Traditional CFD treats air as a continuous fluid. This works beautifully for most engineering applications, from aircraft to air conditioning systems. But what happens when the air becomes so thin that this assumption breaks down?
At high altitudes, in vacuum systems, and at very small scales, gas molecules become sparse enough that we can no longer ignore their discrete nature. This is the domain of rarefied gas dynamics.
The Knudsen Number
The key parameter that determines whether continuum assumptions apply is the Knudsen number:
Kn = λ / L
Where λ is the mean free path (average distance a molecule travels between collisions) and L is a characteristic length scale of your problem.
The flow regimes are typically classified as:
| Knudsen Number | Regime | Approach | |----------------|--------|----------| | Kn < 0.01 | Continuum | Traditional CFD (Navier-Stokes) | | 0.01 < Kn < 0.1 | Slip flow | Modified CFD with slip boundary conditions | | 0.1 < Kn < 10 | Transition | DSMC or hybrid methods | | Kn > 10 | Free molecular | Analytical or DSMC |
Where Rarefied Flow Matters
Rarefied gas dynamics appears in several important applications:
High-altitude flight: At 100 km altitude, the mean free path is about 10 cm. A vehicle at this altitude experiences significant rarefied effects.
Spacecraft reentry: The transition from free molecular flow to continuum during reentry affects heating, drag, and stability.
Vacuum systems: Semiconductor manufacturing, freeze drying, and scientific instruments often operate in rarefied conditions.
MEMS devices: At micron scales, even atmospheric pressure air can exhibit rarefied behavior.
Planetary atmospheres: Mars, with its thin atmosphere, presents rarefied flow challenges for landers and aircraft.
Why Navier-Stokes Fails
The Navier-Stokes equations assume that the gas is in local thermodynamic equilibrium, and that transport properties (viscosity, thermal conductivity) can be defined meaningfully. These assumptions require that molecules collide frequently enough to establish equilibrium distributions.
When Kn becomes significant:
- The velocity distribution deviates from Maxwellian
- Temperature may not be isotropic (different in different directions)
- The concept of viscosity becomes poorly defined
- Heat transfer mechanisms change fundamentally
Applying continuum equations to rarefied flows doesn't just give inaccurate results. It gives qualitatively wrong physics.
The DSMC Method
Direct Simulation Monte Carlo (DSMC), pioneered by Graeme Bird in the 1960s, offers a solution. Instead of solving continuum equations, DSMC directly simulates the motion and collisions of representative gas molecules.
The basic DSMC algorithm:
- Move particles: Advance all particles ballistically for a time step
- Apply boundaries: Handle collisions with surfaces (reflection, accommodation)
- Collide particles: Pair particles within cells and perform probabilistic collisions
- Sample properties: Collect statistics to compute macroscopic quantities
Each simulated particle represents many real molecules. The method is statistically accurate when enough samples are collected.
DSMC Challenges
DSMC is powerful but computationally expensive:
Cell size constraints: Cells should be smaller than the local mean free path
Time step constraints: Time steps should be smaller than the mean collision time
Statistical noise: Results are inherently noisy; long sampling times are needed for accuracy
Computational cost: Scales with the number of molecules, which increases as rarefaction decreases
For transition regime flows (0.1 < Kn < 10), DSMC becomes very expensive. This has motivated development of hybrid methods that couple DSMC with continuum solvers.
Practical Considerations
If you're facing a rarefied flow problem, here's a practical approach:
-
Calculate the Knudsen number: Estimate λ from pressure and temperature, use an appropriate length scale
-
Assess the regime: If Kn < 0.01 everywhere, use standard CFD. If Kn > 0.1 anywhere significant, consider DSMC.
-
Consider hybrid approaches: For complex geometries with varying Kn, hybrid continuum-DSMC methods may be most efficient.
-
Validate carefully: Rarefied flow data is scarce. Build confidence through code verification and comparison with available experiments.
Gas-Surface Interactions
A critical aspect of rarefied flow is how molecules interact with surfaces. Unlike continuum flow where we specify velocity and temperature boundary conditions, in rarefied flow we must model the molecular reflection process.
Common models include:
- Specular reflection: Mirror-like reflection (frictionless surface)
- Diffuse reflection: Random re-emission at surface temperature (rough surface)
- Accommodation coefficients: Partial energy and momentum exchange
The choice of surface model significantly affects predicted drag and heat transfer, especially in free molecular flow where all information comes from surface interactions.
Looking Forward
Rarefied gas dynamics remains an active research area. Current challenges include:
- Multi-scale methods bridging continuum and molecular scales
- Efficient algorithms for near-continuum flows
- Accurate models for polyatomic gases and chemical reactions
- Integration with thermal and structural analysis
As aerospace vehicles push to higher altitudes and smaller satellites become common, rarefied flow analysis will only become more important.
Need help with rarefied flow analysis? Our team has experience with DSMC simulation for aerospace and vacuum applications. Contact us to discuss your project.