Maneuver nodes. When do I burn?

[Szkeptik]

So i usually start my burn when there is T-x time left where x is the burn lenght. So my burn will end right at the node. While this works well for the easiest maneuvers, if I plot a maneuver that is not just pro or retro burn, but a more complex one with inclination and eccentricity etc., I will usually end up quite off the mark. The resulting path will be generally right, but never on the mark even though i burn the exact time in the exact direction. I tried burining from T-x/2 to T+x/2 but got mixed results. Starting the burn at the node is always bad.

So am I missing something, or is it normal that you need to do secondary correctional burns after the more complex trajectory changes?

[Liowen]

I usually split the longer times in half. Say your burn is 1m 20s, or 80 seconds, I would burn 40s before the node and continue until 40s after, or until the m/s is near 0. I am normally off by just a bit, but that can be fixed after the burn. In fact it is better to fix your interceptions after you have left Kerbin's SOI as it will take less DV, as you need smaller adjustments in deep space to make big changes to your path.

[Skorpychan]

I usually split the longer times in half. Say your burn is 1m 20s, or 80 seconds, I would burn 40s before the node and continue until 40s after, or until the m/s is near 0. I am normally off by just a bit, but that can be fixed after the burn. In fact it is better to fix your interceptions after you have left Kerbin's SOI as it will take less DV, as you need smaller adjustments in deep space to make big changes to your path.

This.

The nodes assume you have a big punch of dV change right at the node, so they're never going to be completely accurate. So, yes, for anything complex, you will generally need a correction. Computers are stupid and fail to take account of real-world conditions.

Best-practice for me is a big node, and some estimation as you get to the end, followed by another node to tweak things. Or more if necessary; MechJeb often uses three or four nodes to intercept other craft.

[Smidge204]

It's also worth noting that the time estimate is based on past performance. So if you ran your engines previously at, say, half throttle, then your time will be double what it actually should be. Or more commonly, changed the engine/set of engines but haven't used them yet (staging, disabled something, etc), the estimated time will reflect the last engine used and be totally wrong.

It's unfortunately not based on any sort of calculation of the ship's current capability, but rather some obscure measure of past performance.

=Smidge=

[BlackBicycle]

So am I missing something, or is it normal that you need to do secondary correctional burns after the more complex trajectory changes?

As people before me stated, you should basically make it so that your needed delta V (as depicted by the yellow/green scale and the number) is at the half of the scale when you reach "Node in 0 seconds". That means you have burned the half before, and will burn another half after the exact node spot. That should give you the most efficient maneuver.

However:

The calculations are not exact, neither you nor MechJeb are precise in working the throttle, therefore you will (almost) always need a correction burn (usually in the middle of the transfer trajectory, so that it takes less deltaV)

In case you start the burn exactly at the node, you will be incurring some mistake, which will need correction. But this correction is usually not much compared to the overall deltaV you are spending on the maneuver.

[longhornchris]

I've found 2/3 before, 1/3 after tends to work best for me for burns over 20 sec - so for a 30 second burn start at T-20. I also tend to throttle down when I get to around 5 sec left in the burn so I can steer the craft to stay on the node. For larger burns, I generally need a small correction burn anyway.

Once you drop a node onto the system the dV (direction and magnitude) are based on what it takes to get you onto that new orbit, so it does adjust if you miss the node.

[Kasuha]

Half before is usually good enough as in most cases you either don't need to be very precise or need a correction afterwards anyway.

Exception are "target" burns, for instance orbit matching burn for rendezvous, or suicide burn where you put the maneuver right on the surface - in these cases I usually start either full time (easy to calculate) or 70% ahead.

In many cases the best approach is to delete the maneuver in the middle of the burn and adjust your burning to achieve what you wanted to achieve instead.

For ejection burns I usually stop paying attention to the maneuver somewhere about 2/3 of the total dv and instead burn to match timing and position of the SOI intersect.

Edited December 30, 2013 by Kasuha

[BlackBicycle]

I've found 2/3 before, 1/3 after tends to work best for me ...

The only other thing here to take into consideration, of course, is that if you are burning at full power and your vehicle is comparatively small, your TWR will change noticeably throughout the maneuver, so the second half will take shorter than the first one, at the same throttle setting. This is not very often the case though, for a number of reasons that I cannot be bothered to explain, unless specifically asked to (soon, as it's late at night here : ).

[stelarfox]

Hello, there is a way to do maneuver nodes perfectly but that means you have to do a lot of math pluss you need to copy some data as you do not have during flight. besides that all you need to do is to aply some formulas (easier if no separation is needed in the middle of course).

point 1, if the node is less than 10 seconds, most of the time you can easily just do half before / after. (its kind of harder to time it exactly that what you can deal).

but when its about 1 minute total, its recomended do do a better approach.

how yo do it, 1, you need first to know how much your rocket fuel consumes and, the ISP of your engines (you need an extra formula if ussing different engines at the same time).

after that, with the dv that you are aiming for (you can even PAUSE the game at the middle of a node and use just the remaining one if you notice, of course this way will not be that good)

you do not need the aceleration, and you can get the actual mass from the game (visit ship info in the orbital view and you find the actual mass).

with that you know: dv(total)=ISP*ln(Mass/FinalShipMass), even if this may feel weird, we want to know how much fuel we need to use, no matter how many engines.

for this e^(dv/ISP)=mass/FinalMass -> finalMass=Ship Mass/e^(dv/ISP), and more important Fuel UsedMass = FinalMass-ShipMass this means, so the final formula is.

Fuel Used to Maneuver (in KG)=Ship Mass(1-1/e^(dv/ISP)), since dv/ISP is positive, 1/e^dv/ISP will always be less than 1, so the fuel uses will always be positive (wich is impossible to be otherwise).

with this you will even be able to notice if you have enough fuel for it or, if you need to go for the next stage (you can do this multiple times removing the thrown mass if you know it.

now this is in KG, to pass to Liters (or whatever measured the engine shows), you have to actually divide this value by 0.0111111111111 (as far as i know thats a constant).

So Fuel In Liters= Fuel Used to Maneuver (in KG)/0.011111111

now with the fuels in liters we just divide this by how much the engine uses each second

ManeuverTime=Fuel In Liters / (Number of engines * Liquid Fuel used by engine each second)

this last is the TRUE amount of time the node uses the best so far is using half before half after the node (even if really you may not even reach the node) but doing this way it seems to compensate acurately.

now you may need to change things if you started late, or early (early is easy just cut off your engines, basically making a pasuse), late you will have to fix later or change the aiming constantly.

even so in this method i always follow the pointer (except when reaching the green part).

I was unable to find anything that solves this but, its almost impossible for the ship to know when you actually need to do the change so doing manually for me at least is the best.

 

[Kryxal]

On 2013-12-30 at 9:05 AM, Liowen said:

I usually split the longer times in half. Say your burn is 1m 20s, or 80 seconds, I would burn 40s before the node and continue until 40s after, or until the m/s is near 0. I am normally off by just a bit, but that can be fixed after the burn. In fact it is better to fix your interceptions after you have left Kerbin's SOI as it will take less DV, as you need smaller adjustments in deep space to make big changes to your path.

Actually, fine-tuning outside Kerbin's SOI is not the best way to do things for radial or pro/retrograde adjustments ... those are best done as soon as possible.  Adjustments in the normal/antinormal directions, however, are best done about halfway to your destination or at AN/DN.  Combining plane-change with minor refinements is perfectly valid too.  One other good thing about fine-tuning outside SoI is you have one less SoI transition to clutter the screen.

[Clipperride]

You might also want to take a look at the excellent (IMO) Better Burn Time from @Snark

The mod takes into account the change in mass caused by burning propellant as your fire your engines, which gives a far more accurate result.  It's also nice and small so doesn't effect game performance, even on my ageing laptop.  One other thing I find really useful is that it takes account of the engines "Thrust Limiter" setting (one of the tweakables when right clicking an engine). I find that really handy for small velocity changes.

As a bonus it also gives you a closest approach/burn time as you get near another (targeted) vessel and a time to impact/burn time when approaching the ground.

The only thing it doesn't cover (which was a design decision that helps keep the mod small) is staging during a burn.  I avoid most mods as my system can struggle a little playing KSP, but I'm a complete convert to Better Burn Time.

Edit - just noticed the date on the original post, but the above info may well help others with the same issue

Edited November 15, 2016 by Clipperride
Fixed Link and added note

[DrunkenKerbalnaut]

7 hours ago, Kryxal said:

Actually, fine-tuning outside Kerbin's SOI is not the best way to do things for radial or pro/retrograde adjustments ... those are best done as soon as possible.  Adjustments in the normal/antinormal directions, however, are best done about halfway to your destination or at AN/DN.  Combining plane-change with minor refinements is perfectly valid too.  One other good thing about fine-tuning outside SoI is you have one less SoI transition to clutter the screen.

Glad to see this corrected. To add:

If it's a 4 SOI transfer* do all the corrections you can when your reach Periapsis of the second SOI. 

Then, as @Kryxal has mentioned, do any necessary mid-course corrections at AN/DN during third SOI transit. 

*4 SOI Examples:

Kerbin>Mun>Sun>Duna

Duna>Ike>Sun>Kerbin

(Gravity Assists at Mun / Ike)

Edited November 15, 2016 by DrunkenKerbalnaut