Flying Qualities Research Moves with the Times

Robert R. Gilruth’s key flying qualities contribution was to test a significant sample of airplanes for some flight characteristic such as lateral control power and then to separate the satisfactory and unsatisfactory cases by some parameter that could be calculated in an airplane’s preliminary design stage. The Gilruth method put design for flying qualities on a rational basis, although Chapter 3 tells of some later backsliding, attempts to specify flying qualities parameters arbitrarily.

Modern times have brought the $100 million and more airplane and development costs for new prototypes into the billions of dollars. This has made for a scarcity of new machines that can be tested in the Gilruth manner and an interest in alternate flying qualities methods. The pilot-in-the-loop method surfaced around 1960 as an alternate way of rationalizing flying qualities and to focus attention on the pilot-aircraft combination as a closed-loop system. Pilot-in-the-loop analysis involves adoption of mathematical models for the human pilot as just another control system element.

The three basic concepts of the pilot-in-the-loop analysis method are (McRuer, 1973):

1. To accomplish guidance and control functions the human pilot sets up a variety of closed loops around the airplane, which, by itself, could not otherwise accomplish such tasks.

2. To be satisfactory, these closed loops must behave in a suitable fashion. As the adaptive means to accomplish this end, the pilot must make up for any dynamic deficiencies by adjustments of his own dynamic properties.

3. There is a cost to this pilot adjustment: in workload stress, in concentration of the pilot’s faculties, and in reduced potential for coping with the unexpected. The measure of the cost are pilot commentary and pilot rating, as well as physical and psychological measures.

In this chapter we trace the development of pilot-in-the-loop analysis methods as they apply to airplane flying qualities. Pilot-in-the-loop methods are clearly essential to study closed-loop operations such as tracking, but can they replace or add to the classical Gilruth approach?