Energy Conversion and Management 44 (2003) 2625–2634

www.elsevier.com/locate/enconman

Microscale combustion research for application to micro

thermophotovoltaic systems
a,*
W.M. Yang , S.K. Chou a, C. Shu a, H. Xue b, Z.W. Li a, D.T. Li c, J.F. Pan c

a
Department of Mechanical Engineering, National University of Singapore, 10 Kent Ridge Crescent,
Singapore 119260
b
Department of Mechanical Engineering, California State Polytechnic University, Pomona, CA 91768, USA
c
Faculty of Energy and Dynamics, Jiangsu University, Zhenjiang 212013, China
Received 30 August 2002; accepted 26 December 2002

Abstract
A novel power MEMS concept, a micro thermophotovoltaic (TPV) system, is first described in this work,
which would use hydrogen as fuel and would be capable of delivering 3–10 W electrical power in a package
less than 1 cubic centimeter in volume. A microcombustor is one of the most important components of a
micro TPV system. A high and uniform temperature distribution along the wall of the microcombustor is
required to get a high electrical power output. However, sustaining combustion in a MEMS size combustor
will be largely affected by the increased heat losses due to the high surface to volume ratio, which tends to
suppress ignition and quench the reaction. In order to test the feasibility of combustion in microdevices and
determine the relevant factors affecting microcombustion, numerical and experimental work was per-
formed. The results indicated that a high and uniform temperature could be achieved along the wall of the
flame tube.
Ó 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Micro thermophotovoltaic system; Microcombustion; Photovoltaic array

1. Introduction

With the demand for smaller scale and higher energy density power sources, traditional bat-
teries cannot satisfy the need, which urges the development of micro power devices or power

*
Corresponding author. Tel.: +65-6874-4657; fax: +65-6779-1459.
E-mail address: [email protected] (W.M. Yang).

0196-8904/03/$ - see front matter Ó 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0196-8904(03)00024-4
2626 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634

Nomenclature

A pre-exponential factor for reaction rate

a constant
b constant
Cj concentration of jth gas phase species
Dah Damkohler number
Ea activation energy
m_ mass flow rate
Nr number of gas phase reactions
ng total number of gas phase species
nij reaction rate coefficient of reaction i and species j in reaction power matrix
P pressure
R gas constant
Rk net difference between rkf and rkb of kth reaction
ri rate of reaction i
rkb backward reaction rate
rkf forward reaction rate
T temperature
V volume
w_ j jth species generation rate
Greeks
aij third body enhancement factor of jth species in ith reaction
b temperature exponent
mjk stoichiometric coefficient of species j in reaction k
sresidence residence time
sreaction reaction time

MEMS. These systems use hydrogen or hydrocarbon as fuel and are characterized by thermal,
electrical and mechanical power densities of 1–20 W in sub-centimeter size packages [1–5]. It is
well known that the use of combustion processes for electrical power generation provides enor-
mous advantages over batteries in terms of energy storage per unit mass and in terms of power
generation per unit volume. Furthermore, the advantages of hydrocarbon fuels include low cost,
improved voltage stability, no memory effect and instant recharge.
The micro gas turbine engines, microrockets and micro rotary internal combustion engines are
typical micro power devices being developed by MIT and Berkeley [1–4]. These microengines
employ scaled down versions of existing macroscale devices, in particular internal combustion
engines. However, at microscale, these devices experience more difficulties with heat loss, friction,
sealing, fabrication, assembly etc. than their macroscale counterparts.
The microscale spiral counter flow heat recirculating combustor is another concept of power
MEMS developed by Sitzki et al. [5]. It does not involve any moving parts. Electrical power is
 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634 2627

generated by thermoelectric elements embedded in the walls between cold reactants and hot
products, but the three dimensional structures of a counter flow heat exchanger and combustor
make the fabrication complex. In addition, the current thermoelectric technology only has a
maximum energy conversion efficiency of 4%.

2. Development of micro thermophotovoltaic system

Different from the previous studies just mentioned, the ultimate goal of this research is to
develop a novel micro power system, a micro thermophotovoltaic (TPV) system, which uses
photovoltaic cells to convert heat radiation, e.g. from the combustion of fossil fuels, into elec-
tricity. The concept of TPV energy conversion was first proposed in the 1960s [6,7]. It is only in
recent years that technological improvements in the field of low band gap photovoltaic cells and
high temperature materials have evoked a renewed interest in TPV generation of electricity [8,9].
The micro TPV system we are developing originates from this concept.
The micro TPV system consists of three main parts: a heat source, a micro flame tube
combustor (the wall of the microcombustor would be made of selective emitting materials, such as
Er3 Al5 O12 and Co doped MgO etc.) and a photovoltaic array made of low band gap materials,
such as GaSb (0.72 eV), GaInAsSb (0.5 eV) and so on. Fig. 1 shows the basic design of the micro
TPV system. The volume of the microcombustor is 71 mm3 . Hydrogen and air are mixed in a
micromixer and then enter the microcombustor and combust. When the wall, i.e. the selective
emitter, is heated to a sufficiently high temperature, it emits photons, most of them having an
energy greater than the band gap energy of the photovoltaic materials due to the selective function
of the emitter. Therefore, when they impinge on the photovoltaic array, they evoke free electrons
and produce electrical power output under the action of a PN junction. Because the system does
not involve any moving parts, its fabrication and assembly are relatively easy, and its operation is
expected to be more reliable. At the same time, it possesses relatively high efficiency of energy
conversion [9]. As a result, it can be more commonly used in commercial electronics and micro-
devices, in which convenient and inexpensive production, reliable operation and low maintenance
cost are the key factors of success.
As one of the most important components of the micro TPV system, the microcombustor must
be developed first. The most challenging issue in microcombustor design is maintaining an op-
timal balance between sustaining combustion and maximizing heat output. According to the

photovoltaic
array

selective
emitter
inlet
flame tube
micro combustor
mixer
micro coolibg
fins

Fig. 1. Basic design of micro TPV system.

2628 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634

cube-square law, as the size of the combustor is reduced by a factor of 100, the surface and volume
will decrease by 4 and 6 orders of magnitude, respectively, and thereby, the surface to volume
ratio will increase by a factor of 100. So, sustaining combustion in a MEMS size combustor will
be largely affected by the increased heat losses due to the high surface to volume ratio, which tends
to suppress ignition and quench the reaction. On the other hand, the power output of a micro
TPV system depends on the temperature and the size of surface. High surface to volume ratio is
very favorable to the output of power density per unit volume. This is the most attractive feature
of a micro TPV system. In order to test the feasibility of combustion in microdevices and de-
termine the relevant factors affecting microcombustion, numerical and experimental work on a
micro flame tube combustor was performed.

3. H2 /air combustion mechanism

In microcombustor design, the choice of fuel plays a key role. Compared to a conventional
combustor, a microcombustor is more highly constrained by inadequate residence time for
complete combustion and high rates of heat transfer from the combustor. This fundamental time
constraint can be quantified in terms of a homogeneous Damkohler number Dah , the ratio of gas
residence time to the characteristic chemical reaction time [3]. To ensure complete combustion,
Dah must be greater than unity.
sresidence
Dah ¼ ð1Þ
sreaction
with
VP
sresidence  ð2Þ
m_ RT
½fuel0
sreaction  a b
ð3Þ
A½fuel ½O2  eEa =RT0
In this work, hydrogen was chosen as the fuel because of its high heating value, fast diffusion
velocity and short reaction time [3].
Gas phase kinetics of hydrogen oxidation reactions has been widely studied, and the typical
mechanism (shown in Table 1), consisting of 19 reversible reactions and nine species, is quoted in
our study [10,11]. Five reactions in this mechanism involve third body collisions, and their re-
action rates are expressed as:
Y
ng
n
X
ri ¼ ki Cj ij aij Cj ð4Þ
j¼1

The other gas phase reactions are written as:

Y
ng
n
ri ¼ ki Cj ij ð5Þ
j¼1
 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634 2629

Table 1
Gas-phase reaction mechanism of hydrogen oxidation [10,11]
Reactions A b Ea (cal/mol)
O2 + H ¼ OH + O 5.13E+16 )0.816 16507.0
H2 + O ¼ OH + H 1.18E+10 1.00 8842.7
H2 + OH ¼ H2 O+H 1.17E+09 1.30 3626.0
OH + OH ¼ H2 O + O 6.00E+08 1.30 0.0
H2 + O2 ¼ OH + OH 1.70E+13 0.00 47780.0
H + OH + M ¼ H2 O + Ma 7.50E+23 )2.60 0.0
O2 + M ¼ O + O + M 1.85E+11 0.50 95560.0
H2 + M ¼ H + H + Mb 2.23E+12 0.5 92600.0
H + O2 + M ¼ HO2 + Mc 2.10E+18 )1.00 0.0
H + O2 + O2 ¼ HO2 + O2 6.70E+19 )1.42 0.0
H + O2 + N2 ¼ HO2 + N2 6.70E+19 )1.42 0.0
HO2 + H ¼ H2 + O2 2.50E+14 0.00 1900.0
HO2 + H ¼ OH + OH 2.50E+13 0.00 700.0
HO2 + O ¼ OH + O2 4.80E+13 0.00 1000.0
HO2 + OH ¼ H2 O + O2 5.00E+13 0.00 1000.0
HO2 + HO2 ¼ H2 O2 + O2 2.00E+12 0.00 0.0
H2 O2 + M ¼ OH + OH + M 1.30E+17 0.00 45500.0
H2 O2 + H ¼ H2 + HO2 1.60E+12 0.00 3800.0
H2 O2 + OH ¼ H2 O + HO2 1.00E+13 0.00 1800.0
Rate constants are given in the form of k ¼ AT b expðEa =T Þ.
a
Enhancement factors: H2 O ¼ 20.0.
b
Enhancement factors: H2 O ¼ 6.0, H ¼ 2.0, H2 ¼ 3.0.
c
Enhancement factors: H2 O ¼ 21.0, H2 ¼ 3.3, O2 ¼ 0.0, N2 ¼ 0.0.

aij is unity except as specified in Table 1. Given the rate constant data in Table 1, the species
generation rate can be written as:
X
Nr
w_ j ¼ mjk Rk ð6Þ
k¼1

with Rk being expressed as:

Rk ¼ rkf  rkb ð7Þ

4. Experimental set-up

To investigate the stability of the flame and validate the simulation results, at the same time, for
simplicity of fabrication, three kinds of different stainless steel flame tube combustors (one is
straight tube of 2.2 mm diameter, the other two are flame tubes with a sudden expansion step,
having the diameter of 3 mm) were fabricated and tested. The construction and specifications of
the three flame tube combustors are given in Fig. 2. The mass flow rates of hydrogen and air were
controlled accurately by two sets of electronic mass flow controllers, through which the H2 /air
ratio can also be adjusted. The distributions of temperature, both on the exit plane and along the
wall of the flame tube, were measured by 0.203 mm diameter type K thermocouple.
2630 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634

Fig. 2. Configuration and specifications of three combustors (unit: m  103 ). (a) Tube #1, (b) tube #2 and (c) tube #3.

5. Results and discussion

According to the simulation results and experimental data of the three different flame tube
combustors, when the flow speed at the inlet drops to 1.3 m/s, because of the great heat losses, the
flame extinguishes after a short term combustion, which indicates that combustion cannot be
sustained in such small flame tubes when the flow rate is too low.
When the flow speed exceeds 8 m/s, combustion does not take place inside the straight tube
under any conditions, but it may take place outside of the tube. Therefore, stable combustion can
only be obtained in the straight tube with flow speeds at the inlet varying from 1.5 to 8 m/s, but
the position of the flame core is different for each speed. Furthermore, with increasing flow rate, a
higher H2 /air ratio is required to get stable combustion in the flame tube, or the flame will be
blown out the exit [12]. At 2 m/s, the peak temperatures on the wall occur at about 5, 14 and 25
mm below the exit plane when the H2 /air ratios are 0.45, 0.5 and 0.55, respectively. This indicates
that ignition occurs earlier, and increasing H2 /air ratios can increase the combustion rate.
From the above study, we know that there are some disadvantages with the straight flame tube
combustor: First, it is difficult to control the position of the flame, which keeps changing with the
variation of working condition. Second, combustion cannot be sustained in a straight tube when
the H2 /air ratio is not high enough, especially for high flow rates. So, we designed another kind of
flame tube with a sudden expansion step, which is used to facilitate recirculation along the wall,
thereby enhancing combustion completeness around the rim of the flow tube and ensuring stable
combustion.
Comparing to straight tube #1, tubes #2 and #3 with a sudden step can work steadily under a
much wider flow rate and wider H2 /air ratio. Stable combustion has been obtained with flow speeds
at the inlet varying from 1.5 to 20 m/s and H2 /air ratios varying from 0.45–1.0. At the same time,
the sudden step is very useful in ensuring that the flame occurs in the tube downstream of the step.
It should be mentioned that the tube downstream of the sudden step must be long enough so
that the combustion is finished before flowing out the exit. Fig. 3 shows the temperature distri-
 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634 2631

bution on the axial plane of tubes #2 and #3 when the velocity at the inlet is 8 m/s and the H2 /air
ratio is 0.45. The temperature profiles on the exit plane obtained by numerical simulation and
experimental testing are given in Fig. 4. From Fig. 3(a), we can see clearly that part of the fuel has
not been combusted before flowing out the exit, and thus, the temperature in the centerline of the
exit plane is very low (see Fig. 4(a)). In contrast, the result in Figs. 3(b) and 4(b) is much better.
Furthermore, we can further improve the completeness of combustion by increasing the H2 /air
ratio (see Fig. 5 [12]).

Fig. 3. Distribution of temperature on axial plane (velocity: 8 m/s, H2 /air ratio: 0.45). (a) Tube #2, (b) tube #3.

1600 1600
Temperature T (K)

Temperature T (K)

1200 1200

800 800

400 400

0 0
-0.002 -0.001 0 0.001 0.002 -0.002 -0.001 0 0.001 0.002
(a) radius R (m) (b) radius R (m)

Fig. 4. Temperature profile on exit plane. Simulation results (––), tested results (D). (a) Tube #2, (b) tube #3.

Fig. 5. Distribution of temperature on axial plane (tube #3) (velocity: 8 m/s, H2 /air ratio: 0.75).
2632 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634

Temperature T (K)

Temperature T (K)
1300 1300

1000 1000

700 700

400 400
0 0.002 0.004 0.006 0 0.003 0.006 0.009 0.012
(a) Distance from step L(m) (b) Distance from step L(m)

Fig. 6. Distribution of temperature along the wall (H2 /air ratio: 0.45; velocity at inlet: 8 m/s). Simulation results (––),
tested results (D). (a) Tube #2, (b) tube #3.

From Figs. 3 and 5, we also observe another interesting phenomenon. The combustion takes
place near the wall rather than the centerline of the flame tube at the beginning. Thus, the fuel/air
mixture around the centerline is heated and accelerated by the surrounding combustion products
and flows quickly to near the end of flame tube and combusts there, which is favorable for
maintaining a uniform temperature along the wall. The distributions of temperature on the wall of
the flame tube are given in Fig. 6. Both the simulation and experimental results indicate that the
maximum difference of temperature along the wall is less than 5%. This feature of the micro flame
tube combustor is very important to the design of micro TPV system.
According to Figs. 4 and 6, we know that the temperature distribution profile obtained by
numerical simulation is very similar to that obtained by experiment, and the differences of tem-
perature, both on the exit plane and along the wall, are less than 9%.
The H2 /air ratio is one of the most important factors affecting microcombustion. It not only
affects the combustion rate but also affects the temperature and its distribution. When the flow
rate is constant, with increasing H2 /air ratio, both the temperatures on the exit plane and along
the wall increase drastically, and the position of the peak temperature on the exit plane also moves
towards the centerline. This can be seen in Fig. 7, which shows the test results when the velocity at
the inlet is 5 m/s.
Flow rate is another important factor affecting the temperatures, both on the exit plane and
along the wall [12]. Fig. 8 shows the variation of mean wall temperature with velocity. With in-
creasing velocity at the inlet, the total energy released by combustion increases, and thereby, the
mean wall temperature increases. However, as the flow rate increases further, the effective resi-
dence time of the fuel decreases, worsening the completeness of combustion, lowering the effi-

1400 1050
Temperature T (K)

Temperature T (K)

1200 1000
950
1000
900
800 850
600 800
-0.002 -0.001 0 0.001 0.002 0 0.005 0.01
(a) Radius R (m) (b) Distance from step L(m)
0.45 0.50 0.60 0.45 0.50 0.60

Fig. 7. Distribution of temperature (velocity at inlet: 5 m/s) (a) on exit plane and (b) along the wall.
 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634 2633

Temperature T (K)
1300
1100
900
700
500
0 5 10 15
Velocity at inlet (m/s)
0.45 0.50
0.60 0.70

Fig. 8. Variation of mean wall temperature with velocity.

Fig. 9. The picture of microcombustion.

ciency and the temperature on the wall, even quenching the flame. This situation should be
avoided.
Optimization yields an average temperature of about 1300 K along the wall, which has been
achieved for tube #3 when the flow rate at the inlet is 12 m/s and the H2 /air ratio is 0.95, which is
appropriate as the heat source of the micro TPV system. The photo of microcombustion by digital
camera under these conditions is shown in Fig. 9.

6. Conclusions

High surface to volume ratio is very favorable to the output of power density per unit volume,
though it tends to suppress ignition and quench the reaction in microdevices. The above studies
indicate that stable combustion can be achieved in a small tube under a sudden step with a wider
flow rate and wider hydrogen/air ratio than in a straight tube. Furthermore, the sudden step is
very useful in controlling the position of the flame. Combustion takes place near the wall rather
than the centerline of the flame tube at the beginning. Thus, the part of the fuel/air mixture
around the centerline is heated and accelerated by the surrounding combustion products and
flows quickly to near the end of flame tube and combusts there, which is favorable to keep the
2634 W.M. Yang et al. / Energy Conversion and Management 44 (2003) 2625–2634

uniform temperature along the wall. The H2 /air ratio not only affects the temperatures, both on
the exit plane and along the wall, but also affects the position of the flame core. Flow rate is
another important factor affecting microcombustion in a flame tube. An average temperature of
about 1300 K along the wall has been obtained with a flow speed of 12 m/s at the inlet and the
H2 /air ratio of 0.95, which is appropriate as the heat source of the micro TPV system.

Acknowledgement

This work is supported by NUS grant number R-265-000-066-112.

References

[1] Epstein AH, Senturia SD, et al., Micro-heat engines, gas turbines, and rocket engines––the MIT Microengine
project. American Institute of Aeronautics and Astronautics, AIAA 97-1773, 1997. p. 1–12.
[2] Epstein AH, Senturia SD, et al., Power MEMS and Microengines. In: IEEE Transducers Õ97 Conference, Chicago,
IL, 1997. p. 753–6.
[3] Waitz IA, Gauba G, et al. Combustors for micro-gas turbine engines. ASME J Fluids Eng 1998;120:109–17.
[4] Kelvin Fu, Knobloch AJ, et al. Microscale combustion research for applications to MEMS Rotary IC Engine. In:
Proceedings of NHTC, 2001 National Heat Transfer Conference, Anaheim, CA, 10–12 June 2001. p. 1–6.
[5] Sitzki L, Borer K, et al. Combustion in microscale heat-recirculating burners. In: The Third Asia-Pacific
Conference on Combustion, Seoul, Korea, 24–27 June 2001. p. 1–4.
[6] White DC, Wedlock BD, Blair J. Recent advance in thermal energy conversion. In: 15th Annual Power Sources
Conference, Atlantic City, NJ, 1961. p. 125–32.
[7] Guazzoni G, Kittl E, Shapiro S. Rare earth radiators for thermophotovoltaic energy conversion. In: International
Electron Devices Meeting. Washington, DC, 1968. p. 130–2.
[8] Ferguson LG, Fraas LM. Theroretical study of GaSb PV cell efficiency as a function of temperature. Solar Energy
Mater Solar Cells 1995;39:11–8.
[9] White DC, Hottel HC. Important factors in determining the efficiency of TPV systems. In: First NREL Conf
Thermophotovoltaic Generation of Electricity, Copper Mountain, CO, 1995. p. 425–56.
[10] Markatou P, Pfefferle LD, Smooke MD. The influence of surface chemistry on the development of minor species
profiles in the premixed boundary layer combustion of H2 /air mixture. Combust Sci Tech 1991;79:237–68.
[11] Warnatz J, Allendorf MD, et al. A model of elementary chemistry and fluid mechanics in the combustion of
hydrogen on platinum surfaces. Combust Flame 1994;96:393–406.
[12] Wenming Y. Simulation of micro-combustion in micro-combustor and relevant experiments, Annual report,
National University of Singapore, 2001. p. 1–18.