Inline Drag Occlusion in 1.2

[Aegolius13]

So pretty much every lunar rocket I've been building in 1.2 has some variation of "small 2.5m tank, Terrier, decoupler, larger 2.5m tank."  I had a general sense that I might be taking some kind of drag penalty for the change in form factor, but was not exactly sure how it worked or how large the effect was.  So, following @Gaarst and @Yakuzi's excellent work on nose cone drag, I whipped up a little experiment.  This was my rig:

 

The nosecone, capsule (I edited mass down to 3 tons so I would actually use it in my new career) and top 2.5m fuel tank represent the payload, and will be providing the drag occlusion, if any.  

The middle piece was my independent variable.  I tried four combinations: (1) Terrier alone, as shown here, (2) Terrier with 2.5m interstage fairing, (3) a 2.5m probe core instead of Terrier (selected because it is the same mass as the Terrier, but in the 2.5 form factor), and (4) the 2.5m probe core with 2.5m fairing base.  On #4, I could not actually build a fairing, I just wanted to check the fairing base itself didn't cause weirdness. I also drained fuel from the upper tank to balance out the added weight of the fairing when it was used (within a couple kilograms, at least).  Just to keep extra variables out, I also disabled the reaction wheel on the probe core when it was used.  

The bottom 2.5m tank and Reliant provided the thrust.  On each test, I set the SAS to hold radial out, and fired the Reliant with only the bottom fuel tank available.  I then saw how high each rocket would go after burning through the bottom tank.  

#1, the plain Terrier, only made it a little over 9,000 meters,

#2, the Terrier with fairing, made it over 11,000 meters.  

#3, the probe core, made it over 13,000 meters!

#4, the probe core with fairing base, ended up very close to #3.  So the fairing base appears to have no special aerodynamic effects.

 

Results:  It looks like parts in front do not fully occlude trailing parts if there is a smaller-size part in between. The lower tank may be getting some drag occlusion, but definitely not as much as if the stack was uniformly 2.5m wide.  So all else being equal, it's better to avoid the "hourglass" shape.  Using an interstage fairing considerably improved drag, though surprisingly, the net effect was not as good as a pure 2.5m setup.  Thus, besides their weight and cost, fairings don't appear to be a perfect option occlusion. 

It's also worth mentioning that, even when I did NOT reduce payload weight to compensate for the fairing, it still made the Terrier package go higher.  This suggests that interstage fairings are probably a good idea, at least as long as the fairing base is staged low enough you're not hauling it to Tylo and back or something. In career mode, though, the fairing might or might not be worth the cost. 

One final note: this test did not attempt to measure drag occlusion when the rocket is not facing directly prograde.  But I would expect the penalty for changing sizes to be even worse in that case, since there's not even perfect occlusion from a geometric standpoint.

 

tl;dr: going back and forth between form factors is bad for drag.  If you have to do it, consider an interstage fairing.  

 

 

[Foxster]

That sounds pretty much like I thought it would go.

The only drag occlusion is if you use same-size parts fixed together or use tapered adapters. You might expect that bigger parts would provide a kind of umbrella effect for trailing smaller parts but that is not modelled currently in KSP.

In the craft above, you'd get pretty similar results using C7 Brand Adapters 2.5m to 1.25m tanks or Rockomax Brand Adapters to a FL-T800 tank. You would lose dV if you don't provide  tapered joint at all or if you used a Rockomax Brand Adapter 02 to bridge the gap - as far as I am aware those are only useful for aesthetic purposes.

Not to be confused with heating occlusion which is modelled differently. It works a little more as you might expect. For instance, during re-entry, a large heatshield at the bottom of a craft protects all the parts above it that are within a cylinder rising from it's circumference, regardless of those parts' form-factor or attachment.   

The combination of the two types of occlusion can lead to some useful effects. For instance, you can put air brakes on the top of a craft and deploy them during re-entry. If they are wholly inside the cylinder of the craft's heatshield then they will not heat up but will provide drag.

Edited October 22, 2016 by Foxster

[Gaarst]

Yeah "hourglasses", in addition of being ugly, are quite draggy and your results don't surprise me.

Though I would tend to think that the 1.25 > 2.5 junction (Terrier to 2.5 tank) causes more drag than the 2.5 > 1.25 junction (2.5 tank to Terrier). I would be interested in result you could obtain by sticking adapters above or below the Terrier and comparing the difference in drag due to their position.

Also, remembering the results I obtained on my tests, you might want to try using larger fuel tanks so that your rocket burns longer. The reason is that with flights averaging around 10km I assume that none did overcome transonic speeds (probably due to fuel and not drag); while in my tests the difference between non-aerodynamic shapes and aerodynamic ones was essentially seen when approaching Mach 1: while draggy shapes never made it out transonic, aerodynamic ones felt almost no effect during the transition from sub- to supersonic. You'd get even clearer results if all the flights get through this transonic regime at one point, seeing the effects depending on the shape.

[moogoob]

Thanks for running the tests! Justifies my love of interstage fairings.

[Tyko]

Thanks for the data! I'd kind of thought this was the case - it looks bad anyway    knowing that making my rocket look 'prettier' also has aerodynamic benefits is great.

[magnemoe]

9 hours ago, moogoob said:

Thanks for running the tests! Justifies my love of interstage fairings.

This, note that in 1.2 you can drop fairing then out of atmosphere and your only cost is then the base. 
I found it economical to stuff even pretty aerodynamic stuff into fairings as legs, solar panels. 

In real world they want to close the gap between wagons on high speed trains as gaps generate lots of drag so you use overlapping plates and rubber skirts. 

[Aegolius13]

18 hours ago, Foxster said:

In the craft above, you'd get pretty similar results using C7 Brand Adapters 2.5m to 1.25m tanks or Rockomax Brand Adapters to a FL-T800 tank.

 

14 hours ago, Gaarst said:

I would be interested in result you could obtain by sticking adapters above or below the Terrier and comparing the difference in drag due to their position.

Also, remembering the results I obtained on my tests, you might want to try using larger fuel tanks so that your rocket burns longer.

Per these suggestions, I did a second round of tests.  For this one, I kept the rig generally the same, but doubled the size of the live fuel tank.  I also kept the monopropellant in the capsule, so I could change the level to fine-tune mass (hey, I finally found a use for monopropellant - ballast!).  And for bonus points, on the last test I subbed out the Terrier for the Poodle, and did not attempt to correct for the Poodle's greater mass -- just to see if the easy solution is to go bigger.  I did not try adapters on the bottom of the top tank for this test - these are all under the Terrier, as before.  

 

Anyway, to the data!  Sorted from best to worst:

2.5m probe core: 33,300m

2.5m to 1.25 fuel tank, plus flat 800 fuel tank (instead of the standard 1600 tank): 27,000m

Rockomax version 1 adapter (the pointy one): 26,600m

Terrier with 2.5 m interstage fairing: 26,400m

Poodle, no mass adjustment: 20,200

Terrier with Rockmax version 2 adapter (the flat one designed for the bottom of a tank):  19,900m

Plain Terrier: 19,300m

 

Interpretation: As before, keeping a perfect 2.5m form factor is the best for drag.  The fairing, pointy Rockmax adapter and angled fuel tank are all in about the same territory, so it might come down to personal preference.  (Though both top and bottom adapters might further improve aerodynamics, allowing those options to surpass the fairing).  The flat Rockomax adapter provided only a nominal improvement over the plain Terrier.  So while it might have some use on the back of tanks (might have to check this in the future), it is NOT a good front-side adapter. 

The real shocker, though, was the Poodle.  Despite adding over a ton of weight, the Poodle version made it higher than the plain Terrier thanks to better aerodynamics. And the mass penalty was probably exaggerated by my relatively small, low-thrust test rig.  Starting TWR  was around 1.15 with the other iterations, but only about 1.05 with the heavier Poodle, meaning the Poodle rig only accelerated 1/3 as fast at first.  With a bigger rocket, you'd have a smaller TWR gap, which would further improve the Poodle's relative performance.  I'm amazed, and may start using more Poodle sustainer stages!

 

 

 

[Foxster]

The other, better, way to get an improvement here is to not use a Mk1-2 Command Pod. It is a very massive way to put three crewmen into space. It also is fat and leads to the use of fat parts below it, which increases drag considerably. 

This is a lighter and less draggy way to loft three crew...

 

[Aegolius13]

19 minutes ago, Foxster said:

 

The other, better, way to get an improvement here is to not use a Mk1-2 Command Pod. It is a very massive way to put three crewmen into space. It also is fat and leads to the use of fat parts below it, which increases drag considerably. 

This is a lighter and less draggy way to loft three crew...

 

Well yeah, I am certainly not suggesting anything close to my test rig for actual use.  But if you're, for example, landing 5 kerbals at time on the Mun, that's probably going to take at least some 2.5m parts.  And the best engine for such a job is likely the Terrier.  

[Foxster]

Sure. I just meant that sometimes the way to fix a problem is not to have the problem in the first place.

Personally I avoid 2.5m stuff at the payload end, especially the Mk1-2 capsule, which is overweight to the point of looking bugged. If I really had to put five crew on the Mun then I'd prefer to use a MK2 Crew Cabin or a couple of Mk1s. 

Edited October 23, 2016 by Foxster

[Aegolius13]

 

8 minutes ago, Foxster said:

Sure. I just meant that sometimes the way to fix a problem is not to have the problem in the first place.

For sure- an all-1.25m rocket would have none of these occlusion problems, plus lower base drag.