Starlink quietly lost over $250M in burned satellites this summer | Hacker News Reader

Starlink quietly lost over $250M in burned satellites this summer

tiblur.com

76 pointsmax_hammer76 comments

Lost 200 but launched 345, almost all of which are the latest upgraded version. The satellites are designed to last 5 years (the lifetime is limited by propellant because of atmospheric drag at their low altitude) and the first batches were launched 4 years ago now. As the constellation matures they will be de-orbiting satellites about as fast as they launch them.

Seems to me like the long term viability of Starlink (if it's not already profitable) will be assured if Starship meets its goals. They'll be able to launch way more satellites at once, for less per launch, and with faster turnaround.

How many satellites can be launched at once? I assume that the number of satellites per orbit will be limited and changing the orbit would take extra fuel that either the launch vehicle or the satellite has to provide.

They typically launch 60 on Falcon 9, last time I checked.

The article says average of 50

Launch way more, and I'd imagine the sweet spot for the amount of fuel each satellite gets to shift up considerably: with a launch the size of starship they can saturate the orbits reasonably reachable from that launch for decades (spares remain in higher orbits with less drag I think) and giving one satellite more fuel is cheaper than deploying two satellites with less fuel each. And achieve far more than twice the runtime, because the fuel complement in today's starlinklings is certainly quite far from 100% mass fraction. The current compromise is far quick global reach running on F9, with a generous dose of iteration.

Until all those sats re-enter, burn up and pollute the upper atmosphere

Wasn't there debate around their aluminium frame reacting with the upper atmosphere? The ozone layer maybe?

4 or 5 years of lifetime for literal tons of satellites is definitely not sustainable

Perhaps they could find a way to refuel the constellation?

Unfortunately the physics of this don't work very well. And at least for now the satellites are getting generational upgrades frequently, so the old satellites are obsolete. Maybe in the far future it will make sense to try this, but not anytime soon.

Pie in the sky: in the future when they have the ability to put large starship fuel tankers into orbit, they could match the orbits of starlink and have some kind of drone move between satellites and tanker.

5200 tonnes of micrometeorites hit the earth each year. Star link isn’t going to make a difference.

https://www.scientificamerican.com/article/antarctic-study-s...

With 22.8 tonnes per Falcon 9 launch, and 50 of those per year? That's 1140 tonnes /year of new debris in the upper atmosphere, a 21% increase. I'd say that is quite a significant change.

Almost none of it hits the earth. They burn up in the atmosphere.

"Burn up" is just "into small pieces". The atoms don't magically disappear.

That's what "pollute" implies.

So in ~5 years they've launched 5113 satellites[1]: 2 * 400 kg + 60 * 227 kg + 1665 * 260 kg + 2987 * ~300 kg + 399 * 800 kg = 1663 metric tons

So if they all burned up today it would be around a 6% yearly increase, globally? (1663 / 5200 * 5). And that's probably the most forgiving estimate I could come up with, I think your ~20% number is more accurate for future projection purposes given most of the mass was launched in a much shorter timespan.

Then the real question is what effect does space dust have on the earth's atmosphere (if any).

[1] https://en.wikipedia.org/w/index.php?title=List_of_Starlink_...

There are two more things to consider. 1) The whole satellite doesn’t burn up. Most of it crashes into the ocean. 2) much of the burning up happens in the thicker lower atmosphere where we’re already putting lots of material from jet engines and factory chimneys.

"I'm commuting every day to work with my private jet. 5200 tonnes of micrometeorites hit the earth each year, so my action isn’t going to make a difference."

See how your argument is wrong ?

"It's simple. I started a charity with $10 million that convinces 1 million middle/low class people to keep their heating off during the winter, meaning my jet is now carbon negative. Sure a few people die each winter, but that's another win for the earth."

This is basically the reality of government leaders jetting in to global climate change conferences.

We’re talking about the risks of satellites burning up in the upper atmosphere. The satellites deposit lots of material when they burn up. People are worried that this might be bad.

Lots of people incorrectly think that the upper atmosphere is pristine and you would never normally get metal up there. In fact thousands of tons of micrometeorites hit the earth each year, depositing lots of rock and metal up there.

This means that the burning satellites cause an increase in metals deposited in the upper atmosphere. We’re not going from nothing to many tones. We’re going from many tones to slightly more tones.

There is an open question if the increase is a problem, but it’s less likely to be a problem than if we were going from nothing to something.

Given a choice between bigger fuel containers burning up empty and more copies of all the other parts burning up I know what I would pick. The current very short lived design is a compromise to get to operational coverage fast and for quick rollout of improved designs. Future generations with a larger mass fraction for fuel won't get multiple orders of magnitude more runtime (in orbits that low this would require some bussard drive equivalent..), but a generous multiplication.

What's not clear from the article is whether the ones lost were old ones or newer ones.

If they're losing ones that were close to their max. life anyway that's possibly no great loss, however if they were newer satellites on top of expected attrition that's more of a concern.

> Seems to me like the long term viability of Starlink (if it's not already profitable) will be assured if Starship meets its goals. They'll be able to launch way more satellites at once, for less per launch, and with faster turnaround.

This is just wishful thinking without any data backing it up.

Starship is substantially bigger than falcon 9, with significant more mass to orbit capability. And runs on a cheaper propellant. Also Elon has publicly stated starlink s long term viability requires starship.

Which data are you looking for?

Elon publicly stated many, many, many things…

Starship is part of economic equation (or just a diversion when it comes to starlink, as they were suppose to go bankrupt already if starship wasn’t flying biweekly last year, and people didn’t cancel their holiday plans to report to work, according to the Musk himself), but saying that it solves the profitability of the business, is like saying that low price of fuel solves profitability of the logistics business.

I don’t think anyone said that. It’s just intuitively obvious it makes starlink much more viable when you can launch more/bigger version of it with more fuel for less cost.

Not sure about the rest of that axe you wish to grind.

> I don’t think anyone said that

It’s exactly what op said and caused me to engage in this thread. “Seems to me like the long term viability of Starlink (if it's not already profitable) will be assured if Starship meets its goals”

Yes, viability of starlink. Not profitability of SpaceX.

Starlink isn’t their only income source.

It's unfortunate that this misinformation is being upvoted on this site. The headline is completely incorrect, as is the statistics they're using to base the article on.

https://twitter.com/Marco_Langbroek/status/17055628292254106...

https://planet4589.org/space/con/star/stats.html

There has only ever been 350 starlink satellites that have deorbited, and only 8 within the last two months.

Lost $250M using the advertised launch price of $67M for Falcon 9, which of course is way higher than what it actually costs SpaceX.

The fuel costs about half a million.

Beyond that it's mostly people's times and materials they have stockpiled for economies of scale.

I would guess the lost satellites cost a bit more to manufacture that to launch in a reused Falcon 9.

Don't they waste the upper stage on Falcon 9 launches? (That being one of the reasons they want to transition to using Starship.)

The fairings alone add another million. The whole upper stage is not re-used at all.

They recover the fairings.

does it? why can they then sell it for not much more and run a profit?

Profit is price - cost

There's reason to be skeptical of that. I think their most recent profitable quarter (maybe their only profitable quarter?) was with a profit margin of ~3%.

Granted maybe there's some marginal benefits, eg launching n+1 rockets is cheaper per rocket than launching n rockets

SpaceX puts a lot money into research and development, and build out of infrastructure for future use.

So if the company is breaking even, then its manufacturing & launch business must have good margins.

Which means, self launched rockets cost SpaceX noticeably less than their commercial pricing.

SpaceX posted a Q1 2023 profit of $55 million on $1.5 billion in revenue

And the spend about $2 billion on Starship per year.

So $6 billion in, $2 billion on Starship, $4 billion in other costs.

In 2022 they launched 61 times, which is $65 million per launch on average, they charge $67 million per launch.

They got $2 billion from NASA for 2022.

In 2022 they launched 3 crewed missions, 2 for NASA and 1 private.

They don't charge for their own launches.

when they spend a launch on their own stuff, they lose out on opportunity cost

no because there are not enough clients

They got $4 billion from Starlink users, that should cover it

Not really questionable. They are building Starship, Starship infrastructure, new factories for Starlink, upgrading existing factories, designing new Starlink version, deploying Starlink ground infrastructure, selling Starlink terminals at negative mergin, at a are developing Raptor v3 and commissioning the factory for it. And of course most of those launches are for Starlink so they don't really produce profit.

And while they raised a lot of money, it isn't actually that much. If they had 0 or negative margin on launches the company wouldn't be nearly sustainable.

That they made any profit at all is a damn miracle.

> Granted maybe there's some marginal benefits, eg launching n+1 rockets is cheaper per rocket than launching n rockets

There very clearly is. Every launch company has talked about this. The fixed cost for infrastructure, boats, launch teams and so on is absolutely huge. That is why so many New Space companies bet their company on launching often.

At the same time SpaceX is now recovering fairing on almost all launches. Landings have been perfect. Development is done. And Fixed cost are distributed over a whole lot of launches. They can now fly the newer cores 15-20 times.

The Upper Stage is the biggest cost, people estimate it between 5-15M. And then 10-20M in fixed cost, labor and recovery hardware amortisation. I think it can't really be more then 35M, I would guess less.

...and completely ignoring cost of lost hardware. Aren't those satellites several $100k a piece?

So we have reached that stage of Universal Paperclips

Be careful googling this if you have a weak resistance to incremental games like I do as some hours will fly by quicky...

That's about a 4.4% loss (200 of 4500 satelites which are expected to have a life of 5 years), and the internal cost to Starlink of a new launch is probably far less than the $67 million 'retail' cost of a SpaceX launch.

I think they're prepared to lose satellites, that's the nature of large swarms.

Some losses might be intentional for safety after damage or low propellant. They really can't afford to put any debris on their orbits.

IIRC debris is not a major concern in the orbits Starlink is using. There is a non-trivial amount of atmospheric drag. This will slow down anything without active propulsion, causing them to drop from orbit, and eventually burn up.

Speaking of Starlink, what piqued my interest is that Huawei Mate 60 incorporated satellite call module without being bulky. It is not yet able to do data transfer but in case they can, isnt that practically a portable starlink equivalent?

More like Iridium

Thanks for mentioning that. Its the pro version and it apparently entails voice and text. Which makes it sound more like satellite phones then internet over satellite. Still, quite the interesting feature.

Starlink is low earth orbit, around 550km up, adding best-case latency of under 4ms (550km*2/c ≈ 3.67ms). That allows an excellent service.

Satellite phones (and I’ve confirmed this applies to the Huawei Mate 60) connect with satellites in geostationary orbits, meaning an altitude of about 35,786km, adding latency of around 240ms (35786km*2/c ≈ 238.74ms). This can be acceptable for some purposes, but harms things far more than you might imagine. (Source: personal experience with the Australian NBN SkyMesh or whatever it was called before NBN took it over, only once, about five years back. Look, a lot of the internet is located in the USA, and many things perform vastly worse from Australia than from the USA, e.g. several second page loads instead of under one second, and this satellite connection was basically that but even for stuff in Australia, and several times as bad for stuff in America.)

The mate 60 uses beidou sats which are not quite geostationary but also in a pretty high orbit.

But messaging is only a small part of their task, the main one is geolocation. They're only able to supply short message service where latency isn't really relevant.

I’m out of my depth here, but my research said Tiantong-1 and that that’s geostationary. Adding the keyword BeiDou I do get a few results too, but it looks like BeiDou is GNSS, so I think those might be errors?

Anyway, my point was just that this stuff isn’t equivalent to Starlink. For what satellite phones do, GEO or whatever similar is fine, but the appeal of Starlink certainly requires that it be LEO.

I didn't look into it but I read that in a review, that BeiDou-3 sats have a sidechannel for emergency comms and that this is what's being used. Perhaps that info was wrong, it's hard to get info on this in Europe.

If this is the case it would be too bad though, because Tiantong-1 is indeed geostationary (leading to suitability issues for emergency comms like needing visibility to the Southern horizon) but also only available in the China region.

There's a a Chinese plan for Guo Wang equivalent

https://rusi.org/explore-our-research/publications/commentar...

Is this because they are making a trade-off between surving some amount of solar flare activity vs increasing satellite weight by hardening them to radation?

I imagine it's preventable since the ISS and other large orbiters stay functional forever.

Is this intentional? Is it just happening for newly launched satellites?

If you read the article you can answer both of your questions. The Sun has no intentions, and solar flares don't choose cohorts of satellites.

How do solar flares burn satellites? I thought they just mess up the ionosphere and inhibit communication.

If shielding is insufficient the control systems fry, or at least get wonked.

Dunno if it's what happened in this case, but solar flares make the atmosphere bigger* which increases drag on satellites in low Earth orbit.

That said, the satellites have an expected life of 5 years so it's not necessarily a big deal. Their planned constellation is so large that it's dependent on having a nearly continuous churn of replacement satellites.

* This is a bit of an oversimplification

The atmosphere becomes less geoid-shaped during flares, right?

Day night cycles already do that. The atmosphere expands during solar storms, increasing drag.

Solar radiation pressure is a non-negligible perturbation to an orbit, especially during solar weather events.

Drag is dominant where starlink is.

Indeed, if you don't have enough delta V to maintain your altitude due to atmospheric drag, increased by solar flares. They've lost them before due to this, an entire batch.

Article says 'burned' but also 'burned up', which I infer to mean, burned up during reentry, though it is not precise with the langauge.

As one example, IIRC earlier this year they lost some freshly launched satellites in a batch because of a flare, which made the atmosphere expand in density by up to 50% at their altitudes and didn't allow those satellites to point their solar panels at the sun in time.

Starlink sats are launched into a very low orbit and use their ion thrusters to raise into an operational orbit so any which happen to be nonfunctional deorbit very quickly (in the 2 months between this happening and it being reported, many of the involved satellites had already deorbited) and because it maximizes how many they can put up per launch.

So the satellites were lost, as they needed the panel pointed at the sun to sufficiently power the thrusters to overcome drag, but if they moved out of the safe mode orientation of having the solar panel edge point in the direction of motion, the drag would be too high.

I read the article and have no clarity still. They say it correlates with increased sun activity but no actual proof.

The article doesn't say when the satellites that are being lost were launched.

They could be bringing down the earlier versions of satellites intentionally to replace them with new ones.

Yes, the Sun doesn't care when the satellite was launched, but the current orbit of the satellite and drag profile will decide which ones come back down, if not also applied thrust.

It might not choose but we don't know how similar those cohorts are. It's not the same situation if they already improved the shielding and only old satellites are impacted.

Going forward what is the expected yearly attrition rate? So how much have they spend a year to keep replacing the natural losses of their satellites?

It's unfortunate that this misinformation is being upvoted on this site. The headline is completely incorrect, as is the statistics they're using to base the article on.

https://twitter.com/Marco_Langbroek/status/17055628292254106...

https://planet4589.org/space/con/star/stats.html

There has only ever been 350 starlink satellites that have deorbited, and only 8 within the last two months.

Reply on news.ycombinator.com