KiteModels

Documentation for the package KiteModels.

The models have the following subcomponents, implemented in separate packages:

AtmosphericModel from AtmosphericModels
WinchModel from WinchModels
KitePodModel from KitePodModels
The aerodynamic forces and moments of some of the models are calculated using the package VortexStepMethod

This package is part of Julia Kite Power Tools, which consist of the following packages:

Julia Kite Power Tools

What to install

If you want to run simulations and see the results in 3D, please install the meta package KiteSimulators which contains all other packages. If you are not interested in 3D visualization or control you can just install this package. When you have installed the package KiteSimulators, use the command using KiteSimulators instead of using KiteModels when this is mentioned in the documentation.

Installation

Install Julia 1.10 or later, if you haven't already. On Linux, make sure that Python3 and Matplotlib are installed:

sudo apt install python3-matplotlib

Before installing this software it is suggested to create a new project, for example like this:

mkdir test
cd test
julia --project="."

Then add KiteModels from Julia's package manager, by typing:

using Pkg
pkg"add KiteModels"

at the Julia prompt. You can run the unit tests with the command (careful, can take 60 min):

pkg"test KiteModels"

You can copy the examples to your project with:

using KiteModels
KiteModels.install_examples()

This also adds the extra packages, needed for the examples to the project. Furthermore, it creates a folder data with some example input files. You can now run the examples with the command:

include("examples/menu.jl")

News

Work in progress

a new 5-point model based on ModelingToolkit (MTK) is in development; this will allow to create linearized models around any operation point and to do analysis in the frequency domain.

April 2025

a new model RamAirKite was contributed, based on the package VortexStepMethod

November 2024

the four point kite model KPS4 was extended to include aerodynamic damping of pitch oscillations; for this purpose, the parameters cmq and cord_length must be defined in settings.yaml
the four point kite model KPS4 was extended to include the impact of the deformation of the kite on the turn rate; for this, the parameter smc must be defined in settings.yaml

October 2024

the orientation is now represented with respect to the NED reference frame
azimuth is now calculated in wind reference frame. This allows it to handle changes of the wind direction during flight correctly.
many unit tests added by a new contributor
many tests for model verification added; they can be accessed using the menu2.jl script
the documentation was improved

August 2024

a new kite model, KPS3_3L was contributed. It uses three lines to the ground and three winches for steering a ram-air foil kite.
a first ModelingToolkit based model was added, which shows a much better performance and easier to read code
a new KCU model was added which assumes a linear relationship between the depower settings and the depower angle and thus is easier to configure than the original model.
the drag of the KCU is now taken into account
the drag of the bridle is now taken into account correctly, also if the real kite has more bridle lines than the model
the function to find the initial state is now much more robust

July 2024

a new groundstation / winch type is now supported, the TorqueControlledMachine. It can be configured in the section winch of the settings.yaml file. It uses a set torque as input.
a Python interface is now provided, see: pykitemodels

April 2024

added support for the native Julia DAE solver DFBDF. It is much more accurate and faster than the IDA solver that was used before.

Provides

The type AbstractKiteModel with the implementation KPS3, KPS4 and RamAirKite, representing the one point, the four point kite model and the ram air kite model, together with the high level simulation interface consisting of the functions init_sim! and next_step!. Other kite models can be added inside or outside of this package by implementing the non-generic methods required for an AbstractKiteModel.

Additional functions to provide inputs and outputs of the model on each time step. In particular the constructor SysState can be called once per time step to create a SysState struct for logging or for displaying the state in a viewer. For the KPS3 and KPS4 model, once per time step the residual! function is called as many times as needed to find the solution at the end of the time step. The formulas are based on basic physics and aerodynamics and can be quite simple because a differential algebraic notation is used.

One point model

This model assumes the kite to be a point mass. This is sufficient to model the aerodynamic forces, but the dynamic concerning the turning action of the kite is not realistic. When combined with a controller for the turn rate it can be used to simulate a pumping kite power system with medium accuracy.

Four point model

This model assumes the kite to consist of four-point masses with aerodynamic forces acting on points B, C and D. It reacts much more realistically than the one-point model because it has rotational inertia in every axis.

Four point kite power system model

Ram air kite model

This model represents the kite as a deforming rigid body, with orientation governed by quaternion dynamics. Aerodynamics are computed using the Vortex Step Method. The kite is controlled from the ground via four tethers.

Tether

The tether is modeled as point masses, connected by spring-damper elements. Aerodynamic drag is modeled realistically. When reeling out or in the unstreched length of the spring-damper elements is varied. This does not translate into physics directly, but it avoids adding point masses at run-time, which would be even worse because it would introduce discontinuities. When using Dyneema or similar high-strength materials for the tether the resulting system is very stiff which is a challenge for the solver.

Reference frames and control inputs

a positive set_torque will accelerate the reel-out, a negative set_torque counteract the pulling force of the kite. The unit is [N/m] as seen at the motor/generator axis.
the depower settings are dimensionless and can be between zero and one. A value equal to $\mathrm{depower\_zero}/100$ from the settings.yaml file means that the kite is fully powered.
the heading angle, the direction the nose of the kite is pointing to is positive in clockwise direction when seen from above.
the steering input, dimensionless and in the range of -1.0 .. 1.0. A positive steering input causes a positive turn rate (derivative of the heading).

A definition of the reference frames can be found here .