I've dealt with way too many MHs that have exploded in enclosed fixtures. (thankfully not while i was around) I've slowly been converting those fixtures over to HPS and eventually LED once the prices come down enough to justify it. 

But weren't these the probe start types on CWA ballasts? There the violent EOL is indeed rather common...

Pulse start systems with standard lamps (quartz arc tube) seem to explode quite often, though the fragments are often contained within the lamp. The ceramic metal halide lamps seem much better. CMH lamps are probably the best new lighting out there - excellent color, brightness, efficiency, and life.

I don't often get to work on discharge lighting systems, but I still pay attention to what is out there.

If I were to estimate the number of ruptured MH lamps ive replaced in the field, it would be under 20 altogether since I started working in this trade and out of hundreds if not thousands of MH fixtures ive serviced. Nearly all of those 20 were universal burn lamps in horizontal fixtures like wallpacks and parking lot shoeboxes. And a few of those were due to misapplication (low wattage lamp on higher wattage ballast). I would say 75% of the MH relamping jobs I do are customers who run their lamps until they completely fail. I have never once replaced a ruptured lamp on accounts with the 12,000 hour/3-year relamping plan.

Pulse start types use HX autotransformer type ballast and that does not hold the arc so well as the CWA, so the worn out lamps ends up cycling and then arctube breaking at rather low pressure, so with not enough force to take the outer with it.

The CMH generally operate at lower pressure and the arctube is not that strong as the quartz one. Plus their volume uses to be smaller than the quartz types, so other aspect reducing the energy in the pressurized gas. Consequently the explosion is usually not that energetic either. Plus the CMH failure mode is rather the fill etching a hole through the arctube with consequent leak to the outer. Even when that means the arctube to disintegrate, it takes longer and at that time the pressure had already dropped...

But even with that, no one have ever proven there can not be any situation leading to more energetic explosion, the industry conservatively assumes as it may happen, so requires either the protective shroud and/or enclosed fixture...

Just relying on "It hadn't happened yet, so it won't happen" is just the mentality making the "cheepeese" products featuring well known "cheepeese" safety/quality...

I would have thought that the cheap brands like Plusrite and Satco would be more prone to explosive failure, but I think poorer quality actually reduces chances of explosion, because a good percentage of those lamps wont even reach EOL to begin with. Lots of these cheaper Chinese brand lamps develop a small hairline crack in the arctube pinch end seal (usually occurs after lamp is switched off and cools down), and the lamp fails to start again. This phenomena typically occurs around he 5000-6000 hour mark ive found.

The ones I work on are indeed wall packs, horizontally mounted.  175W 
Usually these:
http://www.amazon.com/case-Sylvania-64479-Metal-Halide/dp/B00WF5OAGS/ref=sr_1_3?ie=UTF8&qid=1453871642&sr=8-3&keywords=sylvania+M175+U

Or these:
http://www.amazon.com/GE-Lighting-47760-MVR175-Halide/dp/B00076IMY8/ref=pd_sim_sbs_60_1?ie=UTF8&dpID=31yWjW7bMcL&dpSrc=sims&preST=_AC_UL160_SR160%2C160_&refRID=0VRCYKH6APKAWWC1G49Q

Ballasts range in age from 20 years to 2 years. 

Fixtures are these, just with 175W Ballasts in them.

http://lighting.freightliquidators.com/product/5/12214/Rab-Wp2ch150psq-Pulse-Start-Quad-Tap-Metal-Halide-Cutoff-Wall-Pack-Light-Fixture-Ed17u-150-Watt-14-000-Lumens-Bronze.html

I would have thought that the cheap brands like Plusrite and Satco would be more prone to explosive failure, but I think poorer quality actually reduces chances of explosion, because a good percentage of those lamps wont even reach EOL to begin with.

Maybe with lamps turned ON once a week for just few hours, where the system even won't reach few k hours before it gets torn down...

Lots of these cheaper Chinese brand lamps develop a small hairline crack in the arctube pinch end seal (usually occurs after lamp is switched off and cools down), and the lamp fails to start again. This phenomena typically occurs around he 5000-6000 hour mark ive found.

Ironically, such hairline cracks or similar defects reduce the risk of any powerful explosion, just because make the arctube to burst way before (during some warmup, so when the pressure is still low) the lamp could ever reach the condition with very high pressure (when it overheats due to blackening).
With good quality lamp an explosion is indeed way less likely, but if it happens, it will be way more likely the powerful one, just because the quality arctube can accumulate higher pressure before it breaks...
It is due to the small defects, propagating mainly when there are large stresses (few seconds after ignition) weakening the arctube, which act like a fuse and allow the arctube to give up without the large pressure, so when the outer can more likely handle it...

Well, most failed MH's with exploded arctube I've seen only the later kind (first cycling for few days, then arctube in pieces and outer intact), but it is in Europe, so on a series reactor ballast. Only once the outer "detached" from the base and lied on the bottom of the fixture, but that I do not count as an outer explosion (who knows what had happened there - it may be just a manufacturing defect in the outer, with consequent seal failure due to the oxygen)

Having left the topic to rippen a bit (making sure that the big guns (Max and James) didn't have anything to add), allow me to add a couple of notes.

First of all, many thanks for all the information and historical stats. They are indeed invaluable, although they are not "stats" in the exact sense of the world.

As many of you know, the big companies create and publish "survival curves" for all major lamp types. Such curves are made by taking a sample, (say 1000 lamps) and subject it to a specific on/off program for a time period until half of the lamps fail to start. The time until half of the population fails to start, is called "useful life".

The really suspicious thing about modern MHs however (which addmittedly have been shown to occasionally explode (as per the previous contributors' writings) is WHY the companies haven't published survival curves for the phenomenon. Of course, such curves won't be "survival" curves, they'd rather be "(uncontained) explosion hazard" curves.

In any case, such curves wouldn't be a big deal to make; the experiment specifics would be the same, except that you'd prolong the experiment until some lamps actually explode. You then deduct the ones that don't start and you'd get what's called an "uncontained explosion hazard" curve.

The problem then seems to be that IF the big companies published such curves, they'd implicitly reveal to the user that the probability of this mode of failure would be positive for each start, therefore, there always exists the danger of an uncontained explosion, which carries itself back all the way to first use.

Meaning, that such and such lamps have a positive probability of exploding, even on first start. If the experiment for the curve, thus, measures even a single explosion, there exists a positive probability that the lamp can explode, even the first time it is tried out.

You don't have to be a Nuclear Physicist to understand that such tables would be no good, from a commercial standpoint. In other words, publishing data that shows that a company's lamp has a positive prob. of exploding, at ANY time, is, naturally, bad for business.

It seems that the companies chose the way of concealing such data from the public, by omitting it altogether, to avoid negative repercussions coming from the publishing of this information, commercial-wise.

From a purely mechanical/engineering standpoint, this practice is indicative of bad engineering, bad design, and indicative of malicious intent to defraud the consumer.

I won't be surprised, if at some point in the future there is some form of litigation against lamp companies, claiming compensation for damages from accidents of such modes of failures.

It's just a matter of time.

Again, thanks to all the contributors.

--
Ioannis@

I won't be surprised, if at some point in the future there is some form of litigation against lamp companies, claiming compensation for damages from accidents of such modes of failures.

It's just a matter of time.

I believe it has already happened in the USA regarding eye damage following an outer envelope rupture that did not stop the arc tube operation although I don't recall the details. Since warnings are posted on the sleeve, I would imagine any lawsuit would be directed to whomever was responsible for the facility and subsequently ordered the unprotected lamps. Building insurance would also be involved if there is damage to the building.

At my high school, they are really bad about maintence in general, esspesily lighting. One of the MH wallpacks has had the cover missing for about a year now. I can see the yellow socket from accros the parking lot! It just started dayburning about 2 months ago, making an explosion even more likely. How dangerous is this?

Quote from: UVIR on March 09, 2016, 01:55:49 PM

I won't be surprised, if at some point in the future there is some form of litigation against lamp companies, claiming compensation for damages from accidents of such modes of failures.

It's just a matter of time.

I believe it has already happened in the USA regarding eye damage following an outer envelope rupture that did not stop the arc tube operation although I don't recall the details. Since warnings are posted on the sleeve, I would imagine any lawsuit would be directed to whomever was responsible for the facility and subsequently ordered the unprotected lamps. Building insurance would also be involved if there is damage to the building.

I think that's a bit different. I was in the States, back in the 90s, when such lawsuits first showed up. They occured mostly in sports and basketball indoor stadiums, where the ball might have accidentally broken the outer envelope of MHs and as a result some players complained of conjunctivitis after some games. The companies (Sylvania, back then) responded with the so called "Lifeguard" design, which employed a carbon safety fillament in series with the lamp. When the outer envelope was broken, the carbon fillament burned out and disabled the lamp circuit.

In this case (i.e., uncontained explosion hazard) it'd be a matter of time until the litigation is aimed at the company itself and not at intermediaries, even under the auspices of correctly placed insurance.

And that's precisely why the companies haven't published such curves: Because they implicitly point at the manufacturer, so any lawsuits would eventually bounce back at them, one way or another.

Here's a nice analogue: Suppose I built a time-machine that works, but with a "small" catch/glitch: For each and every attempted trip, backwards or forward in time, there is an epsilon chance you will disappear, forever, somewhere in the space-time continuum. The chance is really small, say 0.000021 %.

Still, I wouldn't use it. Not even once.
--
Ioannis@

For the lamp explosion risk: I don't think the makers are hiding it, at least with products and documentation intended for Europe. On the contrary, they require all the fixtures to be able to safely contain the eventual explosion, with the only exception being the "protected" lamps. Indeed, they do not put up any statistical figures, but they do require the fixture designers to anticipate the explosion in any time.
I have rather a feeling, than they in reality do not expect the lamps to really explode violently, but still put up that disclaimer just in case it once really happens. So they can then blame the fixture design (many old MV and even some HPS streetlights are designed as open) for all the eventual damage...