A Tandem Rotor Is Not Two Helicopters

When we extended TaskFlight from one airframe to five, the tempting refactor was obvious: take the AH-64 dynamics model, pull its constants into a spec object, and feed each new aircraft different numbers. Heavier mass for the Black Hawk, less inertia for the Little Bird, done. That works right up until the CH-47 — and the way it fails is instructive, because it fails by producing a helicopter that flies plausibly and is wrong about everything a Chinook pilot or mechanic actually needs to understand.

What the tail rotor is actually for

On a conventional single-rotor helicopter, the engine drives the main rotor, and Newton's third law drives the fuselage the other way. The tail rotor exists to cancel that reaction torque, and the pedals set its thrust. This has consequences that thread through the whole flight model: pull collective and the reaction torque rises, so the pilot feeds pedal with power. Our AH-64 model implements that coupling, and one of our better bugs lived exactly there — parked on the ramp with flat pitch, the aircraft slowly swung 23.8 degrees in ten seconds, because the tail rotor trim was computed against a fixed hover reference instead of the torque the main rotor was actually producing. At near-zero lift there was almost nothing to oppose, and the trim pushed the nose around at full hover authority. The fix scales trim by produced lift; drift now measures 0.00 degrees parked.

The point of retelling that bug: anti-torque is not a feature you bolt onto a helicopter model. It is a coupling woven through collective, pedals, power and trim. Which is a problem when your next airframe does not have it.

The Chinook cancels the problem instead of trimming it

The CH-47 carries two rotors turning in opposite directions. Their reaction torques cancel each other continuously — there is no residual torque to trim and no tail rotor to trim it with. So the entire anti-torque chapter of the conventional model is not "set to zero" on a Chinook; it is the wrong chapter. Three control axes work differently:

• Yaw — with no tail rotor, the pedals tilt the two rotor discs laterally in opposite directions. The front disc pulls one way, the aft disc the other, and the couple yaws the fuselage. Yaw authority therefore rides on rotor thrust, not on a separate device with its own thrust curve.
• Pitch — a conventional helicopter pitches by tilting its single disc. A tandem pitches primarily by differential collective: more lift on the aft head pushes the nose down, more on the front head pulls it up, acting across the ~12 m separation between the hubs. The moment arm is the fuselage itself.
• Power-pedal coupling — gone. Pull collective on a Chinook and both torques rise together and keep cancelling. A trainee who has internalised 'power change means pedal input' from the Apache needs to actively unlearn it here.

Branch, don’t parameterise

You can fake a tandem inside a conventional model — set tail rotor effectiveness to zero, add a synthetic yaw input, overdrive pitch damping. We tried sketching it and stopped, because every fake leaves a dead code path that still executes and a coupling that is quietly wrong. Instead the registry in lib/flight/aircraft.ts declares three physics families — conventional rotor (AH-64, UH-60, AH-6), tandem rotor (CH-47) and fixed wing (737) — and the dynamics module branches where the physics genuinely differs rather than pretending one set of equations covers all of them.

Why a maintenance trainer should care

TaskFlight is maintenance courseware first, and the tandem configuration is not just a flight-model curiosity — it reorders the maintenance stakes. On a single-rotor aircraft the transmission and rotor systems are critical, but the two rotors of a Chinook are mechanically phased through a synchronising shaft so the intermeshing blades never occupy the same space. Lose rotor phase and the blades intersect. That makes the sync shaft the most severe entry in our fault catalogue, and rotor phasing discipline threads through both hub records and the transmission content for the CH-47. A flight model that treated the Chinook as "an Apache with different numbers" could never make that consequence real — and making consequences real is the entire reason the flight simulator sits next to the trainer.