Chasing Ap

[RizzoTheRat]

My standard launch profile is to tip over a few degrees when I hit 30m/s, and then follow prograde until the Ap hits 80km.  However deciding what the initial angle should be is something I've never quite got a handle on, and will presumably be a function of TWR and drag.

However most guides recommend using the throttle to change the trajectory and try and keep the Ap a set time ahead of the ship.

As I'm using kOS my launches are easily repeatable so I thought I'd have a play.

This is all with the same ship, an early game 3 stage craft intended for a Mun flyby that is just starting to use it's 3rd stage as it circularises.

I ran a range of turn angles and noted how much dV was left once it has circularised at 80km, then tried the same thing but controlling the throttle so it stayed a set time behind the Ap until the ship reached 40km, where it put the throttle to full until the Ap hit 80km.

9 degrees and 40 seconds overheated and exploded when it throttled up at 40km, and 9 degrees and 45 seconds got pretty hot, so I concentrated of 50s.   11 degrees and 45 seconds, but with the throttle up at 50km instead of 50km, gave the same result as 11 degrees and 50s so not much advantage of 45 over 50.

My big takeaway from this is that throttling back achieves the same orbital dV as a much more aggressive turn angle, so can be used to give a safety factor. 

It's also interesting that it seemed to spend a lot of time at at TWR of about 1.4-1.5, suggesting that could be a sensible design TWR for a second stage, when the first stage has enough power to get the Ap far enough ahead. 

 

Now I just need to come up with a way of estimating a sensible initial turn angle, but initially I think I'm just going to launch everything at 10 degrees as a reasonable compromise.

 

[Vanamonde]

This how-to discussion has been moved to the how-to subforum. 

[paul23]

Using throttle to limit yourself to get a "neater looking path" is really bad idea. A "gravity turn" is better since there are no "steering losses". However steering losses are very very small compared to the gravity losses. So limiting thrust (and thus "burning longer" increases gravity losses. The improvement in steering loss can never offset this.

 

The other loss is of course drag, but even for drag it's better to "steer later and more aggressive" than "lower thrust".

 

Actually a "90 degree initial turn" would be ideal if there would be no gravity. initially one does have a sideway velocity but no vertical velocity, in other words if kerbin would suddenly become a point mass we'd be at the apoapsis of an elliptical orbit, the best way to raise periapsis is not to "travel outwards" but to travel forwards (sideways). The only reason we go "up" first (apart from the very few km for ground clearance is to get out of atmosphere.

Edited December 11, 2019 by paul23

[RizzoTheRat]

4 hours ago, paul23 said:

Using throttle to limit yourself to get a "neater looking path" is really bad idea. A "gravity turn" is better since there are no "steering losses". However steering losses are very very small compared to the gravity losses. So limiting thrust (and thus "burning longer" increases gravity losses. The improvement in steering loss can never offset this.

I'd always assumed throttling back was inefficient too but it appears that isn't the case.  

Both approaches are still doing a gravity turn so the only steering losses are in the initial pitch over, done at 30m/s.  After that it's flying at 0 AoA except for minor yaw course corrections to ensure its pointing East.  Presumably throttling is reducing both the gravity and drag losses as it's moving some of the burn to later in the trajectory where it's more horizontal and the atmosphere is thinner.

However a better optimised rocket is always going to be more efficient than carrying engine capability you're not using.  This is a pretty basic design (only 4th flight in a career game) but I think this has shown me I need a longer burning first stage and a lower powered second stage.  But even with a better optimised rocket it looks like throttling should still save a bit of fuel as you can never design it for a perfect trajectory.

[AHHans]

Interesting study @RizzoTheRat.

I take away from this two things: one is that not turning enough (and early enough) will induce the greatest losses (I guess mostly steering losses during the circularization). And the other is that the your total dV costs fall into a range of 100 m/s, less than 5% of the >3000 m/s that you need to get to orbit. So as long as you keep a somewhat reasonable gravity turn you won't loose too much dV to really worry about.

51 minutes ago, RizzoTheRat said:

However a better optimised rocket is always going to be more efficient than carrying engine capability you're not using. 

Indeed! My personal rule of thumb is (has been already before your study) that a TWR larger than 2 is wasteful.

But that can all be trumped by expediency: it is much easier to just stick one of my standard lifter subassemblies below a payload than fine-tuning it to the task at hand. - Or, now in the late game more likely: stick an oversized lifter onto my payload and go for a high trajectory, than trying to streamline the payload.

[RizzoTheRat]

7 minutes ago, AHHans said:

I take away from this two things: one is that not turning enough (and early enough) will induce the greatest losses (I guess mostly steering losses during the circularization). And the other is that the your total dV costs fall into a range of 100 m/s, less than 5% of the >3000 m/s that you need to get to orbit. So as long as you keep a somewhat reasonable gravity turn you won't loose too much dV to really worry about.

Yeah, it's a relatively small difference, however this is for a pretty small rocket, the payload was probably not much over a tonne.  I need to experiment with wider and heavier rockets when I have the parts.  Bigger rockets tend to have less drag loss but I'm not sure about the gravity losses, and my gut feeling is the saving from throttling should be bigger the more overpowered the rocket is.

I'm thinking launch at TWR of 2 for a first stage that runs long enough to get the Ap out to 50 seconds, and then a sustainer stage that starts at maybe 1.5, which does seem to tie in with a few comments on staging threads I've seen on here.  Dunno if @GoSlash27 is still about but I seem to remember he had some pretty complex design spreadsheets to work out his optimum stage TWR and burn times.  however the TWR is always going to increase so even an optimised 2 stage design might still benefit from throttling.

Once I've unlocked the Clampotron Senior and Vector engine I tend to use VTLO SSTOs , which are obviously massively over powered in the later stage of launch so this could save me quite a bit of fuel.

 

[Fraktal]

From what I experienced, stage 2 doesn't require even a TWR of 1 for the most part. My launch procedure is a 10° pitch-over at 100 m/s, then stick to prograde, full throttle until Mach 1.4 then throttle down to where time to Ap stays between 45 and 50 seconds. This usually leaves me flying near-horizontally at 0.5 TWR for most of the ascent with no noticeable dV loss compared to a more aggressive flight profile and no Mach/heat effects past transsonic. For something like a satellite launch, this means a single Swivel running at 30% thrust.

Edited December 11, 2019 by Fraktal

[AHHans]

1 hour ago, RizzoTheRat said:

Bigger rockets tend to have less drag loss but I'm not sure about the gravity losses, and my gut feeling is the saving from throttling should be bigger the more overpowered the rocket is.

Gravity and steering losses (in dV, not tonnes of fuel burned) are independent of the size of the rocket. A small  and a large rocket with the same TWR, following the same trajectory will need exactly the same dV.

The relative amount of drag also doesn't really depend on the size of the rocket, a well built rocket will have very little drag during a normal gravity turn, i.e. when pointing directly into the wind (= prograde in surface mode). What makes a difference for me is that the stuff that I send up get bigger and bigger in the late game and look less and less like a streamlined pointy cylinder. At some point they don't even fit inside a fairing anymore. Such a payload has large drag, even compared to its mass.

[RizzoTheRat]

32 minutes ago, Fraktal said:

This usually leaves me flying near-horizontally at 0.5 TWR for most of the ascent

Interesting, I'll have to look at what speed mine was at when the first stage burned out, and when it started throttling.  I found it throttled back to about 1.4 TWR and on a version that didn't have enough fuel in the second stage, the third stage with only about 0.2TWR wasn't enough to keep the AP at away. 

I'd guess you're a bit higher or faster than me when you start throttling.  I need to try the same experiment with a bigger first stage and lower thrust second stage, and I'm now wondering if I need to be logging a lot more parameters to see what else effects it.

 

[Superfluous J]

1 hour ago, Fraktal said:

This usually leaves me flying near-horizontally at 0.5 TWR for most of the ascent with no noticeable dV loss

I know why you're not getting any dV losses here: Thrusting sideways you have 0 gravity losses. The only thing you can see is drag losses and you get LESS of those by throttling down, not MORE.

Of course it's better to design your rocket to have 0.5 TWR at that stage naturally, but in KSP with it's lower dV requirements overall that's not always easy.

[CBase]

In science this is known as Goddard problem, although original postulated as question how to maximize final altitude. However once in LKO it is identical to minimize dV required to reach a certain orbit.

There have been published tons of papers and numerical algorithms for this problem and although I skipped over only few there are easy to grab concepts.

In general: more thrust is better. (Read on before comment )

However in order to reduce drag lost, you want to throttle down once you hit a shape depending limit of dynamic pressure (max Q) in a way to stay just below this limit. I did not research for optimal solutions, but noting Q of roughly mach 1 below 3000m works for me. By chance Mechjeb lets you limit on this 

If you watch a Falcon9 launch you will as well notice that exhaust plume gets bigger once they pass Max Q.

The second addition is that burning through implies that you do not follow a gravity turn as raising thrust does not get you into orbit except you coast to a second burn to circulize. If you include this second burn and coast duration into comparison, you will understand why throttling down saved dV in your experiments: the longer the coast, the bigger the circulize burn, the lower in average was your thrust.

So the idea to keep AP close in front is just right, however do not throttle down, but pitch towards horizon to keep time to AP.

The solution to manage heat is changing the time AP is ahead including once you first meet this and turn out of vertical ascent.

Since Kerbin atmosphere does not match Earth, we will not find a solution in science papers I am afraid.

[Fierce Wolf]

Is this a challenge? LKO with the fewer Dv possible

Edited December 11, 2019 by Fierce Wolf

[RizzoTheRat]

1 hour ago, CBase said:

However in order to reduce drag lost, you want to throttle down once you hit a shape depending limit of dynamic pressure (max Q) in a way to stay just below this limit. I did not research for optimal solutions, but noting Q of roughly mach 1 below 3000m works for me. By chance Mechjeb lets you limit on this 

<snip>

So the idea to keep AP close in front is just right, however do not throttle down, but pitch towards horizon to keep time to AP.

The solution to manage heat is changing the time AP is ahead including once you first meet this and turn out of vertical ascent.

I used to throttle to limit MaxQ in earlier versions of KSP, but my understanding now is that it doesn't matter all that much as the drag is much lower in more recent versions.  I've not messed about with it recently although it can be useful for keeping really draggy payloads stable.

Interesting point on pitching down, but on my current profile I'm only a few degrees off horizontal once I hit about 35-40km.  I guess that would be pretty easy for me test the same way by pitching to horizontal once above a certain altitude, the deflection would be fairly small but could make a big difference to the burn time, thus reducing gravity losses a bit more.  As for heat, I don't think kOS can monitor component temperatures so the only way to it automatically would be to have a thermometer somewhere near the nose. 

Although as it's related to skin friction and convective cooling presumably it's going to be related to dynamic pressure, but I have no idea how much heat it would shift through convection to the rest of the ship.

56 minutes ago, Fierce Wolf said:

Is this a challenge? LKO with the fewer Dv possible

I was starting to think that might be an idea too, however I'm not sure the best way to do it.

• Uploading a stock ship for people to try and get to orbit for the least fuel consumption would mean the ship is optimised for the profile I fly in terms of TWR and dV per stage, but easy to compare as you just measure the amount of fuel remaining.
• But defining a payload and having people build their own ship means people using different engines so much harder to compare,  can't just go on dV as it's going to depend on the altitude the individual engine starts at.

Edited December 11, 2019 by RizzoTheRat
spelling again

[Nicias]

You might want to look at this mod:

and this thread:

I tend to go with a start TWR of 2 for a SRB +drop-tank first stage, then 1.5 for a second stage and 1 for the third stage. I'm working at 10x rescale using SMURFF, so I need quite a bit more DV to get to orbit (~8000) I also found via simultation, that (ignoring atmosphere) an optimal gravity turn uses almost the same dv as porpoising to maintain time-to-ap. (what GravityTurn will do) So, once i get out of the lower atmosphere, I don't mind if I end up porpoising.

[Laie]

12 hours ago, RizzoTheRat said:

Now I just need to come up with a way of estimating a sensible initial turn angle, but initially I think I'm just going to launch everything at 10 degrees as a reasonable compromise.

I don't know how you "launch at 10 degrees" so I can't say whether that's sensible or not. Personally, I've got good results with "have prograde be 70° by the time I reach 200m/s" approach (example).

That inherently adjusts for TWR (slower rockets make a slower turn) -- not perfectly, but well enough that one set of values suits a wide range of rockets.

[CBase]

8 hours ago, RizzoTheRat said:

As for heat, I don't think kOS can monitor component temperatures so the only way to it automatically would be to have a thermometer somewhere near the nose. 

Actually once you measure it, you can only throttle down to avoid RUD. The idea is more to optimize next run by partly increasing time to AP, which means you get higher before reducing time to AP and picking up speed.

[Skorj]

I've had to learn to make a much more aggressive initial turn than I had thought was right.  For the TWR 1.65 ship I just built, for example, it's at least 20 degrees at 40 m/s.  Less for lower TWR.  I tune it to see a few temp gauges, but nothing getting high up the gauge.  Once I get to 50s to Ap, I throttle back to hold 50s for a few seconds, then I can just leave the ship alone to 100 km.  This gives me pretty good results, and is easy enough to fly.  I flip from prograde surface to prograde orbit at 24 km.

I avoid any approach where I have to pay too much attention to "this angle at that attitude", as anything beyond the initial turn, prograde, and throttle down at 50s to Ap now seems inelegant.

Edited December 12, 2019 by Skorj

[RizzoTheRat]

8 hours ago, CBase said:

Actually once you measure it, you can only throttle down to avoid RUD. The idea is more to optimize next run by partly increasing time to AP, which means you get higher before reducing time to AP and picking up speed.

Yeah but you should be able to throttle back a bit to maintain a temperature,  and as the altitude increases the skin friction will reduce and you can speed up again.   KER gives a nice read out of critical temperature,  as a percentage,  of the part closest to its maximum temperature,  but I don't think kOS can access it.

[paul23]

On 12/11/2019 at 9:00 AM, RizzoTheRat said:

I'd always assumed throttling back was inefficient too but it appears that isn't the case.  

Both approaches are still doing a gravity turn so the only steering losses are in the initial pitch over, done at 30m/s.  After that it's flying at 0 AoA except for minor yaw course corrections to ensure its pointing East.  Presumably throttling is reducing both the gravity and drag losses as it's moving some of the burn to later in the trajectory where it's more horizontal and the atmosphere is thinner.

However a better optimised rocket is always going to be more efficient than carrying engine capability you're not using.  This is a pretty basic design (only 4th flight in a career game) but I think this has shown me I need a longer burning first stage and a lower powered second stage.  But even with a better optimised rocket it looks like throttling should still save a bit of fuel as you can never design it for a perfect trajectory.

The difference with a gravity turn is indeed not the "steering" - however the gravity loss is similar (equal) to how you try to do a transfer when not doing instantaneous burns at peri/apoapsis. Rather you do them starting from the half pi mean anomaly point up to the apoapsis.

 

I wonder now: you do take the remaining delta-V after circularizing not after reaching apoapsis right?

[Laie]

On 12/11/2019 at 2:22 PM, Fierce Wolf said:

Is this a challenge? LKO with the fewer Dv possible

On 12/11/2019 at 3:12 PM, RizzoTheRat said:

I was starting to think that might be an idea too, however I'm not sure the best way to do it.

• Uploading a stock ship for people to try and get to orbit for the least fuel consumption would mean the ship is optimised for the profile I fly in terms of TWR and dV per stage, but easy to compare as you just measure the amount of fuel remaining.
• But defining a payload and having people build their own ship means people using different engines so much harder to compare,  can't just go on dV as it's going to depend on the altitude the individual engine starts at.

I'd say: go for it.

Provide a ready-made rocket and ask people to optimize the launch profile for this particular thing. There may be only a few participants, but I for one would like to give it a try. Depending on how unhappy I am with your LV, I may swap it with one of my own design after a few tries... it would be good if you gave a few requirements for the LV. Probably lowest cost, but does it need to be reusable? Tech level limitations?

[Empiro]

Some parts of these results are definitely surprising to me. The full throttle being lower than the other two in particular seems to suggest that drag losses are significant when moving too quickly, yet the fact that the more aggressive turns did better seems to imply that minimizing gravity losses is still the way to go. Can you give us an idea of how far the AP went ahead?

I think the biggest takeaway here is that you really don't need very high TWR to be efficient.

[Norcalplanner]

I'm glad that people are still reading that older thread occasionally. 

[Superfluous J]

2 hours ago, Empiro said:

I think the biggest takeaway here is that you really don't need very high TWR to be efficient.

Actually it's more than that. You really DO need a low TWR to be efficient.

[RizzoTheRat]

11 hours ago, Norcalplanner said:

I'm glad that people are still reading that older thread occasionally. 

I'd not seen it before so found it an interesting read, in particular how low your launch TWR is.  My test ship launched at about 2:1 but I reckon I should repeat my tests with the solid boosters limited a bit too to see how that works.  I assume all your comparisons used Vacuum dV?

I got a bit distracted by another kOS problem the last few days (rendezvous with a target in eccentric orbit) so not revisited it again yet.  I'm thinking I might try an do a few launches on different profiles with each new ship I build and see how I can refine my designs.

 

12 hours ago, Empiro said:

Some parts of these results are definitely surprising to me. The full throttle being lower than the other two in particular seems to suggest that drag losses are significant when moving too quickly, yet the fact that the more aggressive turns did better seems to imply that minimizing gravity losses is still the way to go. Can you give us an idea of how far the AP went ahead?

I think the biggest takeaway here is that you really don't need very high TWR to be efficient.

I think it's still more about the gravity losses, by reducing the throttle I'm making the burn last longer,, so it's increasing the horizontal component of the burn while reducing the vertical component.

 

 

[RizzoTheRat]

22 hours ago, paul23 said:

The difference with a gravity turn is indeed not the "steering" - however the gravity loss is similar (equal) to how you try to do a transfer when not doing instantaneous burns at peri/apoapsis. Rather you do them starting from the half pi mean anomaly point up to the apoapsis.

 

I wonder now: you do take the remaining delta-V after circularizing not after reaching apoapsis right?

Yes, remaining dV once it's circularised at 80km, although my circularisation script isn't perfect and on steeper ascents where a longer circularisation burn is needed it can end up a km or two low.