You are on page 1of 10

Oxygen Methane Combustion Simulation Towards Pollution Free Solution For Industry and Transportation

Arni Steingrimsson and Reda Mankbadi


Embry-Riddle Aeronautical University, Daytona Beach, Florida, 32114, USA

Computational uid dynamic combustion simulation of pure methane, CH4 , and pure oxygen, O2 was laid out to investigate the eects of dierent turbulence models, combustion models, advection schemes, average time stepping, time accurate time stepping and mesh sizing. All the results were compared with empirical data for validation and seem to agree quite well. The results show that steady state, lower delity turbulence model is able to show comparable result to high delity turbulence model with time accurate time stepping. Pulsating injection combustion of methane and oxygen was simulated as well and the results compared with non-pulsating combustion. Results show that there is a performance penalty in the combustor, mainly due to pressure drop in the inlet, while C eciency is comparable to non-pulsating combustion.

Nomenclature
m m CH4 m O2 dA
S

Mass Flow Rate Methane Mass Flow Rate Amplitude Oxygen Mass Flow Rate Amplitude Integrated Surface Area Density Frequency Heaviside Step Function Simulation Time Fluid Velocity

f step t U

I.

Introduction

ith more evidence being released for global warming and growing acceptance of the severity and the W impact it will have on our lives it is crucial that big steps are taken to reduce emissions that cause global warming. Electricity, heating and transportation account for almost 60% of carbon dioxide (CO )
2

gas emissions in the us. The biggest pollutants are road based transportation with 21.6% while residence buildings account for 15.3% and 12% for the commercial buildings, the values account for both heating and electric usage. The purpose of this study is to design a pollution free system that can be used in transportation and energy production. The system consist mainly of combusting pure methane and pure oxygen. Water is injected into combustion chamber and super heated steam from combustor enters steam turbine which extracts energy out of the exhaust. The steam turbine spins an electric generator which charges a super capacitor that stores the energy. High level view or the core components of the system is shown in gure 2.
Engineer, Siemens Energy, and AIAA Member. Professor of Aerospace Engineering, Embry-Riddle Aeronautical University, Daytona Beach FL 32114, Florida Center for Advanced Aero-Propulsion Technology (FCAAP) and AIAA Fellow.
Distinguish Aerospace

1 of 10 American Institute of Aeronautics and Astronautics

A more detailed system diagram is shown in gure 3 which shows how a system could be used in ships and boats. What is dierent about this system compared with e.g. diesel generators in some ships today is that there is no greenhouse emission. Once the steam turbine has extracted energy out of the steam the exhaust (H2 O and CO2 ) is cooled through condenser and the two chemical compounds are separated. The separated compounds, H2 O and CO2 , are stored and the water is reused. One of the dicult tasks in this study is to design the combustion chamber. The main task of this study is to investigate the combustion chamber and its combustion behavior. In this study the use of computational uid dynamics was used entirely to investigate the combustion behavior. An empirical data from Salgues et. al.2 was used to validate simulation model and simulation setups. There are limited empirical and simulation data available for combustion of pure methane and pure oxygen and thus CFD simulation was tted to match Salgues et. al.2 empirical experiment. In addition to match empirical data and validate the CFD models a pulsating combustion was studied. Not to be confused with pulse combustion like the known Schmidt, Helmholtz and Rijke type combustor, the pulsating simulation injects fuel and oxidizer in pulses into the combustion chamber. No available data is on this subject thus simulation used the same geometry and models for the pulsating combustion as for the CFD validation case. The ultimate goal is to design very ecient power system that is pollution free.
H2O Tank/Reservoir CH4 Tank

CO2 Tank Condenser

CH4
O2 Tank

Rocket Combustor Steam Turbine Electric Generator

Rocket Combustor O2 Steam Turbine

Electric Generator
Ship Screw Electric Motor Electric Component Super Capacitor

Figure 2: Simple process ow diagram of the core component

Figure 3: Example of ship drive train process design

II.
A. The Validation

Model

Salgues et. al. performed combustion experiments using pure methane and pure oxygen. Their main results are listed in table 4. Other experiments were performed using imaging technology, such as shadowgraph. The geometry used for this study is shown is shown in gure 4 and gure 5 shows the cross sectional sketch of the combustor that Salgues et. al. used.

2 of 10 American Institute of Aeronautics and Astronautics

Figure 1: US greenhouse gas emission ow chart1 The geometry is a rocket type combustor with two inlets, cylinder shaped at the center axis is the oxygen inlet and an annulus inlet for the methane. The combustion chamber is a straight cylinder and its outlet 292.1 mm consist of convergence-divergence sections. Two meshes were created for the geometry, one course another ne. The ne mesh contained boundary layer but is comparible to Boundiers et. al.3 ne LES mesh, where they performed mesh sensitivity study in complex combustor geometry. Typical LES mesh sizes with wall model were used for the ne mesh, see table 2 for typical values. Boudiers et. al. mesh sensitivity study Figure 4: Geometry created ushas shown that in complex geometry the mesh size plays a mayer role in ing Siemens NX showing comcapturing all physics and producing accurate results, comparison of the bustion chamber dimension meshes are shown in table 1. Saluges et. al. measured the massow rate for the methane and the oxygen very close to the inlet and is shown in the table 4. The boundary conditions for the combustion model are shown in table 3. 8 dierent combustion simulations were performed. Each simulation had a unique CFD setup. The CFD simulation setups are shown in table 5 and 6. Simulations were both time accurate as well as time averaged and two turbulence models were tested one for each time model. For the time average simulations k turbulence model with Eddy dissipation and thermal energy heat transfer model was used. Both 1st order and second order advection scheme was used and also for the turbulence numerics. For the time Figure 5: Original schematic of accurate simulation two turbulence models were used, U RAN Sk and WALE-LES. Finite rate chemistry and eddy dissipation combustion modthe validation combustor2 els were used. Total energy and thermal energy heat transfer were also tested. 3 advection schemes were tested, high resolution, specied blend factor and central dierence. 3 dierent transient schemes were simulated as well, high resolution, 1st and 2nd order backward euler. Course and ne mesh was then tested for both the time accurate simulations and the time average simulation. B. Pulsating Simulations
(a) Consine wave pulse
0.05 Mass Flow Rate 0.04 0.03 0.02 0.01 0 -0.01 0 5e-005 0.0001 0.00015 0.0002 Frequency = 10,000

The other task was to study the aect of pulsating combustion. Like mentioned before this simulation test was dierent than the known pulse combustion, Schmidt, Helmholtz and Rijke type. The re-

0.05 Mass Flow Rate 0.04 0.03 0.02 0.01 0 -0.01 0 5e-005 0.0001 0.00015 0.0002 Frequency = 50,000

Mass Flow Rate

3 of 10

0.05 0.04 0.03 0.02 0.01 0 -0.01 0 5e-005 0.0001 Seconds 0.00015 0.0002 Frequency = 100,000

American Institute of Aeronautics and Astronautics


0.05 e 0.04

(b) Square wave pulse

Total num. of points Total num. of cells Max. cell volume Min. cell volume

Coarse mesh 230,118 1,242,086 3.12671 108 1.81795 1011

Fine mesh 7,661,005 43,949,682 4.05748 109 1.1828 1012

My RANS mesh 153,761 1,514,128 4.79152 109 6.4655 1013

My LES mesh 799,983 8,787,466 5.81999 1010 3.04378 1015

Table 1: Comparison of mesh sizes3


DNS 10-15 5 1 3-5 Wall-Resolved LES 50-150 10-40 1 3-5 LES with Wall Model 100-600 100-300 30-150 -

Streamwise x+ Spanwise z + Wall-normal (y + ) Number of points in 0 < y + < 10

Table 2: Typical LES mesh sizes4

sults from the pulsating combustion were compared with the results from the validation simulations thus the simulation setups and models were the same as for the validation. To make the combustion pulsate the fuel and the oxidizer were injected sequentially. To do this the ANSYS CFX boundary conditions were modied. For the validation simulations the inlet boundary conditions are shown in equation 1, where is the density of the uid, U is the velocity, m is the mass ow rate and dA is the integrated
S

inlet area. This equation was then modied to include cosine function and step function which are a function of simulation time t and frequency f . The amplitude is set by the m CH4 and m O2 . Step function, also called Heaviside Step Function, was also implemented, see equation 4 and 5. The step function turns all positive values into 1 and all negative values into 0. The function was then multiplied by the mass ow rate amplitude, m CH4 and m O2 U =
S

m dA

(1)

m = m =

cos(2f t) + 1 m CH4 2 cos(2f t) + 1 m O2 2

(2) (3) (4) (5)

m = step(cos(2f t)) m CH4 m = step(cos(2f t)) m O2

At one point in time the fuel and the oxidizer are injected with full mass ow rate and at another point in time no fuel and oxidizer is injected. The method of injection was the use of wave function and both square wave and cosine wave were simulated. Each function had a frequency parameter and amplitude parameter. The frequency parameter had a value of 10, 000Hz , 50, 000Hz and 100, 000Hz . The 3 frequencies were tested on both the fuel and the oxidizer. The amplitude was set to 0.03923kg/s and 0.1177kg/s for the fuel and the oxidizer respectfully. There were total of 6 simulations performed for the pulsating study and the injection procedure can be seen in gure 7.
4 of 10 American Institute of Aeronautics and Astronautics

Boundary Inlet Oxygen Inlet Oxygen Inlet Methane Inlet Methane Outlet Outlet Wall

Parameter Mass ow (kg/s) Temperature (K ) Mass ow (kg/s) Temperature (K ) Pressure (P a) Temperature (K ) Smooth no slip

Value 0.1177 122 0.03923 300 101325 300

Type of injector Shear Coaxial Shear Coaxial Swirl Coaxial #1 Swirl Coaxial #2 Shear Coaxial Swirl Coaxial #1 Swirl Coaxial #1 Shear Coaxial

Type of measurement LOX mass ow (kg/s) CH4 mass ow (kg/s) Chamber Pressure (M P a) Chamber Pressure (M P a) Chamber Pressure (M P a) Cexp /Ctheo. (%) Cexp /Ctheo. (%) Cexp /Ctheo. (%)

Values 0.118 0.039 4.23 4.03 3.82 95 91 85

Table 3: Model boundary conditions

Table 4: Validation parameters obtained from Salgues et. al.2


Identication Advection Scheme Specied Blend Factor High Resolution High Resolution High Resolution Specied Blend Factor Central Dierence Specied Blend Factor Specied Blend Factor Turbulence Numerics / Transient Scheme 1st Order High Resolution High Resolution High Resolution 1st Order Backward Euler 2nd Order Backward Euler 1st Order Backward Euler 2nd Order Backward Euler Mesh Course Course Fine Course Course Course Fine Fine mesh mesh mesh mesh mesh mesh mesh mesh

Identication Sim V1 Sim V2 Sim V3 Sim V4 Sim V5 Sim V6 Sim V7 Sim V8

Turbulence Model k model k model k model k URANS model WALE-LES model WALE-LES model WALE-LES model WALE-LES model

Combustion Model Eddy Dissipation Eddy Dissipation Eddy Dissipation Eddy Dissipation Finite Rate Chemistry Finite Rate Chemistry Finite Rate Chemistry Finite Rate Chemistry

Heat Transfer Thermal Energy Thermal Energy Thermal Energy Thermal Energy Total Energy Total Energy Total Energy Total Energy

Sim Sim Sim Sim Sim Sim Sim Sim

V1 V2 V3 V4 V5 V6 V7 V8

Table 5: Simulation identication and solver settings 1

Table 6: Simulation identication and solver settings 2

III.

Results

There were two dierent experiments studied, validation simulation and pulsating simulation. A. Validation

8 dierent simulations were performed for validation. Each simulation had dierent setting and they can be seen in table 5 and in table 6. Oxygen combustion is very volatile and almost immediately after the fuel and the oxygen enters the combustion chamber the combustion occurs. The maximum temperature occurs along the chamber axis immediately after the combustion. The comparison of the average chamber temperature is shown in gure 6. The Sim V3, Sim V7 and Sim V8 all use the ne mesh and show the highest average chamber temperature. The average chamber pressure results are shown in gure 7. The empirical results are shown with black horizontal line and all the simulation results show higher pressure. Advection scheme seems to be the biggest factor for the pressure results as can be seen from the results, see Sim V1, Sim V5, Sim V7 and Sim V8. Combustion chamber performance measures are listed in gure 8 and in gure 9 with specic impulse and thrust specic fuel consumption. Combination of advection scheme and mesh size seems to contribute to the results and the use of second order scheme with courser mesh results in higher specic impulse. Another performance and eciency measure for combustion is the characteristic exhaust velocity eciency and its results are shown in gure 10. Again the advection scheme and mesh size are the biggest factors for the results. The turbulence structure inside the combustion chamber is shown in gure 11 and the eddy viscosity were computed by the WALE-LES turbulence model.
4,17 4 4,15 3 10 6
Chamber Temperature (K)

10 6

4.2 Chamber Pressure (Pa) 4.1 4

10 6

106

4,04 1

4.09

10 6

4.09

4.07

10 6

4,000

4.03

4.02

4.05

10 6

3,68 6

3,68 2

10 6

3,63 0

3,64 0

3,65 0

Theoretical 3.9 Experiment 3.8 Sim V1 Sim V2 Sim V3 Sim V4 Sim V5 Sim V6

3,500 Theoretical Temperature Sim V1 Sim V2 Sim V3 Sim V4 Sim V5 Sim V6 Sim V7 Sim V8

Sim V7

Sim V8

Figure 6: Comparison of combustion chamber temperature

Figure 7: Comparison of combustion chamber pressure

5 of 10 American Institute of Aeronautics and Astronautics

3.92 1

3.93

06

1.2

1.18

1.18

1.18

1.18

.96

316 .6

Specic impulse, Isp (s)

.54

.15

.86

312

311

311

310

312

315 310 305 300

TSFC

.82

.54

.01

1.14

1.12

1.1 Sim V1 Sim V2 Sim V3 Sim V4 Sim V5 Sim V6 Sim V7 Sim V8 Sim V1 Sim V2 Sim V3 Sim V4 Sim V5 Sim V6 Sim V7 Sim V8

Figure 8: Comparison of combustion chamber specic impulse

Figure 9: Comparison of combustion chamber thrust specic fuel consumption

100
CExp /CT heo (%)

91.5

95

90.4 7

89.6

89.4

89.2

90.2

85 Sim V1 Sim V2 Sim V3

Experiment Sim V4 Sim V5 Sim V6 Sim V7 Sim V8

87.2

90

87.2 8

Figure 10: Comparison of characteristic exhaust velocity C*

Figure 11: Turbulence structure from the inside of the combustion chamber (Qcriterion) as computed by WALE-LES

B.

Pulsating Simulation

The pulsating results for the pressure inside the combustion chamber are shown in gure 12. From the results the inlet the pressure osculates back and forth from lower to higher pressure and as we increase the frequency the amplitude of the wave gets smaller. After combustion the pressure levels o at about 2M P a. During combustion the oxygen is fully combusted about 1/4 way through the combustion chamber as can be seen in gure 13. The methane mass fraction results are shown in gure 14 where there is about 5 or 6 % of the methane unburnt, this was expected as the mixture was slightly rich. The turbulence structure and eddy viscosity is larger for the lower frequency than the higher as can be seen in gure 15. Figure 16 shows a contour plot of the eddy viscosity and the turbulence structure (q-criterion) as computed by the turbulence model. The plot is only showing the inlet and the small section of the combustion chamber. The temperature is only aected by the pulsation only at the inlet and before the combustion. After the combustion the temperature evens out and all the simulations result at around 4500K . Finally table 8 lists average chamber pressure, average chamber temperature, characteristic exhaust velocity eciency and the specic impulse of all the pulsating simulations and the validation which had the same simulation settings. The results show that the pulsation does not aect the temperature much and the characteristic exhaust velocity. The pressure drop on the other hand tremendous and the average chamber pressure is half of the non pulsating cases. The specic impulse is much lower for the pulsating cases than for the non-pulsating case.

IV.

Conclusion

Validation of uid dynamics and combustion model was compared with Salgues et al.2 empirical results. Not all dimensions were available to design the combustor to be identical to the empirical data and thus some educated guesses had to be included in this study. Because this uid dynamic study and the empirical test were performed completely independently of each other the results for the empirical data were limited. The results show that all the simulations slightly over predict the theoretical chamber temperature. This can be expected when simulating combustion simulations using equation of state as ideal instead of real

6 of 10 American Institute of Aeronautics and Astronautics

1.15

314

1.16

1.16

320

318

1.17

1.18

1.18

4e+006 3.5e+006 3e+006 2.5e+006 2e+006 1.5e+006 1e+006 500000 0 -500000 -0.05

10000 Hz Pulse 10000 Hz Step 50000 Hz Pulse 50000 Hz Step 100000 Hz Pulse 100000 Hz Step

URANS URANS URANS URANS URANS URANS

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 12: Combustion chamber pressure comparison of the pulsating simulations (results given in Pascals) Parameter Non Pulsating Pulse 10, 000Hz Step 10, 000Hz Pulse 50, 000Hz Step 50, 000Hz Pulse 100, 000Hz Step 100, 000Hz Avg Chamber Temperature 3686.26K 3699.21K 3715.25K 3738.83K 3695.56K 3691.10K 3680.79K Avg Chamber Pressure 4015281.75P a 2090637.63P a 2104722.5P a 2079055.25P a 2064676.5P a 2000926.63P a 1994451.88P a % C 89.23% 92.06% 90.12% 89.81% 89.95% 89.15% 89.07% ISP 310.86s 284.08s 283.77s 286.88s 284.89s 283.37s 283.53s

Table 8: Comparison of pulsating combustion results and non-pulsating results

gas. The results for the chamber pressure showed that some simulations were able to predict pressure under the theoretical chamber pressure although all simulations were slightly over-predicting the empirical chamber pressure. It is known that usage of ideal gas for equation of state will cause simulations to over predict temperature and pressure.5 Another factor is that the geometric details were not know, such as the converging diverging exit nozzle and the inlet pipe length etc. The mesh size seemed to inuence the chamber temperature results signicantly, it did not have the same trend on the chamber pressure. The mesh size did not have much or any change on the RANS simulation chamber pressure. For the LES simulations the mesh size reduced the chamber pressure and the results were under the theoretical chamber pressure. Using the specic blend factor advection scheme seemed to inuence the results more than changing the transient model from 1st to 2nd order as can be seen from results. The Sim V7 and Sim V8 which compare 1st order transient model with 2nd order show very little dierence as can be seen in all the results. Sim V1 and Sim V5 seem to agree very closely with each other even though they have very dierent solver settings. Sim V1 is time averaged and uses lower delity turbulence model while Sim V5 is time accurate and uses LES turbulence model. They both run on course mesh and this could be the reason for them agreeing so closely with each other. The mesh is not small enough for the LES model to capture the smaller eddies and thus computing similar results as the k model. If we then compare Sim V3 and Sim V8 the results show bigger dierence and especially in turbulence structure. This comparison should not be taken fervently as turbulence structure is usually not computed for steady state solutions, only time dependent simulations. Pulsating simulations with injection fuel and oxygen alternating either as cosine wave or step function at specic frequency was performed and compared with validated simulations above. The pulsating simulations
7 of 10 American Institute of Aeronautics and Astronautics

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.05

10000 Hz Pulse 10000 Hz Step 50000 Hz Pulse 50000 Hz Step 100000 Hz Pulse 100000 Hz Step

URANS URANS URANS URANS URANS URANS

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 13: Oxygen mass fraction, ratio of oxygen mass to the mass of the total mixture.

were performed using the same simulation setup as Sim V4 and the results show that there is a drop in eciency by pulsating the injection. This is mainly due to the pressure drop in the combustion chamber. Even though characteristic exhaust velocity eciency is very close to the non-pulsating simulations the results show that TSFC and Isp are lower than non-pulsating simulations. It could be feasible, at least with simulation, to increase the injection pressure and see the results of changing the amplitude. This causes another problem though as the injection pressure is already close to the supercritical state of the methane and oxygen. In this study both k turbulence model and LES turbulence model have been validated within acceptable limit. Results have shown that course mesh and k turbulence model will be able to predict acceptable results and much faster than LES with ne mesh. The main reason for this could be that the geometry is relatively simple and it is well known that LES is able to predict very complex geometries while lower delity turbulence models cannot.6 The k model gives similar results to the LES but the ow eld is averaged and does not show the correct ow eld when compared with combustion images such as shawdowgraph or OH-PLIF images.

References
1 Baumert, K. A., Herzog, T., and Pershing, J., Navigating the Numbers Greenhouse Gas Data and International Climate Policy, Tech. rep., World Resources Institude, 2005. 2 Salgues, D., Mouis, G., Lee, S.-Y., Kalitan, D. M., Pal, S., and Santoro, R., Shear and Swirl Coaxial Injector Studies of LOX/GCH4 Rocket Combustion Using Non-Intrusive Laser Diagnostics, American Institude of Aeronautics and Astronautics , , No. 1, January 2006, pp. 114. 3 Boudier, G., Gicquel, L. Y. M., and Poinsot, T. J., Eects of mesh resolution on Large Eddy Simulation of reacting ows in complex geometry combustors, Combustion and Flame , Vol. 155, No. 1-2, October 2008, pp. 196214. 4 Jiang, X. and Lai, C.-H., Numerical Techniques for Direct and Large-Eddy Simulations , Taylor and Francis Group, LLC, 2009. 5 Cutrone, L., Battista, F., Ranuzzi, G., Bonifacio, S., and Steelant, J., A CFD Method for Simulation of Mixing and Combustion in High-Pressure LOx/Methane Rocket Engines, American Institute of Aeronautics and Astronautics , , No. AIAA 2008-949, January 2008. 6 Jiang, X. and Lai, C.-H., Numerical Techniques for Direct and Large-Eddy Simulations , CRC Press Taylor Francis Group, 2009.

8 of 10 American Institute of Aeronautics and Astronautics

0.07

0.06

10000 Hz Pulse 10000 Hz Step 50000 Hz Pulse 50000 Hz Step 100000 Hz Pulse 100000 Hz Step

URANS URANS URANS URANS URANS URANS

0.05

0.04

0.03

0.02

0.01

0 -0.05

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 14: Methane mass fraction, ratio of methane mass to the mass of the total mixture.

0.35

0.3

10000 Hz Pulse 10000 Hz Step 50000 Hz Pulse 50000 Hz Step 100000 Hz Pulse 100000 Hz Step

URANS URANS URANS URANS URANS URANS

0.25

0.2

0.15

0.1

0.05

0 -0.05

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 15: Comparison of eddy viscosity for all the pulsating simulations

9 of 10 American Institute of Aeronautics and Astronautics

(a) 10, 000Hz cosine wave

(b) 50, 000Hz cosine wave

(c) 100, 000Hz cosine wave

Figure 16: Contour plot of the eddy viscosity of the pulsating simulation using cosine wave. The structure is the turbulence structure (q-criterion) as computed by the turbulence model

6000

5000

10000 Hz Pulse 10000 Hz Step 50000 Hz Pulse 50000 Hz Step 100000 Hz Pulse 100000 Hz Step

URANS URANS URANS URANS URANS URANS

4000

3000

2000

1000

0 -0.05

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 17: Combustion chamber temperature results for all the pulsating simulations

10 of 10 American Institute of Aeronautics and Astronautics

You might also like