First International Conference on
Microchannels and Minichannels
April 24-25, 2003, Rochester, New York, USA
ICMM2003-1016
KEYNOTE PAPER
CRITICAL HEAT FLUX IN MICROCHANNELS: EXPERIMENTAL ISSUES AND GUIDELINES
FOR MEASUREMENT
A.E.Bergles1, [email protected]
S.G.Kandlikar2, [email protected]
1
Mechanical Engineering Department
Massachusetts Institute of Technology
Cambridge, MA 02139
2
Mechanical Engineering Department
Rochester Institute of Technology
Rochester, NY 14623
ABSTRACT
Critical heat flux limit is an important consideration in the
design of any flow boiling unit.
Before the use of
microchannels under flow boiling conditions becomes widely
accepted in critical applications, such as electronics cooling and
laser lenses, it is essential to develop CHF data for
microchannels. The experiments required to obtain this
information pose unique challenges as the channel dimensions
become smaller. The issues of parallel channel instability,
experimental control, experimental uncertainty, and conjugate
effects need to be carefully addressed. These issues are
addressed in the present paper, and guidelines helpful in the
design of CHF experiments are outlined.
Figure 1 Chip with integral cooling channels [3].
INTRODUCTION
GENERAL DESCRIPTION OF MICROCHANNEL HEAT
EXCHANGERS
that the channels are cut from one side, a cover plate is
provided on that side (Fig. 1), and inlet and exit headers couple
the microchannel heat exchanger to the flow system.
This paper will be directed toward microchannels, which
invariably involve cooling channels in blocks, as opposed to
mini and larger channels that are usually individual and
thermally well controlled.
Using commonly accepted
definitions of microchannels [1, 2], the hydraulic diameter
Dsubh will be in the range 10-200 micrometer. The length of
the flow passages, L, will be at least 10,000 micrometers. The
channels will usually be cut in a block; for silicon,
microelectronic fabrication techniques will be used [2],
whereas for copper or other metal, an end mill, or an EDM
process, or a lamination and bonding process will be used.
Typically, there will be of the order of 100 parallel channels.
Heat is usually supplied to one side of the block. Assuming
The result of this is that two problems experienced in
conventional heat exchangers are aggravated. The first of these
is flow distribution among many parallel channels – a
particularly serious concern with boiling and evaporating
flows. The second is conjugate effects, circumferential and
axial heat conduction in the material forming the channel.
These complications make it extremely difficult to ascertain the
flow in each channel, and it is virtually impossible to measure
the heat flux and temperature distributions around the periphery
of each channel. The microchannel heat exchanger is perhaps
best treated as a lumped device, unless appropriate microscale
instrumentation can be developed [2].
141
Copyright © 2003 by ASME
There appear to be three cases of interest*:
boundary conditition), and the CHF values re similar to those
for uniform heat flux.
1) Low, thermosyphon-driven flow, with vaporization – as
proposed for heat sinks used for cooling of electronics;
high heat flux
Standard CHF data are not available for microchannels,
because it is impossible to fabricate thin-walled test sections
with such small flow cross-sections. The best that can be done
is to look for CHF data in small diameter tubes (mini channels).
Extensive experiments were conducted with subcooled boiling
in tubes as small as D = 0.3 mm (300 micrometer). As shown
in Fig. 2, CHF increases continuously as the diameter is
reduced to this value, other conditions remaining the same.
Other parametric trends of subcooled CHF are given by
Kandlikar [7].
2) Low, forced flow, with vaporization – as proposed for
refrigerant evaporators; low heat flux
3) High, forced flow, with subcooled boiling – as might be
used for cooling of high power electronics; high heat
flux.
∗It should be pointed out, though, that the high pressure drop
with vaporization in true microchannels may preclude practical
realization of 1) and 2)
120
G = 40,000 kg/m^2-s
G = 25,000 kg/m^2-s
100
(q/A) * 1E-6 (W/m^2)
It should be noted that boiling is desirable with heat sinks for
cooling of electronic components, because the wall temperature
is more likely to be uniform. This is due to the fact that it is
constrained to the saturation temperature or either subcooled or
bulk boiling. In single-phase flow, the wall temperature would
track the fluid temperature, which may rise greatly, due to the
high heat input. Of course, for phase change, high pressure
drop should be avoided, as it results in a drop in the saturation
temperature along the channel.
G = 10,000 kg/m^2-s
80
o
∆ Tsub = 55.0 C
P = 0.60 Mpa
L/D = 20.0-25.0
60
40
20
0
0.00E+00
5.00E-04
1.00E-03
1.50E-03
2.00E-03
2.50E-03
Diameter in Meters
The occurrence of critical heat flux (CHF), in any of these
situations, must be regarded as an undesirable condition, as it
will cause overheating of an individual channel or even the
entire substrate containing the microchannels.
Figure 2 Dependence of CHF on channel diameter for water
[6]
NORMAL CHF
IS THERE CHF IN MICROCHANNEL HEAT
EXCHANGERS?
Several hundred thousand CHF points have been reported in
the boiling literature of the past fifty years. These data were
overwhelmingly obtained with stable flow (see below) in
single, thin-walled, circular tubes. The tubes were usually of
uniform wall thickness, and direct electrical heating was
utilized to simulate the constant heat flux boundary condition.
The electric power (heat flux) was increased slowly until a
vapor blanketing occurred, as evidenced by physical burnout of
the tube or activation of a burnout protection device by the
increased temperature. In the latter case, the power to the tube
was rapidly disconnected before burnout occurred. Numerous
correlations have been proposed for subcooled exit conditions
(e.g., [4]) and for boiling with net vapor generation (e.g., [5]).
There can, no doubt, be CHF in an individual microchannel,
but, as pointed out above, this situation does not exist. In a
microchannel heat exchanger, the conjugate effects play a
major role. Vapor blanketing of the portions of the channels
near the heat input will lead to a major redistribution of the heat
flow toward the opposite portions of the channels. If the
portions of the channels not vapor-blanketed can efficiently
accommodate the heat transfer, there will be no CHF or
burnout condition.
Some rudimentary, two-dimensional,
conjugate solutions are presented by Bergles et al. [8].
CHF WITH CONJUGATE EFFECTS
Some studies of CHF have considered axial and/or
circumferential flux tilts, usually by machining the tube. The
critical heat flux condition is as well-defined as with uniform
heating. Fluid heating of uniform-wall tubes has also been
used (essentially simulating the uniform wall temperature
Although conjugate effects have been considered with singlephase flow in Rectangular microchannels [9, 10], there do not
appear to be any studies that consider conjugate effects with
high flux boiling in true microchannels. Failing that, it is
142
Copyright © 2003 by ASME
instructive to look at the substantial literature on CHF in
channels with one-sided heating, as in the plasma-facing
components (PFCs) for the International Thermonuclear
Experimental Reactor (ITER). These studies have all been
done with circular tubes, and most involve single tubes with
varying wall thickness to create the flux tilts. Of more interest
are the circular tubes surrounded by a large block, heated over
part of the outside perimeter.
1) Constant flow in each of the parallel channels. There
is an inability of the conjugate effects to handle the
heat input
2) Parallel-channel excursive instability. Some channels
void out, and the heat is transferred to the other channels. The
heat input is high enough that the other channels cannot
accommodate the heat transfer through the conjugate effects,
and they also experience vapor-blanketing. The highest heat
flux will be accommodated in the first case. The situation for
1) is discussed above.
Recalling that a uniform flow
distribution is required for that case, an appropriate distributor
is required.
Boyd et al. [11] described a massive cylindrical test section
heated over 180 deg. by five resistance heaters.
Thermocouples placed at 48 stations give the three-dimensional
distribution of the wall temperatures and heat fluxes. See Fig.
3. The system is very complex, but is considered capable of
obtaining boiling curves around the channel circumference,
including the local CHF. Earlier in the program, it was
postulated that three different flow regimes could coexist
within the channel: single-phase turbulent flow, subcooled flow
boiling, and subcooled film boiling. Recently, Boyd et al. [12]
demonstrated this coexistence.
FLOW DISTRIBUTION
As pointed out above, uniform flow is desirable in a
microchannel heat exchanger. Assuming that the pressures in
the inlet and exit headers are uniform, the flow in each channel
is related to the friction, momentum, and gravitational terms of
the pressure drop. For boiling flows, this is strongly dependent
on the applied heat flux. The situation is especially difficult
with a two-phase mixture entering the heat exchanger. The
development of microchannel evaporators for refrigeration
systems will require an enabling technology: a small-scale
refrigerant distributor that will ensure uniform flow/quality to
each channel. Several configurations are noted by Bergles et
al. [8]; however, they are for conventional-size systems, and it
is impossible to configure them to handle the large number of
small channels in a microchannel heat exchanger. Two
approaches have been tried for minichannel and microchannel
heat exchangers: separation of the gas and liquid after the
expansion valve, and bypass of the gas, so that only liquid
(which is easier to distribute) flows through the evaporator
[13]; transformation of the inlet flow to a mist flow, which can
be distributed much more uniformly [14]. It is noted, however,
that even all-liquid channels may be subject to flow
maldistribution [15].
Figure 3 Test section used for local temperature and heat
flux measurements in a PFC simulation [11].
FLOW INSTABILITY
The same phenomenon is expected in blocks with
microchannels, the difference being that the small size makes it
impossible to install a large number of temperature sensors. It
is strongly suspected that the local heat transfer coefficients
cannot be calculated from the few parameters that are
reasonably assumed to be known, namely, the wall heat flux
and one temperature at the channel base near the heat input.
Microchannel heat exchangers invariably consist of an array of
parallel channels conveying the coolant. The temptation is to
design these channels using heat transfer and pressure drop
data for single channels that have flow imposed. When
subcooled boiling is involved, however, this procedure is not
valid, because it fails to capture the excursive or Ledinegg
instability that results from the unique pressure drop
characteristic of a boiling channel.
GENERAL CONSIDERATIONS
The pressure-drop characteristics of a single channel (including
entrance and exit losses) are necessary to predict this
“hydrodynamic” instability. Typical data for this “demand”
curve” are shown in Fig. 4; the pressure drop is given as a
It seems there are two limiting cases of CHF-induced failure in
microchannels:
143
Copyright © 2003 by ASME
function of mass flow rate for a fixed geometry, but varying
inlet temperature, heat flux, and exit pressure. The pressure
drop starts out lower than the isothermal value for pure liquid,
because of the reduced viscosity at the wall with heating. As
the flow rate is reduced, subcooled boiling is initiated,
whereupon the curve starts to turn around. A well-defined
minimum is observed, and the curve rises sharply until
subcooled CHF is observed at “d.” If the experiment were not
terminated by CHF, the curve would rise further, passing the
saturated exit condition, eventually reaching a maximum and
joining the all-vapor curve.
will be realized by creating a more uniform flow distribution
throughout the heat exchanger. Although this is a challenge,
improved microfabrication techniques may make it possible to
build flow restrictions at the inlet of each channel. Each orifice
must be identical. Since the orifice will necessarily be smaller
than the channels, it will require a tighter manufacturing
tolerance than the channel itself.
OTHER INSTABILITY
When there is a significant compressible volume upstream of
the heated section, may be an oscillating flow leading to CHF
[17]. This could be caused by an entrained air bubble or a
flexible hose, but for small channels, the compressibility of a
large volume of degassed liquid is sufficient to cause the
instability.
Even with stable inlet flow, there is a possibility that there can
be oscillatory flow caused by the internal compressibility of the
large L/D channels. While this has been demonstrated in
conventional-size channels [16], there is less evidence for this
behavior in very small channels.
OBSERVATIONS OF INDIVIDUAL CHANNELS OR AN
ARRAY OF CHANNELS
Since it is obviously difficult to observe two-phase flows in
microchannels, most studies have involved minichannels that
should be applicable to the smaller channels.
Figure 4 Flow-rate dependence of subcooled boiling
pressure drop for water, used to illustrate excursive
instability [16]
The flow pattern map of Tabatabai and Faghri [18] is the most
comprehensive that has been reported for tubes small enough to
be strongly influenced by surface tension. Nuno et al. [19]
observed different flow patterns in different channels of a
multichannel (minichannel) array with a two-phase mixture
entering the array. This is due to flow maldistribution among
the channels.
The key to the parallel-channel behavior is the minimum. If
the “supply” curve is that of a centrifugal pump, “A,” there will
be two intersections on the “demand” curve: “a” and “b.”
Actually, the intersection shown is not possible, since operation
beyond the minimum cannot occur; at the minimum, a firstorder instability occurs, since a perturbation in flow rate
(assumed negative) causes an excursion to point “a.” The heat
flux for this “hydrodynamic” instability is considerably lower
than the
heat flux that would be obtained with a stabilizing
pressure drop at the tube inlet, so that the “supply” curve
(pump plus inlet throttle) has a steeper negative slope than the
“demand” curve, “B.”
Kandlikar et al. [20], Hetsroni et al. [21], and Brutin et al. [22]
report flow pattern and pressure drop fluctuations in
multichannel evaporators. Their geometries were, respectively,
6-1 mm x 1mm square channels, 60 mm long; about 24
triangular channels of hydraulic diameter about 0.12mm, 15
mm long; and unspecified number and cross-section, 50-200
mm long. Kandlikar et al. [20] observed reversal of part of the
flow in some of the channels. The internal compressibility
could cause flow instability, as noted above. An upstream
compressible volume (entrained gas, flexible tubing), between
the pump and the heat exchanger, could also be responsible for
this behavior. The apparent flow reversal appears to be due to
rapid growth of a bubble, which pushes the liquid both
upstream and downstream [20]. Instability of this magnitude
Prediction of the minimum in the channel pressure drop v. flow
rate curve is essential, but data have been obtained only for
minichannels, not microchannels. Of course, one of the main
problems in determining the “hydrodynamic CHF” is the
unknown flow rate in each channel. This complexity, once
again, argues for considering microchannel heat exchangers as
lumped-parameter devices. In any case, higher performance
144
Copyright © 2003 by ASME
will certainly lead to CHF. It is noted, though, that this
expected to be more prevalent at low flow rates.
It should be noted that a positive-displacement pump is usually
used to provide the small flow rates required by a
microchannel. This means that the total flow rate will be
constant. If some of the channels are ineffective, the flow rate
will increase in the other channels. The difficulty, though, is
that part of the heat sink is then inactive; hence, the electronic
device could overheat. This leads to the necessity for
observations of the entire heat sink.
GLOBAL OBSERVATIONS
Observations of the entire heat sink are difficult, because they
must be through the transparent cover plate, assuming the
opposite side is covered with the electronic device or an
electrical heater. Looking through the cover plate with highspeed digital video imaging, Hetsroni et al. [21] found that the
fluctuations led to annular flow with no apparent liquid film.
The evaporating liquid film was there, however, with its
efficient cooling, as infrared radiometry (on the side with the
thin-film resistor) did not indicate a sharp increase in the block
temperature.
Figure 5 Boiling curves for microchannel at various inlet
subcooling [23].
The channel surface temperature and heat flux would be
desirable; however, thermocouples or other sensors are not
available to get the local behavior. The smallest practical
thermocouple (36 ga) has a bead size exceeding the upper limit
of channel size (100 micrometer gap for a rectangular channel
of hydraulic diameter equal to 200 micrometer). It is still
desirable to get local temperatures rather than the average for
the whole block, as done in [23].
Hence, multiple
thermocouples should be placed in the heated side, preferably
at the exit of the channel. This will make it possible to sense
whether some channels, or the whole block, are going into
CHF. As microfabrication techniques improve, it should be
possible to get a better indication of temperatures in a
microchannel block.
For instance, the techniques for
microscale thermometry mentioned by Cahill et al. [24] may
someday be adapted to microchannels. At the moment, the
probing is done externally, and the cover plate, as well as the
flowing liquid, would inhibit the measurements.
In a pioneering study of boiling heat sinks, Bowers and
Mudawar [23] installed a Single thermocouple that essentially
monitored the average temperature of the 10mm x 10 mm
copper block with 17-510 micrometer diameter circular
channels. A well-defined CHF was observed, as shown in Fig.
5, suggesting that all channels reached the critical condition.
INSTRUMENTATION
The experience of previous investigators provides meaningful
guidance for the CHF instrumentation of microchannel heat
exchangers. Visual observations of the boiling channels
through a transparent cover plate are useful to infer flow
patterns, flow stagnation, and flow reversal. Given the small
dimensions, though, magnification must be used. It is noted
that conventional magnification may be inadequate, for
instance, in identifying incipient subcooled boiling [8, 20].
Furthermore, the visualization may not distinguish important
features, such as the evaporating film in annular flow [21].
An even more critical problem is to measure the flow in each
channel to determine the degree of maldistribution. There are
no sensors small enough to do this at present; however, carbon
nanotube bundles may have promise for this application [25].
As pointed out by Garimella and Sobhan [26], optical flow
measurements with a “seeded” flow and particle image
velocimetry hold promise. One should not rule out the
possibility of using microfluidic valves to increase the inlet
resistance of the microchannels, thereby stabilizing the flow
[27].
In terms of quantitative measurements, the heat input to the
channel should be accurate. This usually means correcting for
the heat losses to the ambient from actual devices or electrical
heaters. There is little that can be done except to base the heat
flux on the channel area exposed to the fluid (excluding the
cover plate).
145
Copyright © 2003 by ASME
[6] Vandervort, C. L., Bergles, A. E., and Jensen, M. K., 1994,
“An Experimental Study of Critical Heat Flux in Very High
Heat Flux Subcooled Boiling,” International Journal of Heat
and Mass Transfer, 37, Suppl. 1, pp. 161-163.
NOMENCLATURE
P-Pressure,bars
∆P- Pressure drop, kPa
q- Heat flux, Wcm-2
Ts-Surface temperature, °c
Ti – Inlet temperature, °c
w- Mass flow rate, kg/s
[7] Kandlikar, S. G., 2000. “Critical Heat Flux in Subcooled
Flow Boiling – An Assessment of Current Understanding and
Future Directions for Research,” 2001, Multiphase Science and
Technology, 13, pp. 207-232.
[8] Bergles, A. E., Lienhard V, J. H., Kendall, G. E., and
Griffith, P., 2003, “Boiling and Evaporation in Small Diameter
Channels,” Heat Transfer Engineering, 24, No. 1, pp.18-40.
SUBSCRIPTS
i-Inlet
m- maximum (Critical heat flux)
sub- subcooled
[9] Phillips, R. J., 1990, “Microchannel Heat Sinks,” in
Advances in Thermal Modeling of Electronic Components and
Systems, Bar-Cohen, A., and Kraus, A. D., eds., ASME Press,
New York, pp. 109-184.
CONCLUDING REMARKS
[10] Rostami, A. A., Hassan, A. Y., and Chia, S. I., 2000,
“Conjugate Heat Transfer in Microchannels,” in Heat Transfer
and Transport Phenomena in Microscale, G. P. Celata, ed.,
Begell House, New York, pp. 121-128.
This paper has considered the issues associated with driving a
microchannel heat exchanger to the highest possible heat flux.
There are indeed very great opportunities for microtechnology.
This is especially the case in designing for operation near the
critical heat flux, and in devising instrumentation to study the
CHF condition.
[11] Boyd, Sr., R. D., Cofie, P., Li, Q.-Y., and Ekhlassi, A.
A.,2002, “A New Facility for Measurement of ThreeDimensional, Local Subcooled Flow Boiling Heat Flux and
Related Critical Heat Flux for PFCs,” Fusion Science and
Technology, 41, pp. 1-12.
REFERENCES
[12] Boyd, Sr., R. D., Cofie, P., and Ekhlassi, A.,
2002,“Conjugate Heat Transfer Measurements in a NonUniformly Heated Circular Flow Channel Under Flow Boiling
Conditions,” International Journal of Heat and Mass Transfer,
45, pp. 16051613.
[1] Mehendale, S. S., Jacobi, A. M., and Shah, R. K., 2000,
“Fluid Flow and Heat Transfer at Micro- and Meso-Scales with
Application to Heat Exchanger Design,” Applied Mechanics
Reviews, 53, No. 7, pp. 175-193.
[2] Kandlikar, S. G., and Grande, W. J., 2003, “Evolution of
Microchannel Flow Passages – Thermohydraulic Performance
and Fabrication Technology,” Heat Transfer Engineering, 24,
No. 1, pp. 3-17.
[13] Beaver, A. C., Hrnjak, P. S., Yin, J. M., and Bullard, C.
W., 2000, “Effects of Distribution in Headers of Micro-Channel
Evaporators on Transcritical COsub2
Heat Pump
Performance,” Proceedings of the ASME Advanced Energy
Systems
Division, AES-Vol. 40, ASME. New York, pp. 5564.
[3] Tuckerman, D. B., and Pease, R. F. W., 1981, “HighPerformance Heat Sinking for VLSI,” Electronic Device
Letters, EDL-2, pp. 126-129.
[14] Hrnjak, P., 2003, “Developing Adiabatic Two Phase Flow
in Headers – Distribution Issue in Parallel Flow Micro-Channel
Heat Exchangers.” Presented at First International Conference
in Microchannels and Minichannels, Rochester, NY, April 2425.
[4] Hall, D. D., and Mudawar, I., 2000, “Critical Heat Flux
(CHF) for Water in Tubes –II, Subcooled CHF Correlations,
International Journal of Heat and Mass Transfer, 43, pp. 26052640.
[15] Friedrich, M., 2002, Personal Communication, Pacific
Northwest Laboratory, July.
[5] Bowring, R. W., 1972, “A Simple but Accurate Round Tube
Uniform Heat Flux Dryout Correlation Over the Pressure
Range 0.7 MN/mexp2 (100-2500 psia), Report AEEW-R789,
Winfreth, UK.
[16] Boure, J. A., Bergles, A. E., and Tong, L. S., 1973,
“Review of Two-Phase Flow Instability,” Nuclear Engineering
and Design, 25, pp. 165-192.
146
Copyright © 2003 by ASME
[17] Daleas, R. S., and Bergles, A. E., 1965, “Effects of
Upstream Compressibility on Subcooled Critical Heat Flux,”
ASME Paper No. 65-HT 67, ASME, New York.
[18] Tabatabai, A., and Faghri. A., 2001. “A New Two-Phase
Flow Map and Transition Boundary Accounting for Surface
Tension Effects in Horizontal Miniature and Micro Tubes,”
Journal of Heat Transfer, 123, pp. 958-968.
[19] Nino, V. G., Hrnjak, P. S., and Newell, T. A., 2003, “TwoPhase Flow Visualization of R134A in a Multiport
Microchannel Tube,” Heat Transfer Engineering, 24, No.1, pp.
41-52.
[20] Kandlikar, S. G., 2002. “Two-Phase Flow Patterns,
Pressure Drop, and Heat Transfer during Boiling in
Minichannel Flow Passages of Compact Evaporators,” Heat
Transfer Engineering, 23, No. 1, pp. 5-23.
[21] Hetsroni, G., Moysak, A., and Segal, Z., 2001, “TwoPhase Flow in Microchannels,” Proceedings of the 2001 ASME
International Mechanical Engineering Congress and
Exposition, ASME, New York, November 11-16.
[22] Brutin, D. D., Topin, F., and Tadrist, L., 2001, “On
Thermo-Hydraulic Instabilities in Small Channels During Flow
Boiling, in Experimental Heat Transfer, Fluid Mechanics, and
Thermodynamics 2001, Celata, G. P., Di Marco, P., Goulas, A.,
and Mariani, A., eds., Edizioni, ETS, Pisa, Italy, pp. 273-289.
[23] Bowers, M. B. and Mudawar, I, 1994, High Flux Boiling
in Low Flow Rate, Low Pressure Drop Mini-Channel and
Micro-Channel heat Sinks,” International Journal of Heat and
Mass Transfer, 37, pp. 321-332.
[24] Cahill, D. G., Goodson, K., and Majumdar, A., 2002,
“Thermometry and Thermal Transport in Micro/Nanoscale
Solid-State Devices and Structures,” Journal of Heat
Transfer, 124, pp. 223-241.
[25] Ghosh, S., Sood, A. K., and Kumar, N., 2003, “Carbon
Nanotube Flow Sensors,” Science, 299, pp. 1042-1043.
[26] Garimella, S. V., and Sobhan, C. B., 2001, “Transport n
Microchannels – A Critical Review, in Annual Reviews in Heat
Transfer.
[27] Thorsen, T., Maerkl, S. J., and Quake, S. R., 2002,
“Microfluidic Large-Scale Integration,” Science, 298, pp.
580-584.
147
Copyright © 2003 by ASME