RAPIER engine drag in 1.6.1, with tailcone attachments

[fourfa]

The RAPIER engine is unique among jets, as many of us have found when building SSTOs.  It has an attachment node on the back, like most rocket engine parts, but unlike any other air-breathing engine.  The way KSP's aerodynamic drag model works this node, when left open, is meant to look like a draggy, untapered, flat plate.  With vertically launched rockets, that's no big deal - you get out of the atmosphere quickly enough that the drag hardly matters.  Problem is, spaceplanes flying single-stage-to-orbit spend a LOT of time cruising at high speed, suffering from this drag that doesn't afflict any other jet engine.  Since the long-long-ago, many have found that placing a reversed nosecone on that node (essentially giving the engine a smooth tapered profile) and clipping it forward into the engine, so it doesn't block the exhaust, brings the total drag down to a level similar to the other jets.

Is this an exercise in splitting hairs on an already-overpowered part?  Perhaps.  But sometimes the building and tuning of an SSTO (particularly efficient, underpowered ones) can be a delicate thing, and drag reduction always helps.

Anyway I got curious to look at this drag with a variety of possible rear attachments.  So I built a 120 ton spaceplane with seven pairs of RAPIERs, each with a different tailcone, capable of lifting an 36T orange tank to orbit:

  Typical flight pattern for a powerful, heavy, inefficient SSTO - constant steep trajectory (15 degrees, any less and it'll burn up) until flameout, then follow prograde on rocket mode until 100km apoapsis, then circularize and deploy:

On the way up I monitored and screenshotted the aerodynamic drag of all engines and tailcone parts (Alt-F12 -> Physics -> Aero -> Display Aero Data in Action Menus):

then plotted the total drag of each combination through the whole trajectory to space (in logspace to amplify the differences):

Tested options were: bare RAPIER, RAPIER with a small 0.625m nose cone directly on the 1.25m rear node, small nose cone with A10 structural adaptor, circular jet intake, aerodynamic nose cone, advanced nose cone A, and shock cone.  As you'd expect, peak drag occurs at transonic speeds (approaching Mach 1) then declines.  The shape of the curve is plotted against speed directly, though really the drag number reflects both speed and density.  The downward knee in the curve at Mach 4.5 is where jets flame out and switch to rocket mode, then the massive power of 14 RAPIERs pushes the craft into less dense air very quickly.  The final small uptick in drag happens when setting apoapsis, having reached orbital velocity while still in atmo before MECO and coast to apo.

Nevertheless, the interesting result is the differences between these curves.  The bare RAPIER obviously has considerably more drag: all of these tailcones are effective.  There are some unexpected reversals - for instance, the small nose cone does poorly in the initial climb out but is one of the best performers at high speed and high altitude.  Then the small nose cone with the A10 adaptor starts out good, becomes the worst combination at Mach 3, the best at Mach 4, then second worst at high speed.  The shock cone and circular intake perform very well.  The plain nose cone outperforms the advanced nose cone.

None of this takes into account the mass of the parts.  Whether that matters to you will be highly dependent on the craft in question; with this 156T sledgehammer it matters not at all.  On an efficiency-challenge nuclear-powered SSTA, it really might.  The plain nose cone and circular intake do very well for their mass (and cost); the point of the shock cone sticking out in the back undeniably looks the coolest .  The others are probably better avoided.

[I_Killed_Jeb]

Thank you for this. I've been sticking with the nosecones based on some anecdotal information, good to have confirmation that this is still the case.

[Shadowmage]

Really this brings to light the major issues with the drag cube system regarding engines, and the discrepancies between them (or rather, the issue with basing drag on attach nodes at all; it should be entirely geometry based, nodes shouldn't figure into it).  Sadly, it really just makes me want to start putting nosecones on the back end of my rockets' engines as well....

 

 

[DMagic]

The drag isn't caused by the empty attach node on the back of the engine. It is caused by that engine's drag cube, which reflects the very flat surface on the back of that engine. If you look at most of the other jet engines they are tapered more like a cone in the back, and their drag cube values reflect that. There is one other engine, the Whiplash, which has a relatively flat back and a relatively high drag coefficient on its back end, I would guess that one also generates excess drag, though not as much as the Rapier. 

If you want to test this you can just delete the attach node at the back of the Rapier (node_stack_bottom in the config), and compare its drag to the standard version with nothing attached.

Putting a nose cone on the back of the Rapier just blocks the drag on the back of the Rapier and instead presents a nose cone, which is similarly shaped to other engines, to the airstream.

More on drag in the spoiler:

Spoiler

Rapier drag cube, the drag on the "back" of the engine is reflected by the bold values, the middle value in particular, 0.7808, the drag coefficient, is probably the biggest driver of excess drag for this part:

cube = Default, 1.495,0.7777,0.7375, 1.495,0.7777,0.7375, 1.304,0.9318,1.384, 1.304,0.7808,0.7056, 1.499,0.7745,0.738, 1.499,0.7754,0.9021, -7.272E-05,0.3494,0.0002167, 1.292,1.184,1.294

This is the drag cube for the Panther, which has a much more tapered back side, its corresponding drag coefficient value is much lower than the Rapier's, even though the surface area (the first bolded value) is similar :

cube = Default, 1.661,0.7782,0.7165, 1.661,0.7782,0.7165, 1.218,0.9522,0.163, 1.218,0.45,1.493, 1.661,0.7778,0.7165, 1.661,0.7786,0.7165, -2.384E-07,-0.6688,2.384E-07, 1.25,1.463,1.25

The only way to really fix this is for Squad to stop relying on automatically generated drag cubes for almost all of their parts. Engines could be massaged to reduced this tail-end drag (and to fix a number of issues caused by engine variants). And a number of parts are out-right bugged (structural tubes) because they don't have their drag cubes setup right.

Edited April 10, 2019 by DMagic

[Fearless Son]

I feel like KSP needs a kind of "aerodynamic shroud" attachment for sticking on the rear of engines that'll let the exhaust breath through while providing some straking that will break up the air and make less of an induced vortex behind attachable engines like this.

Also, it might look good as an aesthetic feature on some rocket designs.

[DMagic]

@Fearless Son Kind of like these things:

 

[Fraston]

Can’t help you now, 1.7 is out 

In all seriousness, it could be fixed by switching to 1.7.

Edited April 11, 2019 by Fraston

[Fearless Son]

19 hours ago, DMagic said:

@Fearless Son Kind of like these things:

Legitimately, yes.  I can think of other uses for them too.  Imagine a mk 1 passenger module that still has it's rear hatch accessible without causing a ton of drag.  Heck, give it some variable geometry articulation and you can make it a small control surface or airbrake by changing how the edges of it are inclined.  

Would be good for non-gimbling boosters on ascent vehicles too.

Edited April 11, 2019 by Fearless Son

[Foxster]

18 hours ago, Fraston said:

In all seriousness, it could be fixed by switching to 1.7.

How so?

[Aeroboi]

I'm not to fond of this technique due to it having very little savings. IMO the reversed nosecone trick has a effect in practice beyond a graph but it is incredibly negligible and only offers any benefit in terms of drag reduction if the space plane has draggy outer parts to begin with. It is much more effective to change your wing incidence by 1 degree so that your nose is directly through the velocity reticle and you gain much more effect.

There's a general consensus that the weight limit per rapier to push a vessel into orbit is 20 Ton's per rapier, with the expected higher limit is in the area of 25 Tons.
This is false. While it requires many optimizations I found out that I could do 37.4 Tons per Rapier and probably more using more wings.

This vessel has ISRU with RTG's (to visit Jool and Eeloo) with 7700m/s Dv with LF only tanks and can reach moho's surface from Kerbin albeit with a good launch window and a Eve assist. It Weighs almost 75 Tons.
I share the following vessel because it has just enough thrust to stay above sea level after exiting the runway during takeoff using the LV-N as assist. However, it is so aerodynamic that it can get past 400m/s while climbing at 4 degrees when above 300m/s.

The reversed nosecone trick would cut the total Dv budget (regardless of how low weight it is "and 2 extra parts") so I couldn't be bothered. To reiterate another way, the reversed nosecone trick would deduce the Dv budget while not requiring the trick in any case because the takeoff thrust is just barely high enough not to sink into the sea after liftoff. Mind you that I have optimized the craft completely and can reach Minmus with 3/5 Tank in which case it can race to orbit whereby the nosecone has even less effect. That's less reason to bother.

I only wanted to share this so people understand that the reversed nosecone trick isn't always a added benefit. I only use it when a nearly capable vessel fails to accelerate past Mach 1.20. If I try to optimize a craft for drag it never fails to do so, and when it is a interplanetary SSTO/SSTA the nosecone cripples the budget, albeit marginally.

Edited April 12, 2019 by Aeroboi

[IronMaiden]

On 4/12/2019 at 4:23 AM, Aeroboi said:

There's a general consensus that the weight limit per rapier to push a vessel into orbit is 20 Ton's per rapier, with the expected higher limit is in the area of 25 Tons.
This is false. While it requires many optimizations I found out that I could do 37.4 Tons per Rapier and probably more using more wings.

I've found the optimal mass to be 30-35t per RAPIER depending on the design, more mass would require substantially more liquid fuel during the open cycle ascent phase, with the maximum mass around 50t per RAPIER, but by then it's way past the point where it's more efficient to have 2 RAPIERs.

[Aeroboi]

1 hour ago, IronMaiden said:

I've found the optimal mass to be 30-35t per RAPIER depending on the design, more mass would require substantially more liquid fuel during the open cycle ascent phase, with the maximum mass around 50t per RAPIER, but by then it's way past the point where it's more efficient to have 2 RAPIERs.

50Ton's wow, I think I could do that. Such planes would have substantially more wings i'd presume, eventually you'll be adding so much wing's that another rapier is similarly in weight.