Systems Control and Flight Dynamics
SACSO,
Active Suspension for Wind Tunnel Tests
Design Stages
Verifications in a wind tunnel
To study the effect of the cables on the aerodynamic flows, several series of tests (figures 2 and 3) were carried out in the L1 wind tunnel at the Onera's Lille center. These tests verified that the effects of the presence of cables on the aerodynamics of a Mirage 2000 model were compatible with the objectives pursued.
Figures 2 and 3: evaluation of the effects of the cables in the wind-tunnel
Feasibility of simulated free flight
The theoretical studies showed the feasibility of coordinated control of this suspension for simulating free flight. This control is implemented using a coefficient of tension for the required force generated by the cables, for example for simulating the action of the engines. This tension coefficient of forces is also used to modify the direction or amplitude of the weight, and to modify the model's mass or inertia by generating the corresponding inertial forces. Our studies showed that the hard point was not coordinating this redundant assembly, which is simply a well-mastered numerical algorithm problem, but obtaining a high bandwidth for the cables' tension mechanisms, which required advanced know-how about systems, whose complex physical reality is very badly represented by models (hysteresis in the rubbing of the cables on the pulleys, for example).
The design of the mechanical structure
The design studies resulted in several publications ([1], [2] and [3]) and the engineering doctoral thesis of P. Lafourcade [4]. They defined several mechanical architectures, including:
- an architecture based on seven cables with a uniform distribution of anchor points (figure 4)
- an architecture based on nine cables (figure 5) offering a larger volume of motion, better uncoupling of the commands, and requiring less powerful motorization.
This latter architecture was selected.
Figure 4: architecture with 7 cables
Figure 5: architecture with 9 cables
Cable tension control mechanism design
Several test benches (figures 6 and 7) were created for testing the selected design solutions. Finally, to increase the performance of these control mechanisms and reach the current bandwidth (about 35 Hz), we had to model and identify the non-linear phenomena due to the rubbing of the cables on the pulleys.
Figures 6 et 7: test benches of control mechanisms of the tension of cables
Control algorithm design
The algorithms used to control this redundant architecture are the fruit of our experience in robot control. It is a position-force hybrid coordinated control structure. This offers a choice of several types of controls:
- Follow six-DOF trajectories imposed on the model. This is the pure control of position where the cable force tension coefficient is calculated to eliminate the error between the trajectory created and the trajectory required.
- Leave the model in six-DOF free flight. This is the pure control of force where the cable tension coefficient is calculated to simulate the thrust of the engines, and may also compensate for faults in similitude in mass and inertia, but also changes the direction of gravity (which is often the case for the SV4 vertical wind tunnel)
- Impose motion or immobility on (or around) certain axes and let the model move freely on (or around) the others. This is hybrid position-force control in which some components of the tension coefficient implemented by the cables come from position control and others from force control.
These various control methods all result in a desired tension coefficient that the cables must apply. A specific quadratic programming algorithm calculates the tension coefficient for the tensions of the nine cables with a minimum value constraint of 20 Newtons and maximum value constraints of 100 to 200 Newtons, depending on the cables.
The Cartesian coordinated control loops in position and/or in force and the cable tension control mechanism loops are synchronous and activated with the same sampling frequency of 2 kHz.
