# Numerical study of air staging in a coke oven heating system

- 2.2k Downloads
- 3 Citations

## Abstract

The air staging to combustion approach introduced to a coke oven heating system as a primary method of nitrogen oxide (NO) formation reduction is considered in this paper. To numerically investigate the thermal and prompt NO formation, a heating flue model representing the most popular Polish coke oven battery was used. The model was developed and experimentally validated as a transient coupled model for the representative heating flue and the two coke ovens. Numerical simulations were performed to estimate the amount of NO passing into the atmosphere during the operation of such a heating system with and without the secondary air inlets. Various strategies for the secondary air distribution along the flue gas flow as well as the secondary air velocity were studied. The results of the numerical investigation demonstrated the substantial positive effect of the considered air staging on NO formation reduction.

## Keywords

Coke oven battery Heating flue Coke oven gas NO reduction Air staging## Introduction

Considering its emission and significantly toxic characteristics, nitrogen oxide (NO) pollution represents an important issue in modern combustion technologies. This substance contributes to smog formation, causes plant decay and negatively impacts living creatures. A vast majority of the emitted NO is created by burning fossil fuels which cause the crucial impact on the ecological area. The emission limitations of NO are also related to the economical area. An interesting analysis concentrated on an influence of the investment part on the air quality was presented in the literature (Panepinto et al. 2014). Further analysis which concerned the time variable emission was provided by Kumar et al. (2016). In that work, a global perspective of various emission sectors was studied. An illustration of the European Union developed on the basis of NO and SO_{2} emission coming from Spain was presented by García-Gusano et al. (2015). The economic aspects were investigated by those authors, and conclusions based on renewable energy sources were presented. Similar analysis focused on the Polish energetic sector was performed as well. Namely, Ilamathi et al. (2013) analysed renewable sources located in sea-shore district of Poland. Nevertheless, the investigation of this character was not presented for other districts of Poland. However, according to Houshfar et al. (2014), the emission due to pyrolisis and combustion processes takes a position of serious pollutant in such an industrial sectors as those located in the south of Poland.

Currently, NO emission is legally limited in many countries around the world. Each novelisation of greenhouse and toxic gas emission limits in countries of the European Union provides a significant restriction on the permitted amount of pollution released into the atmosphere, as defined in Directive 2001/80/EC in the Official Journal of the European Communities. For example, large combustion plants supplied with hard coal, where the nominal thermal power exceeds 500 MW, are obligated not to emit more than 500 mg/Nm^{3} of NO with a mole fraction of the oxide in the flue gases of 6 %. The most tangible restriction (binding since 1 January, 2016) sets the standard for NO emission to 200 mg/Nm^{3}.

The problem of NO emission has been studied in numerous works. Loffler et al. (2006) considered the contribution of thermal and prompt mechanisms of NO generation on the overall NO emission for the premixed combustion of natural gas. The result of that work was a simplified model of thermal NO formation. Ma et al. (2009) developed biomass co-firing models and described the effects of this process on NO emission. In the work of Miltner et al. (2008), Computational Fluid Dynamics (CFD) methods were used to optimise the work of a solid biomass combustor and predicted the fuel-NO emission. Zhou et al. (2014) analysed the influence of the primary air pipe structure on the NO emission of a swirl burner. A numerical study of NO formation from turbulent lean-premixed flames was investigated by Kang et al. (2009). They used a flamelet model to represent the complex turbulence-chemistry interaction. In the work of Ilamathi et al. (2013), the authors performed an experimental study and artificial neural network computations to estimate NO emission from a pulverised coal-fired boiler. The results of experimental research on NO formation in a PF oxy-fuel firing system are described in the work of Ndibe et al. (2013). In the study of Boyd Fackler et al. (2011), both experimental and numerical investigations were performed for a high-intensity, single-jet, stirred reactor to predict the amount of emitted NO.

Substantial changes in NO emission limits require efficient methods for reducing NO formation. Different approaches have been developed to this end. In general, these methods can be divided into primary and secondary methods. Primary methods for decreasing NO emission (Annamalai and Puri 2006) include rearranging the combustion process itself to obtain a lower amount of NO passing into the atmosphere. Operations of this type are widely described in the literature. In the works of Nimmo et al. (2008), the reburning method, defined as a three-stage combustion process, was presented. A decrease in NO emission based on the use of this approach was achieved by reducing the maximum temperature within the gas domain. A similar method was considered in the work of Hampartsoumian et al. (2003), where the reburning fuel was a pulverised coal. The use of overfire air to achieve combustion in a tangentially fired utility boiler was studied in the work of Diez et al. (2008). The authors developed and then validated a mathematical model that can extend analysis of the implementation of the NO emission reduction using primary methods. Cozzi and Coghe (2012) applied air staging to swirling burner. The authors considered a non-premixed natural gas combustor under overall lean conditions. Air staging and swirl burners were also a main topic of the work of Li et al. (2015). In that work, various secondary air mass flow rates and their influence on carbon oxide and NO emissions were considered. The concept of air staging in large-scale industrial boiler has been widely discussed by Higgins et al. (2010). The authors proposed the RROFA technology for a reduction of NO emission to follow environmental regulations. The work of Wang et al. (2014) shows the possible application of a co-firing process with air staging on the NO emission reduction using advanced CFD models.

The main source of NO pollution is the heavy and power industries. One of the most important branches of heavy industry is coke production. Coke is a fuel used to supply blast furnaces and contributes to a major part of worldwide steel production. This fuel is obtained via the coking process, namely by heating hard coal without air access. The thermal energy required by this process is generally provided through a coke oven gas (COG) or blast furnace gas (BFG). In Polish coke plants, COG is mainly used as a by-product of the coking process. Therefore, a significant amount of NO, i.e. 400 g/Mg to 950 g/Mg of coke, is formed during coke plant operation (Report 2005). This range depends on the battery type, age, technical state and applied emission reduction technologies.

The problem of NO pollution in the coking industry has already been studied. Jin et al. (2013) investigated the influence of BFG staging on NO emission. The authors performed numerical computations on a coupled model of a coke oven furnace and two halves of coking chambers. Their results showed that the staging fuel (as well as its distribution among specific inlets) has a large impact on the resulting NO pollution. Another example is the work of Weiss et al. (2012), in which NO emission reduction in a coke oven furnace was also investigated. These authors investigated the influence of BFG staging enriched with natural and converter gases.

Unfortunately, other than the idea of external flue gas recirculation, modifications to the coke oven battery in operation aimed at NO reduction cannot be easily implemented due to the construction limitations of plants. Namely, an approach that involves air staging requires additional inlets at selected levels. This means that a rebuilding of the furnace wall is necessary to include new air ducts. The wall is made of silica brick, and its operating temperature cannot exceed 1273 K during the coking process. Alternatively, longer processing times than indicated by coking technology exclusion of the combustion process lead to a non-permissible decrease in the wall temperature. Such a temperature drop causes a degradation of the mechanical properties of the brick and consequently results in the destruction of the entire construction. This has a crucial impact on the realistic possibilities of the air staging implementation when considering a coke plant in operation. Taking the cost aspects of this technology into account, it is worth noting that reburning technologies are still a more expensive primary method of NO reduction. Therefore, the idea of staging air combustion seems to be a very interesting approach.

The objective of this paper is to investigate air staging combustion in the most modern coke oven battery (PWR-63) on the Polish coke production market. The investigated type of coke oven is yet to be studied in terms of NO emission reduction. In addition, this technology of an oxidizer supply implemented in, for example, steam boilers cannot be directly compared with this application because of the different combustion conditions. The main differences include the geometry of the combustion chamber resulting in different values of the velocity and turbulence intensity fields and the manner of heat transfer. However, the most important difference is in the air–fuel mixture inlets because simple rectangular ducts are applied in the coke oven battery, and low-emission swirl burners are used in steam boilers.

This paper is also important for Polish emission reduction policy because Poland produces coke at one of the highest rates in the EU, contributing almost 25 %. Many reports (for example, Report of the Polish Ministry of Environment, 2005) show that the emission of harmful substances, including NO, from coke oven batteries is a serious problem. That document also indicated that air staging technologies represent a potentially effective way to decrease such emissions.

To perform this study, a fully developed model of the combustion chamber in the above-mentioned unit was used. The model was first formulated and subsequently experimentally validated in a transient coupled form. The model includes two coke ovens and a heating flue as a representative domain of the entire coke oven battery (Smolka et al. 2016). For this study, the model of the heating flue was used only as described in the work of Smolka et al. (2014). However, the boundary conditions were taken from the transient coupled model and then time averaged. The model was completed with a number of staging air inlets located in the vertical walls between the heating flue and the coke oven furnace.

To apply the most efficient strategy for staging air combustion, various approaches of secondary air delivery are analysed in this work, including an approach considering the height of the inlet location in the wall and a combination of the parallel operation of the air streams and the air velocity at the inlets. Following numerous computations, the most promising distribution of the secondary air in the proposed additional inlets was found. The emission reduction obtained under the final solution is substantial when compared to the case without air staging.

## Model of gas combustion in coke oven furnaces

### Physical model

A coke oven battery is formed by a series of heating flues and coking chambers that are placed in rows to maximise the heat transfer rate. The battery heating system consists of a number of furnaces with upward and downward flues. To endure the high temperatures, the heating walls of the coke oven battery are made of silica brick.

In this system, the air and fuel are introduced into an upward flue. Then, the air–fuel mixture combusts, and the hot flue gases flow upward, pass the bridge window and flow down along a downward flue. The flue gas outlet is located at the bottom of the downward flue. The flue gas flow increases the temperature of the heating walls via convection and radiation. Then, heat is transferred to a coking chamber through the wall. A schematic view of the heating flue is shown in Fig. 1.

The required temperature of the coke in the centre of the coking chamber at the end of the coking cycle is approximately 1273 K. Thus, the temperature of the flue gases should be in the range of 1673 K to 1873 K. To increase the thermal efficiency of the entire process, the enthalpy of hot flue gases is employed to heat the COG and air to combustion using a ceramic regenerator. Additional details of the heating system installed in the coke oven plant can be found in the works of Smolka et al. (2014) and Smolka et al. (2016).

### Mathematical model

- Energy equation solved for temperature, which is coupled with the velocity field for fluids (mixture of gas, air and flue gases) and for the temperature fields in solid walls:$$\frac{{\partial \left( {\rho u} \right)}}{\partial t} + \nabla \cdot \left( {\rho {\mathbf{v}}h} \right) = \nabla \cdot \left( {k\nabla T + \tau \cdot {\mathbf{v}}} \right) -\nabla \cdot \left( {\mathop \sum \limits_{i} h_{i} J_{i} } \right) + S_{h}$$(1)$$J_{i} = - \rho D_{i} \nabla y_{i} \quad h = \mathop \sum \limits_{i} h_{i} y_{i} \quad u = h - \frac{p}{\rho }$$(2)where$$h_{i} = \mathop \smallint \limits_{{T_{ref} }}^{T} c_{{p_{i} }} dT$$(3)
*ρ*is the density,*u*is the internal energy,*t*is the time,*h*is the specific enthalpy,*k*is the effective thermal conductivity,*v*is the velocity vector,*T*is the temperature,*τ*is the effective stress tensor,*J*is the diffusion mass flux,*S*is the source term in the transport equation of the scalar quantity,*D*is the mass diffusion coefficient,*y*is the mass fraction,*c*_{ p }is the isobaric specific heat and*p*is the pressure. - Continuity equation solved for pressure coupled with the velocity vector of the fluids:where$$\frac{\partial \rho }{\partial t} + \nabla \cdot \left( {\rho {\mathbf{v}}} \right) = \mathop \sum \limits_{i} S_{{y_{i} }}$$(4)
*S*_{yi}is the mass source term of the*i*th species. - Momentum equation solved for the velocity vector components coupled with the pressure of the fluids:where$$\frac{{\partial \left( {\rho {\mathbf{v}}} \right)}}{\partial t} + \nabla \cdot \left( {\rho {\mathbf{vv}}} \right) = - \nabla p + \nabla \cdot {\bar{\bar{\tau}}} + \rho \varvec{g} + \varvec{S}$$(5)
is the gravitational acceleration vector and**g**is the source term in the momentum transport equation.**S** - Turbulence (two-equation k-ε realisable turbulence) solved for the turbulence kinetic energy and the dissipation of the turbulence kinetic energy in the fluids:$$\rho \left( {{\mathbf{v}}\nabla \kappa } \right) = \nabla \cdot \left[ {\left( {\mu + \frac{{\mu_{T} }}{{\Pr_{k} }}} \right)\nabla \kappa } \right] + S\left( {\mu_{T} ,\kappa ,{\mathbf{v}}} \right) - \rho \varepsilon$$(6)where$$\rho \left( {{\mathbf{v}}\nabla \varepsilon } \right) = \nabla \cdot \left[ {\left( {\mu + \frac{{\mu_{T} }}{{\Pr_{\varepsilon } }}} \right)\nabla \varepsilon } \right] + C_{{\varepsilon_{1} }} \frac{\varepsilon }{\kappa }S\left( {\mu_{T} ,\kappa ,{\mathbf{v}}} \right) - C_{{\varepsilon_{2} }} \rho \frac{{\varepsilon^{2} }}{\kappa }$$(7)
*κ*is the turbulent kinetic energy,*μ*is the dynamic viscosity, Pr is the Prandtl number,*ε*is the dissipation rate of the turbulent kinetic energy and C_{ ε1}and C_{ ε2}are the model constants. This model was compared with other two-equation models offered in commercial software in the previous works of Smolka et al. (2014) and Smolka et al. (2016) and successfully validated. - Species transport. The number of these equations depends on the gas composition:$$\frac{{\partial \left( {\rho y_{\text{i}} } \right)}}{\partial t} + \nabla \cdot \left( { \rho {\mathbf{v}}y_{\text{i}} } \right) = \nabla \cdot \rho D_{\text{i}} \nabla y_{\text{i}} + S_{{{\text{y}}_{\text{i}} }}$$(8)
- Radiation (Discrete Ordinate model) solved for the radiative heat sources in the fluids and for the radiative heat fluxes on the surfaces of the solid walls$$\nabla \cdot \left( {I_{\lambda } \left( {\varvec{r},\varvec{s}} \right)\varvec{s}} \right) + \left( {a_{\lambda } + \sigma_{s} } \right)I_{\lambda } \left( {\varvec{r},\varvec{s}} \right) = a_{\lambda } n^{2} I_{b\lambda } + \frac{{\sigma_{s} }}{4\pi }\mathop \smallint \limits_{0}^{4\pi } I_{\lambda } \left( {\varvec{r},\varvec{s}^{\prime}} \right)\varPhi \left( {\varvec{s} \cdot \varvec{s}^{\prime}} \right)d\varPhi^{\prime}$$(9)$$\dot{q}_{r} = a\int\limits_{\omega\,=\,4\pi } {\left( {i - i_{b} } \right)d\omega }$$(10)$$\frac{{di\left( {r,\xi } \right)}}{d\xi } = - ai\left( {r,\xi } \right) + ai_{b} \left( {r,\xi } \right)$$(11)where$$\dot{q}_{{{\text{v}},{\text{r}}}} = - \nabla \cdot \dot{q}_{\text{r}}$$(12)
*I*is the intensity,*a*is the absorption coefficient,*I*_{bλ}is the black body intensity, σ is the Stefan-Boltzmann constant,*n*is the refractive index,*s*is the scattering direction vector and Φ is the phase coefficient.

Material properties of the flue gases and heating flue walls (temperatures are given in K)

Property | Fluid subdomain | Solid subdomain |
---|---|---|

Density | Ideal gas model | 1800 |

Thermal conductivity | Mass-weighted average for the flue gases and the polynomial functions of the temperature for the component gases | 0.000,628·T + 0.9334 |

Specific heat | 0.25,116·T + 768.633 | |

Mass diffusivity | – | |

Dynamic viscosity | Molar-weighted average based on temperature-dependent binary diffusivities | – |

Absorption coefficient | Weighted sum of the grey gas model | – |

For external walls at the bottom and top of the furnace, a convection boundary condition was defined. To simulate the influence of adjacent coking chambers, a heat flux boundary condition for the heating walls was employed. However, due to the reversing of the direction of the flue gases flow, the real coking process becomes unsteady. Hence, to simplify the numerical investigation, the value of the heat transferred through the heating walls was assumed to be an average value from the entire coking process. The value of 5000 W/m^{2} was estimated and successfully validated in the authors’ previous work on the coupled model of the heating system (Smolka et al. 2016). As a result of such an assumption, the numerical prediction of a typical amount of NO emitted over the entire cycle of coke oven furnace operation can be employed.

Mass flow boundary conditions for gas and air inlets

Parameter/component | Air | COG | |
---|---|---|---|

Mass flow rate, kg/s | – | 0.002196 | |

Temperature, K | 1473 | 1173 | |

Mass fraction | O | 0.20676 | 0.0209 |

CO | – | 0.1363 | |

H | 0.01592 | 0.0342 | |

N | 0.77732 | 0.1049 | |

CO | – | 0.2305 | |

H | – | 0.0883 | |

CH | – | 0.3090 | |

C | – | 0.0476 | |

C | – | 0.0204 | |

C | – | 0.0079 |

The simulation scenarios for the 60/40 and 70/30 distributions of the total air mass flow rate

Case no/Inlet no | 0 | 1 | 2 | 3 | 4 | 5 | 0 | 1 | 2 | 3 | 4 | 5 |
---|---|---|---|---|---|---|---|---|---|---|---|---|

Reference | 100 % | – | – | – | – | – | 100 % | – | – | – | – | – |

1 | 60 % | – | – | 40 % | – | – | 70 % | – | – | 30 % | – | – |

2 | – | – | – | 40 % | – | – | – | – | 30 % | – | ||

3 | – | – | – | – | 40 % | – | – | – | 0 % | 30 % | ||

4 | – | – | 13 % | 13 % | 14 % | – | – | 10 % | 10 % | 10 % | ||

5 | – | – | 24 % | 12 % | 4 % | – | – | 18 % | 9 % | 3 % | ||

6 | – | – | 4 % | 12 % | 24 % | – | – | 3 % | 9 % | 18 % | ||

7 | – | – | 10 % | 20 % | 10 % | – | – | 7.5 % | 15 % | 7.5 % | ||

8 | – | – | 16 % | 14 % | 10 % | – | – | 12 % | 10.5 % | 7.5 % | ||

9 | 8 % | – | 14 % | – | 18 % | 6 % | – | 10.5 % | – | 13.5 % | ||

10 | 4 % | – | 12 % | – | 24 % | 3 % | – | 9 % | – | 18 % |

#### Chemical reactions and NO calculation model

### Modelling the NO reduction process

Three mechanisms of NO formation may be distinguished: thermal, prompt and fuel mechanisms. The thermal mechanism of NO formation (Zeldovich mechanism) occurs in a large-scale manner when the temperature exceeds 1600 K (Wilk 2002) or even 1700 K (Warnatz et al. 2006). At such a high temperature, nitrogen molecules can dissociate into two atoms, creating an opportunity to form NO molecules. The prompt mechanism relies on reactions between the nitrogen and hydrocarbon molecules. Radicals obtained in this manner can easily be oxidised to NO. The fuel-NO formation mechanism occurs when nitrogen is chemically bonded in a fuel.

A determination of the NO emission rates was performed using post-processing computations based on the temperature and flue gas mole fraction fields. Thermal and prompt NO generation mechanisms are assumed. The former mechanism is mostly responsible for modelling hydrocarbon radicals, such as CH and CH_{2}, which can react very rapidly with nitrogen to produce NO. To simplify the employed numerical simulation, only the global kinetic parameters were used in the presented work to control the rate of NO production (Soete 1975).

_{2}+ O→ NO + N, N + O

_{2}→ NO + O), as proposed by Zeldovich (1946) and further extended by Lavoie et al. (1970) by considering the reaction of N with the radical hydroxyl OH (N + OH → NO + H). An important feature of the thermal NO mechanism is an assumption that the rates of formation and consumption of N atoms are equal. Hence, NO formation is defined as follows:

_{f,1}, k

_{f,2}and k

_{f,3}are the rate constants for the forward reactions, and k

_{r,1}, k

_{r,2}and k

_{r,3}are the rate constants for the reverse reactions. Equation (19) requires that the concentrations of O and OH be known. In the presented work, the partial equilibrium model described by Drake et al. (1987) was used to calculate the [O] and [OH] radicals using the following equations:

*f*is the correction factor,

*E*is the activation energy,

*a*is the oxygen reaction order,

*FUEL*are the fuel species and \(k^{\prime}_{pr}\) is calculated as 6.4

^{−6}(

*RT*/

*p*)

^{ a+1}. The source term added to the NO transport equation due to the presence of the prompt NO is defined as follows:

*f*in the above equation is defined as a function of the equivalence ratio

*φ*= 1/1

*λ*and the fuel-carbon number

*n*. The equation used for calculating the correction factor is as follows:

_{3}and HCN:

*Y*is the mass fraction of NO, NH

_{3}and HCN;

*S*is the source term computed based on different NO mechanisms and

*D*is the effective diffusion coefficient. The sum of the computed mass fraction is the total NO emission. In the case of combustion process modelling using Ansys FLUENT, the transport equations (Eqs. (26), (27) and (28)), similar to the remaining governing equations, are averaged using the density-weighted approach, where the mean turbulent reaction rates \(\bar{S}_{NO}\), \(\bar{S}_{NH3}\) and \(\bar{S}_{HCN}\) are calculated using the probability density function (PDF) theorem defined in terms of the temperature. To calculate the PDF, the cumulative density function in terms of the variance has to be known. The variances can be calculated using a transport equation or via algebraic solution. For computational stability reasons, the second method was employed in this work.

## COG combustion with air staging

Typical values of the air excess ratios used in the heating system of the PWR-63 coke oven battery are in the range of 1.3 to 1.5. In this paper, a single value of the air-to-fuel ratio of 1.5 was used. The theoretical basis for the approach of air staging for achieving combustion used in this paper was developed by Wilk (2002). The staging combustion performed in this study is based on the division of the total oxidiser mass flow rate into a primary and secondary stream. The primary air is used to form the reducing atmosphere in the first part of the combustion chamber. This means that the air-to-fuel ratio at the beginning of the combustion process must not exceed unity.

In this work, a magnitude of the air-to-fuel ratio of 0.8 to 0.95 was assumed for the reference cases. Such an assumption defined the 60/40 and 70/30 distributions of the total air mass flow. Those distributions are typical for air staging in industrial applications (Wilk 2002). This means that 60 or 70 % of the total air mass flow rate was supplied to the bottom inlet, and the remaining 40 or 30 % was distributed among additional inlets. Then, various approaches for the secondary air stream distribution were considered. Namely, five positions of the secondary inlet were numerically tested, as shown in Fig. 1. The air inlets are numbered from the bottom of the heating flue, where Inlet 0 is the primary air inlet and the side wall inlets are labelled from Inlet 1 to Inlet 5.

Table 3 presents the percentages of the air mass flow distributions among all the air inlets for the 60/40 and 70/30 distributions of the total air mass flow rate. In this work, the cases are distinguished based on the distribution of the secondary air stream. For example, in Case 1, 100 % of the secondary air mass flow (for both the 60/40 and 70/30 distributions of the total air mass flow) was supplied to Inlet 3.

## Results and discussion

For the 60/40 distribution of the total air mass flow rate, the lowest rate of NO emission is found in Cases 1, 6 and 7. In Case 1, the NO reduction in comparison to the reference case is approximately 38 %. The highest NO emission is found in Cases 3, 5 and 9. However, the difference between these cases and the reference case remained substantial, exceeding 30 %.

For the 70/30 distribution of the total air mass flow distribution, two tendencies can generally be distinguished. The NO pollution is noticeably greater when the secondary air stream is supplied to one inlet only (Cases 1 through 3) than if the secondary air stream is divided among three inlets (Cases 4 through 10, excluding Case 9). For Cases 5, 6 and 7, the NO reduction exceeds 40 %. In other cases, this value still exceeds 30 %.

The simulation scenario for the Case 6 analysis for the 60/40 and 70/30 distributions of the total air mass flow rate

Case/inlet | 0 | 1 | 2 | 3 | 4 (%) | 5 (%) | 0 | 1 | 2 | 3 | 4 (%) | 5 (%) |
---|---|---|---|---|---|---|---|---|---|---|---|---|

6 | 60 % | – | – | 4 % | 12 | 24 | 70 % | – | – | 3 % | 9 | 18 |

6A | – | – | 4 % | 10 | 26 | – | – | 3 % | 7.5 | 19.5 | ||

6B | – | – | 4 % | 14 | 22 | – | – | 3 % | 10.5 | 16.5 | ||

6C | – | – | – | 14 | 26 | – | – | – | 10.5 | 19.5 | ||

6D | – | – | – | 12 | 28 | – | – | – | 9 | 21 |

In Fig. 4, the higher velocity in Case 1 caused the NO emission to increase for the 60/40 distribution of the total air mass flow and the NO emission to decrease for the 70/30 distribution of the total air mass flow. In Fig. 5, the results obtained for Case 6 are presented. For the 60/40 distribution of the total air mass flow, the height decrease of the secondary air inlet area contributed to almost the same NO emission result. However, the smaller width of the secondary air inlet area caused a substantial NO formation reduction. For the 70/30 distribution of the total air mass flow, the decrease in the secondary air inlet area resulted in an NO emission decrease under both strategies.

To summarise, a negative influence of the decreasing secondary inlet area was predicted for Case 1 with the 60/40 air flow distribution, as shown in Fig. 4. For Case 6, presented in Fig. 5, the inlet width reduction of the secondary inlets resulted in a 10 % lower NO emission in comparison to the case with unmodified additional inlets.

## Conclusions

A numerical investigation of the function of a coke oven heating system with air staging has been performed. A variety of secondary air mass flow distributions were considered for steady-state computations. In this study, the computational domain was limited to the heating flue only. However, to ensure model accuracy, the crucial value of the heat transfer through the heating walls was time averaged based on a fully coupled transient model of the coke oven battery.

The results showed a positive impact on the NO reduction in each considered case. Even the cases with a single secondary inlet obtained a reduced emission of 30 % compared with the case without the air staging. In the case of additional air flow distributed between three secondary air inlets above a half height of the heating flue, the NO emission decreased by an additional 20 %. This result was consistent with the findings in the literature. Weiss et al. (2012) proved that staging on two levels is better than staging only on one level. In another work, Jin et al. (2013) showed that using three levels of staging was better than using only two levels.

In addition, a more beneficial effect on the NO emission for the 70/30 distribution compared to the 60/40 distribution of the total air mass flow rate was presented. The results showed that an appropriate distribution and size of the staging air inlets can reduce NO formation by an additional 10 % compared to the best case with a standard air inlet size.

The presented results are very promising for the Polish coke production sector. It becomes clear that the final distribution and size of the secondary inlets should be a matter of optimisation computations. However, future coke oven batteries or those retrofitted with the implemented air staging will potentially be able to meet stricter emission limits.

## Notes

### Acknowledgments

The first four authors gratefully acknowledge the financial support of the European Union Project “Competent mechanical engineers for energetic sector” no. POKL.04.01.02-00-131/12. The work of JS and WA was also partially supported by the statutory research fund of the Faculty of Power and Environmental Engineering.

## Supplementary material

## References

- Anderson JD (1995) Computational fluid dynamics.The basics with applications. McGraw-Hill, New YorkGoogle Scholar
- Annamalai K, Puri IK (2006) Combustion science and engineering. CRC Press, Boca RatonGoogle Scholar
- Bachmaier F, Eberius KH, Just T (1973) The formation of nitric oxide and the detection of HCN in premixed hydrocarbon–air flames at 1 atmosphere. Combust Sci Technol 7:77–84CrossRefGoogle Scholar
- Boyd Fackler K, Karalus MF, Novosselov IV, Kramlich JC, Malte PC (2011) Experimental and numerical study of NOx formation from the lean premixed combustion of CH4 mixed with CO
_{2}and N_{2}. J Eng Gas Turbines Power 133:121502–121507CrossRefGoogle Scholar - Cozzi F, Coghe A (2012) Effect of air staging on coaxial swirled natural gas flame. Exp Thermal Fluid Sci 43:32–39CrossRefGoogle Scholar
- Diez L, Cortes C, Pallares J (2008) Numerical investigation of NOx emissions from a tangentially-fired utility boiler under conventional and overfire air operation. Fuel 87:1259–1269CrossRefGoogle Scholar
- Directive 2001/80/EC of the European Parliament and of the Council (2001) On the limitation of emissions of certain pollutants into the air from large combustion plants. Off J Eur Commun 309:1–21Google Scholar
- Drake M, Correa S, Pitz R, Shyy W, Fenimore C (1987) Superequilibrium and thermal nitric oxide formation in turbulent diffusion flames. Combust Flame 69:347–365CrossRefGoogle Scholar
- García-Gusano D, Cabal H, Lechón Y (2015) Evolution of NOx and SO
_{2}emissions in Spain: ceilings versus taxes. Clean Technol Environ Policy 17:1997–2011CrossRefGoogle Scholar - Hampartsoumian E, Folayan OO, Nimmo W, Gibbs BM (2003) Optimisation of NOx reduction in advanced coal reburning systems and the effect of coal type. Fuel 82:373–384CrossRefGoogle Scholar
- Higgins B, Yan L, Gadalla H, Meier J, Fareid T, Liu G, Milewicz M, Repczynski A, Ryding M, Blasiak W (2010) Biomass co-firing retrofit with ROFA for NOx reduction. Polish J Environ Study 19:1185–1197Google Scholar
- Houshfar E, Wang L, Vähä-Savo N, Brink A, Løvås T (2014) Characterisation of CO/NO/SO
_{2}emission and ash-forming elements from the combustion and pyrolysis process. Clean Technol Environ Policy 16:1339–1351CrossRefGoogle Scholar - Ilamathi P, Selladurai V, Balamurugan K, Sathyanathan VT (2013) ANN–GA approach for predictive modelling and optimization. Clean Technol Environ Policy 15:125–131Google Scholar
- Jin K, Feng Y, Zhang X, Wang M, Yang J, Ma X (2013) Simulation of transport phenomena in coke oven with staging combustion. Appl Therm Eng 58:354–362Google Scholar
- Kang S, Kim Y, Lee KS (2009) Numerical simulation of structure and no formation of turbulent lean-premixed flames in gas turbine conditions. J Mech Sci Technol 23:3424–3435CrossRefGoogle Scholar
- Kumar A, Patil RS, Kumar AD, Kumar R (2016) Comparison of predicted vehicular pollution concentration with air quality standards for different time periods. Clean Technol Environ Policy March:1–11Google Scholar
- Kuo K (2005) Principles of combustion. Wiley, HobokenGoogle Scholar
- Lavoie GA, Heywood JB, Keck JC (1970) Experimental and theoretical study of nitric oxide formation in internal combustion engines. Combust Sci Tech 4:313–326CrossRefGoogle Scholar
- Li S, Li Z, Jiang B, Chen Z, Zhang X (2015) Effect of secondary air mass flow rate on the airflow and combustion characteristics and NOx formation of the low-volatile coal-fired swirl burner. Asia-Pac J Chem Eng. doi: 10.1002/apj.1923 Google Scholar
- Loffler G, Sieber R, Haraseka M, Hofbauer H, Hauss R, Landauf J (2006) NOx formation in natural gas combustion—a new simplified reaction scheme for CFD calculations. Fuel 85:513–523CrossRefGoogle Scholar
- Ma L, Gharebaghi M, Porter R, Pourkashanian M, Jones JM, Williams A (2009) Modelling methods for co-fired pulverised fuel furnaces. Fuel 88:2448–2454CrossRefGoogle Scholar
- Miltner M, Makaruk A, Harasek M, Friedl A (2008) Computational fluid dynamic simulation of a solid biomass combustor: modelling approaches. Clean Technol Environ Policy 10:165–174CrossRefGoogle Scholar
- Ndibe C, Sporl R, Maier J, Scheffknecht G (2013) Experimental study of NO and NO
_{2}formation in a PF oxy-fuel firing system. Fuel 107:749–756CrossRefGoogle Scholar - Nimmo W, Singh S, Gibs BM, Williams PT (2008) The evaluation of waste tyre pulverised fuel for NOx reduction by reburning. Fuel 87:2893–2900CrossRefGoogle Scholar
- Panepinto D, Brizio E, Genon G (2014) Atmospheric pollutants and air quality effects: limitation costs and environmental advantages (a cost–benefit approach). Clean Technol Environ Policy 16:1805–1813CrossRefGoogle Scholar
- Smolka J, Fic A, Nowak AJ, Kosyrczyk L (2014) 3-D periodic CFD model of the heating system in a coke oven battery. Int J Numer Methods Heat Fluid Flow 24:891–906CrossRefGoogle Scholar
- Smolka J, Slupik L, Fic A, Nowak AJ, Kosyrczyk L (2016) 3-D coupled CFD model of a periodic operation of a heating flue and coke ovens in a coke oven battery. Fuel 165:94–104CrossRefGoogle Scholar
- Soete GD (1975) Overall reaction rates of NO and N
_{2}formation from fuel nitrogen. 15th Symposium on combustion, vol 1, pp 1093–1102Google Scholar - Techniques Report Best Available (2005) Reference document for the coke industry. Ministry of Environment, Warsaw
**(in Polish)**Google Scholar - Wang X, Hu Z, Deng S, Xiong Y, Tan H (2014) Effect of biomass/coal co-firing and air staging on NOx emission and combustion efficiency in a drop tube furnace. Energy Procedia 16:2331–2334CrossRefGoogle Scholar
- Warnatz J, Maas U, Dibble RW (2006) Combustion. Berlin, GermanyGoogle Scholar
- Weiss C, Rieger J, Rummer B (2012) Formation and control of nitrogen oxide in the heating system of a coke oven. Environ Eng Sci 29:555–562CrossRefGoogle Scholar
- Wilk RK (2002) Low-emission combustion. Silesian University of Technology Publishing House, GliwiceGoogle Scholar
- Zeldovich YB (1946) The oxidation of nitrogen in combustion explosions. Acta Physicochim U.S.S.R. 21:577–628Google Scholar
- Zhou H, Yang Y, Liu H, Hang Q (2014) Numerical simulation of the combustion characteristics of a low NOx swirl burner: influence of the primary air pipe. Fuel 130:168–176CrossRefGoogle Scholar

## Copyright information

**Open Access**This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.