Eric Hansen

In the August ICR article, “SNCR Emission Control” by Horton, Linero and Miller, it was suggested, “With mixing air fans to enhance gas turbulence in the system, improved mixing of gasses containing NOx and ammonia may offer possibilities.”  It was also noted that a correlation between plant size and reduction efficiency, that larger plants had lower efficiencies than smaller.  It was speculated that larger cross sectional areas would result in less mixing and direct contact with NH3 and NO2.  However, it was observed that increasing the number of nozzles did not overcome the presumed lack of mixing.  Acknowledging this observation, the authors proposed that the observations are a result of fluid dynamic mixing differences experienced or on stratification present as size increases.  The use of Computational Fluid Dynamic (CFD) modeling is a means to explore the possibilities of improving mixing and the distribution of the reducing reagent prior to conducting full scale testing.
Currently, the addition of ammonia reagents into cement kilns in the 900 C regions is focused on achieving distribution of the ammonia through the use of multiple nozzles.  A limitation of this approach may be what is being observed on large calciners; it is very difficult to uniformly distribute the reagent in large ducts.  Also, there is a possibility the gasses in the calciner are stratified prior to the introduction of the reagent so the gas composition to be treated is different in different areas of the calciner.  An alternative to distributing the reagent with multiple nozzles is to mix the reagent into the calciner gasses and to mix the calciner gasses by a technique developed by Cadence (US Patent no. 6,672,865) of inducing mixing and turbulence by the use of a high momentum air jet.  The amount of energy in this jet is not trivial, requiring a compressor of up to 200 hp on a 3300 tpd kiln.  The results of CFD modeling show that a gaseous reagent introduced in the high momentum jet is rapidly distributed throughout the duct.  Further, if there was pre-existing stratification in the duct, the high momentum jet would mix this into the total gasses in the duct.  This method of mixing can also be employed downstream of the injection of aqueous ammonia to improve its distribution.
An example of this method of mixing was modeled by CFD.  A jet of air containing and equal molar amount of ammonia to the NOx in the calciner was injected at sonic velocity perpendicular to the calciner flow.  Two cases were modeled with a mass flow of this jet of 2% and 6% of the mass flow in the calciner.  This jet induces turbulence in the duct which mixes the ammonia into the calciner gasses and mixes the calciner gasses.  In addition to the introduction of the SNCR reagent, the air jet supplies some of the excess air to complete combustion, lowering the excess oxygen upstream of the air injection resulting in reduced formation of nitrogen oxides.  Also, if stratification is present in the calciner, the reduction of the stratification after the mixing jet would allow reduction of the excess oxygen at the calciner exit for the same CO emission level, allowing further reduction of excess oxygen in the primary combustion zone further contribution to less NOx formation.
The use of the mixing technology not only allows the use of a less costly gaseous reducing reagent like anhydrous ammonia, it can make conventional aqueous injection more effective by creating a turbulent zone to mix the injected reagent more effectively. 
There is a significant cost savings for the use of anhydrous ammonia relative to aqueous ammonia or urea with the magnitude of the cost of anhydrous ammonia often being only 60% or less that of aqueous ammonia or urea.
The mixing technology also enables the application of SNCR to long kilns.  Several long kilns already use the mixing air technology to create staged combustion for NOx reduction.  By adding a manifold to the fan inlet anhydrous or aqueous ammonia can be introduced.
CFD modeling can be a valuable tool to understand the pre-existing condition in a calciner (stratification) and the nature of the distribution of the SNCR reagent.  By understanding the mixing, means can be employed to improve the distribution such as the Cadence mixing technology.

Implementation of CFD Modeling

It is generally accepted that the cement industry is technically very complex with several time dependent and interrelated processes occurring simultaneously. Traditionally, detailed mathematical modeling solutions were perceived as too difficult and more practical approaches have been favored by experienced plant engineers.  However, through close interaction with the industry, complicated problems may be sub-divided into more manageable ones and handled using modern computational fluid mechanics (CFD) provided that reliable inlet and boundary information is available. Cinar has recognized this potential and developed dedicated pre-calciner and clinker mathematical modeling which directly interacts with existing and proven CFD combustion modeling (see, for example, 1-4).  To demonstrate the effectiveness of CFD, a precalciner is modeled using the Cinar multi-fuel combustion code incorporating a calcination model and coupled with a detailed NOx post-processor to account for SNCR reactions.
 
Briefly, the mathematical model is formulated for steady incompressible high Reynolds number flows.  For the numerical solution of the governing equations, the flow domain is discretized by cylindrical/Cartesian and unstructured geometrical mesh types, according to the geometry of the equipment.  An improved version of the two-equation ‘k-e eddy viscosity closure’ is used to model the turbulence transport terms, where the turbulence viscosity and the eddy diffusivity are obtained in the terms of the turbulence kinetic energy and its dissipation rate determined from their modeled transport equations.  The conserved scalar approach is used to model the combustion, where the chemical reactions between the element species, admitted from any number of initially segregated streams, are considered to be governed by the mixing pattern of the streams.  The thermo-chemical state of the flow is then related to a strictly conserved scalar quantity, termed the ‘mixture fraction’, which may be related to the local fuel to air ratio.  The chemical state relations are estimated assuming a binary single-step fast reaction for the fuel element disappearance, which results in a set of piecewise linear functions in mixture fraction space.  To account for the thermal radiation heat transfer within the flow domain, additional balance equations are solved for the enthalpy and radiation fields.  In the computational scheme, the thermal radiation transfer is determined by the numerically exact ‘discrete transfer’ procedure. 
 
The thermal decomposition/calcination of the hot-meal is simulated using a calibrated single reaction model.  The meal prediction method is based on a technique where the combustion calculations, and thus the heat transfer from the flame, are dynamically linked with a procedure that computes the calcination rate and therefore provides the thermal boundary conditions for the combustion calculations.  The calcination reaction consumes the majority of the heat and liberates significant amounts of CO2, all of which are modeled by the combustion/process interactive calculation.
 
The chemistry of nitrogen at high temperature in combusting systems is extremely complex involving well over 200 elementary reactions between molecules, atoms and radicals.  For the purposes of engineering calculations, simplified models are of necessity.  However, the moment one substitutes a model for the fundamental chemistry, validation of the model is essential. Cinar Ltd has carried out model extensive validation exercise, in collaboration with an academic establishment, Imperial College London as well as its industrial clients within the power generation, cement manufacturing and other combustion based sectors. (for example: Ansaldo, Holcim, Total)Figure1 gives a pictorial summary of the NOx formation and reduction mechanisms.

The SNCR reduction chemistry is such that it is only effective within a ‘temperature window’ in the range 900 – 1100 C.  Unfortunately, this range only intersects precalciner conditions at the upper temperature limits of operation. As a result reagent injection in the kiln riser is sometimes preferable, although it is possible to identify effective regions near the burners of a precalciner.  In addition to the temperature, other factors affect the NOx reduction efficiency: the mixing between the reagent and the NOx infected gases; the molar ratio of ammonia to nitric oxide; the residence time of the mixture and the initial or pre-injection NOx level.  If the reduction reactions are too slow because these conditions are not optimized, then unreacted reagent will escape with the exhaust gases. This condition is known as ‘ammonia slip’. Worse still, the unreacted ammonia can react with SO3 and lead to build up problems.  Given that there are so many factors influencing the effective use of SNCR, the daunting question is: how can we ensure successful and optimal usage? 
The optimal injection strategy, which will be plant specific, is far from clear. One could reflect and come to some ad hoc decision to be effected based on identification of the temperature window through gas temperature measurements, but the extent of reactant mixing, the fate of un-reacted ammonia, the use of alternative fuels and the effect on the clinker production rate may only be assessed through CFD flow simulations, which account for all of the relevant physical process changes.

Comments on Results:

The CFD model is a typical inline precalciner with two tertiary air entry locations, one hot-meal and one pulverized coal above and below tertiary air inlet, respectively.  In order to improve mixing and reduce NOx via SNCR, two and six percent of the total combustion air containing NH3 at 1.0 molar ratio to NOx, were injected at sonic velocity through a nozzle, Case 1 and 2 respectively.  Results are presented for a ‘Base case’, without the introduction of the mixing air jet, and two further cases, ‘Case 1 and Case 2’ for two different injection angles and flow rates at the same elevation.  In general, precalciners suffer from flow stratification, as seen from coal particles’ trajectories, volatile release, oxygen and temperature profiles. This results in lower char-burnout (82%) and relatively lower calcination levels (88%) see Figure 2. 

The introduction of the mixing air jet, containing NH3, increases the mixing in both Case 1 and 2 but the performance in Case 2 is much better as compared with the former due to its angle as well as the much higher velocity.  As a result, coal burnout increases to 88 to 95 % and calcination level from 92 to 95 % for Cases 1 and 2, respectively.  The NO exit values are also reduced in Cases 1 and 2, as compared with base case, by some 20 to 35%, with ammonia slip concentrations of 8 and 5 ppm, respectively.  It is believed that further reductions of NOx emissions can still be achieved by reducing the flow stratification as well as a better distribution of NH3.

 References:
1.  Lockwood, F.C, Shen, B. and Lowes, T. "Numerical study of petroleum coke fired cement kiln flames," Presented at the Third International Conference on Combustion Technology. Lisbon, 1995.
 
2.  Lockwood, F.C. and Shen, B. "Performance predictions of pulverized coal flames of power station and cement kiln types," 25th International Symposium on Combustion, The Combustion Institute, pp. 503, 1994.
3.  Yousif, S., Abbas, T. and Lockwood, F. (2002).  Mathematical Modeling of Combustion and Cement/Lime Formation Processes in Rotary Kilns Fired with Conventional and Alternative Fuels.  Proceedings of the 12th International Cement Conference and Exhibition (AUCBM), Vol II, pp. 34, Marrakech, Morocco, 24-27th October 2002.
4.  Abbas. T., Lockwood, F. and Akhtar, S. ‘Plant Performance Improvement Through Mathematical Modeling’ Paper to appear in ZKG, 2006.
5.  Dimitriou, D., ‘Applied Mathematical Modeling of NO Abatement in Pulverize Fuel Furnaces’, Imperial College London, PhD Thesis, 2004.
6.  Sawyer R.F., ‘The Formation and Destruction of Pollutants in Combustion Processes’, Eighteenth Symposium (International) on Combustion, The Combustion Institute, p1, 1981.
7.  De Soete G.G., ‘Heterogeneous N2O and NO Formation from Bound Nitrogen Atmos during Coal Char Combustion’, Twenty-Third Symposium (International) on Combustion, The Combustion Institute, p1257, 1990.
8.  Levy J.M., Chan L.K., Sarofim A.F., Beer J.M., ‘NO/char Reactions at Pulverized Coal Flame Conditions’, Eighteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, PA, p111, 1981.
9.  Brouwer J., Heap M.P., Pershing D.W., Smith P.J., ‘A Model For Prediction of Selective Non Catalytic Reduction of Nitrogen Oxides by Ammonia, Urea, and Cyanuric Acid with Mixing Limitations in the Presence of CO’, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, p2117, 1996.