FAR + Aerobraking != accuracy

[EdFred]

I know that FAR is based on more realistic drag models, but it seems almost impossible to aero-brake and get a landing location or new orbit even close to predictions in Mechjeb. How are you guys handling landing or aerobraking accurately on Eve, Kerbin, Duna, Jool, and Laythe. I went to Duna and ended up almost 1/4 of the way around the planet off from my intended landing spot.

I know MJ probably uses the stock drag model, so is there something out there that will help get me closer to my intended landing spots or aero-brake orbit?

[Jovus]

There's the Atmospheric Trajectories mod. I don't use it myself.

I usually find with winged craft that taking a decent guess and then circling around the target if I overshoot works fine for me, but I haven't flown at Duna or Eve yet.

[jarmund]

There's the Atmospheric Trajectories mod. I don't use it myself.

I usually find with winged craft that taking a decent guess and then circling around the target if I overshoot works fine for me, but I haven't flown at Duna or Eve yet.

What he said. MechJebs predictions are rubbish when FAR is being used, as the different drag and lift model isn't accounted for. The mod mentioned above, however, takes this into account.

[EdFred]

Thanks guys, I will check it out. I don't put wings on my rockets. I'm a 'brake and chute guy.

[pincushionman]

I know the mod accounts for some FAR changes, but does it use just the CURRENT stage configuration, or does it account for the possibility of (currently-un-deployed) chutes cutting your lateral velocity and bringing your impact point closer?

[xXIndestructibleEVAXx]

It doesn't account for chutes (or staging for that matter), but they normally open so low that it wouldn't really matter.

Edited March 26, 2015 by xXIndestructibleEVAXx

[Starwaster]

Even in real life you can't predict with accuracy where something is going to land. Stock uses the same amount of drag for every single part regardless of orientation, size or mass and is highly predictable, unlike real life. Remember that FAR tries to provide realism and the more realistic its drag model gets, the less accurate landing predictions are going to be, even with the trajectories mod.

As an example, refer to the image below. It shows the predicted landing zones for several Mars probes that we've sent over the past few decades. You'll note that it takes the form of an ellipse. The probes were predicted to land anywhere within the designated ellipses. You'll also note that with time we've gotten more accurate with our predictions as we learned more about Mars atmosphere and aerodynamics and as our technology has improved.

Curiosity had the greatest degree of accuracy yet but it was made possible because Curiosity could 'fly out the errors'. In other words it actively steered towards its desired landing zone.

You're already getting far more accuracy in predictions than NASA got

[Yakky]

I use the Trajectories mod with FAR and have found that it sometimes undershoots quite a bit. Frankly I haven't spent enough time investigating why this is, so I can't offer too many thoughts on whether it's correctable or not. I have learned to manually correct for it based on the type of craft and the steepness of the re-entry trajectory. The undershoot seems to be worse on shallow re-entry angles and heavy atmospheres like Eve.

Note that Trajectories has settings to let you define what angle of attack you're going to have at various stages of the re-entry. It's important to adjust those to get meaningful predictions. Re-entering sideways in FAR yields a very different trajectory than head-first or butt-first.

Trajectories also doesn't seem to understand aerodynamic lift too well if you're flying winged vehicles. While it does take entry angle into account, my conclusion is that it still assumes a ballistic flight path (i.e. only thinks about drag effects, not lift). Somebody correct me if I'm wrong here... but anyway... Re-entering a lifting Space Shuttle type vehicle that can actually glide pretty well in the atmosphere can lead to overshooting the Trajectories estimate.

What Trajectories does very well in any event, however, is help map out the impact of planet rotation when you are still many hours or days away from re-entry. The Trajectories "X" is still going to be in the right vicinity and then you can simply correct based on experience to get a pretty good line-up.

None of this is to rain on the Trajectories mod parade... it is a terrific mod and I'm very glad to have it. What it's trying to do is just very hard to get right.

[smjjames]

With the latest MechJeb dev build, it'll actually attempt to steer itself towards the landing point (like Curiosity did I guess), so, the accuracy is a whole lot better.

[sarbian]

With the latest MechJeb dev build, it'll actually attempt to steer itself towards the landing point (like Curiosity did I guess), so, the accuracy is a whole lot better.

I love it when code that I did not change for month gets better by itself

[Wanderfound]

Trajectory in FAR is strongly affected by the behaviour of the craft during the descent; even a bare capsule can manouvre by changing the angle at which it meets the wind. You can't predict an impact trajectory without knowing what control inputs will be made during the descent.