Laminar and turbulent flow

Water moving slowly through a pipe and water moving quickly through the same pipe are not the same phenomenon slightly rescaled. They are structurally different flows, organised in entirely different ways, and every quantity a pipe-flow problem eventually asks for — friction, heat transfer, how fast a contaminant spreads — depends on which one you have before it depends on anything else. The distinction was first made visible, rather than merely inferred, with a glass pipe and a thread of dye.

Reynolds' dye experiment

In 1883 Osborne Reynolds fed water through a glass pipe from a large tank, opened a valve to control the speed, and injected a fine thread of dye along the centreline. At low speed the dye ran the length of the pipe as a single straight filament, barely spreading at all — evidence that the water itself was moving in orderly layers, with nothing crossing between them.

Opening the valve further changed that filament's behaviour before it changed its speed by very much. The thread began to waver, taking on a gentle sinuous shape while still remaining a recognisable line. Push the speed higher again and the wavering gave way abruptly: the dye broke apart into a tangle of eddies and filled the entire cross-section within a short distance of the injection point, the orderly thread gone completely.

Reynolds identified the ratio of quantities that controlled this change — pipe diameter, mean velocity, and the fluid's density and viscosity — and found the transition clustered around a particular value of that ratio, in his apparatus somewhere near \(2300\). That number has survived as the standard textbook figure for pipe flow, and the next section is about why it deserves rather less reverence than it usually gets.

The Reynolds number

Reynolds number is not a dial that reads "laminar" below one mark and "turbulent" above another. It is a ratio of two force scales acting on the same parcel of fluid, and understanding it as a ratio is what makes the number usable outside a pipe, and honest about where its threshold actually comes from.

Inertial force scales as \(\rho V^2 L^2\) — mass times acceleration, expressed through the density, velocity and size that set the problem's scale. Viscous force scales as \(\mu V L\), the shear stress \(\mu\,du/dy \sim \mu V/L\) acting over an area \(L^2\). Divide one by the other and the length scales partly cancel, leaving

which is exactly the dimensional-analysis argument behind dimensional analysis and similarity, applied to this one specific pair of forces. A large Re means a fluid parcel's own momentum carries it onward faster than viscosity can iron out any disturbance to its path; a small Re means viscosity wins that argument before the disturbance can grow. Turbulence, on this reading, is simply what an unstable disturbance turns into once inertia has enough of an advantage to let it grow rather than decay.

The specific value 2300 belongs to a circular pipe, defined on mean velocity and diameter, under realistic laboratory disturbance levels — background vibration, a less-than-perfectly smooth entrance, ordinary imperfections. Reynolds himself later showed that with an unusually undisturbed inlet and a vibration-free rig, laminar flow in a pipe can be coaxed to survive past Re of 100 000. Roughen the wall, or shake the apparatus, and transition arrives earlier instead. A flat plate has its own transition Reynolds number, an open channel another again — the number 2300 is a property of a particular geometry and a particular level of disturbance, not a constant of nature, and treating it as one is the most common misreading of this topic.

Laminar structure

Below the transition, fluid moves in smooth concentric layers — laminae, the word the regime is named for — each sliding past its neighbours without any macroscopic exchange of fluid between them. Whatever momentum crosses from a faster layer to a slower one does so purely by molecular diffusion, the same mechanism that lets heat conduct through a solid, and it is a slow one.

Solved exactly for fully developed flow in a circular pipe, that picture gives a closed-form velocity profile — the Hagen–Poiseuille parabola:

Zero at the wall, by no-slip, and maximum on the centreline. The average velocity across the section is exactly half the centreline value,

with no empirical correction and no chart — a rare case in this subject where the answer really is derivable from first principles rather than measured and fitted. It is worth noticing how much is riding on that word exactly. Nothing about turbulent flow will offer the same guarantee.

Turbulent structure

Above transition, three-dimensional eddies superimpose themselves on the average motion, and those eddies — not molecular diffusion — become the dominant way momentum moves sideways across the flow. The usual way to make that tractable is to split the instantaneous velocity into a time-averaged mean and a fluctuation, \(u = \bar u + u'\), and work with the mean; the fluctuations still matter, but a full account of their motion is not needed to describe the shape of the flow.

Because eddies mix momentum far more effectively than molecular viscosity does, the turbulent core is far more uniform than the laminar parabola — commonly approximated by a one-seventh power law,

measuring \(y\) from the wall. Nearly the entire velocity change is then squeezed into a thin region right at the wall — the viscous sublayer, often a fraction of a millimetre thick — where the no-slip condition still has to be satisfied and eddies cannot survive close enough to the surface to do the mixing they manage everywhere else. That sliver of fluid carries a wildly disproportionate share of the total resistance to the flow for how little of the pipe it actually occupies, and boundary layers takes that structure apart in more detail.

With the core so much flatter, the average-to-maximum ratio rises well above the laminar value:

against 0.5 for laminar flow — a fuller profile, carrying more of the pipe's cross-section close to the bulk speed, at the direct cost of concentrating all the shear where the profile does drop.

Why the distinction cascades

Every one of this course's later formulas asks, implicitly or explicitly, which regime it is being applied to, because the structural difference just described has consequences well beyond the shape of a velocity profile.

Friction is the most immediate. The laminar friction factor \(f = 64/Re\) falls steadily as Re rises, right up until transition — where the turbulent value jumps well above where the laminar line would have continued, because a thin, highly sheared sublayer generates far more wall stress than a smooth parabola ever needs to. Flow in pipes and head losses is built entirely around that friction factor and what decides it.

Convective heat transfer follows the same pattern for the same reason. Fully developed laminar flow in a pipe carries a fixed, unremarkable Nusselt number — 3.66 at constant wall temperature — while a typical turbulent correlation such as \(Nu = 0.023\,Re^{0.8}Pr^n\) routinely returns values many times larger at the Reynolds numbers most engineering flows actually run at. The eddies that mix momentum mix thermal energy just as readily.

And mixing itself, the thing Reynolds' dye streak measured directly, scales the same way: a dye or a contaminant released into laminar flow spreads only by the same slow molecular diffusion that moves momentum between laminae, taking metres to blend appreciably, while a turbulent stream folds a contaminant across its whole cross-section within a handful of pipe diameters. One number sets the regime, and the regime sets nearly everything that follows from it.