation in SCR system mainly focuses on chemical "poisoning", the exploration of physical deactivation of catalysts is mostly ignored.Many scholars have also carried out numerical simulation of smoke flow field in SCR system, but they mainly focus on macroscopic description, without involving turbulence and swirl inside the flow field.In other words, there is basically no exploration on physical deactivation of catalysts from the microscopic characteristics inside the flow field.In this paper, Computational fluid dynamics(CFD) technology was used to simulate the working environment and deactivation risk of catalysts in SCR system under gradient flow field, and experimental data were used to verify the simulation results.
On the basis of macroscopic analysis, the Large eddy simulation (LES) model was established to describe the microscopic turbulence in gradient flow field, and the distribution of turbulent vortices and the mechanism of their effect on catalyst deactivation were obtained.The simulation results of LES are verified by measuring the energy spectrum density in SCR system.

Numerical simulation of gradient flow field
In SCR system, the uneven distribution of flue gas flow field and its alternating characteristics with boiler load are the main reasons affecting the physical deactivation of catalyst.Especially, when the volume flow of flue gas in the system is too small, the local gas velocity in the upper catalyst is low and laminar flow.Fly ash particles in the flue gas will settle on the surface and accumulate to a certain extent to forma dense shell eventually, causing the SCR system to be blocked and inactivated.
CFD technology has been widely used in scientific research and engineering practice for its intuitive and comprehensive advantages.Its visualization advantages can display the global simulation results more vividly, which greatly improves the research efficiency and quality.The process of flue gas denitrification in the SCR system is very complicated, involving fluid flow, multi-component transmission, chemical reaction, heat and mass transfer, etc.The CFD technology is used to simulate the SCR system, which is helpful to analyze the uniformity of the distribution of flue gas field in the system.

Research object
This paper takes the SCR system of 600MW coal-fired power station as the research object, and uses SolidWorks to establish geometric model of the system with actual size.The flue gas enters the SCR system from the exit of the economizer, passes through multiple groups of diversion devices, ammonia injection grid and rectification grid successively, mixes with NH3 and enters the catalyst layer for denitrification reaction, and the denitrification flue gas is discharged from the exit of the SCR system.The SCR system uses a zonal control ammonia spray grille.10 manual butterfly valves respectively control 10 I-shaped ammonia spray grille subunits.Each subunit contains 4 ammonia spray ports with a diameter of 90mm.The catalyst is arranged in two layers, with the height of each layer being 1.65m.The main parameters of SCR system are shown in Table 1. Figure 1   The actual SCR system is very complex.In order to facilitate the simulation calculation, the following assumptions are made about the initial conditions: 1.The flue gas at the inlet of the system is a uniform and incompressible fluid.
2. Denitrification reagent NH3 is an incompressible fluid mixed evenly with air.
3. The components with less interference to flue gas field are ignored, and only ammonia injection grid, diversion layer and catalyst layer are retained.
4. The porous medium model was used to simplify the structure of catalyst layer.
5. The effects of chemical reactions and temperature fields in the reactor are ignored for now.

Governing equation
The flow field in the system belongs to three-dimensional turbulent flow field, and the simulation follows the conservation of mass, momentum and energy.The mathematical equation of fluid flow consists of continuity equation and momentum equation, as shown in formula ( 1) and ( 2).
Where: u is the velocity vector, p is for pressure, ρ is density, ʋ is kinematic viscosity, t is time, xi,j is the coordinate direction.S is the lost momentum term.

k-ɛ turbulence model
The turbulence action of flue gas and ammonia in the SCR system is simulated by the standard k-ɛ two-way equation model.The turbulence motion viscosity is calculated as follows: ɛequation: Where: Gk is the effect of velocity gradient on turbulent kinetic energy, Gb is the influence of buoyancy on turbulent kinetic energy, YM is the effect of compressible fluid on dissipation rate.G1g, G2g, G3g are the experience values.

Porous medium model
The research object is a two-layer honeycomb catalyst system.Considering the complexity of the ( ) catalyst layer, it is simplified to a porous medium region in the numerical simulation.When the fluid flows through the porous medium, it causes inertia loss and viscosity loss, which are reflected in the momentum source term in equation (2).Equation ( 6) was used to calculate.
For the homogeneous porous media model, Equation ( 6) can be simplified as: Where: α is permeability, C2 is the inertial drag coefficient.• The velocity of ammonia gas injection was taken as the boundary condition of the inlet, and the velocity was calculated according to the field data.The model considers that the velocity direction of ammonia injection is perpendicular to the entrance section and the distribution is uniform.Drawing on previous experience, ammonia and air are mixed at 18:1. Figure 3 shows the parameters of NH3 at the ammonia injection port after measurement and calculation.In this paper, Gambit software was used to divide the grid of the research object.In addition, the ammonia spray grid adopted unstructured network due to its complex structure, and the other parts adopted structured grid.As shown in Figure 2, the total grid number is about 3.73 million.

Evaluation criteria for uniformity of flow field
The uniformity of flow field can effectively reduce the probability of physical deactivation of catalyst, so it also becomes the evaluation criterion of physical deactivation of catalyst.Flow field uniformity can effectively reduce the probability of physical deactivation of catalyst, so it also becomes the evaluation standard of physical deactivation of catalyst, named relative standard coefficient.For the flue gas field in the SCR system, the relative standard coefficient is defined as the percentage of the standard deviation of the flue gas velocity at each point on a section in the mean value of the flue gas velocity on the section, which is calculated according to Formula ( 8)-( 10): Where: K is the relative standard deviation of velocity, N is the number of detection points on the section, V is the average velocity of the section, V  is the standard deviation of the velocity.

Experimental data processing
The pressure drop measured experimentally is calculated by the following formula： ( ) Where: Ps is the static pressure, Pd is dynamic pressure, g is the acceleration of gravity, Z is the height, subscripts 1 and 2 are imports and exports, respectively.increase of the heat load, and the relative standard deviation of the velocity calculated by the simulation and measurement also increases with the increase of the velocity, which can be considered as a linear relationship within the allowed range of accuracy.However, it also indicates that with the increase of flow velocity, the uniformity of flow field on the outer surface of the topmost catalyst in the SCR system becomes worse, and the probability of catalysis plus physical inactivation also increases rapidly.Under 600MW load, the relative standard deviation of the topmost catalyst surface reaches 14.87%, almost reaching the limit of physical deactivation of the catalyst.

Test results of gradient flow field
Figure 8 and Figure 9 respectively show the variation trend of pressure drop at SCR system inlet and outlet and catalyst inlet and outlet with increasing load.These operating points are measured by pressure transmitters installed throughout the system and calculated.It can be seen that the pressure difference at the inlet and outlet of SCR has an approximate linear relationship with the load, the pressure difference between catalyst inlet and outlet also increases with the increase of load, and the higher the load, the faster the increase rate.This also confirmed the conclusion of the previous numerical simulation, the higher the flow rate, the higher the risk of physical deactivation of the catalyst.For unsteady and incompressible turbulent motion in SCR system, the governing equations of large eddy simulation include continuity equation and momentum conservation equation: Where: u, t, p, ρ and ʋ are the velocity vector, time, pressure, density, and motion viscosity, respectively.The superscript "_" is the filtering of the LES method, and the box filter is used here.
Subscripts i and j are the directions of the coordinate system, and subscript t is turbulence.The right side of equation 12 is the pressure gradient term, the stress term and the momentum source term.The turbulent viscosity in equation 12 needs to be solved by a sublattice model.At present, the most widely used sublattice model is Smagorinsky vortex viscosity model, which can be calculated as follows: Where: s L is the sublattice mixing length, min( , ) is the subgrid filter size, S is the strain rate tensor, S C is the Smagorinsky constant, when 0.1 are obtained in the previous flow simulation.
In order to reveal the 3D turbulent vortex structure predicted by large eddy simulation more clearly, the isosurface of Q factor is introduced in this paper for display.Q factor is calculated by the following formula: Where: Ω is the vorticity tensor.
3.1.2Transient three-dimensional turbulent vortex structure analysis Figure 10 shows the distribution of three-dimensional vortex structure in gradient flow field under different working conditions, Q=600 1/s 2 .It can be seen that in the low-speed flow field, the turbulent vortices are mainly distributed in the upstream of the ammonia spray grid and near the wall of the pipe before the flue gas outlet, and both are located near the corner.When the load increases to 300MW, the distribution of vortex structure is similar, but the number increases obviously.When the load increases to 300MW, the distribution of eddy current structure is basically the same, but the number increases obviously, and a small amount of eddy current begins to appear at the rectifier grid.
As can be seen from Figure 6, in the low-speed flow field, the flow field at the corner of flue gas inlet is disarranged, and the velocity varies from 2.42 m/s to 8.14 m/s.In addition, at the corner of flue gas outlet, although the velocity also varies from 4.86 m/s to 9.03 m/s, it is laminar flow close to the wall.This is because the flue gas flow field tends to be uniform after the rectification action of the rectifier grille and the diversion device inside the SCR system.By comprehensive analysis, it can be considered that the turbulent vortices are triggered by the wall surface in the low velocity flow field.With the increase of the flow velocity of the flow field, the number of turbulent vortices increases sharply, and the turbulent vortices also appear in more and more locations.This indicates that with the increase of the flue gas velocity, the shear force of the flue gas flow increases, and the turbulent vortices can be triggered not only at the corner edge, but also in the shear jet.Turbulent vorticity can promote the uniformity of NOx and NH3 mixing in flue gas and improve the denitration efficiency of SCR system.But at the same time, it will impact the catalyst layer and lead to the grinding loss of catalyst.In 600MW, it can be seen that a large number of turbulent vortices have broken through the rectifier grid and reached the upper surface of the first catalyst, and started to impact the surface of the catalyst, rapidly increasing its probability of physical inactivation.frequencies, which is called turbulence energy spectrum.Therefore, the flow condition of the flow field in the system can be obtained through turbulent energy spectrum analysis, and the authenticity of LES simulation can be verified.Figure 11 shows the spectral densities of velocities at different monitoring points (Power spectrum densities, PSD) in the high velocity flow field.During the calculation, the data were collected once every 0.001s for a total of 30s(50-80s).The frequency ranges from 0.03Hz to 500Hz, and the minimum frequency is limited by the collection time (1/30s = 0.03Hz).The highest frequency is half of the signal sampling rate, i.e. 0.5×1/0.001s= 500 Hz.Based on the energy spectrum analysis of the 6 measuring points, it can be seen that the high energy area in the flue gas flow process is concentrated in the low frequency range of 0.03-10Hz, and the energy spectrum density decreases gradually with the increase of frequency, and there is an accelerated attenuation phenomenon.Pulsation of different frequencies is mainly caused by turbulent vortices of different scales, and the higher the frequency, the faster the energy dissipation of the flue gas.

Conclusions
1.In the SCR system, the flow rate of the flue gas field is positively correlated with the load.The change of the load will make the flue gas field become an alternating flow field, resulting in the physical inactivation of the catalyst module, especially the upper catalyst, such as blockage, wear and erosion.The numerical simulation and experimental results show that the relative standard coefficients of flow field with characteristic flow velocity are 10.03%, 12.48% and 14.37%, respectively, which means that with the increase of flow velocity, the uniformity of flow field becomes worse, and the risk of catalyst deactivation greatly increases.
2. The CFD simulation characterized the alternating characteristics of the smoke flow field and its influence on the catalyst from the macroscopic perspective, but the local characteristics and flow state of the flow field could not be studied from the microscopic perspective.LES simulates the generation and distribution of turbulent vortices in the gradient flow field, reveals the internal mechanism of the uniformity variation of flow field and its role in catalyst deactivation, which is an effective supplement to CFD simulation.
3. Through LES simulation and turbulent energy spectrum test, it is found that turbulent vortices increase with the increase of velocity in gradient flow field.At low speed, the turbulent vortices are mainly distributed at the corner of the inlet of SCR system and triggered by the near-wall effect in laminar flow.In the high velocity flow field, the turbulent vortices are distributed in more parts, especially on the surface of the first layer catalyst, which has a great influence on its activity.
The triggering mechanism of turbulent vortex also changes.As the shear force of flue gas jet increases, the turbulent vortex can be triggered mainly in the shear jet except at the guide plate or corner edge.

Figure captions
is a three-dimensional flow chart of the research object.

• 2 . 3 . 1
Flue gas inletThe composition content of flue gas at SCR inlet varies under different loads, but this paper focuses on the influence of flow field changes on catalyst physical inactivation.Therefore, the change of flue gas composition is ignored, and only the composition of flue gas at rated load is taken as the boundary condition of the system inlet.The composition content of flue gas under 600MW load is summarized in Table2

Fig. 2
Fig. 2 The 3D grid diagram of the subject 2.4.2Grid independence verification

Fig. 3
Fig. 3 The distribution of velocities near the center line along the x direction 2.7.Simulation results of gradient flow field

Fig. 7 Fig. 9
Fig. 7 The relative deviation curve of the upper surface of the first layer catalyst under the gradient flow field

Figure 11 (
a, b, c) shows the strong pulsation at measuring points 1, 2 and 3, which can also verify the results of LES simulation in Figure10(d).These three places are where a large number of turbulent vortices gather.Due to the action of the rectifier grating, the pulsation of the flue gas field slows down at the measuring point 4. The flue gas pulsation at the measuring points 5 and 6 is greatly reduced, and the flue gas energy is also greatly reduced.This is because the energy dissipation of flue gas is serious after passing through two layers of catalysts, and the flow of flue gas becomes gentle and closer to laminar flow state, which is also consistent with the simulation results in Figure10(d).It can be seen that the turbulent vortices in the flow field will not only scour the upper surface of the upper catalyst and cause catalyst inactivation, but also consume the energy of the flue gas, resulting in energy dissipation in the system.LES simulation can visually describe the microscopic flow characteristics in the gradient flow field, providing a visual tool for preventing catalyst deactivation and turbulence control in the flow field.

Fig. 11
Fig. 11 Turbulence spectrum analysis at each measuring point

Table Ш .
Parameters of NH3 at ammonia injection port

Table captions Table I .
The SCR system structure parameters Table II.Inlet flue gas composition content Table III.Parameters of NH3 at ammonia injection port