2.2.23 · D5Fluid Mechanics
Question bank — Boundary layer separation — adverse pressure gradient
Before we start, a one-line reminder of the two objects every question uses:
- The boundary layer is the thin sheet of slowed fluid hugging the wall (see Boundary layer theory).
- The adverse pressure gradient is : pressure rising along the flow direction . That is linked to speed by Bernoulli's equation.
True or false — justify
Is it true that separation can only happen where the pressure gradient is adverse?
True — with everywhere the near-wall slope never reaches zero, so no reversal. A rising pressure () is the switch that lets backflow begin.
True or false: a favourable pressure gradient () can still separate if the flow is fast enough.
False — speed is not the trigger. A favourable gradient accelerates the near-wall fluid and keeps , so the layer stays attached no matter how fast.
True or false: at the exact separation point the fluid is at rest throughout the boundary layer.
False — only the wall shear is zero. The fluid above the wall is still moving forward; just downstream the near-wall layer reverses.
True or false: an ideal (inviscid) fluid over a cylinder would still separate and leave a wake.
False — with no viscosity there is no slow near-wall layer to stall, so the flow stays attached all the way to the rear stagnation point (d'Alembert's paradox, zero form drag). Viscosity is required to create the vulnerable layer.
True or false: a turbulent boundary layer has more wall friction yet often gives less total drag on a bluff body.
True — turbulence raises skin-friction drag but delays separation, shrinking the low-pressure wake and cutting the much larger form drag; see Drag — form vs skin friction.
True or false: separation always increases drag on any body.
False for a streamlined body where skin friction dominates — there separation is negligible either way. It is true for bluff bodies (cylinders, spheres) where the wake sets the drag.
True or false: an inflection point in the velocity profile guarantees the flow has already separated.
False — the inflection appears as soon as the gradient turns adverse, before reaches zero. It signals a profile prone to reversal, but separation is only at .
True or false: in a converging nozzle the boundary layer separates readily.
False — a converging channel accelerates the flow (, ), a favourable gradient. It is the diverging diffuser that separates; see Diffusers and nozzles.
Spot the error
"Separation happens because the fast free-stream fluid rips the boundary layer off the wall."
The free stream is fast, but it is the slow, momentum-poor wall fluid that stalls under rising pressure. The free stream easily climbs the pressure hill; the tired near-wall fluid cannot.
"Since viscosity slows the wall fluid, viscosity by itself causes separation."
Viscosity only builds the slow layer; the adverse gradient is what stops and reverses it. With viscosity is present but there is no separation.
", so where the flow separates."
Zero curvature at the wall is not the separation condition. Separation is (zero slope). just means the profile has no wall curvature there.
"Golf-ball dimples work by making the surface smoother so air slides off cleanly."
Opposite — dimples deliberately roughen the surface to trip the layer turbulent. Turbulent mixing energises the wall fluid, delaying separation and narrowing the wake.
"In the wall momentum balance the convective terms are dropped because the boundary layer is thin."
They vanish because of the no-slip condition: and exactly at , so there. Thinness gives , a separate fact.
"Separation on a cylinder happens right at the front stagnation point where the flow first hits."
The front half is a favourable gradient (flow accelerating to the shoulder); separation happens past the widest point where pressure rises. Laminar: near ; see Flow over a cylinder and sphere.
"An aerofoil stalls because the air over the top gets too fast."
The fast suction peak is fine; stall occurs when the steep adverse gradient downstream of that peak separates the layer, collapsing lift. Fast flow precedes stall but is not its cause; see Stall on an aerofoil.
Why questions
Why must an adverse gradient force an inflection point into the velocity profile?
At the wall makes the curvature positive, but far out the profile must bend the other way (negative curvature) so smoothly — the sign change is the inflection.
Why does a turbulent boundary layer resist the adverse gradient longer than a laminar one?
Turbulent eddies mix high-momentum outer fluid down to the wall, so the near-wall fluid is energy-rich and can climb further up the pressure hill before stalling.
Why does separation on a bluff body create such large drag?
The separated flow leaves a wide, low-pressure recirculating wake; the pressure imbalance between high-pressure front and low-pressure rear is the dominant form drag.
Why can Bernoulli's equation be applied at the edge of the boundary layer but not deep inside it?
Bernoulli assumes negligible viscous losses; that holds in the fast outer flow but fails inside the layer where viscosity dissipates energy. So is used only on the edge streamline.
Why does a diffuser with too large a cone angle "stall"?
A steeper divergence means falls faster, so is more strongly adverse over a short distance — the layer cannot cope and separates, dumping the flow into a lossy recirculating region.
Why is the separation condition stated as rather than "velocity reverses"?
is the precise onset: the wall slope first touches zero. Reversed flow () exists only downstream of that point, so marks the exact boundary.
Edge cases
Edge case: exactly at the separation point, what is the sign of ?
Exactly zero at the point; positive (forward) just upstream, negative (backflow) just downstream. Separation is the crossing point of that slope through zero.
Edge case: what happens if the pressure gradient is exactly zero () all along a flat plate?
No separation — this is the classic zero-gradient flat-plate layer. The profile keeps everywhere; the layer just thickens with .
Edge case: can separation occur even though the whole surface is in an adverse gradient from the start?
Yes, and it occurs sooner. The layer has no favourable stretch to build momentum, so the near-wall fluid stalls at a smaller downstream distance.
Edge case: for a very streamlined body (thin aerofoil at low angle) is form drag from separation significant?
Usually negligible — the gentle gradients keep flow attached, so skin friction dominates the drag budget instead of wake pressure.
Edge case: at very low Reynolds number (creeping flow), does a cylinder still shed a big separated wake?
No — at viscous forces dominate and the flow stays attached (Stokes flow), with almost no wake. The wide wake is a moderate-to-high phenomenon.
Edge case: if you keep tripping a layer turbulent earlier and earlier, does drag keep dropping forever?
No — once separation is already delayed, further tripping only adds skin-friction drag with no wake benefit, so total drag rises again. There is an optimum, not a monotonic gain.