WorksPhysics and Astronomy
Incompressible Squeeze-Film Levitation
Abstract
Transverse vibrations can induce the non-linear compression of a thin film of air to levitate objects, via the squeeze film effect. This phenomenon is well captured by the Reynolds' lubrication theory, however, the same theory fails to describe this levitation when the fluid is incompressible. In this case, the computation predicts no steady-state levitation, contradicting the documented experimental evidence. In this letter, we uncover the main source of the time-averaged pressure asymmetry in the incompressible fluid thin film, leading the levitation phenomenon to exist. Furthermore, we reveal the physical law governing the steady-state levitation height, which we confirm experimentally.
However, relaxation of the assumptions of the lubrication theory is needed to uncover the underlying physics behind this phenomenon and find out its physical governing law.
In this letter, we show that a stable steady-state squeeze-film levitation of objects can be obtained in incompressible liquid environments. We uncover the fundamental pressure-inducing mechanisms in thin films. Finally, We reveal the physical law that governs the steady-state levitation height, which we validate experimentally.
Consider an axisymmetric system of a sinusoidal vibration source, a free-floating reflector and a thin film of liquid in between as shown in Figure 1. We define the time-dependent film thickness such that , where is an assumed initial film thickness between the source and reflector. The disc radius is such that . The source oscillates with an angular frequency and an amplitude such that . We define the squeeze Reynolds number [21] to be where is the steady-state time-averaged film thickness, is the density of the liquid and is its viscosity. This number gives a measure of the relative significance of inertial and viscous effects.
Figure 1. In-liquid squeeze film levitation. Initially, a film thickness separates the vibrator and the free-floating reflector. This initial film thickness originates from various sources such as the roughness of the two surfaces or misalignment. Upon the start of the vibration , a pressure builds up in the film layer, which pushes reflector away distance until it reaches equilibrium at the time-averaged steady-state levitation distance . The reflector oscillates around its equilibrium position harmonically, denoted by . The over-pressure in the liquid layer is linked to the steady state time-averaged levitation film height.
One of the key assumptions of the lubrication theory is that fluid inertia is negligible and thus fluid behaviour is dominated by viscous effects. This proved to be true for gas squeeze films since the squeeze Reynolds number is typically less than unity () [22, 23]. However, in the case of a typical liquid, such as water, the order of magnitude of the physical parameters of the Reynolds number is as follows: Kg/m, rad/s and Pa.s. Given the experimental data of Hatanaka et al. [7], we can also expect the order of magnitude of to be m. This yields a Reynolds number of suggesting the significance of the film inertia and thus, the lubrication theory’s assumption becomes invalid. Therefore, we revisit the basic fluid governing equations, which take inertia into account.
Given the aforementioned axisymmetric system of a source and reflector and given that , we can safely assume that the pressure gradient across the film thickness is negligible compared to the radial pressure gradient and therefore the pressure is only a function of the radial coordinate [24]. With this assumption, the conservation of momentum and mass equations for this system using cylindrical coordinates, are given by [24, 25]:
where is the liquid pressure, is its density, is its viscosity, and and are the liquid velocity fields along and respectively. To solve for the pressure and velocity fields, we follow the iterative scheme of Kuzma and Jackson [26, 27]; we first approximate the inertial forces (left hand side of Eq.1) by employing the velocity profiles of the classical lubrication theory, which are given by:
Then, by integrating (1) twice with respect to and assuming static boundary conditions , we find the following new expression for the radial velocity field :
The conservation of mass principle requires that the inflow or outflow across the control volume of the film is equal to the volume change due to the source vibration and the reflector levitation. This condition can be expressed mathematically in integral form as follows:
Finally, by substituting the velocity profile expression Eq.4 into the continuity equation Eq.5, we obtain the following expression for the pressure gradient along [26]:
This expression has different coefficients of the second and third terms compared to the work of Li et al. [28], because in that work the term was neglected despite being comparable in magnitude to [26]. This first order iterative solution was shown previously to be stable and to agree with the full numerical solutions [29], unlike the higher order solutions [24]. By integrating the pressure gradient expression with respect to assuming the boundary condition where is the ambient pressure and is a pressure loss term due to the edge effect, we obtain the following pressure field expression:
where
This pressure profile expression suggests three main pressure-inducing mechanisms for incompressible films. The first one is associated with the viscosity of the liquid represented by the term . This term is symmetric and thus, has no contribution to the time-averaged pressure. The second and third mechanisms are associated with the temporal and convective accelerations of the liquid squeezing in and out of the film, represented by the asymmetric terms and respectively. In addition, accounting for liquid inertia results in an inevitable pressure drop at the edge of the squeeze film during negative squeeze motion (), due to the sudden contraction of the liquid at the film entrance. Using Bernoulli’s equation, we can find an expression for the pressure drop term [30, 31] equal to where is a pressure loss coefficient that depends mainly on the geometry of the film entrance (i.e. sharp or round edge). This edge effect term along with the asymmetric temporal and convective acceleration terms contribute to a non-zero time-averaged squeeze-film pressure (refer to the supplemental material [32] for more details about the pressure terms).
To account for the levitated object being free floating, we assume a single degree of freedom system where the reflector is a mass. The mass is connected to the vibrating surface through a liquid layer whose initial thickness is . We introduce two independent coordinates where is the displacement of the reflector and is the displacement of the vibrating surface. The film height is a function of and coordinates as follows:
By analyzing the forces acting on the reflector, we can obtain the following equation of motion:
where ,
where the mass ratio () is a measure of the relative significance of the inertial effects of the free floating object and the liquid film, and the kinematic viscosity () is a measure of the fluid resistance to flow under inertial forces.
To understand the nature of this dynamical system, we conducted experiments in which we measured the levitation displacement of a free-floating mass placed on top of a vibrating surface (40 kHz) in a liquid container at different vibration amplitudes, similar to other experiments from literature [33, 34, 35]. As demonstrated in Figure 2(a) for a mass of (64 g), experiments confirmed the existence of stable squeeze-film levitation in liquid (refer to the supplemental material [32] for details about the experimental setup).
Figure 2. (a) Demonstration of the stable steady-state levitation of a free-floating reflector of mass 64g, at 40kHz excitation. For each vibration amplitude, five trials of measurements were collected and the results are plotted as the mean displacement (solid lines) and the standard deviation (shades). (b) A detailed view of one experimental trial corresponding to the input vibration amplitude , with the vibrator displacement being . We observe that the floating mass levitates to a steady-state position and oscillates around the equilibrium position, where the oscillations are represented as . The response shows two characteristic time scales: (a) time corresponding to the mass reaching steady-state levitation distance (b) time associated with the reflector oscillations .
By looking closely at the response shown in Figure 2(b), we observe the following: the reflector reaches an equilibrium levitation position while oscillating. We notice that the amplitude of the reflector oscillations is considerably smaller than the input amplitude (). In addition, we also find the phase shift between the reflector oscillations and input vibrations to be consistently around . Finally, the dynamic response of the system has two characteristic time scales: (a) time corresponding to the mass reaching a steady-state levitation position (b) time associated with the reflector oscillations , such that . We can exploit this observation to decompose the response into two components; namely, a time-averaged and an oscillating components as follows:
where is the time-averaged steady-state levitation distance and is the oscillation component such that . Since we are mainly interested in the time-averaged levitation component (), we can impose a time average operator on the dynamic equation (10). This time average operator is given by , where is the period of the oscillation. The analytical time-average of (10) exists in the special case when the mass ratio is negligible compared to the term in the denominator of the right hand side of (10). We can find the range of values where the assumption of a negligible mass ratio is valid to be as follows: for masses of few hundred grams , radii of tenth of millimeters and for a typical liquid density of order , the mass ratio is of order compared to an for the term. By imposing this assumption on (10), we obtain the following expression:
A remarkable consequence of the assumption is that all of the terms of (10) become symmetric in (12) except for the convection terms and . It means that within the range of values in which the assumption is valid, the convection effects prevail and becomes the sole source of the steady-state time-averaged levitation. By imposing the time average operator, the symmetric terms , and of (12), by definition, converge to zero which yields the following:
The time average of the film height yields the steady-state component . On the other hand, we can find the time average of the by substituting the derivative of (11) into the expression of (9) and find the time average integral, to be:
Given our earlier experimental findings that the amplitude of the reflector oscillations is () of the input amplitude and that the phase shift is around , we can conclude that the reflector dynamics terms and are negligible compared to the vibration input term . By omitting the negligible terms from (14), substituting it back in (13) and expanding the mass ratio , we find the following expression for the time-averaged steady-state film height :
where is a correction factor that accounts for the energy dissipation due to the edge effect, such that . To validate the theory, we conducted experiments using different combinations of masses (64, 101 and 122 g) with a disk radius (10 mm) at ultrasonic (40 kHz) vibration amplitudes (1-3 ), in a purified liquid of density (1030 kg/m). Using the collected data, we identified the empirical parameters () to be and . As demonstrated in Figure 3, the model shows close agreement with the experimental data. We also observe that the correction factor is not influenced by the vibration amplitude nor the mass, as expected, since the loss coefficient depends only on the geometry of the film entrance. In addition, we see that levitation starts at certain amplitude threshold, similar to air systems [35], which is mainly because for thicker squeeze films, higher amplitude is required to induce sufficient pressure in the film to outweigh the mass of the reflector and vice versa.
Figure 3. Validation of the model demonstrated by the steady-state time-averaged levitation height as a function of the vibration amplitude and mass of the reflector expressed as . The experiments were conducted on three different masses: 64 g (blue parallelograms), 101 g (orange circles) and 122 g (yellow triangles) at vibration amplitudes up to . We identified the initial film thickness to be and the pressure loss coefficient to be . The model (black solid line) shows close agreement with the experimental data (individual plotted points) at different combinations of vibration amplitudes and masses. The inset shows the decomposed experimental data; the steady-state time-averaged levitation height as a function of the vibration amplitude for each mass individually.
In summary, we showed that liquid inertia, manifested as the temporal and convective acceleration of the liquid film, is the source of pressure asymmetry that causes the levitation phenomenon to exist. We demonstrated that in the special case of a negligible mass ratio () compared to a film thickness parameter , the fluid convective inertial effects prevail and become the sole source of the levitation phenomenon. In such a case, we derived a closed form compact formula that predicts the steady-state time-averaged levitation height as a function of the vibration input parameters (), liquid density () and mass () of the levitated object. The findings of this letter could potentially lead to novel physics-informed designs of in-liquid levitation systems for non-contact manipulation, transportation and friction modulation applications.
The data that support the findings of this study are openly available in 4TU.ResearchData Repository [36].
References
References
- [21]N. Brunetière and M. Wodtke, Considerations about the applicability of the Reynolds equation for analyzing high-speed near-field levitation phenomena, Journal of Sound and Vibration 483, 1 (2020).DOI
- [22]A. Minikes, I. Bucher, and S. Haber, Levitation force induced by pressure radiation in gas squeeze films, The Journal of the Acoustical Society of America 116, 217 (2004).DOI
- [23]T. A. Stolarski and W. Chai, Inertia effect in squeeze film air contact, Tribology International 41, 716 (2008).DOI
- [24]J.A. Tichy and W.O. Winer, Inertial Considerations in Parallel Circular Squeeze Film Bearings, ASME Pap , 588 (1970).DOI
- [25]B. J. Hamrock and S. R. Schmidt, Fundamentals of Fluid Film Lubrication Second Edition (CRC Press, 2004) pp. 1–693.
- [26]D. C. Kuzma, Fluid inertia effects in squeeze films, Appl. Sci. Res. 18, 15 (1967).DOI
- [27]J. D. Jackson, A Study of Squeezing Flow, Appl. Sci. Res. 11, 148 (1963).DOI
- [28]J. Li, W. Cao, P. Liu, and H. Ding, Influence of gas inertia and edge effect on squeeze film in near field acoustic levitation, Applied Physics Letters 96, 1 (2010).DOI
- [29]R. J. Grimm, Squeezing flows of Newtonian liquid films, an analysis including fluid inertia, Applied Scientific Research 32, 149 (1976).
- [30]H. Hashimoto, Squeeze Film Characteristics Between Parallel Circular Plates Containing a Single Central Air Bubble in the Inertial Flow Regime, Journal of Tribology 117, 513 (1995).DOI
- [31]Squeeze film, in Hydrodynamic Lubrication (Springer Tokyo, Tokyo, 2006) pp. 137–160.
- [32]See supplemental material at [url will be inserted by publisher] for more details about the experimental setup, analysis and fem simulation.
- [33]A. Minikes and I. Bucher, Coupled dynamics of a squeeze-film levitated mass and a vibrating piezoelectric disc: Numerical analysis and experimental study, Journal of Sound and Vibration 263, 241 (2003).
- [34]P. Liu, J. Li, H. Ding, and W. Cao, Modeling and experimental study on near-field acoustic levitation by flexural mode, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 56, 2679 (2009).DOI
- [35]Y. Wang and P. Guo, Stiffness modeling for near field acoustic levitation bearings, Applied Physics Letters 118, 10.1063/5.0051372 (2021).DOI
- [36]M. Atalla, R. van Ostayen, A. Sakes, and M. Wiertlewski, Raw data for the paper: Incompressible squeeze-film levitation, 10.4121/21523890 (2022).DOI


