How much drag does an engine bell produce?

[Edax]

I've designed spaceplanes with the LV-N in the front for the sake of streamlining and thrust alignment, and I've had some success with it, but the nagging question in my head is, does the engine bell act like a giant air scoop producing massive amounts of drag?  Reality would tell me yes, but I'm not sure if the game models engine bell drag or not.  I could easily stick a air intake or nose cone with decoupler in front of the LV-N, but my spaceplanes would then lose the SSTO status, so I'm wondering if that step is necessary at all?

Picture below, not actual SSTO.

[Streetwind]

Well apparently it doesn't add enough drag to keep you from reaching Minmus in a single stage... so, how much does it really matter?

[swjr-swis]

I do not think engine bells generate any more drag in KSP than if they were simple cylindrical parts; I think it just models the drag of an open top/bottom node.

But if it did work as an airscoop: it would complicate the engine design a bit, but why shouldn't it double as a (slightly less efficient than a dedicated) air intake? All it would need is some type of valve system and some extra ducting to channel the air it catches to the jet at the other end.

 

A few questions though, as I am looking at your picture:

• What kind of maneuver do you pull when you switch from jet to LV-N, presumably still less than halfway through the atmosphere where the jet cuts out?
• Once switched to LV-N and flying through the upper atmosphere in the other direction: how do you make the control surfaces on the 'back' of the wings (not directly visible but there's a shadow) obey normal controls for pitch and/or roll, instead of reversed? Whenever I've placed them 'backwards' at the front of a wing, they always act in the opposite way than I want them to - and 'invert' doesn't change that.
• Lastly: why worry about the cone+decoupler on the LV-N 'revoking the SSTO status'... if you have already staged and decoupled other parts (as deduced from the remnant of radial decouplers on the tips of the swept wings)?

[Val]

I'm assuming you're set on only have one airbreathing engine? Because otherwise I don't understand why you'd need to put the LV-N in front for thrust balance.

Anyway, back in KSP 1.0.4 I used to build like this. Craft file

The bent shape of the fuselage keeps the CoT pretty well aligned with CoM, as long as you  transfer fuel to the forward fuselage regularly. Though it does get a little out of alignment when low on fuel.

In 1.0.4 it could get to orbit with 4 km/s dV, but it might not be able to do that in 1.0.5 due to heating issues.

I've built versions up to 26 t take-off weight with 5.5 km/s dV in LKO. Craft file

I hope this can serve as inspiration for your designs.

[Evanitis]

5 hours ago, Edax said:

I could easily stick a air intake or nose cone with decoupler in front of the LV-N, but my spaceplanes would then lose the SSTO status, so I'm wondering if that step is necessary at all?

I'm just nitpicking here, but semantics-wise if you only jettison the cone after you reach orbit, it's still an SSTO. Though it would invalidate a K-Prize entry, so I see why you hesitate.

I always just used two jet engines radially attached to the rocket fuselage, so I never faced your design-difficulty. But I love the idea of a nuke facing other than backwards. It's inefficient, as you can't use that engine to reach orbit. Well, most likely. And there will be drag, unless you use the nose-cone way.

But Kod gave the kerbals moar boosters to cope with inefficiency and drag, so I wouldn't rule out any option because of these... and the silly looks. I'm somehow tempted to test how would a spaceplane fly that has a nuke diagonally slapped to the CoM. We could call it a chimney if anyone asks..

[Snark]

I don't think the game explicitly models engine bells-- I think it just treats them as a rear-facing, flat-ended cylinder, as @swjr-swis points out.

However, rear-facing flat surfaces do create drag.  Nosecones are useful on the rear as well as the front.  So if you want to put a rear-facing disposable nosecone on there until you're out of the thick part of atmosphere, then yes, it should help you.

[Edax]

I see that the very old picture of my non-SSTO is throwing people off (it was towed in orbit and designed to fly backwards 98% of the time [It had detachable fuel tanks because it was 1.0.2 and LF tanks back then were super light and carried little fuel], and only aerobreak and reentry in the forwards position, with the 1.0.5 update, I'm trying to see if I can make an SSTO variant since I don't need too many air intakes), so here is a different example that I've manged to get to SSTO,  Still needs refinement, but I'm working on a MK1 version at the moment, and I'm trying to kill as much drag as possible, while also trying to keep the spaceplane from getting too long that it'll be too difficult to land on other moons.  The example below can get into space so that I can safely flip it backwards to use the LV-N, (but as I'm noticing with 1.0.5, you can't flip over anymore around the 30000-40000 meter zone).

 

Edited February 1, 2016 by Edax

[Edax]

6 hours ago, Val said:

I'm assuming you're set on only have one airbreathing engine? Because otherwise I don't understand why you'd need to put the LV-N in front for thrust balance.

Anyway, back in KSP 1.0.4 I used to build like this. Craft file

The bent shape of the fuselage keeps the CoT pretty well aligned with CoM, as long as you  transfer fuel to the forward fuselage regularly. Though it does get a little out of alignment when low on fuel.

In 1.0.4 it could get to orbit with 4 km/s dV, but it might not be able to do that in 1.0.5 due to heating issues.

I've built versions up to 26 t take-off weight with 5.5 km/s dV in LKO. Craft file

 

I hope this can serve as inspiration for your designs.

This sounds like the solution to my problem, mounting a LV-N facing the back, but I have a heck of a time trying to keep an non-vectored engine balanced enough not to cause spinning. I've only manged to do it with rapiers on a quadruplane, and I expect it's the vector control that keeps it from spinning out of control.

Edited February 1, 2016 by Edax

[Val]

10 hours ago, Edax said:

I've designed spaceplanes with the LV-N in the front for the sake of streamlining and thrust alignment, and I've had some success with it, but the nagging question in my head is, does the engine bell act like a giant air scoop producing massive amounts of drag?  Reality would tell me yes, but I'm not sure if the game models engine bell drag or not.

By my estimation, using a LV-N as nosecone, is no worse than using a Heat Shield or Shielded Docking Port, which are the 2 parts that allow the most efficient ascent profiles (highest airbreathing speed possible without overheating). And the LV-N gives great thermal protection, probably even better than the aforementioned. It also helps distributing the engine mass, making it a lot easier to keep it balanced when fuel is consumed, I guess.

With the exception, that you have to flip the craft when exiting the atmosphere, it looks to be a workable solution, in my opinion.

But I do believe it is more efficient to have, the LV-N helping during ascent, from 18-20 km, even with the added drag of the additional "nose cones", that would be required to have 3 or 4 rearward facing engines.

I'm considering if I can make 2 equivalent crafts, that would prove my theory. They would need to at least have:

• Same engine count and mix.
• Same take-off mass.
• Same wing area and control surfaces.

I'm not sure about fuel mix. My guess is the one with forward facing LV-N(s) would need some or more oxidizer for the RAPIERs, than one with all rear facing engines, because the the latter can in many cases get to orbit without any oxidizer use at all.

Which also makes me suspect that a craft with forward facing LV-N(s) would need to use a much steeper and inefficient ascent profile in order to get up into the thinner atmosphere, to flip around safely and use the LV-N(s).

Just my thoughts.

18 minutes ago, Edax said:

This sounds like the solution to my problem, mounting a LV-N facing the back, but I have a heck of a time trying to keep an non-vectored engine balanced enough not to cause spinning. I've only manged to do it with rapiers on a quadruplane, and I expect it's the vector control that keeps it from spinning out of control.

The idea with those craft is that the uncentered air breathing engine is only used while you still have plenty of aerodynamic control authority. The first one has no oxidizer at all, uses RAPIER only in air breathing mode. The second one only has a small amount of oxidizer, that should run out before the atmosphere gets too thin.

Edited February 1, 2016 by Val
Spelling

[NathanKell]

It's an issue in the drag cube system, it can't tell if somehting is a bucket or a cone, all it sees are the surface normals' dot products to calculate taper.

[Val]

4 hours ago, Val said:

I'm considering if I can make 2 equivalent crafts, that would prove my theory. They would need to at least have:

• Same engine count and mix.
• Same take-off mass.
• Same wing area and control surfaces.

@Edax, I created 2 almost equivalent craft, to test my theory.

SSTO E-2 Starglider Mk.1	SSTO E-3 Starbeam Mk.1
Take-off mass: 30.1 t Wet mass: 26.5 t Dry mass: 13.3 t Liquid Fuel: 2200 Oxidizer: 440 Total fuel mass: 13.2 t Payload: 3.6 t dV in LKO 80 km: ~1890 m/s	Take-off mass: 29.96 t Wet mass: 26.36 t Dry mass: 13.05 t Liquid Fuel: 2112 Oxidizer: 550 Total fuel mass: 13.31 t Payload: 3.6 t dV in LKO 80 km: ~1750 m/s

The LV-N seemed to have much less drag, than both Shielded Docking Port and Heat Shield, but it was not very heat resistant. I ended up having to take a much steeper ascent to avoid overheating. That and the hassle of having to turn the craft around in the upper atmosphere lost some dV. A better pilot than me, a shorter design or more reaction wheels, would help with flipping around faster, to minimize losses.

I could also have brought more oxidizer or tried an even steeper ascent to get my AP out of atmosphere, before flipping, but the conventional design would also benefit from more oxidizer.

The difference is not huge, but for ease of use, the conventional design definitely wins, in my opinion.

 

[swjr-swis]

6 hours ago, Edax said:

This sounds like the solution to my problem, mounting a LV-N facing the back, but I have a heck of a time trying to keep an non-vectored engine balanced enough not to cause spinning. I've only manged to do it with rapiers on a quadruplane, and I expect it's the vector control that keeps it from spinning out of control.

Depending on how you feel about clipping (or the kind of challenges you wish to enter) you could also simply mount the LV-N on the back, mount a RAPIER on the open node behind the LV-N, then offset it forwards until the backplane of both are level so neither one will harm the other when firing. That would keep both engine thrusts exactly balanced with CoM.

(add a shock cone to the back of the RAPIER to close the last open node, then rotate it to face forward and offset forward to get it out of the exhaust area, and you also have air intake and drag reduction)

[Val]

1 hour ago, swjr-swis said:

(add a shock cone to the back of the RAPIER to close the last open node, then rotate it to face forward and offset forward to get it out of the exhaust area, and you also have air intake and drag reduction)

Just as a nose cone in front of a stack can be rotated to have drag as if it was behind the stack. If you rotate an intake mounted like you suggest, it will have drag as if it was in front of the stack.

You'll effectively have the drag of 2 stacks in one, but no rear node drag. That's an increase in drag.

It'd be more beneficial to just leave the rear node open and add a couple radial ramp intakes, or insert a precooler in front of the engines.

[Edax]

4 hours ago, Val said:

@Edax, I created 2 almost equivalent craft, to test my theory.

SSTO E-2 Starglider Mk.1	SSTO E-3 Starbeam Mk.1
Take-off mass: 30.1 t Wet mass: 26.5 t Dry mass: 13.3 t Liquid Fuel: 2200 Oxidizer: 440 Total fuel mass: 13.2 t Payload: 3.6 t dV in LKO 80 km: ~1890 m/s	Take-off mass: 29.96 t Wet mass: 26.36 t Dry mass: 13.05 t Liquid Fuel: 2112 Oxidizer: 550 Total fuel mass: 13.31 t Payload: 3.6 t dV in LKO 80 km: ~1750 m/s

The LV-N seemed to have much less drag, than both Shielded Docking Port and Heat Shield, but it was not very heat resistant. I ended up having to take a much steeper ascent to avoid overheating. That and the hassle of having to turn the craft around in the upper atmosphere lost some dV. A better pilot than me, a shorter design or more reaction wheels, would help with flipping around faster, to minimize losses.

I could also have brought more oxidizer or tried an even steeper ascent to get my AP out of atmosphere, before flipping, but the conventional design would also benefit from more oxidizer.

The difference is not huge, but for ease of use, the conventional design definitely wins, in my opinion.

 

Yeah, I found a steeper descent necessary, which meant the air-engines need an generous amount of acceleration to achieve it (hence why I'm checking for reasons on how to reduce drag).  The crux of my problem is that the LV-N is suppose to be the main engine for travel once the tanks have been refilled, but every oxidizer tank I add seriously reduces the Dv efficiency (but is very useful for actually getting out of the atmosphere), so I'm having a hard time trying to figure out the right balance.  In essence, the spaceplane with the LF tanks and LV-N is the main payload I'm trying to get into space.  I suppose I could try and make the oxidizer tanks detachable so they can be detached in orbit whilst still maintaining SSTO status.  I'm trying to get better then 4000 vacuum Dv with MK1 parts.

[swjr-swis]

@Val You're one of the leading experts in this forum on spaceplane design, so I am very reluctant to dispute your words, but since my own experiences do not correlate with what you stated, I decided to put this to experimentation.

I designed 3 simple crafts, based on the same basic design. Top to bottom:

• Advanced Cone
• 1.25m probe core
• 1.25m reaction wheels
• 1.25m battery
• precooler (20 of 40 units of fuel so it flames out still well within the atmosphere and before heat damage happens)
• RAPIER set to airbreathing and on manual to avoid it trying to switch

01 - 'my' design: a shock cone on the bottom node, and rotated it 180 degrees to face front, clipped into the rapier,

02 - 'your' design: left rear node empty but added a couple (two) radial ramp intakes in 2x symmetry,

03 - for comparisson, just the basic design without either the shock cone or the ramp intakes.

 

Note that in this test, 'my' configuration is at a distinct disadvantage due to the shock cone weight (0.13t vs 0.025t vs 0.0t). I didn't feel like resolving that by craft file editing, it was just a quick test. Note also that due to the precooler, the RAPIER has 100% air needs met in all three configurations from start to flame out, so any differences are not caused by lack of intake (even though the radial intakes only offer roughly 1/8th the air of the shock cone).

I then ran three tests each: throttle to max, SAS on, stage to launch vertically from the pad, and record the max altitude and top speed (F3) as soon as it starts falling back down.

 

Results:

• 01 - top speed 867m/s ±0m/s, max altitude 37716m ±5m
• 02 - top speed 870m/s ±0m/s, max altitude 35897m ±3m
• 03 - top speed 890m/s ±1m/s, max altitude 35648m ±1203m (very twitchy at top speed resulting in large deviations)

Both 01 and 02 are very stable and consistent in results and keep to a pure vertical trajectory, so I didn't see the point of running more tests. 03 however got very unstable past Mach 2, enough so that SAS had trouble keeping it steady, ending the acceleration with a random vector about 1-5 degrees off pure vertical. Best result was still lower altitude than 01, and two results even lower than 02.

 

So I am conflicted. At first sight the top speeds seem to somewhat agree with your statements, but keep in mind 01 is pushing 0.1t more weight.. so the radials being more beneficial is even then very debatable. I am very inclined to believe that the differences are more caused by the difference in weight than the difference in drag. It certainly does not seem that a shock cone rotated is 'effectively the drag of 2 stacks in one'; if anything it's a very marginal difference. And it definitely seems better than using the radial intakes.

I'm open to hearing the math & physics explanation, but with these results I will be inclined to keep using the rotated shock cone rather than the radials, and leave it open only if all my intake air is already resolved otherwise and I am aiming for absolute max speed at the risk of more instability.

I saved screenshots and craft files here, in case anyone else feels like doing a more statistically sound experiment. It's from a pure stock 1.0.5 install, so should be usable by anyone.

Edited February 2, 2016 by swjr-swis
because 1.25m != 2.5m

[Edax]

Looks like I got a MK1 spaceplane with a front LV-N to finally work.  I've dubbed it The Indagatrix.  Figured out that the MK0 LF fuel tanks are very handy at keeping the plane nice and short.  

Whiplash engines and oxidizer fuel tanks are detachable (and cheap), they'll feed oxidizer to the RAPIER until I hit 1800m/s.

The flip over.  Takes about 20-30 seconds to stabilize. 

Once in orbit, the tanks at detached.  (And self-destructed, for the environment!)  And I'm left with a rather impressive amount of LF fuel, though I'm doubtful that Mechjeb is giving me accurate readings, the Dv was the same before and after the stage detachment.

Edited February 2, 2016 by Edax

[Val]

1 hour ago, Edax said:

The flip over.  Takes about 20-30 seconds to stabilize. 

Once in orbit, the tanks at detached.

I can understand going through the flip maneuver if you're trying to SSTO, but if you're going to detach stuff, then why not just attached it behind the LV-N?

Edit: And if you're truly going for reducing drag, then you should avoid fuel ducts and give your wings some Angle of Incidence. AoI can greatly reduce fuselage drag.

 

5 hours ago, swjr-swis said:

@Val You're one of the leading experts in this forum on spaceplane design, so I am very reluctant to dispute your words, but since my own experiences do not correlate with what you stated, I decided to put this to experimentation.

Thanks for the praise and sorry, I didn't link any references. I'd just gone to bed, was on my mobile and couldn't find the post to back up my claim. But I found it now.

@Yakuzi made a very thorough investigation of drag and the effects of nose and tail cones, back in 1.0.4. It is still valid, but has unfortunately been mangled a bit by the forum update. It also doesn't test exactly the case of a reversed intake on a bottom node, but it can be inferred from this quote.

Quote

Inverted rockets constructed from the bottom node of the LV30 engine as proposed by drewscriver did not result in any substantial performance gains over the cone-less control.

Basically he constructs a 2nd stack by flipping a stack mounted on the rear node of the engine, and offsetting it to the side, but it has no noticeable benefit.

What you're suggesting when you flip the intake on the rear node is exactly the same except for the sideway offset. You're creating a second stack drag wise. Sideways offset does not affect drag.

 

Edited February 2, 2016 by Val
Missing words

[swjr-swis]

49 minutes ago, Edax said:

Looks like I got a MK1 spaceplane with a front LV-N to finally work.  I've dubbed it The Indagatrix.  Figured out that the MK0 LF fuel tanks are very handy at keeping the plane nice and short.

And the Juno comes to the rescue. I love the crap outta that little engine.

Good work, Edax.

I see that you are now doing the flip pretty much when atmosphere is practically irrelevant, so the control surfaces on the 'back' of the wings acting weird should have no effect, but I am still curious. Can you check and let me know if they act in reverse to the control inputs? I am still wondering if this is a problem I am only experiencing with my own ships.

 

[Edax]

18 minutes ago, swjr-swis said:

And the Juno comes to the rescue. I love the crap outta that little engine.

Good work, Edax.

I see that you are now doing the flip pretty much when atmosphere is practically irrelevant, so the control surfaces on the 'back' of the wings acting weird should have no effect, but I am still curious. Can you check and let me know if they act in reverse to the control inputs? I am still wondering if this is a problem I am only experiencing with my own ships.

 

I've not experienced reversed control imputs on the Indagatrix , (but I haven't tested it yet with the reversed command probe yet for when I use manuver nodes with the LV-N) but I think I know what your talking about.  In other ships, when I put some tail fins on a strut (placed at the back of the plane) or FAT-445 flaps on a wing, they'd inexplicably work backwards creating downforce when I'm trying to get off the runway, and inverting them doesn't help at all.  Here's an example I made where the tail control surfaces just wouldn't work in the proper direction.

 

[swjr-swis]

10 minutes ago, Val said:

@Yakuzi made a very thorough investigation of drag and the effects of nose and tail cones, back in 1.0.4. It is still valid, but has unfortunately been mangled a bit by the forum update. It also doesn't test exactly the case of a reversed intake on a bottom node, but it can be inferred from this quote.

Quote

Inverted rockets constructed from the bottom node of the LV30 engine as proposed by drewscriver did not result in any substantial performance gains over the cone-less control.

Basically he constructs a 2nd stack by flipping a stack mounted on the rear node of the engine, and offsetting it to the side, but it has no noticeable benefit.

What you're suggesting when you flip the intake on the rear node is exactly the same except for the sideway offset. You're creating a second stack drag wise. Sideways offset does not affect drag.

 

@Val I think we're referring to different things. The case you refer to, building an entire second stack, is a different thing. Your quote is about that one. But as far as I understand it, Yakuzi does in fact show that the single inverted shock intake (and a whole host of other single inverted parts) on the end node of the engine offers a clear benefit over the non inverted one, in the paragraph and chart right before the one you quote:

Several cones were rotated 180° along the horizontal plane as previously mentioned by RIC and evaluated as described above. Note: the Inverted variants were constructed from the terminal end of the LV30 engine and subsequently rotated 180° and translated upward as described by drewscriver earlier in this thread. The average maximum height and standard deviation (black error bars) of the various configurations are plotted below:

 

The comment from RIC he refers to is this one:

Which is exactly what I meant. And the chart clearly shows a much better performance.

 

Btw, even in the second offset stack a la drewscriver, a 'no noticeable effect' sounds better to me than a clear detrimental effect of attaching anything radially, yes? But although indeed a lot smaller effect in that second chart, it still consistently shows a slightly better result for the inverted one than the regular placed one.

So remains the objection of either method being 'exploitative', which will depend on the player and the situation (like challenge rules).

[swjr-swis]

42 minutes ago, Edax said:

I've not experienced reversed control imputs on the Indagatrix , (but I haven't tested it yet with the reversed command probe yet for when I use manuver nodes with the LV-N) but I think I know what your talking about.  In other ships, when I put some tail fins on a strut (placed at the back of the plane) or FAT-445 flaps on a wing, they'd inexplicably work backwards creating downforce when I'm trying to get off the runway, and inverting them doesn't help at all.  Here's an example I made where the tail control surfaces just wouldn't work in the proper direction.

Placing control surfaces so they are .. trailing, for the lack of a better word, compared to the direction of flight, I've so far always been able to somehow convince them to work the way I need them to. Takes some fiddling sometimes but I've always found a way. The final test to make sure always needs to be while at least slightly in forward motion, because what they do in the SPH/VAB or even when at standstill is sometimes not representative of how they will work once moving.

My problem is where I tried to create a craft that works in both directions, and so I needed to add control surfaces that would work correctly both when trailing and 'leading'. Except that when leading, no matter what I do, they will completely refuse to react to my control inputs the correct way and do the exact opposite. So the only workaround I've found is to effectively make them fixed when I need to reverse flight, which unfortunately is a manual thing that can't be pegged to an action group to do inflight (in stock).

In any case, thanks for checking.

[Edax]

18 hours ago, swjr-swis said:

Placing control surfaces so they are .. trailing, for the lack of a better word, compared to the direction of flight, I've so far always been able to somehow convince them to work the way I need them to. Takes some fiddling sometimes but I've always found a way. The final test to make sure always needs to be while at least slightly in forward motion, because what they do in the SPH/VAB or even when at standstill is sometimes not representative of how they will work once moving.

My problem is where I tried to create a craft that works in both directions, and so I needed to add control surfaces that would work correctly both when trailing and 'leading'. Except that when leading, no matter what I do, they will completely refuse to react to my control inputs the correct way and do the exact opposite. So the only workaround I've found is to effectively make them fixed when I need to reverse flight, which unfortunately is a manual thing that can't be pegged to an action group to do inflight (in stock).

In any case, thanks for checking.

Well, I've run another test with a backwards control module, all the control surfaces reverse themselves the backwards control module has the "control from here" selected.  Have you tried that yet?

[Signo]

Back in the days I used to love this kind of design.

 

If you dig the forums for a user called "Kasuha" you will find he even managed a Jool 5 with a "craft" of this kind.

 

That's why my old craft shared his name.

 

 

Edited February 3, 2016 by Signo