totm nov 2023 SpaceX Discussion Thread

[StrandedonEarth]

 

[tater]

Post static fire fire...

No bueno I think.

Might have stopped

 

Seems OK. Spark ignition test (or issue)?

[CatastrophicFailure]

Eek. 

...well... crap... 

 

2 hours ago, tater said:

Spark ignition test (or issue)?

Too much Taco Bell. 

[CatastrophicFailure]

Wow... so like... I just watched all this on a proper screen, finally. The glowey-drippy part came over a minute after the actual static fire. 

How the heck does stuff get hot enough to still be molten that long after the fiery bit and not make the whole dang thing go up in flames?

[RealKerbal3x]

Oh boy, that didn't look great... I wouldn't be surprised if this set back the schedule a month.

This is why we test, though. Good thing is, it happened on the launch pad, not in flight, and it didn't destroy the vehicle.

[magnemoe]

15 minutes ago, RealKerbal3x said:

Oh boy, that didn't look great... I wouldn't be surprised if this set back the schedule a month.

This is why we test, though. Good thing is, it happened on the launch pad, not in flight, and it didn't destroy the vehicle.

Its an reason why they has 9 and later in the pipeline.

[Flying dutchman]

Perhaps the preburner starter running too lean and the problem is just a valve. The Heat from the event/molten steel burned through one of the pneumatic lines 

 

 

Could i be getting close ?

[RealKerbal3x]

Video of SN8's third static fire including dripping molten metal and comparison with previous static fires:

 

[Spaceception]

If they happen to go forward with SN8, how many more tests would they need to do? Just another WDR, and static fire, then flight? Or more tests than that?

Edited November 13, 2020 by Spaceception
another not nother

[tater]

They need to swap out engines, clearly, then presumably retest.

[RealKerbal3x]

17 minutes ago, Spaceception said:

If they happen to go forward with SN8, how many more tests would they need to do? Just another WDR, and static fire, then flight? Or more tests than that?

Elon did say that they were going to swap out engines, so clearly the damage to the rest of the vehicle is minimal enough to continue testing.

They're probably going to need to do another cryo proof on the header tanks considering the pneumatic issue, and after that probably another round of preburner and static fire tests.

[sevenperforce]

10 hours ago, CatastrophicFailure said:

Wow... so like... I just watched all this on a proper screen, finally. The glowey-drippy part came over a minute after the actual static fire. 

How the heck does stuff get hot enough to still be molten that long after the fiery bit and not make the whole dang thing go up in flames?

Yikes, that drippy bit looked terrifying. This is why I'm still squicked about human landings on Starship.

The only difference between these tests and prior static fires is that they're feeding the Raptors from the LOX header tank. I wonder if the long flow line from the LOX header tank is resulting in an out-of-envelope temperature condition for the LOX. The LOX flow is used to regeneratively cool the LOX preburner, so if its temperature is a few degrees warmer than the LOX that Raptor is used to accepting, it could have insufficient cooling and thus melt parts of the preburner or turbopump.

That would also seem to align with what Elon was saying about the pressure in the header tank rising uncontrollably. There's a GOX line that runs from the heat exchanger to the header tank for autogenous press. If it was coming in too hot and the header tank was already too warm, it could cause boiloff in the tank.

 

Only tangentially related...Elon has COVID and is whining about it.

[magnemoe]

3 minutes ago, sevenperforce said:

Yikes, that drippy bit looked terrifying. This is why I'm still squicked about human landings on Starship.

The only difference between these tests and prior static fires is that they're feeding the Raptors from the LOX header tank. I wonder if the long flow line from the LOX header tank is resulting in an out-of-envelope temperature condition for the LOX. The LOX flow is used to regeneratively cool the LOX preburner, so if its temperature is a few degrees warmer than the LOX that Raptor is used to accepting, it could have insufficient cooling and thus melt parts of the preburner or turbopump.

That would also seem to align with what Elon was saying about the pressure in the header tank rising uncontrollably. There's a GOX line that runs from the heat exchanger to the header tank for autogenous press. If it was coming in too hot and the header tank was already too warm, it could cause boiloff in the tank.

 

One issue with the long pipe might be that you should get some boil off then filling it. 
Assumes if filled with gas oxygen or inert gas before filling. Now you could vent the fist bit if needed, but venting pure oxygen is an fire hazard. 
The pressurized oxygen well it should be pretty simple as in first they have an system to control pressure and I assume they have old fashion safety valves in addition. 
But the LOX in the header tank pipe might a an issue here as it will boil of fast once it stop flowing.  And yes they should have an overflow pipe too, problem is that you will end up with extra lox you need to handle safely. 

[cubinator]

34 minutes ago, magnemoe said:

One issue with the long pipe might be that you should get some boil off then filling it. 
Assumes if filled with gas oxygen or inert gas before filling. Now you could vent the fist bit if needed, but venting pure oxygen is an fire hazard. 
The pressurized oxygen well it should be pretty simple as in first they have an system to control pressure and I assume they have old fashion safety valves in addition. 
But the LOX in the header tank pipe might a an issue here as it will boil of fast once it stop flowing.  And yes they should have an overflow pipe too, problem is that you will end up with extra lox you need to handle safely. 

A higher pressure in the pipe could reduce boiloff.

[RealKerbal3x]

4 minutes ago, cubinator said:

A higher pressure in the pipe could reduce boiloff.

But doesn't increasing pressure increase the temperature? So wouldn't that have the opposite effect?

Apologies if this is a stupid question, my physics knowledge is mediocre at best.

[cubinator]

14 minutes ago, RealKerbal3x said:

But doesn't increasing pressure increase the temperature? So wouldn't that have the opposite effect?

Apologies if this is a stupid question, my physics knowledge is mediocre at best.

I suppose it would - at least while letting the gas in. Once you're at a stable pressure, you can bring the whole thing back down and the boiling temperature for the liquid will be higher.

[magnemoe]

1 hour ago, RealKerbal3x said:

But doesn't increasing pressure increase the temperature? So wouldn't that have the opposite effect?

Apologies if this is a stupid question, my physics knowledge is mediocre at best.

The pressurized oxygen is already hot. They heat it to increase pressure. However it goes into the top of the header tank.  Yes the LOX cool it down but some of the LOX also boil off but assume they will need to keep adding gas. 
I assume they fill the pipe some time before they fire the engines to let it cool and to get out the gas build up. 

[mikegarrison]

2 hours ago, RealKerbal3x said:

But doesn't increasing pressure increase the temperature? So wouldn't that have the opposite effect?

pV=nRT

So yeah, increasing pressure increases the temperature of a gas in a confined volume. But this is the ideal GAS law. It does not apply to liquids.

Boiling happens when the vapor pressure of a liquid is greater than the pressure of the fluid that is resisting it. So pressurized liquids are less likely to boil. (Essentially, it becomes hard for the vapor to make room for itself.)

Edited November 13, 2020 by mikegarrison

[sevenperforce]

1 hour ago, magnemoe said:

The pressurized oxygen is already hot. They heat it to increase pressure. However it goes into the top of the header tank.  Yes the LOX cool it down but some of the LOX also boil off but assume they will need to keep adding gas. 
I assume they fill the pipe some time before they fire the engines to let it cool and to get out the gas build up. 

Remember the ideal gas equation: PV = nRT. Recall, also, that the boiling point of a liquid scales with changes in the temperature of the liquid.

Pumping hot, pressurized GOX into the top of the header tank will not significantly increase the temperature of the LOX itself, because the heat transfer coefficient between gas and liquid is very low. However, the heat transfer coefficient between liquid and solid is higher and so the LOX will warm up a lot while flowing down the feed line; that LOX in the feed line will be significantly warmer than the LOX in the header tank or in the main tank because it has more surface area exposure. Adding insulation to the feed line might help matters.

Another question I'm not sure about -- is there a phase transition in the LOX or the CH₄ between the feed lines and the preburners? The schematics I've seen have both propellants remaining liquid all the way through until they enter the preburners, but if there is a phase transition, then the enthalpy of vaporization would be the primary source of cooling and the entry temperature wouldn't matter.

The more I think about it, the more convinced I am that adding gas accumulators to Starship is going to be necessary.

Just posted about Elon:

 

[RealKerbal3x]

19 minutes ago, sevenperforce said:

Just posted about Elon:

I feel like this would be better suited for the COVID-19 thread.

[tater]

That tiktok is flatly wrong. All medical tests have a false negative and false positive rate.

The true FP rate for PCR is very low, but not 0. A number I have seen is 0.4%. At low prevalence as a prior, a substantial % of tests become FPs.

Due to the exponential growth with each cycle, however, PCR detects virus LONG after people have recovered. At a cycle threshold of 40, out to ~12 weeks. People are not infectious after ~9 days (higher the sicker they are, with the longest times for ICU patients). They have been unable to culture live virus in anyone with a test that required 35+ cycles (most tests are 38-42 in the US), and from what I have read, no one is infectious if the CQ was over ~28-30. There was a talk at Santa Fe Institute I saw that suggested (months ago) that all the tests should be done at MUCH lower cycle thresholds, intentionally missing "cases" where people are already over it in favor of only detecting people actively infectious (very low CQ). So in an attempt to reduce FP rate by more cycles, they detect uninteresting "cases" of COVID-19. Uninteresting because they are long since recovered (or were never in fact ill). If people are infectious for 1.3 weeks, and the test detects for 12 weeks, the rate at which the test detects people we don't care about (not infectious, not sick) is closer to 90%. So True FP of well under 1%, and an effective FP rate closer to 90%, since only infectious people matter WRT testing (those are the only ones who need to isolate to stop spreading).

In short, a positive test is meaningless without concurrent serious symptoms, all that matters is hospitalization for respiratory issues. My wife has operated on a bunch of COVID patients at this point, none of whom were in the hospital for COVID-19, their COVID status was incidental to a positive test after admission for what they required surgery for. They of course end up on the COVID floor, and in the state stats as COVID hospitalizations in spite of receiving no care for their nonexistent symptoms, they are merely recovering from surgery. Not saying there are not real cases, I'm saying that tests are often detecting people who are not actually sick.

EDIT: It's important to remember that PCR doubles the RNA each cycle. So if a sample were to be contaminated by a single SARS-CoV2 RNA in the lab, do the math on 40 doublings. It's not like the local lab running COVID-19 PCR testing is BSL-4 clean. You can have a test that in ideal (nonexistent) conditions cannot FP, but will in real world settings, even being careful, that's why the quoted specificity (when you can find it) varies by an order of magnitude. I've seen 0.4%, and some say 4%. Anyone with a + PCR test should be retested at a lower cycle threshold, ideally under 30. If they still positive, they are likely infectious and should isolate.

 

Edited November 13, 2020 by tater

[CatastrophicFailure]

 

[tater]

^ Luckily for SpaceX recovery conditions for the capsule with crew aboard are probably tighter than for the booster

 

[mikegarrison]

1 hour ago, tater said:

That tiktok is flatly wrong. All medical tests have a false negative and false positive rate.

The true FP rate for PCR is very low, but not 0. A number I have seen is 0.4%. At low prevalence as a prior, a substantial % of tests become FPs.

Due to the exponential growth with each cycle, however, PCR detects virus LONG after people have recovered. At a cycle threshold of 40, out to ~12 weeks. People are not infectious after ~9 days (higher the sicker they are, with the longest times for ICU patients). They have been unable to culture live virus in anyone with a test that required 35+ cycles (most tests are 38-42 in the US), and from what I have read, no one is infectious if the CQ was over ~28-30. There was a talk at Santa Fe Institute I saw that suggested (months ago) that all the tests should be done at MUCH lower cycle thresholds, intentionally missing "cases" where people are already over it in favor of only detecting people actively infectious (very low CQ). So in an attempt to reduce FP rate by more cycles, they detect uninteresting "cases" of COVID-19. Uninteresting because they are long since recovered (or were never in fact ill). If people are infectious for 1.3 weeks, and the test detects for 12 weeks, the rate at which the test detects people we don't care about (not infectious, not sick) is closer to 90%. So True FP of well under 1%, and an effective FP rate closer to 90%, since only infectious people matter WRT testing (those are the only ones who need to isolate to stop spreading).

In short, a positive test is meaningless without concurrent serious symptoms, all that matters is hospitalization for respiratory issues. My wife has operated on a bunch of COVID patients at this point, none of whom were in the hospital for COVID-19, their COVID status was incidental to a positive test after admission for what they required surgery for. They of course end up on the COVID floor, and in the state stats as COVID hospitalizations in spite of receiving no care for their nonexistent symptoms, they are merely recovering from surgery. Not saying there are not real cases, I'm saying that tests are often detecting people who are not actually sick.

EDIT: It's important to remember that PCR doubles the RNA each cycle. So if a sample were to be contaminated by a single SARS-CoV2 RNA in the lab, do the math on 40 doublings. It's not like the local lab running COVID-19 PCR testing is BSL-4 clean. You can have a test that in ideal (nonexistent) conditions cannot FP, but will in real world settings, even being careful, that's why the quoted specificity (when you can find it) varies by an order of magnitude. I've seen 0.4%, and some say 4%. Anyone with a + PCR test should be retested at a lower cycle threshold, ideally under 30. If they still positive, they are likely infectious and should isolate.

 

I find this post confusing.

If the PCR test amplifies the signal as you repeat the cycles, it seems like this means high cycle tests should:

• have fewer false negatives
• have more false positives, unless it is impossible to amplify a false positive signal

Lower cycle tests would therefore increase the false negatives but decrease the false positives, wouldn't they?

[tater]

35 minutes ago, mikegarrison said:

I find this post confusing.

If the PCR test amplifies the signal as you repeat the cycles, it seems like this means high cycle tests should:

• have fewer false negatives
• have more false positives, unless it is impossible to amplify a false positive signal

Lower cycle tests would therefore increase the false negatives but decrease the false positives, wouldn't they?

 

Yes.

They are increasing sensitivity with cycles, but specificity is constant (they are amplifying anything it amplifies). If they truly only reproduce actual target RNA each cycle (not fragments possibly shared with other viruses), then it could in fact be nearly 100% specific (contamination always possible).  If it positives on fragments, then that could be a problem as well (they could be shared).

Flu test specificity (also PCR) is apparently 90-95% for example. The actual false positive rate depends on prevalence, obviously—if 99% of the people are actually negative (use 95% specificity), you will expect to see ~5 in 100 tests as false positives instead of the ~1 actual positive. If 99% of the people are infected, then you will expect 0.05*1 false positive—~0 (5% of the true negatives are rendered FP). Somewhere around 5% prevalence, we'd see 5 true positives, AND ~5 FPs.

Lower CQ means that only people with a lot of virus to start within the sample get detected as positives. You will possibly miss cases (false negatives), but you end up with fewer false positives. More importantly, the people who test positive at 25 are almost certainly infectious. Isolate them!

Early in an outbreak, the incentive is to "catch all the cases" even if you tell someone who isn't sick to stay home, so what, no big deal. When prevalence lowers, the problem really reverses. When by far most people are true negatives (outbreak nearly over), ANY FPs are concerning. If it's 0.4%, and a country does 1 million tests a day, and 0% are actually infected—we'd still see 4000 "cases" a day. If you test everyone admitted to hospital, and 1% of all hospital admissions die, then 4 in every 100,000 hospital admissions will "die of COVID," even if 0% actually have COVID. There are ~100,000 hospital admissions per day in the US on average, every day, every year. Looks like actual hospital death rate is ~2% though. So in an end game of COVID-19, if the test FPs at 0.4%, we will have 8 deaths a day in the US forever (none of which are actually COVID). The obvious solution is to only test people who are sick. They only test 75k people/yr for flu, even though 1000 times as many actually get flu. They test tough to treat people as a rule out. If someone actually sick gets tested, the prior is not at large prevalence (which it is for random testing), it's whatever the doc thinks (if the patient looks like a COVID patient and tests positive, they are probably positive).

 

Edited November 13, 2020 by tater