REVIEW Numerical Analysis of Wall Slip on Nano & Micro Channels

Keywords: Poiseuille flow; Couette flow; slip velocity; no-slip velocity; boundary condition; Knudsen

number; non-equilibrium condition; wet flow; Lagrangian particle method; Backward tracking; kinetic

theory; Newtonian fluids; non-Newtonian fluids; viscoelastic fluids; non equilibrium molecular dynamic

(NEDM) simulation; direct simulation Monte Carlo (DSMC)

Introduction

Richardson (1973) argued that there is no relation motion (no-slip velocity) between the
fluid and the solid at the interface due to viscose fluid flows on the rough surface on a
microscopic scale.
Maxwell (1879) had published the famous “velocity slip boundary condition”. He proved
that in terms of microscopic consideration, the relation motion between fluid and the solid
interface doesn’t usually lead to no-slip boundary conditions. These boundary conditions are
successful in terms of a rarefied gas which could slide over a surface and that inequalities in
temperature could give rise to a force from colder to hotter regions. These two conditions can
lead to slip boundary conditions.
Lockerby et al (2004) claimed that Maxwell’s theory was inapplicable for most complex
surface geometry such as curvature and/or rotational motion; because it neglects that the
velocity normal to the wall can vary in the stream wise direction. They suggest using second
order condition which is more accurate at greater degrees of gas rarefied and is able to
capture micro flow phenomena.
More over Flow near interface at various scales to reduce flow friction are investigated by
Barrat and Bocquet (2007). They discuss the present theatrical understanding of flow past
solid interface and influence of surface slip on interface driven at different length scales.
Bartel (2001) report , today the direct simulation Monte Carlo (DSMC) methods are used in
the area of rarefied gas dynamics or where mean free path phenomena are of interest.
Although current almost universally using of DSMC method to simulating transition
continuum flows, Gallis et al (2003) claimed that DSMC computational cost limits its
applicability to simulation flow in quite simple geometry. In other hand, they argue that the
difficulties in measuring the slip flow accuracy are due to lack of the amount of reliable data
available. However, they suppose this has not halted the development of numerical
techniques for the simulation of such flows. In spite of they believed DSMC is used with
confidence to check the result of experimental measurement, they recommended using an
alternative method , instead of DSMC method, numerical techniques which is a generic term
covering all the methods outline in this literature.

Knudsen Layer in gas flows:

The Knudsen layer is the region of local none equilibrium extending one or two molecular
mean free path from the wall in any gas flow near a surface, this is in most hypersonic and
micro flow applications. Engineering designer is mostly interested in the momentum and
energy fluxes from the region of the Knudsen layer to the boundaries. Due to this reason a
correct description of the Knudsen layer is critical when modeling of such flows. This was
the thought and words of Lockerby et al (2005).
White (1999) and Gad-el-Hak (1999) define Navier-Stocks equations. The Navieat-Stock
equations put into words for momentum and heat transfer in fluids. The equations work very
well for modeling fluids flowing in macroscopic scale where Knudsen number (Kn) is less
than 0.001. The Navier-Stokes equations cannot, model the nonlinear stress/strain-rate
behavior within the Knudsen layer.
A wall function based on a simple Knudsen layer solution within a continuum fluid
formulation for locally non equilibrium gas flow and the Navier-Stock equations for
monatomic gas flow one dimensional planar Coutte and Poisuille system, had been used by
Lockerby et al (2005) to account the Knudsen number in a benchmark by employing
“fictitious” or “macro” slip boundary conditions and the “actual” or “micro” slip velocity at
the boundary. These slip boundary are different in term of their slip coefficient velocity but
in both cases the Knudsen number had been defined as the mean free path to half channel
height. They compared the Navier-Stock equations by using both the macro and micro slip
boundary conditions and the solution was obtained by using the Direct Simulation Monte
Carlo (DSMC) method. They demonstrated that when micro slip applied at the boundary, the
prediction of the velocity both inside and outside the Knudsen layer is poor compare to
fictitious (macro) velocity slip does, at least, an accurate solution outside the Knudsen layer.
Slip coefficient of Fictitious slip boundary conditions, kinetic teary predicted this factor as
1.46665 by Cerecignani (1963) and ones for Actual slip boundary are based on Maxwell’s
value (1879) for slip coefficient to be 1.0. But Lockerby et al (2005) calculated this factor as
0.8. Cercignani’s “second-order slip coefficient” was again clarified by Hadjiconstantinou
(2003).
Gallis et al (2003) claim that the flow velocity at any surface (slip velocity) effects by
differentiated the temperature of the gas near the surface from the surface temperature itself
(temperature jump), this mean’s that the boundary condition normally required by the
Navier-stock equations has to change. This situation becomes more problematic as Knudsen
number rises to the range 0.1-1(transition continuum regime). Aubert and Stephane (2001)
report the Navier-Stock equations remain usable up to kn=0.1, they are generally associated
with Maxwell-Smoluchuski boundary conditions, which express a temperature and a velocity
jump at the wall. When the Knudsen number increases, up to or even slightly above 0.1, one
might think it would be necessary to take in account the higher order. Several authors have
addressed this problem, and taken into account the second order terms..

Second order equations:

Second order equation’s effect on the velocity and temperature jumps at a wall is described
by Deissler (1964). Differences between the first- and second-order conditions on the order
had been compared by experiments. The Navier-Stokes and energy equation in the gas was
shown to be consistent with the use of second-order boundary conditions. The paper used
Burnett equation to analyze second order jump conditions (jumping heat and velocity near
the wall). Moreover it includes analyzing the effective mean free path in terms of second
order equations for momentum and energy transfer from velocity and temperature jumps at a
wall.
The potential of some higher order continuum equations had compared with DSMC (direct
simulation Monte Carlo) data, as well as standard result from kinetic theory Gallis et al
(2005). Neither the Burnett, super-Burrnet, nor Grad’s 3 moment higher order equations can
model the Knudsen layer. But some of them can qualitatively describe the Knudsen layer
structure. They recommend making quantitative comparisons by additional boundary
conditions from kinetic theory to obtain for a unique solution to the higher order equations.
However, they claim the agreement between kinetic theory and DSMC data is good.
Knudsen layer been investigated in term of kinetic theory by Cercignani (1990) and Sone
(2002).
Higher order Burnett equations are used for continuum model for steady- state micro coutte
flow investigation by Lockerby and Reese (2003). Solution to these equations are shown first
order sleep condition agree very well with direct simulate up to Knudsen number 1 (the limit
of the equations validity). And orders of magnitude are quicker to obtain than the
corresponding DSMC data. The poorer agreement with DSMC data when using second order
slip conditions indicated a fundamental problem with higher order micro sleep conditions and
requires further investigation.
Lockerby et al (2004) they suggested to use a higher order boundary condition based on
Maxwell’s general equation and the constitutive derived by Burnett second order equation.
They showed that Maxwell-Burret boundary conditions have reasonable agreement with the
limit experimental Poiseulle flow data available and can predict the phenomenon of thermalstress
slip flow.

Liquids:

Lauga & Stone (2003) say ‘when the fluid does not completely wet an atomically smooth
substrate, then might expect (even rarefied gases or liquids), or at least it has been
conjectured, that the flow may exhibit expect, or at least it has been conjectured, that may
exhibit some manifestations of microscopic slip’.
A spatial case of a liquid that partially wets the solid boundary in Coutte and Poiseuille flow
for the same fluid substrate film consired by Barrat & Bocquet (1999a). They used this to
investigate influence of wetting properties on hydrodynamic boundary conditions at a fluid
interface (1999b). They realized that the boundary condition depend on tow parameters, the
wall location and the slipping length. From non equilibrium molecular dynamic (NEDM)
simulations, it had been realized that when the contact angle between molecules of the fluid
and the substrate is large enough, the boundary condition can drastically differ from no slip
condition. More over they derived an expression for the slipping length and it compared with
the numerical results for the slipping length obtained in NEMD.
Nano-scale grating patterns had been used to measure their effect on liquid flows in micro
channels by Choi et al (2006).They believed the effective slip is clear and larger in flow
parallel to the nano-grating pattern rather than in transverse. They recommended that the
nano-grated surfaces would not only reduce friction in liquid flows under pressure but also
enable directional control of the slip.

Micro flows on corners:

Effects of Newtonian and non-Newtonian fluids on the wall slip in micro and macro fluids
had searched by Sunara et al (2007). A part of their experiment shows how the vortex growth
for the flow in micro and macro channels. More over how slip on the wall is effected by wall
shear stress changing due to Newtonian and non-Newtonian fluids.
Crochet and Purnode (1996) had been investigated the mechanism of vortex improvement
and occurrences of slip vortices in contraction corner sharpness and viscoelastic fluids
(polycrilamide). Another method had been used to investigate convex growth in sharp corner
by Wapperom et al (2000). It was Backward-tracking Lagrangian particle method with
different stress and vertex developing in time of streamlines.

Acknowledgments

I wish to thanks my supervisor, Dr Yongmann Chung, for helpful discussions.