Aircraft and flying taxis: flight models that miss flight tests

High-power electric drives for aircraft are hard to control

What happens. An electric propulsion drive for an aircraft combines high power, low weight and little cooling margin. The inverter and motor are pushed close to their limits, and the control has to deliver torque quickly and cleanly while the temperatures, voltages and currents stay inside them. Experienced drive-control engineers are scarce.

What it costs. Derating in flight, extra weight for margin, noise and losses from poor torque control, and a development schedule held back by a role that stays open.

Why the usual fix fails. Drive controllers from industry were built for other duty cycles. Adding margin costs weight, which is the scarcest thing in an aircraft. Waiting for a new hire delays the whole propulsion chain.

How a fixed-price remote diagnosis checks it. From test-bench logs (currents, voltages, speed, torque, winding and inverter temperatures) and your drawings I look at where the current control gives away power or comes close to limits. The fixed-price remote diagnosis tells you whether predictive drive control can use the limits better, and what it needs on your hardware.

Background: Predictive torque and drive control

Aerodynamic models drift apart across simulation, wind tunnel and flight test

What happens. The aerodynamic model of a new aircraft exists three times: from calculation, from the wind tunnel and from flight. They differ, and every new test result has to be put into one model. The people doing that are few and each update takes weeks.

What it costs. Flight-control design on a model nobody fully trusts, extra flight tests to close the gaps, and decisions about margins that rest on opinion.

Why the usual fix fails. Adjusting coefficients by hand until one test case fits breaks the others. A bigger simulation does not tell which source is right. More engineers on the task mean more versions of the model.

How a fixed-price remote diagnosis checks it. From your simulation results, wind-tunnel tables and flight-test recordings I look at where the sources disagree and whether the differences are consistent. The fixed-price remote diagnosis tells you whether model identification from the combined test data can give one traceable model, and what data it needs.

Flight control is designed, but models are not improved from flight data

What happens. A flight-control team is being built while the first flights produce data. The designed control laws rest on a model from the design phase. Flight data show where the model is wrong, but there is no routine that turns them into an improved model.

What it costs. Control gains that stay conservative, flight tests repeated to see what the data already shows, and a team that spends its time on data handling.

Why the usual fix fails. Hiring helps in a year, not for the next test campaign. Tools from other projects do not know your aircraft. Tuning in the flight test itself costs expensive flight hours.

How a fixed-price remote diagnosis checks it. From flight recordings (inputs, surface positions, attitude, accelerations) and your current model I look at how well the model predicts the flights and where it needs to be corrected. The fixed-price remote diagnosis tells you whether identification plus a model-based control design is a realistic next step for the team.

Flight-control architecture and mode management for transition flight

What happens. A vertical take-off aircraft that flies forward like an aeroplane has to pass from hover to wing-borne flight. That needs a clear architecture: which controller is active, how they hand over, what happens when a sensor or an actuator fails. The architecture grows by additions and nobody can see all the mode combinations.

What it costs. Transitions that are flown carefully and slowly, flight tests lost to mode errors, and a team that struggles to lead and to design at the same time.

Why the usual fix fails. Another mode for each case produces a logic tree that is hard to verify. Separate controllers for hover and forward flight leave the transition to a short blend. Hiring a lead takes months.

How a fixed-price remote diagnosis checks it. From architecture drawings, mode logic and test or simulation logs of transitions I look at where handovers cause jumps and which combinations are untested. The fixed-price remote diagnosis tells you whether one predictive control across the transition would simplify the architecture, and what it would take.

Background: Industrial and aerial predictive control

Simulator flight models break down at stall and in aerobatics

What happens. Training and engineering simulators fly well in normal flight. At the edge of the envelope (stall, spins, aerobatics) the model is thin, because there are few measurements and the physics is hard. Pilots feel that the aircraft behaves differently from the real one.

What it costs. Simulator training that teaches the wrong feel, flight-test preparation that cannot trust the simulator, and certification effort spent on proving fidelity.

Why the usual fix fails. Extrapolating normal-flight coefficients into the stall fails exactly where it matters. Pilot comments lead to patches that fix one manoeuvre and spoil another. Flight tests into the stall are risky and expensive.

How a fixed-price remote diagnosis checks it. From available flight-test data and the current model I look at how well each manoeuvre is reproduced and which parts of the envelope rest on assumptions. The fixed-price remote diagnosis tells you whether data-based model improvement can close the gaps, and which test points are worth flying.

Simulation modules for aircraft systems stay unwritten

What happens. A full aircraft simulation needs models of the systems: fuel, hydraulics, electrics, engines, brakes. Each is a module that has to be written, tested and fit together. Specialists for it are rare, and the work stays on the list for months.

What it costs. Simulators that cannot yet show system failures, test campaigns waiting for the model, and engineers pulled from other tasks.

Why the usual fix fails. A new employee needs months to learn the existing structure. Buying a generic library leaves the interfaces and the aircraft-specific behaviour open. Outsourcing without a clear scope returns modules that do not fit.

How a fixed-price remote diagnosis checks it. From your interface descriptions, existing modules and available system data I look at which modules are missing, in what order they depend on each other and how much of each can come from a physical model. The fixed-price remote diagnosis tells you what a clean split of the work into modules would be.

Requirements and verification of fly-by-wire systems

What happens. In a fly-by-wire aircraft the controls act through a computer. Every requirement has to be written clearly, linked to a test and shown to be met. The control laws change during development, and requirements and tests lag behind them.

What it costs. Late findings in verification, rework on laws that already flew, and review cycles with the certifying body that take long.

Why the usual fix fails. More people on documentation do not make requirements consistent. Late tests find errors when the design is hardest to change. Tools hold the text but do not know the control meaning of a requirement.

How a fixed-price remote diagnosis checks it. From requirement documents, control-law descriptions and test results I look at which requirements can be checked by simulation of the control loop, and where gaps and contradictions are. The fixed-price remote diagnosis tells you what could be verified earlier and how.