Oblique Shock Waves: How Supersonic Air Turns a Corner

The geometry: a shock that decides its own angle

When a uniform supersonic stream at Mach number M₁ meets a rigid ramp deflected by an angle θ (the flow-turning or deflection angle), it cannot turn the corner isentropically the way a subsonic flow would. Instead it forms a straight, planar shock inclined at a wave angle β measured from the upstream flow direction. The flow crosses the shock, turns by exactly θ so it runs parallel to the ramp surface, and emerges at a lower Mach number M₂.

The single most useful fact in the whole subject is the normal-component decomposition: an oblique shock behaves like a normal shock acting only on the velocity component perpendicular to the wave, while the tangential component is preserved unchanged. Define:

• Mₙ₁ = M₁ sin β — the upstream Mach number normal to the shock. This must exceed 1 for a shock to exist.
• Mₙ₂ = M₂ sin(β − θ) — the downstream normal Mach, always subsonic.

So you take every normal-shock relation you already know, feed it Mₙ₁ instead of M₁, and read off the pressure, density, and temperature jumps. The tangential velocity being conserved is why the flow bends toward the ramp: the normal component collapses while the tangential component stays, so the resultant vector rotates into the surface.

The θ–β–M relation and the Rankine–Hugoniot jumps

The link between the three angles is the θ–β–M relation, the workhorse equation of supersonic aerodynamics:

tan θ = 2 cot β · (M₁² sin²β − 1) / (M₁²(γ + cos 2β) + 2)

with γ = 1.4 (ratio of specific heats for air). Given M₁ and the geometric deflection θ, you solve this transcendental equation for β. The property jumps then follow from the Rankine–Hugoniot normal-shock relations evaluated at Mₙ₁:

• Pressure: p₂/p₁ = 1 + (2γ/(γ+1))(Mₙ₁² − 1)
• Density: ρ₂/ρ₁ = (γ+1)Mₙ₁² / ((γ−1)Mₙ₁² + 2)
• Temperature: T₂/T₁ = (p₂/p₁)(ρ₁/ρ₂), consistent with the ideal-gas law
• Downstream normal Mach: Mₙ₂² = (1 + ½(γ−1)Mₙ₁²) / (γMₙ₁² − ½(γ−1))

Worked case: M₁ = 3, θ = 20°. Solving θ–β–M gives β ≈ 37.8°, so Mₙ₁ = 3·sin 37.8° ≈ 1.84. That yields p₂/p₁ ≈ 3.77, T₂/T₁ ≈ 1.56, and M₂ ≈ 1.99 — the flow is compressed and slowed but stays firmly supersonic. Crucially the shock is adiabatic but not isentropic: total temperature T₀ is conserved, but entropy rises and stagnation pressure drops (here p₀₂/p₀₁ ≈ 0.80). That 20% stagnation-pressure loss is the price of one compression, and it scales roughly with (Mₙ₁ − 1)³ at modest strengths — the physical reason engineers prefer several weak oblique shocks over one strong one.

Weak vs strong solutions, and the Mach angle floor

The θ–β–M relation is quadratic-like in behavior: for any deflection θ below the maximum, it returns two valid wave angles. The smaller β is the weak solution; the larger β is the strong solution.

• The weak shock (small β, near the Mach angle) leaves the downstream flow supersonic in nearly all practical cases, produces a smaller entropy rise, and is what actually forms on wedges and ramps in unbounded flow. Nature selects it because the downstream boundary conditions of an external flow favor it.
• The strong shock (large β, approaching 90°) drives the flow subsonic and appears only when a high downstream back-pressure forces it — for example just behind a detached bow shock near the stagnation streamline, or in an over-contracted inlet.

The lower bound on β is the Mach angle, μ = arcsin(1/M₁). At θ → 0 the weak 'shock' degenerates into a zero-strength Mach wave lying at exactly μ (e.g. μ = 19.5° at M = 3, 11.5° at M = 5). The upper bound is β = 90°, the normal shock. Everything useful lives in the band μ ≤ β ≤ 90°, and the whole design game is choosing where in that band to operate.

Shock detachment: the hard limit on turning angle

For every M₁ there is a maximum deflection angle θ_max the flow can negotiate with an attached oblique shock. Try to turn the flow more sharply than θ_max and the shock detaches: it can no longer sit on the wedge tip, so it bows out ahead of the body as a curved bow shock with a locally normal (β = 90°) segment on the centerline and a subsonic pocket behind it.

• θ_max grows with Mach number: about θ_max ≈ 12° at M = 1.5, ≈ 34° at M = 3, and it asymptotes near 45.6° (γ = 1.4) as M → ∞.
• At θ_max the weak and strong solutions merge into a single β; the deflection at which the downstream flow first goes sonic (θ*) is a hair below θ_max, so an attached shock is essentially always weak-solution supersonic.

Detachment is the dominant real-world failure mode. A supersonic inlet designed for an attached ramp shock will, at too-low a flight Mach number or too-aggressive a ramp schedule, throw a detached bow shock that spills flow, collapses mass capture, and can trigger inlet unstart — a sudden, sometimes explosive expulsion of the shock system that on the SR-71 produced yaw excursions violent enough to bang the pilot's helmet against the canopy. Variable-geometry ramps exist precisely to keep θ below θ_max across the flight envelope.

Sizing supersonic inlets: the multi-shock compression

A single normal shock at M = 3 recovers only ~33% of stagnation pressure — unacceptable for a jet engine, whose thrust scales almost linearly with inlet pressure recovery. The fix, formalized by Oswatitsch's theorem, is to stage the compression through several oblique shocks of gradually increasing strength, followed by a weak terminal normal shock. The design procedure:

• Set the total turning. Decompose the required compression into n ramps, each turning the flow a few degrees (typically 6–12°), well inside θ_max at every station.
• Chase equal-strength shocks. Oswatitsch showed total-pressure recovery is maximized when all oblique shocks have equal Mₙ₁. For a 2-ramp external-compression inlet at M = 3, that logic lands each oblique shock near Mₙ₁ ≈ 1.5–1.6.
• Add the terminal normal shock at a throat Mach only slightly above 1 (say M ≈ 1.3), so its loss is tiny.
• Tally the recovery. A two-oblique-plus-normal system at M = 3 recovers p₀₂/p₀₁ ≈ 0.85–0.90 versus 0.33 for a bare normal shock — a 2.5× improvement that translates almost directly into thrust and range.

The classic realization is the mixed-compression, axisymmetric spike inlet of the SR-71/J58, whose translating conical centerbody moved up to ~66 cm fore-and-aft to hold the shock system on the cowl lip from Mach 1.6 to 3.2. Two-dimensional external-compression ramp inlets (F-15, Concorde) do the same job with hinged wedge plates driven by the air-data computer. MIL-E-5008 / AIA reference recovery schedules set the certification bar these designs are measured against.

Real hardware and the shock diamond

Oblique shocks are everywhere the flow is supersonic and something is not perfectly aligned with it:

• Supersonic wing leading edges and diamond/biconvex airfoils generate attached oblique shocks whose pressure jump is the source of wave drag — the extra drag component unique to supersonic flight, quantified for a thin airfoil by Ackeret's linearized theory as C_d ∝ 1/√(M²−1) times the thickness and camber terms.
• Scramjet isolators ride a train of reflecting oblique shocks (a 'shock train') that decelerates Mach-2-plus core flow while resisting the back-pressure from combustion — the shock structure is the combustor's mechanical fuse.
• Over- and under-expanded rocket and jet nozzles produce the iconic shock diamonds (Mach disks). When exhaust pressure differs from ambient, oblique shocks and Prandtl–Meyer expansion fans reflect off the free jet boundary in a repeating diamond lattice; each bright node is a local compression where combustion products re-ignite. F-1, Merlin, and afterburning turbofan plumes all show them.
• Shock–boundary-layer interaction (SBLI) is where oblique shocks meet real, viscous surfaces: the adverse pressure gradient can separate the boundary layer, throwing a lambda (λ) foot shock, causing buzz, unsteady loads, and localized heating spikes — a first-order concern in transonic wing buffet and inlet design.

Measurement is classically done with schlieren and shadowgraph optics, which visualize the density gradient ∂ρ/∂n across the wave, letting engineers read β directly and back out M₁.

Limits, assumptions, and best practice

The clean θ–β–M picture rests on assumptions worth stating explicitly, because every one of them is a place designs go wrong:

• Calorically perfect gas (γ = 1.4). Above roughly M = 5, temperatures behind the shock exceed ~1000–2000 K and vibrational excitation, dissociation, and ionization make γ drop toward 1.2–1.3. Real-gas effects reduce the temperature jump but increase density ratio and can move shock stand-off distance by 10–30% — hypersonic vehicles (X-15, Space Shuttle, reentry capsules) must use real-gas or CFD models, not the perfect-gas chart.
• Inviscid, thin shock. The relations ignore the boundary layer; SBLI can thicken the effective body, shift β, and cause separation the ideal theory never predicts.
• Steady, straight, planar shock. Curved and conical shocks (a cone in supersonic flow) obey the Taylor–Maccoll equations, not the 2-D wedge relation — a cone of the same half-angle makes a weaker, more-attached shock than a wedge because the 3-D relief lets flow spill around it.

Best-practice rules of thumb: keep each ramp turn ≥ 3–5° inside θ_max with margin for angle-of-attack and Mach droop; prefer many weak shocks to one strong one for pressure recovery; provide bleed or boundary-layer diverters at every SBLI location; and always verify against θ_max and the sonic-deflection limit before trusting a two-shot design. When in doubt near θ_max, assume the shock will detach in service before it does on the drawing board.

Oblique shock (weak solution, wedge) vs normal shock at the same freestream Mach number M₁ = 3.0, air, γ = 1.4

Property	Oblique shock (θ = 20°, β ≈ 37.8°)	Normal shock	Why it matters
Normal Mach into shock	Mₙ₁ = 3·sin37.8° ≈ 1.84	Mₙ₁ = M₁ = 3.0	Oblique shock is a 'weaker' normal shock
Downstream Mach M₂	≈ 1.99 (still supersonic)	0.475 (subsonic)	Inlets keep flow supersonic then repeat
Static pressure ratio p₂/p₁	≈ 3.77	10.33	Lower jump = less loss per compression
Static temperature ratio T₂/T₁	≈ 1.56	2.68	Sets heating on leading edges
Stagnation pressure ratio p₀₂/p₀₁	≈ 0.80 (20% loss)	0.328 (67% loss)	Multi-shock inlets beat one normal shock
Flow direction change	Turns 20° into the wedge	None (straight through)	Oblique shock does the corner-turning