Presented in summer heat transfer conference

1. RESEARCH POSTER PRESENTATION DESIGN © 2015
www.PosterPresentations.com
The effective use of coolant in maintaining gas turbine components below
failure limits is becoming increasingly necessary in view of increasing
combustor exit temperatures. A novel incremental impingement
configuration in a pin-finned coolant channel is studied; where
impingement jets are sheltered behind “cut-out” regions of the pin-fins is
studied. Due to the sheltering effect, the deflection of the jet by the cross-
flow is negligible. In the current work, a LES (Large Eddy Simulation)
study is conducted to model the flow for Re = 7500. LES is a well known
turbulence model which resolves the large scale anisotropic structures and
models the rest of the small scale isotropic ones. Two types of cooling
configuration are modeled and compared, one with the jet impinging in a
cut out region, similar to the experiment, and the other with the jet
impinging further downstream along the next row of pin-fins in the wake
region of the pin-fin. All other geometrical and flow parameters are kept
similar in both the configurations for one to one comparison. Since the
stagnation point Nusselt number is not affected by the cross-flow, the
configuration in which the jet impinges in the cut out region of the pin-
achieves more cooling as compared to the configuration in which the jet
impinges further downstream. Flow field and the thermal effectiveness
levels are shown to compare the cooling performance.
ABSTRACT
PROBLEM OBJECTIVE
A pin-fin section used in the experimental facility of Busche et. al. [1]
consists of two plates, a top plate and a bottom plate. The bottom
plate holds eight rows of pin-fins in a staggered fashion, and the top
plate has the impingement holes. The purpose of the top plate is to
impinge coolant into the pin-fin channel. The thickness of the top plate
is 0.25D where D (2.54 cm) is the pin-fin diameter. The coolant is
introduced into the channel through a plenum via the impingement
holes of the top plate. A schematic of the top plate is shown in figure
2. The solid serpentine lines represent the boundaries of each row,
which in the experiments, are separated by an insulating barrier for
undertaking row-by-row heat transfer measurements. The longitudinal
(L/D) and transverse spacing (S/D) of the pin-fin arrangement is 1.046
and 1.625, respectively. The height to diameter ratio (H/D) of pin-fins
is 0.5. The top plate of the configuration has 6 holes in the first row
where holes are located between pin-fins. The plate also has 6 holes
FLOW SOLVER AND BOUNDARY CONDITIONS
this study are 7500 and 0.0018, respectively. The computations are initially
done in 3-4 flow-through cycles to wash away the transients and for the
temporal-periodicity in the flow to be established, and then averaging of
the data is done for 5-6 flow-through cycles to collect second order
turbulent statistics. The time-period for averaging is confirmed when
statistics do not change over a longer period of time.
NUMERICAL RESULTS VALIDATION
CONCLUSIONSFig. 5 (a) and 5 (b) show the iso-surface of Q criterion colored by time-
averaged temperature for case-1 and case-2, respectively. Due to the
presence of cut-out region in case-1, an existence of small helical vortex
can be seen. This vortex is responsible for creating CVP (counter-
rotating vortex pair) downstream of the jet region after getting interacted
with the upstream cross-flow (figure 5 (a) ). These two low-temperature
vortices are mainly responsible for the heat transfer in this configuration.
CONCLUSIONS
[1] Busche, M.L., Moualeu, L.P., Chowdhury, N., Tang, C., Ames, F.E.,
"Heat transfer and pressure drop measurements in high solidity pin fin
cooling arrays with incremental replenishment," Journal of
Turbomachinery, 135(4):041011.
ACKNOWLEDGEMENT
This work was supported by a grant from the U.S Department of Energy
Award Number: DE-FE0011875, under the University Turbine Systems
Research (UTSR) Program with University of North Dakota as the prime
recipient and IIT-Chicago as the sub-recepient. I would also like to thank
LONI and HPC of LSU for providing the required computing resources.
Busche et al. [1] claims that by placing the jets in a specially designed cut-
out region of the pin-fin, a major portion of free jet shear layer is protected
and thus stagnation point heat transfer is not affected. This claim is
numerically verified in this study by comparing two cases. In the first case
(case-1), the jets are placed in the cut-out region of the pin-fin, as the
experiment, and in the second case (case-2), the jets are placed directly in
the wake of the pin-fin. Figure 1 shows the schematic of the numerical
domain for both the cases.
Louisiana State University
Mechanical Engineering Department
Baton Rouge, LA, USA
Susheel Singh
COOLING PERFORMANCE COMPARISON OF INCREMENTAL IMPINGEMENT PIN-FIN CHANNEL
CONFIGURATIONS USING LES TURBULENCE MODEL
Fig. 1 Top view of the incrementally impingement channel a) Impingement
in the cut-out region (case-1) b) Impingement in the wake region (case-2)
EXPERIMENTAL AND COMPUTATIONAL DOMAIN
Fig. 2 Experimental top and bottom plates
in the second, fourth, sixth, and eighth rows, where holes are located
right behind the pin-fin in the cut-out regions. Note that due to the
presence of the cut-out region, the shape of the pin-fins is not circular
in row-2, row-4, row-6, and row-8. The diameter of the jet-1 is 1.054
cm, and the diameter of the rest of the jets is 0.748 cm
Since the flow field is periodic in the span-wise direction, the
section A-A’ of the experiment plate is only modelled. A box-plenum
above the top plate, which feeds the coolant into the pin-fin channel, is
also included in the computational domain. A 3D block-structured
mesh is generated using ICEM- CFD; an O-block strategy is used to
create high quality grid around the pin-fins (figure 3). The
computational domain consists of 766 blocks. After performing the
sensitivity analysis, the mesh size of 4.2 million is chosen for the
simulation.
Fig. 3 Different views of the meshed domain for both case-1 and case-2
Sumanta Acharya
Illinois Institute of Technology
Mechanical, Materials & Aerospace Engineering Department
Chicago, Illinois, USA
An in-house code, Chem-3D, is used to solve the filtered Navier
Stokes (NS) equation. The Chem-3D solver is a three dimensional
block structured compressible flow solver which uses a curvilinear
mesh. The convective term of the NS equation is discretized by the
fifth order WENO (weighted-essentially-non-oscillatory) scheme and
the diffusive term is discretized by the second order central difference
scheme. Time stepping is done by implicit second order method. Sub-
grid scale stresses (SGS) are modelled by Dynamic Smagorinsky
model in which the Smagorinsky constant (Cs) is obtained by Lilly et
al. . The negative value of the constant (Cs) is clipped and averaging is
done over six neighboring cells.
(a)
RESULTS AND DISCUSSION
Fig. 4 Experimental and Numerical Cooling parameter for case-1
A wall resolved LES study is conducted in which Y+ value near all
the wall is less than 1. Wall boundary condition (BC) is used on the top
wall, the bottom wall, pin-fins, jet holes, and the plenum. Constant
temperature BC (309.15 K) is used on all the walls except the plenum
faces where adiabatic BC is imposed. Velocity inlet BC ( 294.95 K) is
imposed at the Y minimum face of the plenum and outflow BC is
applied at the exit of the channel. The velocity is obtained from the
experimentally reported mass flow rate (6.55 x 10-3 kg/s). The
Reynolds number and non-dimensional time step (Δt*Uo/D) used in
A row-by-row numerical cooling parameter is plotted against the
corresponding experimentally obtained row-by row cooling parameter
[1]. The error in numerical cooling parameter at all locations is less
than 5%. The error is close to 8% near last row which is due to the
discrepancies in the pressure drop measurement in the exit section.
Moreover, a same value of Cd is used for all the jets in the experiment
which is not realistic due to the existence of a different kind of the
recirculation zone in the jets.
Fig.. 6 Thermal Effectiveness (ε) along streamwise direction
Fig. 5 Existence of major vortical structures in (a) Case-1 (b) Case-2
(b)
As the jet directly impinges in the cross-flow in case-2, the deflection
of the jet core can be seen which dimishes the jet stagnation point heat
transfer. The resulting vortex, which is formed after the jet and the cross-
flow interaction, is not highly coherent and thus not be able to transport a
large amount of heat from the configuration. The HSV (horse-shoe
vortex ), which also transports heat, is visible in both the cases. The
HSV system in case-2 is not highly coherent as compared to the HSV
system of case-1. Figure 6 shows row-by-row thermal effectiveness
(non-dimensional bulk temperature) plot for both the cases. Case-1 has
the highest thermal effectiveness at all the rows which clearly indicates
the increased level of heat transfer in case-1.
REFRENCES
Two different incremental impingement pin-fin configurations are
simulated using LES turbulence model. A design benefit of placing jet in a
cut-out region is numerically explored. The presence of small helical
vortex and CVP increase the heat transfer in the original experimental
configuration. The cut-out region also helps in avoiding the jet deflection.
Forrest Ames
University of North Dakota
Mechanical Engineering Department
Grand Folk, North Dakota, USA