Designing test rigs for particular engineering problems is a tremendous and creative task in itself.
What were the most complex or innovative?
I’m sure there’s a huge bunch from NASA. I dont have clue about nuke or hardcore military stuff. Even new tests for otherwise “mature” procedures–please jump on me if I’m wrong–such as auto crash tests are tweaked.
The work put into this area is sadly neglected in the public mind.
Do flight simulators count? Or you mean hardware tests, not training?
I only know about testing ICs such as microprocessors. To test whether the design is correct before it is fabbed we have compute ranches with thousands of processors. To make sure the part is made properly takes very expensive (up to a million bucks) automatic test equipment, and software which generates the test vectors.
Because of the speed and density of chips, we put a lot of our testing equipment inside the chip.
I think you hit on an important point when you mentioned automotive crash tests. In cases like testing a new car it’s simply too complex for us to biuld a predictive model. So we do destructive testing. As an old boss of mine was fond of saying “the best way to see how something will break is to break a shitlod of 'em.”
I’ve workied in plants that made industrial pipe fittings and flanges and it was not uncommon to burt test them, run up the pressure untill something ruptured.
I worked in a plant that made electical switches - the kind that flip over to emergency generator power in an instant for places like operating rooms where losing power really isn’t an option. We’d overlod the hell out of those things and throw the switch - makes quite a boom and flash and we’d spend the next few days picking up parts and reassembling the thing. That was fun.
The point is predictive testing can only get so complicated.
Zoid quoted: “the best way to see how something will break is to break a shitlod of 'em.”
That’s the most common sensical common sense I’ve heard in a long while! 
H-bombs excepted.
At NASA Marshall Space Flight Center we have some fairly impressive test facilities. The Saturn-V Dynamic Test Stand is probably the most famous. It’s a vibration/dynamic test stand big enough to hold and shake a fully assembled Saturn-V rocket. I don’t think it’s used anymore though.
For purely linear size, the X-ray and Cryogenic Facility is much larger. (This PDF is more up to date.) It’s a large vacuum chamber (75 ft long and 24 ft diameter), connected to a 1700-ft long vacuum beamline. The original purpose was to test X-ray telescopes; an X-ray generator 1700 ft away is the most cost-effective and reliable way to simulate a distant (astronomical) X-ray source. It was most famously used for testing the Chandra X-ray Observatory. The main test chamber has since been fitted with a cryogenic system that goes down to 35 Kelvin, and currently being used to test the mirrors for the James Webb Space Telescope.
However, I suspect some industries (aircraft, oil, mining, etc) have even more impressive test facilities.
Not nearly as large or complicated as the Saturn V stand, but a lot more blowy-uppy - NASA rigged a Boeing 720 with a remote-control unit so that they could fly it about and crash it. For science.
There’s some video here, and inevitably the idea of a remote-controlled airliner full of dummies being crashed into something with a giant ball of flame has piqued the interest of a certain segment of the population. You know, I always assumed that the test inspired John Varley’s “Air Raid”, but he wrote it in 1977 and the test took place in 1984.
I read something once about a giant wood trestle bridge over 100 feet in the air - assembled entirely with dowel pins, no metal. The purpose was that you could roll a giant aircraft onto it, subject it to simulated EMP to see what effect a pulse would have on the Air Force electronics; without worrying that the ground or surrounding metal was distorting or amplifying results.
Maybe not complex, just hard: a heavy metal coolant for a nuclear reactor. How the heck are you going to continuously cool and cycle molten lead? In a cramped attack submarine even.
Do they then have to design and build test rigs to test the test rigs?
[sub](You just KNEW somebody was going to jump in with that question!)[/sub]
[sub](It’s test rigs all the way down!)[/sub]
Could we consider the *Large Hadron Collider *a test rig for the Universe/Multiverse?
There is a shake table in Japan for earthquake testing that can handle a full size 7 story building.
Si
Could you give a cite/reference for that?
The various engineering tests for the LHC must uav been tremendous. Maybe some people on the board can give a précis.
From Wikipedia
This PDF references the 7 story experiment (it was a reinforced wood structure - the table can handle a fullscale 3 story concrete building).
Youtube video here.
Si
Was it Apollo 10 that (by design) “turned back” after coming within a few km of the surface of the Moon? That could be considered a very complex and expensive “test”, no? True, it was really more like a series of dozens of smaller (but still big) tests.
It is called “Trestle” and still is used (or was a decade or so ago when I was working nearby) You can see it if you take a flight in or out of Albuquerque. It is a few hundred yards south of the runway. It is built in a pit, so even though it is fairly tall, it doesn’t stand much above the surrounding terrain.
The public non-understanding of Zoid maddens me (that is, when they are aware of the design–>test cycle at all).
Say it is reported that an anti-ballistic missile system has failed on a kill, or failed twice.
That’s why they call them tests. The engineering of the system is–so far–operating smoothly.
I liked the improvement program for the Roll Royce Merlin - they took random engines off the production line and ran them until failure. Then they pulled the engine apart, found what broke, and strengthened that part. Rinse, lather, repeat.
The result was an improvement in reliability and performance.
Si
never mind. redundant post.