2-D CFD SIMULATION OF THE HEAT TRANSFER AND FLUID DYNAMICS IN AN
EXPERIMENTAL MODEL OF THE HOT END OF A STIRLING ENGINE
Mounir Ibrahim
Mechanical Engineering Department
Cleveland State University
2121 Euclid Ave, Cleveland, Ohio 44115-2214
ABSTRACT
Experiments were conducted at the University of
Minnesota (UMN) for simulating the hot end of a Stirling
engine. This experimental test rig included the acceptor
(normally a heater), expansion space, regenerator, and
displacer. To establish a temperature difference for the
purpose of measuring convective heat transfer coefficients
between the heater wall and the flow, one was heated and
the other cooled. The expansion space has a 127 mm (5-
inch) diameter dome and a minimum spacing between the
dome and the displacer of 4.8 mm (0.1912 inch), at
maximum displacer amplitude. The regenerator was
simulated as a porous medium. Velocity and temperature
measurements were made at seven different locations (PI,
P2, ...and P7) along the dome surface and on lines normal
to the dome surface that extended a distance from 4.5 to 6
mm (0.18 to 0.24 inches) into the flow field.
In this paper a 2-D computational model was developed
to simulate the UMN experimental test rig. The acceptor
was modeled with a 2-D annular geometry that had three
annuli matching the UMN hydraulic diameter, Dh, and
approximating the UMN flow- and heat-transfer areas.
The CFD (Computational Fluid Dynamics) model has 2-D
geometry, no metal thicknesses (heat capacities), laminar
flow, and one thermal boundary conditions case Ts-315.
These simplifications were made to reduce the
computational (CPU) time required. Fluent, commercial
software version 6.3.26 was used in this effort. The CFD
simulations provided more insight into the details of the
flow and thermal fields in the expansion space, acceptor
channels and porous media. Many of these details were
not obtained via the experiment (particularly for the
acceptor and porous media) and will likely provide
helpful information for future Stirling-device modeling
and design efforts.
I. INTRODUCTION
NASA Glenn Research Center (GRC) has funded
Cleveland State University (CSU) in collaboration with
the University of Minnesota (UMN) and Gedeon
Associates (GA) to develop a reliable Computational
Fluid Dynamics (CFD) code for application to Stirling
engines. This code can predict engine performance with
978-1-4673-1835-8/12/$31.00 2012 IEEE
the goal of significant improvements in accuracy when
compared to one-dimensional (1-D) design code
predictions. The purpose of this work is to help in
developing a Stirling multi-dimensional computational
model. Such a model will help designers get a better
understanding of the critical Stirling design parameters
and will permit improvements in the accuracy of I -D
design and optimization codes. The engine simulation is
intended to calculate the overall heat transfer for the heat
exchangers, the net power output and the efficiency of the
Stirling engine. In addition, the simulation provides
detailed pressure, velocity and temperature fields inside
the engine working spaces. Figure 1 shows a cut-away of
a modern Stirling engine. The figure depicts the
Compression Space (CS), Cooler (CR), Regenerator (RE),
Heater (HR) and Expansion Space (ES) as the major
components of the engine.
From similar but earlier research, done under the SP-I OO
program, an analytical technique for modeling unsteady
flow and heat transfer in Stirling engine heater and cooler
tubes was developed (Ibrahim et al. [I ] and Simon et al.
[2]) which is now an essential component of the I -D
system simulation code, Sage (at that time called
d
GLIMPS), Sage User Manual, 3r ed. [3]. The tests used
for this development showed important features of
oscillatory flow which led to enhanced understanding of
the mechanisms for losses in Stirling systems. Such
features include transition from laminar flow to turbulence
in each cycle of oscillation and the rapid dissipation of the
turbulence which remains from one cycle as a result of the
extremely strong axial straining of the fluid upon temporal
acceleration at the beginning of the following cycle. The
development took place by coordination of the experiment
and 2-D turbulence modeling. This lead to a I -D model
(Ibrahim et al. [I ]) that was incorporated into Sage.
Adolfson et al. [4] conducted an introductory investigation
of the flow features in the expansion space, of the Stirling
engine, by utilizing two parallel disks as a simulation of
the heater head region. Flow visualization and hot-wire
anemometry measurements were performed under
conditions of unidirectional flow and oscillatory flow
(zero mean velocity). Valuable information was offered
by this work towards understanding the fluid mechanics
inside a Stirling engine. Unsteady separation and laminar
to-turbulent transition were shown in this study. Ibrahim
1111
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND
et al. [5] and Ibrahim et al. [6] focused on the
computational analysis of this case study. Additionally,
two papers were written to document steady flow in the
same facility, Adolfson et al. [7], and Zhang and Ibrahim
[8]. Also, two review papers (Tew et al. [9], Tew et al.
[10]) which discuss these tests. Several other research
papers on technique development were presented at the
IECEC04 (Ibrahim et al. [11] [12] and Hoshino et al. [13]).
Linear Alternator
Stationary Magnets
Heater sup plied
by ra dioisotopes
Power Piston Flexures Displacer Flexures
Figure 1. Cut-away of a modern Stirling engine
In this paper the focus is given to the expansion region (an
area east of the heater shown in Figure 1). The heater (or
acceptor) of a Stirling engine, where most of the thermal
energy is accepted into the engine by heat transfer, is the
hottest part of the engine. Almost as hot is the adjacent
expansion space of the engine. In the expansion space, the
flow is oscillatory, impinging on a two-dimensional
concavely-curved surface. Knowing the heat transfer on
the inside surface of the engine head is critical to the
engine design for efficiency and reliability. Jiang and
Simon [14] experimentally investigated (at UMN) the
flow and heat transfer in the expansion space region. Flow
fields and heat transfer coefficients were measured to
characterize the oscillatory flow. This paper focuses on
the CFD simulation of the UMN rig. The overall objective
is to show how these results might be used for heater head
and acceptor region design calculations. Thus improved
CFD codes are needed to design modern Stirling engines
of high efficiency and power output. This study involves
three of the engine components: Regenerator (RE), Heater
(HR) and Expansion Space (ES) with two thermal
boundary conditions: Ts-308 and Ts-315 (indicating
surface temperatures of 308 and 315 K respectively).
However because of space limitation results shown in this
paper are only for ES and thermal BC Ts-315.
II. EXPERIMENTAL SET-UP
The experimental set up for the fluid flow and heat
transfer data is described in details in Jiang and Simon
[14]. A brief description is given here for completeness.
Figure 2 shows a photograph of the UMN test facility.
. . . , M Ibrahim
The left side of Figure 2 shows the cam mechanism,
including an eccentrically mounted plate and spring, used
to impart sinusoidal motion to the displacer piston. The
"suction " and "exhaust " strokes of the displacer
alternately (1) suck air through the expansion space, via
drawing air in from the surrounding room, through the
porous medium and then the acceptor, to the expansion
space, and (2) exhaust air from the expansion space, via
pushing it through the acceptor then the porous medium to
the surrounding room. Some of the internal features of the
test section of Figure 2 are shown in Figure 3. In
particular, note the radially oriented slots in the acceptor
which impart non-axisymmetric (3-D) features to the
experimental geometry and thus prevent a 2-D simulation
from exactly representing the test section.
Figure 2. The University of Minnesota (UMN) Test
Facility
Front and side views of the test-section measuring stations
are shown in Figures 4 and 5, respectively. Linear and
angular positions of each measuring station, relative to the
center of the nearest acceptor slot are given in Table 1.
Internal design parameters of the test section, including
dimensions of the acceptor slots, are summarized in Table
2. General information for the air/working fluid upon
entrance to the test section from the room is given in
Table 3. Operating frequency of the test section was 1.6
Hz.
Thermal boundary conditions of the UMN test section are
defined with reference to Figure 6. A two-inch diameter
heating element was glued to the center of the head (dome
of expansion space) by using an epoxy of good
conductivity. This imposed an approximately constant
heat flux boundary condition in this area of the head. A
copper tube was wrapped around the outer surface of the
head where the acceptor simulator was located. Cooling
by city water kept the temperature of the acceptor at,
approximately, a uniform 303 K. The entire head was
then wrapped in insulating material to reduce heat transfer
to ambient.
2/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND . . . , M Ibrahim

study: A Dell Precision PWS670, Intel (R) Xeon (TM)

CPU 2.8 GHz. After cycle-to-cycle convergence was
established one cycle more was run with 0.5 degree per
time step and all the data needed was recorded (velocity,
temperature, pressure ... etc.).

Table 1 Positions of Measuring Stations on Head of

UMN Test Section (See Figures 4 and 5)
Measuring S e e to x to R
Stations (inch) slot slot (inch)
center center
(inch)
PO 0.556 0' 004055 0.034502 4.875
PI 0.678 5' 0.7305 0.061358 4.8125
P2 1.090 10' 1.0555 0.082899 4.5
P3 1.366 20' 1.7055 0.124648 4.1875
P4 1.722 30' -2.319 -0.14675 3.625
P5 1.926 40' -1.669 -0.09288 3.1875
P6 1.722 50' -1.019 -0.0645 3.625
Figure 3. Some Internal Features of UMN Test P7 2.277 60' -0.369 -0.0133 2.0625
Section
Note: + counterclockwise, - clockwise, e(fin and slot) =
4.67Y (360177), efm=3.21' (0.268 inch), eslot=lo461'
(0.122 inch), x is the distance from the station hole
center on the inner surface to the nearest slot center

,
,
,
,
,
'I
......'
P5

Figure 4. Front View of Measuring Stations on the

Head of the UMN Test Section

III. CFD SIMULATION

The Fluent commercial code [15] was utilized to simulate

Figure 5. Side View of Measuring Stations on the
the above cases. Version 6.3.26 was used with 151,657
Head of the UMN Test Section
cells. Typically we ran the cases with a 0.000625 s time
step which corresponds to 1000 time steps per cycle (i.e.
the displacer moves 0.36 degrees per time step). A
convergence criterion of lE-05 was used for all variables
and the simulation ran for over 15 cycles to get cycle-to
cycle convergence for Case Ts-315. Each time step
required approximately 20 s on the computer used for the

3/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND . . . , M Ibrahim

Table2 UMN Test Section Design Parameters, mm transfer area/unit length (in the flow direction) =

(inches) 41,206.37 mmA2. In order to simulate this 3-D geometry

Displacer Diameter 203(8) in a 2-D domain we made several attempts to match the
Regenerator LD. 207.9(8.184) above three parameters. The easiest one to match is the
hydraulic diameter by choosing an annular geometry with
Regenerator O.D. 255.0(10.04)
spacing = 2Dh =10.57 mm. Then one can try applying
Inner Regenerator Wall
different numbers of annular channels between the inner
Thickness 1.29(0.05)
and outer diameters. As the number of these annuli
Outer Regenerator Wall
increases (with a decreased wall thickness between
Thickness 1.76(0.07)
channels) the heat transfer area will increase. A
Appendix Clearance Gap 1.08(0.04) compromise was made by choosing three annuli with the
Minimum Clearance Between two inner ones having 10.5mm gap and a metal thickness
Cylinder Head and Displacer 4.87(0.192) = 26.8 mm, that leaves the outer annular gap 10.2mm.
=

Heater Length 74.6(2.936) In this configuration the deviations from the original
Heater Channel Width 3.11(0.122) geometry are: 1) Hydraulic diameter, -0.13%, 2) Flow
Heater Channel Height 25.4(1.0) area + 30.8%, and 3) Heat Transfer Area/unit length (in
=

Heater Fin Thickness 6.18(0.243) the flow direction) + 162%.

=

Figure 7a also shows different locations in the 2-D model,

Table3 General Information for Air Flow into the
corresponding to the UMN-test-rig measuring stations,
UMN Test Section
where computational velocity and temperature data were
Working fluid Air
taken. There are seven lines PI , P2,.....,P7 corresponding
Operating temperature 298 K
to locations where the UMN measurements were made.
Operating pressure 1 bar
Seven points (also designated PI , P2,....,P7) are shown at
Kinematic viscosity (v) 15.9xl0-omL/s
the end of these lines. Note that Line P8 and Point P8,
defined in Figure 7a, do not directly correspond to a UMN
measuring station; the reason for definition of this extra
CFD data location will be discussed below. Velocity and
temperature data were obtained as they varied with time
and location during the cycle. On Figure 7a we
documented the main dimensions of the model (in mm) as
well as other important data such as: a) displacer
frequency (t) 1.6 Hz, b) displacer-motion amplitude (A)
=

= 8.78 mm. Thus displacer velocity can be obtained from

(-2*n*f*A*sin (2*n*f*time. The porous-media
parameters used in the CFD model were permeability =

2.14eE-08 mA2 and inertial coefficient 0.19. These data

=

were obtained from the UMN experiments.

Figure 7b. defllles the thermal boundary conditions

examined in this CFD study, Case Ts-315. In this case
only a small section of the dome (2 inch diameter, as in
Figure 6. Setup of Thermal Boundaries of the UMN
the UMN experiment) is heated to 315 K while the rest of
Test Section
the dome surface is adiabatic. As shown in Figure 7a, the
solid surfaces in the acceptor (the heat capacity of these
Figure 7a shows a schematic for the 2-D computational walls was not part of the computation domain) were kept
model used in this study to simulate the UMN-test-section. at 303 K. It should be noted that for Case Ts-315, point P8
The dimensions of the acceptor simulator used in the (shown in Figure 7a) lies close to the mid position of this
UMN experiments are given, again, below: heated section. It is worth summarizing the differences
between the CFD model assumptions and the test
Table 4. Dimension of UMN Acceptor Simulator (3-D) conditions of the UMN test section: 1) The UMN test
Outer Inner Length, Slot Slot Fin section is non-axisymmetric (3-D) while the CFD model
Diameter Diameter width height width is 2-D, axisymmetric, 2) No metal thicknesses (i.e. heat
243.08 206.88 74.6 3.11 17.97 6.18 capacities) were included in the CFD model, 3) The CFD
mm mm mm mm mm mm model assumes laminar flow, 4) Somewhat different
WIth 77 slots III the acceptor we calculated the followlllg thermal boundary conditions were used in the 2-D model.
parameters for the 3-D rig: 1) Hydraulic diameter, Dh =
These simplifications were made to reduce the
5.286 mm, 2) Flow area 4,287.9 mmA2, and 3) Heat
=
computational (CPU) time required. Our experiences

4/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim

show that running 3-D simulations, including the metal 1) The highest velocity magnitudes take place at 90 (Fig.
heat capacity and allowing turbulent flow would have 9) & 270 (Fig. 12) C.A. as expected,
increased the CPU time tremendously. The current CFD 2) Between C.A. 0 & 180 (Figs. 8-10), suction stroke, the
model (with its simplifications) still is computationally flow is jetting out of the annular channels and expected to
intensive. Moreover, the CFD memory size of the model be very different from UMN results due to the fact that
was sufficiently small that parallel-computations were not CSU is a 2-D vs. UMN is a 3-D geometry. Highly
justified (i.e. there is no gain in CPU time running it on 4 recirculated zones are noticed near the acceptor exit (in
parallel nodes compared to a single computer node, for this case will be jet exits). More recirculation activities are
example). seen at 30 & 150 C.A. with more intensity at 150 c.A.
Accel,tOl' Exit This is due to the fact that while the jetting velocities are
Accel,tOl'IPOl'OUS similar (for the two cases) there is more room in the ES
POl'OUS Medi lVIedi hltelf' ce Sttionay ''Vall for the flow to expand into at 150 than at 30,
Exit
65.12 74.6----+_+_ 3) Between C.A. 180 & 360 (Figs. 11-13), exhaust stroke,
=;I;;;t:;=tLine P1 the fluid is flowing into the annular channels (sink flow).
POl'OUS Medi Again the flow is expected to be very different from UMN
LineP2

t
MoviIll(*Sin(2*PI*f*
Wll (mlibtic) results due to the fact that CSU is a 2-D vs. UMN is a 3-D
Ve l -2*PI*f*A
= time) LineP3 geometry. Highly recirculated zones are noticed near the
PI=3,14159 I SoM t 303 K
acceptor exit plane at 210 C.A. as compared to 330 c.A.
f = 1.6 Hz
A=OOS7S 123.5m
LineP4 This is again due to the fact that while the flow velocities

I
are similar (for the two cases) there is more room for the
'B LinePS--- flow before entering the acceptor at 210 than at 330.
.- ., Symmetl'y ------H
is
/

Axia l Figure 14. shows a comparison of the velocity (m/s)

Direction 4.S6 between CFD and UMN experimental data at the end
point along lines PI , P2, P3, P4, P5, P6 & P7 vs. C.A.
Figure 7a. A Schematic (dimensions in mm) for the 2- degrees (zero is when the displacer at TDC ). These points
D Model simulating the UMN Expansion-Space and were picked up by UMN and they are at distances from
used in this study the dome surface that ranges from 4.5 to 6.5 mm.
Therefore they represent flow activities outside the
boundary layer. The computed velocities are in reasonable
Ts-315 Case agreement with the data especially further away from the
acceptor exit plane (e.g. P7). The velocity traces seem to
be in better agreement with the data in the suction stroke
(0-to-180) for P5, P6 and P7 than those for PI , P2,P3 and
Adiabatic P4. On the other hand these velocity traces seem to be in
'Vall better agreement with the data in the exhaust stroke (180-
to-360) for P2, P3 and P4 than those for PI , P5, P6 and P7.
These results are attributed mainly, as discussed above,
due to the 2-D (CSU model) vs. 3-D (UMN model). Also
during the suction stroke there are jet flow coming out of
the acceptor channels into the expansion space, the CFD
Heated Section
model (laminar flow) is better capable to simulate this
Ts=315K flow further away from the jets (P5, P6 and P7) than near
the jet exit. The worst comparison between CFD and
Figure 7b. A Schematic (dimensions in mm) for the 2- experiments do occur during the suction stroke and for PI ,
D Model simulating the UMN Expansion-Space P2, P3 and P4. The CFD data show almost symmetric
showing the thermal boundary conditions examined profiles when one compares suction with exhaust strokes,
with only a small section of the dome is heated to 315 but the experimental data show asymmetry. Finally, the
K, Ts-315 CFD data are close to zero velocities at 0 and 180 C.A.
while the experimental data did not show such a behavior.
IV. RESULTS

A. Flow Field

Figures 8, 9, 10, I I , 12 and 13 show velocity contours

(colored by the velocity magnitude) and velocity vectors
near the acceptor exit for 30, 90, 150, 210, 270 and 330
C.A. respectively. The following observations were made:

5/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND . . . , M Ibrahim
0.396
_3..01

0.0

Y'- Velocity contours (colored by the

0.396 m/s .
\ JDltude) and velocity vectors near the

Figure 11. Velocity contours (colored by the
acceptor exit for 30 CA. velocity magnitude) and velocity vectors near the
acceptor exit for 210 CA.

0.84 mls

0.0 mls

Figure 9. Velocity contours (colored by the

velocity magnitude) and velocity vectors near the Figure 12. Velocity contours (colored by the
acceptor exit for 90 CA. velocity magnitude) and velocity vectors near the
acceptor exit for 270 CA.

0.44

I 0 396 m/s Velocity contours (colored by the

. Figure 13. Velocity contours (colored by the
h.v .. . .Jnitude) and velocity vectors near the
... ..
velocity magnitude) and velocity vectors near the
acceptor exit for 150 CA.
acceptor exit for 330 CA.

6/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim

0.30 II . UMN-data-P1 0.35 I UMN-data-P5 1

'"
1 ,
C;
I. I - CFD-P5
.!!! 0.25
- CFD-P1 .!!! 0.30
II; E
j
D.
E
>. _ 0.25 PI " Il.
,:: 0.20 : 1/ /
IlY /
-;;
t.) '" :r' :'I.. \ 0.20
r )3 III '\ ,A I
""I
0.15 '" '" / "\ / ",.
::::- c: "\ , ::::- c: 0.15
;;:; 0.10 '" .-
:c 0.10 I J{r \ 1
.eN

e. 0.05 I
J
\ / 1,\ Eo
3: 0.05
I r\ r-' 7 JJ
c:
w 0.00 V \V c:
w 0.00 \/ \
0 60 120 180 240 300 360 0 60 120 180 240 300 360
CA Degrees CA Degrees

UMN-data-P6 1
== 0.50
I
UMN-data-P2 1 C;
0.30
1

OA5 - CFD-P6
- CFD-P2
.!!! 0.25
!I II
il>
E
I\
N I
D. OAO
j:....:: 0.35 1\'.. ,:>:: II; E
; i= : I ,:: 0.20
'(j ri \

0.30
0.25 ./1 ;: i' H\ IA
...
t.) '"
o
Qj '" 0.15 J
V "". V ,.,. r\
.: 0.20
" L.--
I' '"" , IV
::::- c:
J0.10 "'f\.
:c

e.
0.15
0.10 r
/
"-
"
J' 1.\
0.05
e. I I '-
-I,r
II>
c:
w
0.05
V v c:
w
\/ V
0.00 0.00
0 60 120 180 240 300 360 0 60 120 180 240 300 360
CA Degrees C.A. Degrees

UMN-data-P7 1
==
II>
OA5
OAO II UMN-data-P3
CFD-P3
1 C;

0.25

- CFD-P7 1
li; E . Ii;: EII>
Ion! :.Ii,,, If 1\
0.35 0.20
:v
,:: 0.30
0.25 JV<
I' f >;
>- -
g '" 0.15 /h
0.20
J- I
/ "
,
/ \.'rr "
f V \1\
-
'" '"
.: '/ \ / J \
0.10

e. 0.05 II "'- ,.6V \1

0.15 - ..,
'\ // \
I\,
.e
Eo 0.10
I ":J
./V
"'
II>
c: 0.05 c:
II>
w 0.00 rv '\1' w 0.00
0 60 120 180 240 300 360 0 60 120 180 240 300 360
CA Degrees C.A. Degrees

== 0.50
Figure 14. Comparison of velocity (m/s)
OA5 UMN-data-P4 1 between CFD and experimental data at the
'ot E
II>
OAO I CFD-P4
D.
I' end point along lines PI, P2, P3, P4, P5, P6
j:....:: 0.35
'(j 0.30 j \ & P7 versus CA. degrees (zero is when the
0.25 ttl ... \ " displacer at TDC )
.: 0.20 IJ /
:c 0.15
'\ \ V"I?'" \1 \
e. 0.10 / '\ " \
0.05 I I/' ..lI
W
0.00 /
o 60 120 180 240 300 360
CA Degrees

7/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim
Figure I S. shows comparison between CFD &
experimental data for V absolute, m/s at line P7 for C.A. =
30, 90, I SO, 210, 270 and 330 degrees. The best
agreement between CFD and the experiments appears at
210 c.A. Next to that come the results at 90 and 270 C.A.
which is followed by results at 30 and I SO with the worst
scenario takes place at 330 C.A.
The pressure field calculated by the CFD model shows the
highest pressure drops occurring at 90 and 270 C.A.,
where the absolute velocities are greatest, with most of it
occurring (as expected) in the porous media (87.S% of the
total pressure drop).
The overall flow structure exhibited high recirculation
zones in the expansion space, straight flow in the acceptor
channels and fairly uniform flow in the porous media.
Highly recirculated zones are noticed near the acceptorlES
plane at I SO degree C.A. At that time during the cycle the
jets are issuing into a larger expansion space, than earlier
in the cycle, and flow is decelerating.
The major difference between the UMN test section and
the simplified CSU 2-D model is due to the non
axisymmetric, 3-D, features of the radially-oriented UMN
acceptor channels. Despite this 2-D model simplification
there is overall good agreement between CFD and
experimental velocities, and excellent agreement at some
locations. Generally, good agreement is obtained further
away from the acceptor/ES plane (in the expansion space),
e.g. line P7, during the suction stroke (0 to 180 degrees
c.A.) when flow is jetting from the acceptor to the ES.
However during the exhaust stroke (from 180 to 360) with
sink flows into the acceptor channels from the expansion
space, better agreement was obtained in the expansion
space near the acceptor channels.
The overall conclusion of the flow field comparisons was
that the 2-D CSU CFD model provided a reasonably good
flow field in the expansion space, the only component for
which experimental data was available for comparison.
This provided confidence for the model to be utilized for
studying the thermal field and heat transfer coefficients.
B. Thermal Field
The flow field (as obtained above via experiments and
CFD) is highly agitated and is expected to be only weakly
affected by the thermal boundary conditions. In the UMN
experiments a heating element (about 2 inches in
diameter) is glued to the center of the dome. The acceptor
wall temperatures were virtually kept constant at 303 K
via cooling water channels wrapped around the acceptor
outer surface. In the CFD simulations a thermal boundary
conditions was chosen Ts-31S K. In this case only a small
portion (2S.4 mm radius) was kept at 31S K while the rest
of the dome is kept adiabatic. The acceptor wall
temperatures were kept at 303 K.
Figures 16, 17, 18, 19, 10 and 21 show temperature
contours near the acceptor exit for 30, 90, I SO, 210, 270
and 330 C.A. respectively. The contours show how the
temperature field for this thermal boundary conditions
change during the cycle. The dome surface temperature
changes monotonically from 31S K near the dome center
and decreases to 303 at the acceptor location. Also as a
result of the specified displacer motion (speed and
frequency) only a small portion of the hot thermal wave
travels towards the acceptor and is not able to penetrate all
the way to the acceptor exit plane. Vice versa the cold
wave from the acceptor exit does not travel very far
towards the dome center.
Since the dome is chosen to be adiabatic, in this case,
except near the center, we added another line P8 in the
middle position of the heated section (see Figure 7a). The
heat transfer coefficient was computed from the wall heat
transfer rate (from CFD in W/ml\2) and the temperature
difference between the wall and the free stream. Since our
CFD results showed good agreement (in the most part)
with experiment in the temperature difference near the
wall (not shown in this paper) we used the free stream
temperature at y = 2.3 mm away from the wall (as was
done in the experiment).
Figure 22. shows CFD data for the temperature (K) along
Line P7 for different C.A. degrees, Ts-31S. In this case a
large temperature variation at the wall (at P7), from 304.S
to 308 K, occurs which is not shown in the experimental
data. This is attributed to the dome surface being adiabatic
and the lack of the presence of a metal wall. Accordingly,
we chose to compare the temperature difference between
the wall and any y location (away from the wall) along
line P7 as shown in Figure 23. The CFD data and the
experiments are in close agreement near the wall but
different away from the wall with CFD showing
maximum of 1 K difference while the experiments show
close to 3 K variation.
Finally, Figure 24 shows comparison between CFD and
UMN experimental data for the heat transfer coefficient
for different C.A. degrees for Ts-31S (at P8). The highest
difference is about 7S%, while the lowest about 23.S%.
V. CONCLUSIONS
The pressure field calculated by the CFD model showed
the highest pressure drops occurring at 90 and 270 C.A.,
where the displacer and fluid velocities were the greatest.
Most of the pressure-drop occurred (as expected) in the
porous media (87.S% of the total pressure drop).
The overall flow structure exhibited more recirculation
zones in the expansion space (as might be expected due to
its larger volume). Between C.A.s 0 and 180 degrees, the
suction stroke, the flow jets from the annular acceptor
channels into the ES; the detailed flow should be expected
8/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim
to be quite different from UMN data due to the fact that
the CFD simulation of the acceptor is a 2-D
approximation of UMN's non-axisymmetric (3-D)
acceptor. Highly recirculated zones are noticed near the
acceptor/ES interface (in this case they are driven by
exiting jets of gas). More recirculation activities are seen
at 30 & 150 c.A. with more intensity at 150 C.A. '"
Accordingly more uniform temperature fields were .c
f!
observed for C.A. 150 compared to 30.
Between C.A.s of 180 and 360 degrees, the exhaust stroke,
the fluid is flowing into the annular acceptor channels
(sink flow) from the ES. Highly recirculated zones are
noticed in the ES near the acceptor/ES plane at 210 C.A.
as compared to 330 C.A. This is again due to the fact that Figure 15. Comparison between CFD & experimental
while the flow velocities are similar (for the two cases) data for V absolute, m/s at P7 for CA 30, 90, 150, 210,
there is more room for the flow before entering the 270 and 330 degrees
acceptor from the ES at 210 than at 330
In the case of Ts-315, a hot wave starts from the dome [1]
center area and attempts to travel all the way to the
acceptor/ES plane (during the exhaust stroke). Similarly a
cold wave starts from the acceptor/ES plane and attempts
to travel all the way to the dome center area (during the
suction stroke). [2]
The major difference in the CFD simulation, compared to
the test section, is the 2-D simplification of the
experiment's non-axisymmetric (3-D) acceptor. In spite of
this 2-D simplification there is overall good agreement
between the experiments and the CFD results, with [3]
excellent agreement at some locations. Generally, good
agreement is obtained further away from the acceptor/ES
interface, e.g. line P7, particularly during the suction [4]
stroke (0 to 180 degrees c.A.).
In addition to the geometrical (2-D) simplifications, there
are also thermal boundary conditions simplifications.
However, there was also overall good agreement between [5]
the experiments and CFD results for the heat transfer
coefficient near the center of the heated area of the dome
(with agreement in the range from 23 to 75%).
Comparing CFD results on the basis of temperature [6]
differences (between the wall and the fluid) good overall
agreement were obtained for Case Ts-315.
VI. ACKNOWLEDGEMENTS
The work described in this paper was performed for the
NASA (National Aeronuatics and Space Administration) [7]
Science Mission Directorate (SMD) and the Radioisotope
Power System (RPS) Program, Exploration Systems
Mission Directorate (ESMD) , which provided funding for
these projects. Also, the opinions expressed in this paper
are those of the author and do not necessarily reflect the
views of NASA. We are grateful for sponsorship of this
work by the NASA Glenn Research Center under grant
NNC046B62G. The contract monitor was Dr. Roy Tew.
0.25
330
0.20
90
0.15
0.10 270
0.05
210
0.00 :j==;-c;zt======::I====C-- 0.005
150
=
7. REFERENCES
Ibrahim, M., Baur, c., Simon, T. and Qiu, S. 1994.
"Modeling of Oscillatory Laminar, Transitional,
and Turbulent Channel Flows and Heat Transfer,"
Proc. of the 10th Int'1. Heat Transfer Conf., Vol. 4,
pp. 247-252, G. F. Hewitt, Ed.
Simon, T. W., Ibrahim, M., Kannaparedy M.,
Johnson, T. and Friedman, G. 1992. "Transition of
Oscillatory Flow in Tubes: An Empirical Model for
Application to Stirling Engines," Proc. of the 27th
IECEC Conference, Vol. 5, pp. 495-502, paper #
929463.
Sage User Manual, 3rd ed., Gedeon Associates,
16922 South Canaan Road, Athens, OH 45701,
USA.
Adolfson, D., Simon, T. W., Ibrahim, M. B. and
Gedeon, D., 2003, "Unsteady Fluid Dynamics
Simulation of a Stirling Engine Heater Head,"
Paper # AIAA-2003-5934, Presented at the 2003
IECEC Conf., Portsmouth VA.
Ibrahim, M. B., Kembhavi, S., Simon, T. W., and
Gedeon, D., 2002, "A 2-D Axisymmetric CRD
Modeling of Oscillatory Flow with Separation, "
IECEC 2002 Paper No. 20121, 2002 IECEC Conf.
Ibrahim, M. B., Zhang, Z., Kembhavi, S., Simon,
T. W., Tew, R., and Gedeon, G., 2003, "A Model
of 90 Degree Tum Oscillatory Flow, " Paper #
AIAA-2003-5935, 2003 International Energy
Conversion Engineering Conference, Portsmouth
VA.
Adolfson, D., Simon, T. W. and Ibrahim, M. B.,
2004, "Measurements, Computation and Analysis
of a Semi-Constrained, Axi-symmetric Jet," ASME
GT2004-54215, 2004 Int'1. Gas Turbine
Conference, Vienna, Austria.
9/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim
315K
3.15e+02 ----------- 3_15e+02 -------

3.14e+02 ___________
:::: -------
.............
3.14e+02

3.13e+02 3.13e+02

3.12e+02
3.12e+02
3.11e+02
3.11e+02
3.11e+02
3.11e+02
3.10e+02
3.10e+02
3.0ge+02
3.0ge+02
3.0Se+02
3.08e+02
3.08e+02
3.08e+02
3.07e+02
3.07e+02
3.06e+02
3.06e+02
3.05e+02
3.05e+02
3.051':+02

3.05e+02 3.04e+02

3.04e+02 3.03e+02

3.03e+02 3.02e+02

3.02e+02 3.02e-t02 Figure 19. Temperature

3.02e+02
Figure 16. Temperature 3.01e+02
contours in the ES for
3.01e+02 3.00e+02

3.00e+02 contours in the ES for 30 210 CA., Ts-315

CA., Ts-315 300K
300K
Figure 16. Temperature
contours in the ES for 30
315K C.A., Ts-315 315K
3.15e+02 -- ---------
3.15e+02 ________mJCJ:l!IElf.!
::::: -----
:::::: ------- 3.13e+02 ............
3.13e+02 ..........._ 3.12e+02

3.12e+02 3.11e+02

3.11e+02 3.11e+02

3.11e+02 3.10e+02

3,10e+02 3.0902+02

3.0ge+02 3.08e+02

3.08e+02
3.08e+02

3.07e+02
3.08e+02
3.06e+02
3.07e+02
3.05e+02
3.06e+02
3.05e+02
3.05e+02
3.04e+02
3.05e+02
3.03e+02
3.04e+02
3.02e+02
3.03e+02
Figure 17. Temperature 3.02e+02 Figure 20. Temperature
3.02e+02

3.02e+02 contours in the ES for 90 3.01e+02

contours in the ES for
3.o0e+02
3.01e+02
CA., Ts-315 270 CA., Ts-315
3.00e+02

300K 300K

315K 315K
3.15e+02 -----------
3.15e+02 ----------- :.::::: -------
3.14e+02 ___________ 3.13e+02
3.14e+02 3.12e+02

3.13e+02 ............ 3.11e+02

3.12e+02 3.11e+02

3.11e+02 3.10e+02

3.11e+02
3.0ge+02
3.10e+02
3.08e+02

3.0ge+02 3.08e+02

3.08e+02 3.07e+02

3.08e+02 3.0Be+02

3.07e+02 3.05e+02

3.06e+02 3.05e+02

3.05e+02 3.04e+02

3.05e+02 3.03e+02

3.04e+02 3.02e+02

3.03e+02 3.02e+02

3.02e+02 Figure 18. Temperature 3.01e+02

Figure 21. Temperature
3.02e+02
contours in the ES for 150 3.00e+02 contours in the ES for 330
3.01e+02

3.00e+02 CA., Ts-315 CA., Ts-315

300K
300K

10/11
 2-D CFD SIMULATION OF THE HEAT TRANSFER AND ... , M Ibrahim

308.00 [8] Zhang, Z_ and Ibrahim, M. (2004a), "CFD Studies

307.00
-
-----
l- I on A Large Diameter Jet Impingement Flow ",
Paper-HT-FED-56058, Proceedings of 2004
'"
----- ---- CFD-P7-30
306.00 - CFD-P7-90 ASME Heat Transfer/Fluids Engineering Summer
- -CFD-P7-150
e
---- Conference Westin Charlotte & Convention
-CFD-P7-210

305.00
- --
CFD-P7-270 Center, Charlotte, North Carolina, USA, July 11-
-- CFD-P7-330
304.00
- 15, 2004_
303.00
l-
I [9] Tew, R. c., Cairelli, 1. E., Ibrahim, M. B_, Simon,
0 0.001 0.002 0.003 0.004 0.005 T. W_ and Gedeon, D., 2003, "Overview of NASA
Y, Location, m Multi-Dimensional Stirling Convertor Code
Development and Validation Effort, " Space,
Figure 22. CFD data for the temperature (K) along
Technology and Applications International Forum
Line P7 for different C.A. degrees (Zero is when the
(STAIF-2003), Albuquerque, NM, February 2-6,
displacer at TDC), Ts-315
2003_
[10] Tew, R. c., Ibrahim, M. B., Simon, T. W., Mantell,
S., Gedeon, D., Qiu, S_ and Wood, G., 2004,
0.50 P7-30Deg

"Overview 2003 of NASA Multi-Dimensional
- P7-90Deg
0.00 P7-150Deg Stirling Convertor Code Development and DOE &

P7-210Deg
NASA Stirling Regenerator R & D Efforts," Space,

-0.50
P7-270Deg
r---
--===:=:

'"
-1.00 t.:::: P7-330Deg Technology and Applications International Forum
r- -CFD-P7-30
-1.50 : , -CFD-P7-90 (STAIF-2004), Albuquerque, NM, February 8-12,
"
....0 . -CFD-P7-1S0 2004_
I
-2.00 CFD-P7-210

-CFD-P7-270 [11] Ibrahim, M.B., and Wang, M., Tew, R.C. and
It

-2.50
. CFD-P7-330 David Gedeon "Experimental Investigation of
I

-3.00 Oscillatory Flow Pressure and Pressure Drop

0 0.001 0.002 0.003 0.004 0.005
through Complex Geometries, " AIAA 2nd
YI Location, m
International Energy Conversion Engineering
Figure 23. Comparison between CFD and UMN Conference, Providence, RI, August 16-19, 2004_
experimental data for the temperature (K) difference Paper Number AIAA-2004-5560.
(Ts-T) along Line P7 for different C.A. degrees (Zero [12] Ibrahim, M.B., Mittal, M_, Simon, T. and Gedeon,
is when the displacer at TDC), Ts-315 D_, "A 2-D CFD model of a Free Piston Stirling
Engine for Space Applications with Annular Heat
Exchangers, " AIAA 2nd International Energy
:::s::: 20
Conversion Engineering Conference Providence,
RI, August 16-19, 2004, Paper Number AIAA-

[13]
2004-5583_
Hoshino, T., Watanasirisuk, Po, Ibrahim, M.B. and
......
.. Simon, T. "Dynamic Temperature Measurements
I- - I.-Ar- in Stirling Cycle Machines Using Two
h-CFD-Ts-315 thermocouple Technique, " AIAA 2nd International

I
-

_abs(h)-UMN Energy Conversion Engineering Conference

Providence, RI, USA during August 16-19, 2004_
o 60 120 180 240 300 360 Paper Number AIAA-2004-5559.
C.A. Degrees [14] Nan Jiang, N. and Terrence W. Simon, T.W. "Heat
Transfer and Fluid Dynamics Measurements in the
Figure 24. Comparison between CFD and UMN Expansion Space of a Stirling Cycle Engine, "
experimental data for the heat transfer coefficient for Paper IMECE2006-15631 Proceedings of
different C.A. degrees (Zero is when the displacer at IMECE2006 2006 ASME International Mechanical
TDC) for, Ts-315 Engineering Congress and Exposition November 5-
10, 2006, Chicago, Illinois, USA.
[15] FLUENT, Inc., 2005. Fluent 6.3 - User Guide.

11111