Kenneth J. Fewel, John H. Conroy
Peerless Mfg. Co.
Dallas
Proper ammonia injection grid (AIG) design reduces original equipment costs and operating expenses of selective catalytic reduction (SCR) NOx-removal systems. AIG design is, however, largely neglected in the engineering design stage. Good AIG design reduces costs in several ways:
- Ammonia- Less will be used in the reduction process.
- Catalyst-Both original purchase (less catalyst is required with good ammonia Distribution) and replacement (good distribution should lead to longer catalyst life).
- Ammonia flow control system - Smaller pipes, smaller blowers, and smaller evaporators.
- Reactor - Can be smaller because less catalyst material is required.
- Pressure loss - Because less catalyst is required with good ammonia distribution, less pressure is lost and horsepower, or throughput, is increased. Field results prove the importance of careful and precise ammonia injection system design.
NOX EMISSIONS
Fossil fuels and other combustible products are consumed in industrial processes, in the generation of electricity for example, or in numerous other ways. During such processes, nitrogen oxides (NOx) - a combination of nitrogen dioxide and nitric oxide - are formed as components of the exhaust gas.
NOx sources have been classified by government regulatory agencies as mobile sources (such as automobiles, trucks, and mobile diesel powered equipment), and stationary sources (including gas turbines, industrial boilers, and refinery heaters). While air chemistry is very complex the emissions of NOx into the atmosphere have proven negative impacts upon the environment-first, by direct exposure, and subsequently by contributing to the formation of acid rain and photochemical oxidant (ozone), and dry acid deposition. Fig. 1 illustrates examples of predictable environmental impacts of NOx emissions.
SCR PROCESS
A number of technologies exist to reduce the NOx emissions generated by these sources. One such technology, selective catalytic reduction, has been successfully applied to stationary combustion sources. SCR can reduce NOx emissions from a single source by as much as 95%.
Although SCR technology has been available since the late 1950s, most industries and companies did not implement the process on a wide scale until it was determined to be the best and most efficient way to reduce NOx emissions to required levels.
Fig. 2 illustrates the major components of an SCR system. The NOx laden exhaust gas passes over the ammonia-injection grid, where ammonia vapor is dispersed. The manifold and external piping transport the ammonia vapor from the ammonia flow control unit (AFCU).
The mixed gas and ammonia vapor then enter the reactor and pass through a catalyst bed, where a chemical reaction occurs. This reaction reduces NOx to harmless nitrogen gas and water vapor. Operation of the entire system is precisely regulated and monitored by the control system.
PROCESS STEPS
- Ammonia evaporation
Anhydrous ammonia (99.5% pure ammonia) and aqueous ammonia (a solution of 25-30% ammonia) normally are stored in liquid form. Aqueous ammonia is much safer to handle, store, and transport than anhydrous ammonia.
Several grades of both anhydrous and aqueous ammonia are available, and selection of the appropriate type and grade is the first step in the design of an SCR system. Aqueous ammonia is evaporated in a special evaporator tower; the mixture of air and ammonia is usually about 5% ammonia and 95% air.
- Ammonia injection
The ammonia mixture is injected into the exhaust gas to mix with the NOx. This must be as even as possible; the concentration of ammonia at the catalyst face cannot vary more than 10% (depending upon the application). This is more challenging when mixing must be done in an extremely short distance.
- Catalytic reaction
As the mixed NOx and ammonia pass through the catalyst, several chemical reactions occur (Table 1). Because the reactions are accelerated by the catalysts selection of an appropriate catalyst material is essential to NOx-reduction efficiency.
If untreated exhaust gas seeps out of the reactor that holds the catalyst, NOx-reduction efficiency is diminished and unreacted pollutants are released into the atmosphere. The reactor, therefore, should be designed and manufactured to be airtight.
- Ammonia slip
A control system uses an NOx sensor in the exhaust gas stream to precisely control the amount of ammonia injected into the stream. This scheme prevents the emission of unreacted ammonia into the atmosphere. An airtight reactor design is critical to the prevention of ammonia slip.
SCR CHEMISTRY
The SCR reaction requires the presence of oxygen, a reducing agent such as ammonia, and a catalyst, to produce the desired result-maximum reduction of NO.. The temperature window for the reduction reactions ranges from 475 to 1,100 F.
The temperature variation is so great because of the many catalyst compositions available for a wide variety of applications. Several of the competing chemical reactions that occur in this environment are shown in Table 1. The first four reactions must be driven to the right and the oxidation rate of ammonia to NOx (shown in the fifth reaction), must be minimized. To achieve this, proper temperature, velocity, and NOx and ammonia concentration profiles must be maintained at the catalyst face as the exhaust gases pass over the SCR catalyst bed.
MANIFOLD, AIG DESIGN
Although these chemical reactions appear simple, real-world SCR systems are quite complex in design. And because thorough mixing of the ammonia and exhaust gases is the most critical element of complete NOx reduction, AIG design is one of the greatest challenges in engineering an SCR system.
Once the ammonia mixture is ready for injection, the object is to inject it as evenly as possible into the duct upstream of the catalyst. This requires a manifold and distributor. This distributor is commonly referred to as the ammonia injection grid. It is usually made from pipe or tubing with perforated orifices. These orifices create jets which inject the ammonia mixture into the free-stream exhaust gas. The pipe or tubing generally is stainless steel because of the elevated temperatures at typical AIG locations.
Good distribution requires careful engineering of the manifold and AIG to ensure evenness of flow to all jets. This is not difficult, given ample supply pressure to the grid and proper placement of the AIG from the catalyst face. Achieving the least pressure loss requires an iterative melding of design and analysis. This, of course, reduces long-term operating costs. The equations governing the even distribution of gases to all parts of an AIG can be derived from equations of momentum and pressure loss in pipes and across orifices. There are two parts of this analysis.
First, the flow distribution to the orifices must not be affected by the length of the distributor pipe. If the pipe is exceedingly long with respect to its diameter (a high L/D ratio), the nearest distributor orifices will inject the most ammonia, starving the downstream orifices because of the pressure drop along the length of the pipes. Second, the momentum of the ammonia within the distributor pipes must be small in comparison to the momentum in the orifices. The higher the velocity in the pipes, the greater the likelihood that the flow will concentrate in the last orifices of the distributor. An elegant pair of equations has been developed by Senecal. The equations derive from a simple pair of rules based on momentum and pressure drop:
- Momentum in the pipe of the AIG, expressed as paV2, must be less than or equal to one tenth of the pressure drop across the average orifice.
- The friction loss in the pipe of the AIG must be less than or equal to one tenth of the pressure drop across the average orifice. Equations 1 and 2 (see Equations), which assume fully turbulent plug flow, express these rules mathematically. The better the distribution of the ammonia in an AIG system, the less expensive the operating costs. Ammonia maldistribution results in wasted ammonia.
Suppose that the ammonia is mal-distributed to 30% over the AIG orifices. Some spots experience 30% less ammonia than necessary to reduce NOx. To increase the flow of ammonia to these dilute regions, the control system must increase the overall flow to the AIG by 30%.
As a result, the catalyst will experience 30% excess ammonia-ammonia which will not be reacted and will pass to the atmosphere. This can result in a loss of overall efficiency and increased ammonia slip (emissions of unreacted ammonia). Ammonia slip is being monitored much more closely today by the Environmental Protection Agency (EPA), and ammonia has been listed as an air toxin.
In addition, the better the ammonia distribution, the less catalyst is required. Additional catalyst means higher costs for original and replacement catalyst, and results in greater pressure loss. Also, it is believed that uneven distribution causes premature depletion of the catalyst.
JET DISPERSION
The jets that emanate from the AIG grid force the ammonia mixture to be mixed into the exhaust stream. This is accomplished using two mechanisms: free turbulence and forced stirring.
Free turbulence occurs because of the turbulence of the exhaust stream and the turbulence generated by the interaction of the AIG distributor pipes with the injected jets (Fig. 3),
Forced stirring is created using an airfoil or blunt body to stir the jets into the exhaust stream. A stationary appendage can accomplish this by using the flowing energy of the free stream. The result is a slight increase in pressure drop but an increased rate of mixing. The most common AIG design depends on free turbulence from the flow and grid to mix the gases. The orientation of the jets is important. Certain patterns, like the one shown in Fig. 4, are optimum for jet dispersion and mixing,
The objective is to create as even a pattern as practical from a given spacing of lances. The design in Fig. 4 creates a square pattern of jet plumes in cross section, just 10-20 orifice diameters downstream of the grid lances. These are optimally positioned for free turbulence mixing before encountering the catalyst.
The trajectories of the jet plumes can be computed using a correlation of the form developed by Rudinger (Equations 3 and 4).
These equations are useful for designing the AIG grid orifice pattern. The orifice pattern can be optimized for a given lance spacing.
The factor K, and the exponents a and b must be determined from experimental tests or computational fluid dynamics (CFD) models of the AIG lances in crossflow. Fig. 5 illustrates a jet trajectory.
MIXING LENGTH
Once the jets are dispersed, turbulent mixing ensues. The transverse jets create their own turbulent vortices which spiral off both sides of the jet.
After the jet plumes have turned 90, they interact with the turbulent wake of the lances. Although seemingly chaotic in nature, the effects of this mixing process can be predicted using proven calculations. Correlations have been developed to estimate the mixing length required to reduce the concentration fluctuations to a desired amount. Breidenthal, et al., developed one such correlation based upon experiments with helium and nitrogen. An aspirating probe was used to measure the concentration fluctuations. This work supports the general theory that large-scale turbulent eddies govern mixing rates.
Mixing rates are a function of flow geometry and jet/free-stream momentum ratios. Surprisingly, these factors create observable eddy patterns which are independent of the Reynolds Number. These eddy patterns are critical to the rate of effective mixing. Molecular diffusion is very rapid in comparison and does not limit the mixing process in the shear layer between the two fluids.
Breidenthal's correlation is useful for determining the size ratio of an AIG to the catalyst mixing chamber. Central to the formulation is his jet momentum ratio, JB (Equation 5).
From experimentation, the correlation given in Equation 6 has been developed to find the concentration fluctuation at any point downstream of the injection grid, The K2 value is a function of the geometry of the injector grid and mixing chamber. Its value can vary greatly.
The significance of this equation is the organization of the important variables that affect the required mixing length. The art of mixing-chamber design can be summed up in two important variables, K2 and J.
The lower the value of K2 the shorter the required mixing length. Good mixing-chamber design results in K, values below 0.4. Conversely, the higher the value of J, the shorter the mixing length. The value of j is dependent on the pressure available to the AIG manifold, and is thus an energy cost to operations.
CFD IN MIXING DESIGN
Computational fluid dynamics can aid in the design of the AIG mixing chamber. The interaction of the injected jets of ammonia mixture and the exhaust-free stream can be simulated using the "finite difference" technique. Comparisons between mixing-chamber geometries can be made to determine the effectiveness of various forced-mixing devices.
CFD uses a gradient-diffusion model which is known to be accurate for mixing in turbulent, shear layers. The effective mixing rate in turbulent flows is estimated using the turbulent viscosity or eddy viscosity. This eddy viscosity is found using a turbulence model of isotropic form.
The turbulent viscosity can be thousands of times higher than molecular viscosity and thus the mass diffusion of the gas is greatly affected. At the shear-layer interface between the jet and the free stream, the turbulence can be quite high, which result in a greatly increased mixing rate. Equation shows the formula for computing mass diffusion.
This mass-diffusion equation yields accurate estimates of time-averaged concentration distribution at every finite volume in the model. Thus, a CFD model can identify regions of uneven concentration at any point in the flow field, including the catalyst face. In addition, a CFD model provides two or three-dimensional velocity and turbulence fields. These are useful for designing an SCR system with good exhaust-flow distribution to the catalyst, which also is important to efficient operation.
CFD RESULTS
Results of a CFD study provide two significant facts about the mixing design:
- Velocity profile
A three-dimensional contour map of gas velocity at the catalyst face will reveal the maximum velocity. The maximum velocity divided by the average velocity yields the maldistribution ratio (which should always be less than 1.1). Fig. 6 illustrates the profiles and contours of velocity magnitude before the catalyst. Good velocity distribution is required to optimize the SCR system design. Maldistribution of velocity leads to premature replacement of the catalyst, poor NOx reduction, and excess ammonia slip. The result shown is from a model of free turbulent mixing for an operating SCR system.
- Ammonia concentration in exhaust gas
A time-averaged ammonia concentration result at the catalyst face is depicted in Fig. 7. The regions of high concentration are darker for clarity. These high concentration regions correspond to the jet positions in the grid. The areas of high concentration propagate downstream in a shadowlike fashion. The flattening illustrated by the profiles results from turbulent mixing. Several reliable CFD benchmark studies have been published regarding turbulent mixing. Peerless Mfg. Co. engineers have performed a benchmark study using experimental measurements by Rathgeber, and the CFD results confirmed the mixing-duct measurements (Fig. 8).
Computational fluid dynamics has emerged as the most credible method outside of physical testing to confirm mixing-systems design.
CASE STUDY
Peerless Mfg. Co. retrofitted a poorly designed ammonia injection grid system using the design criteria in this article. The customer had reported high ammonia consumption and emissions (slip) of both NOx and ammonia that greatly exceeded allowable EPA limits.
A detailed investigation revealed an ineffective ammonia/air mixture, along with an ammonia injection grid that was obviously not distributing ammonia evenly to the catalyst face. A traverse of the duct concluded that the temperature, velocity, and NOx profiles at the catalyst face were not the causes for the reported problems.
The original supplier's ammonia mixer and injection grid were replaced with designs based on the guidelines discussed here.
As a result of replacement with the newly designed equipment, ammonia consumption has been lowered, and EPA requirements for both NOx reduction and ammonia slip are being met or exceeded.
BIBLIOGRAPHY
Senecal, Industrial Engineering Chemistry, Vol. 49, pp. 993-997, (Circa 1957).
Rudinger, Textron (Bell Aerospace) ASME Symposium on Non-equilibrium Two-Phase Flow, W. A. M (1975).
Breidenthal, et at., AIAA journal, Vol. 24, pp. 1867-1869, November 1986.
Cobb, David, et al., "Application of Selective Catalytic Reduction (SCR) Technology for NOx Reduction From Refinery Combustion Sources," Environmental Progress, Vol. 10, No. 1, February 1991, pp. 49-59.
Ichiki, Masayoshi, et al, "Development of New Type De-NOx Catalyst-Reactor Design and Performance in the Actual System, "The Hitachi Zosen Technical Review, Vol. 52, No. 3, November 1990.
Ichiki, Masayoshi, "Development of New Type De-NOx catalyst-Catalyst Structure and Basic Characteristics," The Hitachi Zosen Technical Review, Vol. 2, No. 2, December 1990.
Fluent Inc., Fluent Vol. 4, User's Manual, p. 223, (1992).
Hartman, et al., AIAA journal, 91-0576, 29th Aerospace Sciences Meeting.
Groot, et al., AIAA journal, 25, 8, pp. 1142-1144.
Rathgeber, D.E., Thesis, Department of Chemical Engineering, Queen's University at Kingston, 1974.
Bosch, H., and Janssen, F., AIChE Series: "Catalysts & Reactors for Emissions Control," AIChE, 1989, New York.
Copyright 1993 Oil & Gas Journal. All Rights Reserved.
