If you're landing a rocket then waiting to the last minute slamming on the brakes actually saves fuel over slower, more considered approaches to landing.
The atmospheric drag also greatly contributes to the efficiency of a suicide burn. By waiting to the last second to begin the burn, you maximize the amount of drag placed on the rocket which will reduce the velocity at burn time and thus will require less fuel.
It's most efficient to land with the wind going upwards, slowing you down. So the best way to land is near the eye of a hurricane. The disadvantage: You might have to wait quite long for your landing window.
I really enjoy your "Things Kerbal Space Program Doesn't Teach" series! Just the right mix of 'real' rocket science and the fun gaming version of rocket science!
Agreed! Just enough Kerbal to make it relatable. It's nice to start from an example I've actually tried myself, and then layer the real world on top of it.
What's more impressive than a rocket landing itself? Your rocket splitting into 3 pieces and each piece landing itself XD. Well 2 out of 3 isn't bad either.
It's really brilliant if you think about it. They needed it in smaller pieces to transport it back safely. They just took it apart really really quick.
There's a lot more to decoupling than that. While decoupling is sometimes just an explosive charge detonated to bring the two pieces farther apart, there are much more sophisticated mechanisms for times when the rocket needs to allow for transfer of resources between the two halves before the staging happens.
calculating a suicide burn can get incredibly complicated especially when you add: air friction fuel use making the rocket lighter gradually while you descend but the more fuel you use the more this comes to effect non vertical approach trajectories gimbal control reducing thrust etc approaching with 3 enignes and shutting down 2 justb efore landing gives a huge advantage because it means you have a rather significant point of control relatively close to the landing, so you can do final adjustments by controling when you shutdown the two extra engines and these are close to the actual landing so the deviation after this correction is small - and shutting down engines is a lot more precise and reliable hten igniting them
that would be so easy if gravity wouldn't keep pulling on you, so your path curves making your descent rate nonlinear meaning that now its really complicated to figure out how long exactly your descent is gonna take - its not too hard to figure out when you have tiem and apen and a paper but when trying to do it on the fly it gets a bit tricky
Julian Danzer I think the problem is the wind. You can't measure how the wind behaves perfectly and an unexpected squall or in the case of landing on a boat the change of the angle due to waves might ruin an otherwise perfect landing.
I think 'suicide burn' sounds WAY cooler than 'Hoverslam,' but I get why they'd want to avoid that terminology. It makes it sound reckless and poorly thought out/uncalculated rather than reckless and extremely well thought out and calculated.
Yeah it seems a lot like Elon cares about terminology. You know he is calling his dream rocket that will fly to Mars in 2024 (hopefully) the BFR, right? It stands for Big Fucking Rocket, and I am not even lying
i think its the other way around : Elon: hey guys, we found a way to slow the cores to land! PR: great! what shall i tell the public its called? Elon: THE SUICIDE BUUUUURRN PR: NONONO-
@@enricobianchi4499 He probably figured that since the rocket was empty it would be as flammable as a used firework. Of course, "Empty" doesn't mean "Not still soaked with explosive propellants" in this case. To say nothing of any remaining usable fuel after the suicide burn.
Ngl the maths is semi weak because acceleration is velocity with respect to time and using a=T/M there is no time component nor velocity component. Thrust is just a force and mass is just a mass.
There is another advantage of the 3 engine sequence they used for the falcon heavy launch. Shutting down an engine is much easier than starting, and thus possible with higher accuracy. You stop the fuel flow, done. Having 3 engines running and then shutting down 2 gives you a precise and very capable degree of freedom. Shut them down at just the right combination of height and velocity, where the single engine can do the remaining job.
I land my racing drones like this. Free fall from about 50 meters, use the OSD to level the airframe with the horizon, then pin the throttle and steadily roll off... it's so rad when I nail it.
This. Especially since racing drones have such high power to weight ratio, if you punch out at the correct moment the drone LITERALLY stops without any delay, it really looks like CGI.
Lol I usually just fly low, then turn around quickly with yaw, a quick throttle pulse. That eliminates horizontal speed and then just descend. (I use 40° of uptilt on my FPV cam). Or I simply crash
lol. That seems... Risky. Then again, it's just a drone I guess. I've been learning to fly a plane, and well, when you're on the thing that's landing you tend to be a little more risk averse. Though I guess fixed wing aircraft have rather different dynamics, since you can land just fine with no power at all. Turns out the primary approach for powered landings though basically amounts to flying the aircraft into the ground then killing your vertical velocity at the last minute. Then you cut all the power and coast just above the ground. As you slow down, you lose lift and drop onto the runway... I guess you can't really do anything comparable to a suicide burn in a fixed-wing aircraft... XD
I did. Though the code I'm still working on. So far I've made the Grasshopper fully autonomous. Flies up, hovers at 1000m, then angles and lands itself on the launchpad. It's* using a simple calculation to figure out how much time it'd take to decelerate at 1.5g's of force. It usually gets up to about 50m/s in vertical speed before it starts to suicide burn. I'm still actively working on it. I don't like the lat/long measuring to stop overburning horizontally, and I'm still trying to perfect it before moving onto the Falcon 9/Heavy/BFR. If nothing else, it's been fun figuring the code out. Just started learning kOS a few weeks ago.
Lol, well yes and no. I am high, but the response got held for review because I'm an idiot and tried to give a quicklink for anyone to view. I'll repost without links: "I did. Though the code I'm still working on. So far I've made the Grasshopper fully autonomous. Flies up, hovers at 1000m, then angles and lands itself on the launchpad. It's* using a simple calculation to figure out how much time it'd take to decelerate at 1.5g's of force. It usually gets up to about 50m/s in vertical speed before it starts to suicide burn. I'm still actively working on it. I don't like the lat/long measuring to stop overburning horizontally, and I'm still trying to perfect it before moving onto the Falcon 9/Heavy/BFR. If nothing else, it's been fun figuring the code out. Just started learning kOS a few weeks ago."
Great video Scott! Out-done yourself, extremely interesting content peppered with nice side notes as well (like the fact the boosters aim for the sea not the barge), keep up the good work :)
Surprisingly, I remember a very basic 'simulator' Deutsche bahn released on their website a while ago that was a PR thing showing the best way to drive a high speed train for energy efficiency. One of the things was quite obvious; when dealing with a hill, let the train slow down uphill, then recover the energy downhill. (though it demonstrated this was tricky to do while keeping to schedule and not breaking line speed limits on the downhill section.) But the one that I'm reminded of here is that the program and the web page to go with it showed that the most energy efficient way to drive a train is to accelerate as quickly as possible, and decelerate as quickly as possible as late as possible. In other words, it's a similar operating principle to the suicide burn rocket landings - quick, sudden velocity changes are more energy efficient than slower, gradual ones. (though both for rockets and trains there are a bunch of limiting factors that make this less practical in a real world scenario.)
Oh, yes! Energy-efficiency optimisation often leads to bizarre results. One more niche application, very closely related to the problem with trains: given identical lap-time target, a more powerful racing car will use less fuel (assuming equally efficient engines). Accelerate as fast as you cant out of corners and then lift-and-coast for a while before hitting hard on the brakes. (Modern F1 with its ”hybrids” likely uses this approach, obviously, on how to best use the recovered braking energy.)
I love Space-X. Not because of what they do, how successful or not they are, or because Elon Musk...no... I love Space-X simply because they are a private company building spacecraft, and publicising every test, every launch, every landing(attempt) so you can just watch it. It's fascinating to me, and is somehow way different from the clinical approach NASA launches used to have.
yeah. Thats the best part of spaceX. Elon musks bussiness model is to create and industry and to create services to profit off a industry, which is actually really smart. Thats why with tesla hes striving to become the biggest car battery manufacturer so he can sell said batteries once other companies start building electric
I quite like their approach to R&D. Put in the research and design to make it workable, then test, evaluate, redesign, and repeat, instead of endless paper shuffling. I take the same approach to building model planes: If it works, great. If it fails and I learn something, well and good. If it crashes and burns, I'm only out a few dollars of foamboard and an hour of my Saturday afternoon to build it.
Falcon 9 block 5. If they're really able to be flown 10 times, we could wind up seeing a SpaceX launch every single week. In particular, seeing a block 5 Falcon Heavy fly.
After seeing a stray frame inserted somewhere around 2:30 I thought you may have pulled a Tyler Durden on us. After slowing the video down and several attempts at pausing it at just the right time, I found it was just your face. Love you videos!
Hey Scott, just wanted to say that as a Computer Science student, you've inspired me to take a class completely unrelated to my major - Astrodynamics! Thanks for the years of great videos!
As for naming terminology; Even though suicide burn is perfectly appropriate for what the maneuver is; hoverslam sounds not as cool, why don't we call it the HALO burn? Since special forces also don't use their parachute until the very last moments before landing, that also sounds cool.
people have probably commented on this before, but I really love how Scott's house / video set is just mayhem. Every other YT'er has manicured at least 1 corner in there house, but Scott is like "f that!"...what a champion.
+Virginia Hansen True fans of the Christmas tree will celebrate at least through St. Patrick's Day, possibly redecorating it as an Easter tree when the time comes. =D (Admission: we actually had a 4th of July tree once, but even I have to admit that was a touch excessive.)
Yup. The only thing it doesn't teach in all of this is that engines cannot be throttled, and need time to start up. But that was already part of a previous "things KSP doesn't teach" video.
@slopedarmor Sure, but there are limits. They need a certain amount of fuel flow to sustain combustion and operation, and that means there's a minimum thrust they can put out. Between that and full throttle you can probably throttle them pretty smoothly, but there's that minimum - which KSP doesn't model without mods.
That's true. You can go to like 1% thrust output in ksp but not in real life D:
6 лет назад
@Scott Manley, I think there's an easier way of understanding how reducing the burn time saves fuel. If the rocket is falling at a terminal velocity of 200 m/s, and you do an instant burn to reduce speed, you'd only need 200 m/s of deltav to brake. If you instead take X seconds, gravity keeps acelerating your vehicle during those X seconds, so you instead require 200 + 9.81 * X m/s of deltav. The longer the burn, the higher the deltav needed (roughly linearly).
I put down the datasheets to take a relaxing entertaining break from electronics and math...(how bout an interesting space video from Scott)... "...I like math..."..".. here's a graph" (facepalm)
For those wondering, as I was until I worked out a likely reason for myself, why can't real rocket engines throttle like in Kerbal, it's most likely because of one concept. The more moving parts you add to any system, the more points of failure you introduce. In order to throttle the way KSP does, they'd have to have a computer that controls valves opening and closing in increments and calculated to properly mix the oxidizer with the fuel. Thats introducing quite a bit more points of failure and is probably a risk they didn't want to take, and money they didn't want to spend.
I am curious why a suicide burn landing would be more desirable than landing the booster with a parachute (maybe I am just too lazy to research it). Or even a combination drogue chute and suicide burn, but I guess excess thrust is not a problem in the examples in the video. I would use this combined approach in Kerbal, but the unrealistic engine throttles make it so easy. IRL, sure you have to haul a parachute with you on the way up (weight), but you have to carry extra fuel for the suicide burn. Maybe you could use a suicide burn to slow the craft to a relatively low speed and use a larger low speed (low quality) chute at the end. Just wondering, because coming from Kerbal, I use chutes on every thing. Even experimental aircraft on the runway can be saved by chutes if they lose control on takeoff. Obliviously, Kerbal is not real life, but I do not know the real life pros/cons of using parachutes for landing.
You have to haul the parachute, it wouldn't be as precise, I imagine you might need some engine thrust for a soft landing anyway, plus you still need to reignite your engines for a possible boostback and a mandatory entry burn, and at that point you really might just carry a little extra for a final landing burn instead of dealing with the extra complexity and the downsides of chutes. A lifting body or winged vehicle might be able to skip the entry burn, and I think that's what the BFR upper stage/spacecraft is plannd to do, but it'll still land propulsively.
Robert G I'm guessing that flooding a rocket with salt water makes it much harder to reuse said rocket. It's pretty difficult to land something at a fixed location using a parachute and it's much more prone to wind blowing it off course
Accuracy is the biggest problem. With parachutes, your landing zone would stretch for miles, and wind can easily blow it off course. Then there is the problem of the drag of parachutes. Rockets are designed to take the compression forces, but a stretching force would require a new structure which would also be heavier.
Are engines that can reliably perform suicide burns pure sci-fi or do we have some ideas for tech that would pull it of with close too 100% reliability?
OH senpai noticed me! Well I was thinking that the tech seems really good for unman crafts at the current stage of development but will is realistically be used for manned flights in our lifetime? Seemed from the video that we are at the "good enough for unmanned"-stage not the "I'd take a ride in a craft that can only land via hoverslam"-stage.
GreatgoatonFire The boosters wouldn't have people in them... so a manned falcon 9/heavy is pretty reliable now. Put a reproduction mercury capsule on autopilot (I'm no steely eyed missle man) on the nose of one... I'd ride it tommorow given the chance. Rocket engines are reliable. The challenge with the falcon booster engine return is they run on LOX/RP1 So they need a reliable method of reignition (they probably use TEB, Triethylborane. Which is pyrophoric liquid that spontaneously ignites (quite violently) on contact with oxygen.) This isn't a new concept... the SR71 was using TEB to light its engines and afterburner since the 60s. Each engine had enough TEB to relight its afterburner 15 times. after that... nada. Carbon deposits on the TEB injector inside the J57 engines would sometimes cause multiple shots of TEB to be needed to light the afterburner... wonder if SPACE X is having a similar issue? If the Falcon was a hypergolic first stage, the relight would be simple, and almost foolproof... hypergolics are nasty, nasty, chemicals though. Anyone know why they went with LOX/RP1 for the boosters? As for landers... how do you think the apollo LEM landed on the moon... parachutes? Came down on its engines on a suicide burn, with enough reserve to adjust the landing site as needed to avoid terrain on final. It was said the most reliable part of the Apollo moon system... was the Lander. Now, a powered manned decent on a planet with a thick atmosphere like earth... I see it as a needless risk, chutes work fine, and are lighter than the fuel needed to pull off a powered decent. I think it could be done, certainly... but why work when you have drag to do the job for you?
They could use something inspired by the approach Rocket Lab takes (who use an electrically-driven turbopump, which provides extremely fast startup and throughput modification), to help start it spinning quickly upon startup. This could perhaps also provide a means for small but highly responsive adjustments to the turbine rotation speed, so as to significantly reduce the latency between targeted pump throughput and what is actually achieved - as well as maintain it, counteracting any disparities which arise. This could be achieved by coupling electric motor(s) to the turbine shaft(s); supplying power would increase turbine speed, and drawing power from it (by using it as a generator) would decrease turbine speed - performing these actions as required for control. The key part is the level of near-instantaneous control this should enable in control of turbopump throughput. *Important:* The proposed electric motor addition is only a _compensatory_ mechanism which aims to decrease latency and short-timescale variances between what the flight computer commands and their real-world outcomes - it is NOT proposed as the sole energy source for pump operation... it is merely a fast-acting and precise augmentation to provide responsiveness and consistency to the operation of existing components. It should consume only very modest amounts of energy, nothing _remotely_ like that required of purely electric-driven pumping like that used by the Rutherford engine used by Rocket Labs. ... Come to think of it, this would theoretically be even more effective for the operation of Raptor engines instead of Merlin engines, as full-flow staged combustion cycle engines like Raptor depend upon highly precise, well coordinated pressure differentials throughout the preburners and the turbines that feed them - especially during startup - and this level of control should be particularly useful in this context.
Do while (landing) { If (Rocket Going To Crash) { Don’t Crash } } Someone tell Elon that I’ll accept a million dollars to my PAYPAL & whatever the latest Tesla is sent to my door.
Wise words from a wise man on an island of his own inside a block-chain of knowledge, one moment to decentralized thought that begins day by day. Don't give up!
Tis the drawback of making a video months in advance. But judging the work that went into making the awesome video a couple of months for good editing isn't a bad thing!
Way back when I was in high school in 1971 we had an interesting program in a computer programming class. The school had an IBM 1620 computer and one of the computer programs available was Lunar Lander. The input/output on this computer was a built-in typewriter, so it was typed information only. You had to use your imagination for any "graphics". The Lunar Lander program would type your altitude, fuel remaining and velocity. You would type in the desired setting for the decent rocket motor. The goal being to end the program with zero altitude, near zero velocity and some fuel remaining. If you started braking too soon, you ran out of fuel and crashed, just like they would in real life. We quickly learned that the only successful flight profile was to do a hard burn at the last moment. I seem to remember that a few short burns along the way helped to keep the velocity from climbing too high (no atmospheric drag on the moon). It was fun, and I don't think we appreciated we were learning something. Fast forward 45 years. We watched the amazing flight and landing of the two boosters of the Falcon Heavy at work. In discussing the booster landing, it was neat to be able to share why the booster has to fly the "hoverslam" profile, it's the only flight profile that will work. Good job on your video and explaining this. Yes, it is rocket science..
Scott, just a remark, if you're interested (on an otherwise terrific video yet again): At the Falcon "rapid unscheduled disassembly" section, there's some serious information-overload going on, which can only be fully tackled by repeated pausing, rewinding and rewatching: - there's your explanation, which alone requires some focused attention - there's the amazing footage of that "technically, a landing", which grabs most of the attention, too - and there're the hilarious subscripts there, too, which can only be properly enjoyed if paying due attention. So, that part is a three-man job to watch actually. ;) The video could've been a bit longer, I wouldn't mind at all (as in fact it was, adding all the replays). Thanks, cheers!
That video of the F9 sliding around on the droneship is one of the most stressful things I've seen. Every single time I see that, my heart rate jumps. Even though I know the rocket is "okay!"
"Hoverslam " sounds cooler than "Suicide burn" but describes the process much worse (Edit: Scott talks about this in the video, I should've watched the video before commenting)
Hi Scott, You mentioned, "every second spent using engines against the force of gravity is wasting delta V equal to the force of gravity" This is a good concept for rocket fans to understand and in a bit more detail. It can more fully be explained by pointing out that hovering a rocket, is the intuitive example of how 'using engines against the force of gravity is wasting delta V'. The flight time to orbit is more fuel efficient the quicker it is completed. And similarly the booster landing burns are most fuel efficient when they can happen quickly. Having both the launch and landing burns a short period of time is accomplished by having them at the highest acceleration tolerable. Now, where does the wasted fuel go? That is a little less intuitive but knowing the answer gives you a more complete understanding of orbital dynamics. Well, the fuel used to fight gravity is used to move the barycenter of the Earth-rocket pair. That means the rocket is acting as a space tug during ascent and decent burns. But you can act as a space tug for a minimum amount of time ( and thus waste less fuel ) if you use a high acceleration to shorten the time of the needed burn.
Hey there, just a little note: MechJeb and KerbalEngineer *do* calculate the fuel burn just fine. The real reason is that while the math for a completely vertical suicide burn is trivial, it gets insanely nasty for a burn with an initial horizontal velocity. As far as I understand, there is no exact analytical solution for that. MechJeb does have a neat approximation though and comes pretty close, at least with the timing itself. More discussion here: forum.kerbalspaceprogram.com/index.php?/topic/155599-suicide-burn-code/
Actually, the intuition that a suicide burn saves fuel was a thing Kerbal Space Program *DID* teach me. Having said that, this is the Scott Manley series that I enjoy the most. More, please.
Had an interesting idea for a new series or mini series. Kerbal Budgetary Concerns. Basically every episode you have to unlock something and after the first mission you roll a die every episode. 6 means you can spend as much money as you have, 5 means you can spend 50%, 4 means 40% and so on.
Can we just appreciate how many explosions we saw in this video. Like even a what seems like a gentle lovetap by the booster is a huge explosion. We need a mod for this.
Back in the 70‘s there was this very early computer game „Moon Lander“, originally developed on PDP-8. To land the moon lander successfully, you had to use the limited fuel really late and close to the surface, but not too late. It was the same principle.
@3:52 "using more engines for landing . . = less fuel" Ok here is how I understand this: >engines = >thrust. - Thus you need < time for (total) thrust - Thus < thrust time = < time to combat gravity during burn. (hover) *Result: the savings of the amount of fuel needed to hover the rocket [for the difference between > engines vs. < engines] Did I get that right?
Love the video! And I would love to see a video in which you explain the mathematics of the more realistic suicide burn as an example. Great work as always!
When I started playing this video the title was “How to do a suicide burn” and when I finished watching it the title suddenly changed to “how to do a hoverslam”