M.

Greiner
Two-Dimensional Simulations of
ASME Mem.
Professor of Mechanical Engineering,
University of Nevada,
Enhanced Heat Transfer in an
Reno, NV 89557
e-mail: greiner@[Link]
Intermittently Grooved Channel
P. F. Fischer Two-dimensional Navier-Stokes simulations of heat and momentum transport in an inter-
Mathematics and Computer Science Division, mittently grooved passage are performed using the spectral element technique for the
Argonne National Laboratory, Reynolds number range 600⭐Re⭐1800. The computational domain has seven contigu-
Argonne, IL 60439 ous transverse grooves cut symmetrically into opposite walls, followed by a flat section
e-mail: fischer@[Link] with the same length. Periodic inflow/outflow boundary conditions are employed. The
development and decay of unsteady flow is observed in the grooved and flat sections,
H. M. Tufo respectively. The axial variation of the unsteady component of velocity is compared to the
Department of Computer Science, local heat transfer, shear stress and pressure gradient. The results suggest that intermit-
University of Chicago, tently grooved passages may offer even higher heat transfer for a given pumping power
Chicago, IL 60637 than the levels observed in fully grooved passages. 关DOI: 10.1115/1.1459730兴
e-mail: hmt@[Link]
Keywords: Channel Flow, Computational, Enhancement, Forced Convection, Instability

Introduction of two. This suggests that three-dimensionality strongly affects the

transport characteristics of these flows, especially drag.
Engineering devices frequently employ enhanced heat transfer
Experimental measurements in a flat passage downstream of a
surfaces 关1兴. Fins are typically used to extend surface areas while
grooved channel were performed for Reynolds number range
offset strips are commonly used to promote thin boundary layers. 1500⭐Re⭐5000 关15,16,17兴. These measurements show that the
In recent years, a number of configurations that increase fluid heat transfer coefficient remained high for a substantial distance in
mixing by triggering flow instabilities have been considered. the flat region. The pressure gradient dropped down to the flat
Transversely grooved channel 关2–5兴 passages with eddy promot- passage value much more rapidly, especially for Re⬎2500. As a
ers 关6,7兴 and communicating channels 关8兴 all contain fairly large result, the heat transfer for a given pumping power was even
features whose sizes are roughly half the channel wall to wall greater in the first five hydraulic diameters of the decay region
spacing. These structures are designed to excite normally damped than in the grooved passage itself.
Tollmien-Schlichting waves at moderately low Reynolds num- Three-dimensional Navier-Stokes simulations in a flat passage
bers. downstream of a fully developed grooved channel were performed
The current authors have presented a series of articles on heat for 405⭐Re⭐764 关17兴. The grooved channel had transverse
transfer augmentation in rectangular cross section passages with grooves cut symmetrically into both walls. Two different compu-
contiguous grooves cut into the walls. Experimental flow visual- tational sub-domains were employed. The first represented one
izations in a long grooved channel downstream of a laminar flat periodicity cell of a continuously grooved passage. It had periodic
passage show that two-dimensional waves appear after an initial inflow/outflow boundary conditions in order to simulate fully de-
quiescent development length 关9兴. Unsteadiness is first observed veloped flow. The second sub-domain consisted of a single groove
thirty-five hydraulic diameters downstream of the first groove at a cell coupled to a flat passage at the downstream end. The inflow
Reynolds number of Re⫽350. As the Reynolds number is in- conditions to the grooved/flat sub-domain were taken from the
creased, the onset moves upstream and the flow behavior at a outflow of the fully developed domain. Unsteady flow from the
given location becomes increasingly three-dimensional. Experi- grooved region persisted several groove-lengths into the flat pas-
mental and numerical results in a passage with eddy promoters sage. This unsteadiness increased both local heat transfer and
indicate that the instability that leads to unsteady flow is convec- pressure gradient relative to steady flat passage flow. Moreover,
tive rather than absolute in nature 关10兴. Measurements using air the heat transfer for a given pumping power in the first three
show that fully developed heat transfer is enhanced relative to groove-lengths of the flat passage was even greater than the high
laminar flat channel flow by as much as a factor of 4.6 at equal levels observed in a fully developed grooved passage. However,
Reynolds numbers and by a factor of 3.5 at equal pumping powers the numerical Nusselt number decayed more rapidly in the flat
关11,12兴. passage than was expected from measurements.
Numerical simulations of fully developed convection in trans- The favorable heat transfer versus pumping power performance
versely grooved passages were performed using the spectral ele- of flat passages downstream from grooved channels suggests that
ment technique for Re⭐2000 关13,14兴. Those simulations em- intermittently grooved passages, in which flat regions separate
ployed three-dimensional computational domains that represented grooved sections, may have significant advantages in engineering
one periodicity cell of the contiguously grooved passage. The heat transfer devices. However, the development of unsteady flow
pressure gradient and heat transfer results were within 20 percent in grooved regions as well as the decay of unsteady flow exiting
of the measured values. At Re⫽1000 two-dimensional simulations from a short grooved section must be investigated before the de-
gave Nusselt number values that were 20 percent below three- sign of intermittently grooved passages can be optimized.
dimensional results while friction factors were smaller by a factor The current work is a two-dimensional numerical investigation
of heat transfer in an intermittently grooved passage for the Rey-
Contributed by the Heat Transfer Division for publication in the JOURNAL OF
nolds number range 600⭐Re⭐1800. The grooved portions of this
HEAT TRANSFER. Manuscript received by the Heat Transfer Division February 6, passage have seven right-triangular slots cut symmetrically into
2001; revision received October 17, 2001. Associate Editor: M. Faghri. opposite walls. The flat portion is also seven groove-lengths long

538 Õ Vol. 124, JUNE 2002 Copyright © 2002 by ASME Transactions of the ASME

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 Fig. 1 Computational domain and spectral element mesh

and its wall to wall spacing is the same as the minimum spacing in linear terms. Then, at each time step, the implicit Stokes problem
the grooved section. The study of an intermittently grooved pas- is solved with the nonlinear terms treated explicitly, in the absence
sage requires a very long computational domain. As pointed out of any mean pressure gradient. To this second solution, we add a
earlier, three-dimensional simulations predict the pressure gradi- multiple of the first that yields the desired flow rate. The multi-
ent in fully developed grooved passages much more accurately plier corresponds to the mean pressure gradient.
than two-dimensional calculations 关13,14兴. However, the re- The thermal problem for the periodic domain requires careful
sources to perform three-dimensional simulations in this large do- treatment. If one simply specifies zero-temperature conditions on
main are not available at the current time. The current two- the walls then the solution eventually decays to zero. To produce
dimensional simulations afford an opportunity to learn more about the desired spatially fully-developed state requires that the tem-
this flow and provide guidance to future three-dimensional perature profiles at the inlet and outlet be self-similar, that is,
calculations.
T 共 x⫽L d ,y,t 兲 ⫽C T 共 x⫽0,y,t 兲 , (1)
Numerical Method with T⭓0 and C⬍1. The solution technique for computing the
fully developed temperature field for constant temperature bound-
Computational Domain. Figure 1 shows the two- ary conditions follows the analysis of Patankar et al. 关23兴. The
dimensional spectral element mesh employed in this work. The energy equation 共neglecting viscous dissipation兲 and associated
upper and lower boundaries are solid walls, and the flow is from initial and boundary conditions are
left to right. The domain consists of seven grooves, each of length
⳵T
b⫽0.024 m and depth d⫽0.012 m, followed by a flat section of ជ •ⵜT⫽ ␣ •ⵜ 2 T
⫹U (2a)
length 7b. The total domain length is L d ⫽14b and the minimum ⳵t
passage wall to wall spacing is H⫽0.01 m. The groove length, b,
was chosen to be compatible with the wavelength of the most T 共 x,y,t⫽0 兲 ⫽T init共 x,y 兲 (2b)
slowly decaying Tollmien-Schlichting waves of the outer channel T 共 x,y,t 兲 ⫽0 on the walls (2c)
flow 关2兴. Moreover, the groove and passage dimensions are the
⫺cL d
same as the geometries studied in our earlier work on decaying T 共 x⫽L d ,y,t 兲 ⫽e T 共 x⫽0,y,t 兲 (2d)
unsteadiness downstream of a grooved passage 关12,16,17兴. How- ជ ⫽(u, v ) is the convecting velocity field determined by
ever, the current domain uses periodic inlet/outlet boundary con- where U
ditions and thus models fully developed flow in an array of alter- the hydrodynamic part of the computation. Equation (2d) corre-
nating grooved and flat channels. sponds to the fully developed condition where the temperature
In the spectral element method 关18,19兴 the velocity, data and profile is self-similar in each successive domain in the periodic
geometry are expressed as tensor-product polynomials of degree sequence, that is T(x⫹L d ,y,t)⫽e ⫺cL d T(x,y,t) for all (x,y,t),
N in each of K quadrilateral spectral elements, corresponding to a where e ⫺cL d ⫽C. The constant c is unknown and is a parameter to
total grid point count of roughly KN 2 . Numerical convergence is be determined as part of the computation. The fact that each do-
achieved by increasing the spectral order N. The present calcula- main independently satisfies the homogeneous set in Eq. 共2兲 and
tions were carried out at a base resolution of K⫽1960, N⫽7 共Fig. that we are considering fully developed solutions that are indepen-
1 shows the K spectral elements but not the KN 2 grid points兲. dent of T init implies that the solution to Eq. 共2兲 for each domain
Resolution tests were performed for Re⫽1200 and Re⫽1800 at would yield the same value of c. Hence, c cannot be a function of
N⫽8 and N⫽9, respectively. The present simulations use consis- x. Moreover, an energy balance on the computational domain
tent approximation spaces for velocity and pressure, with pressure shows that c is proportional to the log-mean Stanton number.
represented as polynomials of degree N⫺2 关19,20兴. The momen- Since the Stanton number is constant c is not function of time
tum equations are advanced by first computing the convection even when the flow is itself unsteady.
term, followed by a linear Stokes solve for the viscous and pres- Any function satisfying the above self-similar condition has the
sure terms. The decoupling allows for convective Courant num- unique decomposition T(x,y,t)⫽e ⫺cx ␪ (x,y,t), where ␪ (x
bers greater than unity while maintaining second-order accuracy ⫹L d ,y,t)⫽ ␪ (x,y,t) is a periodic function. Thus, the computa-
in time. Full details of the method can be found in 关20兴. tion of T is reduced to the computation of the periodic function ␪,
and the constant c. Substituting this decomposition into Eq. 共2兲
The Periodic Domain. The flow is driven from left to right in yields:
the periodic domain by applying a uniform body force. The level
⳵␪ ⳵␪
of forcing is adjusted at each time step to ensure that the mass ជ •ⵜ ␪ ⫺ ␣ •ⵜ 2 ␪ ⫽ 共 ␣ •c 2 ⫹uc 兲 ␪ ⫺2 ␣ •c
⫹U (3a)
flow rate through the domain is invariant with time. The approach ⳵t ⳵x
is outlined in Ghaddar et al. 关21兴 and Fischer and Patera 关22兴. It
exploits the linearity of the implicit Stokes problem. One first ␪ 共 x,y,t⫽0 兲 ⫽ ␪ init共 x,y 兲 (3b)
computes, in a preprocessing step, velocity and pressure fields for ␪ 共 x,y,t 兲 ⫽0 on the walls (3c)
the Stokes problem that result from application of unit forcing,
corresponding to a mean pressure gradient, in the absence of non- ␪ 共 x⫽L d ,y,t 兲 ⫽ ␪ 共 x⫽0,y,t 兲 (3d)

Journal of Heat Transfer JUNE 2002, Vol. 124 Õ 539

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 Fig. 2 Snapshot contour plots of periodic temperature ␪ at ReÄ600, 1200 and 1800

Since the fully developed solution is independent of the initial end of each trial period, we adjust c until convergence is attained.
condition we may arbitrarily assign ␪ init , which is typically set to Typical values of cL d over the range of Re considered are 0.55 to
unity when starting from rest, or to a prior converged result when 1.0, corresponding to 42 percent to 63 percent drops in mean
starting from an existing flow-field. Equation (3a) is solved using temperature over the domain length.
a semi-implicit time-stepping procedure similar to that of our The simulations at Re⫽600 were initialized using u⫽0 and ␪
Navier-Stokes solver. The diffusive terms are treated implicitly ⫽1. Subsequent cases were initialized from converged results at
while the convective terms are treated explicitly. In addition, all lower Reynolds numbers. Because of the extreme length of the
terms on the right of Eq. (3a) are treated explicitly using the domain 共14b, versus b for our earlier computations 关13,14兴兲, very
latest available value for c. long time integrations were required to reach a quasi-steady-
In the steady state case ( ⳵ / ⳵ t⫽0), Eq. 共3兲 constitutes an eigen- periodic state at the higher Reynolds numbers. For example, the
problem for the eigenpair (c, ␪ ). The constant c corresponds to Re⫽1800 case was initiated from the Re⫽1200 final solution and
the decay rate of the mean temperature in the x-direction. As such, run for a physical time of 2.9 sec 共corresponding to 9.6 convective
a larger value of c implies more rapid decay and more effective passages through the domain based on mean flow-rate and domain
heat transfer 共larger Stanton number兲. In the convection- length兲 before ‘‘steady state’’ statistics were calculated. The sta-
dominated limit where the Peclet number U m D h / ␣ is large, Eq. tistics were calculated based on a simulation with spectral order
(3a) becomes a linear eigenvalue problem. In this case, standard N⫽9 and a time step of 0.000016 sec 共corresponding to a con-
iterative methods for computing the lowest value of c 共corre- vective Courant number of 3.0兲. Solution files were extracted ev-
sponding to the most slowly decaying mode in x兲 can be used ery 100 time steps, and a total of 580 such files were used to
even when the nonlinear (c 2 ) term in Eq. (3a) is not identically compute the time-averaged and rms data. This sampling rate cor-
zero. We find that this method accurately computes the decay rate responds to roughly 50 samples per oscillation in the solution
and Nusselt numbers for steady flows in square and round ducts signal.
关24兴. Adequacy of the spatial resolution was determined by compar-
For steady-periodic flows with period ␶, the temperature is pe- ing the results for the most difficult case (Re ⫽ 1800) computed
riodic in time, implying T(x,y,t⫹ ␶ )⫽T(x,y,t). Since c is inde- using polynomial order N⫽7 with those obtained using N⫽9.
pendent of time, this implies that ␪ (x,y,t⫹ ␶ )⫽ ␪ (x,y,t). If the Figures 4 and 5 show the axial variations of the fluctuating veloc-
value of c is not chosen correctly, this condition will not be sat- ity and the bulk Nusselt number. The results for N⫽7 and 9 are
isfied. Unfortunately, ␶ is not known a priori but is a result of the virtually indistinguishable. The maximum difference in the fluc-
hydrodynamic part of the calculation. A robust approach to com- tuating velocity is less than 4 percent of the average value. The
puting c is obtained by multiplying Eq. (3a) by ␪, integrating Nusselt number is based on the derivative of temperature. The
over the domain ⍀, and simplifying to yield: maximum difference is always less than 3 percent, except at the
singular points located at the groove peaks. Given the unsteady
1 d
2 dt 冕⍀
␪ 2 dV⫽ 冕
⍀
关共 ␣ c 2 ⫹uc 兲 ␪ 2 ⫺ ␣ ⵜ ␪ •ⵜ ␪ 兴 dV (4)
nature of these flows, this close agreement indicates both adequate
spatial resolution and sufficiently long time integration to obtain
meaningful statistics.
While we do not expect the time derivative of the average tem- The simulations were performed on 8, 16, and 32 processors of
perature 共represented by the left-hand side of Eq. 共4兲兲 to be iden- a 96 processor SGI Origin2000. Each processor is a MIPS
tically zero, it will in general be less than the time derivative of ␪ R10000 running at 250 MHz and shares 24 GB of memory. The
at any one point in the domain. Moreover, if we integrate the Re⫽1800, N⫽9 computation required 2.5 CPU sec/step on 32
right-hand side of Eq. 共4兲 from time t to t⫹ ␶ , the resultant quan- nodes.
tity must be zero due to the temporal periodicity.
This suggests a two-tier strategy for computing c in the un-
steady case. Initially, we determine c such that the right hand side Results
of Eq. 共4兲 is identically zero at each time step. This permits a Figure 2 shows three contour plots of the dimensionless peri-
relatively coarse but quick determination of c and ␪. Once ␶ is odic temperature ␪. These plots are typical snapshots at Reynolds
well established, we use this value of c to advance ␪ for one or numbers Re⫽600, 1200, and 1800. In this work the Reynolds
more periods, and monitor the decay or growth of 兰 ␪ 2 dV. At the number is Re⫽UmDH /␯. The average velocity through the mini-

540 Õ Vol. 124, JUNE 2002 Transactions of the ASME

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 Fig. 3 Contour plots of the root-mean-squared component of periodic temperature ␪ rms at ReÄ600, 1200, and 1800

mum channel cross section is U m ⫽( 兰 ⍀ udA)/(14bH), where the grooved portion of the domain, while 7⭐x/b⭐14 represents the
integration is taken over the entire area of the computational do- flat section. Two traces are included for Re⫽1800 with spectral
main, ⍀. The minimum channel hydraulic diameter is D H ⫽2H, orders of N⫽7 and 9.
and ␯ is the fluid kinematic viscosity. Other researchers have used At Re⫽600, the velocity unsteadiness reaches maximum values
the Orr-Sommerfeld definition of the Reynolds number, which is of less than 2 percent near the end of the grooved section. For
ReOS⫽(3/8)Re 关2兴. Re⫽1200, the unsteadiness grows in the first four grooves (0
For Re⫽600, temperature contours lines in the open passage ⭐x/b⭐4), drops off slightly in the next groove (4⭐x/b⭐5),
are virtually parallel to the x-axis. Streamline plots 共not shown兲 and then increases in the two last grooves. The unsteadiness re-
indicate that the central portion of the passage has essentially no mains near the high values observed in the grooved section for the
transverse motion and the grooves contain slowly turning vortices. first half-groove-length of the flat region. It drops off very rapidly
The effect of the vortices on the contour lines in the grooves is for the next two groove-lengths and then decreases at a much
evident. In contrast, long contiguously grooved channels exhibit slower rate. For Re⫽1800, the unsteadiness grows rapidly in the
two-dimensional waves for Re⭓350 关9,13,14兴. The steady behav- first three grooves, drops off in the fourth groove, grows again in
ior of the current intermittently grooved passage indicates that the the next two grooves and then drops off slightly in the last groove.
development length for unsteady flow at Re⫽600 is longer than Once again the unsteadiness remains high in the first half-groove-
the groove section length, 7b. This result is consistent with the length of the flat region before dropping off. Time dependent
convective nature of the instability 关10兴. streamline plots show that the velocity field exhibits a traveling
At Re⫽1200 a wavy structure develops in the third groove and wave structure. In Fig. 4, the local rises and dips in the time-
its amplitude grows in the x-direction. This transverse motion averaged data indicate that the flow field exhibits standing waves
persists for the remainder of the grooved section and for several as well.
groove-lengths into the flat region. At Re⫽1800, the transverse The thinner solid lines in Fig. 5 show bulk Nusselt number
motion is stronger and more irregular than it is at Re⫽1200. It versus axial location and Reynolds number. A dashed horizontal
develops more rapidly in the grooved section and decays more line in the region 7⭐x/b⭐14 shows the Nusselt number in a fully
slowly in the flat region. developed flat passage. The thicker line in the domain 5⭐x/b
Figure 3 shows contours plots of the root-mean-squared com-
ponent of the dimensionless periodic temperature ␪ rms , for Re
⫽600, 1200, and 1800. While the isotherms for Re⫽600 in Fig. 2
are nearly parallel to the x-axis, Fig. 3 shows that some unsteadi-
ness develops in the fifth groove and persists roughly three groove
lengths into the flat region. This unsteadiness is concentrated in
the region across the groove opening, and does not penetrate
deeply into the grooves. The contour plots for Re⫽1200 and 1800
show that as the Reynolds number is increased, unsteadiness ap-
pears closer to the first groove and it persists further into the flat
region. Moreover, the flow exhibits high levels of unsteadiness
deep in the grooves. For Re⫽1800 a significant level of unsteadi-
ness is present at the end of the flat section 共entrance to the groove
region兲.
Figure 4 shows dimensionless axial velocity unsteadiness
u ⬘ /U m versus location and Reynolds numbers. This unsteadiness
is defined as u ⬘ /U m ⫽(1/D h ) 兰 (u rms /U m )dy, where u rms is the
root-mean-squared deviation of the axial velocity from its local
time mean value, and the integration is taken from the bottom to Fig. 4 Unsteady component of axial velocity versus location
the top of the channel. The region 0⭐x/b⭐7 corresponds to the and Reynolds number

Journal of Heat Transfer JUNE 2002, Vol. 124 Õ 541

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 Fig. 6 Groove-length averaged Nusselt number versus loca-
tion and Reynolds number
Fig. 5 Bulk Nusselt number versus location and Reynolds
number

passage value after three groove lengths as its wall gradients ap-
⭐12 shows results at Re⫽640 from a three-dimensional simula- proach the fully developed levels. Examining the Re⫽1200 and
tion of a flat passage downstream of a fully developed grooved 1800 traces shows that both the level of heat transfer enhancement
channel 关17兴. and the length of the flat region where augmentation is observed
The bulk Nusselt number is defined as NuB ⫽(D h /k) increase with Reynolds number. The steeper velocity and tempera-
⫻( 具 q ⬙ /⌬T B 典 ), where the pointed brackets ( 具 典 ) indicate a time ture gradients as well as the higher levels of unsteadiness cause
average. In this expression the fluid thermal conductivity is k and these increases. For Re⫽1200 and 1800, the heat transfer drop off
⌬T B is the local temperature difference between the surface and is inflected one-half of a groove-length downstream from the
bulk fluid. The bulk temperature at any axial location x and time grooved section. This corresponds to the location where the ve-
t is defined as T B (x,t)⫽( 兰 uT dy)/( 兰 u dy), where both integra- locity unsteadiness in Fig. 4 begins to decrease. At Re⫽1800, heat
tions in the y direction are taken from the bottom to the top of the transfer enhancement persists for the full length of the flat region.
domain. The heat transfer to the fluid per unit projected surface The three-dimensional results for Re⫽640 共thicker line, Greiner
area is q ⬙ (x,t)⫽⫺k(dT/dn) wall /m, where T is temperature, n is et al. 关17兴兲 give heat transfer levels that are substantially higher
the direction normal to the wall, m is the wall surface direction than the current Re⫽600 data. As mentioned earlier, the three-
cosine, and the temperature gradient is evaluated at the wall. The dimensional results are for a flat passage downstream of a fully
direction cosine in the flat region is m⫽1, while it is m⫽0.7071 developed grooved channel. The flow in a fully developed
in the grooved region. Two traces are included for Re⫽1800 with grooved channel is unsteady for Re⭓350, while the current inter-
spectral orders of N⫽7 and 9. mittently grooved passage is essentially steady at Re⫽600. The
The strong singularities at x/b⫽0, 1, 2, 3, 4, 5, 6, and 7 are unsteady structure of the three-dimensional simulation increases
caused by the sharp edges of the groove peaks. At all other loca- the heat transfer level well beyond that predicted by the current
tions NuB increases with Reynolds number. The heat transfer in work for Re⫽600. In fact, its level is closer to the current Re
the first groove is very similar for all three Reynolds numbers. We ⫽1200 results. Moreover, the three-dimensional simulations do
see that the Nusselt number on the downstream 共windward兲 sur- not predict the inflection at x/b⫽7.5 or the secondary vortices in
face (0.5⭐x/b⭐1) is significantly higher than that on the up- the grooves.
stream 共leeward兲 side (0⭐x/b⭐0.5). This is due to the location Figure 6 shows the bulk Nusselt number averaged over differ-
and rotation direction of the groove vortex flow. The upstream ent groove-length regions of the domain. A thicker solid line
surface exhibits a local plateau centered at x/b⫽0.3. The groove shows the fully developed flat passage Nusselt number. The hori-
vortex impinges against the wall at that location. zontal line segments with solid squares represent fully developed
For Re⫽600, the shape of the Nusselt number profile is similar grooved channel results from two-dimensional simulations at Re
in all seven grooves. At Re⫽1200 the profile shape in the second ⫽600, 1200 and 1800 关14兴. Those simulations employed a com-
groove is similar to that in the first. Its shape then changes sub- putational domain that represents one periodicity cell of the
stantially in subsequent grooves and its magnitude increases. The grooved channel and periodic inlet/outlet boundary conditions.
heat transfer on the windward side of each groove is greater than For Re⫽600 the average Nusselt number increases slightly for
the level exhibited at Re⫽600. Moreover, the groove vortex im- the first four grooves and then reaches a fully developed value.
pingement plateau grows stronger in the third through fifth This value is below the flat passage level. The thermal resistance
groove. Its shape is essentially the same in the fifth, sixth and of the slowly turning groove-vortices causes this reduction in heat
seventh grooves, indicating that the heat transfer has approached transfer. For Re⫽1200, the average heat transfer in the first
its fully developed condition. The profile shape has a number of groove is the same as that at Re⫽600, but its value increases
local peaks especially on the leeward face. This suggests the time- substantially in the second, third and fourth grooves. After reach-
averaged flow field has small secondary vortices that impinge ing a local maximum the heat transfer drops slightly in the fifth
against the walls at the locations of the peaks. For Re⫽1800, the notch and then rises slightly in the sixth and seventh grooves.
Nusselt number profile in the second groove is substantially dif- However, the heat transfer level in the last four groove-lengths is
ferent form its shape in the first groove. Moreover, its shape does fairly uniform. Moreover, this level is greater than the value for a
not appear to stabilize until the sixth or seventh groove, although fully developed flat passage. The unsteady mixing at Re⫽1200 is
it is difficult to say whether it would continue to change if the sufficient to overcome the thermal resistance of the groove
grooved section were longer. vortices.
The heat transfer at the inlet of the flat region (x/b⫽7) is well At Re⫽1800 the heat transfer in the first groove is slightly
above the fully developed flat passage value for all three Reynolds higher than the level exhibited at Re⫽600 and 1200. This may be
numbers. Very steep velocity and temperature gradients near the caused by the unsteadiness at the exit of the flat section 共Figs. 3
walls cause this. The heat transfer for Re⫽600 drops to the flat and 4兲. The heat transfer exhibits a substantial rise in the next two

542 Õ Vol. 124, JUNE 2002 Transactions of the ASME

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 Fig. 7 Dimensionless axial shear stress versus location and Fig. 8 Dimensionless momentum flux gradient versus loca-
Reynolds number tion and Reynolds number

grooves. It drops in the fourth groove and rises for the next two drag characteristic that affects the pumping power per unit pas-
grooves before dropping slightly in the final groove. The average sage volume required of the prime mover, (dp/dx)U m . This
Nusselt numbers in all but the first groove are greater than the prime mover power is of great importance to many engineering
fully developed flat passage value. The rise and fall of heat trans- devices. We now relate the Fanning pressure gradient f P
fer in the interior grooves at Re⫽1200 and 1800 are closely cor- ⫽(⫺dF p /dx)(1/2␳ U m 2 ) to the wall shear stress. In this expres-
related with the unsteady velocity levels described in Fig. 4. sion the time averaged pressure force is F p ⫽ 具 兰 p dy 典 , where p is
At Re⫽600, the heat transfer in the seventh groove is substan- the local pressure and the integration is performed from the top to
tially less than the level predicted for fully developed flow. The the bottom of the channel. A time-averaged force balance on a
unsteadiness present in the fully developed simulation causes this control volume of axial length dx shows that the Fanning pressure
difference. At Re⫽1200 and 1800 the heat transfer in the seventh gradient is composed of wall shear stress and momentum flux
groove is, respectively, 2 percent and 7 percent higher than the gradient components, that is f P ⫽ f M ⫺ f S . The axial gradient of
fully developed values. the momentum flux is defined as f M ⫽d/dx 关 兰 ( 具 u 2 典 /U m 2 )dy 兴 ,
In the flat region (7⭐x/b⭐14), the heat transfer begins at very where 具 u 2 典 is the time average of the square of the x-velocity, and
high levels. It decreases as distance from the grooves increases, the integration is performed from the bottom to the top of the
eventually approaching the fully developed rate. The heat transfer passage.
development length in the grooved region 共as demonstrated by a Figure 8 shows the axial gradient of the momentum flux f M
rise in heat transfer兲 decreases as Reynolds number increases, versus location and Reynolds number. In the first groove, the mo-
while the heat transfer decay-length in the flat region increases. mentum flux gradient is similar for all three Reynolds numbers. In
As a result, the length of the region of enhanced heat transfer subsequent grooves, the magnitude of f M increases with Reynolds
increases with Reynolds number. numbers. This indicates that the fluid experiences large levels of
Figure 7 shows the dimensionless x-component of shear stress acceleration and deceleration. In the flat region downstream of the
f S ⫽2 ␶ x / ␳ U m 2 versus axial location and Reynolds number. In this grooves, the axial variation of the velocity is small and f M drops
expression the x-component of wall shear stress is ␶ x to zero very quickly. Comparing Figs. 7 and 8 共and noting the
⫽⫺ ␮ (du t /dn) wall /m, where u t is time average component of y-axis scales兲, we see that the magnitude of the momentum flux
fluid velocity tangential to the wall, ␳ is the fluid density, and ␮ is gradient is much larger than that of the shear stress in the grooved
the fluid dynamic viscosity. The strong negative singularities are regions, while the opposite is true in the flat portion of the pas-
caused by the sharp groove peaks at x/b⫽0, 1, 2, 3, 4, 5, 6, and 7. sage. The local Fanning pressure gradient is the difference be-
For Re⫽600, the shape of the shear stress profile in the grooves tween the traces in these figures, f P ⫽ f M ⫺ f S .
(0⭐x/b⭐7) has a number of similarities to the Nusselt number Figure 9 shows the product f P,avg Re versus location and Rey-
profiles seen in Fig. 5. For instance, the shear stress is signifi- nolds number. In this product, f P,avg is the Fanning pressure gra-
cantly higher on the downstream surface of each groove than it is dient averaged over different groove-length regions. The dotted
on the upstream side. Moreover, the shear stress exhibits a plateau
0.3-groove-lengths downstream from the leading edge of each
groove. These similarities are caused by the analogous behavior of
heat and momentum transport in the absence of strong pressure
gradients.
At Re⫽600 the shear stress is positive throughout the grooved
region 共with the exception of the singular points兲. This indicates
that the fluid near the groove surfaces is always moving in the
negative x-direction. For Re⫽1200 and 1800, on the other hand,
the shear stress is negative in certain locations. This implies that
fluid in these regions moves in the positive x-direction, indicating
the existence of secondary vortices, as described in connection
with Fig. 5. In the flat region downstream of the grooves, the
dimensionless shear stress is in the negative x-direction. The mag-
nitude of the shear stress in the flat region is much higher than it
is in the grooved region.
Comparing Figs. 5 and 7 shows that the wall shear stress is Fig. 9 Groove-length average Fanning pressure gradient ver-
analogous to heat transfer. However, the pressure gradient is the sus location and Reynolds number

Journal of Heat Transfer JUNE 2002, Vol. 124 Õ 543

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 horizontal line in the region 7⭐x/b⭐14 shows the fully devel- program of the Office of Advanced Scientific Computing Re-
oped flat channel value f P,avg Re⫽24. In the first groove f P,avg Re search, U.S. Department of Energy, under Contract W-31-109-
is negative for all three Reynolds numbers indicating that the di- Eng-38. The work of H.M. Tufo was supported by the Department
mensional pressure actually increases across this region. This of Energy under Grant number B341495 to the Center on Astro-
pressure rise is caused by fluid deceleration. The expansion of the physical Thermonuclear Flashes at University of Chicago, and by
channel cross section as the flow exits a flat passage and enters the the University of Chicago.
grooved region causes this deceleration. For Re⫽600, the friction
factor becomes positive in the second groove and continues to
increase with distance from the inlet of the grooved section. The Nomenclature
increment in f P,avg Re decreases with increasing x/b until the last b ⫽ groove length
groove. The pressure decrease across the last groove is larger than c ⫽ decay constant
in the internal grooves because the flow accelerates as it exits the d ⫽ groove depth
grooved section and enters the flat region. Dh ⫽ minimum hydraulic diameter, 2H
For Re⫽1200, f P,avg Re increases for the first five grooves, then fM ⫽ dimensionless momentum flux gradient
decreases in the next groove, and finally increases in the last fP ⫽ fanning pressure gradient
groove. The pressure gradient for Re⫽1800 increases for the first fS ⫽ dimensionless axial shear stress
four grooves, decreases in the next groove, and rises in the final H ⫽ minimum channel wall to wall spacing
two grooves. We see that the pressure gradient development k ⫽ fluid thermal conductivity, 0.0263 W/m°C
length in the grooved section decreases with Reynolds number. K ⫽ number of spectral elements
In the first groove-length of the flat region, f P,avg Re is very Ld ⫽ domain length, 14b
high for all three Reynolds numbers due to the high shear stress N ⫽ spectral element order
共sharp velocity gradients兲 at x/b⫽7. The pressure gradients then NuB ⫽ bulk Nusselt number based on projected area
approach their fully developed values after only three groove Pr ⫽ fluid molecular Prandtl number, 0.70
lengths. The pressure gradient in the last groove-length of the flat Re ⫽ Reynolds number, U m D h / ␯
region is below the fully developed flat passage value. This is due t ⫽ time
to the decelerating flow caused by the expanding cross sectional T ⫽ temperature
area at x/b⫽14. Tb ⫽ bulk temperature
Comparing Figs. 6 and 9, we see that while the heat transfer u, v ⫽ velocity components in the x and y directions
remains enhanced for up to six groove-lengths in the flat region, u⬘ ⫽ axial velocity unsteadiness
the pressure gradient drops back to the flat passage value in only Um ⫽ mean x-velocity at the minimum channel cross-
three groove lengths. Moreover, while the decay length for heat section
transfer increases with Reynolds numbers, the pressure gradient
decay length is rather insensitive to Reynolds number. Greek
␣ ⫽ thermal diffusivity, 2.63⫻10⫺5 m2 /s
Conclusions ␯ ⫽ fluid kinematic viscosity, 1.84⫻10⫺5 m2 /s
␪ ⫽ periodic temperature
Two-dimensional Navier-Stokes simulations of heat and mo- ␳ ⫽ fluid density, 1.006 kg/m3
mentum transport in an intermittently grooved passage were per-
␶ ⫽ period of local time variations
formed using the spectral element technique for the Reynolds
␶x ⫽ x-component of wall shear stress
number range 600⭐Re⭐1800. The computational domain had
⍀ ⫽ computation domain
seven contiguous transverse grooves cut symmetrically into oppo-
site walls, followed by a flat section of the same length. This
domain employed periodic inflow/outflow boundary conditions. References
The flow is essentially steady at Re⫽600. However, traveling
关1兴 Webb, R. L., 1994, Principles of Enhanced Heat Transfer, John Wiley & Sons,
waves develop near the inlet of the grooved section at Re⫽1200 New York.
and 1800 and persist several groove lengths into the flat region. 关2兴 Ghaddar, N. K., Korczak, K., Mikic, B. B., and Patera, A. T., 1986, ‘‘Numeri-
The axial variation of unsteady velocity within the grooved sec- cal Investigation of Incompressible Flow in Grooved Channels. Part 1: Stabil-
tion is closely correlated with increases in heat transfer, shear ity and Self-Sustained Oscillations,’’ J. Fluid Mech., 168, pp. 541–567.
关3兴 Greiner, M., 1991, ‘‘An Experimental Investigation of Resonant Heat Transfer
stress and pressure gradient. In the grooved region the develop- Enhancement in Grooved Channels,’’ Int. J. Heat Mass Transf., 24, pp. 1383–
ment lengths for heat transfer and pressure gradient both decrease 1391.
with increasing Reynolds number. While the length of the flat 关4兴 Roberts, E. P. L., 1994, ‘‘A Numerical and Experimental Study of Transition
region where heat transfer augmentation is observed increases Processes in an Obstructed Channel Flow,’’ J. Fluid Mech., 260, pp. 185–209.
关5兴 Lee, B. S., Kang, I. S., and Lim, H. C., 1999, ‘‘Chaotic Mixing and Mass
with Reynolds number, the pressure gradient returns to the flat Transfer Enhancement by Pulsatile Laminar Flow in an Axisymmetric Wavy
passage value in roughly three groove lengths for the entire Rey- Channel,’’ Int. J. Heat Mass Transf., 42, pp. 2571–2581.
nolds number range considered in this work. 关6兴 Kozlu, H., Mikic, B. B., and Patera, A. T., 1988, ‘‘Minimum-Dissipation Heat
The current two-dimensional simulations in an intermittently Removal by Scale-Matched Flow Destabilization,’’ Int. J. Heat Mass Transf.,
31, pp. 2023–2032.
grooved passage exhibit a number of differences from three- 关7兴 Karniadakis, G. E., Mikic, B. B., and Patera, A. T., 1988, ‘‘Minimum-
dimensional results in fully developed grooved channels and Dissipation Transport Enhancement by Flow Destabilization: Reynolds Anal-
downstream flat passages. However the current calculations will ogy Revisited,’’ J. Fluid Mech., 192, pp. 365–391.
provide guidance for future three-dimensional calculations in in- 关8兴 Amon, C. H., Majumdar, D., Herman, C. V., Mayinger, F., Mikic, B. B., and
Sekulic, D. P., 1992, ‘‘Experimental and Numerical Investigation of Oscilla-
termittently grooved passages. Finally, these results suggest that tory Flow and Thermal Phenomena in Communicating Channels,’’ Int. J. Heat
intermittently grooved passages may offer very favorable heat Mass Transf., 35, pp. 3115–3129.
transfer versus pumping power performance for engineering de- 关9兴 Greiner, M., Chen, R.-F., and Wirtz, R. A., 1990, ‘‘Heat Transfer Augmentation
vices. Through Wall-Shaped-Induced Flow Destabilization,’’ J. Heat Transfer , 112,
pp. 336 –341.
关10兴 Schatz, M. F., Tagg, R. P., Swinney, H. L., Fischer, P. F., and Patera, A. T.,
1991, ‘‘Supercritical Transition in Plane Channel Flow with Spatially Periodic
Acknowledgments Perturbations,’’ Phys. Rev. Lett., 66, No. 12, pp. 1579–1582.
关11兴 Greiner, M., Chen, R.-F., and Wirtz, R. A., 1991, ‘‘Enhanced Heat Transfer/
National Science Foundation Grant CTS-9501502 supported Pressure Drop Measured From a Flat Surface in a Grooved Channel,’’ J. Heat
this work. The work of P.F. Fischer was supported by the Math- Transfer, 113, pp. 498 –500.
ematical, Information, and Computational Sciences Division sub- 关12兴 Wirtz, R. A., Huang, F., and Greiner, M., 1999, ‘‘Correlation of Fully Devel-

544 Õ Vol. 124, JUNE 2002 Transactions of the ASME

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]
 oped Heat Transfer and Pressure Drop in a Symmetrically Grooved Channel,’’ 关19兴 Maday, Y., and Patera, A. T., 1989, ‘‘Spectral Element Methods for the Navier-
J. Heat Transfer, 121, pp. 236 –239. Stokes Equations,’’ State of the Art Surveys on Computational Mechanics, A.
关13兴 Greiner, M., Spencer, G., and Fischer, P. F., 1998, ‘‘Direct Numerical Simula- K. Noor and J. T. Oden, eds., ASME, New York, pp. 71–143.
tion of Three-Dimensional Flow and Augmented Heat Transfer in a Grooved 关20兴 Fischer, P. F., 1997, ‘‘An Overlapping Schwarz Method for Spectral Element
Channel,’’ J. Heat Transfer, 120, pp. 717–723. Solution of the Incompressible Navier-Stokes Equations,’’ J. Comp. Phys.,
关14兴 Greiner, M., Faulkner, R. J., Van, V. T., Tufo, H. M., and Fischer, P. F., 2000, 133, pp. 84 –101.
‘‘Simulations of Three-Dimensional Flow and Augmented Heat Transfer in a 关21兴 Ghaddar, N. K., Karniadakis, G. E., and Patera, A. T., 1986, ‘‘A Conservative
Symmetrically Grooved Channel with Constant Temperature Walls,’’ J. Heat Isparametric Spectral Element Method for Forced Convection: Application to
Transfer, 122, pp. 653– 660.
Fully Developed Flow in Periodic Geometries,’’ Numer. Heat Transfer, 9, pp.
关15兴 Greiner, M., Chen, R.-F., and Wirtz, R. A., 1995, ‘‘Augmented Heat Transfer
277–300.
in a Recovery Passage Downstream From a Grooved Section: An Example of
关22兴 Fischer, P. F., and Patera, A. T., 1992, ‘‘Parallel Spectral Element Solutions of
Uncoupled Heat/Momentum Transport,’’ J. Heat Transfer, 117, pp. 303–308.
关16兴 Huang, F., 1998, ‘‘Experimental Investigation of Fully-Developed Augmented Eddy-Promoter Channel Flow,’’ Proc. of the European Research Community
Convection in a Symmetrically Grooved Channel,’’ Masters of Science Degree on Flow Turbulence and Computation Workshop, Lausanne, Switzerland,
thesis, University of Nevada, Reno, NV, May 1998. Cambridge University Press, pp. 246 –256.
关17兴 Greiner, M., Fischer, P. F., Tufo, H. M., and Wirtz, R. A., 2002, ‘‘Three Di- 关23兴 Patankar, S. V., Liu, C. H., and Sparrow, E. M., 1977, ‘‘Fully Developed Flow
mensional Simulations of Enhanced Heat Transfer in a Flat Passage Down- and Heat Transfer in Ducts Having Streamwise Periodic Variations of Cross-
stream From a Grooved Channel,’’ J. Heat Transfer, 124共1兲, pp. 169–176. Sectional Area,’’ J. Heat Transfer, 99, pp. 180–186.
关18兴 Patera, A. T., 1984, ‘‘A Spectral Element Method for Fluid Dynamics; Laminar 关24兴 Kays, W. M., and Crawford, M. E., 1993, Convection Heat and Mass Transfer,
Flow in a Channel Expansion,’’ J. Comp. Phys., 54, pp. 468 – 488. 3rd edition, McGraw-Hill, New York.

Journal of Heat Transfer JUNE 2002, Vol. 124 Õ 545

Downloaded 16 Dec 2012 to [Link]. Redistribution subject to ASME license or copyright; see [Link]