1 BOUNDARY ELEMENT METHOD FOR EVALUATION OF THERMAL CREEP FLOW IN MICROGEOMETRIES C. Nieto 1 , H. Power * 2 and M. Giraldo 1 1 Universidad Pontificia Bolivariana, Grupo de Energía y Termodinámica, Facultad de Ingeniería Mecánica 2 The University of Nottingham, Faculty of Engineering, Department of Mechanical, Materials and Manufacturing Engineering. * Corresponding Author ABSTRACT Flow in rarefied gases can be caused by a temperature gradient in the tangential direction without the presence of any pressure gradient. In that case, the fluid is subjected to a driven force that allows its flow from colder to hotter regions. This phenomenon is known as thermal creep and has gained importance in recent years in connection with gas flow in micro scale systems. Examples of Micro Electro Mechanical Systems (MEMS) based on the thermal creep phenomenon, like Knudsen type compressor and gas vacuum pumps, can operate subjected to temperature gradients without moving parts. Prediction of flow behaviour in micro systems had been done by using continuum based models (i.e. Stokes system of equations) under appropriate boundary conditions. Maxwell boundary condition accounts the slip velocity due to tangential shear rate and tangential heat flux effects. These boundary conditions has been misapplied to predict flow behaviours reported in some numerical works dealing with flow fields confinement by curved or moving surfaces, due to wrong evaluation of the local tangential shear rate at the wall surfaces. In this work a boundary integral equation formulation for Stokes slip flow, based on the normal and tangential projection of the Green’s integral representational formulae for the Stokes velocity field, which directly incorporates into the integral equations the local tangential shear rate and heat flux at the wall surfaces, is presented. The tangential heat flux is evaluated in terms of the tangential gradient of the temperature integral representational formulae presenting singularities of the Cauchy type, which are removed by the use of an auxiliary field. These formulations are used to evaluate microchannel flow performance with complex geometries often encountered in MEMS subjected to jump temperatures. KEYWORDS Linear slip boundary conditions, boundary element method, rotating mixers, thermal creep 1. Introduction The role of Nano-scale fluids has become more prevalent after the introduction of chemical and biological analysis and micro electrometrical system fabrication technologies, as they are the key to the mechanism that takes place within these new fields of technologies. When geometry devices are scaled down, the surface-to-volume ratio increases dramatically so that the surface related phenomena become increasingly dominant, e.g. micro heat exchangers and micro mixers present higher heat and mass transfer rates than macro systems of equal capacity [1]. Some new features emerge when mechanical structures are sufficiently small, and it becomes important to understand the various types of interactions that arise between the fluid flow constituents and the solid surfaces that contain it. For instance, a phenomenon known as the slip flow regime could emerges as consequence of an insufficient number of molecules in the sampling region [2] or hydrophilic and hydrophobic recovering quality of surfaces in contact with fluids [3]. Neto et al [4], on a review article, highlights the need of properly describing the flow near the fluid-solid interfaces, because of its relevance to a wide range of applications, from lubrication to micro fluidics. In the case of micro fluid flow of rarefied gases, the thermal creep or transpiration also appears as a consequence of inequalities in temperatures that force the fluid to slide over a surface from colder to hotter regions [5]. Rarefied gas flows are general associated with low-density conditions, such as high-
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
BOUNDARY ELEMENT METHOD FOR EVALUATION OF THERMAL CREEP FLOW IN
MICROGEOMETRIES
C. Nieto1
, H. Power*2
and M. Giraldo1
1
Universidad Pontificia Bolivariana, Grupo de Energía y Termodinámica, Facultad de Ingeniería
Mecánica 2
The University of Nottingham, Faculty of Engineering, Department of Mechanical, Materials and
Manufacturing Engineering.
* Corresponding Author
ABSTRACT
Flow in rarefied gases can be caused by a temperature gradient in the tangential direction without the
presence of any pressure gradient. In that case, the fluid is subjected to a driven force that allows its flow
from colder to hotter regions. This phenomenon is known as thermal creep and has gained importance in
recent years in connection with gas flow in micro scale systems. Examples of Micro Electro Mechanical
Systems (MEMS) based on the thermal creep phenomenon, like Knudsen type compressor and gas
vacuum pumps, can operate subjected to temperature gradients without moving parts. Prediction of flow
behaviour in micro systems had been done by using continuum based models (i.e. Stokes system of
equations) under appropriate boundary conditions. Maxwell boundary condition accounts the slip velocity
due to tangential shear rate and tangential heat flux effects. These boundary conditions has been
misapplied to predict flow behaviours reported in some numerical works dealing with flow fields
confinement by curved or moving surfaces, due to wrong evaluation of the local tangential shear rate at
the wall surfaces. In this work a boundary integral equation formulation for Stokes slip flow, based on the
normal and tangential projection of the Green’s integral representational formulae for the Stokes velocity
field, which directly incorporates into the integral equations the local tangential shear rate and heat flux at
the wall surfaces, is presented. The tangential heat flux is evaluated in terms of the tangential gradient of
the temperature integral representational formulae presenting singularities of the Cauchy type, which are
removed by the use of an auxiliary field. These formulations are used to evaluate microchannel flow
performance with complex geometries often encountered in MEMS subjected to jump temperatures.
KEYWORDS
Linear slip boundary conditions, boundary element method, rotating mixers, thermal creep
1. Introduction
The role of Nano-scale fluids has become more prevalent after the introduction of chemical and biological
analysis and micro electrometrical system fabrication technologies, as they are the key to the mechanism
that takes place within these new fields of technologies. When geometry devices are scaled down, the
surface-to-volume ratio increases dramatically so that the surface related phenomena become increasingly
dominant, e.g. micro heat exchangers and micro mixers present higher heat and mass transfer rates than
macro systems of equal capacity [1]. Some new features emerge when mechanical structures are
sufficiently small, and it becomes important to understand the various types of interactions that arise
between the fluid flow constituents and the solid surfaces that contain it. For instance, a phenomenon
known as the slip flow regime could emerges as consequence of an insufficient number of molecules in
the sampling region [2] or hydrophilic and hydrophobic recovering quality of surfaces in contact with
fluids [3]. Neto et al [4], on a review article, highlights the need of properly describing the flow near the
fluid-solid interfaces, because of its relevance to a wide range of applications, from lubrication to micro
fluidics.
In the case of micro fluid flow of rarefied gases, the thermal creep or transpiration also appears as a
consequence of inequalities in temperatures that force the fluid to slide over a surface from colder to
hotter regions [5]. Rarefied gas flows are general associated with low-density conditions, such as high-
2
altitude and vacuum. However, the small length scales commonly encountered in micro fluid flows imply
that rarefaction effects will be important at much higher pressures, for more details see [6].
Taking into account the advantages of the Boundary Integral Methods for the numerical simulation of
micro fluid flow under slip condition, this work presents a numerical approach based on the use of the
normal and tangential projections of the velocity integral representation formula for Stokes flows,
resulting in weakly singular mixed system of integral equations of the first and second kinds for the
normal and tangential components of the surface traction. The proposed approach is used to study flows
in a different geometries regular present micro fluid flow systems (i.e. eccentric Couette rotating cylinder,
cavity flow, stepped channels) under linear slip conditions with the thermal jump effect at surfaces. The
development of these formulations permits the inclusion of linear slip conditions into the boundary
integral expressions directly, allowing the evaluation of micro flow in plane and curved geometries not
subjected to symmetry conditions. Validation of numerical results is done by direct comparison with
exact solution for the tangential heat flux and interpolation of the temperature derivatives in tangential
direction from the temperature filed.
2. Governing equations
The fluid flow of an incompressible viscous Newtonian fluid with viscosity , can be modelled by the
Navier-Stokes equations of motion, written in dimensionless form as follows:
(1)
being the Reynolds number, defined by where and are some characteristic velocity and
scale length respectively. Further, the fluid flow also satisfy the continuity equation, written in
dimensionless form
(2)
being the velocity vector, the pressure and the Kronecker delta. At micro scale conditions, the
fluid flow occurs at small Reynolds numbers , since the characteristic length scale of the
systems is below 100 µm and the velocity flow is of the order of 1 mm/s [7]. Then, the above system of
equations reduces to its linear quasi-static form defined by the Stokes system of equations, written as
(3)
The use of governing equations based on the continuum for the study of micro and nano flow, requires the
definition of appropriate velocity slip conditions at the wall surfaces to still satisfy the continuum
hypothesis. In order to simulate micro fluid flow of rarefied gases, besides the slip condition due to the
tangential projection of the wall shear rate, the thermal creep effect must be considered. Thermal creep is
a consequence of inequalities in temperatures that forces the fluid to slide over a surface from colder to
hotter regions, [5]. Rarefied gas flows are normally correlated with low-density conditions, for instance
high altitude and vacuum. Nevertheless, the small length scales are commonly encountered in micro fluid
flows, which imply that rarefaction effects will be significant at much higher pressures, for more details
see [6]. In this case, Maxwell slip condition relates the tangential gas velocity at the boundary contours to
the tangential shear stress and heat flux, as:
(4)
with
(5)
where and are respectively the i components of the normal and tangential vectors to a boundary
surface and is the dimensionless tangential heat flux at the boundary contours. In the
3
above expression, is a coefficient constant proportional to the product of the Reynolds number times
the square of the Knudsen number and inversely proportional to the Eckert number.
For the implementation of the slip condition in the velocity integral representational formula for Stokes
flows, the tangential shear rate at solid-fluid interfaces can be evaluated in terms of the surface traction
force which its tangential projection can be expressed as:
(6)
where for consistence with our formulation, the above expression has been written in dimensionless form,
with a characteristic traction force . To complete the boundary conditions at the solid surfaces, the
following non-flux condition across any points x on the boundary surfaces needs to be considered:
(7)
where the following integral relation needs to be satisfied, according to the conservation of mass for an
incompressible fluid:
(8)
with S as the union of the external and internal surfaces and , respectively. For the case of
pressure driven flows, the inlet boundary is defined by a pressure inlet in the normal direction as follows:
(9)
The solution of equation (3) under boundary condition (5) requires the evaluation of the temperature field
with the aim of obtaining the temperature gradient in the tangential direction at the boundary contours.
The temperature field is found by solving the energy equation, written here in dimensionless form as:
(10)
where is the Peclet number and the thermal conductivity. The Peclet number can
be expressed in terms of the Reynolds number through the Prandtl number as . For
cases of , our previous assumption of small Reynolds number implies that , reducing
the above energy equation (at the first order of approximation in ) to the Laplace equation, i.e. quasi-
static approximation;
(11)
In the slip flow region, the temperature field is also subject to a jump condition at the boundary contours,
[8], given by the following equation:
(12)
with as the normal heat flux to the surface and is a coefficient constant function of the
Knudsen and Prandtl numbers and the energy accommodation coefficient. The above jump condition
represents a Robin type boundary condition for the temperature field.
3. Integral formulation for the normal and tangential heat flux with thermal creep
The integral representation formula for the energy equation defined in (11) (i.e. Laplace equation) is
defined as:
(13)
The kernels of the above integral representational formula are for the surface single
layer potential and for the surface double layer
potential, which are weak singular and regular, respectively, where is the Euclidean distance between
point and , i.e. . The temperature jump condition (12) is replaced into the integral
4
representational formula for the temperature (13), leading to the following boundary integral equation for
the unknown normal heat flux:
(14)
The above integral equation can be solved by a classical BEM procedure, quadratic scheme in the present
work, been possible to be written in matrix form as:
(15)
where is the number of elements with surface and the matrices of influence coefficients
and
are defined as:
(16)
(17)
where the integral densities along the element are approximated by quadratic elements,
(18)
being , , the values of on the three nodes of the element with as the local coordinate
of an isoparametric element and interpolation functions given by
(19)
The surface heat flux in the normal direction obtained through the solution of the equation (15) can be
substituted into the temperature jump condition (12) to find the corresponding fluid temperature at contact
with the solid surfaces.
The tangential heat flux at the boundary contours is obtained by taking the limiting value of gradient of
the temperature field, at an internal point, approaching a boundary point, and then multiplying the
resulting expression by the surface tangential vector. In this way the surface gradient of the temperature
field is given by:
(20)
where the first integral is a singular integral of the Cauchy type and the second a hyper singular integral.
However by multiplying the above relation by the surface tangential vector, the resulting integral relation
for the tangential derivative only possesses singularities of the Cauchy type:
(21)
where the limiting value of the above kernels as the distance r tends to zero is given by:
5
The evaluation of the above integrals needs to be considered in the sense of Cauchy principal value.
Various regularization methods to reduce the order of singularity of this type of integrals are available in
the literature; most of them are based on expanding the kernel around the singular point and subtraction of
the most singular part as original suggested by Miklin [9]. A simple alternative is to find the integral
representational formulae of a known potential field, with the same singularity than the field in
consideration. Subtraction of both integral equations removes the most singular part of the integral
operators (for more details see [10]). In general this is not a simple task, but in the present case it is
possible to define the potential
(22)
solution of the Laplace equation , with values of the constants and equal to the
values of the temperature field (13) and its gradient at a given evaluation point . Therefore at the
evaluation point and . By subtracting the integral
representation formula of the tangential derivative of both fields, we obtain:
(23)
At an evaluation point , the above integral relation reduces to the following regular integral equation:
(24)
Given the values of the temperature and its normal derivative , obtained from (12) and (14),
equation (24) provides a linear relationship between the directional derivatives of the temperature at the
evaluation point , i.e. between the values of and . A second relation between
these two values is given by the known normal derivative at , i.e. . From these two expressions, it is possible to obtain the directional derivatives of the
temperature field at each boundary point and find the corresponding tangential derivative to be used in the integral equation (35) and (36) for the normal and
tangential components of the surface velocity in order to determine the corresponding effect of the
thermal creep on the flow field.
4. Integral formulation for slip Stokes flows with thermal creep
The Stokes velocity field has the following direct integral representation formulae for an arbitrary point
in a closed domain filled with a Newtonian fluid [11]:
(25)
where is a constant dependent on the position of the source point. For internal points , for
point at a smooth boundaries and for external points .
The kernels of the above integral representational formula are the Stokeslet and its corresponding surface
tractions or Stresslet, which two dimension expressions are given by:
6
(26)
(27)
Normal and tangential projections of the direct boundary integral representational formula for the Stokes
velocity field can be obtained after multiplying equation (25) by the local normal and tangential vectors
and , at a surface point, and expressing the velocity and surface traction vectors in terms of their
normal and tangential components , and , , i.e. and with and , ), the following surface
integral equations are obtained, for the normal and tangential projections, respectively [12]:
(28)
(29)
These two equations represent a system of surface integral equations for the normal and tangential
velocities and surface tractions , , that can be written in matrix form as:
(30)
where and correspond to the normal and tangential projections vectors respectively and the
matrices of influence coefficients
and
are defined as:
(31)
(32)
where the integral densities along the element are approximated by quadratic elements like in previous
section. The equation (30) is finally written as:
(33)
taking into account that
is defined by:
(34)
The limiting value of the integral kernels in (28) and (29) as the radius r tends to zero is given by:
7
with only logarithmic singularities on and
. In the above limiting
values, the term tends to as the radius r tends to zero, where is the curvature at a
point x on the boundaries (see Courant and Hilbert [13], Vol. 2, page 299).
The equation (33) can be solved when the normal velocity (i.e. boundary condition defined in (7)) and
pressure (i.e. boundary condition defined in (9)) are known, been possible to solve for the unknown
values of , , . The corresponding values of and are directly obtained from the
relations and .
In order to consider the slip regime with thermal creep the slip condition with thermal creep effect,
equation (4), is substituted into the boundary integral equation (28) and (29) leading to the following
system of integral equations for the normal and tangential components of the unknown surface traction:
(35)
and
(36)
that can be written in matrix form as:
8
(37)
and the matrices of influence coefficients
and
are those defined in (31) and (32), and the
coefficient is defined as:
(38)
The additional matrix of influence coefficients
for the thermal creep term is defined as follows:
(39)
Finally, the velocity field is obtained by solving equations (35) and (36) at the boundaries subjected to
slip condition with thermal creep (i.e. solid wall boundaries). The tangential heat transfer to be replaced at
the last term on the right of equations (36) and (37) is evaluated after solving equation (24) for the
temperature gradient components, and , where the values of the normal heat flux
and the temperature jump are obtained through the solution of (14) and the evaluation of (12),
respectively. At the inlet and outlet boundaries, like in the case of pressure driven flows, the velocity and
pressure values are obtained by solving equations (28) and (29), while the temperature is obtained
thorugh the solution of the energy equation (13).
5. Numerical results
In this section is presented the performance of the proposed numerical scheme in terms of the normal and
tangential projections of the direct boundary integral formulation for Stokes velocity fields to predict the
behaviour of slip flow with thermal creep in micro geometries. Initially, a temperature field that satisfy
Laplace equation is used to evaluate the behaviour of the boundary integral formulation proposed for the
solution of the tangential heat flux evaluation in both curved (circle) and corner geometries (square).
Having validated the proposed numerical approach for the tangential heat flux, numerical results for the
eccentric Couette rotating mixer, the cavity flow, and the backward and forward facing steps are
presented with the aim of show the performance of the formulation under mixed boundary conditions.
Additionally, the tangential heat flux at the boundary contours obtained with the integral formulation is
compared with that obtained by direct differentiation of the temperature field in the tangential direction of
the surface interpolation functions, at each element. Since the temperature at the elements in terms of
the interpolation functions is given by:
(40)
then the directional derivative of the temperature field along its surface can be written as:
(41)
where the super index in the temperature denotes the node position along the quadratic elements, and the
interpolation functions derivatives are given by:
(42)
9
From the expression (40), the tangential heat flux can be evaluated from the temperature field at each
boundary point. These results can be used to evaluate velocity field defined by (3) under the thermal
creep effect when is replaced in boundary condition (4) as shown before.
5.1 Assessment of tangential heat flux boundary integral formulation
The assessment of tangential heat flux boundary can be done analytically by using a proposed expression
for the temperature field that satisfy the Laplace equation and then take the tangential derivatives to
compare with results obtained through the numerical approach. In this order of ideas, the following
temperature field is a solution of the Laplace equation (11):
(43)
where the direction derivatives in and directions are, respectively:
(44)
Two cases are proposed with the aim to probe the behaviour of the numerical approach for the evaluation
of the tangential heat flux in curved and corner geometries. The first one is a circle of radios and
centre at the origin, defined by: . For this geometry, the normal and tangential vector are
and , leading to the following expressions for the normal an tangential
heat flux, respectively:
(45)
(46)
The second case is a square centred at the origin with sides’ length , where the normal and
tangential vector and its respectively normal and tangential derivatives at each face are presented in Table
1.
Table 1. . Normal and tangential vectors and temperature derivatives for a square centred at the
origin
Face
Results for temperature, normal and tangential heat flux at the circle’s boundary are plotted in Figure 1,
where open circles represent the numerical solution by the boundary integral approach and the solid line
correspond the expression (45) and (46). In Figure 2 is plotted the same results for the square centred at
the origin. In both cases, the boundary integral approach satisfactory matches the exact solution of the
tangential heat flux. In that way, we can verify that the boundary integral formulation defined in equation
(24) for the tangential heat flux can be used for both types of curved and corner flows.
5.2 Eccentric Couette flow between rotating cylinders
Since the numerical approach for the evaluation of the tangential heat flux has already validate with the
previous examples and the normal and tangential projections of the boundary integral approach has been
used before for the study of linear slip flow (in [12] we present a comparison between numerical results
and exact solution for both concentric and eccentric Couette mixers under the linear slip condition), in
this case is considered the flow between eccentric cylinders as shown in Figure 3. The eccentric Couette
mixer has been previously studied by several authors (see [14], [15] and [16]) due its ability to develop
complex laminar flow: e.g. pressure gradients, Taylor vortices, and eddies. In the present work, an
extension to the numerical approach of the micro fluid flow with BEM is done with the aim to study the
effect of the thermal creep with linear slip condition over the flow behaviour.
10
(A) (A)
(B) (B)
(C) (C)
Figure 1. Assessment of boundary integral
formulation for the evaluation of the boundary
temperature (A), normal heat flux (B) and tangential
heat flux (C) at a circle of radii . (solid line:
exact solution; open circle markers: boundary integral
approach)
Figure 2. Assessment of boundary integral
formulation for the evaluation of the boundary
temperature (A), normal heat flux (B) and tangential
heat flux (C) at a square of side 2 by 2 (solid line:
exact solution; open circle markers: boundary integral
approach)
Figure 3. Eccentric Couette rotating cylinders for evaluation of slip boundary and thermal creep effect
Figures 4 to 6 show the obtained fluid temperature, normal and tangential heat flux distribution at the
rotor and housing boundaries of the eccentric Couette mixer with a given constant temperature at the solid
walls. Two different cases are reported in the figures, corresponding to a hotter and colder rotor, i.e.
and respectively. Values for the same variable under no slip-no thermal creep
conditions are also plotted, showing how the consideration of the thermal creep effect, drastically modify
the patterns of the heat transfer variables. The temperature profiles in Figure 4 (A) and (B) show the
development of temperature of profile under the thermal creep effect. It can be noted that as the no
thermal creep effect is considered, the nondimensional temperature shift down. This can be atribuitted to
the increase in the jump temperature at the wall, and then less heat is transferred from the wall to the
11
adjacent fluid as can be observed in Figures 5 (A) and (B). In Figure 6 can be observed how the results
for the tangential heat flux through the boundary integral approach (filled circle markers) match those
obtained by interpolation evaluation (solid line), allowing us to conclude that the first one satisfactory
reproduce thermal creep effect, since the last one results from the direct differentiation of the
interpolation functions as presented before.
(A)
(B)
Figure 4. Temperature profiles for the eccentric Couette mixer ( with thermal creep and
(solid line with open markers) and non thermal creep effect and (solid line with
open markers): (A) and , (B). Circle markers: inner rotor; triangle markers: static
housing
(A) (B)
Figure 5. Normal heat flux for the eccentric Couette mixer ( with thermal creep and
(right side - solid line with open markers) and non thermal creep effect and
(left side - solid line with filled markers): (A) and , (B). Circle markers: inner rotor;
triangle markers: static housing.
12
(A)
(B)
Figure 6. Tangential heat flux for the eccentric Couette mixer ( with thermal creep and
(solid line with open markers): (A) and , (B). Circle markers: inner rotor;
triangle markers: static housing.
In Figure 7, we report the corresponding streamlines inside the mixer in the case of a rotating rotor and
the interaction between both the slip regime and the thermal creep induced fluid motion. Figure 7 (A)
presents the reference case of the no slip-no thermal creep condition with the characteristic vortex
formation at the mixers bottom. The consideration of the slip condition tends to reduce the vortex size due
to the reduction in momentum difussion at both rotating and stationary boundary walls (see Figures 7 (B))
and tends to dissapear as the slip condition is increased (e.g when is aumented from to
as shown in Figure 7 (C)). In the other hand, the inclusion of the thermal creep with no slip condition
conduce to the original recirculation zone observed under only no slip condition, plus an asymmetric
recirculation region around the rotor side, where the position of the recirculation changes according to the
direction of the heat flux, i.e. or (see Figures 7 (G) and (M)). Similar recirculation
patterns are reported by Lockerby et.al. [6] in the case of eccentric circular cylinders, however in their
analysis they do not use a jump temperature condition, as in (12), but instead they use a high order
thermal stress slip condition, where the tangential velocity is proportional to the tangential derivative of
the normal heat flux. As the slip over boundaries is considered, the original recirculation zone disappear
at values of , while the second recirculation induced by the thermal creep increase its size
detaching from the rotor’s head and moves to the mixer´s bottom (see Figures 7 (H) and (N)). Finally,
when the slip length is increased from to , a recirculation zone corresponding in position
to that one observed in the reference case (i.e. the no slip one condition) reapers while the thermal creep
one remains as beforehand (see Figures 7 (I) and (O)). The direction and position of the recirculation
zones are associated with temperature gradient direction. In Figures 7 (D) to (F) and (J) to (L) the
streamlines for the both conditions, i.e. or , are poltted. The flow pattern in both cases
shown iddentical behviour but the flow direction (see vector field) is inverted since the temperature
gradient is inverted too.
13
(A) (B) (C)
(D) (E) (F)
(G) (H) (I)
(J) (K) (L)
(M) (N) (O)
Figure 7. Streamlines for the eccentric Couette mixer. (A) to (C) are for , , (
(A), (B), (C)); (D) to (F) are for , , : ( (D),
(E), (F)); (G) to (I) are for , , , ( (G), (H),
(I)); (J) to (L) are for , , , : ( (J), (K), (K)); (M) to
(O) are for , , , ( (M), (N), (O)).
14
5.3 Cavity flow
To test the developed numerical scheme for slip flow with thermal creep in corner geometries, we also
considered the problem of a shear driven cavity flow (see Figure 8). In this type of flow, the motion is
caused by horizontal velocity imposed at the upper horizontal wall. In our case, the slip flow condition is
considered and the thermal creep effect is prompted by a difference in temperatures between the
horizontal and the vertical boundaries. In this work two cases are considered: and , where
refers to the horizontal flat walls and is defined at the vertical flat walls. The effect of the variation
in the heat flux will be analysed over the velocity field.
Figure 8. Cavity flow for evaluation of slip boundary and thermal creep effect. Th temperature at
horizontal boundary condition and Tv temperature at vertical boundary condition
In Figures 9 and 10 results for the temperature, normal and tangential heat flux are plotted for both
temperature cases, respectively. The temperature profiles plotted in Figure 9 (A) and Figure 19 (A)
present a similar behaviour than for the eccentric Couette flow, since the nondimensional temperature
shift down as the thermal creeping appears due to reduction in the normal heat transfers between the walls
and the adjacent fluid (see Figure 9 (B) and Figure 10 (B)). It possible to make a similar conclusion
regarding the tangential heat flux since it was evaluated by both boundary integral and interpolation
approach of the tangential heat flux, showing a perfect match of the first one in relation with last one, as
can be observed in Figure 9 (C) and Figure 10 (C).
The streamlines for the shear driven cavity flow are plotted in Figure 11. Figures 11 (A) to (C) presents
the results for the flow without the thermal effect, where Figure 11 (A) correspond the well-known no slip
condition that had been studied for other authors before [17]-[18]. As the slip is considered the
recirculation pattern is slightly modified respect to the no slip condition due to the reduction in the
momentum diffusion, tending to produce an almost symmetric recirculation vortex (see Figures 11 (B)
and (C)). Once the thermal creep is considered it leads to the formation of four recirculation zones into
the cavity for both the no slip and slip conditions. The dominated recirculation zone is up to the direction
of the heat flux, i.e. or , as can be observed in Figures 11 (G) to (I) for the case when
and in Figures 11 (M) to (O) for the case when . For both the no slip and slip condition
with slip length of the recirculation cells are asymmetrical, while when the slip length is
increased to , the four recirculation zones becomes almost symmetrical showing that the thermal
creep induced movement can overcome the shear driven flow. The degree of asymmetry in the vortex
formation (e.g in the cases when and ) can be explained by the flow formation associated
to the only thermal creep driven flow ( ) as shown in Figures 11 (D) to (F) for and in
Figures 11 (J) to (L) for . Identically to the eccentric Couette flow, in the case the flow pattern for
both temperature cases are equivalents but the flow direction is inversed as the temperature gradient is
opposite leading to a flow behaviour principally dominated by thermal creep effect in both stationary and
moving wall cases.
15
(A) (A)
(B) (B)
(C) (C)
Figure 9. Evaluation of dimensionless temperature (A) and normal (B) and tangential
(C) heat flux with boundary integral formulation (open circles) and interpolation
differentiation (solid line) approaches for the cavity flow with: = 1. Temperature of
horizontal boundaries is greater than the vertical boundaries
Figure 10. Evaluation of dimensionless temperature (A) and normal (B) and tangential
(C) heat flux with boundary integral formulation (open circles) and interpolation
differentiation (solid line) approaches for the cavity flow with: = 1. Temperature of
vertical boundaries is greater than the horizontal boundaries
16
(A) (B) (C)
4
(D) (E) (F)
(G) (H) (I)
(J) (K) (L)
(M) (N) (O)
Figure 11. Streamlines for the cavity flow. (A) to (C) are for , , ( (A),
(B), (C)); (D) to (F) are for , , , : ( (D), (E),
(F)); (G) to (I) are for , , , : ( (G), (H), (I));
(J) to (L) are for , , : ( (J), (K), (L)); (M) to (O) are
for , , : ( (M), (N), (O))
17
The horizontal velocity distribution at vertical central line and vertical velocity distribution at horizontal
central line are plotted in Figure 12 (A), that shown a comparative behaviour with results presented in
previous works [17]-[18] for the no slip condition . As the slip condition is considered, the
velocity profiles at both vertical and central lines are modified due to the expected reduction in
momentum transfer at the fluid-walls interface. The result plotted in Figure 12 (B) correspond to the case
when thermal creep is analyzed at the central lines. The velocity profiles for this cases are the same for
the both temperature gradient cases due the tangential heat flux presents a zero value at this points (see
Figure 9 (C) and Figure 10 (C)) corresponding to a maximum and minimum values in the temperature
distribution as shown in Figure 9 (A) and Figure 10 (A), not allowing any thermal creep driven force at
this points.
(A)
(B)
Figure 12. Horizontal velocity distribution at vertical central line and vertical velocity distribution at
horizontal central line with slip condition and (A), with slip condition with thermal creep
(B).
5.4 Backward and forward facing steps
Micro channel has been studied previously by a few authors [19]-[20] due to their significant role in
micro fluidic industry. In this case, the simulation of micro fluid flowing through backward and forward
facing stepped channels with the geometry as shown in Figure 13 is studied due to its strong presence in
various applications such as Y and T type micro mixers [22],[23], heat exchangers [8], Knudsen type
compressor and gas vacuum pumps [21]. The analysis of both stepped channels is considered under linear
slip condition coupled with the thermal creep effect. In this type of problem geometry, we will
concentrate on the flow behaviour under three slip length conditions ( ) and will see the effect
of the thermal creep movement induced over the fluid flow behaviour. In a similar way like in previous
problems, two different temperature conditions will be imposed (temperature of the lower boundary, ,
greater than the upper boundary , , and temperature of the lower boundary greater than the upper
boundary) to allows the study of the heat flux direction effect over velocity patterns. At the inlet and
18
outlet boundaries a Neumann type boundary condition is applied , meaning that we will
consider that temperature field will be fully developed inside the stepped channels.
(A) (B)
Figure 13. Backward facing step (A) and forward facing step (B) channels for evaluation of slip flow
with thermal creep effect
The temperature profiles in Figure 14 (A) and (B) show the development of temperature of profile under
the thermal creep effect. Similar than to the previous results for the eccentric Couette and the cavity flow,
it can be noted that as the no thermal creep effect is considered, the nondimensional temperature shift
down. The reduction in the temperature can be associated to two major aspects: the slip condition implies
that more fluid passes through the channel (i.e., the heat source at the wall must heat a greater amount of
fluid); as the thermal creep effect is considered (i.e. is changed from 0 to 1), the jump in temperature
at the wall will increase, which means that less heat is transferred from the wall to the adjacent fluid).
(A) (B)
Figure 14. Temperature profiles at the facing steps (backward and forward) under different temperature
gradients: (A) and (B). Solid line open circles are for no thermal creep results; dashed
line filled circles are for thermal creep results ( ).
The flow pattern for the no slip-no thermal creep flow is shown in Figures 15 (A) and 16 (A) for the
backward and forward facing step channels, respectively. The results follow a very reasonable trend since
we are dealing with the incompressible viscous Stokes flow and no recirculation zones are expected after
(before) the backward (forward) step. As the slip length increases, the velocity of the flow increases
proportionally with the slip length, tending to completely attach to the step’s walls despite the flow
direction (backward and forward) as can be observed in Figure 15 (B) and 16 (B).
Figures 15 (E) and (F), illustrates the effect of thermal creep as recirculation appears for the backward
facing step when . In Figure 15 (E) can be observed two thermal creep induced recirculation zones
downstream the step (one at the bottom and another at the top) for the settings and and
, which can be explained with the flow pattern present in the no pressure gradient-thermal
creeping flow plotted in Figure 15 (C). This behaviour tends to reduce its effect when the slip condition is
included leading to the elimination of the upper vortex (see Figure 15(F) for and and
), despite the induced vortex observed when the temperature gradient is the only driven force
19
(Figure 15 (D)). This can be understood when the flow direction is observed in both cases when
and or , since the former is against the flow while the latter contributes to it.
When the heat flux direction is inverted (i.e. ) and the no slip condition and the thermal creep are
considered, one recirculation zone appears downstream the step´s bottom rear as can be observed in
Figure 15 (I). As the thermal creep effect is included alongside the slip regime (Figure 15 (J)), the
previous behaviour is inverted and the thermal creep recirculation moves upstream the step. This explain
by the inversion in flow velocities respect to the previous case when thermal creep is the only driven
force, as can be observed in Figures 15 (G) and (H). Similar behaviours are observed in the case of the
forward facing step. However, in these cases the formation of recirculation is given in a manner contrary
to what happens in the backward facing step as can be observed in Figure 16.
(A) (B)
(C) (D)
(E) (F)
(G) (H)
(I) (J)
Figure 15. Streamlines for the backward facing step. Results in (A) and (B) are under linear slip
condition, pressure gradient ( ) and no thermal creep condition , ( (A),
(B)); (C) and (D) are under linear slip condition, no pressure gradient ( ) and thermal creep
condition , with : ( (C), (D)); (E) and (F) are under linear slip
condition, pressure gradient ( ) and thermal creep condition , with : (
(E), (F)); (G) to (H) are under linear slip condition, no pressure gradient ( ) and thermal
creep condition , with : ( (G), (H)); (I) and (J) are under linear slip
condition, pressure gradient ( ) and thermal creep condition , with : (
(I), (J))
20
(A) (B)
(C) (D)
(E) (F)
(G) H)
(I) (J)
Figure 16. Streamlines for the forward facing step. Results in (A) and (B) are under linear slip condition,
pressure gradient ( ) and no thermal creep condition , ( (A), (B)); (C)
and (D) are under linear slip condition, no pressure gradient ( ) and thermal creep condition
, with : ( (C), (D)); (E) and (F) are under linear slip condition,
pressure gradient ( ) and thermal creep condition , with : ( (E),
(F)); (G) to (H) are under linear slip condition, no pressure gradient ( ) and thermal creep
condition , with : ( (G), (H)); (I) and (J) are under linear slip
condition, pressure gradient ( ) and thermal creep condition , with :
( (I), (J))
6. Conclusions
In this work, has been evaluated the thermal creep effect at micro scale flows. This phenomenon usually
takes place due to micro fluid flow of rarefied gases and could affect the velocity field due to a heat flux
and a temperature difference into the domain. An integral equation approach based on the normal and
tangential projections of the direct boundary integral representational formula for the Stokes velocity field
is developed for the numerical simulation of creeping flow under linear slip boundary conditions and
applied to analysis the performance of a three different cases under variable boundary conditions. The slip
condition was included in the boundary integral formulation by expressing the tangential shear rate in
terms of tangential component of the surface traction vector. The projection of the velocity integral
representational formula on the normal and tangential directions smoothest the singularity of the integral
kernels resulting only in a weak singular kernel of the logarithm type, which can be numerical integrated
by using Telles' transformation and standard Gaussian Quadrature formulae.
21
The integral formulation has been modified to account the thermal creep effect which is function of the
tangential heat flux. For the latter, an integral representational formulae was obtained by taking the
limiting value of gradient of the temperature field and multiplying the resulting relation by the surface
tangential vector, presenting only singularities of the Cauchy type. The order of singularity of this type of
integrals was reduced by finding the integral representational formulae of a known potential field, with
the same singularity than the field in consideration, reducing the expression to a regular integral equation
at the evaluation point. The tangential heat transfer attained through this formulation where satisfactorily
validated in the first instance with results obtained for a circular and square region shape where exact
solution was developed for a temperature field proposed as solution of the Laplace equation.
Additionally, a second comparison was done by evaluation of the temperature derivatives in tangential
direction by using the surface interpolation functions, at each element and the temperature field
obtained through the boundary integral representation of Laplace equation with temperature jump. In the
case, the boundary integral formulation proposed to evaluate the tangential heat transfer also matched the
results obtained by the interpolation approach values allowing us its use for the study of flows confined
by even both curved and corner geometries. The formulation gave appropriate results for geometries with
moving boundary conditions as well as with input and output boundary conditions.
ACKNOWLEDGEMENTS
The first author acknowledges the support of the PhD program at the Universidad Pontificia Bolivariana
and COLCIENCIAS, Colombia, under the support provided with the project "Desarrollo de geometrías
para aplicación industrial en microintercambiadores de calor" code 1210-479-21999, contract 436-2008.
REFERENCES
[1] G. Hu and D. Li, Multiscale phenomena in microfluidics and nanofluidics, Chemical Engineering
Science, 62 (2007), 3443–3454.
[2] P. A. Thompson and S. M. Troian, A general boundary condition for liquid flow at solid surfaces,
Nature, 389 (1997), 360–362.
[3] O.I. Vinogradova, Slippage of water over hydrophobic surfaces, Int. Journal of Miner. Process., 56
(1999), 31–60.
[4] C. Neto, D.R. Evans, E. Bonaccurso, H. Butt and V. Craig, Boundary slip in Newtonian liquids: a
review of experimental studies, Rep. Progr. Phys., 68 (2005), 2859–2897.
[5] G. Karniadakis, A. Beskok and N. Aluru, “Microflows and Nanoflows: Fundamentals and
Simulation,” 1st edition, Springer, New York, 2005.
[6] D. Lockerby, J. M. Reese, D. R. Emerson and R. W. Barber. Velocity boundary condition at solid
walls in rarefied gas calculations, Physical Review E 70, (2004), 017303-1 - 017303-9.
[7] N. Nguyen and S. Wereley, “Fundamentals and Applications of Microfluidics”, 2nd edition, Artech House, Norwood, 2006.
[8] S. Kandlikar (editor), “Heat transfer and fluid flow in minichannels and microchannels”, 1st edition, Elsevier, Boston, 2006.
[9] S.G. Mikhlin, Multidimensional Singular Integrals and Integral Equations. Pergamon Press, New
York, 1957.
[10] H. Chen, J. Jin, P. Zhang and P. Lu, Multi-variable non-singular BEM for 2-D potential problems,
Tsinghua Science and Technology, Vol 10, 1, 43-50.
[11] H. Power and L.C. Wrobel, “Boundary Integral Methods in Fluid Mechanics,” 1st edition, Computational Mechanics Publications, Southampton, 1995.
[12] C. Nieto, M. Giraldo and H. Power, Boundary Integral Equation Approach for Stokes Slip Flow in
Rotating Mixers, Discrete and Continuous Dynamical Systems - Series B, In review.
[13] R. Courant and D. Hilbert, Methods of mathematical physics, 3rd
edition, Cambridge University
Press, Cambridge, 1962.
[14] Z. Ping, W. Cheng-wei and MA Guo-jun, Nonlinear boundary slip of fluid flowing over solid
surface, J. Cent. South Univ. Technol., (2007), 30-33.
[15] J. Maureau, M.C. Sharatchandra, M.Sen and M. Gad-el-Hak, Flow and load characteristics of
microbearings with slip, J. Micromech. Microeng., 7 (1997), 55–64.
[16] V. Hessel, H. Lwe, F. Schnfeld, Micromixers: a review on passive and active mixing principles,
Chemical Engineering Science , 60 (2005), 2479–2501.
22
[17] Ghia U., Ghia K.N., and Shin C.T., High-Re solutions for incompressible flow using the Navier-
Stokes equations and a multigrid method, Journal of Computational Physics, 48 (1982), 387–411.
[18] Erturk E. and Gökcöl C., Numerical solutions of 2-d steady incompressible driven cavity flow at high
Reynolds numbers, Intenational Journal for Numerical Methods in Fluids, 48 (2005), 758–769.
[19] Yin Xie-Yuan, Qin Feng-Hua and Sun De-Jun, Perturbation analysis on gas flow in a straight
microchannel, Physics of Fluids, 19-2 (2007), 027103-027103-14.
[20] Z.A. Zainal, Pradeep Hegde, K.N. Seetharamu, G.A. Quadir, P.A. Aswathanarayana and M.Z.
Abdullah, Thermal analysis of micro-channel heat exchangers with two-phase flow using FEM,
International Journal for Numerical Methods in Heat & Fluid Flow, 15 (2005), 43-60.
[21] Vargo, S. E.; Muntz, E. P.; Shiflett, G. R.; Tang, W. C., Knudsen compressor as a micro- and
macroscale vacuum pump without moving parts or fluids, Journal of Vacuum Science & Technology
A: Vacuum, Surfaces, and Films, 17 (1999), 2308-2313.
[22] E.A. Mansur, Y. Mingxing, W. Yundong and D. Youyuan, A state-of-the-art review of mixing in
microfluidic mixers, Chin. J. Chem. Eng., 16 (2008), 503–516.
[23] V. Hessel, H. Lwe, F. Schnfeld, Micromixers: a review on passive and active mixing principles,
Chemical Engineering Science , 60 (2005), 2479–2501.