Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
Networking

Networking in Extreme Conditions? 89

Posted by Cliff from the this-article-will-self-destruct-in-5-seconds dept.
222 asks: "Mission: Create an intermediate distribution frame. Difficulty: A few feet away, industrial equipment will be generating roughly 2000 degree heat. Bonus: Keep the network switches inside the IDF from melting. Does anyone have experience in making IT work in such extreme conditions? Is there an enclosure in existence that can handle this type of abuse? This is essentially what I've been asked to accomplish, and now I'm asking my fellow readers for help: Can it be done?"
This discussion has been archived. No new comments can be posted.

Networking in Extreme Conditions? More

Networking in Extreme Conditions?

Comments Filter:

  • hmm (Score:2, Interesting)

    by mastershake_phd ( 1050150 )
    Your going to need more than just an insulated box. I cant think of any other way to do this than a water cooled box. ie a box with some pipes in the walls then you put in your hardware. Not sure refrigeration would even work in these conditions.
    • We need to start using this stuff [wikipedia.org]. I'd like to see it at Home Depot eventually.
      • It probably will sooner rather than later. This would make 1 heck of a home insulation.

      • Re: (Score:2)

        by Nutria ( 679911 )
        We need to start using this stuff. I'd like to see it at Home Depot eventually.

        To heck with that fancy stuff! A couple of inches of asbestos will do the trick quite nicely, and is a lot more durable than aerogel...

    • Maybe a ceramic surface on the enclosure will help. I know it's used in exhaust piping and other high-heat applications, but whether it's viable I don't know -- I have no idea what (and how big) an IDF is, or how you would go about insulating it.

      But if you can't move the heat away from the equipment, can you move the equipment away from the heat? Why is it that it needs to be that close? Can't you put only a subset of the equimpent (sensors, antennae, &c) close by and move the bulk of it to a safer loca
      • I guess another question might be, who is goign to operat this thing and what is going to protect them? Remote access will not do 100% or the time.

  • Easy (Score:2, Interesting)

    by DavidHOzAu ( 925585 )
    Ask an avionics engineer.

    • Re: (Score:1)

      by Geminus ( 602334 )
      Having worked on avionics, I'd recommend quite a bit of usage of peltier junctions with an avionics oil cooled system. Your biggest threat isn't going to be heat, it's going to be condensation. You'll no doubt be using some seriously shielded fiber as well.

  • Consultant? (Score:5, Insightful)

    by pdbaby ( 609052 ) on Friday January 12, 2007 @06:54AM (#17571228)
    Cisco make some pretty hard-wearing equipment, but I'm not sure they make anything capable of withstanding 2000 degree heat. Perhaps you should be talking to a consultant that specialises in industrial and extreme condition networking instead of slashdot?

    • Re: (Score:3, Funny)

      by baldass_newbie ( 136609 )
      Perhaps you should be talking to a consultant that specialises in industrial and extreme condition networking instead of slashdot?

      One of two reasons why he'll eventually be fired.

  • Try a fire safe... (Score:1, Informative)

    by mad zambian ( 816201 )
    as a starting point. These beasts are supposed to be fireproof / fire resistant, so they should protect the stuff inside, however, then there is the problem of internal heat build-up.
    There are peltier based coolers available designed to cool sealed boxes, but how they would cope with what sounds like a severe radiant heat source near by, who knows?
    You are also going to have serious problems keeping the cat-6 from melting which implies metal ducting insulated from the heat source as well.
    There are bound to b
    • No. Fire safes give off a kind of steam when they are being heated. The steam keeps the papers inside from burning up. The fire safe stops being effective after a certain period of time when it runs out of material to create the steam with.

      Therefore, the fire safe has two problems for this application.

      Limited duration for effective fire protection.

      Steam that will short out anything electrical in the safe.

      Fire safes are designed to keep paper from burning. Nothing else.

  • Watch out for interference (Score:5, Informative)

    by FooHentai ( 624583 ) on Friday January 12, 2007 @07:26AM (#17571388) Homepage
    Having worked on and in a networked environment for a 24-hour steel mill, I can predict that you'll have just as much trouble with interference as you will with heat. Assuming you overcome the heat issue, whatever's generating it will also be generating so much interference you'll be lucky to get 10mbits from a gigabit line.

    Our solution was to use fiber for pretty much everything, to the point that we were using so much that it was simpler to use fiber for every tasks than mix in cat5.

    Regarding the heat issue, if you're based as an IT specialist in an industrial environment, then you have to liase with the site engineers for this kind of task. It's outside of your experience/training/knowledge, so don't succumb to vanity and assume that you have the intelligence to pull through it unaided. Such a place will already have other heat and interference-sensitive equipment which has been installed by engineers on-site, and they'll be able to do the same for your equipment.

    Expect a hefty price tag for installing a suitable thermal casing and ventilation system. Or after due consideration and consultation, if it can't feasibly be done then that's the end of the discussion.
    • The parent summed up the entire scenario for you perfectly.

      I trained as a mechanical engineer, and I work now in IT.

      Once you get into an industrial environment, recognise you don't know what you are doing.

      The site engineers should be quite happy for you to outline the locations and temperature range you need, and work out a solution to that problem. Be specific about the scope of your situation and don't try to solve the problem for them.

      And if you have to, work back up the chain a bit and see if they have alternate paths to send the cable, maybe that will help... The fibre is possibly a necesity too. I'm finishing up an airport project ATM, and everything that is important that needs to travel distances goes by fibre.

      Just remember, YOU'RE the clueless client now. Be clear and specific with exactly what you need achieved, and they should be able to help. :)
      • I'm a cubical-jockey IT guy. I often end up helping out installing IT equipment at some of my companies plants, and although it's nothing near the enviromental setup as the poster described; you gotta realize that the plants have Plant Engineers that know thier setup just like you know the setup of your network or whatever.

        I'll also agree with the people that stated you better have a pretty real budget here. Industrial-hardend equipment is expensive for a reason.

        • Re:I second endorse the above comment. (Score:5, Informative)

          by Mysticalfruit ( 533341 ) on Friday January 12, 2007 @10:18AM (#17572630) Homepage Journal
          My experience with plant engineers is that they've always got people coming around and messing with their setup. So, they'll appreciate it when you come to them and say right to their face "I'm out of my element here and I'm like your expertise".

          Building interpersonal bridges has never hurt anybody's career (we'll except for those people who attented Duke Cunningham's luncheons, but that's a different story!)

          Also like other posters said, more than likely they'll have a host of questions and potential problems (such as the whole places power systems are all DC) that you hadn't even thought about.

          • Re: (Score:2, Interesting)

            by Hemogoblin ( 982564 )
            My Grandpa used to run his own refridgeration engineering company. He told me a story about one job of his he did for a meat-packing plant. Meat needs to be hung in a refridgerated environment for a certain length of time. After installing the new refridgeration system for the room, a few weeks later he received a phonecall. Apparently the system "wasn't working right" and he'd better come and fix it. It turns out a supervisor had cut a damn huge hole in the wall and was using the cold air to cool off the w

    • Re: (Score:2)

      by mkosmo ( 768069 ) *
      Could you have used cat6 stp and gotten better results? I can understand that being a problem (interference) in an area with that kind of constant metal and ever changing electro-magnetic properties. I would think using stp instead of utp may help some of that though. If you were still having a problem... wow. I am impressed. Or coax. 10base2 still may have its uses :)

      • Re: (Score:1)

        by eta526 ( 833281 )
        STP, if installed right, will lessen interference, but fiber will eliminate it entirely. The OP was right to go with fiber, given that they had the budget for it.

    • Re:Watch out for interference (Score:5, Informative)

      by Panaqqa ( 927615 ) * on Friday January 12, 2007 @08:37AM (#17571774) Homepage
      I agree with this. Having installed networking in an ironworks, I can tell you that the way to go is fibre - it will withstand both extreme temperature and extreme EMC/RFI interference. Why don't you take a quick look at Moxa [moxa.com] and what they have to offer? They can be a starting point for you.

      Good luck! And I hope your client/employer has the budget for this one - we're not talking commodity priced stuff here.

    • Re: (Score:2)

      by udderly ( 890305 ) *
      Regarding the heat issue, if you're based as an IT specialist in an industrial environment, then you have to liase with the site engineers for this kind of task. It's outside of your experience/training/knowledge, so don't succumb to vanity and assume that you have the intelligence to pull through it unaided. Such a place will already have other heat and interference-sensitive equipment which has been installed by engineers on-site, and they'll be able to do the same for your equipment.

      Having dealt with
    • The OP is correct. Heat is much simpler to handle than the EMC issues you might see here. If there is, for example, inductive heating going on then you could be looking at some EMC of mind bending levels. You would need to pay special attention to near field effects as most RF enclosure only focus on containing the minor RF originating from what is in the enclosure, not from some massive magnetic field coming from a ziga-watt inductive heater, transformer, etc. And ESD in industrial environments can loo

      • Re: (Score:2)

        by Andy Dodd ( 701 )
        I agree wholeheartedly.

        I'm an EE who is working on networking in an extreme environment myself (although with very different requirements as this person), and I agree - find an expert not just in extreme environments, but in YOUR extreme environment.

        Fortunately for myself, I'm surrounded by experts with years of experience with the environment I'm dealing with.

  • Repeat? (Score:2, Informative)

    by Anonymous Coward
    There was recently a similar Ask Slashdot asking us readers how we would handle severe cold environments. Most people said to keep the equipment as far away as possible.

    I suggest the same, since I've never heard of solder that can stay solid at those temperatures, much less sillicon (or whatever crazy elements are being used nowadays) not turning into the Magical Blue Smoke that makes machines run.

    You're asking for a great deal of problems (or intense job security) if you put anything electronic near that h

    • Re: (Score:2)

      by mikael ( 484 )
      I suggest the same, since I've never heard of solder that can stay solid at those temperatures, much less sillicon (or whatever crazy elements are being used nowadays) not turning into the Magical Blue Smoke that makes machines run.

      One computer lab I used to work in, was located in an old chemistry room - it had single glazing with continental style slatted windows (a safety feature to stop fumes from building up). Heating in the entire block was on from 8.00am to 5.00pm. In Winter, just about every PC woul

  • Why ? (Score:5, Informative)

    by SomethingOrOther ( 521702 ) on Friday January 12, 2007 @07:49AM (#17571516) Homepage

    OK I'll bite.

    I work with industrial networks...
    WTF do you want networking kit so close to such a heat soure?

    The motors/sensors etc on the machinery (ie, your heat source) should fead back into PLC [wikipedia.org] controalers which are *bombproof*.

    Only then well away from the machinery do you fead the signals from your PLC, into a computer to network and process data. (Or better than this, a robust NI fieldpoint network module [embeddedstar.com] to network the data, and process the data well away from the factory floor.)

    Putting network switches in such a harsh environment in taking the piss. I'm not saying it can't be done, but you need to look at how everybody else reliably does it, and think why you are trying to do it another way.

    Oh and BTW, 2000 F = 1093 C Most of us use SI units :-)

    • Re: (Score:2)

      by 222 ( 551054 )
      It's not a matter of feeding data directly into our cabinet; (as you suggested) have that fed to a PLC. The situation is a bit different than described in the summary. The IDF is already in place, and the decision has been made to place a new furnace in this area. My options include moving the IDF, which would require an impossible amount of downtime, setting up a new IDF with identical equipment and run both in parallel, which is more expensive but tolerable regarding downtime, or "fireproofing" our curren

      • > The IDF is already in place, and the decision has been made to place a new furnace in this area.

        Bloody hell!
        Was whoever made this decision in full possetion of all the facts when they made it?

        Seriously, you are going to struggle with this, no matter what solution you come up with.
        With interference on the line, heat shielding will be the least of your problems. Personaly I'd aproach whoever made this decision and thell them it can't be done without spending some serious money. Good Luck :-)
      • I don't know jack about IDFs, but it seems to me that staying in the existing location (by the furnace) is the proverbial Damocles' Sword: one mishap and BOOM. I imagine that that would result in intolerable downtime as well. Personally, I wouldn't want that hanging over my head. :-)

        It sounds to me that your only good option is to build a second IDF in parallel and then switch over to it BEFORE the furnace goes in.

        Good luck!

        • Re: (Score:1)

          by eta526 ( 833281 )
          Given the new info that the furnace is going in after the IDF, which must then be retrofit, this is probably the best option yet. Your equipment in this location is NOT going to last, so get another unit set up before the current one fails. You'll wind up spending less time and money in the long run by far. Although geeks do make miracles happen every day, there's a cost-benefit analysis to consider. It sounds to me like you have four options:
          1) Put the furnace elsewhere
          2) Put the IDF elsewhere
          3) Prepare to
      • Not knowing the costs (or the particulars of networking in an industrial environment), I'd look into the cost of a new IDF in parallel to designing the fireproofing solution.

        You may very well end up getting a more reliable service for the same or less money by setting up a new IDF in a sane environment, than going forward with the fireproofing.

        Good luck!

      • It's not a matter of feeding data directly into our cabinet; (as you suggested) have that fed to a PLC. The situation is a bit different than described in the summary. The IDF is already in place, and the decision has been made to place a new furnace in this area. My options include moving the IDF, which would require an impossible amount of downtime, setting up a new IDF with identical equipment and run both in parallel, which is more expensive but tolerable regarding downtime, or "fireproofing" our curren

      • Re: (Score:2)

        by afidel ( 530433 )
        The good thing about doing a parallel install is if you can use identically (or near so) speced equipment you get a twofor, you get your new IDF and you get equipment to use in your DR planning. Remember DR isn't just about offsite moves, it can also be about what do you do if your critical equipment dies.
    • He may be working in a Foundry, or other metalcasting location. 2000F (1380C) is around the melting point of IRON.

      Although, I'd be hard pressed to know what kind of computer equipment would be located so close to a furnace that can burn that hot.

      Maybe, he works at Nasa?

  • Is that Fahrenheit, Celsius or Kelvin? Since even NASA is metric now, can I have that in some SI unit?

    • Re: (Score:2)

      by Ashtead ( 654610 )

      2000 F = 1093 C = 1366 K

      2000 K = 1726 C = 3140 F

      2000 C = 2273 K = 3632 F

      It is damn hot for the electronics no matter which kinds of degrees are used.

      My suggestion would be to try and work towards relocating the whole networking kit somewhere else on the site where there is less heat. Or use some of the same tricks that the presumably adjacent electronics already use, whatever these are.

  • Shielding (Score:5, Informative)

    by onyxruby ( 118189 ) <onyxrubyNO@SPAMcomcast.net> on Friday January 12, 2007 @07:59AM (#17571584)
    Did some work in an industrial environment that ran pretty hot some years ago. Hot enough that the ambient air temp in the general area ran 160f in the winter when the outdoors temp was 10f, the doors were wide open and the building was unheated.

    I agree with the FooHentai about fiber, you want that, don't even bother with cat 5. You also want to work with your plant electrician and draw on their experience dealing with heat.

    You are also going to want some kind of shield to run in front of the enclosure, even if it's just a piece of metal that maintenance installs. That shield will get fairly got, so keep it a good foot or so from your equipment. This will also stop some of the interference that whatever equipment your working will put out.

    Get a proper industrial enclosure to put your equipment in, and expect to pay heavily for it. You can also get ruggedized switches like the Cisco Catalyst 2955S-12 that are designed for extreme heat conditions to begin with. Do as much shielding as you can, it can make a big difference on how effective your equipment works.

    • The Cat 2955 has an operating temperature range of -40F to 140F (-40C to 60C) (I just deployed two of these myself, and am looking in the Hardware manual as I type this)
    • Also, make sure the "peice of metal" you put there is not tin, or aluminum, brass, bronze, copper, or gold.

      2000F (1300C) will easily turn any of those metals to liquid.

      I would recommend a graphite thermal barrier, with some high-temp refractory cement/fireclay. Oh, and you'll need at least a couple of feet thick of this material to keep the radiant heat from melting your equipment.

      Let us know how this turns out, because it's a very interesting problem.

    • Re: (Score:2)

      by Andy Dodd ( 701 )
      "You can also get ruggedized switches like the Cisco Catalyst 2955S-12 that are designed for extreme heat conditions to begin with."

      You can even get that in extra-ruggedized form (well, not quite yet), see the Parvus DuraNet 2955. It's a repackaged Cisco 2955 with MIL-C-38999 connectors, among other nice ruggedization features.

  • Are you the Devil's network guy in hell? (Score:1, Funny)

    by Anonymous Coward
    The only reason that I can see to have network switch equipment that close to 2000 degree heat (units? please?) is that you must be part of the Devil's IT team in Hell and you are in the network group.

    • No! (Score:3, Funny)

      by Dr. Hok ( 702268 )
      The only reason that I can see to have network switch equipment that close to 2000 degree heat (units? please?) is that you must be part of the Devil's IT team in Hell and you are in the network group.

      No way! We all know that there are lakes of molten brimstone in hell, which limits the temperature to its boiling point, which is 444C. I quote from http://www.fifer.net/quotes/ [fifer.net]:

      The temperature of Heaven can be rather accurately computed. Our authority is Isaiah 30:26, "Moreover, the light of the Moon s

      • Re: (Score:3, Funny)

        by lostboy2 ( 194153 )
        We have, then, that Heaven, at 525C is hotter than Hell at 445C.


        Yeah, but it's a dry heat.

  • Small Things (Score:3, Informative)

    by pionzypher ( 886253 ) on Friday January 12, 2007 @08:53AM (#17571888)
    What we've done at my place is to shield the boxes with fiberfrax board on the heat-facing side. Typically on a warm summer day, 10-15 feet from the furnaces it can reach 300 - 400 degrees. Keeping lateral air circulation is key (i.e. a cooler air flow along the side of the furnace that hits the side of the box) helps. Even with all that we end up replacing components at least once a year. Water cooled can help, but here in colorado the winters have frozen numerous water jackets even a few feet from the furnace (2800 degree). In most cases, cooled air brought in to the box via lines from an ac or fan works well enough without worrying about freezing or leaks.

    Really the best solution is to keep all components with the exception of the cable itself in a control room or mcc room where there is some form of climate control.
  • Give it a few years, and perhaps global warming will make this standard operating conditions?
  • Aerogel. It's been covered on /. before. You can read about it here. [nasa.gov]

    There are companies marketing it now. You can buy it in cut sections. It would be perfect for lining the outside of an IDF. One thing though--it will hold in heat equally as well as keep it out. You might have to have an AC unit of some kind.
    • You might have to have an AC unit of some kind.

      Lots of water cooling plus aerogels would seem to be a winning combination. It's going to have to move pretty fast to keep from being turned to steam...
  • A lot of people have mentioned Cisco, but I'm not sure how many of them have actually worked in chemical plants. For industrial plants, you want your switches with all your other 'sensitive' equipment -- in the PLC enclosure or the remote station. You can purchase these enclosures at different NEMA ratings for your industrial application. For switches, we use Phoenix Contact [phoenixcon.com], which are DIN rail devices that go in the PLC panel.

    Industrial control is a whole different ballpark. What works for business offices
  • Why don't you ask the people that installed whatever it is your plugging into the switch, as they obviously had the same problem, if they didn't why are you putting the switch beside the industrial machinery?
  • Use a bit of science! You can make the box a little cooler by reflecting infra-red light away from the box. When we human beings "feel heat" from things like the sun and electric heaters, we are sensing the infra-red electromagnetic "light waves" radiating from the source. Since that is light, even though we humans can't see it, finding a mirror that reflects infra-red well should help keep the box cooler, if it is near the heat source. Of course, you will still need to cool the box since the ambient air te

    • Re: (Score:2)

      by yancey ( 136972 )
      Mind you, the mirror itself will heat up over time, even if only from the ambient air temperature. So if the mirror is hot, it should not be touching the networking enclosure. For good effect, the mirror should probably be placed at least two centimeters away from the enclosure, so as not be touching it in any way. You want air-flow between the two. In addition to reducing heat transfer through radiation, you will also want to reduce heat transfer through conduction and convection as much as possible.
    • Keep in mind, a mirror of any sort (even just a piece of polished aluminum) will only work while it is clean.

  • Easy (Score:1, Interesting)

    by Anonymous Coward
    You were misled.

    No one is asking you to do what you think they are asking you to do.

    Tell us what the industrial equipment is and we'll tell you if it will actually affect network equipment a few feet away.

    For example, I have installed networked data acquisition systems within feet of plasma torches that reach over 10000K, no special considerations with regard to heat were required at all.

    Tell us what the equipment is and we'll tell you if you actually have a problem or not.

    captcha: panties (As in, don't ge
  • You should contact the "project manager" from this job..... http://thedailywtf.com/Articles/A_Secure_and_Well- Ventilated_Location.aspx [thedailywtf.com]
  • Fire bricks are usually rated for 2200F, within your range limit. I like the idea of reflecting back some of that heat, assuming you can find a reflector that won't be affected by the temps. Mounted against a wall of fire brick, and the other side should be well within temp limits for off-the-shelf gear.

    How do I know this? My father owns a home in the Poconos, with a real nice fireplace. The fire bricks keep the house from burning down, even when the fire gets a few chunks of dry wood. Heck, the firepl
  • I'm a mechanical engineer for a glass plant, and we have leveling lasers with microprocessors built into the sides of our furnaces (typically >1500 C) - they have a case with a water jacket built around them, and the water is cycled through a forced-air convective system outside. The lasers are typically kept at about 45 C with no problem. Firms do exist to design such systems.

  • Talk to an engineer (Score:3, Informative)

    by pclminion ( 145572 ) on Friday January 12, 2007 @01:42PM (#17576564)

    As others have said, you really need to talk to an engineer about this. Having said that, remember the three ways that heat can be transmitted: radiation, conduction, and convection. Insulating the box takes care of conductive transfer. To prevent radiative transfer, the box should be wrapped in several layers of reflective material, with air gaps between the layers. Then, to prevent convective transfer, blow cool air through the spaces between the layers of reflective material.

  • Build a huge (very tall -- because it can not be very wide) heat sink in between the heat source and the equipment. Make the heat sink filled with water, and continuously exchange out the heated water with cool water. Add a refrigration system to the water, if the water volume is not enough by itself. Also, try and get a heat sink on the heat producer to direct heat away from the equipment. Ie.) outside. The water in the heat sink should keep the metal that contains it from melting. Water is pretty am
  • How about moving the IDF underground? (or rather, below the heat-generating equipment)
    If you build a small "basement" as a networking closet, you won't need to do all that much to shield it from the heat.
    just a thought...
  • CAN in vehicle. In most cars, at lease 10 sensors, keep feeding data to the ECU. Chips are placed behind firewall (whatever seperates the compartment and engine room), others like wire bundle, fuse and relays reside with the heat source. Due to harsh requirement (4 cyln engine ignite 4 times every ms @ 6000 rpm, oxygen sensor with 300 degree celcius working temperature mount on the catalytic convertor operate @ 900 degree celcius, wire 18 or less awg prefer.
  • If you are really in a 2000 degree uncontrolled environment, you will find all your cables melt. The only thing that might last is the Teflon cables, which are really expensive. Keep in mind any spilled molten steel/aluminum/magnesium/whatever will melt through stuff that the computer industry thinks is indestructible. Also, the industrial guys like moving equipment around. Plan on them rearranging anything that looks like it can be remotely moved.

    My solution: after experimenting with conduit covered wi

Slashdot Top Deals

"Atomic batteries to power, turbines to speed." -- Robin, The Boy Wonder

Close