Folders and files

You can now run the simulation. Jump here to see how, or continue reading to learn about the airframe and the relevant reference frames first.

The simulated airframe has a twin-boom fuselage and a wing with upward cranked tips. The total wing span is 1.5m and the take-off weight is 1.56kg (actual glider equiped with on-board computer and temporarily installed electric motor for testing / take-off). Center of gravity has been found to be at 92mm from the leading edge of the main wing. Via GPS measurements a gliding velocity of about 45km/h was confirmed (at roughly zero elevator deflection). The glider uses two actuators: elevator and rudder. The rudder is asymmetrically attached to the left of the two vertical stabilizers.

Below is the airframe as defined for the vortex lattice method computation with Tornado:

The airfoil JR001 features a planar pressure side which simplifies the build procedure for the wing. The profile was designed to work well with low Reynold's numbers and to provide friendly stall characteristics. It wasn't designed with gliding performance in mind.

Further drawings related to the airframe can be found here and here. The Tornado definition of the airframe is here.

When specifying forces, moments, or angles a body-fixed reference frame is used. The usual convention is shown in the figure below on the left. This is the convention as introduced in [2] and also as used in Matlab. The Tornado implementation [1] uses a slightly different reference frame, see below on the right.

Another reference frame which is sometimes used is a set of axes for which the x-axis is parallel to the velocity vector for an equilibrium state (e.g. trimmed gliding). Such axes are called stability axes. Choosing the principal axes in this way simplifies some equations when computing longitudinal and lateral linear systems from given aerodynamic coefficients (see also [2] pp. 45). But since the solution implemented here, finds the corresponding LTI systems by linearizing a non-linear model around an equilibrium state, this is not really an advantage.

The stability axes reference frame is not used in this implementation.

The codebase and the Simulink models can be used to:

1. Compute aerodynamic properties and coefficients using the Tornado VLM implementation.
2. Run the non-linear flight simulation using previously computed coefficient matrices.
3. Extract the linear time-invariant systems for the trimmed gliding state.

Aerodynamic coefficient matrices are computed by calling the script:

Note, the Simulink models contain pre-computed coefficient matrices, and the result of above script is added here. So, you don't need to re-compute the coefficients if you just want to to run the simulation or compute the LTIs. The resulting matrices are visualized here.

Input to compute the coefficients is:

1. Tornado airframe definition (geometry and airfoil),
2. center of gravity position
3. air speed [m/s]
4. air density [kg/m^3]
5. range of alpha and beta values for which coefficients shall be computed (alpha denotes the angle of attack, and beta the sideslip angle).

Output is for each coefficient a 2-dimensional matrix which contains the coefficient's value for all specified [alpha, beta] configurations. Computed coefficient matrices can be found here. The coefficient naming convention is summarized below.

• CX: force coefficient in body-fixed longitudinal direction.
• CY: force coefficient in body-fixed lateral direction.
• CZ: force coefficient perpendicular to CX, CY (body-fixed 'lift').
• Cl: roll moment coefficient.
• Cm: pitch moment coefficient.
• Cn: yaw moment coefficient.

For each force and moment coefficient, a damping coefficient is computed. These coefficients specify how each coefficient changes when the aircraft rolls (P), pitches (Q), or yaws (R). The resulting matrices are CX_P, CX_Q, CX_R, CY_P, CY_Q, CY_R, CZ_P, CZ_Q, CZ_R, Cl_P, Cl_Q, Cl_R, Cm_P, Cm_Q, Cm_R, Cn_P, Cn_Q, Cn_R.

For each control surface, coefficient derivatives *_d are computed. They denote how much the respective coefficient changes when the respective surface is deflected. The resulting matrices CX_d, CY_d, CZ_d, Cl_d, Cm_d, Cn_d are 3-dimensional where the index of the third dimension denotes the control surface.

Tornado can also be used to estimate the neutral point of the entire airframe. The neutral point is the point on the vehicle's x-axis where the aerodynamic moment Cm remains constant independently of the angle of attack. The distance between the center of gravity and the neutral point is called the stability margin. In a classic fixed-wing configuration, the center of gravity has to be placed before the neutral point (in flight direction). If the the stability margin approaches zero, the airplane becomes unstable. For the considered airframe, Tornado calculated the neutral point to lie at 49% MAC (mean aerodynamic chord), which is for the given wing geometry simply 0.49*0.25[m].

Open the ExperimentalCarrierSimulink.prj located here. This will start FlightGear and open a couple of Simulink models. Select and run the Plant model. Switch to the FlightGear window and use v to change the view perspective.

To control the glider the elevator and rudder surface deflection can be adjusted with two sliders as shown below.

To find the trimmed gliding state and calculate the lateral and longitudinal linear systems, run the script:

Results are shown here.

The resulting linear systems correspond to the ones derived in [2] (see Sect. 5.2 and Sect. 5.3), except that they are calculated for the coordinate system of the standard body-fixed reference frame and not in the reference frame of the equilibrium axes.

Datum coefficients (forces and moments).

Elevator control surface derivatives. Describe how datum coefficients change when the elevator is deflected.

Rudder control surface derivatives. Describe how datum coefficients change when the rudder is deflected.

Asymmetric values appear due to the asymmetry of the vertical stabilizers (only the stabilizer on the port side carries the rudder control surface).

For remaining coefficients, see here.

Output of mainComputeLTIs.m:

Longitudinal system:

Lateral system:

Short period and phugoid modes (poles) of the longitudinal system. Dirac and step response to elevator deflection. A negative pitch moment curve for increasing alpha values (angle of attack) confirms the glider's positive stability margin.

Spiral, rolling, and dutch role modes (poles) of the lateral system. Dirac and step response to rudder deflection.

[1] Melin, Tomas. Tornado, a vortex lattice MATLAB implementation for Linear Aerodynamic Wing applications, Masters thesis, Royal Institute of Technology (KTH), Sweden, December 2000.
[2] Caughey, David A. Introduction to Aircraft Stability and Control, Course Notes for Mechanical & Aerospace Engineering, Cornell University, New York, USA, 2011.