Anda di halaman 1dari 2

Flame Propagation and Knock

September 2006

Numerical and Experimental Study of Flame Propagation and Knock in a Compressed Natural Gas Engine
Author: Lmmle Christian

Background

One of the major objectives during the development process of new products is to reduce costs and time to market. Increasing computational power and continuous improvements of models for internal combustion engine applications show promise with respect to replacement of some optimisation steps by computer simulations. A prerequisite for such a substitution is that trends can be reasonably predicted and that calculations adequately incorporate the physics. The flame propagation and the knock behaviour of compressed natural gas engines have been studied in the present work. The aim is to improve the physical understanding on one hand and to develop physically based models for cycle simulation tools on the other hand. The flame propagation model has finally been verified using 3D-CRFD (computational reactive fluid dynamics) tools. These models and simulation tools have been used to optimise a new engine concept which combines ultra-low emissions, high efficiency and driveability where various driving strategies considering compression ratio, turbocharging and exhaust gas recirculation have been tested as shown in figure 1.

Fig. 2: Visualisation of the flame propagation the phenomenological combustion model.

turbulence. Furthermore, a turbulence model is applied, accounting for the turbulence generation due to the intake and squish flow and the change of turbulence intensity due to density changes and dissipation. The geometrical properties of the combustion chamber are described by a flame front area function. The flame propagation model was used to interpret results obtained at the engine test bench and to predict different operating conditions and strategies. It was found that the flame impinges the piston early in the combustion period as the compression ratio was increased (see figure 2). This result was validated through 3D-computations as shown in figure 3. Furthermore, the burn rate was

Fig. 1: Typical optimisation task in the engine development process.

Flame propagation model

Fig. 3: Visualisation of the flame propagation using 3D-CRFD.

The turbulent flame propagation is described in a phenomenological way, including submodels for the laminar and turbulent flame speed considering in-cylinder pressure, temperature, amount of exhaust gases, equivalence ratio and

predicted and therefore the overall engine performance has been analysed. As a result, figure 4 compares the crank angles of 50% burned for the experiments and the computations.

25 Crank angle at 50% burned [CA ATDC] 20 15 10 5 0 5 0

Experiment Simulation

around 3% accuracy. The crank angles of knock occurrence can be predicted well for all 20 operating points. Similar results have been achieved for the other gas compositions.
CA of knock onset simulation [CA ATDC] 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 CA of knock onset experiment [CA ATDC] 18
2

Identification Verification R =0.36497

Fig. 4: Experimentally determined and calculated crank angle at 50% burned for 30 operating points.

10 15 20 Number of operating point

25

30

Knock model approach

The knock phenomena in spark ignition engines results from the self-ignition of part of the unburned gas ahead of the propagating flame front. The engine can be damaged due to high cylinder pressure rise and pressure waves if knock occurs. Research and Development activities often focus on increasing the efficiency of spark ignited engines, but many modifications leading to higher engine efficiency in part load operation lead to higher risk of knock occurrence at full load. These contradictory requirements clearly indicate the necessity of accurate physical formulations of knock phenomena. In the present work a knock model dedicated to compressed natural gas engine applications was developed. The model is based on a one step chemistry approach leading to the so called knock integral method. Due to the varying gas composition of compressed natural gas five well-defined compositions of synthetic gases which are listed in table 1 - have been tested to investigate the influence of the individual components. Furthermore, the model considers different operating conditions of an engine.
G20 G4E G8E G44 G25 Component Methane Ethane Propane Nirogen Unit Vol.-% / Mass-% Vol.-% / Mass-% Vol.-% / Mass-% Vol.-% / Mass-% 100 / 100 ---96 / 92.76 4 / 7.24 --92 / 85.99 8 / 14.01 --92 / 83.26 4 / 6.79 4 / 9.95 -86 / 77.86 --14 / 22.14

Fig. 5: Final results for the predicted crank angle of knock occurrence and comparison with experimentally determined crank angle of knock onset (gas: 100% methane).

Conclusions

A typical optimisation task in the engine development process with concern to new IC engine concepts was presented in the beginning of this work. The complexity and requirements will further increase in the future. Experimental work combined with computations will significantly accelerate the development process and support the understanding of the results obtained. Simulations therefore should not be interpreted as a competing but as a complementary process to measurements and are necessary to reduce costs and time to market.

Award

In October 2006 the CEV-project in which these models have been applied won the German Innovationspreis der deutschen Gaswirtschaft 2006.

References

Tab. 1: Gas compositions used for the knock model development process.

Figure 5 compares the computed crank angles of knock occurrence with the experimentally determined ones. The dashed lines indicate an arbitrary range, meaning that the crank angles are within

[1] Christian Laemmle, Konstantinos Boulouchos, Christian Bach: Numerical and Experimental Study of Knock Behaviour in a Turbocharged Compressed Natural Gas Engine, International Journal of Engine Research (review process) [2] Christian Lmmle, Konstantinos Boulouchos, Christian Bach: Prediction and Interpretation of Combustion Processes in Natural Gas Engines - A comparative Overview of Simulation Methods for Practical Applications, 4. Dessauer Gasmotorenkonferenz, Juni 2005

Contact

Dr. Christian Lmmle LAV - Institut fuer Energietechnik ETH Zurich CH-8092 Zurich tel: +41 44 632 46 18 email: laemmle@lav.mavt.ethz.ch

Swiss Federal Laboratories for Materials Testing and Research, EMPA Dbendorf Volkswagen Bosch Corning Engelhard Swiss Gas and Water Industry Association SVGW German Technical and Scientific Association for Gas and Water DVGW Austrian Gas and Water Industry Association VGW Swiss Federal Office of Energy (BFE)

Project Partners