Have you heard about HAWK, the software that allows instantaneous wind calculation? To know a bit more about it, I interviewed Prof. Dr. sc. techn. Heinrich Meyr, who has developed the Hawk with his PhD student Peng Huang.
L'article en français est disponible ici.
Could you explain how Hawk has been created?
We had developed this software as part of a larger project that has to do with drones. We wanted to be able to model the wind in 3D: the airmass.
For years, the wind is calculated using a GPS position, the speed of the aircraft and a compass that gives the “attitude” of the aircraft. That option wasn’t available to us as electronic compasses are extremely sensitive to electromagnetic disturbances and error in the IGRF. This can lead to large errors in the wind calculation. Also, all algorithms used to today require long averaging times. They do not deliver “live wind”.
Being a glider pilot, I knew about different options with sensors but none of them seems to be both reliable and precise enough for what we needed to do: live wind. So, we decided to use an algorithm, based on an extended Kalman filter (EKF). I used this algorithm when working for the space program. It is the same fundamental algorithm that was used to land Perseverance on Mars. This algorithm is using GPS data, the total and static pressure, 3-dimensional gyro and acceleration sensor. With this information, and based on the successive triangle strategy, it is possible to estimate precisely the “attitude” of the glider and consequently, the air mass movement in 3D. In addition, the artificial horizon is included in the software, as we need to generate this information during the calculation.
It was challenging to prove mathematically to the scientific community that you don’t need a compass. We tested the algorithm with 6 pilots in 6 different planes for a total of around 100 flights. But we also record all the information from these flights, from all the sensors and use them in simulation to analyse and improve our algorithm on the ground. Without this simulation, it would not have been possible to improve the algorithm as much as we wanted. Indeed, all the variables used are interlinked.
Instantaneous measure of the wind during a flight of Klaus Ohlmann from Seynes with a weak north wind. You can see the wind variation from the Durance valley (on the right) to Serre. The instant wind information allows you to visualise local wind phenomena.
With the HAWK, you can have the instant wind, but also a better variometric information?
Yes, as you know total energy compensated varios give false reading due to acceleration and horizontal wind burst.. It is a lot of intuition to compensate for that. After, it is ok, in the thermal, the classic information is good. Our system is not based on energy conservation so we don’t have this problem.
Our system shows you the air mass movement. We have a real netto, not simply the value of the variometer minus the sink of the glider according to the polar. We had a pilot, a very good pilot, flying a lot in competition and as he tested the system, I could analyse the data on the ground and I told him: “look, your thermalling is not optimum.” and we showed him the “two needles''. With that you can see if the difference between the air mass movement and your glider vertical movement is too large, it means that you can fly your plane better. He compared when he flew symmetric versus when he had the string outside and concluded that you are not climbing well when you fly asymmetrical at 12-15 degrees. It was really interesting to record and be able to see everything of what was happening during the flight. It is in the manual; we call it the two needles approach. If pilots find that our vario is too optimistic, it may be that they can improve the symmetry of the glider.
Comparison of the HAWK and the classic total energy variometer.
Any other advice for pilots before they are starting their free trial?
We made it as easy as possible. All these parameters we are using can be confusing, mainly when flying. So, they are all integrated. The only things that the pilots need to choose is the integration time. Indeed, if the HAWK is really “nervous” making data difficult to interpret, so we are using a dampening system: the SIGWIND that smooths the data mathematically. That parameter is configurable by each pilot to decide how nervous vario and wind reading will be.
Example of the wind speed readings over time (in second) using four different values of SIGWIND
It is not dependent on the TE probe, that we know can be leaky. Any idea on how Hawk will react on different installations?
We are not using the compensation probe. We are using the total pressure. For now, and using 6 different gliders we found their installation allows very reliable measure. Obviously if the glider has leak, it will not work: “Garbage in”, “garbage out”, but for now, we didn’t have problems on these gliders.
Why this start by implementing on the variometer LXNAV S10 and S100 and not on the main LX90xx gliding computers?
It was very easy to implement Hawk into the S10 and S100 as they have all the sensors that we needed in one place. It will be done in a few months for the LXxxxx line.
Thanks a lot for your time
You can find here, an article more technical (in German) published by Prof. Heinrich Meyr in Segeliegen Magazine.
You can find the Hawk here.
You can try the HAWK for a month for free. Click here.