Automatic control for airborne wind energy systems

We develop new observation and control algorithms for airborne wind energy systems – from control theory to on-site experiments

Let’s start with a classical wind turbine (what’s currently the best at converting wind into electricity) and make it better.

It’s the tip of the blade that creates the most mechanical energy (because it spins the fastest).

Let’s keep that and delete the rest.

Well, we’ve deleted a lot of fiberglass. Nice.

Let’s place the generator on the ground. Easier maintainance.

We might as well delete the steel mast and the concrete foundation.

Well, there isn’t much left. We need to change the blade design now. And add a tether.

Some use a glider (because its efficient, durable and the airodynamics are well known)

Others use a kite (because it can takeoff at lighter winds and is cheaper)

Instead of turning, we follow figures of eight, to prevent tangling of the tethers

As the kite flies, it pulls on the tether and unwinds it from the drum

Distance times force is work, converted to electricity by the generator. First phase.

When the tether is unwound, transition to the slack zone. Force goes down. Second phase.

When the tether is slack, wind it back.

Some energy is consumed. Third phase. Start again!

Design of a low-tech ground station

We design a new ground station for a small-scale, ground-gen, soft wing airborne wind energy system.

Low-Tech

Just what’s needed : only two motors and inverters for the whole system.

Low wind takeoff

All actuators are on ground: only sensors remain on board, making the airborne system lighter, and allowing for takeoff at low wind speeds.

Versatile

The ground station turns into a bicycle trailer to be towed to the test site.

Realistic testing

Designed to produce 1.5 kW, the systems will embed batteries and a discharge resistor acting as a load, as well as a spooling system able to stow 150m of tether.

Standard components

Each motor is driven by a STM32 Nucleo board, and the central computer is a Raspberry Pi.

Experimental data

All sensor data will be automatically stored as timeseries on a USB memory stick using CSV files. This will allow for a fast post-processing.

Advanced automatic control

Using Lie group theory, we design precise tether and kite models to improve estimation and control algorithms for AWE systems

Lagrangian dynamics on manifolds

Global model for high fidelity
No sigularities
Compact equations and geometric properties

Nonlinear control

Robustness to out-of-nominal conditions
Exploiting geometric properties of models
Formal guarantees of stability

Detailed models

Exploiting static and dynamic properties of Dyneema
Experimental validation upcoming