Combustion Simulation and study of water injection with fuel on NOx and soot formation in Ansys Fluent

Combustion Simulation

1. Introduction:

Combustion is basically defined as a chemical reaction between substances, usually including oxygen, and usually accompanied by the generation of heat and light in the form of flame. CFD solvers are often used to simulate the combustion problems for engineering and research purposes.

All combustion solvers use the complete chemical kinetics involved in the process which is simulated. For engineering simulations mechanisms have 100-200 reactions and up to 150 species whereas for research simulations the numbers can go up to 1000. A compact notation is used to feed the reactions into the computer.

1.1 Possible types of combustion simulations available in the Ansys Fluent

1.1.1 Based on mixing

1. Non-premixed combustion (direct injection, Late injection problems)
2. Premixed combustion (engines using a carburetor)
3. Partially premixed

1.1.2 Based on phase

1. Fluid phase (volumetric reaction)
2. Wall surface reactions
3. Particles (surface reactions)
4. Porous region (ex- after treatment systems)

Non-premixed combustion is like one in a diesel engine where the fuel is injected directly right before the combustion. Premixed combustion is one like in petrol engines where a carburetor is used to mix the fuel and air in metered quantity before supplying into the combustion chamber.

1.2 Basics of combustion simulation

Equations solved:

1. Mass
2. Momentum

In addition to these two conservation equations if turbulence is present then the turbulence parameters like K and epsilon, for the K-epsilon model, need to be solved. Sometimes the transport equation also needs to be solved.

In this study, a non-premixed combustion case is analyzed in Ansys Fluent. The study has two parts, the first part is a simulation of combustion in a combustor model and calculation of the mass fraction of various species inside the domain. The second part deals with the addition of water in the fuel in the range of 5 to 10% and analysis of its effects on the generation of pollutants NOx and soot.

2. The first part of the study.

2.1 Geometry:

A CAD model of the combustor was provided as shown in the Figure:

Fig 2.1.1: CAD model cross-section view,

Since the 3D simulation of the same would have been computationally very expensive, the problem was converted to Axysymmetric for the purpose of simplification. The axisymmetric approach can be applied to a problem if it's symmetric about an axis and the interpretation of the results are valid for the entire 3D model which can be generated by the rotation of the planar model about the axis.

In our case as clear from the figure the model is having rotational symmetry about the X-axis, a 2D model as shown below was prepared to simulate the case:

Fig 2.1.2: Axisymmetric 2D model used.

2.2 Meshing:

The prepared surface body was meshed in mesher with following settings:

Named Selections:

Fig 2.2.1: Named selections used for assigning boundary conditions.

Following sizing settings were used:

The proximity function was used to capture the flow properly near the fuel inlet. The local refining of mesh near the fuel inlet is clearly visible in the meshed model from below figure.

Fig 2.2.2: Meshed model.

The above meshing strategy resulted in 17477 nodes and 17012 elements.

Fig 2.2.3: The mesh quality

Mesh quality obtained was above 50% and most of the elements were of the quadrilateral type which is a good mesh for the case.

2.3 Setup:

General settings:

Solver Type	Pressure Based
2D Space	Axisymmetric
Time	Steady

Pressure based solver was used as the flow velocities involved are not much higher and the compressibility effects can be safely ignored. The axisymmetric scheme was used due to reasons explained in section 2.1. The steady-state solver was used as we are more concerned about the end results of the simulation like the mass fraction of pollutants and temperature distribution.

Models used:

Viscous model	K-epsilon (2 eqn)
K-epsilon model	standard
Near wall treatment	Standard wall function
Species	Species transport
Reaction	Volumetric
Mixture material	methane-air-2 step
Turbulence chemistry interaction	Eddy dissipation

The K-epsilon viscous model was selected so as later on the eddy dissipation interaction could be selected. The methane-air-2 step mixture material was selected as it enables the addition of water in the fuel which is required for the second part of the study.

Finite rate eddy dissipation model: It is a combination of the finite rate chemistry model and mixing model. In it the turbulence time scale which is the ratio of k and epsilon is used to determine the occurrence of burning.

NOx and Soot models used:

For the calculation of mass fractions for the NOx and Soot particles were turned on as follows:

NOx: Only Thermal NOx was turned on and as the fuel used does not contain any nitrogen in it so Fuel NOx, as well as the Prompt NOx, were not used. A single step Soot model was used for Soot formation calculation.

Boundary Conditions:

Table 2.3.1: The Boundary Conditions used

Calculation of inlet mass fraction:

• CH4+ ar(O2+3.76 N2) --> a. CO2 +b. H2O+ c.N2
• Where a,b and c are coefficients and ar represents no of moles of air: as air contains 21% O2 and 79% N2 on Volume basis so a fractional representation was made as [(1.)O2 + (79/21 = 3.76) . N2)]
• Further for balancing when we solve for coefficients, we get: ar=2, a=1, and b=2
• So for the air, we have 2 moles i.e. 2 moles of O2 and (3.76*2= 7.52 ) moles of N2
• Hence mole fraction of O2 is 2/(7.52+2)=0.21 and 7.52/9.52= 0.79 for N2 which is true and our calculation is correct.

Further, these fractions when calculated on a mass basis, come out to be 0.23 for O2 and 0.77 for N2 which was used for the inlet.

Note: in the inlet species window there is no mention of N2 as its an inert gas and shows up directly in the RHS term of the reaction without affecting the mechanism.

For the fuel inlet, as we wanted to introduce only methane, we used a mass fraction of 1 for it.

2.4 Solution and Results:

The solution was initialized with the hybrid initialization option and 1500 iterations were run.

Fig 2.4.1: The residuals after 1500n iterations.

The oscillations of the residuals have become quiet periodic in nature which indicates that the solution is converged.

2.4.1: Mass fraction of different species inside the domain.

The mass fraction of different species were plotted on probe lines to investigate the presence of pollutants and unburnt fuel particles.

Table 2.4.1.1: Location of line probes inside the domain used in CFD Post

Fig 2.4.1.1: Line probes inside the domain.

2.4.1.1. Mass fraction of CH4 plotted on all the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.2: Variation of CH4 on the probe lines

Note: Overlapping of 0.4 m line, 0.6 m line and outlet lines can be observed in the domain.

Fig 2.4.1.3: Variation of CH4 on the domain

2.4.1.2. Mass fraction of CO2 plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.4: Variation of CO2 on the probe lines.

Fig 2.4.1.5: Variation of CO2 on the domain

2.4.1.3. Mass fraction of H2O plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.6: Variation of H2O on the probe lines.

Fig 2.4.1.7: Variation of H2O on the domain

2.4.1.4. Mass fraction of O2 plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.8: Variation of O2 on the probe lines.

Fig 2.4.1.9: Variation of O2 on the domain

2.4.1.5. Mass fraction of N2 plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.10: Variation of N2 on the probe lines.

Fig 2.4.1.11: Variation of N2 on the domain

2.4.1.6. Mass fraction of NOx plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.12: Variation of NOx on the probe lines.

Fig 2.4.1.13: Variation of NOx on the probe lines.

2.4.1.7. Mass fraction of Soot plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.14: Variation of Soot on the probe lines.

Fig 2.4.1.15: Variation of Soot on the domain

2.4.1.8. Mass fraction of CO plotted on the probe lines (mass fraction Vs vertical distance on line)

Fig 2.4.1.16: Variation of CO on the probe lines.

Fig 2.4.1.17: Variation of CO on the domain

2.4.3 Temperature contour

Fig 2.4.3.1: Temperature contour plotted on the domain.

The maximum temperature generated inside the combustor is about 2490 K.

2.4.4 Inference:

Based on the above-plotted charts following points can be made:

• CH4 mass fraction goes on decreasing as we proceed towards the outlet which is a good sign that no unburnt fuel is escaping.
• Most of the CO2 production is happening near the first line i.e. around 0.15 m from the inlet and
• NOx production is apparently concentrated around the region up to 0.13 m from the inlet.
• The maximum soot produced is at the outlet.
• The maximum temperature attained in the combustor is around 2490 K i.e. 2200 deg Celsius, so a material selection decision needs to be based on sustaining up to this temperature. Also, a cooling system can be provided to cool down the surface for selecting a material with lower melting points based on the application in which the combustor is being used.

3. The second part of the study

This part of the study comprises of simulation of the same set-up with an injection of water in the fuel which is aimed at the reduction of NOx and soot produced.

Methodology up to the meshing was similar and a parametric study was done to assess the impact of water addition in the Ansys Fluent.

3.1 Parameter formation:

In the Fluent while assigning Boundary conditions following parameters were created:

• At the Fuel inlet :(1) mass fraction of CH4 and (2) mass fraction of water were created as input parameters.
• Lines were created in the Fluent Post Processing tab with coordinate values as provided in the table 2.4.1.1.
• The maximum values of NOx and Soot mass fraction were made into output parameters on these lines.
• A total of five cases were solved with water content from 5% to 30 % in the fuel.

3.2 results and discussion:

Fig 3.2.1: The output of the parametric study (P1 and P2 are input parameters while the rest are the output parameters)

The data from the above parametric study was exported to excel and charts were plotted to get an idea of the trend.

Table 3.2.1: The exported data set

Chart 3.2.1: Plot of the maximum value of NOx against the percentage of water added at various probe lines.

Chart 3.2.2: Plot of the maximum value of Soot mass fraction against the percentage of water added at different probe lines.

3.3 Inferences

• Based on the above two plots it can be concluded that at 20% water content the NOX production is at peak and at 10% water addition the soot mass fraction is maximum.
• Although the range of 15-25% is having the least soot production, but the mass fraction of NOx is very high in this range.
• Based on both the above plots the range of up to slightly above 5% seems to be favorable if we consider the NOx as well as the soot production for a less polluting combustor.

4.Conclusion

• This study was a very simplified one although it is capable of providing a basic understanding of combustion simulation in the Ansys Fluent.
• A more detailed study would involve the use of proper thermodynamic data file and mechanism files with appropriate data which requires a more detailed study of the involved chemical kinetics.