Turbulent Combustion Modelling with Conditional Moment Closure

monkeyresultMechanics

Feb 22, 2014 (3 years and 1 month ago)

46 views

introduction
Due to their high effi ciency, Diesel engines
are widely used in power generation, heavy-
duty off-road application, passenger and freight
transportation as well as ship propulsion. In
addition to the traditional development goals
- high reliability, fuel effi ciency and power density
- modern Diesel engines need to comply with
increasingly stringent emission legislations. Some
of these development targets exhibit a trade-
off behaviour and therfore require a complex
optimisation process. New development tools
and approaches are called for to help reduce the
number of tests and thereby development time
and costs [1].
Computational Fluid Dynamics (CFD) has
been used extensively in the past to investigate
the phenomena occurring in the combustion
of liquid fuels. Due to the complexity of the
processes in sprays undergoing atomisation and
secondary break-up, evaporation and subsequently
chemical reactions, a high degree of modelling and
simplifi cation of the underlying physical processes
is is necessary. Many existing models have known
limitations due to their ‚over-simplification‘
primarily dictated by the available computational
resources. Therefore, substantial calibration of
model constants and validation with experimental
data is required and the range of validity of the
predictions in general limited.
Conditional moment closure
Models in presently available CFD software
for the simulation of non-premixed combustion
are generally based on characteristic time-scales,
i.e. eddy break-up concepts and employ a one-
step irreversible reaction for the description of
the fuel oxidation and product formation. For
accurate predictions of pollutant formation during
hydro-cabon combustion however, an improved
description of the chemical processes occuring at
very different time-scales is required.
The methodology employed is based on the
Conditional Moment Closure concept developped
by [2] which belongs to the so-called presumed
probability density function (PDF) approaches.
The model enables the usage of arbitrarily
complex kinetics and parametrises the infl uence
of turbulence on the chemistry. Figure 2 shows
a schematic representation of the fully coupled
solution procedure of the two interfaced codes.
Results
Model validation is performed by means of
experimental data from two well documented
high pressure high temperature spray combustion
chambers. The fi rst system is an constant pressure
open chamber at Diesel engine relevant conditions.
The data set from[4] includes spray penetration
and ignition delays for n-Heptane at different
temperatures with and without initial turbulence
in the surrounding air for which a comparison is
presented in fi gure 3. The importance of the spatial
transport terms in in the CMC equations during
the ignition phase and, in particular, the fl ame
propagation phase was demonstrated in [5].
July 2006
Turbulent Combustion Modelling
with Conditional Moment Closure
Author: Yuri M. Wright
Turbulent Combustion Modelling
Fig. 1: hydrocarbon oxidation time-scales
0 0.1 0.2 0.3 0.4
η [−]
0
5
10
15
20
β−PDF [−]

2
,k,,
,
P
P,
t
 

 



 

 
n
i i
i 1
h h T Y





STAR-CD
CMC code
solve flow field
and compute CMC
“parameters”:
 



2
P P,,

    


solve species and enthalpy
equations in physical and
conserved scalar space
   
1
0
i i
Y Y P d

 

 
return mean values
by weighting with
presumed ȕ-PDF
0 0.1 0.2 0.3 0.4
η [−]
250
500
750
1000
1250
1500
1750
temperature [K]
fuel
(C
x
H
y
)
small molecules
C
2
H
2
, C
3
H
6
radical pool
T, Y
i
t
H
2
O
CO
2
T
Fig. 2: Interfacing CFD-CMC [3]
A further validation study was performed [6]
for the closed high pressure temperature cell of
the Aerothermochemistry and Combustion Sys-
tem Lab. at ETH. The extensive experimental data
set comprises pressure traces for the validation
of ignition delay and heat release predictions for
the various stages of the combustion as shown
in fi gure 4.
By means of Schlieren, Mie scattering and
chemilunimescence imaging data the temporal
evolution of the spray penetration and the ignition
location can be verifi ed as can be seen from fi gures
5 and 6.
References
[1] Weisser, G.; Schulz, R.; Wright, Y.M.; Boulouchos, K.:
Progress in Computational Fluid Dynamics (CFD)
Applications for Large Diesel Engine Development,
CIMAC 2004
[2] Klimenko, A.Y.; Bilger, R.W.: Conditional moment
closure for turbulent combustion, Progress in
Energy and Combustion Science 25, 1999
[3] Mastorakos, E.; Wright, Y.M.: Simulations of
Turbulent Spray Auto-ignition with Elliptic
Conditional Moment Closure, Proceedings of the
European Combustion Meeting 2003
[4] Koss, H.J.; Brüggemann, D.; Wiartalla, A.; Bäcker,
H.; Breuer, A.: Investigations of the Infl uence of
Turbulence and Fuel Type on the Evaporation and
Mixture Formation in Fuel Sprays, Final Report of
Joule Project on IDEA, 1992
[5] Wri ght, Y.M., De Paol a, G., Mastorakos, E.,
Boulouchos, K.: Simulations of spray auto-ignition
and fl ame establishment with two-dimensional
Conditional Moment Closure, Combustion and
Flame 143, 2005
[7] Wright, Y.M.: Numerical investigation of turbulent
spray combustion with Conditional Moment
Closure, PhD Thesis No. 16386, ETH Zürich, 2005
Contact
Dr. Yuri M. Wright
LAV - Institut fuer Energietechnik
ETH Zurich
CH-8092 Zürich
tel: +41 44 632 46 16
email: wright@lav.mavt.ethz.ch
Project Partners
Dr. E. Mastorakos
University of Cambridge, UK
Wärtsilä Switzerland Ltd.
Swiss Innovation Promotion Agency
Swiss Federal Offi ce of Energy




780 785 790 795 800 805 810 815 820 825
Temperature [K] ?
1.1
1.3
1.5
1.7
1.9
2.1
ignition delays [ms]
Experiment
MOL
OS − VODPK/CHEMEQ2 (Δt=1.0E−7s)
Fig. 3: Measured vs. predicted ignition delays for
method of lines (MOL) and operator split-
ting (OS), without (upper) and with (lower)
initial air turbulence
780 785 790 795 800 805 810 815 820 825
Temperature [K] ?
1.2
1.4
1.6
1.8
2
2.2
2.4
ignition delays [ms]
Experiment
MOL
OS − VODPK/CHEMEQ2 (Δt=1.0E−7s)
Fig. 5: Two-phase flow-field: Schlieren images
(lower), simulated fuel vapour distribution
with overlaid droplets (upper)
Fig. 4: pressure rise, simulation vs. experiment
0 0.001 0.002 0.003 0.004
time [ms]
79.75
80
80.25
80.5
80.75
81
81.25
pressure [bar]
Fig. 6: Ignition location: chemiluminescence si-
gnal (left), OH radical (right) at the time of
ignition