一枚火箭飞完一趟就整根扔掉,等于每飞一次就报废一架全新的波音747。
2011年,马斯克在美国航空航天学会(AIAA)的演讲里给了一个更精确的比例:猎鹰9号(Falcon 9)单次任务的总成本里,推进剂只占大约3%。剩下的97%压在火箭本体上,而它在把载荷送进轨道之后,要么坠进海里,要么在大气层里烧成碎片。
这就是他整场演讲的题眼。火箭贵,贵的根源在于它只用一次,燃料本身几乎不值钱。想明白这一层,后面关于火星、关于发动机、关于载人飞船逃逸系统的所有讨论,就全都串起来了。
推进剂只占3%,另外97%在替一次性买单
那个3%的数字值得多停一会儿。
马斯克的原话是,猎鹰9号上的推进剂成本约占任务总成本的3%。这意味着,假如能把整枚火箭完整收回来、加满燃料再飞第二次,成本效率至少会提升两个数量级,也就是百倍这个量级。
这个换算很有意思。它说明航天的高成本来自商业模式的选择,物理学并没有强制这个价格。燃料便宜到可以忽略,真正昂贵的是那些被一次性消耗掉的精密结构、发动机、航电系统。整个行业几十年来的成本曲线,本质上被“用完即弃”这四个字锁死了。
这笔账还能反过来算。马斯克提到,如果要把“在另一颗星球上建立生命”这件事的开销分摊到全球GDP上,让你拿出GDP的3%去做,这件事还有可能被摆到桌面上讨论;让你拿出GDP的100%,那它根本不会发生。人们还要吃饭,还要过日子。所以把成本砍下来远不止是锦上添花,它直接决定了这件事到底能不能成。
我后来又把这个3%在脑子里过了几遍。它其实是一个很残酷的对照:几十年来,人类把最顶尖的材料、最精密的加工、最聪明的头脑,全都投进了一个一次性的容器里。每次发射成功,都伴随着一堆价值数亿美元的硬件沉进大西洋。整个行业一边为这个结果欢呼,一边默认它是天经地义的。
梅林发动机的第四次迭代
演讲开头,马斯克先盘了一遍SpaceX手上的家底。
猎鹰1号(Falcon 1)、猎鹰9号、猎鹰重型(Falcon Heavy),构成了运载能力的梯队。这些收入来源之所以重要,是因为它们要给未来的研发输血。航天公司没法靠讲故事活着,得靠一次次发射合同把现金流续上。
然后是梅林发动机(Merlin)。当时它即将进入第四次迭代,也就是梅林1D。马斯克给它的定位是:在飞的碳氢燃料发动机里效率最高的一款。
这句话有分量。碳氢发动机(烧煤油的那一类)在推重比和成本上有优势,效率却一直是个短板。梅林1D能把效率做到那个位置,意味着SpaceX在最硬的那条赛道上,已经摸到了第一梯队。后来梅林1D果然成了猎鹰9号的主力发动机,一路迭代,支撑了几百次发射。
更值得留意的是他紧接着说的那句话:大概今年年底或者明年年初,会公布一款全新的发动机,超高效率的分级燃烧(staged combustion)发动机。
分级燃烧是液体火箭发动机里技术难度最高的一档。它要把预燃室里那部分富燃或富氧的燃气送进主燃烧室再烧一遍,好处是比冲高、效率高,坏处是涡轮泵要在极端温度和压力下工作,材料和工艺都是硬骨头。苏联人当年在这一路上走了很远,美国这边长期绕道走。SpaceX 要碰这个,等于公开宣布自己要从“便宜可靠”往“效率顶尖”再上一个台阶。几年之后,猛禽发动机(Raptor)把这条路走通了。
马斯克还补了一句很有意思的话:我们要先把所有鸭子排好队(all ducks in a row),再公布这个研发计划。这句英文俚语的意思是,等所有前置条件都稳妥了再对外说。一个以高调著称的人说出这种话,反而说明这个项目在他心里的分量。
我们到底要不要成为多行星物种
发动机聊完,马斯克把话头拉到了最根本的地方。
他说,从大历史尺度看,最重要的问题是:我们有没有走在成为多行星物种的路上。
如果我们没有,那未来就不太光明。我们只会待在地球上,直到某场最终的灾难把我们带走。(马斯克)
这句话听着像科幻,可它是他整套商业逻辑的地基。小行星撞击、超级火山、核战争、某种失控的病原体,这些事件的概率单看都很低,时间尺度一拉长,概率就趋近于必然。地球上的文明如果只有一份备份,它的期望寿命就是有限的。
所以SpaceX的KPI落在“我们有没有让这件事往那个方向挪一点”上,而不在“今年发了多少次”。这是马斯克自己说的判断标准,他把这句评价标准当着几百个航天工程师的面讲了出来。
接下来是一个很硬的约束条件:要在火星上建立自给自足的文明,你得能运送几十万人,甚至可能上百万人,还有几百万吨的货物。
这个量级靠优化现有机型解决不了。它需要根本性的突破。
最大的卡点:快速且完全可重复使用
马斯克认为,最根本的那个突破是“快速且完全可重复使用”(rapidly and fully reusable)的火箭。
注意这里有两个限定词,他说得很清楚:partial(部分)不行,not rapidly(不够快)也不行。
这两个词是他整段论述里最容易被忽略、也最容易被误解的地方。很多航天项目都宣称自己在做可重复使用,落到这两个标准上,大部分都不合格。
部分可重复使用,意味着你还是要扔掉一部分硬件。比如只回收一级,扔掉二级和整流罩。那样成本能降,降不到质变的程度。
不够快,意味着回收之后的翻修周期太长。一枚火箭回来要拆开检修半年,它一年就只能飞两次,摊销下来跟一次性火箭的差距被抹平了。
航天飞机是理解这一点最好的例子。它确实能重复使用,飞了三十年,一共执行了一百三十多次任务。可每次任务结束之后,返回的轨道飞行器要经过大规模翻修:发动机要拆下来检修,隔热瓦要一片一片检查,有些要重新粘。整个流程要花几个月,成本核算下来,每次发射的开销反而比一次性的运载火箭还高。
所以航天飞机的历史意义,在马斯克眼里恐怕是个反面教材:它证明了“能重复使用”和“值得重复使用”之间隔着一整个世界。
汽车、飞机、自行车,全都是“用完还能再用”
为了把这个道理讲透,马斯克搬出了生活里的类比。
他说,你看看其他所有机动交通工具,自行车、汽车、飞机、轮船,全都是快速且完全可重复使用的,除了偶尔的保养项目。
这就是它们能运转良好的原因。
假如你有一辆车,每开一趟就得换一次轮胎,或者一周只能开一次,那它就不是一辆好车,也不是一架好飞机,什么都不是。(马斯克)
这个类比朴素,杀伤力却很强。它把“可重复使用”从一个技术指标,拉回到了常识判断。没人会觉得一辆每次出行都要换轮胎的车是合理的,可因为火箭行业一直以来就是这样,大家反而习以为常,把它当成了物理定律。
行业惯例有时候比技术难题更顽固。技术难题有人去攻,惯例没人去质疑。
马斯克还举了莱特兄弟的例子。他说,莱特兄弟发明的并非“会飞的东西”,他们发明的是受控的动力飞行。真正让飞机变得有用的,是这个。
这个区分很关键。会飞的东西,之前有过,热气球会飞,滑翔机会飞。只有受控的动力飞行,才把飞机变成了交通工具。同样地,能回收的火箭之前也有过,航天飞机就是。只有快速且完全可重复使用,才能把火箭变成交通工具。
他接着说,这件事要么SpaceX来做,要么别的什么机构来做,总得有人做出来。这是让航天发生革命的必要条件。
为什么非要垂直起降
聊到回收方式,马斯克的偏好很明确:垂直起降(vertical takeoff and landing)。
理由很实在。你要去太阳系里的其他地方,月球上没有空气,也没有跑道,机翼就是个累赘。靠滑翔着陆这一套,在地球上或许还能玩,出了地球就没戏了。
这个思路解释了为什么猎鹰9号的一级回收是垂直落回来的,而不是像航天飞机那样滑翔降落。垂直起降的通用性更强,它对目的地的要求只有一个:有块平地。
火星有大气,密度只有地球的1%左右,降落伞的效率大打折扣,靠气动减速加反推着陆才是正解。月球干脆连大气都没有。垂直起降在方案比选里能胜出,靠的是它能同时覆盖多个目的地,地球上省不省那点推进剂反倒是次要的。
从1500人开始,还在快速扩张
演讲结尾有一段是招人。
马斯克报了几个数字:去年人员规模增长了50%,现在大约1500人,今年还在继续快速扩张。业务分布在佛罗里达、得州、加州,还有华盛顿特区到弗吉尼亚一带。
这些地点不是随便选的。佛罗里达有卡纳维拉尔角的发射场,得州有麦格雷戈的测试场,加州有霍桑的总部和制造基地,DC一带负责和政府、NASA打交道。四个点,覆盖了研发、测试、发射、公关四条线。
他说,如果有兴趣加入SpaceX,把简历发过来。这话说得实在。2011年的SpaceX确实处在急剧扩张期,猎鹰9号刚首飞不久,龙飞船还没完成第一次国际空间站任务,人手是硬缺口。
放在当时的行业背景下看,1500人是个小数目。波音、洛马、ULA这些老牌玩家的雇员规模都在几万人以上。一家一千多人的公司要去动几十万人建立起来的行业惯例,靠的只能是工程效率,而不是资源堆叠。
火星架构:知道配方,还没细到每一克
进入问答环节,第一个问题就是火星。
有人问,你们有没有一套完整的火星任务架构?
马斯克的回答很坦率:有,但不能说已经细化到了很具体的程度,我们知道配方是什么。
配方是什么?一艘能把“可观质量”送到火星表面、并且还能返回地球的飞行器。
他给了具体的量级。猎鹰重型大概能把12到15吨的载荷送到火星表面,这个数字不算差。可如果要做正经的火星任务,你需要的运载能力大概在50吨这个量级,而且必须完全可重复使用。只有到了那一步,整个方程才成立。
他还补了一句:把这件事做成的路径不止一条(more than one way to skin that cat)。这句英文的意思是“剥猫皮的方法不止一种”,也就是殊途同归。
这段话透露的信息量不小。2011年的猎鹰重型还在图纸上,首飞要等到2018年。马斯克已经在用“12到15吨”和“50吨”这两个数字,划出了从“能做点事”到“能成事”之间的差距。这个差距,正是后来星舰(Starship)要填的坑。
还有一个容易被忽略的细节:他强调的是“送到火星表面”并且“返回地球”。这两件事叠在一起,难度比单纯送上去高一个档次。火星的引力虽然只有地球的38%,要从火星表面起飞回地球,仍然需要一套完整的上升级,还得在火星上就地生产返程推进剂。这套设想后来演化成了星舰在火星表面加注甲烷和液氧的方案。
一年两三百次发射,靠谁买单
第二个问题很尖锐。提问者说,他看过很多可重复使用飞行器的研究,关键是要做到一年两三百次发射量级,才能把成本压下来。那么,什么商业需求能支撑起这么大的发射频次?
马斯克的回答分三层。
第一层是类比。他说,我猜在飞机的早期,人们也想象不到后来会有这么多航班。这句话的意思是,需求往往在供给出现之后才长出来。
第二层是否定。他不认为卫星发射能撑起这个量级。也不认为在月球上开采氦-3(helium-3)这类事情有理性基础。他原话是,他很难想象在任何一个地方开矿,再把东西运回地球,因为地球上几乎什么都有。
这句话值得细品。很多人对太空经济的想象是“去太空挖矿运回来”,因为小行星上有稀有金属。马斯克的判断是,这条路在成本上算不过来。地球本身就是一个巨大的物质库,从引力深井里往上运设备、往下运矿石,光摆脱地球引力这一关,成本就吃掉了所有利润空间。
他还加了一句挺有画面感的话:地球很大(Earth is big)。言下之意,人类还没把脚下这颗星球用明白,先别惦记别人的石头。
第三层是肯定。他认为真正的市场,是那些想搬到月球或者火星去住的人。
这个需求听起来很虚,体量却是实打实的。如果真有大批人想移居,每年几千次航班的需求就撑得住。这也解释了马斯克为什么一直在讲“多行星物种”。情怀之外,这也等于在给一个每年几千次的发射市场做需求侧铺垫。
龙飞船离载人还差什么
第三个问题转向龙飞船(Dragon)和载人运输。
马斯克的回答很有技术密度。他说,猎鹰9号和龙飞船在设计之初,就是按照NASA公开的载人评级设计指南(human rating design guidelines)来做的。
具体到细节,他列了几条。
安全系数上,用的是1.4倍,而NASA的标准是1.25倍。这个差别不是小数,整枚火箭的结构重量都要为此增加,运力要为此打折。多出来的每一公斤结构,都是从载荷里扣出来的。
故障容错上,龙飞船可以在任意时刻失去任意两台发动机,依然安全。对接空间站的时候,还能正常平移和旋转。这是多重故障容错(multi-failure tolerant)的要求。
隔热上,主防热盾后面还藏了一层备份防热盾。
航电系统全部做到双故障容错(two-fail tolerant)。
甚至窗户都留了。
然后他话锋一转:那它还缺什么?
主要就一样,逃逸系统。
逃逸系统的另类解法
这一段是整场演讲里技术含量最高、也最能体现SpaceX工程思路的地方。
传统的载人飞船逃逸方案,是在火箭顶端装一个逃逸塔。发射初期一旦出事,逃逸塔点火,把飞船拽离火箭。这个方案成熟可靠,却有个短板:逃逸塔在飞行的某个阶段就得抛掉,因为带着它上轨道太浪费运力。抛掉之后,飞船就失去了高速逃逸能力。
SpaceX 的做法另走一路。他们省掉了逃逸塔,直接用飞船自带的推进剂当逃逸动力。这种推进剂是单组元肼(hydrazine)类燃料,平时用于在轨机动。
马斯克的解释是:你要么需要在轨机动,要么需要逃逸,你不需要同时做这两件事。所以同一套推进剂可以两用,发射段当逃逸动力,入轨后当机动动力。这个设计在质量效率上非常划算。
更妙的是,它把高速逃逸能力一直带到了轨道上。传统方案里,逃逸塔一抛,飞船在飞行的中后段就失去了快速脱离的能力。SpaceX 这套方案没有这个断档。
然后是他顺带提到的一个“附赠功能”:同一套发动机,还能用来做推进式着陆。
四层冗余,顺便还能在火星上软着陆
有人可能会问,那降落伞呢?
马斯克说,降落伞还是保留的,而且是冗余配置。
所以龙飞船实际上有四层冗余:着陆发动机本身是冗余的,主降落伞也是冗余的。要出人命,得四个环节接连失效才行。
这个设计的额外好处是通用性。同一套推进式着陆能力,可以拿到别的地方用,比如火星,比如月球。马斯克说,他觉得龙飞船有潜力成为一个通用的科学载荷投送平台,服务于太阳系里的其他任务。
这段话放在2011年听,像是远景规划。把它和后来的事实对照着看,思路的连续性非常清楚:垂直起降、推进剂两用、冗余分层、平台通用化。这些原则后来在星舰上被放大到了另一个量级。
顺带说一句,这场演讲举办的时间点很有意思。2011年7月,航天飞机执行完最后一次任务退役。美国突然失去了把宇航员送上太空的能力,此后近十年只能买俄罗斯联盟号(Soyuz)的船票。同一时期,一家一千多人的私营公司在讲可重复使用、讲火星移民、讲逃逸系统的另类解法。十年之后,龙飞船把宇航员送上了国际空间站,猎鹰9号的一级回收成了常态。回头看,那场演讲更像一份公开的技术路线图,只是当时没什么人当真。
这套逻辑对今天做产品的人有什么用
马斯克这场演讲讲的是一家火箭公司,可里面有几条思路,放到别的行业一样成立。
- 先算清楚成本结构里,哪一部分被“一次性”这个假设锁死了。猎鹰9号的3%和97%就是一次彻底拆解。很多行业的成本高,原因常常是某个环节默认了“用完就扔”,物理学并没有下这个命令。
- 把“可重复使用”拆成两个可验证的指标,别停在模糊的口号上。快速、完全,两个限定词缺一不可。做产品做服务也一样,光说“可复用”没用,得说清楚复用一次要多久、复用之后性能衰减多少。
- 用生活里的常识去检验行业惯例。每趟换轮胎的车不是好车,这个判断不需要任何专业知识。当你发现一个行业里所有人都接受某个荒谬的默认设置时,那通常就是机会所在。
- 给技术路线加上通用性约束。垂直起降能胜出,理由是它能同时覆盖月球和火星。做架构设计时,多想一步“这套东西换个场景还能不能用”,往往能省掉后面的重构。
- 用冗余分层代替单点加固。龙飞船的四层冗余靠的是让四套独立系统各自可靠,把一套系统做到极致可靠那条路它没走。工程上,分散失效点通常比强化单点更划算。
- 需求等不来。飞机早期的先驱没法预测后来的航班密度。如果你在做一件基础设施级别的事,别指望市场调研能告诉你天花板在哪,你得自己把供给做出来。
- 把大目标翻译成可核算的数字。几十万人、几百万吨、50吨运力、每年几千次航班,这些数字听着粗糙,却能让讨论从“值不值得”变成“差多少”。空谈愿景的人很多,能把愿景换算成工程参数的很少。
0:00and all of these things are important revenue streams because they they're they're needed to fund future developments and you know we've got obviously Falcon I and Falcon heavy we've got the Merin engine which is about to go through its a fourth iteration with the Merin 1D and it's it's the the highest efficiency hydrocarbon us hydrocarbon engine ofc flight and then we we'll probably be a be able to announce maybe late this year or early next year a a new engine development that we're going to do which is going to be very exciting and that that that'll be super high efficiency stage combustion engine and we want to make sure we have all L ducks in the row before we
0:52announce announce that development but I think that's going to be very exciting and and also something like that's very important for you know if you if we want to take significant amounts of of cargo and people to to Mars because I think ultimately the the the thing that is super important in in the grand scale of history is are we on a path to becoming a multiplet species or not and and if we're not
1:21well that's that's not a very bright future you know we'll simply be hanging out on Earth until some eventual Calamity claims us so so that that's really the important thing and so that that's kind of you know how I'm judging SpaceX like are we are we helping move things in that direction and in order to have a self-sustaining civilization on Mars you have to be able to transport hundreds of probably hundreds of thousands maybe millions of people and and millions of tons of cargo so so clearly very significant breakthroughs are are needed and I I think the most fundamental breakthrough is a rapidly and fully reusable rocket and
2:13those are important qualifiers it doesn't help if it's partially reusable and it doesn't help if it's not rapidly reusable because if you think of any other motor transport like U really any any motor transport from bicycles cars planes boats they're all rapidly and completely reusable I mean apart from occasional maintenance items so and and that that's why they function well you know if if you had a car if you had a car where you had to say you know replace the tires on every trip or could you could only use it once a week that wouldn't be a very good car or plane or anything so it's got to be fully and rapidly reusable just like when you know when the R Brothers
2:59invented create the airplane they actually didn't didn't invent something that flu they invent they invented controlled power flight that that was really those those are really what what made an airplane very useful and so so that that's that's something that either SpaceX or or some some organization has to invent I think that's a critical element in revolutionizing space you
3:27if you look at say the say the Falcon 9 the cost of propellant on falcon9 is about 3% of the cost of of the mission so you know the that means if if we could fully reuse the the vehicle I mean it's at least a two order of magnitude increase in the cost efficiency of the of the flight you
3:51know or sort of another way you know it's one thing to ask for say 3% of of the GDP P to establish life on another planet it is another thing to ask for 100% of the GDP that it will simply not occur people need food and think you know basic stuff like that so that that's something that I think is also an important long-term thing that has to occur and and in that regard I'm I'm a fan of vertical takeoff and Landing because obviously if you're going to go to to other places in the solar system you know there there no air on the moon and there's no runways so wings are not a great idea and yeah so
4:37so those are sort of the things that we hope to do in in in the future and we're hiring a lot of people last year we we grew by 50% in in Personnel we're continuing to grow rapidly this year so we're now about 1,500 people and yeah and we have significant operations in Florida Texas and California and in the DC Virginia
5:05area so Ro always looking for great talent so if if anyone's interested in joining SpaceX please send us your resume and let's say I think I think with that I'd be happy to open the floor to any questions you have sure so the question is do do do we have an architecture planned out for M trips
5:45we we do I wouldn't say it is fleshed out to a detailed level but we know what the ingredients are you know which is a vehicle that's capable of of delivering substan stantial Mass to the surface of Mars and and then returning to Earth as well and but I think we'll take this in steps certainly something like Falcon heavy could deliver you know decent payload to the surface of Mars it's sort of on the order of 12 to 15 metric tons so that that's not bad I I think you probably want a vehicle that can deliver something a loader of 50 metric tons you know and be able to do that in a fully reusable manner then then the whole equation Works and there's more than one way
6:39to skin that cat but yeah you that that's that's what I think yes I have a question on the reusable vehicle that you talked about I I've seen a lot of studies of these vehicles and the key to to making it work seems to be to get on the order of two or 300 missions a year to keep the cost down what do you see as the the Commercial Drive that will will result in the need for two or 300 launches a year yeah I mean it that's always a tough question to answer and and I suspect in the early days of of airplanes people didn't imagine that there would be this many airplane trips but I I think the biggest driver for that is is really a base on mars or you know
7:31potentially based on the moon I don't think there's really a need for that many satellites for example and I I I also don't think it's that there's a rational basis for like you know think things like mining helium 3 on the moon and or really I have a hard time imagining mining anything anywhere and bringing it to Earth because Earth has got a lot of stuff Earth as Christ has got almost every I mean Earth is Big but I do think there is a market
8:05for people who would want to move to the moon or move to Mars and and and that that I think is is is interesting and it could support a huge number of flights that thousands of flights a year can you tell us your plans for the driving capsule and time get sure question just about our Dragon spacecraft and what we're going to do with respect to carrying transporting astronauts and whatnot so the we actually designed the Falcon 9 and dragon to meet the published
8:52NASA human rating design guidelines so you know things like subd designing to a 1.4 factor of safety instead of 1.25 having a multi being multi failure to tolerance drag a Dragon for example can lose any two engines at any at any time and still be safe with respect to the space station can still translate and rotate as needed and we actually have a sort of a secondary heat shield behind the primary heat shield the avionics all two
9:26fail tent all that sort of stuff it's it's got window so the question is what doesn't it have really the main thing is just the Escape system so so in terms of upgrading it we need to add a launch Escape and in this we're taking a a sort of a different path than has been done before which is to tap the onboard propellant which is ANH as Escape propellant and because you either need to maneuver on orbit or you need to escape you don't need to do both and this this makes for a very eff Mass efficient design it and it also means
10:06that you can carry High Velocity Escape capability all the way to orbit as opposed to having to dump a launch Tower kind of early on and and then a sort of optional bonus is we can use those same engines to land propulsively now we're still going to retain redundant parachutes on dragons so you still have redundant you know parachutes so effectively you'd have almost four levels of redundancy because the The Landing engines are themselves redundant and then you've got redundant main parachutes so like four things would have to fail before actually only on yeah the yeah four things in a row would have to fail in to for loss of life but then the neat
10:57thing about that is you could use it to propulse land in other places like Mars or moon or somewhere so I think it's it's got potential as a generalized science delivery platform for for other you know missions in the solar system
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