Not sure if bug or feature. Excessive drag/heating on some Mk1 parts.

[Trogdor!]

Zero mods KSP 1.7 on win 8.1.

While trying to make the most streamlined flyers possible, I've found that certain mk1 parts seem to generate excessive drag. This Picture shows what I'm talking about.

the 1.25m cargo bay is responsible for almost *half* the drag of the entire vessel, even though it's attached in line to the rest of the ship. As near as I can tell, this is because of the tiny lip shown here highlighted in purple.

The NCS adapter is showing 0.34Kn of drag and the cargo bay directly attached behind it is showing 4.79Kn

This effect is also noticeable when other parts are swapped into the same spot such as the advanced inline stabilizer, RC 001S Remote guidance unit, and 1.25m heat shield. Are these slightly larger than MK1 parts supposed to be generating this much drag and therefore heat? Or are they supposed to be streamlined? If a heat shield is placed before the cargo bay, stabililzer and probe core, those parts generate much less drag, but the heat shield actually wears away even while shrouded due to drag and heat being generated..

While the pictures show effect happening at ~21Km and 1700m/s, it can also be generated at 100m and 700m/s with a wheesley turbofan.

Craft file is here

[OHara]

I think you have it figured out perfectly.

In version 1.2 there was a change to make sloped surfaces have less drag :

Quote

• There is an additional drag curve whose function is to raise drag coefficient to a power based on the mach number. This leads to blunter things having (relatively) more drag supersonic than they did and pointier things having (relatively) less than they did. Pods will slow down better on steep reentries and planes will produce drag in better proportion to how streamlined they are (less overpowered airplane drag).

That has the effect that any blunt surface, like a tiny lip that is not sloped, contributing a large fraction to the drag.   
For SSTO efficiency challenges after 1.2, the stacking order of parts became important, to try to avoid forward-facing lips.

(There is a chance that you accidentally connected the reaction wheel and remainder of the craft to some contents of the cargo bay, and not the outer rear connection point of the cargo bay.  But that makes an even more obvious effect.
I don't use Steam for KSP so can only 'subscribe' to your link, not actually download the craft file to see myself.)

[Trogdor!]

1 hour ago, OHara said:

For SSTO efficiency challenges after 1.2, the stacking order of parts became important, to try to avoid forward-facing lips. 

(There is a chance that you accidentally connected the reaction wheel and remainder of the craft to some contents of the cargo bay, and not the outer rear connection point of the cargo bay.  But that makes an even more obvious effect.
I don't use Steam for KSP so can only 'subscribe' to your link, not actually download the craft file to see myself.)

I wasn't really sure how to attach the craft to the post, so I just used steam. It's a pretty simple ship and can be recreated completely with 14 or so parts. The wings are structural wing type Ds tilted at 5° and the horizontal and vertical stabilizers are elevon 4s, with COL balanced slightly below and behind the COM. I checked numerous times on the cargo bay attachment and am pretty sure that's not the case, but regardless, the effect still happens without any cargo bay at all. Most notably the 'shrouded' heat shield acting as if it were exposed to atmosphere and the reaction wheels overheating and eventually exploding if attached directly behind the NCS.

Is there any sort of guide for the correct order for these 'mk1.01' parts? I had some success rearranging things by putting the engine nacelle in front of the cargo bay, but it also gets really hot. That tiny lip is presumably responsible for twice as much drag as the wings, which seems... odd. Perhaps we need to send Bill out there to perform some rocket surgery with an angle grinder or a very large lathe.

[TBenz]

From what I have heard, unused attachment nodes contribute to drag. Try putting nosecones on the top and bottom attachment nodes inside the payload bay, to see if that helps for that. You could feasibly have the same problem on the heat shield, as it has two different attachment nodes (one with the shroud, one without). As far as the reaction wheel, that probably just comes down to it being larger than the other parts. You probably don't need that large a reaction wheel on such a small craft, you might be better served putting some smaller reaction wheels inside your craft.

[OHara]

This forum is text only, so images are linked from other hosts.  kerbalx.com is great for  KSP craft files that you are ready to share, or works-in progress that you will eventually want to share.

On 5/23/2019 at 1:01 PM, Trogdor! said:

Is there any sort of guide for the correct order ?

Smaller in front, larger in back, so that the only big lip is rear-facing.  The surface areas used for drag are very close to what you see. 
There is a way under Alt-F12 to show the in the right-click menus the (exposed) surface areas and coefficients of drag of each part, which details you might or might not enjoy looking into.

I can see some ways the simplified aerodynamics model of KSP might be good for a game; it gives no incentive to spend time tucking parts into each other, because all that matters the the parts orientation and  connections at those green nodes.  But, there were some un-physical game-play adjustments that have not-fun effects like the one of this thread. 

If you want more-physical aerodynamics, there is a mod, FAR,  that simulates the airflow around the shape of the craft, and it works rather well.

Edited May 25, 2019 by OHara
missing 'not'

[OHara]

44 minutes ago, TBenz said:

Try putting nosecones on the top and bottom attachment nodes inside the payload bay, to see if that helps for that.

Yep, that's it.  Putting any node-attachable part on the bottom floor of the service bay --- even though that part is shielded from airflow and contributes nothing to drag --- counts as covering an equivalent area of that forward facing exposed lip around the outside of the service bay.  An octahedral strut inside the bay is big enough to cover the lip on the craft shown above.

Unused attachment nodes do not always cause drag, but the show an opportunity to remove drag by covering the area of a draggy surface.  Here, the exposed outer top wall of the was draggy, but the inside floor has no drag, but covering the inner floor counts as covering forward-facing area.  Makes no sense, but I can imagine someone writing code that would account this way.   I'll update my notes on notes on drag once I recover from the disappointment.

[Trogdor!]

4 hours ago, TBenz said:

From what I have heard, unused attachment nodes contribute to drag. Try putting nosecones on the top and bottom attachment nodes inside the payload bay, to see if that helps for that. You could feasibly have the same problem on the heat shield, as it has two different attachment nodes (one with the shroud, one without). As far as the reaction wheel, that probably just comes down to it being larger than the other parts. You probably don't need that large a reaction wheel on such a small craft, you might be better served putting some smaller reaction wheels inside your craft.

While I appreciate the information, it doesn't only happen on the cargo bay. I rebuilt another from scratch without ever adding a cargo bay or heat shield and ran it again.

As you can see, this time the RGU is behind the NCS, and it's drag value is 2.24kN, while the NCS is 0.31kN, the wings are 0.74kN and the stabilizer seems to be in the RGU's shadow with 0.13kN. The RGU is also a toasty 1927K.

I'm set up to launch vertically, so the reaction wheel is a leftover from Wheesley testing. Kind of hard to stay balanced on the tip of it for long enough to get going without it.

I guess I'll have to track down all the 'mk1.01' parts and test which order to put them all in. Perhaps it's possible to start with a cargo bay packing nosecones and taper down from there?

20 hours ago, OHara said:

 my notes on notes on drag

Thanks to Ohara's most excellent notes on drag, I've compiled a spreadsheet on all the mk1 parts. It seems size is the determining factor. Apparently there is no sort of automatic 'taper' when two parts are mated. If anything is left over, the remainder is counted as a completely flat surface. I suppose this is just a leftover from way back when the mk1 parts were created where the size wasn't actually that critical to the design of the model.

Also, just looking at the numbers, the heat shield with a faring is 1.44m² and without it is 1.27m² . Which leads to the awkward realization that the *shielded* heat shield is therefore more exposed to the atmosphere and subject to ablation due to friction on the fairing. Kinda counterintuitive.

I suppose these quirks aren't really that important in a game about space, but I started down this rabbit hole because I wanted to find the smallest atmospheric probes that I could use on Laythe. As an aside, it's totally possible to go supersonic with a tiny juno drone who's only control surfaces are reaction wheels. Too bad you can't tuck any science instruments on it because they all count as exposed, no matter where you put them.

 

[FleshJeb]

On 5/23/2019 at 3:50 PM, OHara said:

Yep, that's it.  Putting any node-attachable part on the bottom floor of the service bay --- even though that part is shielded from airflow and contributes nothing to drag --- counts as covering an equivalent area of that forward facing exposed lip around the outside of the service bay.  An octahedral strut inside the bay is big enough to cover the lip on the craft shown above.

And I thought I had come close to mastering drag-voodoo... This is scary.

@Trogdor! I've found that the service bays are consistently some of the LEAST-draggy parts. My guess was that it's the reaction wheel and probe core causing all the trouble. If I were to modify your craft, I would put a 0.625m probe-core and reaction wheel inside a second service bay. I've pulled off Jool dives that reach 2x terminal velocity on very similar craft.

[Trogdor!]

Update:

So after extensive testing, I think I've found a pretty decent part order for the least amount of drag on mk1 parts. I uploaded the original test drone to kerbalx here, and have a new craft with much reduced mk1 part drag here.

On 5/23/2019 at 6:01 PM, OHara said:

Smaller in front, larger in back, so that the only big lip is rear-facing.  The surface areas used for drag are very close to what you see.

I've found that you should build up to the largest, leading with the parts that create the least amount of drag, then tapering down. Because the engine nacelle is larger than both the probe core and the cargo bay, it shields both of them really well, but you have to build up to the probe core with a faringless heat shield mounted in front of an empty cargo fairing.

Incidentally, the SAS module has less rear facing drag than the cargo bay, so it acts as a step down.

I'm not sure if this is the most optimal order for mk1 parts, but it's fairly optimized. I also could have inserted an inline cockpit between the probe core and the nacelle for even less rear facing drag, but this craft maxes out at 1,751m/s already. You could try the 'mount an intake behind the engine trick' for even less drag, but you won't actually go any faster thanks to the engine flaming out.

On 5/28/2019 at 5:07 PM, FleshJeb said:

 

@Trogdor! I've found that the service bays are consistently some of the LEAST-draggy parts. My guess was that it's the reaction wheel and probe core causing all the trouble. If I were to modify your craft, I would put a 0.625m probe-core and reaction wheel inside a second service bay. I've pulled off Jool dives that reach 2x terminal velocity on very similar craft.

It's not the probe core or the reaction wheel, it's apparently how mk1 parts are calculated. Some mk1 parts are larger than others, and apparently order is very important. You can take away the probe and the SAS wheel and the cargo bay will still generate significant drag at high speed, because of a tiny 0.078 m² forward facing lip that has a C_d,0 of 0.91. I've found that if you shield it with something larger, like an inline cockpit or an empty faring, you will eliminate all of it.

Edited June 2, 2019 by Trogdor!