Environmental Engineering Reference
In-Depth Information
con
gurations, simulations, and data are
given by Ajmani et al. (
2014
); here we will provide only the summary of the results
and future plans.
The RANS simulation of the 13 mixer modules (Fig.
43
) by using a tetrahedral
mesh with more than 17 million elements to resolve important features of the
geometry. A two-equation turbulence model was used, namely cubic k-
gurations 3, 7, and 9. Details on the con
ʵ
model
with variable C
along with dynamic wall functions with pressure gradient effects.
The governing equations for the liquid phase were solved based on a Lagrangian
formulation along with Jet-A fuel properties and turbulence transport effects. Initial
spray conditions came from measured data and empirical knowledge with the spray
divided into 10 droplet groups, discretized into 32 spatial streams along the 360
0
circumference. A compact 14 species with 18 chemical reaction steps for Jet-A fuel
combustion and NO
x
predictions was used along with the other transport equations.
The results shown in Fig.
45
have provided motivation to develop and calibrate the
next generation of computationally affordable kinetics mechanism with
ʼ
20 species
to be spearheaded by Prof. Fred Dryer based on his most recent outlook (Dryer
2014
).
The data and calculations in Fig.
45
cover a broad range of operating conditions,
viz. con
≤
ed as 3C1, 3C2, 3S1, and 3S2; here C and S are
abbreviations, respectively, for the cruise and sea-level takeoff operating conditions;
and 1, 2, and 3 as three fueling options as described. Similar short description given
as legends in Fig.
45
helps one to appreciate the highlights of con
guration 3 identi
gurations 7 (viz.
7C1, 7C2), 9 (viz. 9C1, 9C2, 9S1, 9S2), and several baseline (
(flat) dome operating
conditions listed as BC1, BC2, BC22A, BS1, BS1A, BS2, and BS2A. The range of
NO
x
data is between 2 and 8EI. Comparison between data and predictions can be
broadly be made by noticing that the ideal match is along the red line, followed by
two bounding lines of
fl
30 %. This level of
agreement is considered a reasonable start for the RANS simulations because
Kumud Ajmani plans to continue with the TFNS calculations with his current
kinetic scheme as well as with the improved kinetic scheme to be developed by
Prof. Fred Dryer in the coming years.
The author fully appreciates the critical importance of CFD in the design process
even when we have several examples of successfully implementing advanced
technology concepts through empirical means only, e.g.,
±
20 % accuracy and the outer band of
±
the CFM-TAPS
rst
engine demonstration, and the two LDI-2 con
gurations of Fig.
43
. He does not
need any convincing or discouragement from the summary given in Table
3
.
However in order to get the acceptance of the turbulent combustion models by the
“
(Table
3
), models should meet the accuracy metric derived from two
directions, namely the measurement repeatability (
nonbelievers
”
˃
) and semi-empirical correlation
˃
˃
levels. The average
values of the 128 samples from the ICAO engine emissions
-
database of the landing
takeoff hydrocarbon, CO, and NO
x
emissions (DP/F
00
) are,
respectively, 0.89, 2.61, and 1.33 g of pollutants/kN or 13.4, 5.0 and 2.8 % of the
average DP/F
00
. Therefore, the gas turbine prediction models
goal
proposed by Mongia has been: 3 % of takeoff EINO
x
, 7.5 and 15 %, respectively, of
idle CO and HC emission indices.
'
long-term
˃
Search WWH ::
Custom Search