22-Mec-B6 Advanced Fluid Mechanics · December 2016
Question 4 of 6: Rigid-body rotation of water in a spinning container
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. National Exams, December 2016 —
07-Mec-B6 Advanced Fluid Mechanics. Three hours, OPEN BOOK, any approved
Sharp or Casio calculator permitted. Six questions are printed; page 1 states that any five
of them constitute a complete paper for 100 marks, each question carries an equal 20 marks, and the
item weights are shown in the left margin. No aid sheet is bound into the paper — the open-book rule is the candidate’s table source, so every compressible-flow ratio below is quoted in closed form rather than read off a chart. All six questions are solved here.
Reference texts. The solutions follow the conventions of the texts the
EGBC syllabus recommends for this subject:
F. M. White, Fluid Mechanics, 8th ed. — potential-flow building blocks
(§4.4, §8.2–8.3), rigid-body motion of a fluid (§2.9), flow in
non-circular ducts and minor losses (§6.6–6.7), dimensional analysis
(§5.2–5.4), the momentum integral (§7.4), compressible nozzle flow and normal
shocks (§9.4–9.6).
F. M. White, Viscous Fluid Flow, 3rd ed. — exact Navier–Stokes solutions in
cylindrical coordinates (§3.2–3.4) and integral boundary-layer methods
(§4.5).
J. D. Anderson, Modern Compressible Flow, 3rd ed. — quasi-one-dimensional nozzle
flow, choking and normal shocks (Ch. 3, §3.3 and §3.6).
P. K. Kundu, I. M. Cohen & D. R. Dowling, Fluid Mechanics, 6th ed. — the
complex potential and elementary singularities (Ch. 6); boundary layers (Ch. 9).
H. Schlichting & K. Gersten, Boundary-Layer Theory, 9th ed. —
Kármán–Pohlhausen integral method for accelerating external flows
(Ch. 8).
R. W. Fox, A. T. McDonald & J. W. Mitchell, Introduction to Fluid Mechanics,
10th ed. — the repeating-variable procedure and pipe-system energy accounting
(Ch. 7, Ch. 8).
Question 4: Rigid-body rotation of water in a spinning container (20 marks)
Given. An incompressible Newtonian liquid of constant density
ρ and viscosity μ, filling a cylindrical container that has been spinning
long enough about the vertical z axis for the liquid to turn with it as a solid body at
constant angular velocity ω. The velocity field is therefore
$v_r=v_z=0$, $v_\theta=\omega r$, the motion is steady and axisymmetric, gravity acts in the
−z direction, and the free surface is exposed to a uniform ambient pressure
pa.
Find. The reduced r- and z-momentum equations with a
justification for discarding the viscous terms, the pressure field p(r, z)
throughout the liquid, and from it the equation of the free surface.
Figure 4 — The spinning container. Once the transient has decayed the liquid turns as a rigid body and the free surface is a paraboloid of revolution whose vertex sits on the axis; the origin of the coordinates is placed at that vertex.
Approach. Insert the rigid-body velocity field into the cylindrical
Navier–Stokes equations, show that the deformation-rate tensor vanishes identically so the
viscous term is exactly zero, then integrate the two surviving pressure gradients and apply the
constant-pressure condition at the free surface.
Part (a) — reduce the r-momentum equation. With
$v_r=v_z=0$, $v_\theta=\omega r$, and $\partial/\partial t=\partial/\partial\theta=0$, the only
surviving inertia term in the radial direction is the centripetal one:
$$-\rho\frac{v_\theta^{2}}{r}=-\frac{\partial p}{\partial r}+\mu\left[\ldots\right]
\quad\Longrightarrow\quad \frac{\partial p}{\partial r}=\rho\frac{v_\theta^{2}}{r}=\rho\omega^{2}r .$$
Physically, a radial pressure gradient is what supplies the centripetal acceleration of each fluid
ring.
Reduce the z-momentum equation. There is no vertical acceleration and no
vertical velocity, so the vertical equation is a pure hydrostatic balance,
$$0=-\frac{\partial p}{\partial z}-\rho g \quad\Longrightarrow\quad
\frac{\partial p}{\partial z}=-\rho g .$$
The θ-momentum equation is satisfied identically because nothing varies with
θ and there is no azimuthal forcing.
Show that the viscous stresses vanish exactly, not merely approximately. The
only candidate viscous term acts on vθ, and in cylindrical coordinates it is
$$\mu\left[\frac{\partial}{\partial r}\left(\frac{1}{r}\frac{\partial (r v_\theta)}{\partial r}\right)\right]
=\mu\frac{\partial}{\partial r}\left(\frac{1}{r}\frac{\partial(\omega r^{2})}{\partial r}\right)
=\mu\frac{\partial}{\partial r}\left(2\omega\right)=0 .$$
The deeper reason is kinematic: in rigid-body motion every fluid particle keeps its position relative
to every other, so the rate-of-strain tensor is identically zero. A Newtonian fluid generates viscous
stress only in response to deformation, never to rotation, so a fluid in solid-body rotation
carries no viscous stress at all. This is why the answer is exact rather than a high-Reynolds-number
approximation, and it is also why μ never appears in the final result.
Part (b) — integrate the radial equation. Holding z fixed,
$$p(r,z)=\int \rho\omega^{2}r\,\mathrm{d}r = \frac{\rho\omega^{2}r^{2}}{2}+F(z),$$
where F is so far an arbitrary function of z alone.
Integrate the vertical equation to pin down F(z). Differentiating the last
result with respect to z and matching step 2,
$$\frac{\mathrm{d}F}{\mathrm{d}z}=-\rho g \quad\Longrightarrow\quad F(z)=-\rho g z + C .$$
Hence the pressure everywhere in the liquid is
$$p(r,z)=\frac{\rho\omega^{2}r^{2}}{2}-\rho g z + C .$$
Fix the constant of integration. Place the origin at the vertex of the free
surface, where r = 0, z = 0 and the pressure equals the ambient value
pa. Then C = pa and
$$\boxed{p(r,z)=p_a+\frac{\rho\omega^{2}r^{2}}{2}-\rho g z}$$
The result is the familiar hydrostatic law with one extra term, and that term is the centrifugal
contribution: at fixed depth the pressure rises quadratically outward.
Part (c) — impose the free-surface condition. The free surface is by
definition the locus on which the pressure equals the ambient value, so set
p = pa in the boxed result:
$$p_a=p_a+\frac{\rho\omega^{2}r^{2}}{2}-\rho g z_{\text{surface}}
\quad\Longrightarrow\quad \rho g z_{\text{surface}}=\frac{\rho\omega^{2}r^{2}}{2}.$$
Read off the shape. The density cancels, leaving
$$\boxed{z_{\text{surface}}=\frac{\omega^{2}r^{2}}{2g}}$$
as required. The surface is a paraboloid of revolution, its shape independent of which liquid is in
the container. Note also that the isobars are the family
$z=\dfrac{\omega^{2}r^{2}}{2g}-\dfrac{p-p_a}{\rho g}$ — congruent paraboloids stacked
vertically, of which the free surface is simply the one at ambient pressure.
Put numbers on the result. For a container of radius R = 0.15 m spun at
ω = 10 rad/s, the rim of the surface stands
$z_s(R)=(10)^{2}(0.15)^{2}/(2\times9.81)=0.1147$ m above the vertex. Because the volume under a
paraboloid is half that of the enclosing cylinder, the surface drops
$\omega^{2}R^{2}/4g=0.0573$ m at the axis and rises the same 0.0573 m at the wall relative to the
liquid level at rest — so at least 57 mm of freeboard is needed if the container is not to
overflow. The gauge pressure at the wall in the plane of the vertex is
$\rho\omega^{2}R^{2}/2=998(100)(0.0225)/2=1123$ Pa.
Item
Result
(a) r-momentum
∂p/∂r = ρω2r
(a) z-momentum
∂p/∂z = −ρg
(a) Viscous stresses
identically zero — rigid-body motion has zero rate of strain
(b) Pressure field
p = pa + ρω2r2/2 − ρgz
(c) Free surface
zsurface = ω2r2/2g (paraboloid)
Worked illustration (ω = 10 rad/s, R = 0.15 m)
rim rise above vertex 0.1147 m; axis drop 0.0573 m; wall gauge pressure at the vertex plane 1123 Pa