What is the Rayleigh number and why does it matter?

The Rayleigh number Ra = gβΔTL³/(νκ) is the ratio of buoyancy-driven destabilising forces to the stabilising effects of viscosity and thermal diffusion. For Ra below the critical value of about 1708, the fluid remains in a stable conduction state. Above 1708, convection rolls form. As Ra increases further, the flow transitions through oscillatory convection to turbulence.

What numerical model does the simulator use?

The simulator uses a simplified 2D Boussinesq convection model. The temperature field is evolved with an advection–diffusion equation; the velocity field is updated from the buoyancy forcing (temperature anomaly). A semi-Lagrangian backtracing scheme advects temperature on an 80×40 grid using Float32Array for performance.

Why do convection cells form regular roll patterns?

When instability sets in, all spatial perturbation modes compete to grow. The fastest-growing mode has a preferred wavelength roughly 2–3 times the layer thickness, and this mode dominates to form regular rolls. This is a classic example of spontaneous pattern formation or self-organisation. As Ra increases, these ordered rolls break down into more complex, eventually turbulent motion.

› Convection Cells Simulator Back

Thermal Fluid Simulator

Convection Cells Simulator — Rayleigh–Bénard Convection

Q: What is Rayleigh–Bénard convection?

Rayleigh–Bénard convection is the buoyancy-driven flow that occurs when a fluid layer is heated from below and cooled from above. Hot fluid near the bottom becomes less dense, rises by buoyancy, and is replaced by cooler descending fluid from above, forming organized roll-shaped convection cells. The phenomenon was observed experimentally by Henri Bénard in 1900 and theoretically analysed by Lord Rayleigh in 1916.

Vary the Rayleigh number (Ra) and watch convection cell formation, development, and turbulent transition in real time

Parameters

Rayleigh Number Ra (log scale)

Ra = 2512

Aspect Ratio (width / height)

2.0

Presets

Temperature field

Velocity vectors

Streamline overlay

Max Temperature

—

Min Temperature

—

Max Flow Speed

—

Convection State

Computing...

Hot

Cold

What Is Rayleigh–Bénard Convection?

🧑‍🎓

When I heat a pot of water, I can sometimes see fluid moving in circular patterns near the bottom. Are those the convection cells you're talking about?

🎓

Exactly! Those are Rayleigh–Bénard convection cells. Heating from below makes the bottom fluid less dense, so it rises by buoyancy. The cooler, denser fluid at the top sinks to replace it. This creates periodic roll-shaped circulation patterns. Bénard first observed this beautifully in whale oil in 1900, and Lord Rayleigh provided the theoretical framework in 1916. It's one of the most studied phenomena in fluid dynamics.

🧑‍🎓

Sometimes when I warm water only slightly nothing visible happens. What determines whether convection starts or not?

🎓

That's exactly what the Rayleigh number Ra tells us. Ra compares buoyancy-driven destabilisation against the stabilising effects of viscosity and thermal diffusion. Below the critical value of Ra ≈ 1708, the fluid stays in pure conduction mode — any perturbation gets damped before it can grow. Above 1708, convection kicks in. Try the "Stable" preset in the simulator and watch how the cells disappear, leaving only a smooth temperature gradient.

🧑‍🎓

Why do the cells organise into such regular roll patterns? I'd expect more random mixing.

🎓

Great intuition. Initially, random perturbations of all wavelengths compete to grow. But there's an optimal wavelength — about 2–3 times the layer height — that grows fastest and wins. The system self-organises into this preferred length scale. It's a beautiful example of spontaneous symmetry breaking and pattern formation. Increase Ra further and the ordered rolls eventually become chaotic, then turbulent. You can see the whole transition in this simulator by dragging the Ra slider.

Physics and Governing Equations

Rayleigh number (dimensionless parameter):

$$Ra = \frac{g\,\beta\,\Delta T\,L^3}{\nu\,\kappa}$$

$g$: gravity, $\beta$: thermal expansion coefficient, $\Delta T$: temperature difference, $L$: layer height, $\nu$: kinematic viscosity, $\kappa$: thermal diffusivity

Temperature advection–diffusion equation (Boussinesq approximation):

$$\frac{\partial T}{\partial t}+ \vec{u}\cdot\nabla T = \kappa\,\nabla^2 T$$

Velocity updated from buoyancy forcing (temperature anomaly drives vertical flow).

Boundary conditions: bottom $T=1$ (hot), top $T=0$ (cold), lateral: periodic.

Real-World Connections

Meteorology: Cumulus cloud formation, thunderstorm updrafts, and mesoscale convective systems are large-scale Rayleigh–Bénard convection driven by solar heating of Earth's surface.

Earth's interior: Mantle convection — the driver of plate tectonics — operates on a Rayleigh–Bénard mechanism across geological timescales, with Ra ~ 10⁷.

Electronics cooling: Natural convection in server racks and heat sink design depends directly on understanding when and how convection cells form.

The Sun: The granules visible on the solar surface are convection cells ~1000 km across in the Sun's outer convective zone, driven by the steep temperature gradient.

Common Misunderstandings and Points to Note

When you start using this simulator, there are a few common pitfalls to watch out for. First, it's easy to assume that "a higher Rayleigh number always means more vigorous convection," but it's not that simple. While convection does begin once the critical value (approximately 1708) is exceeded, increasing Ra leads through stages: from neat roll patterns, through a "secondary instability" phase where rolls distort or split, and finally to turbulence. For instance, if you increase Ra from 10^4 to 10^5, you can observe the rolls starting to fluctuate over time instead of remaining stationary. This is the gateway to "transient chaos," so the trick is to change parameters gradually; changing them too abruptly might cause you to miss the transitions between fundamental patterns.

Next, don't underestimate the importance of the aspect ratio (width/height) setting. For example, setting the aspect ratio to 2 (width is twice the height) naturally results in two convection rolls side by side. However, if you set it to an intermediate value like 3.5, the number of rolls is no longer an integer, which can easily lead to unstable patterns. In nature or real devices, the "width" of a layer is often fixed, so remember that changing the aspect ratio in a simulation is essentially equivalent to "changing the size of the experimental apparatus or observation area."

Finally, don't forget that this simulation assumes a "2D" model and uses the "Boussinesq approximation." In three dimensions, hexagonal Bénard cells can appear instead of rolls. Also, this approximation breaks down when temperature differences are extreme or when fluid compressibility cannot be ignored. Understand that this is an idealized model for learning fundamental principles.

Related Engineering Fields

The principles of Rayleigh-Bénard convection underlie a wider range of engineering fields than you might think. For example, cooling design for electronic devices. Beneath a smartphone's circuit board or a CPU heat sink, you essentially have a "bottom-heated" situation. To optimize heat dissipation via natural convection, you need to understand how these convection cells form. If you experiment with the aspect ratio in the simulator, you'll see the number and size of the convection rolls change, right? This concept is directly applicable to optimizing heat sink fin spacing (if too narrow, flow is obstructed; if too wide, effective area is reduced).

Another crucial application is process control for "crystal growth" or "thin-film formation" in chemical and materials engineering. When growing crystals from a stationary solution or molten metal, uncontrolled internal convection (driven by temperature or concentration differences) prevents uniform crystal formation. "Suppressing this convection" is key to quality control, and the Rayleigh number is precisely the indicator used. For instance, experiments to grow high-quality crystals on the space station (where microgravity effectively lowers Ra) are a practical application of this principle.

Furthermore, it's applied in architectural environmental engineering. The air gap in double-glazed windows or within building cavity walls is an environment prone to Bénard convection. In winter, when the indoor-side glass pane is heated by indoor warmth, the bottom of the air layer is heated, potentially creating a convection loop that increases heat loss. If you extend this simulator to also allow changing the "top temperature," the underlying concepts could serve as a preliminary design tool for insulation.

For Further Learning

Once you're comfortable with this simulator, as a next step, make sure you firmly grasp the concept of "dimensionless numbers." The Rayleigh number (Ra) is a dimensionless number representing the competition between buoyancy, viscosity, and thermal diffusion. Similarly, the Prandtl number (Pr = ν/κ, the ratio of kinematic viscosity to thermal diffusivity) is another dimensionless number that determines flow characteristics. For example, with water (Pr≈7) and air (Pr≈0.7), even at the same Ra, the speed of pattern transitions and the shape of the rolls differ. For your next learning step, consider "what happens if you change the Prandtl number?" to understand the importance of fluid properties.

If you want to delve a bit deeper mathematically, challenging yourself with linear stability analysis

Finally, as a future topic, I recommend exploring "double-diffusive convection." This is a phenomenon where not only temperature differences but also differences in salinity or species concentration act as driving forces. In oceans, even in "stable layers" where warm, salty water overlies cold, fresh water, convection can occur due to differences in diffusion coefficients (the "salt finger" phenomenon). Your understanding of Rayleigh-Bénard convection provides the perfect foundation for learning about these more complex and realistic mixing and convection processes.