Aerodynamic models

FlapSim's aerodynamics are additive per-wing models evaluated by one engine (flapsim.aero.AeroEngine). The engine computes each wing's relative-flow kinematics once per step, asks every model for its loads (a force at a point plus a pure moment, all in the wing aero frame), multiplies the sum by the modifiers' factors and writes the wrench to MuJoCo. Plugins expose the engine: aero/quasi_steady is the single-model stack, aero/stack composes models.

- type: aero/stack
  models:
    - quasi_steady
    - rotational_lift
    - {type: added_mass, include_rotational_inertia: true}
    - {type: wake_capture, C_wc: 0.8}
  modifiers:
    - {type: ground_effect, model: exponential, gain: 0.3}

Signals: aero/F_world, aero/lift_total, aero/rho, per wing wings/<name>/F_N (total normal force), alpha, U_i, U_o, F_world, torque_world, cop_world, and per model wings/<name>/<model>/<key>.

Models

Model	Physics	Parameters	Reference
quasi_steady	Translational normal force 1/2 rho c C_N(alpha) (U^2 integrals) at the centre of pressure, pitching moment through the chordwise CoP, rotational damping M_rd	options (AeroOptions overrides)	Whitney & Wood 2010; Dickinson et al. 1999 fits
rotational_lift	C_rot rho theta_dot integral c^2 |U| dr (flow-speed magnitude, so the force flips between strokes), positive when pitching up into the flow, at the half chord	C_rot (default pi (0.75 - x0)), point_fraction	Sane & Dickinson 2002
added_mass	Explicit force integral (rho pi c^2 / 4) dU_n/dt dr from the prescribed stroke and body accelerations; the added rotational inertia about the feathering hinge is added to the joint armature (implicit, stable for light wings)	include_feathering_inertia	Sedov; Sane & Dickinson 2002
wake_capture	Empirical transient C_wc 1/2 rho S U_rev^2 exp(-t/tau) after each stroke reversal	C_wc, tau_chords, point_fraction, min_speed	Dickinson et al. 1999 (phenomenon)

Modifiers

Modifier	Factor	Parameters
ground_effect	1 + gain exp(-h / (length_scale R)) or Cheeseman-Bennett 1 / (1 - (R / 4h)^2)	model, gain, length_scale, ground_height, max_factor

Related plugins

• atmosphere/standard: density (ISA at altitude_m or rho), steady wind, a gust (one_minus_cos, sine, step) and Ornstein-Uhlenbeck turbulence. Publishes atmosphere/rho and atmosphere/wind_world, which the aero stack and body drag read.
• force/body_drag: -1/2 rho C_d A |v| v per body axis relative to the wind.
• aero/cycle_averaged: rigid-body hover model, lift = k f^2 along the gimbal-tilted thrust axis plus control moments and damping. Parameters come from flapsim sweep (flapsim.aero.identify), which runs the fixed-base per-step model as a virtual wind tunnel: frequency sweep for k, gimbal deflections for thrust tilt and moments per radian, imposed body velocities and rates for damping.

Writing a model

from flapsim.aero import AeroModel, WingLoads, register_model
import numpy as np


@register_model
class MyModel(AeroModel):
    name = "my_model"

    def compute(self, wing, state, kin, flow, options):
        force_n = ...  # along the wing normal
        point = np.array([0.0, state.side * kin["span_CoP"], -0.5 * wing.mean_chord])
        return WingLoads(
            self.name, np.array([force_n, 0, 0]), point, np.zeros(3), {"k": 1.0}
        )

kin holds the shared kinematics (U_i1, U_i0, U_o1, U_o0, alpha, span_CoP, theta_dot, ...), flow the air density, wind and torso velocity, wing.params the planform (R, c, integrals, hinge geometry). Models shipped in another package register through the same decorator and become available by name in aero/stack configs.