Well, my presumption is that the system is designed with sufficient redundancy and backup that you’re likely to be able to detect and respond to a problem while more than enough control remains to effectively do so. It’s certainly true that if enough independent sensors or actuators simultaneously fail, unpleasantness reliably ensues.
It probably has a boiler safety valve capable of doing that. (Even fully manual steam engines need some way short of an explosion to deal with boiler overpressure.)
Most steam engines have a pug in the top of the fire box that is supposed to melt during an overheat situation. The idea is, the plug melts, and the remaining steam in the boiler will discharge into the fire box and put the fire out.
In real life, the boiler, as it ages, becomes weaker. At some point, the melting point of the plug may be too high to offer any protection. Also, the inside of the engine can become coated with scale and deposits to such an extent, that even if the plug melts, the hole is still clogged.
All boilers have at least one relief valve. They are subject to malfunction too. Eventually, if you have enough poorly maintained engines, you will get an unfortunate combination of operator error and equipment failure and the darn thing will attempt to launch itself into orbit.
I don’t know about steam, but there are certainly autonomous rail locomotives.
The airport rail shuttles at the Orlando, Florida airport appear to be autonomous, though they are probably monitored by a control center.
The trains of the Washington Metro are, or were, autonomous except for door opening/closing and station identification. There is an operator inside. They may have increased the level of manual control since the fatal crash late last decade.
I’m not sure if there are any autonomous trains on the general rail networks, though it may be a combination of company policy, law, industry best practices, and the rail engineer’s union that is preventing automated coal service between the mine and the plant.
You can not just shut down a coal fired boiler by turning off a switch. To shut down a boiler the fire has to be racked, that is the burnning coals pulled accross the grates and broken until they all fall through the grates and out of the fire box.
And that doen not shut down the boiler right away. The fire brick will still be red hot giving off heat to he boiler.
Modern day steam plants be they coal, oil, nuke, or wast heat are fully automated.
They are fully computer operated. Some plants designed to the point to start and operate without human hands on any of the controls. But they still have human operators incase of emergency.
Absolutely false. I work at coal power plants, and have been to nearly 2/3 of all the coal power plants in this country. Have you ever seen a startup, even a warm start? The amount of human interaction needed is very high, as is the amount of continuous human interaction during operation.
Oil plants have less interaction needed but they by no means can startup or shutdown automatically.
If you know a utility coal plant which can “start and operate without human hands on the controls”, please post the name of the plant.
To deal with a runaway locomotive, the first job is is to rapidly reduce the energy directed into driving the train. Lifting the safety valve (which certainly isn’t a switch) does just that. There will then be time for an orderly shut down.
Two things. Remember there is no one on the train we are discussing to lift the safety. And lifting a safety on a running locomotive is very dangerous and at a time of low water could cause the level to drop lower increasing the chance of a boiler explosion.
It would be a strange sort of automatic steam locomotive that omitted this capability.
Right. As noted above, it will always be the case that you can posit enough simultaneous failures to overwhelm an automatic control system. Much the same is true of systems controlled by humans.
I think it’s safe to say that the automated control technology of the late 1800s or early 1900s would not have been up to operating a steam locomotive of the day up to the safety standards of the day.
Since then our automated control technology has gotten vastly better. And our demanded safety levels have gotten vastly, vastly more demanding as well.
Once factor which is usually ignored but lurks just beneath the surface of all failure mode discussions is this:
So for the time being, we leave humans in ultimate control, not because they can always do a better job, but because ultimately we are more comfortable, at the limit case, with humans failing & killing people than we are with machines failing and killing people.
As **Xema **points out, you can always posit a failure setup just beyond the design capabilities of your automated systems. The practical problem today is we can readily build useful things like airplanes and powerplants for which we can’t *affordably *build in the sensor and controller redundancy and automation logic to handle all plausible failure modes.
It’s still cheaper to write manuals and conduct training and have skilled people on-site than it is to design & certify & maintain the “100% failure proof” automatic system which is the minimum standard demanded by society.
Engineers are working today towards even more reliable sensors and even “smarter” control systems. But until societal attitudes & the law catch up, they form a strong barrier to fully automating control of complex systems with dangerous failure modes.
And when, in extremis, the skilled people prove inadequate to the task and disaster ensues, you can always blame them (or their estate) rather than pointing the finger back at either the less than perfect systems, or society’s unrealistic expectation of perfection.
Said another way, human crews make a nice fuse in the liability circuit. This is an important consideration in complex systems design involving lots of big companies & government agencies.