Rethinking clipping...

[swjr-swis]

A small example of the kind of clipping I think is fully warranted in KSP.

I've been playing around with a small and fast long-range Mk2-body jet. I wanted to have just the cockpit, a single engine, and the minimum parts required to make it fly and perform the way I want. Long-range to me means being able to go anywhere on Kerbin and return without refueling. Fast means a maximum flight time of around 1 hour. Additionally, I want a probe core included if at all possible - I play with G-forces on, so I need a safety net for when pilots momentarily pass out at high loads. Preferably, I want this jet to look the part as well... it needs to be nice and sleek.

I decided to try with the Wheesley first, since it's a champion at low fuel consumption. I can reach cruising speeds of Mach 1.7 - 2 @ between 12-14km with Mk1 body planes, which allows them to circumnavigate on as little as 200-300 LF. But I had my mind set on Mk2. Mk2 body drag increases very rapidly when off prograde, and the Wheesley is not exactly high performance which leaves very little margin, so this means I have to balance the plane very carefully to navigate the corridor that will allow it to punch into supersonic and maintain high cruise from start to end. It also means I need to keep this plane down to a single stack - the additional drag of even tiny nacelles would be too much.

After some unsuccessful iterations that told me I couldn't even afford to use radially attached intakes (and that there most definitely is a 'right' size of wings), I ended up with a stack of Mk1 cockpit, long Mk2-1.25m adapter, Mk1 diverterless intake, 1.25m probe core, and Wheesley.

The Mk2 adapter would've been enough fuel on its own, but as the cockpit is so much heavier than the engine there's no good way to keep this balanced through fuel usage. And I needed intake air. The DSI offers extra fuel tankage, plenty of air intake, and the possibility to offset for balance, all while still being part of the single stack. DSI is then offset as far to the fore as possible, clipping it almost entirely inside the Mk2 part of the body, with the core offset into the exact CoM and the Wheesley offset further back again to visibly 'fuse' with the 1.25m end of the Mk2 adapter. Both the DSI and the Mk2 adapter are then only filled partially, to the exact amounts that make CoM stay exactly put from full to empty.

Bonus: the DSI scoop is outside the body (no 'magic' air) and integrates beautifully for a visually very plausible design. Cherry on top: since I am only using part of the tank volume of both DSI and Mk2 adapter, I can even 'justify' the clipping of the parts, for those to whom such things matter.

Pics or it didn't happen:

Spoiler

Some parts clipped for looks and fuel balancing. DSI intake is attached to end of Mk2 adapter, then probe core, then Wheesley. Offset is then used to move them to their respective places while still being part of the same stack. Fuel is balanced, air intake and probe core are provided, and everything integrates into a visually sleek and plausible fuselage.

A rather narrow range of 'wing area' was needed to allow this plane to work - a good bit more than what I wanted for the visual profile, but not so much as to prevent the Wheesley from powering through transsonic due to parasitic drag. So I had to iterate through a few different wing section combinations until I found one that I could clip in such a way that it presented the type of small swept wing I wanted, yet still left CoL where I needed it for balance/performance.

End result: a small Mk2 jet that can cruise at Mach 1.82 @ 12-13km using around 0.03 LF/s, which on a default fuel load allows it to circumnavigate with 40% fuel to spare.

Craft file: https://kerbalx.com/swjr-swis/Mk2-Mini-2b

 

[Lt_Duckweed]

On 9/12/2020 at 2:05 AM, sturmhauke said:

I might try a test with a stock Rapier, a stock Rapier with the cone trick, and a modified Rapier with the aft node removed via Module Manager. 

Removing the node via module manager will do nothing to reduce the drag.

Drag in ksp is calculated off of "drag cubes" which have surface area and "pointiness" factors for each of 6 different faces.  Each node on a part is assigned to one of the 6 faces.

Drag cubes can only ever be modified by node attatchment, NOT by deleting a node in the part deffinition. When you node attach two parts, the drag cube faces tied to each node are compared. The ways this can turn out are:

1. The smaller face (by surface area) is removed entirely.  The larger face has the surface area of the smaller face subtracted from it, and it's pointiness factor adjusted to reflect the shape of the surface area still exposed. 

2. If the two faces are exactly the same size, both get removed.

3. If they are close in size, but don't quite match, its a bit inconsistent but sometimes the pointiness of the larger face gets set to 0, making it infinitely pointy, and thus effectively removed.

[sturmhauke]

Gotta love those weird physics approximations...

[Klapaucius]

I don't understand why anyone would not part clip.  It allows me to build very funky things and sculpt, as a real designer would. The Jaguar E-Type would never have existed if all the builders could do was bolt square fenders onto a chassis.  It allows me to build insane craft, which is what I find most fun.

Having said that, I do set some rules for myself. I never double up engines nor try to get ridiculous amounts of fuel by putting one tank inside the other. All my clipping is done for aesthetic and/or weight reasons. Never intentionally for fuel or power.

I also once practiced the opposite. Extreme non-clipping.  It is like flying an exploded diagram.

 

Edited September 19, 2020 by Klapaucius

[Rocket Witch]

On 9/12/2020 at 7:56 AM, swjr-swis said:

It is absolutely still a factor. Don't take anyone's word for it, regardless of their online following; just try it - takes all of 2 minutes and one revert to prove it to yourself.

It wasn't anyone's word; as I recall, that thread had a lot of data posted in it.
Anyway, one revert doesn't paint the whole picture. Results from firing a 1.25m rocket straight up:

Top Part                    Mass      Altitude

Unoccupied size1 Node       5680 kg     855 km
Communotron 16              5685 kg     844 km
Small Nose Cone             5690 kg   12891 km
Aerodynamic Nose Cone       5710 kg   17670 km
Surface attached A.N.Cone   5710 kg     658 km

I thought the Communitron 16 would be the best to control for node occlusion by itself being a factor, as it's a tiny 'physicsless' part with one size0 node. The relative performances of the parts suggest that node occlusion is present but only applied in proportion to the top part's actual diameter.

Full disclosure: I did this on a modded install (including modifying the ANCone to allow surface attach for comparison), but don't have any aero mods.

Edited September 21, 2020 by Rocket Witch

[Kernel Kraken]

I don't get the dislike for performance clipping some people seem to have. I build lots of replica planes, and sometimes the exact same layout of parts an actual plane has will simply not produce enough thrust in KSP, so I usually clip intakes for reasons of "There's no way an air intake 1.75 meters in diameter only pulls in 2 liters per second of air". When I performance clip, it's for realism's sake.

If a challenge doesn't allow clipping, or if the clip is just absurd (i.e. making up for simply poor design with 600 clipped type d wings) I won't do it. But light clipping (say, 3 intakes in the space of one) is perfectly fine by me.

Edited September 23, 2020 by Kernel Kraken

[swjr-swis]

3 hours ago, Kernel Kraken said:

I build lots of replica planes

I think clipping in the build of replicas specifically is completely accepted. The limited library of parts we have, and the complete lack of sculpting tools (possibly with the exception of creative uses of fairings) means there's really little other way to closely reproduce the forms and lines of the RL counterparts. And due to the way stock aero and physics works, once you go that route, you're almost forced to also use clipping in other ways to get a reasonable performance out of the replica.

Building replicas is a form of art all by itself. Getting the right visual is paramount. Getting it to perform is a second consideration, but a definite bonus. Clipping, in any form, to achieve either is pretty much a tool of the trade.