Skip to content
Tutorials Center
My learning

Topics

Lesson 6 of 46 ยท Continuum and Kinematics

The Continuum Hypothesis and the Knudsen Number

Why molecules can be replaced by smooth fields: the continuum hypothesis, the mean free path of air, the Knudsen number and the flows where the idea fails.

A cup of water holds about 7ร—10ยฒโด molecules, each jostling at hundreds of metres per second. Nobody solves for each of them. Instead, fluid dynamics assumes that you can speak of "the density here" and "the velocity there" as smooth functions of position. This assumption is the continuum hypothesis, and almost everything in this course depends on it. When is it justified, and when does it fail? This lesson answers with numbers.

Molecules with a zigzag free path on the left, a smooth continuum field on the right, and a Knudsen number length scale below
Left: the molecular picture. Right: the continuum picture. Below: the Knudsen number says which one is appropriate.

What you will learn

  • What a fluid particle (a continuum "point") is and how big it must be.
  • How to estimate the mean free path of a gas.
  • How to compute the Knudsen number and read the four flow regimes.
  • Examples of flows where continuum equations break down.

Key terms

Continuum hypothesis: treat the fluid as a continuous medium whose properties (density, velocity, pressure) are smooth functions of position.

Fluid particle: a small volume that contains very many molecules yet is much smaller than the flow features; the continuum "point".

Mean free path ฮป: the average distance a molecule travels between collisions.

Knudsen number Kn = ฮป / L: mean free path divided by the flow's length scale L.

What is a "point" in a fluid?

Suppose you want the density at a point. You take a small box around it, count the molecules, multiply by their mass and divide by the box volume. If the box is too small, it contains a handful of molecules, and the count jumps wildly as molecules enter and leave. If the box is too big, it averages over real variations of the flow. There is a window of sizes in between where the answer is steady. A box in that window is a fluid particle.

For air at 20 ยฐC and one atmosphere, there are about 2.5ร—10ยฒโต molecules in every cubic metre. A cube of side 1 ยตm (a thousandth of a millimetre) has volume 10โปยนโธ mยณ, so it holds about 25 million molecules. Random fluctuations in the count are of order 1/โˆšN, about 0.02 percent. That is smooth enough. A cube of side 10 nm holds about 25 molecules, and fluctuations are of order 20 percent: no longer a good "point".

The mean free path

In a gas, a molecule flies in a straight line until it collides with another, then flies off in a new direction. Kinetic theory gives the average distance between collisions:

ฮป = k T / (โˆš2 ฯ€ dยฒ p)

with the Boltzmann constant k = 1.380649ร—10โปยฒยณ J/K, the absolute temperature T, an effective molecular diameter d (about 3.7ร—10โปยนโฐ m for air), and the pressure p. At T = 293 K and p = 101,325 Pa this gives ฮป โ‰ˆ 66 nm, about 7ร—10โปโธ m. A molecule of air travels about 200 molecular diameters before it hits another. Lowering the pressure lengthens ฮป in proportion: at one thousandth of an atmosphere, ฮป is about 66 ยตm.

In a liquid the molecules are packed shoulder to shoulder, so the mean free path is about one molecular size, roughly 0.3 nm. For water, the continuum model works down to scales of a few nanometres.

The Knudsen number

Compare the mean free path with the size L of the flow (a pipe diameter, a wing chord, a gap):

Kn = ฮป / L.

If Kn is tiny, a molecule collides many times while crossing the flow, so the gas behaves as a continuum in local equilibrium. The usual classification (gases):

Kn Regime Model
below 0.01 continuum Navier-Stokes with no-slip walls
0.01 to 0.1 slip Navier-Stokes with slip velocity at walls
0.1 to 10 transition Boltzmann equation, DSMC simulation
above 10 free molecular molecules rarely collide; ballistic motion

The numbers are guidelines, not sharp edges.

Worked examples

A 10 mm air duct. Kn = 6.6ร—10โปโธ / 0.01 โ‰ˆ 7ร—10โปโถ. Deep in the continuum regime.

A 1 ยตm channel in a microfluidic chip with air. Kn โ‰ˆ 0.07: slip regime. Air slips along the wall slightly, and standard no-slip pressure-drop formulas under-predict the flow.

A 50 nm pore. Kn โ‰ˆ 1.3: transition regime. Gas flow in shale rock and in nano-filters lives here.

A spacecraft at the top of the atmosphere. At altitudes around 100 km the mean free path of air is of the order of 0.1 m, comparable to the size of a capsule. Re-entry vehicles pass through all four regimes, from free molecular to continuum, on the way down. Simulations in this transition region use particle methods rather than Navier-Stokes.

For liquids the same idea applies at a far smaller scale. Water in a carbon nanotube of 1 nm diameter is not described by the usual equations; flow in biological channels in cell membranes needs molecular simulation as well.

Why no-slip works in a continuum

At a solid wall, the continuum model assumes that the fluid sticks: its velocity equals the wall's velocity. Molecularly, a molecule that hits the wall is usually trapped for a moment and re-emitted in a random direction, so on average it loses its tangential momentum. After a few collisions per mean free path, the layer next to the wall has zero average velocity. When Kn becomes large enough that molecules reach the wall without colliding with others, they can keep some of their motion, and the wall appears to slip.

Common misconceptions

"The continuum model assumes the fluid has no molecules." It assumes molecules are so numerous and collide so often that their average behaviour is smooth. The molecules are still there.

"Kn depends only on the fluid." Kn depends on the flow too. Air is a continuum around a skyscraper (Kn ~ 10โปโน) and not in a 20 nm pore (Kn ~ 3).

"Liquids are always continua." At nanometre scales, liquids show layering near walls and are not smooth. The scale is just much smaller than for gases.

Try it

First, compute the mean free path from the kinetic theory formula and plot the Knudsen number against length for three pressures. The shaded bands are the four regimes.

Python Knudsen number against length

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Second, count how the density fluctuation of a sampling box shrinks as the box grows. For a box with an expected N molecules, the typical relative fluctuation is about 1/โˆšN.

JS How big must a fluid particle be?

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. What does the continuum hypothesis assume?

    The molecules exist, but we average over a fluid particle containing very many of them, giving smooth fields.
  2. 2. The mean free path of air at 1 atm and 20 degrees C is about

    About 66 to 70 nm, roughly 200 molecular diameters.
  3. 3. A gas flow has Kn = 0.5. Which regime is this?

    0.1 < Kn < 10 is the transition regime, where neither Navier-Stokes nor free-molecule models are accurate.
  4. 4. If the pressure of a gas at fixed temperature is cut by a factor of 100, the mean free path

    lambda is inversely proportional to pressure at fixed temperature.
  5. 5. Which flow is least likely to satisfy the continuum hypothesis?

    Kn = 66/20 = 3.3 for the pore, well into the transition regime.

Summary

  • The continuum hypothesis replaces molecules by smooth fields defined on fluid particles that contain very many molecules.
  • For air at 1 atm, ฮป โ‰ˆ 66 nm; the number density is about 2.5ร—10ยฒโต mโปยณ.
  • Kn = ฮป/L sorts flows into continuum (below 0.01), slip, transition and free molecular regimes.
  • Navier-Stokes with no-slip walls is the right model for Kn < 0.01, which covers almost everything in this course.

Next: Eulerian and Lagrangian Views