Handling Qualities in Roll: Simulator – CHR-transform method – Flight

By

Björn Kullberg

Saab Aerospace

Abstract

Fixed-base and in-flight experiments are conducted to estimate a correlation between on-ground and in-flight roll handling qualities for different control system configurations and maneuvers, and to verify TsAGI-developed method used for transforming pilot ratings from fixed based simulators to real flight. It is shown that roll fixed-based Cooper-Harper ratings (CHRs) considerably differs from those obtained in flight, and the method tested allows an adequate predicting real flight CHRs.

Introduction

One of the most expensive and risky activities when developing a new aircraft is developing of pilot handling qualities. The existing handling qualities (HQ) criteria are often ineffective for the modern high-augmented aircraft. Thus ground-based simulator experiments play a significant role in aircraft HQ selecting. Most of such experiments are conducted on fixed-based simulators, at least for fighters. Since fixed-base simulators do not reproduce motion cues, the question about the correlation between simulator and in-flight results arises.

Comparative fixed-base and in-flight experiments are rather expensive and, thus, rare. There is known the only carefully study described in [1], which compares fixed-base and in-flight CHRs in roll. It demonstrates that simulator results considerably differ from in-flight ones. It is evident that no ground-based simulation can provide an exact similarity on acceleration effects, as well as on other flight factors – visual cues, manipulator feel system, etc. Thus a question arises, is the discovered difference between simulator and in-flight results regular or accidental? In other words, will the difference remain in other comparative experiments with other simulator and aircraft characteristics? One of the goal of the present work was to verify a regular character of the difference between on-ground and in-flight roll CHRs shown in [1].

The data available in [1] and obtained in the course of the present study, show that the degree of the difference between simulator and in-flight CHRs is considerable and in a complicated way depends on aircraft characteristics. Depending on the aircraft characteristics, simulator CHRs can be better or worse than in-flight CHRs. Experiments alone can not be sufficient to predict the CHR difference for all possible aircraft characteristics. In order to reduce costs and risks connected to flight tests, it would be helpful to have a calculation method for transforming pilot ratings from fixed based simulators to real flight. Such a method has recently been developed at TsAGI (we call it the CHR-transform method) and described in [2]. Thus the second goal of the present study was to verify the CHR-transform method.

1. Description of experiments

Aircraft. Swedish Airforce, Air Defence Fighter JA37 Viggen. This aircraft was chosen for the evaluation of the CHR-transform because it’s an operational and highly augmented aircraft with well-known handling qualities.

Figur 1. Test Aircraft JA 37-301 used in the test of the CHR-Transform. Now retired and standing in front of the Swedish Airforce Museum.

Fixed-based simulator. The simulator is an ordinary fixed based development simulator for the JA37 Viggen. The cockpit contains the real hardware and the forces in the control stick represents the real stick forces. The visual scene has a delay for about 80 ms which is acceptable for this kind of testing.

Figur 2. The JA 37 development simulator used in the test of the CHR-transform. The simulator is located at Saab Aerospace Simulator Centre.

Three types of lateral tests (see table 1) where created for determination of the pilots ratings. These test types where then flown with different modes in the control system and at various speeds in order to be able to obtain the function CHR=f(control system gain) for different test types and pilots. Those functions where obtained both from the simulator and from flight tests and the CHR-transform where applied to pilot ratings obtained in the simulator and thereafter compared with the pilot ratings obtaioned from the flight tests.

A more detailed description of test type design and results can be found in reference [3].

Tabel 1. Test Matrix

Test Types	Altitude	Speed/M	C.S Mode	OP
Large Bank Angle Capture	4 km	0.7	Normal	CO
Large Bank Angle Capture	4 km	0.7	Aiming	CO
Large Bank Angle Capture	4 km	0.7	Back Up	CO
Small Bank Angles Capture	4 km	0.7	Normal	CO
Small Bank Angles Capture	4 km	0.7	Aiming	CO
Small Bank Angles Capture	4 km	0.7	Back Up	CO
Small Bank Angles Capture	4 km	340	Normal	CO
Small Bank Angles Capture	4 km	340	Aiming	CO
Small Bank Angles Capture	4 km	400	Back Up	PAL
Formation Flight	4 km	0.7	Normal	CO
Formation Flight	4 km	0.7	Aiming	CO
Formation Flight	4 km	0.7	Back Up	CO
Formation Flight	4 km	340	Back Up	CO
Formation Flight	4 km	340	Back Up	CO
Formation Flight	4 km	400	Back Up	PAL

2. CHR-transform method and its verification.

It is well known that high g-loads and persistent turbulence accelerations can considerably worsen pilot performance. The reason for this is obvious since the pilot’s physiological state is changing. The CHR-transform method deals with small accelerations created by a pilot while controlling an aircraft in common flight modes (take-off, landing) at calm atmosphere. Such small accelerations also affect the pilots’ performance but these effects are less understood. The investigations at TsAGI [2] showed that in some cases the acceleration was beneficial and improved the Cooper-Harper ratings, but was negative in other. Thus the acceleration effect is bilateral and may give additional cues to control aircraft or its biodynamic effects may be felt as negative by a pilot. The change in pilot rating can be summarized as:

PR_flight – PR_fixed-base =,

where PR_flight and PR_fixed-base are Cooper-Harper ratings obtained, correspondingly, in flight and in fixed-base simulations, DPR ^– and DPR ⁺ are pilot rating increments due to, correspondingly, negative and beneficial acceleration effects.

Beneficial acceleration effect.

Beneficial acceleration effect is determined in accordance with the following expression:

where w_BW (bandwidth frequency) is determined from the phase of the roll rate transfer function :

Negative acceleration effect.

Negative acceleration effect is determined from the expression:

or from the curve in fig.1 (fig.5, indexes 'calc' are omitted).

Parameter l depends on aircraft characteristics. In case roll transfer function is defined as:

parameter l can be determined from the formula

where h is a pilot position relative to the rotation axis, w_c is a pilot-aircraft system cutoff frequency at an aircraft characteristics gain K_p*, T is a prefilter time constant, t_R is a roll mode time constant.

In case roll transfer function is a more complicated, parameter l can be determined from the following formula:

, (1)

where

Verification of the CHR-transform method.

Beneficial acceleration effect DPR⁺ was calculated with the help of MATLAB.

The beneficial effects is divided into high or low speed and large or small stick inputs. The large stick inputs represent large bank angle capture at both high and low speed. The small bank angles capture as well as formation flight are characterized by small stick inputs at corresponding speeds.

It turned out that calculation of the negative acceleration effects accordingly to formula (1) presents difficulties due to difficulties in determining amplitude/frequency characteristics of roll rate transfer function at high frequencies. The point is that the JA37 control system contains nonlinearities and, thus, we need to know the input amplitude and frequency to adequately determine . Although there is no recommendation in [2] on the matter but the following should be mention. The negative acceleration effect (turbulence is absent) is possible at certain aircraft characteristics. The negative acceleration effects, if any, are discovered and eliminated during aircraft flight tests. Since the JA37 fighter has been operated for such a long time, it was not likely that we would find any negative effects. For these reasons no negative effects were added to the ratings from the simulator (DPR ^– = 0).

The final results were calculated for each individual pilot as well as the average values. The average values of all roll tasks are presented in fig.3.

In 11 of 15 cases the difference between calculated and experimental CHRs is less than 0.5, the maximal difference does not increase 1.5. This difference can be explained by the pilot rating random character, since for each configuration only two ratings were used for averaging.

Fig.3 presents CHRs averaged over all roll maneuvers, i.e. over thirty CHRs. It is seen that the difference between calculated and in-flight CHRs is 0.08 in this case.

Figur 3. Average data for all maneuvers and pilots. Note how well transformed simulator data corresponds to flight test data.

Conclusions.

The experiments conducted on fixed-base simulator and repeated on JA37 Viggen have shown that roll pilot ratings obtained in flight considerably differ from those obtained on ground.

Fixed-base CHRs were transformed in accordance with the method developed in TsAGI and then compared with in-flight CHRs. It turned out that calculated CHRs almost coincide with the experimental CHRs. It is recommended that the method should be used in future designs of lateral flying qualities.

Further work should be made to give recommendations on input amplitudes and frequencies to adequately determine amplitude/frequency characteristics in (1).

References.

1. J.R.Wood. “Comparison of Fixed-Base and In-Flight Simulation Results for Lateral High Order Systems”. AIAA-83-2105, August 15-17, 1983, Gatlinburg, Tennessee.

2. White A.D., Rodchenko V.V. “Motion Fidelity Criteria Based on Human Perception and Performance”. AIAA-99-4330, Portland, August, 1999.

3. Magnus Tormalm, CHR Transform Method – Final Report. Saab, February 16