The Blasius Flat-Plate Solution

1. Setup and the similarity guess

Uniform stream U over a flat plate at zero incidence: the outer pressure gradient vanishes, and the only length available at station x is the local diffusion length √(νx/U). Nothing else can matter, so the profile must be self-similar:

The stream function ψ satisfies continuity identically; v follows from v = −∂ψ/∂x.

2. Reduction to the Blasius equation

Substituting into Prandtl's boundary-layer momentum equation, every x cancels and a single ODE remains:

Third order, nonlinear, no closed-form solution — but one number characterises it completely. Numerical integration (shooting on f″(0)) gives

3. Everything follows from that number

Wall shear, boundary-layer thickness and drag coefficient all come out of the similarity structure:

The layer grows like √x; the wall shear decays like 1/√x. These formulas hold up to transition near Re_x ≈ 5×10⁵.

4. Why this problem matters beyond plates

Blasius is the template: pick the similarity variable from the physics, collapse the PDE to an ODE, pin the answer to one wall derivative. The stretching-sheet problem repeats the recipe with a moving wall — and there, remarkably, the ODE solves in closed form.

5. Extended worked examples

Example 1 — full drag of a plate. Air at U = 5 m/s over a 1 m × 1 m plate: Re_L = 5×1/1.47×10⁻⁵ = 3.4×10⁵ (laminar). C_D = 1.328/√Re_L = 2.28×10⁻³. Drag per side D = C_D·½ρU²A = 2.28×10⁻³×0.5×1.225×25×1 = 0.035 N; wetted both sides, 0.070 N. Seventy millinewtons for a square metre — laminar skin friction is genuinely small, which is precisely why laminar-flow control obsesses aircraft designers.

Example 2 — reading the Blasius table. The universal profile f′(η) (standard published values): f′ = 0.330 at η = 1; 0.630 at η = 2; 0.846 at η = 3; 0.956 at η = 4; 0.992 at η = 5. So the "edge" δ₉₉ sits near η ≈ 5 — the origin of the 5.0 in δ = 5.0x/√Re_x. One dimensionless curve contains every laminar flat-plate profile that has ever existed: measure u/U against y√(U/νx) in any lab and the data collapse onto it, one of the most-repeated validation experiments in the subject.

Example 3 — local versus average friction. At x = 0.25 m in Example 1: Re_x = 8.5×10⁴, C_f = 0.664/292 = 2.28×10⁻³; τ_w = 0.035 Pa. Note the neat factor: the drag coefficient 1.328/√Re_L is exactly twice the trailing-edge C_f — because C_f ∝ x^{−1/2} and the average of x^{−1/2} over [0, L] is twice its endpoint value. The leading edge, where the formula diverges, contributes finite drag: integrable singularity, physically a region where boundary-layer theory itself needs the leading-edge (Navier–Stokes) correction.

Example 4 — the layer breathes outward. The transverse velocity at the layer's edge tends to v_e = 0.860 U/√Re_x — at x = 0.25 m above, v_e ≈ 0.0148 m/s. Small but nonzero: the plate displaces the outer flow as if the body were thickened by δ* = 1.721x/√Re_x. Aerodynamicists add exactly this displacement body to potential-flow models — the viscous and inviscid worlds negotiating through one number.

6. Common misconceptions

"Exact solution means closed-form formula." Blasius's solution is exact in the sense that the reduction from PDE to ODE is without approximation; the ODE itself is then solved numerically to any desired accuracy. "Exact" describes the mathematics' honesty, not the presence of elementary functions.

"f″(0) = 0.332 is measured, so it could be refined by experiment." It is a mathematical constant of the ODE, computable to arbitrary digits (0.332057…). Experiments confirm the model; they do not define the number.

"Blasius applies to any plate flow." Zero pressure gradient, laminar, steady, sharp leading edge, no suction or heating that alters properties. A pressure gradient sends you to Falkner–Skan; transition sends you to turbulent correlations; both exits are well signposted.

"The wall shear grows downstream because the layer grows." Opposite: τ_w ∝ x^{−1/2} decays — a thicker layer spreads the same velocity jump over more distance, weakening the gradient at the wall. The leading edge is where the shear is fiercest.

7. Where this shows up

Every CFD boundary-layer solver, commercial or academic, publishes a Blasius comparison as its birth certificate — grid convergence to f″(0) = 0.3321 and the profile table is the community's entrance exam. Ship-hull and aircraft skin-friction estimates start from the flat-plate laminar/turbulent formulas with Blasius anchoring the laminar branch. Hot-wire anemometer calibration rigs, MEMS shear-stress sensors, and the "leading-edge laminar run" that racing-yacht and glider designers polish for are all priced in Blasius units. The solution even seeds turbulence research: the laminar profile it predicts is the base state whose instability (Tollmien–Schlichting waves) begins the transition story.

8. Practice problems

P1. Water at 1 m/s: compute δ, δ*, θ at x = 0.1 m. Solution: Re_x = 10⁵; √ = 316; δ = 1.58 mm; δ* = 1.721×0.1/316 = 0.545 mm; θ = 0.664×0.1/316 = 0.21 mm; shape factor H = δ*/θ = 2.59 ✓ (the laminar signature).

P2. At what η does the flow reach half the free-stream speed? Solution: interpolating the table between η = 1 (0.330) and η = 2 (0.630): η ≈ 1.57.

P3. Total drag (both sides) of a 0.3 m × 0.3 m plate in water at 1 m/s. Solution: Re_L = 3×10⁵; C_D = 1.328/548 = 2.42×10⁻³; D = 2×2.42×10⁻³×0.5×998×1×0.09 = 0.218 N.

P4. Show from δ ∝ √x that the layer's growth rate dδ/dx at x = L equals δ(L)/2L, and evaluate for Example 1. Solution: differentiate; dδ/dx = 4.3 mm/2×0.5 m ≈ 0.0043 — the layer climbs 4 mm per metre: a wedge of slope a quarter of a degree.

9. Going deeper

Add an outer flow U(x) = Cx^m — flow past a wedge of angle πβ/(2−β) with β = 2m/(m+1) — and the same machinery yields the Falkner–Skan equation f‴ + ff″ + β(1 − f′²) = 0: Blasius is β = 0, stagnation flow is β = 1, and small negative β describes decelerating flow, with separation appearing exactly at β = −0.1988 where f″(0) → 0. The family thus maps the whole attached-flow landscape with one parameter. Its reversed-wall cousin — the stretching sheet — swaps the driving from outer stream to moving boundary, and there the ODE finally surrenders a closed form.

10. Historical context

Paul Richard Heinrich Blasius was twenty-four when he solved the flat plate — the problem was his doctoral work under Prandtl at Göttingen, published in 1908, four years after Prandtl's founding paper. He computed the profile by matched series expansions (a near-wall power series stitched to a far-field asymptotic form) entirely by hand, obtaining f″(0) ≈ 0.332 with arithmetic that modern replays confirm was essentially error-free. Töpfer (1912) found the scaling trick of §11 that removes the iteration altogether; Howarth (1938) produced the standard high-precision tables; today the constant is known to dozens of digits and the ODE's series' convergence properties are themselves a research literature. Blasius left research early for a teaching career at Hamburg — his other immortality is the Blasius friction correlation for turbulent pipe flow — but the seventh-decimal agreement between his hand computation, modern collocation codes, and wind-tunnel measurements of the profile remains one of applied mathematics' cleanest full-circle stories.

11. Another way to see it: Töpfer's trick — a BVP solved without shooting

The Blasius equation (in the normalisation 2f‴ + ff″ = 0, common in texts) has a scaling symmetry: if F(η) is any solution, so is f(η) = λF(λη) for any λ > 0. Exploit it. Solve the initial-value problem with F(0) = F′(0) = 0 and a guessed F″(0) = 1; integrate outward and read off the limit F′(∞) = C (a definite number, ≈ 2.0854). The rescaled f = λF(λη) has f′(∞) = λ²C, so choosing λ = C^{−1/2} enforces the true boundary condition f′(∞) = 1 exactly — no iteration, no root-finding. The physical wall shear follows as f″(0) = λ³F″(0) = C^{−3/2} ≈ 0.332. One IVP, one rescaling: the boundary-value problem dissolves. The trick works precisely because the equation and boundary conditions share a one-parameter symmetry group — a miniature of the Lie-group story behind similarity itself, and a standard first exercise in "use the symmetry before you use the computer."

12. Frequently asked questions

Different books show 2f‴ + ff″ = 0, f‴ + ½ff″ = 0, f‴ + ff″ = 0 — which is right? All: they differ by a √2 in the definition of η. The physical outputs (δ, C_f, drag) are identical; only tabulated constants shift. Check a source's η definition before borrowing its numbers — the classic cross-book error in student theses.

How well does experiment match? Superbly, in clean conditions: hot-wire traverses collapse onto f′(η) within instrument error for Re_x up to transition. Discrepancies appear exactly where assumptions fail — near the leading edge, near transition, under free-stream turbulence — making Blasius a diagnostic as much as a prediction.

Is there a closed-form solution? None known in elementary or standard special functions. The power series at the wall has finite radius of convergence (a complex-plane singularity limits it), which is why series plus asymptotic matching — or numerics — is the historical and practical route.

Why does the profile have an inflection-free shape, and does it matter? With zero pressure gradient, f‴(0) = 0 but the profile stays convex (no interior inflection point). By Rayleigh's criterion, inflectional profiles are inviscidly unstable — adverse-gradient (Falkner–Skan, β < 0) profiles develop inflections and trip to turbulence far earlier. The shape of one curve encodes the plate's comparatively long laminar run.

What is the displacement thickness used for, concretely? Wind-tunnel walls diverge slightly along the test section — by exactly the calculated δ* growth — so the effective free-stream stays uniform; nozzle and inlet designers do the same. A theoretical construct with a machining tolerance attached.

13. Further practice

P5. Using C = F′(∞) = 2.0854 in Töpfer's construction, verify f″(0). Solution: C^{−3/2} = 2.0854^{−1.5} = 0.3321 ✓ — the tabulated constant reproduced from the IVP route.

P6. A glider wing section holds laminar flow to Re_x = 3×10⁶ at 30 m/s (air). How far is that from the leading edge, and what is δ there? Solution: x = Re ν/U = 3×10⁶×1.47×10⁻⁵/30 = 1.47 m; δ = 5×1.47/√(3×10⁶) = 4.2 mm — a metre and a half of drag savings hanging on millimetres of smoothness.

P7. Show that between stations x and 4x, δ doubles while τ_w halves. Solution: both scale as √x and 1/√x respectively: √4 = 2 ∎ — the two facing scalings that define the layer's downstream life.

14. Worked exam problem

Problem. A survey drone's wing: chord 0.15 m, span 1 m per side, speed 12 m/s at sea level. Treating each surface as a flat plate: (a) confirm the laminar assumption, (b) find the friction drag of one wing (both surfaces), (c) find δ and δ* at the trailing edge.

Solution. (a) Re_L = 12×0.15/1.47×10⁻⁵ = 1.22×10⁵ < 5×10⁵ ✓ laminar throughout — small, slow aircraft live in Blasius's regime. (b) C_D = 1.328/√(1.22×10⁵) = 3.80×10⁻³; D = 2×C_D×½ρU²(cL_span) = 2×3.80×10⁻³×0.5×1.225×144×0.15 = 0.101 N per wing. (c) √Re = 349.6: δ = 5×0.15/349.6 = 2.1 mm; δ* = 1.721×0.15/349.6 = 0.74 mm. Add both wings and the tail and this two-line friction budget is a genuine first sizing of the drone's cruise power — Blasius as flight-performance engineering.

15. Key takeaways

Blasius reduced the flat-plate boundary layer exactly to f‴ + ff″ = 0 (normalisations vary), whose universal profile carries the constants of the trade: δ = 5.0x/√Re_x, C_f = 0.664/√Re_x, C_D = 1.328/√Re_L, H = 2.59. "Exact" refers to the reduction; the ODE is solved numerically — or by Töpfer's rescaling, no iteration required. The solution is the community's benchmark: codes and experiments alike are graded against it.

16. Where to go next

Pressure gradients generalise it to Falkner–Skan; a moving wall replaces it with the stretching-sheet problem, where a closed form finally exists; the numerics that solve all of them are examined in shooting and bvp4c. The Reynolds number calculator evaluates every trailing-edge quantity from this chapter live.