Friday, February 18, 2011

Don't Put Your Tongue On That Capacitor, You Don't Know Where It's Been!

 Lately, because the lack of a mill has back-burnered the final bits and pieces for my Mad Scientist Light Switch, I've been working on the railgun project instead.  From the very beginning, one of my philosophies for this project was that I was going to everything right.  Because time after time, I had seen other amateur railguns which were almost right in most respects, but wherein the builder got one or two crucial details wrong, and it hurt them. The details really do matter, especially when you're trying to take something which works better at large scales and scale it down.

 I've been thinking a lot about safety.  It's one of those details that a lot of a amateur mad scientists don't think about much.  Honestly, it astonishes me that only a small handful of amateurs have killed themselves working with Tesla coils or big pulse caps.  And I think I've had just about all the second chances with high voltage and lasers and explosives that the fates are likely to grant me. I'm going to try to avoid unplanned excitement for a while.

 Government labs are different.  They care about safety, and they spend a lot of money on it.  PhDs don't grow on trees.  But sometimes even smart people miss a possibility and something bad happens.  I shall describe one such incident in hopes that we'll all learn from it.

 This is entirely from memory, because I can't find the paper in which it is described.  I do have a dead-trees copy somewhere in a box in the basement. Therefore, if I find that or (an online) copy, I'll correct any errors I make and point to a copy you can read.

 Back in the 1950s, the biggest homopolar generator in history (so far) had recently been completed at Australia National University.  And it wasn't long before this 500 MJ monster was connected to a new railgun.  Now railguns want a lot of current, and they have a fair amount of voltage drop across the muzzle- several kilovolts in the plasma armature devices that were in vogue back then.  That's a problem if you're using an HPG for your power supply, since they are great at putting out a lot of current into a dead short, but they aren't good at developing much voltage. So, back then, the standard method was to do pulse compression, much as we do today for high powered pulsed-power projects.  They would discharge the HPG into an inductive transmission line which had the railgun connected to the other end.  Then they'd use a series of short circuiting or opening switches to compress the magnetic field of the line, causing the voltage to soar during the pulse.

 The devices used had to be very fast, and had to handle millions of amperes.  They included special fuses, exotic circuit breakers, and mechanical switches driven by high explosives.

 During preparations for one shot, a heavy equipment cart with metal wheels was rolled over a hydraulic hose which provided high pressure oil to the bearings of the HPG. No apparent damage was caused, and nobody thought anything of it until later. That was because there was no pressure on the hose. But the hose HAD been damaged, and a tiny leak had been created.

 All was made ready in the run room, personnel left, and the heavy steel door between the gun room and the control room was closed and locked.

 Various power supplies were charged, trigger generators armed, and then the HPG was spun up to speed, which included pressurizing the bearings. The high pressure oil leaked out of the supply hose in a very fine mist.  Unknown to the operators, and not visible on the remote cameras, oil vapor was now filling the gun room.

 When the shot was fired, the various plasma clouds and sparks promptly ignited the vapor in the room, causing a significant explosion which blew the heavy steel blast door from its hinges, injuring the technician seated nearest it.

 The point being: ask yourself a LOT of questions that begin, "what would happen if..." before you ever apply power to anything.

 Now I've just completed a 1,000 PSI gas injector system which will pre-accelerate the armature before it reaches the rails. This simple "system" consists of:

• a storage plenum (2" Sch.80 steel pipe + fittings)
• an electrically operated valve with high CV*, or flow capability
• a gauge and utility valves for filling and emptying the plenum without opening the main valve
• coupling plumbing and hardware to fit railgun breach-block

Here's a pic:

Later, after commissioning tests, it will be desirable to place a DeLaval nozzle between main valve and gun breach, to increase gas speed within the gun.

 One of the "gotchas" I discovered recently was in the breachblock assembly, which gives me a way of connecting things to the back end of the gun, such as a high pressure injector system.  The breach closure must be designed so it can be easily removed for cleaning, inspection, disassembly, and re-loading.

 The breachblock assembly must contain 1,000 PSI of gas pressure.  Further, the plug which closes it is 2" in diameter, so now we're seeing 3,140 pounds of force on the breach-plug.  Don't worry, I designed it to take that and then some.  It's massive.  But that was only the forces within the two-part system of breach-plate and breach plug-plate.

 Then I looked at the interface -- and more significantly, the gas seal -- between the breach plate and the rest of the gun.  What if my original seal design -- with two seals, one being a "backup" -- would fail at the inner seal?  The additional exposed area inside the outer seal raises the force on the breach plate to about 20,000 pounds-force. I'm not sure I want to load the bolts that much. So I eliminated the outer seal! Now if the single remaining seal fails, all that happens is that gas at design pressure leaks out undramatically.  I don't expect that inner seal to fail, but no one ever expects the Spanish Inquisition!

The big pulser-without-a-name had been designed with interlocks originally, but they were all defeated, broken, or missing by the time it came to me.  I will of course be repairing / replacing them.  I need to sit down at some point and work through the entire sequence / time-line of a shot preparation, firing, and safing.  Eventually, I'll probably even have a check list and some rotating beacons, because those things not only improve safety, they're just good fun for any red-blooded geek.

"Arming test lights on, one through four?"

"Arming test lights on, one through four!"

Ya know, all this work would be a lot easier if I had a team and a budget.  Sure wish I'd gone to school.

* (CV = valve coefficient, which equates to flow restriction through the valve. In COTS valves, it is usually driven by the valve's smallest opening, typically the valve seat)

No comments: