马斯克在2023年国际宇航大会(IAC)上讲星舰,最值得停下来看的,是那套用机械臂接火箭的方案。他描述这件事的时候,语气里带着自己也觉得离谱的意思:在得克萨斯州的博卡奇卡,SpaceX 立了一座巨型塔架,塔上装着两条能开合的机械臂,发射时张开让火箭穿过去,助推器从太空回来、悬停在塔边的那几秒,机械臂合拢,把它夹住。
原话是这样的:
我们有一座定制设计的巨型塔架,配着巨大的机械臂,会真的去抓助推器、抓飞船。听起来很疯狂,我连一部这么拍的科幻电影都没看过。但理论上讲,它应该能成。(马斯克)
“连科幻电影都没拍过”这句话没有客套成分。火箭入轨是工程问题,把一只从太空回来的、几十吨重的金属筒用两根胳膊在半空接住,是另一种问题。马斯克自己也没给成功率。他的表述是:成功不是一组可以逐一列出的结果,但它落在可能性里面。这话听着含糊,其实交代了他的判断方式,能不能成本身就是一个概率问题,而他认为概率不为零。
不用吊车,是因为有风的日子吊车干不了活
有人会问,抓它干嘛,用吊车吊回发射台不行吗?马斯克说,技术上可以。在风小的日子,用巨型吊车能把助推器和飞船吊到发射架上,只是“相当笨拙”。塔架加机械臂的好处是,风力中等甚至偏大的天气里,它照样能把助推器和飞船吊起来。
他的推理很顺:既然这两条胳膊能把飞船吊到发射架上、把飞船吊到助推器头顶,那它应该也能用来接住从天上掉下来的助推器。同一套肌肉,正向用是吊装,反向用是捕捉。
支撑这个判断的,是 SpaceX 十多年的着陆技术积累。马斯克提到早期的“草蜢”(Grasshopper)测试:猎鹰9号(Falcon 9)的助推器起飞,悬停在100米高度,横移100米,再移回来,然后落地。那是十多年前的事。用推力器让火箭在空中悬停并不高效,推进剂哗哗地烧,但这件事能做到,早就验证过了。所以他当时的想法很简单:让火箭飞回来,在塔边轻轻一跃,两条胳膊合上,夹住它。
真正的目标是快速可重复使用
马斯克在这次访谈里反复强调一句话:重点落在快速复用上,复用本身只是前提。这两个词的差别,决定了整套设计为什么长成现在这样。
逻辑很直白。Falcon 9 的助推器现在回收已经成了常态,一枚助推器能飞17次、18次,这个数据放在十年前是天方夜谭。但 Falcon 9 有一处硬伤:二级没有复用设计,每次发射都要把上面那一截扔掉。用马斯克的比喻,这就像寄快递每次连箱子一起丢。箱子可以回收,可箱子上半截是一次性的,成本就压不下去。
而且即便助推器回收到位,翻修也要时间。Falcon 9 发射完,助推器落回来,至少要几天整备才能再次上场。几天听起来不长,但和星舰想做的事比,差了一个数量级。
星舰的设计逻辑是把这个时间压到极限,压到助推器几乎不用离开发射场。
助推器回家的路,四到五分钟
具体有多快?马斯克给了一串数字。
当前星舰的推重比在1.3到1.4之间。助推器和飞船的推进剂比例,眼下大约是3比1,也就是助推器带的推进剂是飞船的三倍左右,未来会往2比1靠。这个趋势意味着速度增量(Delta V)越来越多地交给飞船那一侧,助推器承担的那一段越来越短。
结果就是助推器烧得很快,飞行时间大约只有100秒。100秒之后它就开始掉头,翻转、点火返航、落回发射场。从发射到被机械臂接住,整个过程大约四到五分钟。马斯克自己也觉得这个数字有点野,“pretty wild”。
这里面有个物理约束值得说清楚。助推器推重比高,意味着它回来的速度也快,没法慢慢飘回来。五分钟里它要么被胳膊接住,要么砸在地上。二选一,没有缓冲。
飞船完全不同。它要绕地球一圈才能回到发射场,最快也要一个半小时。马斯克还补了一句限制条件:能不能一圈回来,取决于发射方位角和倾角(inclination)。如果轨道倾角合适,飞船正好从发射场上空经过,一圈就够;不合适,就得等更久。他半开玩笑地说,“那请一个半小时后再来”。
一个是五分钟,一个是一个半小时。这组对比说明星舰的运力节奏设想:助推器像公交车,几分钟一班;飞船像长途大巴,一个半小时一班。
抓之前,先让它在海上落准
机械臂接火箭听着刺激,实际操作顺序完全相反,先求稳,再求快。
马斯克说得很明确:飞船那一侧,先要确保它能完整无损地回来,并且精确落在太平洋上的某个点,才会尝试在发射场用机械臂接。之前的所有飞行,都要尽可能排除伤人和损毁财产的风险。所以飞船的捕捉会晚一些,得先看到它在水面上落准了,才动手。
助推器这边信心高得多。Falcon 9 的助推器已经无数次飞回卡纳维拉尔角空军基地着陆,只不过用的是着陆腿,不是机械臂。同样的反推着陆逻辑,SpaceX 已经练了十几年。所以马斯克判断,只要发射许可批下来,一年之内、甚至不到一年,就有可能接住第一枚助推器。飞船的捕捉乐观估计在明年年底。
发射许可在这里是一个变量,不能当成客套话听过去。星舰的试飞节奏一直卡在监管审批上,马斯克把它和工程进度并列,说明在他眼里这两件事同样是瓶颈。
星链V3:先发卫星,再啃最难的那块骨头
访谈里有人问,什么时候开始用星舰部署卫星。马斯克的回答很具体:大约一年之后,开始部署星链V3(Starlink V3)卫星。
这里有个容易被忽略的逻辑。星舰最难的部分是飞船安全再入和着陆,这件事要花最长时间去解。但发卫星不需要等这个难题解决。因为即便飞船最后是要丢弃的,发卫星这件事本身也成立。Falcon 9 的二级本来就是一次性的,照发不误。所以星舰的路径是:先用一次性模式把 V3 卫星送上去,同时并行攻再入和复用。
这个安排很务实。星链是现金流业务,卫星越早上去越好;飞船复用是长期命题,急不得。两条线并行,互不阻塞。
顺带说一句,星链V3的尺寸和重量都比现在的大得多,Falcon 9 的整流罩装不下,只有星舰的货舱能装。所以 V3 上线的时间点,实际绑在星舰的运力上,跟复用进度关系不大。
星舰为什么造这么大
很多人第一眼看到星舰的反应是:这也太大了。马斯克的解释是,尺寸算不上目的,它更像结果。
真正比尺寸更深刻的,是它被设计成完全、快速地可重复使用。之所以造得这么大,是因为我们要造的东西,得能在月球上建一个永久基地,在火星上建一座城市。否则我们可以把它做小得多。(马斯克)
换句话说,星舰的吨位是被任务倒推出来的。要在火星建城市,一次发射就得带上足够的人和货,运输成本才能摊薄到可承受。小火箭在数学上走不通,造小了根本到不了。
这也解释了为什么 SpaceX 愿意在快速复用上死磕。去火星要反复往返,持续不断地运人运货。如果每次发射后要修几个月,火星城市就是空谈。复用率决定单位成本,单位成本决定这件事能不能成。机械臂接火箭看着是技术炫技,本质是在压缩两次发射之间的间隔。
火星:三到四年,无人先落
被问到星舰什么时候能无人登陆火星,马斯克的回答是:
我觉得三到四年吧,四年,差不多是这样。不过我得以地球和火星的轨道同步为准,大约每26个月才有一次窗口。你不能在火星跑到太阳另一边的时候发射,那没法弄。(马斯克)
26个月这个数字,是火星任务的硬约束。地球和火星的相对位置每26个月才对齐一次,对齐的这段时间叫火星转移窗口,飞船必须在这个窗口里出发。错过就得再等两年多。
所以马斯克说的“四年”,实际是两个窗口的事。他自己也用了“可行”(feasible)这个词,语气落在可能性上,没有打包票。按他的时间表,未来四年内做一次无人的火星着陆测试是可行的。注意是无人、测试性质。载人还要往后。
把这个时间表和前面几段串起来看,逻辑是通的:一年内接住助推器,明年底可能接住飞船,同期开始发 V3 卫星,再往后三到四年做无人火星着陆。每一环都压在前一环上,助推器复用没跑通,后面的时间表都站不住。
这套方案的赌注在哪
把马斯克这次讲的内容摊开看,赌注集中在两处。
第一处是机械臂能不能接住高速返回的几十吨金属。火箭悬停本身不新鲜,草蜢项目十多年前就做到了。新鲜的是在塔架旁边悬停、让两条胳膊在几秒内合拢夹住,容错空间极小。助推器推重比1.3到1.4,回来的速度快,悬停窗口短,姿态稍偏就是塔毁箭亡。
第二处是飞船再入。马斯克自己承认这是最难、最耗时的部分。助推器回收有 Falcon 9 十几年的数据垫底,飞船从轨道速度再入、经受高温、精确落点,没有先例可循。星链 V3 可以先发,正是因为这一步可以往后放。
这两处赌注的赔率不一样。助推器捕捉是工程难度高但路径清晰,飞船再入是路径本身还在摸索。所以时间表上,助推器一年内、飞船明年底,排这个先后顺序有它的道理。
还有一个隐性的赌注:发射许可。马斯克把许可和工程进度并列,说明在他眼里,监管节奏和造火箭一样能决定时间表。这一点在后面的星舰试飞里反复应验。
给关注这件事的读者,几条能落地的判断
这场访谈信息密度不低,落到具体行动上,可以这么用:
- 看星舰试飞时,把“接住助推器”和“飞船再入成功”当成两个独立节点跟踪。前者一年内、后者明年底是马斯克自己给的时间,两个节点分开看,才不会被整体进度带偏。
- 判断星舰商业化节奏,盯星链V3的首次部署,别只盯载人火星。V3 大约一年后开跑,这一步不依赖飞船复用,是最早能看到实际收入的信号。
- 评估火星时间表时,先在地球和火星26个月窗口的日历上标出节点,再看马斯克说的“三到四年”落在哪个窗口。错过一个窗口,整体往后推两年多,这个换算比记年份有用。
- 看“可重复使用”的宣传时,追问一句翻修周期是几天还是几小时。Falcon 9 的助推器能飞17、18次,但两次之间要整备几天;星舰想做到的是几分钟到几小时,这两个数字才是成本差异的真正来源。
- 关注发射许可的审批进度。星舰的试飞节奏多次受制于监管,许可批得慢,工程进展再快也上不了天,这个变量值得单独盯。
0:00but we have we have a giant custom designed Tower with massive mechanical arms that will literally try to catch the booster and catch the ship which it sounds insane I mean I haven't even seen a sci-fi movie that does this you know but Theory it should work let's just say success isn't a set of possible outcomes I'm not so I'm sure what the probability is but success is somewhere in that it's a success possible yes I think it's possible in terms of catching it I think
0:37well for this for the ship side we obviously want to make sure that it actually comes in a fully intact and and lands at a precise location in the Pacific before we try to catch it at the launch site because we we we're taking every proportion we can for so we do not risk any any human lives or destruction of property so it'll be a few flights so so for the ship it'll be when we see the ship landing at a precise position in the in the water that's when we will try to catch the ship with our mechazilla on the tower the booster obviously booster fights we've done many times on falcon 9. so we're much more familiar and have much higher confidence with booster recovery
1:23we've actually had the booster boost back to land and land at Cape Canaveral Air Force Base many times so I'll be with Landing legs not with the max alarms so booster I I I I I think I think there's a decent chance depending on when our licenses are granted that we would catch the booster within the next year or maybe less than a year and then hopefully if you get lucky we won't catch the ship towards the end of next year and where does the catch take place is it Willie Mays in the middle Outfield over her shoulder or is it Florida somewhere no it's so the the both the booster and the ship come back to the launch site okay fantastic yeah that's what I mean by the the the
2:14in fact I mean the the thing that is Success we need a giant Tower with customized arms to lift the booster and the ship onto the launch pad we're gonna absolutely need it we can technically do it with with humongous cranes on a low wind day but that's quite unwieldy the the the Tower with the arms is capable of lifting the booster and the ship even on on a on on a very windy day oh moderately windy day so then it just seems to me that well if we can lift the the ship and the booster the ship onto the onto the launch stand or the Boost front of the launch stand and the ship onto the booster with those same arms we should be able to catch the
3:03the Boost ownership with those same models you know we've gotten pretty good with with with the thruster-based landing and in fact we can make this we can make the the rocket hover in mid-air in fact we were able to do that many years ago if you look at the old Falcon 9 test videos which we were called grasshopper where we'd actually take the Falcon 9 booster and we'd have it just go up and and hover at 100 meters and then translate over another 100 meters then translate back and then come back and land so we're able to do that really over a decade ago it's it's not obviously very efficient we're propellant to have a rocket hover but it can be done
3:48so that was I was like okay let's just have the rocket come back and you know hop up briefly and have the then the arms come together and catch it so that's that's the general idea is going back to what I was saying it's not it's not just reusability it's rapid reusability and and it doesn't get more rapid than bringing it back to the launch site and so in principle the booster comes back very fast by the way one way or another that booster is coming back to land or it's going to land fast because with with the high thrust to wait that we're waiting for which is sort of on the order of 1.3 to 1.4
4:30and and a staging ratio which is currently about three to one in favor of the booster so propellant to they're propellant on booster to propel my chef is about three to one on on the current version but it's trending closer to two to one on with future versions it means that we're shifting more and more of the Delta V rodent to the ship side but that means the the booster actually uses up its propound quite quickly and we'll Trend towards about only about 100 seconds or so of a booster flight and the booster will immediately flip around boost back to their site and land and so it really we're talking about the booster being back at the lower site in
5:21about four or five minutes which is pretty pretty wild I think it's like five five if you know if a five minute boost so basically it's about it's it's it's it's landed somehow with see the crashed or it's landed on the it's been caught by the arms one of the two and within five minutes and so so so then you then Lance back on launch stand and you can then refill propellant the the booster there's a short side obviously is going to take a minimum of an hour and a half to get around the planet something pretty fast but you go to Circle globe and and obviously that depends on what inclination and so what once you launch Azimuth what's your inclination
6:06of the booster as to whether it has a flight coming back over the launch site or not if it it is technically possible to do a single orbit depending upon your launch inclination very good so please come back in an hour and a half sorry the ship could come back in about an hour and a half any any prediction on when you're gonna start deploying satellites with Starship yeah I think we will start deploying
6:42I think there's a good chance we start deploying starlink V3 satellites next year in roughly roughly a year from now because we we before we are confident that the ship like I said the the the the hardest part about this or the part that will take the longest is this is is solving for ship safe ship reentry and landing and but before we solve
7:12that we can launch the satellites because in any case with Falcon 9 the the upper stage is expendable so you know it's actually fine to start launching satellites even before we saw for worship reusability that that is the hardest part of the equation so with Falcon 9 we've gone pretty far with reusability you know the the booster it's now highly unusual for the booster to not come back and land it's it's gotten quite normal for the booster to come back to land we now have a couple of boosters that are I've done 17 I think 18 flights at this point and and and the fairing is also recovered so the fairing reusability is also solid but but the Falcon 9 design does not
8:01allow for a reusability of the upper stage so and without Falcon 9 just it while reasonably rapid if you especially if you've got a return to launch site Landing still takes at least a few days to refurbish before you can play the game and so with with with Starship actually more profound than the size is the fact that it is fully it is designed to be fully and rapidly reusable the reason for the Absurd size is that we're trying to build something that is capable of of creating a permanent base on the moon and a city on Mars that's that's why it is so large otherwise we can make it much smaller any predictions on when Starship might land on Mars without crew maybe a crude
8:54flight any any prediction there well I think three or four years four years yeah something like all right I have to check with the Earth Mars you know otherwise the you know get have orbital synchronization about every 26 months so you can't just go apply to Mars when it's on the other side of the sun from Earth that's unwieldy so roughly every 26 months the orbits
9:35are in the right relative position and then you then you have the Mars transport window so I I think you know but I think it's sort of feasible within the next four years to do a an uncrewed test test Landing there yeah
资源与出处
- 视频在 YouTube 收看
- 档案silicon-valley.tech 档案页本页视频与文稿数据来自 Elon Musk Archive
- 精读本篇精读 Markdown 版