Control Methods for Aiding a Pilot During STOL Engine Failure Transients (1977)

Ernest R. Nelson's report, with Daniel B. DeBra as principal investigator, studies automatic recovery from engine-failure transients in a short-takeoff-and-landing aircraft. Most of the document is application-specific: aerodynamic data, reduced and full state models, modal approximations, feedback-path choices, simulations, and implementation considerations.

The part used by GTK #

Section 7.2, printed pages 48–51, explains why choosing $Q$ and $R$ is a design problem rather than a mechanical step. It records Bryson's initial normalization:

$$Q_{ii}=\frac{1}{x_{i,\max}^2},\qquad R_{jj}=\frac{1}{u_{j,\max}^2}.$$

The report also says plainly that no foolproof weight choice exists and describes using closed-loop eigenvectors to identify which state components dominate an undesirable mode. That is retained as a diagnostic heuristic, not a theorem guaranteeing monotonic pole motion.

The report's distinction between a full-state optimum and a controller constrained to available feedback paths motivates GTK's separate fixed-structure topic. Deleting an unmeasured feedback path from an LQR gain is not generally the solution of the newly constrained optimization problem.

Visual audit and scope #

All 180 physical pages were included in the inspection ledger. The numerous aircraft drawings, pole plots, simulation traces, and listings are not reused: they are specific to a historical STOL configuration and do not improve the general optimal-control explanations. GTK retains only the weight-selection argument above, with exact page provenance, and independently draws any general diagrams.