Servers by jDrones

UnATRaP - Part 1: Selecting the airframe

unatrap
testbed

(George Zogopoulos Papaliakos) #1

Design criteria

We might be part of the Mechanical Engineering department, but our laboratory does mostly control systems, not mechanical design. Thus we weren't going to build our UAV from scratch. Instead, we wanted to find a solid airframe and convert it to a flying robot.
We thought long and hard about the projects requirements, which would drive the airframe selection. We wanted:

  • A payload capacity of 1-2kg, to avoid the need to miniaturize prototype payload
  • Ample payload volume, for the same reason
  • A flight time of more than 30mins

Do you even lift?

A crucial but difficult question rose from the beginning of the market research procedure: How could we know how much an arbitrary airframe could lift in payload safely, i.e. without altering its flight characteristics significantly or risk a stall in every turn?
We couldn't find any clear answers, neither online nor in our local hobby-stores. The RC manufacturers don't publish payload specs for their models, and professional UAV manufacturers usually state pre-selected payload options.
The general impression was that an airframe could lift about 30% of its empty weight in payload, but that was sounding as an extreme generalization, which certainly couldn't cover the full range of aircraft, from STOL to jets.

In the end, we didn't come up with a definitive answer, but we can talk about UnATRaP's performance. VQ states that the empty-weight of the Porter is about 7kg. After our modifications, the UnATRaP's emtpy weight is 10.5kg and with 0.5lt of gasoline it needs at least half flaps to land safely. Once we forgot to deploy them and the UAV stalled 1m above the runway during landing!

With an extra 1.5kg of payload, the maximum airspeed has dropped 5m/s to 30m/s in order to maintain altitude and the aircraft definitel feels at its limit.
So I'd say that we reached maximum overhead at an additional 5kg than what the manufacturer specified, a 70% increase!
Naturally, ours was a low-wing load, STOL airframe, so your results may vary.


Ok, so which airplane should I buy?

Initially we looked at tailor-made UAV airframes, such as the Penguin B, the Silvertone Flamingo and the Arcturus series. However, these pushed the airframe budget into 5-digit sums, which we couldn't justify, plus they would require significantly more paperwork to obtain them.

So, we turned back to the familiar RC market. The old and trusted Skywalkers were out of the question. They simply couldn't carry as much as we needed. Plus, back in 2013 the only large-ish foamie cruisers available were the Skyhunters, which did have significant payload capabilities but weren't exactly up to spec. Today's Titans, Clouds and My Twin Dreams might be tempting options, but they would still be compromises against our initial requirements.

Inevitably, we went one category up, to the wooden models. Familiar contenders here were the Telemasters and the many Piper Cub choices. These could definitely carry our payload, but their fuselages didn't have proper access ports and/or were too narrow.
3D aerobatic planes commonly have ample space with easy access from a removable top hatch, but we were a bit concered with having to deal with such critically stable and fast airframes.

In the end (around 2014), we chose the Pilatus Porter PC-6, built from VQ model, and ordered it from a local vendor. Plus, we had had heard of another (Some people named Canberra UAV) team who had successfully used this airframe.

That is not to say that this option isn't without its cons. Wooden airplanes are harder to modify, easier to damager and have REALLY ugly crashes. That's why from the get go we knew we would have to do our best to avoid crashes at all costs.


Power Plant Choice

The PC-6 has the option of an electric power plant, with a 6S battery configuration. However, we saw two drawbacks with going the electric path:
1. It would be difficult to go more than 15 minutes of flight time at a time.
2. The batteries would be a very expensive item in the BoM and require special handling. Even worse, to fly consecutive missions we would need to either charge them on the field or have multiple sets.
In addition to the above, we wanted to try out a gas engine, since up to that point we only had experience with brushless motors.

That isn't to say that we expected things to go smoothly. Having been around the RC scene for a while, we had seen far too many people struggling to start up their IC engines. Endless woes tuning the low and high mix, heating up the engine, only to have it stall mid-air. Thankfully, none of the above has happened to us. The DLA 32 was tuned once and since then starts every time and never stalls, with two exceptions:

Once, after a winter of not having flown at all, it didn't seem to draw any fuel from the fuel tank. We had to remove the sparkplug, pressurize the tank and rev up with the starter to get it to flow through the engine block. After that, it sparked immediately.

The other bad feature of this engine (and also all low-grade 2-stroke ICs) is that its throttle to intake manifold to RPM curve is HIGHLY non-linear. At 50% throttle servo, the engine has almost maxed out its RPM range, which is problematic from a control systems perspective. We can't really use a custom curve from the transmitter, because the ArduPlane is agnostic to that. We have tried some geometric compensation, but we have still to see any results.

Also, the engine can't run in idle for long. After half a minute or so there is audible reduction in RPM, as residue builds up in the chamber and if left on idle for long it will stall. Occasional rev-ups prevent this from happening while the airplane is on the ground. On the air we never had such problems, since the engine runs higher than 4000RPM usually.


Propeller Choice

We are still in the process of finding the right propeller for our UAV. One constraint is that we can't go above 20 inches in diameter: the propeller will strike the ground. Another factor is that we want to target endurance, not engine response.

We have tried 20x8 propellers and find that the engine spins them far too easilly, maxing out the RPM range at 50% throttle. This leads us to believe that we can get the same thrust with lower RPMs (less fuel consumption) if we use greater step. With a 18x10 propeller there is not much difference. The next logical step is to go for a 20x10 and see what changes.


This concludes this part of the UnATRaP blog series. If you have comments or recommendations, please let me know. Until next time!

Previous article: Part 0: Introduction
Next article: Part 1: Fuselage Front Section