Persistent Yaw during launch

[Pahimarus]

Hi all,

Been playing for a while. I tend to design my rockets using the same general approach since I've found something that works. My designs seem to work fine with the 1.25m rockets but when I get into the bigger rockets I really have trouble. 

The way I like to launch is to accelerate directly upward to about 100m/s with SAS on and then do a 5-10 degree eastward pitchover. I then turn the SAS off and allow gravity to pull me through my gravity turn, adjusting throttle to keep my TWR ~1.5. Once I hit about 35,000m I turn the SAS back on and pitch over to full horizontal acceleration. On some of my 1.25m rockets this works beautifully allows me to fly a smooth gravity turn all the way to 80-100,000 meters altitude and puts me in a nice 0 degree inclination orbit. On my bigger rockets I have problems with persistent yaw and sometimes roll during the portion of the gravity turn when SAS is off.

Attached is an example design I've been having trouble with. Basically it starts to yaw at about 10,000m and puts me into a highly inclined orbit unless I leave SAS on and fight it all the way to orbit.

I've confirmed COT is midline and CoM is ahead of COL. I've tried more fins, less fins, fins with and without control surfaces with no apparent affect (though I did notice that when I had control surfaces they seemed to be fighting the engine's thrust vectoring and it steered better without control surfaces). I've tried every conceivable symmetry of fins.

At first I was having trouble with the rocket  pointing straight up in flight (basically partway up around 10k meters it swings like a pendulum and starts flying straight up). I've never figured out the cause exactly of this problem conceptually but I encounter it commonly with my rocket designs. It seems some how to be like the reverse of flipping. Adding fairings to complex payloads seems to solve it. 

Then I was having trouble with the rocket flipping so I changed the fuel prioritization to burn the posterior tanks first followed by the anterior tanks and this was solved. 

Nonetheless the northward yaw persists and I can't think of anything else to try to fix it other than scrapping my design. I don't mind starting over but this seems to come up again and again and I'd really love to understand what is happening here. 

Thanks in advance.

Screenshot

**Update**

Thanks all for the many responses. I have done some more extensive testing with very little success so I figured I'd report the results of some experiments.

1.) Tried resetting trim. Trim was already zero'd (flying with keyboard not joystick so no dice). Tried trimming out the yaw. Too finicky.

2.) Tried removing the side booster fins. Tried more fins, bigger fins, w/ and w/o control surfaces. Nothing consistently made a difference. Enough fins would make it fly straight but then it couldn't be steered.

3.) Adjusted the fairing to cover the knuckle in the fuselage. This got rid of the vertical stability issues (flipping, penduluming) but didn't fix the yaw/roll problems.

4.) Got rid of the asymmetrical fuel lines. This did reduce the torque on the stages. 

5.) Carefully removed one part at a time to try to figure out where the S0 and S4 torque in KER is coming from. Guess what, it is coming from the engines themselves.

6.) Tried gimble locking the bobcat and steering with control surfaces. Steers better in SAS but no change to the problems when SAS is off.

7.) Switched to a skipper (in spite of the size mismatch) thinking maybe it was the bobcat. Slight improvement? Not enough to matter.

8.) Tried stabilizing the side boosters with different configurations of struts.

Current thought: could there be inherent torque from the engines and my rocket is just too short thus doesn't have enough of a lever for my fins to counteract the torque? Tried switching to a skipper and adding two additional of the tall 1.875m fuel tanks. And it's...better. Still yaws to the north and rolls a bit but it can fly all the way to 35km when I reengage my SAS while only accumulating a few degrees of inclination. Of course it is way more DV than I need for this payload.

Current

Added query - the reason the original was so short is that the bobcat doesn't have enough thrust to lift more central fuel tanks since it still has a full fuel loadout after the side boosters drop because of the asparagus staging. Is it just not viable as a core stage engine with asparagus staging?

**Final Update**

See my reply to OHara below. Here is the final design.

TL;DR: 

- Fuel lines in radial symmetry introduced asymmetry. They were functionally unnecessary

- Rocket was too short

- Payload was too draggy

- Side booster fins were exaggerating any asymmetric forces that developed during flight

- Bobcat has slightly asymmetric thrust, consider reserving for very stable rockets and/or using SAS

- 4 symmetric fins as far to the rear as possible on the center stage, with control surfaces worked best

Edited July 18, 2020 by Pahimarus
Final Testing results and TL;DR added

[OHara]

Everything about that rocket looks fine to me.  Some accidental asymmetry is usual in KSP (most obviously in aircraft if you let them wheelbarrow on takeoff) and the unwanted yaw that you describe does not seem unusual.

It looks like you used radial symmetry (selected by the pie-slice symbol in the editor) to join the boosters and fuel lines.  When you pitch east, then the starboard side of the craft has the fuel line facing east, while the port-side fuel line faces west, so there is a little asymmetry there.  You might try re-attaching those parts with mirror symmetry (selected by the mirror-image symbol in the editor) to see if you like that better.

Other than that, you might just use SAS or trim out (alt-W or alt-S, then later alt-X to cancel) the yaw.

You might also like to simplify the craft.  Struts, for example, are a little magic in KSP, as one strut fixes parts in all three dimensions, but they weight a lot at 50kg each, so you can use fewer.   Fewer fins work if you can slide them further aft.  Fuel lines are no longer needed in the situation you have them, now that the decouplers will cross-feed fuel.  Maybe some simplification will accidentally give a craft that flies better for you.

Edited July 17, 2020 by OHara
also suggest simplifications

[Streetwind]

Your issue is likely in the fins on the side-mounted boosters. Your craft goes in the same general direction as this one - a short-ish, draggy center core that is stabilized by fins mounted on side boosters which start flexing around the radial decoupler joint as soon as aerodynamic forces come into play. Now, you do also have fins on the center core, which certainly helps, but the fins on the SRBs may still resist active steering and begin flexing that way.

Check out my post in the linked thread for more detail, and a bullet-point list of recommendations that might help with rocket construction. They're not the only right way to do things, but they are a right way. Hopefully you can take something useful away from them

 

(By the way: confirming that the CoM is ahead of the CoL does not help on rockets. You are not concerned about the center of lift, but rather about the center of aerodynamic pressure. That is different, and unfortunately, the game does not have a means of displaying it for you. So basically, you must work with the CoM marker alone, and try to move it upward as best as you can while estimating where your rocket creates drag.)

[jimmymcgoochie]

Have you tried holding prograde instead of switching off SAS altogether? It will make little difference to your launch profile but could help stabilise a wayward rocket without any need to change the design.

[AlexinTokyo]

7 hours ago, OHara said:

Fuel lines are no longer needed in the situation you have them, now that the decouplers will cross-feed fuel.

Without wanting to hijack the thread; is this correct? 

The fuel lines allow fuel to pass one way only, so the outer tanks will feed the inner and outer engines, but once exhausted, fuel from the inner stack will not feed to the outer engines (which can then be discarded along with the empty tanks.
With decouplers set to allow crossfeed, doesn't the fuel flow both ways, meaning that your centre stack feeds the outer engines once the outer tanks are emptied?

Can you get somehow get the decouplers to have the same behaviour as the fuel lines (allowing onion staging with no lines), or am I misapprehending some aspect?

[MrSystems]

Here's another possibility: try rotating the Bobcat 90°. It has some pretty intense thrust vectoring, but only along one axis. Right now it's vectoring on the North-South axis, and it might help to have it on the East-West axis instead. (You could also, instead of rotating it, try turning the Gimbal setting down a bit and see if that helps.)

A third option is to install Kerbal Joint Reinforcement.

[Spricigo]

5 hours ago, AlexinTokyo said:

Can you get somehow get the decouplers to have the same behaviour as the fuel lines (allowing onion staging with no lines), or am I misapprehending some aspect?

Maybe with a MM patch but you are correct on how stock decoupler crossfeed works.

In any case it just means you need to watch the tanks resource level (looking at PAW or something) to know they are empty instead of getting that clue from boosters stopping to boost.

[Meecrob]

12 hours ago, Streetwind said:

Your issue is likely in the fins on the side-mounted boosters.

I second this. Aerodynamically they are too large for their moment-arm from the CofG and will introduce twisting forces. For example using the pic you provided - when you pitch downrange, lift will be generated by the fins on the boosters to point the nose where you want it to go. It will also pitch both boosters' noses further downrange (closer to 0*/horizon) than the core of the rocket AND will make the boosters rotate towards the "dorsal" side of the craft.

 

 

[Spricigo]

Since  KER show  torque >0kN for all stage it is evident you have some asymmetry in your payload. Take a look at what it may be and try to offset it to the centerline, if that is not possible have something to act as counterweight.

[Vanamonde]

Are your other designs having similar trouble? Hit alt-X to zero any trim settings you might have accidentally applied, and if you have a joystick plugged in, make sure it is calibrated to rest at zero. 

[OHara]

On 7/16/2020 at 8:31 PM, Pahimarus said:

5.) Carefully removed one part at a time to try to figure out where the S0 and S4 torque in KER is coming from. Guess what, it is coming from the engines themselves.

The Making History mod did have some engines with misaligned thrust --- the big misalignments have been repaired --- but here the torque shown by KER was only 0.8kN-m from 400kN thrust, so the Bobcat thrust is only mis-centered by 2 mm.
The lift from just the passive action of fins should counter that torque with non-noticeable mis-pointing of the rocket.  Deflection of the steerable portion of the fins would give on the order of 10kN or so on a 20-m lever (a couple 100 kN-m).   Even KSP's reaction wheels give several kN-m (unrealistically strong, but it made a popular game so I won't complain).
You might expect that in vacuum you always need to aim the Bobcat 2mm off the center of mass to compensate, just 0.02° on a 20-m rocket.

I think you're using the Bobcat as intended, as you need only thrust-to-weight around 1-even after the boosters are done. (some player opinions in this thread)

I am stumped why the rocket is pulling north, but I bet someone else will have ideas.  (One side effect of editing your own first post is that the thread is not marked as "having new material", but now I've added a new post so it will jump to the top of everyone's list.)

Edited July 18, 2020 by OHara
Trying the Bobcat now, after you pointed out the misalignment, I can notice the slight pull to the north (with Bobcat in default orientation) and can see how just breaking the symmetry could start a turn and the drag of a large payload fairing enhance it.

[Pahimarus]

11 hours ago, OHara said:

I am stumped why the rocket is pulling north, but I bet someone else will have ideas.  (One side effect of editing your own first post is that the thread is not marked as "having new material", but now I've added a new post so it will jump to the top of everyone's list.)

So I've done some additional testing and settled on a "final" design (do we every really stop iterating lol). This one flies pretty darn straight all things considered. It still has a slight tendency to roll a touch but it is stable enough until I get to an altitude where the aerodynamic forces are reduced such that I don't mind turning the SAS back on and correcting any misalignment.

Some key points (many of the troubleshooting ideas I got from the thread Streetwind linked):

1.) You and others were right, the side booster fins were causing a significant problem. I couldn't tell before because it was just one problem among many but as the design got more refined and I solved other problems this became apparent. Basically what I think is happening is that because the rocket is symmetrical in only one dimension rather than two, even the slightest roll induced by connections wiggling or slight aberrations in my initial pitchover maneuver is then exaggerated by the aerodynamic forces on the side booster fins since the rocket is no longer exactly aligned.

Solution: Removed side booster fins.

Take Homes: Side booster fins = bad. Large side boosters = sub-optimal (especially low in atmo without two dimensional symmetry).

2.) The rocket was inherently unstable for several reasons.

         a.) Too short. Solution: Switched to skipper engine for more center stage thrust, added more fuel tanks in the center stage, added more payload (who doesn't need more DV eh?)

         b.) Payload was still very draggy. Solution: Refined the payload fairing shape further to smooth it.

3.) A word about fins. I ran many, many, many experiments with different configurations of fins. Basically more fins can somewhat stabilize an otherwise unstable rocket (usually draggy payload, COM toward the rear, or short rocket). However, they come at a cost, not just of weight and $$, but also the more fins you have the more straight the rocket wants to fly which reduces your control authority. The more ways you try to add the control authority back the more places there are for something to induce asymmetrical flight. Similar to the issue with the side fins, once the rocket had fewer issues, the difference of moving them just half a fin length to the rear to overhang the engine was noticeable in flight. When appropriately symmetrical, aerodynamic control surfaces do seem to serve better for my initial pitchover than engine gimbal or SAS.

Take homes: Fins are a goldilocks problem, you want them to be just right.

Solution: Four symmetrical delta winglets in the center stage positioned as far to the rear as possible. 45 degrees off from the cardinal directions simply due to the presence of the side boosters in the cardinal directions. 

4.) Bobcat does seem to have more asymmetry than the other engines. This doesn't mean it can't serve well, just that my combination of a tenuously stable rocket and my stubborn insistence on not using SAS during the initial gravity turn made it not a good fit for this particular rocket.

Solution: Changed to skipper for this rocket, added fuel to the center stage.

Thank you everyone for your insights. What a great welcome to the community.

Edited July 18, 2020 by Pahimarus
Edited to reflect linked thread was by streetwind