Question 6 of 6: Question 6 (Part B, Question B3): Two Immiscible Liquids in a Pressure-Driven Two-Dimensional Channel
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. National Exams, May 2013 — 07-Mec-B6
Advanced Fluid Mechanics. Three hours, open book, any non-communicating
calculator permitted. Part A holds three questions of 20 marks each (40 per cent of the
paper) and Part B three questions of 30 marks each (60 per cent); the candidate answers
any two in each part. All six questions are worked here, because the set is a
study resource rather than a three-hour sitting. The paper supplies an aid sheet of compressible-flow,
boundary-layer, Navier–Stokes and potential-flow relations, and the coefficients quoted below
are taken from that sheet so that the arithmetic matches what a candidate had in front of them.
Reference texts.
F. M. White, Fluid Mechanics, 8th ed. — Ch. 5 (dimensional analysis and
similitude), Ch. 7 (external flow), Ch. 8 (potential flow), Ch. 9 (compressible flow),
Ch. 10 (open-channel flow).
F. M. White, Viscous Fluid Flow, 3rd ed. — Ch. 3 (exact solutions of the
Navier–Stokes equations), Ch. 6 (turbulent wall flow).
J. D. Anderson, Modern Compressible Flow, 3rd ed. — Ch. 3 (normal shock waves),
Ch. 5 (quasi-one-dimensional nozzle flow).
P. K. Kundu, I. M. Cohen and D. R. Dowling, Fluid Mechanics, 6th ed. —
Ch. 6 (irrotational flow), Ch. 9 (laminar flow).
H. Schlichting and K. Gersten, Boundary-Layer Theory, 8th ed. — Ch. 21
(plate drag, mixed laminar/turbulent layers).
R. W. Fox, A. T. McDonald and J. W. Mitchell, Introduction to Fluid Mechanics,
10th ed. — Ch. 7 (similitude), Ch. 11 (open-channel flow).
Question 6 (Part B, Question B3): Two Immiscible Liquids in a Pressure-Driven Two-Dimensional Channel (30 marks)
Find. The boundary conditions, the two-layer velocity profile,
the shear stresses at both walls and the interface, the interface speed, and the location and
value of the maximum velocity.
Check: the question text states dP/dx = −1 Pa/m while the label inside Figure B3 and its caption read −1 kPa/m. The written question governs, and it is also the only self-consistent choice: at −1 Pa/m the peak velocity is 17.5 mm/s and the channel Reynolds number is 365, comfortably laminar, whereas −1 kPa/m would give 0.0 m/s and Re ≈ 365,152, which is fully turbulent and would invalidate the very laminar analysis the question asks for. Every result below is exactly linear in the pressure gradient, so if the figure is preferred, multiply every velocity and every stress by 1000.
Approach. With the flow steady, fully developed and
two-dimensional, the Navier–Stokes equations reduce in each layer to a balance between the
constant pressure gradient and the viscous term; integrate twice per layer and fix the four
constants with no-slip at the two walls plus continuity of velocity and of shear stress at the
interface.
Figure 6.1 — Computed velocity profile. The origin is at the interface, the walls are at y = ±h, and the thin, low-viscosity top fluid carries much the faster stream, pushing the peak above the centreline.
Part (a) — reduce the governing equations and state the boundary
conditions. The channel is long, horizontal and two-dimensional, and the flow is steady
and fully developed, so $v = w = 0$, $u = u(y)$ only, and the $x$-momentum equation in each layer
collapses to
$$\mu_i \frac{\mathrm{d}^{2}u_i}{\mathrm{d}y^{2}} = \frac{\mathrm{d}P}{\mathrm{d}x}
= \text{constant}$$
Four conditions are needed. Because the liquids wet the non-porous walls, no-slip applies at both:
$u_1(h) = 0$ at the top wall and $u_2(-h) = 0$ at the bottom. At the interface the two fluids
neither separate nor slip past one another, so the velocity is continuous,
$u_1(0) = u_2(0) = u_I$; and a massless interface can support no net force, so the shear stress is
continuous,
$$\mu_1\left.\frac{\mathrm{d}u_1}{\mathrm{d}y}\right|_{0}
= \mu_2\left.\frac{\mathrm{d}u_2}{\mathrm{d}y}\right|_{0}$$
Note that it is the stress, not the velocity gradient, that matches: the gradients
differ by the viscosity ratio of ten, which is what makes the profile kinked.
Part (b) — integrate twice in each layer. Writing
$G = -\mathrm{d}P/\mathrm{d}x = 1$ Pa/m for the favourable gradient, integration gives
$$u_i(y) = -\frac{G}{2\mu_i}y^{2} + A_i y + u_I, \qquad i = 1 \text{ (top)},\ 2 \text{ (bottom)}$$
where the common constant term already imposes velocity continuity at $y = 0$. Stress continuity
then requires $\mu_1 A_1 = \mu_2 A_2$, so $A_2 = (\mu_1/\mu_2)A_1$.
Part (b) — apply the wall conditions and solve. Substituting
$u_1(h) = 0$ and $u_2(-h) = 0$ and subtracting eliminates $u_I$, leaving one equation for $A_1$:
$$A_1 = \frac{G h}{2}\,\frac{\dfrac{1}{\mu_1}-\dfrac{1}{\mu_2}}{1+\dfrac{\mu_1}{\mu_2}}
= \frac{(1)(0.01)}{2}\,\frac{1000-100}{1.1} = 4.0909\ \text{s}^{-1}$$
whence $A_2 = 0.1 A_1 = 0.40909$ per second. Back-substitution into either wall
condition gives the interface speed, and the profile is complete:
$$\boxed{\;u_1(y) = -500\,y^{2} + 4.0909\,y + 0.009091
\quad (0 \le y \le h)\;}$$
$$\boxed{\;u_2(y) = -50\,y^{2} + 0.40909\,y + 0.009091
\quad (-h \le y \le 0)\;}$$
both in metres per second with $y$ in metres. Each parabola vanishes at its own wall, as required:
$u_1(0.01) = 0$ and $u_2(-0.01) = 0$.
Part (c) — shear stress at the interface. The Newtonian stress is
$\tau = \mu\,\mathrm{d}u/\mathrm{d}y$, and at $y = 0$ the gradient in the top layer is $A_1$:
$$\tau_I = \mu_1 A_1 = (0.001)(4.0909)
= \boxed{4.091\ \text{mPa}}$$
The bottom layer returns the identical value, $\mu_2 A_2 = (0.01)(0.40909)$,
which is the stress-continuity condition satisfying itself and a useful check on the algebra.
Part (c) — shear stress at the two walls. Differentiating each
profile at its wall,
$$\tau_{top} = \mu_1\left[-\frac{G}{\mu_1}h + A_1\right] = (0.001)(-10 + 4.091)
= -5.909\ \text{mPa}$$
$$\tau_{bot} = \mu_2\left[\frac{G}{\mu_2}h + A_2\right] = (0.01)(1 + 0.4091)
= 14.091\ \text{mPa}$$
The signs simply record that the two walls drag on the fluid from opposite sides; both retard the
flow. Their magnitudes are the physically useful numbers,
$5.91$ and $14.09$ mPa. A global
check closes the question: the pressure force per unit wall area on a slice of channel is
$G(2h) = 0.02$ Pa, and the two wall stresses sum to
$0.020$ Pa, exactly balancing it. The more viscous bottom
fluid carries the larger wall stress even though it moves more slowly.
Part (d) — speed at the interface. The constant term found in step
3 is precisely the interface speed:
$$u_I = \frac{G h^{2}}{2\mu_1} - A_1 h = 0.05 - 0.04091
= \boxed{9.091\ \text{mm/s}}$$
Computing it instead from the lower layer, $u_I = Gh^{2}/2\mu_2 + A_2 h
= 0.005 + 0.004091$, returns the same value, confirming the solution.
Part (e) — where the maximum sits. The maximum is where the shear
stress vanishes, so it lies in whichever layer contains the root of
$\mathrm{d}u/\mathrm{d}y = 0$. In the top layer,
$$\frac{\mathrm{d}u_1}{\mathrm{d}y} = -\frac{G}{\mu_1}y + A_1 = 0
\;\Longrightarrow\; y_{max} = \frac{\mu_1 A_1}{G}
= \boxed{4.091\ \text{mm above the interface}}$$
which does fall inside $0 \le y \le h$, whereas the bottom layer's root lies outside its own
domain, so $u_2$ increases monotonically from the lower wall to the interface. The peak value is
$$u_{max} = u_1(y_{max}) = 17.459\ \text{mm/s}$$
almost twice the interface speed. As the figure shows, the profile is a pair of parabolas joined
with a kink at the interface, strongly biased toward the thin, low-viscosity upper fluid; the
maximum sits above the centreline for exactly that reason. Integrating each layer gives volume
flows of 1.288e-04 and 5.379e-05 cubic metres per second per metre
of width, a total of 1.826e-04, so roughly seventy per cent of the throughput is
carried by the upper layer.