SpaceX COTS Demo 2/3 Updates (THREAD 2)

[JayP]

5000 watts is actually a lot of power. A pretty powerfull laptop computer runs on about 250w. You could probably run a good sized enegy efficient house on 5000w. The incandescent light bulb is one of the most inefficent devices ever invented by man. A 60w bulb only emits about 6w of light. The rest of it is wasted as heat.

[Comga]

5000 W is a lot of power.  It is about twice what I would estimate one could get from that area of panels using commercial grade silicon cells.  Higher efficiency is possible with GaAs cells, which should be resistant to higher radiation doses, longer exposure.  However, I am unaware of GaAs in what looks like the 8" wafer size in the closeup of the panel being tested.

(And 6W of "light" from a 60 W light bulb is much higher than what is achieved, if one tries to use watts in that context, say by making a watt equal to ~300 Lumens which is a good average across much of the spectrum.)

From a quick search this site offers 25% efficient GaAs cells, but only up to 60 mm across, which is much smaller than the cells appear in the SpaceX photo.

[ChefPat]

I guarantee you can run a not so efficient small home on 2.5kw. I live in one.
In my research for a Grid Tied Solar/Wind system I learned a bundle about how much juice I draw & how I draw it. Granted this is only a one bedroom Park Model mobile home, it's quite a bit larger than a Dragon.

[krytek]

Just out of curiosity, how much of an improvement in power output would you estimate a generic solar cell can achieve in space, versus the same cell here on earth?

[Robotbeat]

~30% efficient space-qualified solar cells are commercially available these days and have been for a few years.

(and a solar cell on Earth usually sees around 1000W/m^2 during the daytime without clouds... in space, that's 1361W/m^2... multiply that figure times efficiency and times cell area to get the power output)

I'd seriously like to see proof that these cells would only last 2 months.

And I'm willing to bet they have a significant margin in power.

[Comga]

Just out of curiosity, how much of an improvement in power output would you estimate a generic solar cell can achieve in space, versus the same cell here on earth?

This page shows commercially available solar panels getting up to 17% cell efficiency and 15 W/sqft  (a really above average panel), where as this page shows Cubesat GaAs panels with 28% cell efficiency and ~27 W/sqft and you can do better.

[Comga]

~30% efficient space-qualified solar cells are commercially available these days and have been for a few years.

I'd seriously like to see proof that these cells would only last 2 months.

You know that's not how space qualification works.  The cells will last, particularly if they are GaAs, but the reliability of the attached electronics, protection diodes and the like, would be insufficient to assure high 90's percent probability of functionality for the duration of the mission.  That's why many missions, like the Voyagers and the Mars Rovers, live for many times their design lifetimes.  YMMV, and it should.

[Robotbeat]

Just out of curiosity, how much of an improvement in power output would you estimate a generic solar cell can achieve in space, versus the same cell here on earth?

This page shows commercially available solar panels getting up to 17% cell efficiency and 15 W/sqft  (a really above average panel), where as this page shows Cubesat GaAs panels with 28% cell efficiency and ~27 W/sqft and you can do better.

Right, space-qualified cells (multi-junction GaAs, usually) usually have much higher efficiency than typical polycrystalline silicon terrestrial consumer-grade cells where cost per watt (about $1/Watt at the cell level) is the most important metric. If you're willing to spend $20-50/Watt, you can get much, much better efficiency. For GaAs cells, the material costs are in the range of $2-5/Watt, which is a large figure for terrestrial uses (non-concentrating) but very, very small for celestial uses. The efficiency gains of using GaAs are worth far more for space-rated cells than their extra raw material costs.

But I don't see how using the better cells is somehow going to cost $30 million extra per flight. That's just an absurd number, IMHO.

[Robotbeat]

~30% efficient space-qualified solar cells are commercially available these days and have been for a few years.

I'd seriously like to see proof that these cells would only last 2 months.

You know that's not how space qualification works.  The cells will last, particularly if they are GaAs, but the reliability of the attached electronics, protection diodes and the like, would be insufficient to assure high 90's percent probability of functionality for the duration of the mission.  That's why many missions, like the Voyagers and the Mars Rovers, live for many times their design lifetimes.  YMMV, and it should.

You're not taking into account the context of my post, which is pretty usual on the internet. A previous poster claimed that it was the cells themselves that accounted for the much higher costs, which seems outrageous to me.

I agree that if DragonLab were qualified for the same duration and in the same manner as a typical commsat (with a design life of 10-15 years), then $30 million extra isn't an outrageous number for making sure every part of it can meet that lifetime spec, but we're talking about the difference between 2 months and 1-2 years for the cells themselves, which doesn't make sense to me and was my only point in making that statement.

[Comga]

You're not taking into account the context of my post. A previous poster claimed that it was the cells themselves that accounted for the much higher costs, which seems outrageous to me.

Perhaps not, for which I would apologize.  We are in agreement that $30M can't be blamed on upgrading about 1,500 solar panel cells. I bet much of that cost is for upgrades other than the solar panels.

[Robotbeat]

You're not taking into account the context of my post. A previous poster claimed that it was the cells themselves that accounted for the much higher costs, which seems outrageous to me.

Perhaps not, for which I would apologize.  We are in agreement that $30M can't be blamed on upgrading about 1,500 solar panel cells. I bet much of that cost is for upgrades other than the solar panels.

Yes, definitely agree.

And qualifying hardware for long-duration isn't always easy. Modern US GPS satellites are qualified for 15 years, whereas the Russian Glonass satellites only had a lifetime of about 3 years (so they have to build and launch them constantly or the constellation degrades rather quickly). If qualifying for long-duration were easy, you can bet the Russian birds would've been qualified for 10-15 years a lot earlier (though I believe the newer Glonass birds are qualified for 10). The very first Glonass birds only lasted about 16 months, the next batch 22 months, the next 3 years, then 7 years (current) and the next-gen 10 years. Only the latest Glonass birds use unpressurized electronics, and I am sure DragonLab would use cheaper and easier to develop pressurized electronics.

[MikeAtkinson]

I only reported what Richard Godwin said and I never said that the solar cells cost $30M

A major factor in the cost differential of short duration DragonLab ($90M) and long duration ($120M) is the cost of the solar panels [1].

[1] BIS talk on DragonLab by Richard Godwin 1 Dec 2011.

The other major factors he mentioned were the added cost of the ops, and increased data costs. He said that TRDS was being investigated, but was expensive for 2 year DragonLabs. There are probably other increased costs, but those were the ones he mentioned.

My guess is that SpaceX can probably do the solar panels fully qualified for 2 years LEO for about $1000/W and that the complete power system including power electronics, distribution and batteries probably adds $5-10M to the cost of a 2 year DragonLab mission.

[Robotbeat]

I only reported what Richard Godwin said and I never said that the solar cells cost $30M

A major factor in the cost differential of short duration DragonLab ($90M) and long duration ($120M) is the cost of the solar panels [1].

[1] BIS talk on DragonLab by Richard Godwin 1 Dec 2011.

The other major factors he mentioned were the added cost of the ops, and increased data costs. He said that TRDS was being investigated, but was expensive for 2 year DragonLabs. There are probably other increased costs, but those were the ones he mentioned.

My guess is that SpaceX can probably do the solar panels fully qualified for 2 years LEO for about $1000/W and that the complete power system including power electronics, distribution and batteries probably adds $5-10M to the cost of a 2 year DragonLab mission.

Thanks for the response.

I'm glad my challenge (for clarification/justification) was met. I still think that that is too high of a number as far as cost per watt for only a 2-year qualification (not the 15-year commsat qualification, remember) compared to a 2-month qualification, but it really depends. Your figure seems pretty close for first developing the capability for the first couple long-duration DragonLabs. But once the capability is developed and a few have flown, I have a feeling that it wouldn't add that much to the marginal per-flight cost compared to the 2-month version. But, that's merely a feeling and I don't have the experience in building satellites to back that feeling up.

The mission-support costs seem about right, though.

Thanks for the response, and I'm sorry if my post was unnecessarily abrasive.

[Prober]

And I'm willing to bet they have a significant margin in power.

That's what I'd like to know more about.  What margin has been built in? 

For Example one side ( of 4 panels) of the panels wouldn't deploy, would the Dragon have enough power for a full mission?

[Robotbeat]

And I'm willing to bet they have a significant margin in power.

That's what I'd like to know more about.  What margin has been built in? 

For Example one side ( of 4 panels) of the panels wouldn't deploy, would the Dragon have enough power for a full mission?

A good question. It probably depends on what cargo is being brought up and what level of keep-alive power is needed. I suspect that for the COTS demo 2/3 mission that very little (or no) keep alive power is required for the payloads (i.e. tang, t-shirts, and toilet paper), but that won't be true for subsequent missions.

Given the much greater power capability compared to Soyuz/Progress, I suspect Dragon might be able to operate for long periods with just one array deployed, though careful power management may be required.

[apace]

Is there somewhere more information about their solar panels? Are the motors to open the panels in the segments itself and one at the trunc to rotate the panels? Are the solar panels on both sides of the segments or only on one (as I think makes more sense if you can rotate the panels)?
Can the segments folded back? Are the panels rotate also during attachment at the ISS?

[MP99]

I guarantee you can run a not so efficient small home on 2.5kw. I live in one.
In my research for a Grid Tied Solar/Wind system I learned a bundle about how much juice I draw & how I draw it. Granted this is only a one bedroom Park Model mobile home, it's quite a bit larger than a Dragon.

Implied, but not explicit in your response is that the 5KW array will be operating on a 50% duty cycle (day/night).

Presuming there are charging losses on the batteries, Dragon will see a bit less than 2.5KW of continuous power. Anyone have any data what those losses are likely to be (ie estimate the actual continuous power available to Dragon systems from that array).

cheers, Martin

Edit: "a bit".

[docmordrid]

Dragon PDF says 1,500-2,000 W average, 4,000 W peak

[kevin-rf]

Implied, but not explicit in your response is that the 5KW array will be operating on a 50% duty cycle (day/night).

Presuming there are charging losses on the batteries, Dragon will see a bit less than 2.5KW of continuous power. Anyone have any data what those losses are likely to be (ie estimate the actual continuous power available to Dragon systems from that array).

Assuming perfect panel illumination, 50%. More likely much less than 50% in practice. Don't forget depending on the season and inclination of the orbit, you can get much less than 50% daylight per orbit.

[simonbp]

In addition to being a vertically integrated rocket company, they're becoming a vertically integrated satellite manufacturer ... Probably more profit margin there, anyway, compared to being a launch company.

What do you think is keeping OSC in the black? It's certainly not Pegasus and Taurus...