Ducted fan
- 6 state
- 2 control
- 1 variable
- Bolza cost
- free final time
- 1 state box
- 2 control box
- 1 variable box
- 12 boundary
- OptimalControl valid
- OptimalControl, scalarized valid
- JuMP valid
The ducted fan problem is a classical nonlinear benchmark in optimal control with multiple input and state constraints. It models the planar motion of a ducted fan aircraft, described by its horizontal and vertical positions
The objective is to steer the fan from the origin to a horizontal position of
Mathematical formulation
with
Qualitative behaviour
The free end-time formulation yields a time–energy trade-off, governed by
: large emphasises short transfer time, small emphasises reduced input effort. The thrust constraints induce saturation, often resulting in bang–bang control profiles in
and . The angular bound
limits manoeuvrability, shaping feasible trajectories.Flatness-based analysis shows that the system is differentially flat, with outputs that allow explicit trajectory design (Graichen & Petit, 2009).
Characteristics
Nonlinear dynamics.
Benchmark for testing constrained nonlinear OCP solvers.
References
Graichen, K., & Petit, N. (2009). Incorporating a class of constraints into the dynamics of optimal control problems. Optimal Control Applications and Methods. DOI: 10.1002/oca.880 This work presents a methodology for incorporating constraints into system dynamics, exemplified on ducted fan trajectory optimisation.
Yu, J., & Babai, J. A. (2001). Comparison of Nonlinear Control Design Techniques on a Planar Model of a Ducted Fan Engine. Automatica, 37(5), 741–748. Demonstrates practical nonlinear control designs on a planar ducted fan model.
Ohanian, O. J. (2011). Ducted Fan Aerodynamics and Modelling, with Applications of Flow Control. Provides detailed modelling of ducted fan aerodynamics and control-relevant effects.
Parameters
| Parameter | Default value | Unit | Meaning |
|---|---|---|---|
t0 | 0 | s | initial time |
r | 0.2 | m | distance from the centre of mass to the point where the force u₁ applies |
J | 0.05 | kg·m² | moment of inertia |
m | 2.2 | kg | mass |
mg | 4 | N | weight |
μ | 1000 | – | weight of the final time in the cost |
α_l | -0.5235987755982988 | rad | lower bound of the angle α |
α_u | 0.5235987755982988 | rad | upper bound of the angle α |
u₁_l | -5 | N | lower bound of the force u₁ |
u₁_u | 5 | N | upper bound of the force u₁ |
u₂_l | 0 | N | lower bound of the force u₂ |
u₂_u | 17 | N | upper bound of the force u₂ |
tf_l | 0.1 | s | lower bound of the final time |
x₁_t0 | 0 | m | initial horizontal position |
v₁_t0 | 0 | m/s | initial horizontal velocity |
x₂_t0 | 0 | m | initial vertical position |
v₂_t0 | 0 | m/s | initial vertical velocity |
α_t0 | 0 | rad | initial angle |
vα_t0 | 0 | rad/s | initial angular velocity |
x₁_tf | 1 | m | final horizontal position |
v₁_tf | 0 | m/s | final horizontal velocity |
x₂_tf | 0 | m | final vertical position |
v₂_tf | 0 | m/s | final vertical velocity |
α_tf | 0 | rad | final angle |
vα_tf | 0 | rad/s | final angular velocity |
Every modeling takes other values with the keyword parameters, merged with these: ducted_fan(backend; parameters=(t0=…,)).
Reference objective
The best objective among the solves that converged, in each modeling, on the default grid of the modeling: the modelings are independent, each has its own reference. See how the references are computed.
| Modeling | Scheme | Grid | Reference objective | Solves | Distinct solutions | Configurations agree |
|---|---|---|---|---|---|---|
| JuMP | :trapeze | 250 | 1831.827858 | 8 | 1 | yes, within 7.5e-08 |
| OptimalControl | :trapeze | 250 | 1831.827858 | 18 | 1 | yes, within 7.7e-08 |
| OptimalControl, scalarized | :trapeze | 250 | 1831.827858 | 3 | 1 | yes, within 6.2e-16 |