Unit content
Lubrication approximation for thin viscous gaps
When a viscous fluid occupies a gap whose thickness is much smaller than its length, the geometry creates a strong separation of scales. The lubrication approximation exploits that separation to simplify Navier-Stokes flow.
Let
- $H$ be a characteristic gap thickness;
- $L$ be the characteristic distance along the gap;
- $U$ be a characteristic tangential speed.
Define the small aspect ratio
$$\boxed{\varepsilon=\frac{H}{L}\ll1}.$$
The fluid is assumed Newtonian and incompressible.
Why transverse viscous gradients dominate
For a velocity component $u$ along the gap,
$$\frac{\partial^2u}{\partial y^2}\sim\frac{U}{H^2},$$
while
$$\frac{\partial^2u}{\partial x^2}\sim\frac{U}{L^2}.$$
Their ratio is
$$\frac{U/L^2}{U/H^2}=\left(\frac{H}{L}\right)^2=\varepsilon^2\ll1.$$
Thus viscous diffusion across the thin direction is much stronger than viscous diffusion along the gap.
When inertia is negligible
The axial inertial scale is approximately
$$\rho\frac{U^2}{L},$$
while transverse viscous stress contributes a momentum scale
$$\mu\frac{U}{H^2}.$$
Their ratio is
$$\frac{\rho U^2/L}{\mu U/H^2} =\frac{\rho UL}{\mu}\left(\frac{H}{L}\right)^2.$$
Defining
$$Re_L=\frac{\rho UL}{\mu},$$
inertia is small compared with transverse viscosity when
$$\boxed{Re_L\varepsilon^2\ll1}.$$
Lubrication flow can therefore be viscosity-dominated even when a Reynolds number based on the long length $L$ is not itself extremely small.
Leading-order equations
For a slowly varying gap $y=h(x,t)$ with $h/L\ll1$, the leading-order axial momentum balance becomes
$$\boxed{0=-\frac{\partial p}{\partial x} +\mu\frac{\partial^2u}{\partial y^2}}.$$
The transverse momentum balance implies, to leading order,
$$\boxed{\frac{\partial p}{\partial y}\approx0}.$$
So pressure is nearly uniform across the thin gap:
$$p\approx p(x,t).$$
The key idea is local fully developed flow. The gap can vary slowly with $x$ and time, but over each short axial slice the velocity profile is approximately the parallel-plate Couette-Poiseuille profile corresponding to the local gap thickness and local pressure gradient.
Local velocity profile
Take the lower wall at $y=0$ moving in the positive $x$ direction with speed $U$, while the upper wall at $y=h(x,t)$ has zero tangential speed. No slip gives
$$u(0)=U, \qquad u(h)=0.$$
Using the local Couette-Poiseuille solution,
$$\boxed{ u(x,y,t)=U\left(1-\frac{y}{h}\right) -\frac{1}{2\mu}\frac{\partial p}{\partial x},y(h-y)}.$$
The first term is wall-driven Couette flow. The second is pressure-driven Poiseuille flow.
Local flow rate
For a two-dimensional film, define the volumetric flow rate per unit width
$$q(x,t)=\int_0^{h(x,t)}u(x,y,t),dy.$$
Integrating the local profile gives
$$\boxed{q=rac{Uh}{2} -\frac{h^3}{12\mu}\frac{\partial p}{\partial x}}.$$
This compact relation is the central output of the lubrication approximation. It converts a two-dimensional velocity field into a one-dimensional relation among
- local gap thickness $h$;
- wall motion $U$;
- pressure gradient $\partial p/\partial x$;
- local flow rate $q$.
Worked local calculation
Suppose a film has
$$h=0.50,\mathrm{mm}=5.0\times10^{-4},\mathrm m,$$
$$U=1.0,\mathrm{m/s},$$
$$\mu=0.050,\mathrm{Pa,s},$$
and local pressure gradient
$$\frac{\partial p}{\partial x}=-2.0\times10^5,\mathrm{Pa/m}.$$
The wall-driven contribution is
$$q_C=\frac{Uh}{2} =\frac{(1.0)(5.0\times10^{-4})}{2} =2.50\times10^{-4},\mathrm{m^2/s}.$$
The pressure-driven contribution is
$$q_P=-\frac{h^3}{12\mu}\frac{\partial p}{\partial x}$$
$$=-\frac{(5.0\times10^{-4})^3}{12(0.050)}(-2.0\times10^5) \approx4.17\times10^{-5},\mathrm{m^2/s}.$$
Hence
$$\boxed{q\approx2.92\times10^{-4},\mathrm{m^2/s}}.$$
Both mechanisms drive flow in the positive $x$ direction in this example.
Scope of the approximation
Lubrication theory requires more than simply calling a gap "small." Its central assumptions are a slender geometry, slowly varying boundaries, dominant viscous transport across the gap, and sufficiently weak inertia. Abrupt geometry changes, strong separation, large slopes, turbulence, or inertia-dominated motion can invalidate the approximation.
When its assumptions hold, lubrication theory replaces a difficult multidimensional flow problem with a local Couette-Poiseuille profile plus a one-dimensional mass-conservation problem along the film.