sevenperforce

I’m pretty sure the tiles are VERY light. Like styrofoam and fiberglass light. All the more reason to embrace the gaps. Trapezoids can work if you allow staggered positions. Correct. Deposited primarily during the entry burn.

The idea would be to have a single RCC “filler tile” shape that would fill the regularly-shaped gap created by iterations of this pattern. If the pattern repeats, the filler tile shape will repeat as well. So just three tile shapes: hex, filler, and nosecone. But I like the trapezoid idea better. I think you would run into issues with size and attachment mechanism. Plus, a vibration-induced crack or latent manufacturing defect could destroy the entire nosecone and result in an instant RUD, whereas tiles are more fault-tolerant. Sort of a Dragonskin body armor approach, eh? It’s not a bad idea but I don’t think it would work out. Attachment becomes an issue because there is less space to work with. You are introducing an entirely new degree of vibrational freedom which needs to be accounted for. And hot spots will form between overlapping tiles because the tiles do not conduct heat well (that is their purpose after all).

Without something like transpirational cooling, I do not think an entirely unshielded stainless steel Starship could survive re-entry. It can survive the loss of a few tiles because that each hot spot will wax off heat to the cooler shielded areas around it, but it does need some degree of shielding. Just far less than the Shuttle. Reinforced carbon-carbon for the nose, the trailing edges of the fins, and any weirdly-shaped transition points. Makes perfect sense. I think they could get away with just four variants: hex files, truncated hex tiles, trapezoids, and right triangles. The right triangles edge the trapezoids. Gap size tolerances are pretty high tbh. At the tile thicknesses we’re talking about, the aero properties are not going to be significant. Especially given airspeed.

Oh, that’s a VERY good idea. Especially because **gaps are okay** and this can take advantage of that property. The ogive will change in curvature but you can pick a trapezoidal shape that can be used in a number of ways. Trapezoids can be flipped alternately to create a mostly-even ring for the less curved sections and they can be stacked edge-to-edge to create a more curved section. If that doesn’t make sense here’s what I’m thinking.... For the less curved sections: ∆ ∇ ∆ ∇ ∆ ∇ ∆ ∇ ∆ ∇ For the more curved sections: ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ (those are triangles not trapezoids but you get the point)

The tiles have gaps between them, so it's fairly straightforward to wrap the tile matrix around a cone as long as you mind the gap. As you can see, the angle starts four tiles down but the gaps are large enough that the bent-over tile sections (unbent tiles on right) don't collide. Wherever there is a collision you take out that tile and you use a large, single, customized tile to fill that particular space. If you don't want to use a customized piece then you can just tile the space irregularly and allow larger gaps.

One possibility would be to go larger, not smaller. Continue the same hex pattern as far up as possible, but have a large, single-piece ceramic shield at the top which fills in all the oddly-shaped gaps. Or, at most, a handful of them. Right. However, you can tile a curved surface with regular hexagons up to a certain degree depending on gap tolerance. Use irregular, custom shapes to handle any gaps too large to tolerate.

Yeah, you're right. Lever arm is just differential torque from a particular engine combination. With the kick-flip, you ideally want to use the engines which are balanced on either side of the CoM (B & C below), because using one side engine and the "top" engine (A) is going to produce a yaw moment, which will have to be corrected with gimbal but then induces a roll moment, and...well, it can be done, but it's messy. So the decision tree looks something like this: Relight C Relight B Relight A Initiate kick-flip Check Relight C Check Relight B IF(C_Relight = OK || B_Relight = OK), Cutoff A ELSE Check Relight A IF(C_Relight = OK || A_Relight = OK), Cutoff B ELSEIF(B_Relight = OK || A_Relight = OK), Cutoff C Birds are also known for being, well, fluffy. I imagine the supersonic shockwave from a re-entering booster would tend to swat them away before they even made physical contact.

SLS will be the most powerful launch vehicle in history. It will also be very stupid. SLS is not powerful enough to do anything useful beyond low earth orbit and it is far too powerful to do anything useful in low earth orbit. Orion was designed to be ferried into low lunar orbit by Altair, making it too heavy and too underpowered to act as a proper command module. Nevertheless, it is what we are working with. So, your challenge is to use KSP to design a capsule and launch vehicle exactly as bad as SLS and Orion. Requirements: Your capsule must carry four Kerbals Your capsule must have 0-0 abort capacity with splashdown. You need an upper stage with one engine, a core with four engines that is too weak to take off under its own power, and two solid boosters. Your core stage should not reach orbit. All propellant tanks should be full. The closer in size and shape to SLS, the better. Your upper stage should provide orbital insertion. After this, your spacecraft should deploy its solar panels before the translunar injection burn, also from the upper stage. Your spacecraft should be able to do an eccentric Munar orbit insertion and return, but not much more. Bonus for aesthetics!

Let's see, the equation is something like: ForThreeEngineStartup: Risk = RiskOfBadRestart * 3 * RiskThatBadRestartCascades ForTwoEngineStartup: Risk = RiskOfBadRestart * 2 So if the risk of a bad restart cascading to otherwise-healthy engines is greater than 2/3, you should only light 2. If it is less than 2/3, you should light all 3.

Pretty sure sure anything that ends up anywhere near the business end of a Merlin -- or a Raptor -- will end up atomized. Not to challenge your brilliance but my brilliant 8-year-old said the same thing as soon as I told him why the rocket went all asplodey.

Originally it was supposed to be a thruster flip but they determined that the pitch authority on the engines is so much better than on the thrusters that it's better to just use the engines for the whole thing.

The flaps do not generate lift but the fuselage itself generates significant body lift. They use the flaps to control attitude and then they use the body lift vector to control where they are headed. It’s the same with the first stage on Falcon 9. They use the grid fins post-entry-burn to tilt over the cylinder to a fairly ridiculous angle of attack and end up with a cylindrical body lift of nearly 1. With Starship the body lift can be significantly higher.

I only recently learned that there was a steam-powered centrifugal cannon deployed during the Civil War. A steam engine was used to spin a barrel assembly which was fed ball bearing rounds from a hopper. A spring-loaded door was used as a trigger to release rounds from the end of the barrel. It did not work particularly well. It boasted a much higher potential rate of fire than any other projectile weapon of the time -- up to 200 rounds per minute -- but its muzzle velocity was substantially lower than the cannons and rifles of the day and the shots were not particularly likely to go in the direction intended by the operator. During World War One, diesel-powered centrifugal guns were developed which claimed rates of fire in excess of 2,000 rounds per minute and muzzle velocities as high as Mach 1 but they were heavy and unreliable.

I think one of the risk-retiring issues for SN8 was controllability at the flip. Starship isn't aerodynamically stable in unpowered prograde so they were worried a bit about getting into that nice smooth bellyflop. Once they proved it was no trouble for SN8, there was no reason to worry about doing it gently for SN9. I also saw the more aggressive nose-down maneuver. Guessing they wanted to experiment more with the range of attitudes and the type of body lift that Starship generates.

Is this from an exam? The primary disadvantages are weight and complexity. All other things being equal, a regeneratively-cooled nozzle extension will be heavier and more complex (and thus more expensive and prone to failure) than a radiatively-cooled one. The advantages are potential for reuse (if you're going to try to recover and reuse) and preheated fuel that lowers the stresses on the engine. Reinforced carbon-carbon, niobium superalloy, or some other ceramic. The niobium alloy is better for the nozzle extension. It melts at 2624 K while Ti6Al4V melts at a paltry 1299 K. Remember that heat rejection via radiation is proportional to the fourth power of the absolute temperature, so a radiative nozzle made out of a niobium alloy will allow almost 17 times as much radiative cooling as the titanium alloy would. Close the throat with the titanium alloy; it has better convective heat transfer properties (since it will be regeneratively cooled) and a higher strength-to-weight ratio.

This slow-mo is amazing. I was certain that the white cloud was the header tank. Looks like I was right.

I really only worried about two things: aerodynamic control authority during unpowered descent and the engine plumbing reliability at relight. At least the first seems to be a non-issue. My best guess is that there’s another propellant flow issue. With the 1-2 relight they do, it almost looks like the torque from the first engine startup might have introduced unexpected slosh in the lines to its companion. Fluid management is hard.

Looks like a late relight on the second engine for the flip. Getting those engines to spin up at exactly the right time appears to be challenging.

There’s the tri-vent.

Nowhere close. Just loading props right now. We will see frost rings on SN9 well before the tri-vent and engine chill.

Got my second COVID vaccine!

Yep. Let's do the gorram math then. SLS Block 1 with ICPS can send 27 tonnes to TLI. Orion is 26.52 tonnes at TLI injection so that's the limit of ICPS. Let's say FH can throw 23 tonnes to TLI, which is probably conservative. It costs 730 m/s of dV to get to low lunar orbit and 430 m/s of dV to get to NRHO. With 23 tonnes of injection mass and a 300 s hypergolic insertion stage, it takes 5.1 tonnes of propellant to reach LLO and 3.1 tonnes of propellant to reach NRHO. Borrowing mass fractions from the Delta-K (950 kg empty & 6004 kg props = 15.8% stage mass to prop mass), that means a stage mass of approximately 490 kg for the NRHO insertion stage and 806 kg for the LLO insertion stage. These numbers are conservative because you don't need NEARLY this much thrust for cislunar burns, so your engine mass is going to be lower (300 s is the specific impulse of small pressure-fed Draco thrusters so this is also conservative). So FH can send at least 17.1 tonnes to LLO and at least 19.4 tonnes to NRHO. This is dramatically more than can be co-manifested with EUS and Orion **plus** it arrives at the destination without borrowing any dV from Orion. But wait, there's more! What if EUS was used in an Earth Orbit Rendezvous architecture? Orion's service module carries 8.6 tonnes of propellant. Assuming it goes to NRHO, it needs to return itself to Earth entry interface, which costs 410 m/s of dV. Orion's dry mass is (26.52 - 8.6) 17.92 tonnes, so getting back to an Earth entry from NRO is going to cost it 2.54 tonnes of propellant at 316 seconds of specific impulse. Getting from TLI to NRHO costs 430 m/s which is a propellant fraction of 13%. This means the most Orion could actually deliver to NRHO while still retaining enough propellant for return (with NO margin) would be 20.1 tonnes. That's a whole half-tonne more than if you just sent your cargo direct to TLI on FH with an insertion stage! But hold on. TLI costs 3.2 km/s. The ICPS will have a dry mass of 3.49 tonnes, of which the engine is 301 kg. If the ICPS tankage is 3.189 tonnes to 27.22 tonnes of propellant (11.72% of prop), then the EUS's 129 tonnes of prop mass should correspond to roughly 15.12 tonnes of tankage plus 230 kg*4 = 920 kg of engines for a dry mass of 16.04 tonnes. If EUS can send 40 tonnes to TLI at 460.1 s, then its m1 would be 40 + 16.04 tonnes which gives an m0 of 113.95 tonnes in LEO or 57.91 tonnes of propellant in LEO. Tracing the calculations backwards, we know that the staging mass of EUS + 40-tonnes-payload = 185.04 tonnes, which means a burn of 2.19 km/s past staging. So if EUS launches with only Orion (26 tonnes), it will have a staging mass of 16.04 + 26 + 129 tonnes and reach LEO with an m1 of 105.3 tonnes, of which 63 tonnes is excess propellant. If you do EOR and already have your extra cargo loitering in LEO, what can you send to TLI with that 63 tonnes of propellant? Getting to TLI requires a propellant fraction of 50.821% at 460.1 seconds, so your maximum mass out of TLI is 123.96 tonnes. Hurrah! You can now co-manifest (with a separate LEO launch) up to 18.66 whole tonnes! Which is less than Orion's carrying capacity and less than FH can send to TLI independently. EUS is stupid. Because SLS is stupid.

It's basically exactly the same as the 5-meter version of the DCSS. Minimally modified. 13.74 meters long, 5 meters wide, 3490 kilograms empty, 27.22 tonnes of hydrolox, and one RL10B-2.

I could do the math on what that would look like if I wanted to......

Not necessary a depress