Reynolds Number

The Reynolds number (Re) is a dimensionless quantity in continuum fluid mechanics that quantifies the relative magnitude of fluid inertia to internal viscous shear stress within a moving fluid. Defined as Re = ρUL/μ = UL/ν—where ρ is fluid density, U is a characteristic flow velocity, L is a characteristic geometric length scale, μ is dynamic shear viscosity, and ν = μ/ρ is kinematic viscosity—the parameter serves as the primary scaling criterion for predicting whether a flow remains smooth and laminar or transitions into chaotic, three-dimensional turbulence. Although the mathematical ratio was first introduced by George Gabriel Stokes in 1851 while analyzing pendulum damping in viscous media, the dimensionless group was popularized by Osborne Reynolds in his 1883 Manchester pipe-flow experiments and formally named in his honor by Arnold Sommerfeld in 1908.

Across engineering hydrodynamics, biological transport, and geophysical fluid dynamics, the Reynolds number emerges naturally from the non-dimensionalization of the Navier-Stokes equations. Two geometrically similar flows operating at identical Reynolds numbers exhibit dynamical similarity: their non-dimensional velocity and pressure distributions follow identical governing equations regardless of absolute physical size or fluid composition. This scaling invariance allows aerodynamicists to test scale aircraft models in pressurized wind tunnels, bioengineers to design microfluidic lab-on-a-chip channels governed by Stokes flow, and climate physicists to parameterize sub-grid planetary boundary layer theory processes across fifteen orders of magnitude in spatial scale.

Derivation from Momentum Conservation

To compare fluid flows across disparate physical scales, the dimensional variables in the incompressible Navier-Stokes equations are normalized by reference scales intrinsic to the geometry and boundary conditions of the problem. Selecting a reference length scale L (such as pipe diameter D, airfoil chord c, or boundary-layer depth δ), a reference velocity U (such as mean cross-sectional velocity or free-stream speed U_∞), an advective time scale t₀ = L/U, and a dynamic pressure scale p₀ = ρU² defines the dimensionless variables:

Substituting these scaled variables into the dimensional incompressible momentum equation—ρ(∂u/∂t + (u · ∇)u) = −∇p + μ∇²u—and dividing every term by the characteristic inertial force per unit volume (ρU²/L) yields the dimensionless Navier-Stokes equation:

where ∇* = L∇ is the dimensionless gradient operator and the sole remaining control parameter is the Reynolds number:

In this form, Re expresses the ratio of the non-linear advective inertia term, |(u · ∇)u| ∼ U²/L, to the linear viscous diffusion term, |ν∇²u| ∼ νU/L². Equivalently, Re represents the ratio of the viscous diffusion time scale (τ_visc = L²/ν, the time required for molecular momentum diffusion to smooth velocity gradients across distance L) to the advective transit time scale (τ_adv = L/U).

Asymptotic Limits: Creeping Flow vs. Inviscid Regimes

The mathematical structure of the dimensionless momentum equation changes character between the asymptotic limits of low and high Reynolds numbers:

• Viscous-Dominated Limit (Re ≪ 1): When Re approaches zero, the coefficient (1/Re) multiplying the Laplacian operator becomes large. Rescaling pressure by the viscous stress scale p_visc = μU/L eliminates the non-linear convective term (u · ∇)u, reducing the dynamics to the linear, time-reversible Stokes flow equations (∇p = μ∇²u, ∇ · u = 0). Because inertia is negligible, fluid parcels adjust instantaneously to boundary motion, and reversing the direction of boundary forcing retraces the exact fluid trajectory without net displacement (Edward Purcell's 1977 "scallop theorem").
• Inertia-Dominated Limit (Re ≫ 1): As Re becomes large, the viscous term (1/Re)∇*²u* becomes small throughout the bulk interior of the flow. Dropping the viscous operator altogether yields the first-order Euler fluid equations, where pressure gradients and convective acceleration balance along streamlines according to Bernoulli's principle. However, because (1/Re) multiplies the highest-order spatial derivative (∇*²u*), the limit Re → ∞ is a singular perturbation: near solid boundaries where the no-slip condition (u = 0) applies, steep wall-normal velocity gradients restore viscous shear within a thin sheet of thickness δ ∼ L / Re^1/2, as described by boundary layer theory.

The Manchester Dye-Filament Apparatus

In 1883, Osborne Reynolds conducted a series of hydrodynamic stability experiments at Owens College in Manchester to determine the conditions under which water flowing through cylindrical glass tubes ceases to move in parallel laminae and breaks into eddying motion. Reynolds submerged horizontal glass pipes of three different diameters inside a large glass-walled tank filled with quiescent water, fitting each tube with a polished bell-mouth trumpet intake to minimize entry disturbances and a downstream valve to regulate mean discharge velocity U. A fine streak of carmine-dyed water was injected at the centerline of the intake bell.

At low flow speeds, the dyed filament remained a sharp, coherent thread extending down the entire length of the tube, demonstrating steady laminar Hagen–Poiseuille flow with a parabolic axial velocity profile u(r) = 2U(1 − r²/R²). As the control valve opened and U exceeded a critical threshold, the dye filament abruptly developed oscillatory sinuous instabilities several diameters downstream of the inlet before bursting into three-dimensional eddies that mixed the dye across the entire pipe cross-section. By varying both tube diameter D and water temperature (which alters kinematic viscosity ν from 1.79 × 10⁻⁶ m² s⁻¹ at 0 °C to 0.66 × 10⁻⁶ m² s⁻¹ at 40 °C), Reynolds demonstrated that transition always occurred at an invariant value of the dimensionless group UD/ν.

Subcritical Instability, Turbulent Puffs, and Friction Factor Scaling

In standard engineering pipe networks with finite entrance perturbations and wall roughness, the laminar-to-turbulent transition begins around a critical Reynolds number of:

Between Re ≈ 2,000 and Re ≈ 4,000, the flow resides in a transitional intermittency regime characterized by localized patches of turbulence—termed Wygnanski "puffs" and "slugs"—interspersed with laminar segments. Remarkably, linear stability analysis of cylindrical Hagen–Poiseuille flow proves that the parabolic velocity profile is linearly stable to infinitesimal axisymmetric disturbances at all Reynolds numbers. Consequently, pipe transition occurs via subcritical finite-amplitude bypass mechanisms driven by transient non-normal mode growth; when inlet disturbances and wall vibrations are carefully suppressed, laboratory pipe flows have been maintained in a metastable laminar state up to Re ≈ 100,000. Once fully developed turbulence is established (Re > 4,000), the Darcy–Weisbach friction factor f_D = Δp / [(L/D)(½ρU²)] departs from the exact laminar law f_D = 64/Re and follows the implicit Colebrook–White relation (plotted on the Moody chart), where pressure drop becomes governed by turbulent Reynolds stresses and relative wall roughness ε_r/D rather than molecular viscosity alone.

Geometric and Dynamical Scaling in Experimental Fluid Mechanics

Dynamical similarity requires that all relevant dimensionless ratios match between a prototype system (subscript p) and a scale model (subscript m). For incompressible single-phase flow without free-surface gravity waves or compressibility effects, matching the Reynolds number alone ensures dynamical equivalence:

If a 1:10 scale aircraft model (L_m/L_p = 0.1) is tested in an atmospheric wind tunnel using air at ambient pressure and temperature (ν_m = ν_p), matching Re would require U_m = 10U_p, pushing the test flow into supersonic compressible regimes where the Mach number (Ma) no longer matches. Experimentalists overcome this constraint by testing in cryogenic pressurized wind tunnels—such as the European Transonic Windtunnel or NASA's National Transonic Facility—where cooling nitrogen gas toward its boiling point (90 K) and pressurizing it to 4–9 bar reduces kinematic viscosity ν by more than an order of magnitude, preserving full-scale chord Reynolds numbers (Re_c ∼ 10⁷–10⁸) at subsonic velocities.

Quantitative Comparison Across Fluid Regimes

Across natural and engineered systems, the Reynolds number spans more than seventeen orders of magnitude, partitioning fluid behavior into distinct dynamical regimes:

Drag Crisis and Geophysical Eddy Viscosity

For bluff bodies such as smooth spheres or circular cylinders immersed in an external stream, the drag coefficient C_D = F_D / (½ρU²A) varies systematically with Re. Below Re ≈ 1, Stokes' law yields C_D = 24/Re, so drag force scales linearly with velocity (F_D = 3πμUD). Between Re ≈ 10³ and 3 × 10⁵ (the Newton subcritical regime), flow separates laminarly on the upstream hemisphere of the sphere, creating a wide low-pressure wake and a nearly constant drag coefficient C_D ≈ 0.47. At Re ≈ 3.5 × 10⁵, the attached boundary layer transitions to turbulence prior to separation; enhanced turbulent momentum mixing energizes the near-wall fluid against the adverse pressure gradient, delaying separation to the rear hemisphere, shrinking the wake, and causing C_D to drop abruptly to ~0.10—a phenomenon discovered by Gustave Eiffel (1912) and explained by Ludwig Prandtl (1914) as the drag crisis.

In geophysical fluid dynamics, molecular Reynolds numbers reach Re ∼ 10¹⁰–10¹², ensuring that atmospheric and oceanic flows sustain a continuous Kolmogorov turbulence cascade from synoptic forcing scales (L ∼ 1,000 km) down to millimeter Kolmogorov dissipation microscales (η = (ν³/ε)^1/4). Because numerical climate models cannot resolve the N ∼ Re^9/4 ≈ 10²⁵ grid points required for direct simulation, modelers define an effective turbulent Reynolds number Re_T = UL/ν_E using an eddy viscosity ν_E ∼ 1–10⁴ m² s⁻¹ that parameterizes momentum transport by unresolved Kelvin–Helmholtz instabilities and convective eddies.

Why is the critical Reynolds number different for pipe flow versus flow over a flat plate?

The Reynolds number depends on the chosen characteristic length scale L and the baseline stability of the velocity profile. In circular pipe flow, L is the fixed internal diameter D, and transition occurs around Re_D ≈ 2,300. For external flow over a flat plate, L is the streamwise distance x from the leading edge, and Tollmien–Schlichting wave instabilities grow over a much longer fetch before breaking down into turbulent spots at Re_x = U_∞x/ν ≈ 5 × 10⁵ (corresponding to a displacement-thickness Reynolds number Re_δ* ≈ 520).

How do microorganisms swim at very low Reynolds numbers?

At Re ≈ 10⁻⁵–10⁻⁴, inertia is absent and the governing Stokes flow equations are linear and time-reversible. As formulated in Edward Purcell's scallop theorem, any reciprocal swimming stroke that simply opens and closes along the same configuration path produces zero net displacement over a cycle. Microorganisms overcome time-reversibility by breaking kinematic symmetry: bacteria such as Escherichia coli rotate rigid chiral helical flagella driven by proton-motive rotary motors, whereas eukaryotic spermatozoa and ciliates propagate non-reciprocal traveling bending waves along flexible axonemes.

Why do golf balls have dimples?

A smooth sphere traveling at typical golf drive speeds (U ≈ 70 m s⁻¹, D ≈ 4.3 cm) operates at Re ≈ 2 × 10⁵, just below the smooth-sphere drag crisis (Re ≈ 3.5 × 10⁵), resulting in early laminar boundary-layer separation and high pressure drag (C_D ≈ 0.47). Surface dimples act as distributed roughness elements that trip the laminar boundary layer into turbulence at a lower Reynolds number (Re ≈ 4 × 10⁴). The turbulent boundary layer remains attached further around the back of the ball, narrowing the low-pressure wake and cutting total aerodynamic drag nearly in half (C_D ≈ 0.23).

1. Reynolds, O. (1883). "An experimental investigation of the circumstances which determine whether the motion of water shall be direct or sinuous, and of the law of resistance in parallel channels." Philosophical Transactions of the Royal Society of London, 174, 935–982. doi:10.1098/rstl.1883.0029
2. Rott, N. (1990). "Note on the history of the Reynolds number." Annual Review of Fluid Mechanics, 22(1), 1–11. doi:10.1146/annurev.fl.22.010190.000245
3. Purcell, E. M. (1977). "Life at low Reynolds number." American Journal of Physics, 45(1), 3–11. doi:10.1119/1.10903
4. Avila, K., Moxey, D., de Lozar, A., Avila, M., Barkley, D., & Hof, B. (2011). "The onset of turbulence in pipe flow." Science, 333(6039), 192–196. doi:10.1126/science.1203223