Dumpling range challenge

[zolotiyeruki]

Reminds me of the old saying about DC-3's--"a collection of parts flying in loose formation"

What optimizations did you do to extend the range like that?

[Tyr Anasazi]

11 hours ago, zolotiyeruki said:

What optimizations did you do to extend the range like that?

Your simple question requires a rather long answer! Fortunately I have time, because GD MK1's flight is now unpaused and still has a long cruise ahead. When Valentina has safely landed back at KSC I will provide a report with photos, but for now here are the technical details.

GD MK1 was a complete rebuild from scratch rather than modification of the LEP design. The key performance goal was speed. To make a Global Dumpling based on the LEP engine, I needed a plane which could average above 250m/s for more than 4 hours, without increasing power draw.

Looking at LEP, I knew I had to:

• Eliminate any parasitic electricity draws
• Reduce drag and improve the lift/drag ratio
• Increase propeller thrust and power, at all speeds

Electricity parasites were the Stayputnik and OKTO probe cores in the engine. Once the engine is started, they can be hibernated, reducing power draw by nearly 8%, giving an additional 15 minutes of burn time. That may sound small, however 15 minutes at 254m/s equals 228km, just over 21 degrees longitude.

To reduce drag, the pilot was placed into an inline service bay rather than exposed to the airstream. At 150 m/s, a kerbal produces about 1.5 kN of drag - about 1/3 of the total aircraft drag. The fairing is configured as an inline fairing and so produces nearly zero drag despite its appearance. The little nose cone on which it closes is offset inside the fairing  for zero drag. To close off the tail I tested many items and concluded that the Aerodynamic nose cone produces less drag than anything else. LEP had elevons on each wingtip. I eliminated them by replacing a wing connector and elevon with an advanced canard, giving similar lift with less drag. The canard provides both lift and control. Control surfaces (and a pilot) are necessary because the airframe has no power for reaction wheels. The wing position, pitch, and span were all tweaked to improve the lift/drag ratio. Total airframe drag at cruise altitude and speed is approximately 3.8 kN.

Finally, the position, angle and pitch of the prop blades was optimised. The way the blades are angled results in more net force pushing forward as opposed to an off angle. This can be seen in the aerodynamic overlay, especially in early stages of the flight.  The prop operates at very high revs (up to 1750 deg/sec). Failure to monitor and control RPM results in engine failure (many of which occurred). The prop can feather its blades, resulting in nearly zero prop drag during the final glide phase. Prop pitch is modulated manually using the elevon authority limiter. +90 degrees for takeoff, +150 for initial climb, reducing as airspeed and altitude increase, -20 for maximum altitude (17,500m+)  -8 for cruise altitude and speed (12,350m at 254m/s), and -150 to feather. GD MK1 can glide more than 180km, from 12km cruising altitude.

GD MK1 aerodynamic optimisations:

[zolotiyeruki]

That's simply awesome   Now I have a few follow-up questions:

If the fairing is closed onto the small nosecone, how is it so small in the front? 
Does offsetting the nosecone into the fairing actually eliminate its drag?
In the past, the shock cone intake had the lowest drag of any 1.25m part.  Has that changed?
Why did you opt for the biggest reaction wheel?  Is it more efficient than the smaller ones?

[vyznev]

6 hours ago, zolotiyeruki said:

Why did you opt for the biggest reaction wheel?  Is it more efficient than the smaller ones?

That one I can answer. If you look at the reaction wheel stats in the game or on the wiki, you can see that the bigger wheels indeed produce more torque per electric charge (50 vs. 20 kN⋅m⋅s per unit of charge), as well as more torque per mass (150 vs. 100 N⋅m/kg).

Assuming that propeller thrust is directly proportional to torque, we can even calculate the acceleration per electric charge, at least up to an unknown constant of proportionality, as (torque / mass) / charge consumption rate. This turns out to be worse for the bigger reaction wheels, but only if you don't account for the mass of the rest of the craft. I'm not sure exactly how much the Global Dumpling weighs (and in any case the mass goes down as the fuel is consumed), but from the aero forces display in the screenshot above it seems to be about 6.30 kN / 9.81 m/s² ≈ 642 kg at the time the screenshot was taken, or about 442 kg excluding the mass of the reaction wheel. Plugging that extra mass into the formula indicates that the largest reaction wheel is almost twice as efficient as the smallest one (angular impulse of 77 vs. 41 m²/s per charge unit) for this craft (and indeed remains so even if we subtract the total fuel mass of 110 kg from the estimated vehicle mass, indicating that the amount of fuel in the tank makes no difference in this case).

Edited May 20, 2019 by vyznev
units

[Tyr Anasazi]

Greetings fellow dumpling experts 

Firstly I want to apologise for failing to update with photos of Valentina's triumphant Dumpling circumnavigation. The truth is that I spoke too soon and so I am eating humble pie - GD MK1 is until now the NQGD (Not Quite Global Dumpling), due to engine failures that have occurred very late in multiple flights. As the dumpling gets lighter, the plane rises to higher altitudes and accelerates, resulting in engine overspeed. The last failure occurred when GDMK1 was flying at longitude -134, having flown just over 300 degrees around Kerbin. The level of frustration has been high!

The bearing has been tweaked, resulting in a more failure-resistant engine and slightly faster plane which cruises between 255 and 260m/s. Valentina is staunchly confident that GD MK1 will work as advertised. GD MK1 is cruising on its fifth attempt as I write this. Rest assured, Val will not rest until she completes her historic flight!

Now to answer questions above by @zolotiyeruki: 

• The fairing is closed on the nosecone so is an inline fairing (when doing this, "close fairing" appears in blue instead of green)
• When this is done, the nosecone is offset backwards into the fairing. This eliminates the drag of the nosecone.
• I tested many parts for the tail including (I think) the shock intake and the least drag was achieved using the 1.5 short nosecone.
• The biggest reaction wheel is the most efficient in terms of electricity/torque. All other torque parts in the game are less efficient by a big margin as @vyznev explained. The key efficiency factor for this application is how much electricity is consumed per unit of torque. The advanced reaction wheel module large uses 0.020 units of electricity/kNm. the next most efficient part is the Advanced inline stabiliser, which uses 0.030 units electricity s/kNm - 50% more!

 

Edited May 22, 2019 by Tyr Anasazi

[Tyr Anasazi]

Aaaand touchdown! Well, kind of - Due to pilot error, Global Dumpling MK1 glided all the way down to KSC runway then crash landed. Too steep and fast. Doh. I was thinking of going again just to record a perfect landing, but seriously it's a 4 hour flight. 
I can now declare that GDMK1 successfully circumnavigated Kerbin.

Flight album: https://imgur.com/gallery/3NO9kfY

Craft file: https://kerbalx.com/Tyr_Anasazi/Global-Dumpling-MK1

Some points:

• When calculating the distance for circumnavigation we must add the average altitude of flight. GD MK1 averaged 12,350m altitude, so (600 +12) * 2 * pi = 3845km
• The inability to use time warp with stock props makes flight testing of long range craft VERY time consuming!
• I was afraid that GDMK1 wouldn't make it over the mountains near KSC, but it turned out to be an excellent glider and easily passed far above them. I think that it may have been able to continue all the way to the island runway.
• GD MK1 reached 17,500m+ during its flight - Therefore I also submit this as an entry for the prop altitude record
• Trying to carry out this flight without Mechjeb or other autopilot is not feasible in my opinion.

[Pds314]

On 5/5/2019 at 2:48 PM, zolotiyeruki said:

One important warning: the distance travelled shown in the F3 menu is NOT accurate. I don't know the exact source or magnitude of the inaccuracy, but it seems to be related to the rotation of Kerbin while you're airborne, or something like that. We ran into that a lot during the Circumnavigation Challenge.

It would be a lot more accurate to record one's longitude on landing/crashing.

 

I believe it is exactly double. But I have yet to completely prove there are no weird geometrical issues with 3D flight or turning that could exacerbate that, or issues with Krakensbane above Mach 2.

[zolotiyeruki]

It sounds like some experimentation is in order, both on the aero side, and also the "distance traveled" side.

I'm running some drag tests right now, to see how much drag is generated by various parts.  So far, it appears that the aerodynamic nose cone is OP, compared to the advanced nose cone (which seems like it should be far less draggy) and the shock cone, which previously held the crown for lowest drag.  How OP?  The NC has consistently 1/4 - 1/2 the drag of the Adv. NC, depending on AoA, and 50-85% of the drag of the shock cone intake.

I've also discovered that the "stick a shock cone on the back of a RAPIER and offset it into the engine" trick doesn't work any more.