Why isn’t anyone suggesting the super computer? A space shuttle’s operation, or any vehicle for that matter, cannot be as complex as what goes on in a supercomp’s numerous processors. Each micro processor is a technological triumph in itself.
Modern supercomputers are really nothing special at all. Basically enormous numbers of high end PC processors, some with top end GPUs, and a very few with FPGAs. All off the shelf components. The communications networking is more custom on some, but that really typically comes down to a single custom chip. You could buy all the bits that go into a typical supercomputer from any on-line computer store. You just need the money to buy lots and lots of them.
As noted above “highly engineered” is not well defined. It is probably not the same as “well engineered” or “best possible engineering”. Maybe it means the largest application of the engineering arts. It gets really hard to include distributed artefacts like the power grid or phone network.
The ISS would probably be the front runner for that. More engineering man hours have vanished into that than just about anything else. But it is also a lump of bits that has accreted parts over time.
Engineering isn’t construction. The Burj Dubai was largely constructed by cheap labour and cranes. The engineering component is the mix of design and project management. Burj Dubai is just another big building. A team of civil engineers working with CAD systems will design the base building. The fit-out will be done by another sets of engineers - again mostly working with CAD systems, working out the electrical, lifts, water, safety and so on. Probably the biggest challenge is coordinating the construction. There are specialist engineering companies that do that. But again, it is just a very big building. Much of the design and construction is just cookie cutter repetition with small changes.
Cassini was built in a regime where every component, just about down to every nut bolt and screw was individually specified and analysed. Almost every component will have been bespoke in some manner. A great many parts will have been designed from the ground up for this one purpose. And they will have been designed to perform new jobs at the edge of scientific capability in a very hostile environment and yet be as light as possible. The manner in which such systems are built is engineering at its finest and hardest. And it takes a lot of very skilled people a very long time to perform. Extreme engineering techniques are needed. Semi-skilled workers wielding gas axes need not apply.
I suspect the WST might count as a higher level of applied engineering now.
The LHC probably deserves a mention too. But LIGO and Gravity Probe B are some of my favourites.
Another criterion could be sensitivity to performance. If as Mr Vaughan says supercomputers are nothing much, and just amalgams (amalga?) of bits, then they are no different to an average car.
A measure of sensitivity would be how little it takes for any component to fail and completely disable the function of the whole - in other words something that has to be more than the sum of its parts.
Cassini would rate here , but so would a big clock like Big Ben. Lots of components, all having to work in complete precision or the failure of a single one neutralises the overall effect of the remainder. I don’t know much about computers or satellites but would every single component have the potential to completely kill the machine if it failed?
I get your point, but probably not. Mean Time To Failure for any component is… well… a statistical average - so you typically can’t design an item that will last exactly X long, you design it so that the probability of critical component failure within X duration is acceptably low (and ‘acceptable’ is a variable dependent on a lot of factors, such as budget, the outcome of failure, redundancy of parts and function, difficulty of servicing, etc)
You pick your place on the bell curve - at one end, you made it cheap and light, and it might survive, but it probably won’t. At the other end, you made it so that it almost certainly will survive, but it’s no longer cheap and light - it’s always a compromise and a betting game.
I agree we have a rather vague set of criteria here but I’d put forward a modern jet engine.
Simply from a point of view of multiple components, challenging material requirements, harsh environment, long life and mass production. It is effectively a commodity in a way that some of the other (perhaps even more complex) low-volume items are not.
Its got to be the LHC, large hadron collider.
Not only for the number of parts, the total size, the total hours of design, but also for the accuracy, the energy involved and so on.
It is never quite so simple.
There’s a joking saying in racing that the ideal racecar disassociates into a cloud of dust a couple feet past the finish line. Any greater durability is wasted weight you can’t afford.
Seriously, Mangetout nailed it.
If you want 99.999% probability of a complex device lasting X years under Y assumed conditions, then (assuming Y turns out to have been valid assumptions) the thing will eventually fail when the first component gets down to (roughly) 50% probability.
Which will be long after that same component passed through the <99.999% probability threshold on the way to its eventual failure.
There’s more than a hint of truth in that. The classic Lotus 49 (still the most beautiful racing car IMHO) was always on the ragged edge of survival due to Chapman’s innovative brilliance and mantra of “add lightness”
and the turbo era cars of the mid-80’s were insane. The tiny 1500cc engines were boosted in qualifying trim to 1300+ bhp and all they’d be looking to do is run a few laps in order to secure pole and then replace and rebuild. Any more than half a dozen laps in quali trim and they went boom (and often did so anyway)
Well, obviously, the best answer to the posted question, is another question. That is, to call upon the person who asked, to define what they mean by all the terms. Engineering itself, requires that you set specifics, before the task is begun, otherwise it isn’t engineering, it’s just mucking about and trusting to luck.
One of the things we in the tech support world run into all the time, is projects where we can tell that the engineers were put in charge of the entire task, and problems resulted because they DID want to engineer everything. We end up with new products which would have been vastly better and much cheaper, had the engineers NOT insisted on designing an entirely new latch, or screw, or whatever in order to hold things together. There is such a thing as TOO MUCH engineering.
There is no single way to measure this, but a 22-core CPU chip with 7.2 billion transistors is one of the most complex, highly-engineered things ever made by mankind. They are much more complex than a 747 or Space Shuttle, in fact the Intel R&D budget required to facilitate these chips is several times that of Boeing: How Intel Makes a Chip - Bloomberg
The total semiconductor R&D budget is about $54 billion per year. Samsung’s new fab costs $14 billion by itself.
19 new fabs were scheduled to begin construction in 2016 and 2017. From an investment and technology standpoint, it’s like building the Manhattan Project Oak Ridge facilities every single year.
Absolutely. Burj Dubai cost $1.5 billion to build, according to Wikipedia. Presumably that includes engineering cost as well as materials and construction. Cassini cost about the same ($1.4 billion) to build, not including launch vehicle and science operations.
But those numbers pale in comparison to some other industries. The F-35 cost $55 billion just for RDT&E (research, development, test & evaluation) (PDF cite). Boeing reportedly invest $32 billion in the 787 program - this page says $13.4 billion just for development.
Well, of course the Deacon’s Masterpiece is fiction, but if it were real, it would be a truly wondrous piece of engineering indeed.
And the story is told that part of Henry Ford’s genius was in emulating the Deacon’s Masterpiece as closely as possible. After the Model T had been on the road for a while, he sent inspectors out to all of the junkyards, to see what had failed on each Model T to send it there. On learning of a part that had never failed, he ordered that that part be made cheaper, because it was clearly more robust (and hence expensive) than it needed to be.
On computer chips, it’s hard to say that they’re the most engineered thing, given that they’re components of so many other things. A computer chip can’t be more engineered than a 747, if the 747 itself contains one of those computer chips.
That’s not really how engineering works. Designing a car to last 100,000 miles doesn’t mean it will break down at 100,001.
It’s more like, if you need a car to have a 100% chance of lasting 100,000 miles, regardless of driving conditions, weather, maintenance (or lack thereof), you really need to design it for 500,000. Plus all sorts of redundant and expensive automated systems to ensure that it will continue to run even if the owner never changes the oil and leaves it out in the snow.
If you just need a car with a 99% chance of lasting 100,000 miles, assuming regular maintenance and care, well, that’s a much cheaper car.
Well, this thread did give us a factual answer. As it turns out, the answer is entirely dependent on which engineering field one works in.
I can’t find the post, and my memory could be failing me here, but I remember Stranger on a train saying that automobile tires are the most highly engineered objects ever.
Hopefully, Stranger will stop by and correct/update/validate my poor memory.
the notion of crowning something the “most highly engineered thing ever” is silly in and of itself. Not least because someone saying that likely can’t even explain why they say it.
I suppose there’s some validity to that. They last for years despite constant, massive stress and hardly ever fail.
C’mon - it’s only one molecule away from plastic!
Wait, car tires are made out of margarine?