Tolerances on control inputs for testing


Because of unavoidable differences between the sim tests and the real world (gusts, turbulence… that cannot be accurately characterized in the reference flight test data) an exact replication of all control inputs usually tends toward divergent behaviour even in a model otherwise perceived as fully representative by the pilots. Which tolerances/deviations might be acceptable for those control inputs not being the direct object of the validation test performed (i.e. lateral/directional for a longitudinal step response check)?


It might be because the A/C data is not detailed enough, meaning you either have not recorded enough parameters or the parameters do not have a high enough resolution to see exactly what the A/C is doing. Regardless, if the sim inputs are accurately following the A/C and the model is good, then the sim outputs should follow within tolerances. However, there are often differences between simulator tests and real world data due to wind velocity etc. as you state. Even with the correct control inputs the response can be divergent in long period tests. One testing method to overcome this is to use a closed loop feedback system to maintain the profile; deviations outside the tolerance parameters can be axplained as a note to the test result. See Vol I of RAeS Handbook for examples of these conditions & the explanations provided for the differences. The amount of acceptable deviation on the inputs we are afraid is down to "good engineering judgement". Any systematic error across multiple tests should be clearly explained (or justified) in a rationale. (We assume that you are discussing the parameters that are those mandated for the test in the Table of Validation and not those additional parameters listed in the RAeS Handbook, which are listed to assist in any fault analysis, and in which deviations might be acceptable).