Section 4: Examples Using the Navier-Stokes Equation

In general, these equations are handy to have as they establish a starting point for going
about modeling fluid flow. When it comes to analytically deriving models (as in using
pen and paper), it is orders of magnitude more diffucult when you deal with fluid that
move in more than one direction. By that, I mean that if fluid flow has direction partially
in the x-direction and the y-direction, you get a partial differential equation (PDE),
as opposed to just in either x or y or z, which would result in an ordinary differential
equation (ODE), which is what we will be mainly working in the following example. Not
even all ODE systems can be derived analytically, so whenever you deal with such
systems you would usually then go to modeling softwares like COMSOL, ANSYS, or
FEniCS, or computational fluid dynamics (CFD) softwares. As mentioned earlier,
systems like pipes and reactors tend to be done in cylindrical coordinates (z, r, θ).
Systems like flow between two plates, or most microfluidic systems would more often
use Cartesian coordinates (x, y, z).

Here’s a fairly simple example: steady-state fluid flow through a pipe with radius, R, and
length, L (sorry about the poorly drawn axes).

ΔL
r
θ
R

L=0 L=L

STEP 1: Make assumptions

Assumptions are very common in engineering if you ever want to do stuff in a timely
fashion. While people might argue you want to be as accurate as possible, often times
you might just be looking for trends or look for maximums/minimums, for which
absolute accuracy is not always beneficial.
Here are some pretty common assumptions we will make:
- We have an incompressible, constant-viscosity, Newtonian fluid
- Fluid flow is laminar (as in it flows nice and neatly, as opposed to turbulent)
- Flow is driven by a pressure difference at either end of the pipe (typically true)
- No-slip condition at the fluid-solid interface (meaning the fluid has no velocity at
the borders)
STEP 2: Identify direction of fluid flow and flux
Next, we need to determine in what directions fluid will flow and flux will occur. Based
on how we defined the axes (in this case there’s really only one way that you could define
it to make any sense), we can see that fluid will flow in the z-direction. For the direction
of flux, given that we said that fluid moves in the z-direction, will the velocity change as
we move in the radial direction (r) or the angular direction (θ)? The picture below is a
cross section of the pipe. If you were to sample the velocity of the fluid at each dot, would
there be any difference in the blue dots? How about the green dots?

We can reason that velocity in the z-direction is only a function of radius, or vz(r). Since
it won’t depend on θ or z, we can similarly say that vz(θ) = vz(z) = 0. It should also be
mentioned here that since the fluid doesn’t flow in the radial or angular directions,all
variants of vr and vθ are also 0.

STEP 3: Choose the appropriate equation and simplify

We have established that for our pipe, fluid flow is only in the z-direction and is also
only a function of the pipe radius. We can now look to the Navier-Stokes equation for z-
momentum in cylindrical coordinates. We don’t even have to bother with r or θ because
those entire equations are just 0. Here’s what we start with (from Appendix B6):

𝜕𝑣𝑧 𝜕𝑣𝑧 𝑣𝜃 𝜕𝑣𝑧 𝜕𝑣𝑧 𝜕𝑃 1𝜕 𝜕𝑣𝑧 1 𝜕 2 𝑣𝑧 𝜕 2 𝑣𝑧

𝜌( + 𝑣𝑟 + + 𝑣𝑧 )= − + 𝜇[ (𝑟 )+ 2 + ] + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2

Ah big scary equation! But don’t worry, because it’s about to get much simpler. Since vz
only depends on r, we can pretty much just eliminate every term that isn’t vz or
describing the change in vz with respect to r. That means:

𝜕𝑣𝑧 𝜕𝑣𝑧 𝑣𝜃 𝜕𝑣𝑧 𝜕𝑣𝑧 𝜕𝑃 1𝜕 𝜕𝑣𝑧 1 𝜕 2 𝑣𝑧 𝜕 2 𝑣𝑧

𝜌( + 𝑣𝑟 + + 𝑣𝑧 )= − + 𝜇[ (𝑟 )+ 2 + ] + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2
Every term in red is just zero, which greatly simplifies our equation. Furthermore, we
are looking at the steady-state solution, so there is no change in velocity with respect to
time, nor is there a gravity component in the z-direction (we’ll assume it is negligable).
Repeating the same sort of exercise:

𝜕𝑣𝑧 𝜕𝑃 1𝜕 𝜕𝑣𝑧
𝜌( )= − + 𝜇[ (𝑟 )] + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟

All that we’re left with is:

𝜕𝑃 1𝜕 𝜕𝑣𝑧
0= − + 𝜇[ (𝑟 )]
𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟

STEP 4: Determine the velocity profile

Now comes the fun (and slightly more difficult) part because we now get to do calculus.
Our ultimate goal is to have a function that describes the velocity of fluid in the z-
direction as a function of r. This means, we will have to rearrange, integrate, and apply
some boundary coniditions to get our final result.

It’s important to also note that there are actually only two variables in this equation –
vz and r. The pressure differential that drives fluid flow is constant, and the viscosity of
the fluid was assumed to be constant. Let’s go!
First, we’ll reorganize and collect our constants on the right side of the equation:

1𝜕 𝜕𝑣𝑧 1 𝜕𝑃
(𝑟 )=
𝑟 𝜕𝑟 𝜕𝑟 𝜇 𝜕𝑧

From here forwards, you can kind of think of this problem as peeling layers of an onion.
You have a bunch of r-terms and derivatives that you’ll have to deal with one by one
until you get to vz. It can be a little tedious, but this one isn’t too bad.

We’ll first simultaneously deal with the 1/r and the derivative with respect to r (red).
1 𝜕𝑃
Since 𝜇 𝜕𝑧
is a constant with respect to r, if we multiply both sides and integrate, we
would essentially be integrating just ‘r’ with a constant.

1𝜕 𝜕𝑣𝑧 1 𝜕𝑃 𝜕 𝜕𝑣𝑧 1 𝜕𝑃
(𝑟 )= → ∫ (𝑟 ) 𝑑𝑟 = ∫ 𝑟 𝑑𝑟
𝑟 𝜕𝑟 𝜕𝑟 𝜇 𝜕𝑧 𝜕𝑟 𝜕𝑟 𝜇 𝜕𝑧
This gives us:
𝜕𝑣𝑧 1 𝜕𝑃 1 2
𝑟 = ∙ 𝑟 + 𝐶1
𝜕𝑟 𝜇 𝜕𝑧 2

‘C1’ is some constant that arises from doing an integral. It has a subscript because I
know (and you don’t know this yet) that there’s another constant coming up. We now
technically have two choices now; we can either deal with the constant, or keep trucking
along and deal with the next derivative. Without any particular reason at all, let’s deal
with the constant first.
Turns out, there’s a really crafty way to determine the constant using what I consider
arguably a boundary condition. To figure it out, figure divide both sides by ‘r’:

𝜕𝑣𝑧 1 𝜕𝑃 1 𝐶1
= ∙ 𝑟+
𝜕𝑟 𝜇 𝜕𝑧 2 𝑟

While this above equation is 100% correct, what happens if at r = 0 (as in the very middle
of the pipe)? The right side of the equation ends up as 0 + ∞ = ∞. Clearly, that’s not true
because you don’t have infinite momentum at the centre of a pipe (although that’d be
pretty cool), so there’s only one possible explanation for C1. C1 must equal 0. That’s pretty
dandy, and now we only have a couple steps left as we’re at the last derivative to deal
with.
Evaluating the integral again, we get the following equation:

𝜕𝑣𝑧 1 𝜕𝑃 1 𝜕𝑃 1 2
∫ 𝑑𝑟 = ∫ 𝑟 𝑑𝑟 → 𝑣𝑧 (𝑟) = ∙ 𝑟 + 𝐶2
𝜕𝑟 2𝜇 𝜕𝑧 2𝜇 𝜕𝑧 2

All we have left to deal with is the second constant (I told you there’d be another one).
This time, we actually have a boundary condition to apply, which comes from the no-
slip condition we assumed at the very beginning. Since the fluid has no velocity at radius
of ‘R’, the boundary condition we can apply is vz(R) = 0. Plug that into our equation, and
we can determine C2:

1 𝜕𝑃 2 1 𝜕𝑃 2
𝑣𝑧 (𝑅) = 0 = 𝑅 + 𝐶2 → 𝐶2 = − 𝑅
4𝜇 𝜕𝑧 4𝜇 𝜕𝑧

Lastly, we plug in our newfound C2, do some simplification, and then we’re done!
 1 𝜕𝑃 2 1 𝜕𝑃 2 1 𝜕𝑃 2
𝑣𝑧 (𝑟) = 𝑟 − 𝑅 → 𝑣𝑧 (𝑟) = (𝑟 − 𝑅2 )
4𝜇 𝜕𝑧 4𝜇 𝜕𝑧 4𝜇 𝜕𝑧

𝜕𝑃
One last quick note about 𝜕𝑧
. This pressure term is what determines which way the fluid
flows. Again, derivative terms represent ‘Out – In’, but pressure drives from high to low,
so if the ‘Out’ has a higher pressure than ‘In’, fluid actually flows in the negative z-
direction (as in it flows to the left). Taking that into acount, we can rewrite equation to
be a little more intuitive using P0 (pressure at the left end, at L = 0) and PL (pressure at
the right end, at L=L), over the total length of the pipe, L.

1 𝜕𝑃 2 1 (𝑃𝐿 − 𝑃0 ) 2 1 (𝑃0 − 𝑃𝐿 ) 2
𝑣𝑧 (𝑟) = (𝑟 − 𝑅2 ) = (𝑟 − 𝑅2 ) = − (𝑟 − 𝑅2 )
4𝜇 𝜕𝑧 4𝜇 𝐿 4𝜇 𝐿

Applying the negative sign into the difference between radii:

(𝑃0 − 𝑃𝐿 ) 2
𝑣𝑧 (𝑟) = (𝑅 − 𝑟 2 )
4𝜇𝐿
(𝑷𝟎 − 𝑷𝑳 ) 𝟐 𝒓 𝟐
𝒗𝒛 (𝒓) = 𝑹 (𝟏 − ( ) ) ∎
𝟒𝝁𝑳 𝑹

While it might be a little less clear, this is still a quadratic relationship. From the
𝑟 2
(1 − (𝑅) ) term, you can see that as ‘r’ approaches the value of ‘R’. this term scales down
the velocity profile, eventually equaling zero when r = R (as in the no-slip condition is
held true mathematically!). The following picture shows the development of a fluid profile
through a tube/pipe.
Additional Practice:
1.

2. You have an upright cylindrical (of radius R2) chamber with a stationary solid rod
(of radius R1) in the middle. The outer cylinder rotates with an angular velocity of ω.
What is the steady-state velocity profile, if there is no pressure gradient across the
ends of the cylinders? Top-down cross section below:

ω R2
r
θ
R1

3. You have a horizontal cylindrical (of radius R2) chamber with a stationary solid rod
(of radius R1) in the middle. There is a presure gradient driving fluid flow through this
tubular system.

z
R1 R2