Version HTML de base
36
8
e
CFA - Tours 2006
Acoustique
&
Techniques n° 45
areas are envisaged: product-level cooling for portable
electronic devices; and package-level cooling for a range of
electronic systems. For the latter applications, integration of
the fan with the package is proposed in order to achieve heat
removal at source.
A key challenge in the development of small-scale fans is the
reduction of efficiency with scale, a limitation of turbomachinery
at low Reynolds numbers. In order to assess this phenomenon,
a set of three geometrically-similar axial flow fans was created,
based on a 120mm diameter fan – full scale, 1/3rd scale
and 1/20th scale. The 6mm diameter fan shown in figure
2 – from Grimes et al [10] – was fabricated using a micro
Electro Discharge Machining (mEDM) process, and driven by
a brushless DC micro-motor.
Pressure-flow characterisation tests were performed on the
three fans at 0.049 specific speed in order to compare their
performance. To maintain this specific speed, the full scale,
1/3rd scale and 1/20th scale fans were tested at 2,240rpm,
6,720rpm and 44,800rpm respectively.
Figure 3, from Grimes et al [11], shows non-dimensional static
pressure rise for the three fans. Characteristics of the full scale
and 1/3rd scale fans are similar, but the performance of the
1/20th scale fan is lower in terms of both pressure and flow
rate – a difference attributed to inaccuracies in scaling tip
clearance. In dimensional terms, maximum volumetric flow
rate of the 6mm diameter fan is 0.263 m3/hr, a delivery which
yields an average velocity of 3-4 m/s at outlet. The physical
scale of this fan is suitable for integration into portable devices,
and the volumetric flow rate is sufficient to achieve system-
level heat removal. Moreover, the delivery velocity is adequate
to enhance package-level cooling – particularly in conjunction
with extended surfaces.
Particle-Image Velocimetry (PIV) was also performed on the full
scale and 1/3rd scale fans in order to investigate the influence
of scale on the exit flow field. Two specific speeds (0.049
and 0.061) and five pressure rises (0, 5, 10, 15, and 20 Pa)
were assessed, and a sub-set of the results – from Quin et al
[12] – is presented in figure 4. For these results, the full scale
and 1/3rd scale fans were run at 2,800 rpm and 8,400 rpm
respectively in order to achieve a specific speed of 0.061.
Two pressure rises – 0 Pa and 20 Pa – are featured in order to
illustrate the range of flow behaviour from maximum delivery
to maximum pressure.
At zero pressure – maximum delivery – the flow from the full
scale fan is predominantly axial, with some divergence into the
dead zones in front of the fan hub and blade tip. An increase
in pressure induces partially radial flow in the full scale fan,
the influence of changing centripetal forces in the jet due to
an increase in tangential flow. The flow from the 1/3rd scale
fan has significant radial components even with zero pressure
rise. Magnitudes of the velocity vectors are lower than for
the full scale fan, and more air recirculates in front of the fan
hub: these phenomena are more strongly evident at higher
pressure.
It is clear that the flow from small scale axial fans is significantly
radial in character: there would appear to be advantages in
Fig. 2 : 6mm diameter axial microfan
Fig. 3 : Non-dimensional fan characteristics for 0.049 specific speed
Thermal Management of Electronic Systems: Emerging Technologies and Acoustic Challenges
Gestion thermique des systèmes électroniques : Technologies naissantes et défis acoustiques