[dotnet00]

Raptor is also a lot smaller. F-1s were intentionally designed as large and high thrust because it'd have been simpler to build, test and control a handful of F-1s at the time, compared to say, 30 smaller but more efficient engines.

The useful metric on this front would be thrust density, where Raptor 1 is a bit under twice that of F-1A, Raptor 2 is a little over twice of F-1A, and Raptor 3 should be ~2.5x of F-1A.

[capitainenemo]

But with more engines they can tolerate a failure or two (or even more) per launch. If a larger engine fails and you only have 3 of them, you're going to have a bad day.

And, they do seem to test the heck out of their engines, even with 30 of them on a ship.

[sbradford26]

Yes those benefits can now be realized now with modern controls. Back when the Saturn V was designed the control systems necessary to manage 30 engines didn't really exist. Digital control was in it infancy and was only really implemented with a backup on the whole Apollo stack.

Trying to manage that many engines while technically possible with controls of the era (check out the N1) means your control system would be introducing reliability issues instead of adding fault tolerance through redundancy.

[capitainenemo]

Didn't the soviets give it a try? I'd swear they had a large number of engine design way back when also for fault tolerance. Ok, they didn't get it working but I'm pretty sure it wasn't due to lack of digital control... Surely they wouldn'tve even attempted it if it was impossible :)

[edit] ah. That was the N1 you referred to. Ok. So you're saying it was possible, but it introduced more failure points.. So is that why it failed...

[dotnet00]

N1 had a bunch of problems, the engines could not be fired several times, so they tested them by producing them in batches, then test firing one from the batch and assuming all engines in the batch were the same as that one. This obviously isn't how things work, so engines could just be defective from the start.

The second was, as mentioned, that the control systems of the time were not that great, so they had issues properly compensating for engine failures, causing them to cascade until too many engines were lost to get to orbit.

[avmich]

> then test firing one from the batch and assuming all engines in the batch were the same as that one. This obviously isn't how things work

Curious what approach you'd propose in their place?

> The second was, as mentioned, that the control systems of the time were not that great

True. The control system was also cutting edge, and evolved together with engines, and also was much better by the 5th flight - which was scrapped - than it was at the 1st one.

[dotnet00]

>Curious what approach you'd propose in their place?

The approach used nowadays, make engines that can be fired (at least on the ground) multiple times. As far as I'm aware, all current generation American rockets can be static fired on the ground to verify that they work.

Edit: Although, come to think of it, not necessarily true with vacuum engines, but even then, they can test the turbopumps and have enough sensors to find potential issues before launching (at least once enough experience has been built up on the engine).

[ElFitz]

You also need strict QA and minimal deviation both from specs and between engines.

That was another issue the Soviets had a hard time dealing with.