A Degree Makes a Difference - or How I Learned To Stop Worrying About Drag and Concentrate on Gravity Losses

[Norcalplanner]

6 minutes ago, Streetwind said:

Thank you! I've been preaching this for forever over in Gameplay Questions, where most other people giving advice still tell newcomers to dial back their thrust, avoid aerodynamic effects and generally fuss a lot about drag losses. Heck no, people! That stopped being a thing with 1.0, get with the times!

What surprised me though was the giant difference between drag and gravity losses that you recorded. I always operated under the assumption that gravity losses were about five times drag on a typical ascent, but your numbers show more along the lines of 20 times. Perhaps that stems from how the stock atmosphere is scaled in RSS? After all it's tuned for FAR, which generally has more drag and also scales differently.

Streetwind, thanks for the props.

The standard atmosphere in RSS was done by @OhioBob  and it's supposed to be pretty good.  Unfortunately, I don't know enough about it to speak articulately on the matter.   

Edited July 11, 2016 by Norcalplanner

[Streetwind]

Ah? Well then maybe the Bob can

[SyzygyΣE]

1 hour ago, Norcalplanner said:

Any chance you could post a photo of your craft?  It sounds like it's all LFO in the first stage.

Oh, I'm using RSS, RO, RealFuels and RP-0 so it doesn't use LFO. The first stage uses Kerosene + Liquid Oxygen and that fuel feeds 5 Vanguard X-405 Boosters. Here's the link to the rocket. Hope this gives you an idea.

Something went wrong with MechJeb and it shows 8700m/s of dV. In reality the "Vessel" block at the bottom left is right. The rocket has 10 000m/s dV.

Edited July 11, 2016 by SyzygyΣE

[Laie]

On 7/10/2016 at 7:04 AM, SyzygyΣE said:

The biggest problem I find is that as the fuel tanks near depletion and mass is reduced, the TWR starts cranking up. I just got into orbit in my RSS and RO career but the first stage has a maximum TWR of almost 9, making it pretty difficult to control near burnout. How I manage to deal with this is to use an engine cluster of 5 instead of a single, powerful engine. When the TWR reaches about 5, I switch off all 4 of the outer engines and leave the central burning, which "resets" the TWR to 1. That way, upon burnout, the maximum TWR changes to a manageable 3.5 with only the single engine. Many of the engines cannot be throttled so I just kind of have to deal with it this way.

Congratulations, you re-invented the Atlas stage-and-a-half scheme. There's nothing wrong about that, and even the performance hit -while real- isn't all that bad.

As an alternative approach, have a look at the Soviet R7 (a.k.a. Sputnik, Sojus,...) which utilizes kerolox boosters in a way like we today would do with solids.

Or, simple but boring, just use the fiver as a first stage and put another single-engine stage on top of it.

Finally, my common advice for newbie RP-0 players: whatever you're struggling hard to accomplish with vanguards, it will be positively easy with just one more level of rocketry tech.

[OhioBob]

9 hours ago, Streetwind said:

Thank you! I've been preaching this for forever over in Gameplay Questions, where most other people giving advice still tell newcomers to dial back their thrust, avoid aerodynamic effects and generally fuss a lot about drag losses. Heck no, people! That stopped being a thing with 1.0, get with the times!

What surprised me though was the giant difference between drag and gravity losses that you recorded. I always operated under the assumption that gravity losses were about five times drag on a typical ascent, but your numbers show more along the lines of 20 times. Perhaps that stems from how the stock atmosphere is scaled in RSS? After all it's tuned for FAR, which generally has more drag and also scales differently.

Several years ago I did a Saturn V launch simulation and I computed drag losses of only 48 m/s, which is a pretty paltry amount.  Of course the Saturn V has a really low TWR, but even for higher TWR launchers, drag losses are pretty small compared to gravity losses.  I agree that people around here tend to make too much of a fuss about drag losses.  My advice has always been to just make the vehicle reasonably streamlined, punch the throttle, blow through the thick lower atmosphere, and don't fret about drag.

As far as RSS's atmospheres are concerned, they are custom developed models of the real life atmospheres.  They are not scaled from stock, nor are they "tuned" to any particular drag model.  The atmospheres' temperature, pressure and density versus height matches real life.  How the game uses that data to simulate drag, lift, heating. etc. is a separate function from the atmospheric model itself.

 

[pandaman]

I must say I have found this thread fascinating and informative.

Although I play stock (with dabbles into HalfRSS recently) I expect the same basic principles apply.  It has caused me to look more closely at my launch profiles and evaluate to try and optimise.

[p1t1o]

Using the Dr. Strangelove "How I learned to stop worrying and XYZ" trope deserves 1 rep in-and-of-itself, here you go

[foamyesque]

This is a question whose specific answers depend a great deal on the characteristics of the rocket. HIgher TWRs allow for sharper tipovers, with the lower gravity losses that result,  but -- with extreme ones -- will run into major aerodynamic challenges easily up to 15km (or higher) and higher drag losses. In addition, assuming reasonable streamlining, the smaller the rocket the higher drag losses will be, given the same TWR and launch profile, a result of the square-cube law. RSS naturally tends to encourage relatively large and low TWR machines on liftoff, hence these results.

[Norcalplanner]

I did one last set of runs before removing FAR and getting back to my career.  For this set, the rocket was identical to the other FAR runs, with the SRBs set at 75% for an initial TWR of 1.53 for all runs - the only variable is the amount of tipover.  I used the exact same technique used in the comparable stock aero version in an earlier post (30 m/s vertical, then tip to the east) to allow a direct comparison.  As with before, MechJeb doesn't tabulate drag losses with FAR installed. 

A Degree Makes a Difference - 1.53 TWR, FAR Edition

Turn Angle	Gravity Losses	Delta V to Orbit	Delta V Remaining
5 degrees	1,357.7 m/s	8,787.5 m/s	1,550 m/s
6 degrees	1,419.2 m/s	8,846.7 m/s	1,490 m/s
7 degrees	1,253.0 m/s	8,694.8 m/s	1,641 m/s
8 degrees	1,196.5 m/s	8,657.9 m/s	1,680 m/s

So the big shocker here is that 6 degrees is less efficient than either 5 or 7.  I repeated the runs for 5, 6, and 7 degrees, and got comparable results each time.  I'm honestly at a loss to explain this.  The only thing I can think of is that shapes of the ascent curves for 5 and 6 degrees lie on either side of some crossover point in terms of drag, atmospheric density, critical mach number, or something else that I don't fully understand.  If anyone well-versed in FAR or real life aero wants to weigh in, I'd love to hear what you think might be going on.

Edited July 12, 2016 by Norcalplanner
Added link to earlier post

[Terwin]

Last night I was launching a scansat towards Jool and I thought I would try out some of what I read on this thread.

Reusing an existing launcher that in the past required circularization be finished by the 0.03 TWR transit engine, I tirned ~ 5 degrees at 60m/s and got to space with ~400m/s left in the launch core.

I also found that trying to turn at 30m/s did not work for this launcher but 80-100m/s worked just fine to still have a few hundred m/s in the launch core in orbit.(not very aerodynamic with a Mk2 core, 2 Mk2 drop-tanks and radiators+panels+scanners+communicatron bolted on the outside)

Stock aerodynamics by the way.

(The first launch was probably more efficient because I tried to hide my scanners and panels in a pair of fairings attached to a reversed docking port, but when I tried to disengage the docking port it apparently decided that it was afraid of heights and would not let go, and while trying to figure that one out I also realized that stock fuel switcher(1.1.3 compatible) does not respect Modular fuel tank(not yet updated) changes and my NERVA powered probe had large stored of oxidizer, so I just reverted and tried again)

 

[MatBailie]

On 12/07/2016 at 8:54 PM, Norcalplanner said:

I did one last set of runs before removing FAR and getting back to my career.  For this set, the rocket was identical to the other FAR runs, with the SRBs set at 75% for an initial TWR of 1.53 for all runs - the only variable is the amount of tipover.  I used the exact same technique used in the comparable stock aero version in an earlier post (30 m/s vertical, then tip to the east) to allow a direct comparison.  As with before, MechJeb doesn't tabulate drag losses with FAR installed. 

A Degree Makes a Difference - 1.53 TWR, FAR Edition

Turn Angle	Gravity Losses	Delta V to Orbit	Delta V Remaining
5 degrees	1,357.7 m/s	8,787.5 m/s	1,550 m/s
6 degrees	1,419.2 m/s	8,846.7 m/s	1,490 m/s
7 degrees	1,253.0 m/s	8,694.8 m/s	1,641 m/s
8 degrees	1,196.5 m/s	8,657.9 m/s	1,680 m/s

So the big shocker here is that 6 degrees is less efficient than either 5 or 7.  I repeated the runs for 5, 6, and 7 degrees, and got comparable results each time.  I'm honestly at a loss to explain this.  The only thing I can think of is that shapes of the ascent curves for 5 and 6 degrees lie on either side of some crossover point in terms of drag, atmospheric density, critical mach number, or something else that I don't fully understand.  If anyone well-versed in FAR or real life aero wants to weigh in, I'd love to hear what you think might be going on.

Just looking at the numbers it seems that something peculiar happened regarding gravity losses.

• The gravity losses report as 61.5 m/s higher
• The dV to Orbit reports as 59.2 m/s higher
• The dV remaining reports as 60 m/s lower

Increasing the turn angle should always (to my knowledge) reduce gravity losses, but you say the following did not vary?

• Initial TWR
• Exact Speed/Height at Start of Turn
• Rate of Turn (how fast/slow you clicked the -1's)
• Apoapsis before Circularisation
• etc, etc?

How gravity losses went up with a shallower ascent I don't know.  (I understand that the maths changes if you're so shallow that you reach orbit within the atmosphere, but that would make 7 and 8 degrees worse too.)

Also, the total losses increased by less than the gravity losses increased by?  So there were savings in Steering Losses or Drag Losses?

You clearly know more about this than I do, and you practised that test a lot, and repeated it.  It still strikes me as human error.

Which is another way of saying "I have no idea"

[Norcalplanner]

17 minutes ago, MatBailie said:

Just looking at the numbers it seems that something peculiar happened regarding gravity losses.

• The gravity losses report as 61.5 m/s higher
• The dV to Orbit reports as 59.2 m/s higher
• The dV remaining reports as 60 m/s lower

Increasing the turn angle should always (to my knowledge) reduce gravity losses, but you say the following did not vary?

• Initial TWR
• Exact Speed/Height at Start of Turn
• Rate of Turn (how fast/slow you clicked the -1's)
• Apoapsis before Circularisation
• etc, etc?

How gravity losses went up with a shallower ascent I don't know.  (I understand that the maths changes if you're so shallow that you reach orbit within the atmosphere, but that would make 7 and 8 degrees worse too.)

Also, the total losses increased by less than the gravity losses increased by?  So there were savings in Steering Losses or Drag Losses?

You clearly know more about this than I do, and you practised that test a lot, and repeated it.  It still strikes me as human error.

Which is another way of saying "I have no idea"

It was a puzzle to me too, which is why I mentioned it.  This only cropped up in the FAR runs, which had additional fins at the bottom (which the stock aero did not).  Since I don't usually play with FAR, I didn't pursue the matter further.  I will say that I considered human error, and that's why I repeated these three tests several times.  Only when I got the same results after repeating the tests did I think that it was something in FAR itself, since the stock aero numbers didn't have anything like this. 

[MatBailie]

1 minute ago, Norcalplanner said:

It was a puzzle to me too, which is why I mentioned it.  This only cropped up in the FAR runs, which had additional fins at the bottom (which the stock aero did not).  Since I don't usually play with FAR, I didn't pursue the matter further.  I will say that I considered human error, and that's why I repeated these three tests several times.  Only when I got the same results after repeating the tests did I think that it was something in FAR itself, since the stock aero numbers didn't have anything like this. 

I'm just puzzled as to how FAR can cause a "bump" in the Gravity Losses rather than the Aero Losses, freaky!

[Norcalplanner]

Just now, MatBailie said:

I'm just puzzled as to how FAR can cause a "bump" in the Gravity Losses rather than the Aero Losses, freaky!

Yep.  It was really annoying with FAR installed, since MechJeb has a function with stock aero to tabulate gravity and aero losses on a launch.  It's buried in the menus, but it enabled a lot better data with stock compared to FAR.

[problemecium]

Can we either get some people to verify that this applies to the stock game or change the thread title to indicate that it's about RSS?

[MatBailie]

6 hours ago, parameciumkid said:

Can we either get some people to verify that this applies to the stock game or change the thread title to indicate that it's about RSS?

The OP makes it specifically clear that this was tested in RSS.

There is, however, another similar post that deals specifically with Stock.  (Same learning points, different exact numbers.)

Perhaps that could be linked to by the OP?

[Norcalplanner]

10 hours ago, MatBailie said:

The OP makes it specifically clear that this was tested in RSS.

There is, however, another similar post that deals specifically with Stock.  (Same learning points, different exact numbers.)

Perhaps that could be linked to by the OP?

Good idea. I'm on mobile atm, but I'll link the other thread in the OP tonight.

[Norcalplanner]

OP has been updated with the link to the stock thread, but the embedded imgur album was a casualty since the forum and imgur don't get along well anymore.  I added a link to the bottom of the OP pointing at the album.

[problemecium]

20 hours ago, MatBailie said:

The OP makes it specifically clear that this was tested in RSS.

I didn't miss that, hence my request to change the thread title.

[DasValdez]

A little necro, but I did anyone specify that this was only true for very aerodynamic rockets... unlike most things KSP people make?  The exact opposite is true for "Kerbal" style rockets with lots of radial parts, unclosed nodes, clipped parts, wing surfaces, etc... if the rocket is kerbal-draggy (not to be confused with *looking* aerodynamic) you'll use a lot more to drag than gravity losses with early pitches.

In addition, I've always wondered if the "drag loss" counter also takes into account off-vector forces like body/wing/control surface lift... or does it just read out what you get from aero gui? That would stack more drag than your charts record based on craft design...

Apparent dV consumption will also vary based on the ISP curve of your chosen engines, engines with a wide ASL-VAC swing like the KR-2L will benefit differently from various trajectories than a solid ASL lifter like the Mammoth. 

Long story short, the best trajectory will vary widely depending on each rocket design, and I generally teach the middle ground between an immediate pitchover and a straight up, left at ABQ.