Jan 12, 2010, 01:54 PM
Registered User
Data

# F3J launch data

Hello everyone,
I've been using an Eagle Tree FDR aboard my Pike Perfect for about two years, I've been collecting a large amount of data I'd like to share with you.
My primary focus for this thread will be how to use the logged data to analyze the mechanics of the launch of an F3J model.
First of all, a little bit of technical background. The model is a Pike Perfect, C68 version. Wingspan 3.55 m, weight about 2200 grams. The data recorder is a unit from Eagle Tree, the "Glide" FDR. I managed to install it in the Pike Perfect fuselage by removing the plastic case and wrapping it in two layers of heat shrink tubing, it's a snug fit but it's ok. I made a combined pitot-static probe with some spare brass and steel tubing, and I installed it on a spare canopy. My FDR has no accessible static port for the airspeed sensor, so my choice was to leave the static port from the probe open in the space under the canopy: it's far from ideal, but I hope it can help in equalizing the internal pressure with the external static pressure.

To begin with, I am attaching a chart plotting airspeed and altitude versus time during a recent launch. Last Sunday I went flying at my local field. I used an electric winch (ok, not a real F3J launch), the one we use for F3J practice in Italy, with a 0.8 kW starter motor. Turnaround was at 100 m from the winch (it's a small field and it's a good training practice), there was a light breeze at 1.5-2 m/s, temperature was about 3 C.

More data analysis to follow.

Francesco

CORRECTION: after posting I noticed that the airspeed scale claims to be in km/h, but it's really in m/s. I will correct in future postings.

### Images

Last edited by Francesco; Jan 12, 2010 at 02:08 PM.
 Jan 12, 2010, 03:56 PM Registered User Wow, this is interesting But we want to know the extremes! Next I would like to see data from the 230 meter + launch, your Perfect is capable. With Alti2 I have logged taller than 125m zooms from and F3B winch (150m to turnaround) and 100m from 2 man F3J tow. This was in better conditions, though. Anyway, I want to know how much over 200km/h the speed is at moment of release
Jan 12, 2010, 03:57 PM
Registered User
Quote:
 Originally Posted by Tuomo Wow, this is interesting But we want to know the extremes! Next I would like to see data from the 230 meter + launch, your Perfect is capable. With Alti2 I have logged taller than 125m zooms from and F3B winch (150m to turnaround) and 100m from 2 man F3J tow. This was in better conditions, though. Anyway, I want to know how much over 200km/h the speed is at moment of release
That's really good advice, Tuomo. But I need to put my winch on steroids, or wait for a couple of towmen.
For now, I will stick with the low-power winch and short lines (the reason is compliance with the rules of the Italian winter league).

Anyway, rest assured that I did not put up the data to boast my launch technique or performance (quite the opposite, in fact

What is also interesting in having a data logger on board is that with airspeed and altitude data you can start making more educated guesses about your plane performance. For instance, you can integrate the data point-by-point and compute an estimate of the actual flight path (i.e. path of the model in a vertical plane, altitude versus position and not versus time). Playing a little bit with Excel, you get the kind of information you can see in the attached chart. The magenta curve is the plane position (the origin being the release point), the blue curve is the distance from the model to the turnaround at each x position. Scale is in metres.

### Images

Last edited by Francesco; Jan 12, 2010 at 04:06 PM.
 Jan 12, 2010, 04:27 PM Registered User I see, a very clever graph. We could use that method to estimate how efficiently the plane climbs in lauch. But still... I remember the "winch"
Jan 12, 2010, 04:28 PM
Registered User
Quote:
 Originally Posted by Tuomo I see, a very clever graph. We could use that method to estimate how efficiently the plane climbs in lauch. But still... I remember the "winch"
Oh yes, I forgot about it! you had the rare opportunity to experience our winch! Sorry about that...
 Jan 12, 2010, 04:38 PM Registered User No sorry about anything, your winch was fun... ...but also different. Anyway, I am convinced that your method could a help in trimming (and undestanding!) F3J launch. Keep on going! And I was serious about getting data from real life situations using 2 towmen in different conditions.
Jan 12, 2010, 04:43 PM
Registered User
Quote:
 Originally Posted by Tuomo I see, a very clever graph. We could use that method to estimate how efficiently the plane climbs in launch.
Good idea. One of the problems in estimating efficiency during climb is estimating the wind contribution. Let me explain.
Take a look at the blue curve. I didn't even try to launch properly: for sake of repeatability, I just grabbed the fuselage behind the TE, build up tension and let the plane go. This means that the plane must accelerate a lot before producing enough lift to have a good rotation and to balance line tension. In fact, you can see that for the first 20 meters the plane simply "falls" towards the turnaround (plane-turnaround distance drops by 20 meters in the first 22-23 metres of X displacement!). I purposely use the verb "fall" because I tend to think of the towing part of the launch as gliding under the effects of a centripetal force field due to the line tension force instead of the uniform gravity field.
After the first 20 meters or so, the distance to the turnaround seems to level out (i.e., in my opinion, the model starts "flying"), then something strange happens. Distance starts increasing, and it keeps increasing until the zoom. This is strange, because it means that the plane "climbs" instead of "descending" (in the centripetal force reference frame) towards the turnaround. My guess is that we are seeing wind in action here: by increasing the aerodynamic force acting against line tension, it actually stretches the towline (1.15 mm mono) and lets the model build distance from the turnaround. It was just less than 2 m/s at ground level, but it is possible that the model experienced some gradient while climbing.
 Jan 13, 2010, 02:08 AM Registered User Looking at the second "climb" graph, it also easy to understand why throw is so important part of F3J launch. I am taking about the 20-23 meters... In the autumn after Bovec I worked with this part of launch. My throw is not yet in the same category as, say, Marco's, but I got it improved a lot. Would like to see data of this improvement... The key element is to get climbing cleanly straight out of hand. Otherwise it will "fall" towards the turnnaround, as you describe. I am also sure that we could use this fall towards turnaround / fly in the line -phenomena to better understand and trim launch flap settings. At the moment of throw we should use enough flap to minimize the "fall". On the other hand having too much flap inreases drag and reduces speed. Or is it so that line aerodynamic drag is more apect of F3J launch?
Jan 13, 2010, 03:36 AM
Registered User
Quote:
 Originally Posted by Tuomo Looking at the second "climb" graph, it also easy to understand why throw is so important part of F3J launch. I am taking about the 20-23 meters... In the autumn after Bovec I worked with this part of launch. My throw is not yet in the same category as, say, Marco's, but I got it improved a lot. Would like to see data of this improvement... The key element is to get climbing cleanly straight out of hand. Otherwise it will "fall" towards the turnnaround, as you describe.
I agree totally. In fact I am working on my throwing technique as well. There's no point in giving up those 20 meters...
Jan 13, 2010, 04:40 AM
Registered User
Quote:
 Originally Posted by Tuomo I am also sure that we could use this fall towards turnaround / fly in the line -phenomena to better understand and trim launch flap settings. At the moment of throw we should use enough flap to minimize the "fall". On the other hand having too much flap inreases drag and reduces speed. Or is it so that line aerodynamic drag is more apect of F3J launch?
In fact you raised a key point. Line drag is a big part of the total drag budget, and a correct estimate for it plays a big role in deciding launch setup.
I need a little math for this, because you can't treat the line simply as if it were a rigid body moving through the air. A better approximation is to see the line as having a moving end and a fixed end (the turnaround... I know it moves, but it moves much slower than the other end so -in the worst case- we underestimate its drag). For a line of diameter $h=r33yOuFoLDeI0bDPgE2O0Q$ and length $h=KSh8y5M3NP_89CfexAqEmw$, with the "free" end moving through the air with airspeed $h=IyFq48Q3WwnXNRcP1BfIPg$, we have that every line element $h=LbQ1b-NumZVW8MHmZBTdqg$, at a distance $h=J_VFgsmKa_jW99kHwjFVng$ from the turnaround, gives a $h=tsKkNGKNJULMnMTDUhx2ng$ contribution to aerodynamic drag:
$h=wVQdmbHwzj6k0C_M4MbJ6Q$
and when we integrate over the whole line, we get:
$h=oRxIIuH0wUs-k-ouG84T3w$
if we sum model drag $h=bh5eRXAREZ7D30yclBcqlw$ and line drag, we get:
$h=NRKUPa7bNyd2tyqdo6Lh4g$
The expression between parentheses is the new "total" (line+model) drag coefficient w.r.t. the model wing surface. A line with circular cross-section slicing through the air has a drag coefficient of about 1; therefore a 150 m monofilament line with 1.15 mm diameter, compared to a model having a wing surface of 75 sq dm, gives an additional contribution to drag coefficient of about 0.076, which is roughly twice the drag coefficient of an F3J model when gliding at the optimum L/D!
As a consequence, if -for instance- we want to climb at the best L/D during tow, we need to consider that CD on tow is much greater than it is during the remaining part of the flight, so we need to increase CL considerably. In other words, flaps are your friends during launch.
Last edited by Francesco; Jan 13, 2010 at 01:29 PM.
 Jan 13, 2010, 12:36 PM Roman Vojtech Hi Francesco, very interesting discussion. Please could you explain more in details how from speed (TAS/IAS) and altitude do you calculate distance of model from pulley (turnaround) and from pilot (point of release)? I also agree that drop of altitude in very first part of launch and "fall" to pulley is really problem Im thinking very long time about. But no idea were is optimal solution. The more flaps at the beginning, the bigger speed and altitude drop then... Maybe not to wait that 2 seconds (after model release) and change flap position already after aprox. 1 second?
Jan 13, 2010, 01:25 PM
Registered User
Quote:
 Originally Posted by lomcovak. Hi Francesco, very interesting discussion. Please could you explain more in details how from speed (TAS/IAS) and altitude do you calculate distance of model from pulley (turnaround) and from pilot (point of release)?
The method I use is a bit rough... it is simply an estimate of trajectory from one point to the following one, by using Pythagorean theorem for right triangles given the Y-position of the two points and the distance in straight line between them (velocity times sampling interval), like in the figure I attach to this message. It requires a simple boundary condition, i.e. the position at release which is well known.
There are some approximations I take in this approach. First of all, I assume that the plane moves purely in a vertical plane, which is not too far from reality for F3J towing. Second, the climb path is computed in the airmass reference frame, which is generally in motion w.r.t. the ground (turnaround) reference frame. This requires wind speed to be negligible compared to the average model speed (>20 m/s). Strictly speaking, one should also take into account the vertical wind gradient, but if wind is weak the vertical gradient is also limited.

### Images

 Jan 14, 2010, 03:58 AM Roman Vojtech Thank you very much for answer. May I have another few questions? 1. Why this aproximation is not valid for case when plane could move in more horizontal plane? (F3B launch) 2.Could you zoom-in on altitude log first 3 seconds? I see big drop of speed but no drop of altitude. 3.After launch and transition to gliding phase there is drop of speed to nearly 1-2m/s? I doubt if Pike can fly so slow? If I would see drop in altitude I would say "fine it stalled, it's usuall mistake", but I see no fall... 4.I see your Pitot tube placement. I'm not sure if this is the best position due to "shadow" turbulences from nose of model. Or do you have different opinion? Thanks Roman Last edited by lomcovak.; Jan 14, 2010 at 04:08 AM.
Jan 14, 2010, 06:12 AM
Registered User
Quote:
 Originally Posted by lomcovak. Thank you very much for answer. May I have another few questions? 1. Why this aproximation is not valid for case when plane could move in more horizontal plane? (F3B launch)
If I relax the single-plane condition I'd have to go to a three-dimensional case, and in this case I would lack one variable to perform integration (I'd either need lateral position or vector orientation).
Are you by any chance Mr. Lomcovak of the LoLo altitude logger? if so, do you think it could be possible to interface an inertial measurement unit (IMU) to the logger, so to be able to directly compute position out of acceleration and angular velocity? I see small IMUs are available nowadays, they may not be precise in long-term but for short interval they might be OK.
Quote:
 Originally Posted by lomcovak. 2.Could you zoom-in on altitude log first 3 seconds? I see big drop of speed but no drop of altitude.
See attachment. You're right: there is a big drop in airspeed. My interpretation is that, due to my launch technique (poor), the model quickly accelerates under line tension while rotating, and when it starts "flying" it also builds up drag and so it slows down. But I'm very interested in better interpretations... what do you think?
Quote:
 Originally Posted by lomcovak. 3.After launch and transition to gliding phase there is drop of speed to nearly 1-2m/s? I doubt if Pike can fly so slow? If I would see drop in altitude I would say "fine it stalled, it's usuall mistake", but I see no fall...
I let the model stall deliberately, again to ensure better repeatability. There is actually a 7 m altitude drop.
Quote:
 Originally Posted by lomcovak. 4.I see your Pitot tube placement. I'm not sure if this is the best position due to "shadow" turbulences from nose of model. Or do you have different opinion?
Good question. I had calculated boundary layer thickness in that point (less than 1 cm), and the Pitot tube should be well outside the BL, at least for typical airspeed and angle of attack.

### Images

Last edited by Francesco; Jan 14, 2010 at 06:38 AM. Reason: Attachment added
Jan 15, 2010, 05:50 AM
Roman Vojtech
Francesco,
thank you for answers. Yes, that's me Roman "Lomcovak" Vojtech. Bellow I will comment discussed points.

-Inertial Measurement Unit (IMU) is device I'm thinking about very long time. Never had enough time and resources to design and build one. To new LOLO5 is possible to interface nearly anything.

Anyhow I had very short chance to test IMU with looger (sorry article is only in Czech). The blue box on fuselage is IMU with logger and metal cylinder is third party altitude sensor. On this picture you can also see my Prandtl probe (this is the biggest one, I have many and some much smaller). It's connected to modified LOLO5. LOLO5 altitude logger is hidden behind receiver (it's so small). To sum up, in this flight there was LOLO5 (altitude, voltage, servo), LOLO5 modified for differential pressure sensor, IMU with third party altitude logger (one device).

On pictures you can see some acceleration loggs. I can even send you data and application for playing with records. You could evaluate if there is chance to use such IMUs for our purpose. Looks you are in math 100x better than me

In detailed picture you can see some limits of this device when acceleration in Z axis was bigger than 16G (zoom/pull).

-As regards initial launch speed drop. Yes, my interpretation is same, but I don't like this phenomena. I would guess that speed development would be more efficient as is drawn on my speed log bellow (green line).

On point 1 you see altitude drop which I was claiming not to be seen on your EagleTree log. I'm thinking why...

On point 2 you see critical part of this phenomena we are talking about. I would guess green curve would be much better. I have to test it and mainly find a way how to reach this "green line". Unfortunatelly I was starting intensively flying with my speed logger at late autumn last year and since that time there is quite bad weater and terrible cold i.e. I have not too much data. Every +1m/s before zooming makes huge difference in terminal altitude. At least in F3B launch. I'm not experienced with F3J launch, but you are on the same boat, I'm sure...

On point 3 is terminal speed increase I'd expect when we avoid speed peak at point 2. But maybe I'm completely wrong, I have to test it.

-as regards boundary layer near by Pitot tube, yes you are correct, that's good point. What's going on with boundary layer during zoom/pull, speed turning (F3B) and other very fast transitional moments, I'v no idea and I'm not sure about that 10mm thickness.

Anyhow your Pitot tube position is at least much more user friendly than mine.