Finding landing site when descending through atmosphere.

[dlrk]

I'm trying to work out where I will land when descending through an atmosphere. I'm using FAR, so if the use of that makes this more appropriate for add-on affairs, I disagree, but move the topic if you'd like.

FAR will give me the ability to find out the drag and lift produced by my vehicle, and Kerbal Engineer Redux and MechJeb can provide any other necessary info.

I'm trying to figure out how to decide where to place the periapsis when descending through atmosphere.

[Zephram Kerman]

Well, the kerbal way is to 'just try it'. Find out the performance of your craft 'in simulation', then do it 'for real' using those numbers.

Hopefully someone comes up with a better answer than this!

[jaredkzr]

There are so many different factors involved with atmospheric landing that it would probably be easier to just quicksave and use trial-and-error.

There won't be a universal trick to landing at a specific spot on all atmospheric planets since each one has a different atmosphere. You can do a semi-powered descent with rockets and parachutes to help fine-tune your descent but you must be very careful to not lose your parachutes.

[dlrk]

I think I'm going to try checking the lat/lon of the periapsis before atmospheric entry and the lat/lon of the landing site, to find out the change in lat/lon.

[Raptor831]

Set up some tests on a particular body. I had to test a Duna mission a ton "in simulation" before launching officially. Got it down to where I knew from a particular orbit with a particular delta-V whereabouts I would end up. MechJeb isn't all that accurate with atmospheres, and adding FAR into the mix will only complicate things. I'd just place your trajectory a bit long and try to use MechJeb's landing feature for a baseline.

[dlrk]

Does the measuring the lat/lon change make sense?

[dlrk]

I recorded this info from a successful, on target reentry that resulted in a landing northwest of KSC.

65 dv

from

101km

to

35 pe

pe long:

16.01

arg of pe

28.9

inc

1.477

From FAR, I've recorded that the coefficient of drag ranges from .29 to .316.

From this info, is there any info I can collect after or immediately prior to landing that I can use to get a more general landing solution for this vehicle on Kerbin?

[Specialist290]

To put it briefly, the mechanics behind aerobraking are a little difficult to predict without either having a "feel" for it from experience or having a calculator do the math for you.

A while back a fellow by the name of alterbaron crunched the numbers and put together some nifty charts that might help you out with your mission, as well as this demonstration of their effectiveness. Of course, they're more intended for stock aerodynamics, where most of the time the drag coefficients for non-winged craft tend to be close to the same, but it might serve as a helpful jumping-off point.

I've also been helping another user debug an aerobraking calculator that looks very promising at the moment. It's not yet ready for public release, so I can't say too much more about it, but the user in question has used it to perform some analyses here, which should be very informative.

As for information you can collect: In theory, you could just take your craft's current orbital velocity and altitude, your orbit's periapsis altitude (if it's within atmo), your direction of travel (clockwise vs. counterclockwise), and the height and scale height of the relevant body's atmosphere to get a general idea of what the final orbit (or suborbital trajectory) looks like. (Not that this is easy to do accurately by hand, mind you. Believe me, I've tried...) However, predicting an exact landing site takes more information than the stock game really gives you; you need to have (or be able to derive) information about all six of the orbit's basic parameters, as well as the atmospheric data and rotation period of the body you're wanting to land on.

At least, that's if you want to do it mathematically. The best way to do it practically is to just practice over and over again and get a feel for what seems to work best. Set up some kind of target on the ground, get something in orbit, deorbit it multiple times (quicksave / quickload works well for this), and pay attention to what's going on. Eventually you'll start to get an eye for where the ship needs to be in relation to the planet.

Disclaimer: I'm hardly an expert on the subject myself, but I've managed to get a few capsules within eyeshot of KSC before. Like everything, it takes practice and familiarity with your craft and controls.

Edited October 2, 2013 by Specialist290

[dlrk]

Thanks for that very helpful and informative post.

I don't need to calculate very precise landing point, if I can get the landing in between the two coasts of the KSC continent, I can handle it from there. I'll take a look at those links.

How would I use the velocity, altitude, periapsis, direction and atmosphere to get a rough look at the resulting subortial trajectory? That sounds like it would be exactly what I need!

To provide some more info, I'm using a fairly complex vehicle(I'm actually planning on posting it in spacecraft exchange later). It has a Planetes inflatable heatshield, and uses three B9 airbrakes to keep it stable retrograde during the descent while the heatshield is out. After the heatshield is discarded, I can use the airbrakes and engines to control the trajectory more. So, if I can get roughly within those two coasts, I can get a decent landing.

Edited October 2, 2013 by dlrk

[Mirai The Kerbonaut]

here is the answer just be advance of your target landing area simple

[Meithan]

Have you tried MechJeb2's landing autopilot? I haven't personally, and I hear people say it's still a bit buggy.

But that's an autopilot, perhaps you just want a way to predict the landing site to do things "manually". Well, Specialist290 and I have been working on a piece of code that was originally meant to simulate aerobraking, but recently we've also made it work with a promising level of accuracy to predict surface impacts. It wouldn't be complicated to include the option of simulating the deployment of parachutes at a given altitude and have the program include the additional drag. However, at the moment the simulation is 2D, so that would restrict its usability to close to equatorial (zero inclination) trajectories. It would have to be expanded to full 3D for inclined trajectories.

I can imagine that one way in which it could work is this:

1) You input your current orbital parameters (your current speed, altitude and periapsis are sufficient) while out of the atmosphere but on a reentry trajectory, and you gauge (by eyeballing) how far your current orbit's predicted impact point (shown ingame by the blue line, which doesn't account for atmosphere) is from your desired landing site, inputting this distance as perhaps an angular distance (something like: "my current orbit would impact 10° ahead of my intended landing site").

2) The program then simulates the reentry a few times and tries to determine the periapsis that would minimize the actual impact point distance to the desired landing site. This would account for planetary rotation.

3) With this information, you change your current periapsis to what the program reported by means of a burn perpendicular to your current trajectory.

However, this would be feasible right now using the stock aerodynamics. Using FAR the problem is much more complicated, as drag coefficient (edit:) and cross-sectional area depend on ship orientation and geometry I believe. That would require calculating the torques produced on the aircraft, which is a whole different ballgame.

Also, the process I just described, while feasible, is rather cumbersome in that you have to pause the game and switch to a different program (or use a second computer). What would be awesome would be writing a mod that can show the aerodynamic prediction ingame. If anyone has some mod-writing experience (I have none) and believe it's possible to have the game display arbitrary trajectories calculated by a mod, I'd be happy to collaborate to adapt the simulator for that task. I imagine activating the mod and have the game display the predicted airborne trajectory as a dashed line or something, in addition to the current orbit already displayed. Then, as you applied thrust, it could re-compute the prediction in real-time and show it to you.

Edited October 2, 2013 by Meithan

[dlrk]

I actually have tried MJ2's landing autopilot. It doesn't work too well.

Using FAR's flight data, I've found the coefficient of drag for my ship. It's quite consistent, since I reenter with the same orientation. Could that be plugged into the program?

[Meithan]

Only if that's FAR's only change to the aerodynamics. I suspect that FAR also corrects the aerodynamic drag formula to account for mass and cross-sectional area. If such were the case, and assuming your craft's mass doesn't change during the reentry (probably true), we'd need to make sure the cross-sectional area isn't changing, which could again be achieved by maintaining the same orientation relative to the airflow (keeping your craft always pointed in the direction of motion, for instance). If that could also be guaranteed, and assuming those are the only differences in FAR's aerodynamic model, we could use the simulator if we knew the cross-sectional area of your craft. It's a lot of IFs, to be honest.

[dlrk]

Meithan:

My ship's mass doesn't change, and the area does change once, however FAR does give me the different cD for that. I'm using the spacecraft I discussed here: http://forum.kerbalspaceprogram.com/threads/51326-RULOS-ReUsable-Low-Orbit-Shuttle-%281-5-STO-launcher-reusable-2k-DV-vehicle-DR-FAR%29 . The Cd starts at .29, once the heathshield is deployed and the airbrakes are raised it goes to .36.

Since I do have that information, is there any chance we could try it with your simulator?

[Van Disaster]

Mass has nothing to do with aero drag so that doesn't matter. Shape on the other hand does matter, and your shape will depend on your angle of attack which will also affect how much lift you're getting, and the angle of your lift & drag vectors which are way harder to predict & depends on atmosphere density too - at reentry speeds even a minor change in AoA will have a big effect. If you can guarantee not to wobble too much you should get a ballpark prediction though.

[dlrk]

Van,

All I'm looking for is a ballpark prediction. I don't really wobble too much in reentry, and I have the coefficients of drag and lift, as well as AoA from FAR.

With that data, how can I get a prediction?

Thanks

[Meithan]

My ship's mass doesn't change, and the area does change once, however FAR does give me the different cD for that. I'm using the spacecraft I discussed here: http://forum.kerbalspaceprogram.com/threads/51326-RULOS-ReUsable-Low-Orbit-Shuttle-%281-5-STO-launcher-reusable-2k-DV-vehicle-DR-FAR%29 . The Cd starts at .29, once the heathshield is deployed and the airbrakes are raised it goes to .36.

Since I do have that information, is there any chance we could try it with your simulator?

To be honest I'd have to look first at how FAR computes aerodynamic forces to be sure it's what I think it is. Does far compute lift for parts other than wings? Or is there only drag in that case, like in stock KSP?

I guess I could try inputting the drag coefficient, cross-sectional area and mass of your ship into the simulator just to see what it outputs.

Mass has nothing to do with aero drag so that doesn't matter. Shape on the other hand does matter, and your shape will depend on your angle of attack which will also affect how much lift you're getting, and the angle of your lift & drag vectors which are way harder to predict & depends on atmosphere density too - at reentry speeds even a minor change in AoA will have a big effect. If you can guarantee not to wobble too much you should get a ballpark prediction though.

Uhm, please correct me if I'm wrong, but while the drag force is indeed independent of mass (depending only on object geometry and flow conditions), in order to integrate the equations of motion I need the acceleration, and since a=F/m, the mass does matter in the simulation. Picture two ships with exactly the same size and geometry, moving at the same speed in the same air, but with different masses. The drag force will be the same, but the lighter one will experience a higher acceleration, and thus its trajectory will be different.

[dlrk]

Meithan,

I'm almost entirely sure the FAR computes lift parts other than wings, but I'm not sure. In any case, the coefficient of lift it's giving for my ship rounds to .00.

I think it would be a worthwhile experiment to try inputting the drag coefficnet and such. How do I get the cross-sectional area?

[Meithan]

Meithan,

I'm almost entirely sure the FAR computes lift parts other than wings, but I'm not sure. In any case, the coefficient of lift it's giving for my ship rounds to .00.

I think it would be a worthwhile experiment to try inputting the drag coefficnet and such. How do I get the cross-sectional area?

Doesn't FAR report it? I can't run the "real" drag formula without the cross-sectional area. About the coefficient of lift, if it's nearly zero for your craft then I can probably disregard lift, so my calculations should work.

[Van Disaster]

Uhm, please correct me if I'm wrong, but while the drag force is indeed independent of mass (depending only on object geometry and flow conditions), in order to integrate the equations of motion I need the acceleration, and since a=F/m, the mass does matter in the simulation. Picture two ships with exactly the same size and geometry, moving at the same speed in the same air, but with different masses. The drag force will be the same, but the lighter one will experience a higher acceleration, and thus its trajectory will be different.

In terms of computing landing predictions, then of course mass matters however a massier object of the same shape does not cause any more aerodynamic drag. The mass in your calculations is going into the opposite of drag - the paragraph was concerned with FAR's cD and overall drag calcs.

FAR does calculate lift for non-wing items, I'm not sure how it goes about that though.

[dlrk]

FAR is giving me a surface area of 1 square meter. It looks like FAR's surface area isn't quite accurate for unwinged vehicles, but for mine it considers to be one square meter. The coefficient of drag is .36 and the coefficient of lift is essentially zero.

[Virtualgenius]

Mechjebs landing autopilot works you just sometimes have to correct it a little but usually its within 10m to 150m of the target site and i have had it land right on top of the flag as i set this as a target

[Meithan]

In terms of computing landing predictions, then of course mass matters however a massier object of the same shape does not cause any more aerodynamic drag. The mass in your calculations is going into the opposite of drag - the paragraph was concerned with FAR's cD and overall drag calcs.

Ah, I see. I didn't mean to say that the drag coefficient depends on mass. What I was saying is that if I'm to simulate dlrk's craft path through the atmosphere I'll need to know the mass of the ship too (unlike with stock drag, in which case the ratio A/m is assumed to be a fixed constant and thus the trajectory becomes independent of the mass of the craft -- which is unrealistic of course).

[Meithan]

FAR is giving me a surface area of 1 square meter. It looks like FAR's surface area isn't quite accurate for unwinged vehicles, but for mine it considers to be one square meter. The coefficient of drag is .36 and the coefficient of lift is essentially zero.

Assuming that's the area that's going into the drag calculation, then if you tell me the mass of the ship I can run the simulator. This would assume constant coefficient of drag and constant cross-sectional area.

If you want to simulate the parachutes, we'd need to know if opening them changes the cross-sectional area FAR uses (is it computed in a part by part basis?) to compute drag. If they don't, but only change the drag coefficient, then I could simulate them by changing the ship's overall drag coefficient at a specified altitude, and using that new constant value subsequently. To be honest I have no idea whether this will capture FAR's aerodynamics at all, but I can certainly try.

What kind of initial orbit do you have initially?

I could do the following. Assuming that the craft is initially in a circular orbit of given altitude, I can play around setting the periapsis to several values within the atmosphere, simulating the descent, and reporting how short the actual landing site was of the impact site predicted by the game. That could function as a rule of thumb in the game: set up a maneuver node, lower the periapsis to the one specified by the simulation, and move the node until the predicted impact site in the game is X km ahead of your intended landing site.

[dlrk]

Mass is 16723 kg, according to KER. Cd is .36, area is 1 sqM. I'm not concerned with the parachutes, if I can figure out roughly where I will be when I'm at 10k or below, I'm happy.

I'm starting at a rough 200x200 orbit.

If you could do that, that would be awesome. I'm going to try to measure the difference between the periapsis longitude and the landing longitude, and see if that gives a meaningful result as well.

Edited October 3, 2013 by dlrk