This article presents research at Colorado State University on retrofitting
the direct-injection natural gas engine. It is an example of a successful
engine development study using STAR-CD.
Large bore natural gas engines have been used for many years in stationary
applications such as gas compression and electric power generation. A common
configuration of the engines is a two-stroke cycle with direct injection
of natural gas into the cylinder. Developing successful retrofit technologies
to improve engine performance and to reduce pollutant generation has become
an important issue as more stringent air emission regulations are enacted.
Poor in-cylinder mixing due to ineffective fuel delivery is believed to be
problematic in these natural gas engines. In addition, cyclic combustion
instability due to slow early-flame-growth is one of key contributors to
NOX and CO emissions. The retrofit technologies discussed in this article
are categorized into two areas: ‘Fuel-air mixing enhancement’,
and ‘Alternative ignition systems’.
Fuel-air mixing enhancement
The specific engine modeled in this study, the GMV Cooper Bessemer engine,
is widely used in the gas compression industry, primarily in 10 cylinder
versions. It is a large engine, with a 38 cm bore and stroke. A computational
model, shown Fig. 1, incorporating a moving grid simulation of the scavenging,
compression, combustion, and expansion processes of the engine was developed
and validated through comparison with optical experimental results. The CFD
computations are compared with the PLIF (Planar Laser Induced Fluorescence)
results in Fig. 2. The PLIF images are on the right hand side, and the CFD
results are on the left hand side of the pairs of images. The scavenging
flow bends the fuel jet slightly toward the exhaust ports, so that the fuel
jet hits the piston top slightly off center, producing non-symmetric mixing
in the combustion chamber. These images indicate that the computations using
STAR-CD can model the in-cylinder flow induced by scavenging and the actual
injection and mixing processes quite accurately.
| Top Dead Center |
Bottom Dead Center |
 |
Fig 1. Computational mesh of natural gas fired 2 stroke engine |
 |
Fig 2. CFD validation with Planar Laser Induced Fluorescence experiments |
One of the promising mixing enhancement technologies is high-pressure fuel
injection. Natural gas is typically injected at low pressures, 1~3 bar
above manifold pressure. Since natural gas pipelines operate at pressures
of the order of 35 bar, it is of interest to explore the use of pipeline
gas at high pressure as the source for injected fuel. However, the cost
of three dimensional engine simulations becomes considerable if the computational
model has to capture the details of the complex supersonic flow structures
in and near the intricate geometry of the injection valve. So, a supersonic “virtual
valve” was designed for a 3D engine CFD model to reproduce the actual
downstream jet characteristics, which were of crucial importance in the
macroscopic engine performance. In Fig.3, the high-pressure fuel injection
and mixing during the compression stroke are compared with conventional
low pressure injection. The gridded dark colored regions in the figures
represents a mixture richer than the lean limit of flammability, f = 0.5.
The jet sweeps toward the intake ports, around the top of the cylinder
volume toward the opposite side of the cylinder, and along the outer edges
of the piston top. In the high-pressure injection case the flow patterns
are similar, but the fuel moves with higher momentum so that the most of
the volume in the cylinder is flammable at top dead center.
Low
Pressure Injection
|
High
Pressure Injection
|
|
|
Fig 3. Fuel Injection and Mixing comparison |
Alternative ignition systems
Lean burn combustion is a common solution for emissions reduction. However,
if the combustion occurs in a very lean regime or mixing is not sufficient,
CO and hydrocarbon emissions become unacceptably high due to ignition misfires.
PCC (Pre-combustion chamber) ignition and laser spark ignition are potential
retrofit technologies for obtaining stable ignition. Since a flame jet provides
the ignition in the main chamber, PCC ignition is less affected by lean regions
or poor mixing around the spark plug, resulting in greater stability of combustion.
With a laser-based system, the spark can be positioned at any location in
the cylinder.
Knowledge of the initial flow and concentration fields is required to determine
the subsequent flame propagation and pollutant generation during combustion.
Because of the non-homogeneity of the mixture system in the cylinder of this
type of engine, the spatial fuel distribution, mean flow field and turbulence
quantity field at ignition timing are critical factors determining the characteristics
of heat release during the expansion stroke.
Fig. 4 presents the equivalence ratio distribution at 10 degrees before TDC,
the ignition timing of the conventional spark plug. The location of the spark
plug is marked as SP in the figure. As seen in the figure, there is a lean
region on the spark plug side, and richer regions in the crevice around the
edge of the cylinder. Selected section plots including velocity magnitude
contours and vector plots are shown in Fig. 5. The scavenging induced cylinder
vortex dominates the cylinder flow pattern at ignition timing. An energetic
narrow bulk flow exists on the plane of symmetry flowing from the left to
right side toward the spark plug. This flow direction is opposite to the
direction of the flame propagation. On the other hand, the flow direction
in the crevice is in the same direction with the flame propagation.
 |
Fig 4. Fuel distribution at 10 degrees before TDC |
 |
Fig 5. Flow Field at 10 degrees before TDC |
Comparison of the flame propagation in conventional spark
ignition, PCC ignition, and laser spark ignition is presented
in Fig. 6. With spark ignition, the flame mainly propagates
along the rim in the azimuthal direction, not across the
center. The flame propagation during PCC ignition is quite
different. The flame propagates faster across the center
than in the azimuthal direction. In this case, the flame
jet overcomes the adverse flow field of the main chamber.
For the laser ignition computation, the laser spark location
chosen was slightly off-center to the intake side. The
flame initially propagates toward the chamber center, then
the flame front moves radially outward. The duration of
heat release is about 25 degrees of crank angle for the
PCC and laser spark ignition systems, and about 30 degrees
for the conventional spark case.
 |
Fig 6. Comparison of flame propagation |
In Fig. 7, the NO formation of the conventional spark ignition case and laser
spark ignition case are compared. Knowledge of the NO formation region
is important for NO reduction technologies such as water injection. With
the conventional spark ignition, NO is mainly formed in the crevice region
where the richer mixture burns at a higher temperature. On the other hand,
since the combustion in the laser spark ignition case starts at chamber
center, NO is mainly formed near the cylinder center.
 |
Fig 7. Comparison of NOX formation |
For further information, contact Gi-Heon Kim, kgheon@lamar.colostate.edu
http://lamar.colostate.edu/~kgheon
|
Your route to this page:
