The Shuttle was a very marginal vehicle in every regard. As Stranger has explained many times, it just barely worked as a rocket. It also just barely worked as an airplane.
That’s not a dig at anyone; the thing was a miraculous stretch to, and in many cases past, the state of the art. The failure, if there was one, was in management not fully owning up to just how far out on a limb they’d built this thing.
A few thoughts:
In general, deep delta wings are a bitch to land. They react really weird to ground effect. Which means the pitch forces get squirrely from 50-ish feet down and the vehicle will either want to pitch up or nose over, with the trigger point being pretty unpredictable.
As well, delta winged aircraft touch down at a very high deck angle which means that after the mains are rolling on the ground you have a long way to lower the nose before the nose gear makes contact. The whole time you’re doing that the vehicle is very susceptible to weathervaning into crosswinds. As well the whole time you’re doing that the speeds and pitch forces are still changing. You don’t want to dawdle, because your ability to hold the nose up is quickly failing as the speed decays. But you can’t hurry or it’ll hit too hard.
If you ever wondered what a too-steep landing attempt with insufficient (or ineffective) flare at the end looks like: Crash landing McDonnell Douglas rips off tail - YouTube
That’s roughly what the end of a shuttle arrival would look like if they ran *slightly *too low on energy. But instead of the tail popping off the vehicle would break just aft of the cockpit, at the forward edge of the cargo compartment.
Just a bit less energy than that and you get more like impact & splat.
The shuttle approach from 100,000 feet down to a couple thousand feet is silly-steep. They fly a more or less 3/4ths circular approach to short final. The computer has a model of the winds at various altitudes (provided by chase airplanes that get there 10 minutes early & take all the data).
It then dynamically adjusts the guidance so the circle is the right size to burn off the right amount of energy; not too much and not too little.
The end result is they fly an arcing path down the inside of an imaginary funnel ending at the runway. If that is miscomputed or is inaccurately flown the result is a crash.
There was a mechanical instrument failure in the simulator once. The needle showing the pilot how he was doing left/right got stuck for a few seconds. So he thought he was in perfect shape as he was deviating off the circle about halfway around. It took them and the reentry control crew who also are practicing the event about 15 seconds to notice & recognize the problem & a few more to raise the alarm.
The pilot whacked the instrument with his glove, the needle unstuck & slid went waay out to the side. Holy **** Batman!!! What now??!?
The computer was recalculating the whole time and offered guidance back towards nominal. But they had spent about 99% of their energy margin. The pilots got back into alignment with the revised plan and made the runway with a few inches to spare. Good thing this was a simulator! But the actual vehicle was equally subject to the same malfunction.
A pretty hefty investigation ensued. This event was a big part of the impetus to replace the electromechanical cockpits of the early shuttles with the fully electronic ones like the last shuttles had gotten right from the factory.
Had this been real they may well have lost the orbiter. All for a sticky needle that lasted 15-20 seconds.
All in all, the Shuttle was a cast-iron bitch. NASA’s original calcs were that they’d destroy one in 25. And back when they started the project that was thought to be an acceptable loss rate. As in acceptable to NASA, the astronauts, the public and the politicians.
Somehow over the intervening 20 years it took to get the thing off the ground, all these groups changed their ideas of what was acceptable risk. And NASA HQ’s solution was to ignore the difference between what they’d computed and what they now believed the outside world would find acceptable. That and to hope.
In fact they did better than their original calcs indicated. How much of that was due to tightened margins and vast amounts of overtime spent double-checking stuff and how much was simple good luck I can’t say.
