I have heard people speaking about how, unlike Hubble, the James Webb Space Telescope is "one of a kind", "not designed to be repaired", "not serviceable", and the like, and that's the reason its complicated deployment process must go exactly to plan and thus why it was so long delayed (to improve reliability).

I don't believe in absolutes like that, so let's say that there is/was* a failure, at any point in this process:

  1. Launch phase (to LEO) [will obviously require a rebuild, though debris could be used]
  2. Launch phase (boost to Halo orbit)
  3. Fairing separation
  4. Launch vehicle separation
  5. Solar array deployment
  6. High-gain antenna deployment
  7. Forward Sunshield deployment
  8. Aft Sunshield deployment
  9. Tower deployment
  10. Momentum flap deployment
  11. Sunshield membrane cover release
  12. Sunshield boom deployment
  13. Sunshield membrane tensioning
  14. Mirror support structure deployment
  15. Aft radiator deployment
  16. Right mirror wing deployment
  17. Left mirror wing deployment
  18. Orbital insertion/stationkeeping

And regardless of their previous positions, people with the agency to do so decide(d*):

NO. This is not the end of the mission. We WILL repair this telescope regardless of the cost, regardless of how long it takes, and regardless of how many new technical innovations are required to do so.

So. What would be the """easiest""" point of failure among these to rectify, and how could/would it be done? What would be the hardest, and ditto? I would reckon it would depend on if certain later tasks can be done without former tasks being fully successful, and the functionality of the telescope as an instrument or spacecraft if imperfectly deployed.

*This being submitted a bit more than 12 hours before the scheduled launch time, this is bound to be somewhat alternate-historical, though the close time could allow contrasting "immediate revival attempt" and "later revival attempt" answers.

(Note: Especially because the title was originally written within 250 characters for this to be posted on Quora and then had to be truncated to below 150, if you need any clarification, please ask.)

  • 2
    $\begingroup$ If the \$10,000,000,000 space observatory were safe and healthy and the only thing wrong were that the secondary got stuck (unlikely) it's certainly reasonable to consider a hammer and WD-40 toting smallsat mission that cost 1 part-per-thousand of the original investment. I think this is a great question! +1 $\endgroup$
    – uhoh
    Dec 25, 2021 at 1:12
  • 2
    $\begingroup$ Better suited for Quora, this is quite hypothetical. The premise is counterfactual. $\endgroup$ Dec 25, 2021 at 1:32
  • 4
    $\begingroup$ The problem with "if it were ordered regardless of the cost and technical hurdles" is that the answers can be absurdly ridiculous. "First I'd build a space habitat for a thousand people at L5 and staff it with engineers, then they would..." The scope of possible answers to this question is so large that it isn't a good fit for this site. Can you narrow this down at least somewhat? Otherwise it may end up just being closed as likely to attract answers that are primarily opinion-based. Thanks! $\endgroup$
    – uhoh
    Dec 26, 2021 at 2:55
  • 1
    $\begingroup$ Here's a somewhat similar question, but you can see how it is more carefully constrained Suppose Hubble needed to be fixed by replacing an accessible module again and NASA called for help, what are the most likely mission proposals? and yet it still remains unanswered. $\endgroup$
    – uhoh
    Dec 26, 2021 at 4:08
  • 1
    $\begingroup$ The obvious step #1 is to start shuttle operations again. If money's no object. $\endgroup$
    – Richard
    Dec 26, 2021 at 11:36

3 Answers 3


In 2007 it was decided to add a docking ring to JWST. (1) This was not for any particular servicing capability. Rather the idea was: who knows what capabilities may exist in the future. (The instruments are not designed to be serviced or replaced).

In 2004, NASA began a study of a possible robotic servicing mission for the Hubble. (2, 3). The result of that study was that a robotic mission was too costly and not practical at the time (4). However, while we were working on the LUVOIR telescope decadal survey study, we had a meeting with some of the folks that had been involved in serving the Hubble. One takeaway from that meeting was, what was not practical in 2004, was not necessarily impractical in 2020.

'Bottom line, if something catastrophic happened to JWST, such as the deployment of the mirror assembly getting stuck halfway open, (stealing a phrase from The Princess Bride) JWST would not necessarily be dead, only mostly dead.

  1. https://spacenews.com/nasa-adding-docking-capability-james-webb/

  2. https://www.nasa.gov/home/hqnews/2004/jun/HQ_04173_hubble_robotic.html

  3. https://spacecraft.ssl.umd.edu/publications/SM4R_white_paper_r1.pdf

  4. https://www.space.com/579-study-hubble-robotic-repair-mission-costly.html

  • 3
    $\begingroup$ +1 for the well-sourced answer! "In 2007 it was decided to add a docking ring to JWST." How does the story end? In 2021 does JWST actually have a docking ring? $\endgroup$
    – uhoh
    Dec 29, 2021 at 9:16
  • $\begingroup$ There's no reference to a docking ring on the spacecraft bus. That one article seems to be the only source to reference a docking ring. $\endgroup$
    – Schwern
    Dec 29, 2021 at 10:58
  • $\begingroup$ Schwern, see: space.com/… and reddit.com/r/jameswebb/comments/rjjvoh/… and nbcnews.com/id/wbna18825023 $\endgroup$
    – Vince 49
    Dec 29, 2021 at 17:24
  • $\begingroup$ @Vince49 Yes, they're all referencing the same 2007 report by Brian Berger. I don't doubt the veracity of his report, but I can't find any mention of a docking ring again. The report seems to indicate some confusion between investigating adding a docking ring and planning to add a docking ring. Nowhere does it say they actually added one. $\endgroup$
    – Schwern
    Dec 29, 2021 at 21:50

The question is very broad covering everything from failure to launch to a stuck part requiring a gentle nudge. Someone can address all the engineering solutions to particular problems, but I'm going to address the hard reality that...

Failure is always an option.

Any exciting space rescue has to be cheaper, simpler, and less risky than just building a new one. "At any cost" includes building another JWST and fixing whatever went wrong before the second launch.

When the Cluster mission blew up on takeoff in 1996, after returning what was left to the science team, they built and flew a second one, Cluster II, in 2000. The explosion was considered a four-year delay.

Much of the \$10 billion cost came from development and a 14-year delay. The original cost was \$500 million. Building a second JWST would cost somewhere in between.

Even if it costs another \$10 billion, that may be cheaper and less risky than mounting a repair mission. Robotic repair missions are brand new, and JWST is on hardcore mode.

We have to diagnose the problem(s) from 1.5 million km away on a spacecraft designed to shield itself from the Earth. We have to devise fixes for a spacecraft not designed to be fixed. We have to design and outfit a robotic spacecraft with everything it needs to perform the repair semi-autonomously in zero-G without further damage to JWST and send it 1.5 million km away. While we're developing all that, JWST could be incurring further damage.

JWST is risky and expensive, but it isn't doing anything we haven't done before; it's just doing them all at once.

First, JWST can try to repair itself.

If it turns out to be a snag during deployment, JWST has options.

It can heat and cool the part causing it to expand and contract and wiggle itself loose. It can do this by turning heaters and equipment on and off, and exposing the part to sunlight and shadow.

It can run various motors and gyroscopes to attempt to flex the part and wiggle it loose.

It can use its own thrusters to apply force to the part. It can rotate creating a centrifugal force on the part. It can oscillate to try and shake the part lose. And it can accelerate putting a linear force on the part.

These have all been done on spacecraft before with various levels of success.

If that doesn't work, the likelyhood of a successful repair drops rapidly.

Then, send an observation probe.

Before mounting an expensive and risky rescue mission, send a smaller, cheaper probe to observe JWST and perform diagnostics.

We can't just take a peek at JWST through a telescope. JWST is 1.5 million km away and, if oriented correctly, always pointed away from the Earth. It is not designed to look at itself. A small craft with cameras would help diagnose the problem for a limited cost.

Only then would you assess whether a robotic servicing mission is worth the cost, or if it would be cheaper and simpler to make a new one.

This alone would take months to design, build, test, launch, and travel to L2. During this time JWST may be taking additional damage because...

JWST is very fragile.

Hubble was more like a traditional telescope, a long tube with a mirror inside. It had lens caps which could be closed to protect the mirror.

JWST lets it all hang out. All its parts are exposed to space. It has an enormous mirror, and a very, very, very thin Sun shield.

Any time spent out of orientation or without its sun shield deployed can potentially damage JWST from the Sun. Any time spent in LEO can potentially damage it with debris. Any spacecraft maneuvering near it risks damaging its mirror and sun shield with propellant.

This complicates any repair. The longer JWST has to wait for repair, the more likely it will take additional damage. Any spacecraft which approaches to repair risks further damaging JWST.

Maneuvering near JWST risks spraying the mirror with propellant. In order to apply any direct force, it would have to securely grapple JWST on some make-shift grapple points risking damage to itself and JWST, and that grapple would have to be strong and secure enough to resist the applied force.

Any use of thrusters near the JWST would be done off-angle to avoid sending the exhaust straight at JWST. For example, if it needs to move away from JWST it would simultaneously fire thrusters 45 degrees above and below the direction of JWST creating a net force away.

If the problem requires contact with JWST, it might be cheaper and simpler to make a new one.

This is a job for a robot.

We can't send an astronaut to fix JWST.

The last time a human went beyond LEO (about 300km up) was almost 50 years ago, Apollo 17 in 1972. They went to the Moon 300,000 km away. The whole trip was a little over 12 days. We don't plan on returning to the Moon until 2025 the earliest, and that's costing \$35 billion.

JWST is 1.5 million km away. It took 30 days to get there, one way.

We'd have to send astronauts, and all their stuff to survive in deep space for at least 60 days, and all the fuel to get there and back. This is closer to a Mars mission than a trip to the Moon.

It has to be a robot, and a semi-autonomous one. We've never done that.

The robot has to be semi-autonomous

While robotic servicing missions have been considered for Hubble, that is in LEO where the robot can be controlled from the ground. JWST is at L2 5 light-seconds away. Ground control will be working with a 10 second lag. The robot will have to be semi-autonomous.

This is not unheard of, Mars rovers have a lag of minutes, but it sure complicates things.

Robotic servicing is in its infancy

To repair JWST we need to develop and mature many new technologies.

Robotic servicing and refueling missions are only in the technology demonstration phase.

Right now NASA is working on OSAM-1: On-orbit Servicing, Assembly, and Manufacturing. It will "rendezvous with, grasp, refuel and relocate a government-owned satellite to extend its life". OSAM includes Space Infrastructure Dexterous Robot (SPIDER) with a 5 meter robotic arm. "SPIDER will assemble seven elements to form a functional 3-meter communications antenna".

Those are known problems, specially designed to be performed by robots, in LEO. In contrast, JWST is at L2, it is not designed to be repaired, the fault may not be fully understood, and it is very fragile.

That's a tall ask for a technology only in the demonstrator phase.

Doing repairs in zero-G is very hard

Let's say the robot has a wrench it wants to undo a bolt on JWST. On Earth, when it applies force to the wrench gravity and friction hold the robot and JWST in place; the bolt turns.

In space, noting is holding the robot nor JWST in place. There's only inertia. JWST weighs 6000 kg. When the robot applies force to the wrench, unless its firmly attached to JWST, the robot turns.

JWST has no grappling points nor docking ring designed to secure another spacecraft to. The articles claiming they added one all refer back to one 2007 mention of a plan to add a docking ring for Orion which does not appear to have happened. So the robot has to figure out how to attach itself firmly to JWST in a position where it can affect repairs without damaging itself or JWST.

Fix it without touching it?

Very, very minimal force may be required to release the snag. If that's the case, a robotic mission could apply force without contacting JWST.

It could use a laser to rapidly heat and cool the snagged part causing it to expand and contract and eventually wiggle itself loose.

It could use gyroscopes to spin while spooling out a tiny weight on a filament. Done very precisely, the weight can be made to gently knock lose the stuck part.

Or instead of a weight, a hook to snag the stuck part (doing this near the sun shield is terrifying); once snagged the spacecraft can reel in the line fast and slow (relatively) using its own inertia to jerk the part free. Or it can carefully use thrusters, again angled above and below to avoid sending exhaust directly at JWST.

Which is all to say, if it comes down to sending a repair robot it might be easier and cheaper to just build another JWST. FWIW Scott Manley agrees.

Here's some details about failure at various stages.

Any failure before upper stage jettison

De-orbit it and build a new one.

Why? Let's assume after the failure in the first stage, upper stage, or fairing separation (see below) you can somehow get JWST into a stable orbit and keep it there during the months and years to prepare a serving mission.

The Ariane 5 upper stage is designed for a 30 minute flight. While its waiting in orbit, its fuel and fluids are freezing, leaking, and evaporating. In addition to repairing whatever went wrong, that servicing mission now has to fix and refuel or replace the upper stage. That has never been done.

After fairing separation its mirror and sun shield are exposed. Its flight pattern is specifically designed to protect it from the Sun. While in low Earth orbit waiting for rescue, JWST will inevitably be damaged by the Sun and debris adding to the repair hassle. Like buying a used car that's been sitting unprotected outside, you can fix what you know is wrong, but meanwhile everything is degrading.

If it misses L2.

Learn from your mistakes, declare it lost, and build a new one.

While waiting for the rescue mission to be mounted, JWST will be drifting further and further adding to the delta-V necessary to bring it back to L2. It risks damage from the Sun due to being out of position, orientation, or having its sun shield not deployed.

The repair mission has to chase JWST. Once its caught it, it has to dock firmly with a spacecraft that was not designed for docking, though a gravity tractor is possible. It has to avoid further damaging JWST. And it has to have sufficient remaining delta-V to bring JWST and itself to L2.

This is all before it conducts repairs. Though perhaps the repair should be done by a second spacecraft.

The 6000 kg JWST coasted to L2 with no fuel and minimal propulsion. This pushed the limits of our heavy launch capability. This rescue craft would have to carry fuel for itself and JWST. Due to the tyranny of the rocket equation this is A LOT more fuel.

  • $\begingroup$ JAXA rebuilt a spacecraft, too. (Astro-E2, aka Suzaku) $\endgroup$
    – Joe
    Dec 29, 2021 at 22:46
  • $\begingroup$ Cluster is in no way comparable to JWST though. It doesn't seem to involve any spectacularly unusual hardware, its novelty is mostly in the formation flying, which seems to be mostly a software challenge. Even hardware-wise, the makeup of four spacecraft to begin with means they would have put some more thought into small series production anyway. 4→8 is a much more straightforward expansion than 1→2. $\endgroup$ Dec 29, 2021 at 22:58
  • $\begingroup$ @leftaroundabout It is comparable in that loss of spacecraft does not end the mission; building a new one is an option. If a repair mission is costlier and riskier than building a new one, you build a new one. People seem to forget that. $\endgroup$
    – Schwern
    Dec 29, 2021 at 23:11

Too big for a comment; however not a complete answer. The following cases are all LoV scenarios no matter the cost or complexity:

  • Launch phase (to LEO)


  • Launch phase (boost to Halo orbit)

Likely total loss given that the upper stage makes only one burn. A failure of the upper stage engine will probably not leave the spacecraft in a serviceable orbit. If it's somehow in a stable orbit and got a clean separation despite failure of the upper stage engine a repair mission could be done.

  • Fairing separation

LoV due to unable to generate power. By the time you encounter it it will be tumbling end over end and simply too dangerous to attempt to grapple.

  • Launch vehicle separation

LoV due to unable to unfold and gain ability to use spacecraft RCS resulting in (again) a tumbling vehicle. The upper stage has a short lifetime and cannot hold attitude until a repair vehicle arrives.

  • Solar array deployment

LoV due to unable to generate power. (Same as above.)


Not the answer you're looking for? Browse other questions tagged or ask your own question.