Rocket Airbrake and Fin Aerodynamics

Project Waxwing is HyPower Bristol's entry into the 3 km Liquid category at the European Rocketry Challenge 2025, built out of BristolSEDS at the University of Bristol. It is a student researched and developed bi-propellant vehicle carrying a deployable airbrake system used to trim apogee.

My part was the external aerodynamics. Two things needed answering. The airbrakes needed a drag model before their controller could be tuned, because a controller that modulates drag has to know how much drag it is commanding. And the fins needed a section and a planform chosen on evidence rather than preference. Both problems sit across the subsonic and transonic range, since the vehicle reaches Mach 0.9 early in the burn.

The vehicle reaches Mach 0.9 about five seconds into flight, with apogee near 3850 m. The airbrakes operate on the coast, at velocities up to 150 m/s.

Compressibility becomes significant above roughly Mach 0.3, which is around 100 m/s at these conditions. The sweep was capped there on purpose, so that the model stayed a clean incompressible one rather than a compressible one with nothing to check it against. The cost of that decision is real and worth stating: the top of the deployment envelope, 100 to 150 m/s, is not covered by these results.

Two flap geometries were on the table, a slot design and a rounded petal. The choice was not simply which one produced more drag. A brake that sheds a large separated wake risks that wake reaching the fins, which would put stability at risk at exactly the moment the vehicle is being slowed.

The rounded petal produced greater braking force but separated further from the body. The slot design gave less drag with a more predictable separation for a constant base length. The configuration taken forward was a compromise between the two rather than either one outright.

A second constraint came out of the geometry rather than the flow. The number of flaps had to match the number of fins, so the two sit clocked apart and the wake from a brake does not run straight into a fin downstream.

Each case ran to 1000 iterations on a segregated solver, by which point continuity, energy and momentum residuals had fallen to between 10⁻⁷ and 10⁻¹⁰.

A drag number is only worth handing to a control team if there is some independent reason to believe it. Three checks were run on the solution before any of it was used.

Drag rises with both deployment and velocity, and the sensitivity to deployment grows with speed. At 20 m/s the entire deployment range moves Cd by under 0.005. At 100 m/s the same range moves it by more than 0.1. That spread is the reason a single drag figure would not have been sufficient for the controller.

Those recirculation regions close quickly, and the flow is reattached by the time it reaches the fins. At this condition the brakes do not compromise fin effectiveness, which was the clearance the mechanical design needed before the geometry could be frozen.

The fins see both subsonic and transonic flow, since the vehicle passes Mach 0.9 on the way up. A section that behaves well in one regime is not automatically right for the other.

The section tested was a hybrid: a NACA leading edge to keep the flow attached subsonically and delay shock induced separation transonically, then a sharp wedge trailing edge to fix the rear stagnation point and cut wave drag. It was compared against standard NACA sections at three thicknesses, evaluated at the trajectory average condition rather than at a single design point.

The hybrid at 10 percent thickness gave the lowest drag and was selected. The margin over the equivalent standard section is about 1.3 percent, which is small, and it is worth reporting as small rather than dressing it up. The CFD drag figures were cross checked against a MATLAB panel method written to run over the same geometries, so the ranking rests on two independent methods rather than one.

The planform was chosen separately, through a weighted decision matrix over five candidates. The weightings came from a pairwise comparison of the requirements rather than being assigned directly, and structural stiffness at maximum dynamic pressure, static stability and flutter margin carried three quarters of the weight between them.

Swept trapezoidal won. The elliptical planform lost on manufacturability in composite rather than on aerodynamics, which is worth saying plainly, because a downselection is more useful when the losing reason is specific. Three fins were used rather than four, so the airbrake wake has fewer chances to meet a fin downstream.

The chosen fin was built from three triaxial carbon plies in a [-45, 0, 45] layup. For the structural check I took the worst case flight condition, Mach 0.9 at 850 m with a conservative 15 degree angle of attack, and reduced it to an equivalent uniform pressure of 146 kPa through dynamic pressure and a finite wing lift curve slope, which is what was applied to the Abaqus model of the fin.

These are the assumptions recorded at the time, not ones reconstructed afterwards.