Updating Simulator Prediction with Flight-Test Experience

Test pilots who "flew” the early simulators were skeptical of the results that they observed, because there was usually some aspect of the sim­ulation that did not match the real airplane. Stick forces and control surface hinge moments were often not properly matched on the sim­ulator, and thus the apparent effectiveness of the ailerons or elevators was often higher or lower than experienced with the airplane. For pro­cedural trainers (used for checking out pilots in new airplanes) mathe­matical models were often changed erroneously based strictly on pilot comments, such as "the airplane rolls faster than the simulator.” Since these early simulators were based strictly on wind tunnel or theoretical aerodynamic predictions and calculated moments of inertia, the flight – test community began to explore methods for measuring and validating the mathematical models to improve the acceptance of simulators as valid tools for analysis and training. Ground procedures and support equip­ment were devised by NASA to measure the moments of inertia of small aircraft and were used for many of the research airplanes flown at DFRC.[725]

A large inertia table was constructed in the Air Force Flight Test Center Weight and Balance facility at Edwards AFB for the purpose of measuring the inertia of large airplanes. Unfortunately, the system was never able to provide accurate results, as fluctuations in temperature and humidity adversely affected the performance of the table’s sensitive bearings, so the concept was discarded.

During the X-15 flight-test program, NASA researchers at Edwards developed several methods for extracting the aerodynamic stability

derivatives from specific flight-test maneuvers. Researchers then com­pared these results with wind tunnel or theoretical predictions and, where necessary, revised the simulator mathematical models to reflect the flight-test-derived information. For the X-15, the predictions were quite good, and only minor simulator corrections were needed to allow flight maneuvers to be replicated quite accurately on the simulator. The most useful of these methods was an automatic computer analy­sis of pulse-type maneuvers, originally referred to as Newton-Raphson Parameter Identification.53,54 This system evolved into a very useful tool subsequently used as an industry standard for identifying the real-world stability and control derivatives during early testing of new aircraft.[726] The resulting updates are usually also transplanted into the final train­ing simulators to provide the pilots with the best possible duplication of the airplanes’ handling qualities. Bookkeeping methods for determin­ing moments of inertia of a new aircraft (i. e., tracking the weight and location of each individual component or structural member during air­craft manufacture) have also been given more attention.

Characteristically, the predicted aerodynamics for a new airplane are often in error for at least a few of the derivatives. These errors are usually a result of either a discrepancy between the wind tunnel model that was tested and the actual airplane that was manufactured, or a result of a misinterpretation or poor interpolation of the wind tunnel data. In some cases, these discrepancies have been significant and have led to major incidents (such as the HL-10 first flight described earlier). Another source of prediction errors for simulation is the prediction of the aeroelastic effects from applied air loads to the structure. These aeroelastic effects are quite complex and difficult to predict for a lim­ber airplane. They usually require flight-test maneuvers to identify or validate the actual handling quality effects of structural deformation. There have been several small, business aircraft that have been built, developed, and sold commercially wherein calculated predictions of the aerodynamics were the primary data source, and very little if any wind tunnel tests were ever accomplished. Accurate simulators for pilot

training have been created by conducting a brief flight test of each air­plane, performing required test maneuvers, then applying the flight-test parameter estimation methods developed by NASA. With a little bit of attention during the flight-test program, a highly accurate mathematical model of a new airplane can be assembled and used to produce excellent simulators, even without wind tunnel data.[727]