Thermally driven flows: thermal transpiration

At the present moment we are focusing attention on the possibilities of using thermal transpiration as means for building an operational micro pump prototype. Different configuration options are being studied such as the Ratchet and Serpentine Knudsen pumps geometries.

Furthermore thermal transpiration is being studied from a more fundamental aspect. For the first time investigation is being carried out on the subject by means of a non intrusive velocimetry experimental method, that is the Molecular Tagging Velocimetry technique.

Wall-bounded flows may be efficiently controlled by appropriately modifying the boundary layer structure. It is thus possible to reduce the drag or increase the lift of an aircraft wing, to favor mixing in a combustion chamber, to reduce the aero-acoustic noise, or to improve heat transfer. The nature of perturbations that need to be introduced in the boundary layer mainly depends on the flow characteristics: Reynolds and Mach numbers, type of instabilities in the boundary layer, etc. For high Reynolds numbers and for compressible flows, active control methods, based on local momentum injection in the near wall flow thanks to dynamic systems composed of sensors and actuators, have proven to be efficient in particular for flow separation control.

Several types of mechanical microactuators have been developed for active control applications (thermal microactuators, micro magnetic flaps, micro balloons, etc.) but fluidic solutions have the advantage, important for reliability, of having no moving part in direct contact with the external flow and allowing a simple control.

Our work focuses on the numerical simulation and experimental analysis of the behavior of fluidic actuators such as synthetic jets, able to inject momentum in the controlled flow without added mass, or fluidic oscillators able to generate pulsed jets without any moving part. The efficiency of the new actuators designed and developed in our team is tested and experimentally characterized on a separated flow over a ramp. These works are performed in collaboration with the University of Orléans (Laboratory PRISME) in the framework of the French Research Network "Flow separation control") (GDR 2502: http://www.univ-orleans.fr/GDR2502).

A new micro molecular tagging velocimetry (μMTV) setup has been developed to analyze velocity fields in confined internal gas flows. MTV is a little-intrusive velocimetry technique. It relies on the properties of molecular tracers which can experience relatively long lifetime luminescence once excited by a laser beam with an appropriate wavelength. The technique has been validated for acetone seeded flows of argon inside a 1 mm depth rectangular minichannel.

The photo-luminescence effects of gaseous acetone excited by UV light, implemented in MTV, have been analyzed in various pressure conditions. As a result, the acetone phosphorescence shows a drastic decrease with pressure and becomes non measurable for pressures lower than 1 kPa. On the other hand, fluorescence shows a slower decrease and remains clearly detectable at pressures as low as 10 Pa. The motion of tracer molecules within the carrier flow has been studied. The analysis of the displacement of the tagged molecules has shown the strong influence of molecular diffusion, this influence being increased with the gas rarefaction.

An important step forward on MTV has been achieved by realizing a reconstruction method based on the advection diffusion equation. this method allows to extract the velocity profile from the analysis of the displacement and deformation of the tagged region, taking into account the diffusion of tracer molecules. This reconstruction method has been successfully implemented to analyse a Poiseuille flow in a rectangular millimetric channel, under atmospheric pressure conditions, and the capability of MTV to accurately extract the local velocity in confined gas flows has been demonstrated.

In the field of suspension flows, the pioneering experiments of Segré and Silberberg highlighted the fact that neutrally buoyant finite sized particles migrate across the tube flow streamlines towards an equilibrium ring located at r = 0.6R (R being the tube radius). After these observations, many theoretical studies tackled the problem of inertia-driven particle migration in order to determine the origin of this phenomenon. All the evidence suggests that the particle cross-streamline migration is related to the interaction of the particle Stresslet (rigidity constraint of the finite sized particle) with the curved velocity profile of the pressure-driven flow, and that the speed of the migration process, as well as the particle equilibrium positions, depends exclusively on the Reynolds number and on the particle-to-channel (or tube) size ratio.

In the last decade, the particle migration due to flow inertia has been extensively used for successful particle separation and sorting in microfluidic devices. Square channels are often considered due to the ease of their manufacturing process. However, it is complex to establish theoretical predictions of the particle-flow interaction in rectangular 3D flow geometries. At finite inertia, the experiments and numerical simulations agree on the fact that, in a square channel flow, particles first undergo a cross-streamline migration (similarly to the Segré-Silberberg phenomenon) until they reach an equilibrium ring parallel to the velocity iso-contours.

In our labs, the migration of neutrally buoyant finite sized particles in a Newtonian square channel flow is investigated in the limit of very low solid volumetric concentration, within a wide range of channel Reynolds numbers Re = [0.07-120]. In situ microscope measurements of particle distributions, taken far from the channel inlet (at a distance several thousand times the channel height), revealed that particles are preferentially located near the channel walls at Re > 10 and near the channel center at Re < 1. Whereas the cross-streamline particle motion is governed by inertia-induced lift forces at high inertia, it seems to be controlled by shear-induced particle interactions at low (but finite) Reynolds numbers, despite the low solid volume fraction (<1%). The transition between both regimes is observed in the range Re = [1-10]. This experimental work is compared to numerical simulations made by research partners in the "Laboratoire de Génie Chimique de Toulouse".

In order to exclude the effect of multi-body interactions, the trajectories of single freely moving particles are calculated thanks to numerical simulations based on the force coupling method. With the deployed numerical tool, the complete particle trajectories are accessible within a reasonable computational time only in the inertial regime (Re > 10). In this regime, we can show that (i) the particle undergoes cross-streamline migration followed by a cross-lateral migration (parallel to the wall) in agreement with previous observations, and (ii) the stable equilibrium positions are located at the midline of the channel faces while the diagonal equilibrium positions are unstable. At low flow inertia, the first instants of the numerical simulations (carried at Re = O(1)) reveal that the cross-streamline migration of a single particle is oriented towards the channel wall, suggesting that the particle preferential positions around the channel center, observed in the experiments, are rather due to multi-body interactions.