OLD! Explosions with Hypothesis, CLOSED. But mods, keep this open for discussion.

[Ekku Zakku]

please imbed the images using: [ IMG]http://i.imgur.com/joePYfl.png[ /IMG]

Ignore the spacing between [ and IMG] and [ and /IMG]

You mean [noparse]? Just replace "img" with "noparse" and you can post BBCode tags no problem! BOLD! ITALICS!, see? Quite useful =3 [/noparse]

[Whirligig Girl]

good to know.

[KevgloB]

My first post on the KSP forums. Here I will write on my work building rovers with rockets on them for leaps over obstacles and so on. This started as an attempt to recreate the M35 Mako from mass effect, but changed into an attempt to build a flying rover.

Question: Is it possible to build a wheeled rover with some vertical capability?

Difficulty 1: The biggest problem is (especially in my attempts to build the Mako), if the thrust is not perfectly lined up with the center of mass, the rover simply flips over, as the off-center thrust creates angular force rotating the rover.

Solution: Building more symmetrically largely eliminates this difficulty, but I have found that by using A.S.A.S. coupled with RCS thrusters it is possible to build asymmetrically. Balancing by moving around interior components (generators, batteries and so on) is possible, but be careful that the fuel tanks are balanced so that as you use your fuel you don't unbalance your rover.

Difficulty 2: From my testing, it seems that even small vertical forces on rover wheels causes them to break, making a "hopping and rolling" rover difficult to maintain without downtime replacing the wheels between jumps.

Solution: I have found it is possible to mount landing legs directly onto the wheels (I was using ruggedized wheels). These legs, while on the wheels, won't rotate with the wheels. The legs can be folded down while in flight or on the ground, and can be retracted, putting the rover back onto its wheels once one has landed. This gives a much larger margin for landing speeds.

I believe that this idea has some practical use, for jumping over obstacles, as well as some game-use, such as mapping the engines to allow flips and other stunts. I have found my rover capable of jumping on Kerbin and the Mun, and for a model capable of jumping on high gravity worlds, simply adding more engines is easily done.

Thank you for reading my submission,

Kevglob

[Whirligig Girl]

Interesting entry, but it does not show the scientific method. There are no tests, or at least no pictures of tests done to the rover.

[Agent30632]

GregroxMun, I updated my post with another CSM+Lander Duo test. Please say if I need to do more tests.

Thanks.

[Whirligig Girl]

That's much better, thank you.

[tssn1611]

I plan to submit an entry about the most efficient landing profile from Munar orbit (assuming you don't really care where you land). Preliminary results: MechJeb "Land at Target" takes 755 m/s of delta-V from a 30 km orbit. More to come tomorrow (it's 4:45 AM here ...)

[tssn1611]

Efficiencies of Munar Landing Trajectories

tssn1611 and Edfel Kerman

Submitted to the Journal of Kerbal Spaceflight

Introduction

Achieving an efficient landing trajectory on a Munar mission is important for mission efficiency, cost and ultimate crew survival on a mission. Which trajectory is most efficient has been somewhat up for debate, however. Some prefer a steeper trajectory, even going so far as to eliminate all horizontal velocity on the deorbit burn and focus on destroying vertical velocity on final landing approach. Others prefer a shallower trajectory that removes horizontal and vertical velocity at the same time. The purpose of this entry is to offer insight into which of these landing profiles is most efficient.

Craft and Procedure

The craft used in these experiments is a Betelgeuse A lander (see Figure 1) manufactured by Koeing Spacecraft Corporation. It has a mass of 8.48 tons fully loaded and has 3832 m/s of ∆v, enough to conduct missions with potentially large ∆v requirements and ensure crew return. Its main descent engine is an LV-909 model manufactured by Jebediah Kerman's Junkyard and Spaceship Parts Company, whose vacuum I_sp is 390 s. In addition, it has MechJeb 1.9.7 installed for automated maneuvers. The craft was launched by a Klaturn booster on Year 1, Day 131 at 2317:49 UTK and placed into a 75 km Kerbin orbit. After booster jettison, a Hohmann transfer orbit to the Mun was initiated, and the Betelgeuse was placed in a 30 km x 30 km Munar orbit.

Figure 1: Photograph of the Betelgeuse landing craft.

From this 30 km orbit, various tests were conducted varying the size of the deorbit burn. The spacecraft's MechJeb module was used to keep the craft pointing retrograde for the duration of the burn. After deorbit burns were complete, MechJeb's landing autopilot was engaged for the remainder of descent. After each descent, the Almighty Quickload was invoked to bring the Betelgeuse back to its 30 km orbit in preparation for the next test. The sizes of the deorbit burns ranged from 50 m/s to 500 m/s. Additionally, a test was conducted using a "skimming" burn, one that barely lowers the periapsis of the orbit below the surface (12 m/s from a 30 km orbit), and a test was conducted in which the deorbit burn removed all velocity relative to the Munar surface (521 m/s at a 30 km orbit). Finally, a test was conducted using the MechJeb's "Land at Target" feature, which performs its own deorbit burn.

Each trajectory was scored by its ∆v used, as obtained from MechJeb's Vessel Information readout. It is known that the elevation of the landing site has an effect on ∆v requirements for the landing. For these experiments, the test was repeated until a landing site was obtained that is near 1500 m elevation, though the actual landing site may differ in elevation by up to 250 m for reasons of pilot sanity.

Results

In examining Figure 2, it is apparent that the lower ∆v deorbit burns give a lower ∆v for the overall landing maneuver. However, the effect is non-linear such that deorbit burns of less than 250 m/s do not use significantly different amounts of ∆v. The KHV burn is easily the least efficient of all the three, requiring 852 m/s of ∆v, and as such, that strategy should be avoided when landing on the Mun. The "Skim" burn and the 50 m/s burn are the most efficient, requiring 682 m/s of ∆v. The difference between the most and least is 24.9%, which is a significant difference in ∆v.

Figure 2: Plot of ∆v as a function of burn size. "Skim" denotes a skimming burn (12 m/s), and "KHV" denotes a burn to kill horizontal velocity (521 m/s).

As a comparison, the MechJeb "Land at Target" feature used 755 m/s ∆v, which places it between 350 m/s and 400 m/s burns. However, its landing strategy was less efficient overall, as it was constantly running the engine at at least a low level. Additionally, on final descent, it forced the craft into large yaw oscillations about its intended trajectory, which wasted fuel.

Conclusions

We have demonstrated that, at least for Munar descent, shallower trajectories are more efficient. However, for deorbit burns less than about 250 m/s, the difference in efficiency is relatively minor. Conversely, there is a large difference in efficiency between the most efficient ("Skim" deorbit burn) and least efficient ("Kill Horizontal Velocity" deorbit burn). For future work, we plan to try different starting orbits and descents different bodies. Of particular interest are Duna and Laythe, as they have atmospheres that help to slow the craft.

[Agent30632]

And I was expecting more competition. Come on people, you can do it! There's only 6 entries!

[kahlzun]

I have an idea that I am working on:

The Progressive testing on the effect of increasing TWR on atmospheric launches

I'll see if I can post something later on.

Also: I created this a while ago, is it of any use?

Experiment with airship parts: optimal launch assist

Edited July 14, 2013 by kahlzun

[Aqua667]

I'm working on an entry:

Study of the Upper Limits of Parachutes

Tests are done on Kerbin with various chutes and crafts of various sizes. This probably goes in rocketry or general.

Basically I'm launching dead weights of different masses (1 t, 2, 3, 5,) with either 1 standard chute or 2 radials, and if I have time maybe a drogue chute. Launches are always conducted with a single small solid booster.

Edited June 30, 2013 by Aqua667
more details

[kahlzun]

EXPERIMENT BRIEF 1

RESEARCHER:

Samuel Birchenough “Kahl'Zunâ€Â

EXPERIMENT TITLE:

Progressive testing on the effect of increasing TWR on atmospheric launches

HYPOTHESIS:

Increases in efficiency when launching with high-TWR vehicles will eventually reach a peak level of efficiency, and provide minimal benefit after this point.

METHOD:

A rocket has been designed with a range of engines which, while otherwise identical, have differing thrust values, allowing fine control of the TWR of the rocket while maintaining all other variables (Isp, mass, drag etc) as constants.

This will allow us to isolate the true effect of changing the TWR of the rocket, and will allow us to make more informed decisions when constructing launch vehicles.

The fuel for the engine is sufficient to provide approx 6000ms of dV, which should allow all designs to reach LKO, defined here as 100km.

Mechjeb 1.9.8 will be used to automate the launches, with settings as follows:

Turn start: 10km, Turn End: 70km, Ascent angle: 42 degrees, inclination 0)

CONTROL:

As Mechjeb will be used to show the dV consumed, we will need to calibrate the instrument.

To do this, I have recreated the 0.17 Aerospike engine (250kN, 390Isp) to provide a stable Isp throughout the launch, and allow us to check Mechjeb's figures.

Fig 1.

TEST 1:

Rocket from Fig 1 has been fitted with Toriodal Aerospike OS, which has 250kN thrust and an Isp consistant across both atmospheric and vacuum flight of 390.

Fig 2.

Rocket on launchpad

Fig 3.

Initiating gravity turn

Fig 4.

Rocket 'coasting' to apoapsis

Fig 5.

After circularisation, with 'stats' shown.

Some troubling results begin to present themselves:

Mechjebs 'Ascent Stats' gives the total dV burnt as 4559, however the dV at launch was 5966, and which means there should be only 1407 ms remaining. Vessel information clearly shows 1410 remaining.

Furthermore, adding up the figures on 'Ascent Stats', we arrive at a different total:

Gravity Losses: 1875

Drag Losses: 611

Steering Losses: 42

Speed Gained: 2043

Total: 4571

Also, when the numbers are calculated by hand, first using Mechjebs weight info:

Launch Mass: 20.26T

Final Mass: 6.16T

Ratio: 3.28896

LN: 1.19057

(390*9.81): 3825.9

Result: 4555.0

Then calculating from hand:

Fuel remaining = 171F 209O.

Because they have the same density (0.005U/T), they can be treated as one fuel: 380U.

380*0.005= 1.9T, plus 'dry mass' of tank (2T) = 3.9T

Launch Final

Mk1 landercan: 0.60T 0.60T

X3200 tank: 18.00T 3.90T

Standard Canard (4) 0.16T 0.16T

Aerospike 1.50T 1.50T

Total Mass: 20.26T 6.16T

These weight figures match those provided by mechjeb, so the previous calculation of 4555.0 was correct.

Conclusion:

Mechjeb is not accurate at providing dV information. There is a variance of almost 20dV between the 3 figures (hand, Ascent stats, Intial-remaining), which equates to an accuracy level of ~0.43%.

END EXPERIMENT 1, MORE TO FOLLOW.

Edited July 1, 2013 by kahlzun

[kahlzun]

I'm working on an entry:

Study of the Upper Limits of Parachutes

Tests are done on Kerbin with various chutes and crafts of various sizes. This probably goes in rocketry or general.

Basically I'm launching dead weights of different masses (1 t, 2, 3, 5,) with either 1 standard chute or 2 radials, and if I have time maybe a drogue chute. Launches are always conducted with a single small solid booster.

I have often wondered if there is any difference in the behavior between the Mk16 and Mk16XL, as they state the same drag value on the tooltip.. Is there any way you can expand the test to also check this?

[SpaceOddity]

I hope it's not too late. Here is my entry:

Various staging methods as means of increasing the efficiency of single-engine rockets within SIMPL (standard) and advanced (FERRAM) aerodynamics model.

https://docs.google.com/file/d/0B3DkWZBXa0eRM0ZFVk1WbFhubTA/edit?usp=sharing

It's written from the in-game perspective, but it should fit the requirements.

Edited July 1, 2013 by SpaceOddity

[tavert]

I have often wondered if there is any difference in the behavior between the Mk16 and Mk16XL, as they state the same drag value on the tooltip.. Is there any way you can expand the test to also check this?

FYI, the drag value from the tooltip is per unit mass. So the difference between the Mk16 and the Mk16XL is the latter is more mass of high-drag parachute. Hopefully Aqua667 can do some experimental tests to verify this.

Regarding your post and delta-V variance in MechJeb, this is entirely expected due to the different methods MechJeb uses to calculate the different delta-V readouts. The Vessel Information delta-V numbers are based on the rocket equation, and (for multi-stage rockets) a fuel flow/staging simulation. The ascent stats delta-V numbers are calculated by numerical integration, which always incurs slight errors due to the finite step size of the integration (and I don't think MechJeb's using any particularly fancy high-order method here). And be careful simply adding the gravity, drag, and steering losses, as each of these is computed by its own numerical integration. When you add these numbers, you're combining the inaccuracy of each of these separate integrations, in addition to roundoff error from the readout only indicating to the nearest 1 m/s. 0.43% is reasonably small, as these types of errors go.

These entries are getting good though, it's nice to see experimental data verifying many of these efficiency concepts.

Edited July 1, 2013 by tavert

[Whirligig Girl]

Your daily Bump, Delivered.

[kahlzun]

FOLLOWING FROM PREVIOUS EXPERIMENT

"Effects of increasing TWR on dV required for surface-to-LKO launch"

It's probably pretty obvious, but I think this goes under "Rocketry". I am open to submission under any category, however.

Javascript is disabled. View full album

In order to normalise the results, I will be using the same model of engine (LV-909) and adjusting the thrust to match a TWR figure previously calculated:

00.60T Mk1 Command Pod

18.00T X1600 fuel tank

00.16T Standard Canards

00.50T LV-909

---------

19.26T Total

Multiplying this by 9.81 (standard gravity) we get the following TWR equivalences:

[table=width: 500, class: grid]

[tr]

[td]TWR[/td]

[td]Thrust (kN)[/td]

[/tr]

[tr]

[td]1[/td]

[td]188.9406[/td]

[/tr]

[tr]

[td]1.5[/td]

[td]283.4109[/td]

[/tr]

[tr]

[td]2[/td]

[td]377.8812[/td]

[/tr]

[tr]

[td]2.5[/td]

[td]472.3515[/td]

[/tr]

[tr]

[td]3[/td]

[td]566.8218[/td]

[/tr]

[tr]

[td]3.5[/td]

[td]661.2921[/td]

[/tr]

[tr]

[td]4[/td]

[td]755.7624[/td]

[/tr]

[tr]

[td]4.5[/td]

[td]850.2327[/td]

[/tr]

[/table]

CONTROL:

Using the 'worst-case scenario', the first launch will have the bare minimum for success: starting at 1.00 TWR. (188.9406 kN)

Stats:

Standard launch mass (SLM): 19.26T (will remain static across all subsequent tests)

Final mass: 4.85T

dV expended: 4948 (As per ascent stats)

As the mass Mechjeb displayed in ascent stats in the previous 'proof-of-concept' test was accurate, will use that for 'final mass'

TWR 1.5:

SLM.

Final: 5.63 T

dV: 4472

TWR 2.0:

SLM

Final: 5.67 T

dV: 4450

(Already we are seeing a much lower increase of efficiency)

TWR 2.5:

SLM

final: 5.65 T

dV: 4467

(Surprisingly, there is an actual DECREASE in efficiency??)

TWR 3:

SLM

Final: 5.62 T

dV: 4483

Even greater decrease in efficiency!

At this point I got bored...

TWR 100:

engine launched off the ship

TWR 48.38:

engine launched off the ship (no picture)

TWR 26: (highest stock possible)

engine launched off the ship (even when slowly incrementing thrust)

TWR 3.5:

SLM

final: 5.61

dV: 4494

TWR 4:

SLM

final: 5.60

dV: 4502

TWR 4.5:

SLM

final: 5.59

dV: 4507

RESULT:

Even though Mechjeb automatically adjusts throttle to terminal velocity to prevent excess drag, for every TWR higher than 2, there was a reduction in efficiency, which seemed to be stabilising around the 4.0-4.5 mark (variance of only 5 ms, a 0.1% decrease).

The efficiency did stabilise as predicted (due to the auto-adjusting), however the decrease in efficiency after 2.0 was unexpected.

Currently, I have completed the battery of test planned for this experiment. If further information is needed, it is designed to be very simple to do further tests.

Edited July 5, 2013 by kahlzun

[Whirligig Girl]

good work Kahlzun.

[SirJodelstein]

How on Kerbin did you manage that??

Well, i wanted to do a more impressive presentation than just a 4x4 floating plate before releasing the how-to, so sorry for this delayed answer.

Here is an advanced example of what is possible with the Skyhook construction method, and a Tutorial explaining how to do it

[kahlzun]

Well, i wanted to do a more impressive presentation than just a 4x4 floating plate before releasing the how-to, so sorry for this delayed answer.

Wow, that is an impressive use of glitching. So obvious now I've seen the video, but would not have thought of it prior!

[Whirligig Girl]

FASCINATING! Indeed!

[Agent30632]

Gregrox, I see that my name in the post of contestants is still red. I feel I have finished my experiment as there are 2 options for the Lander+CSM duo, and the one option for direct ascent. Please fix that if you may.

Unless you just haven't updated that in a while and you don't want to for now...

[Aqua667]

I have often wondered if there is any difference in the behavior between the Mk16 and Mk16XL, as they state the same drag value on the tooltip.. Is there any way you can expand the test to also check this?

Sure, I could probably do that. I should probably hurry before .21 breaks save though.

[kahlzun]

Anyone got any further developments in their experiments?

[kahlzun]

I've discovered that 4x"24-77" engines seem to be more efficient than 2 Aerospikes in vacuum in almost every situation..