Furnace door open 'x' minutes = 'y' scale
Furnace door open 'x' minutes = 'y' scale
(OP)
I'm writing a procedure on how to handle tray wrecks in a pusher carburizing furnace. History has told me that wrecks will never be completely eliminated, but it is how you handle them that minimizes loss of product in the furnace.
The furnace operates at 1700F with endo+NG atmosphere. Like I said above, when a wreck occurs the endo is replaced with nitrogen, preferably after some cooling. At some point a furnace door will have to be opened to assess the wreck and how easily it can be cleared.
If time was not a factor and I was the production manager I would let the furnace cool to below 1000F and then analyze the wreck. However this is the auto business and I'm just the metallurgist
, so things don't always work the way I like it. This means the wreck sometimes must be assessed AND cleared while hot, and parts are cherry-red.
The rule that I learned from the old-timers years ago was to work at the wreck no more than five minutes, then close the door to allow the N2 to recover. I'm being asked to publish a maximum time that a door can be opened which I'm reluctant to do on just my opinion. Five minutes always worked and even if some decarb formed it was easy to rework.
You may have figured out by now that the door was opened for far longer than five minutes, creating a lot of rework and loss of product due to excessive scale...What I'm looking for is any published evidence of an empirical relationship between duration of air entry and how much decarb or scale forms in the process. The material is 8822 or 8620 steel. Thanks for your help!
The furnace operates at 1700F with endo+NG atmosphere. Like I said above, when a wreck occurs the endo is replaced with nitrogen, preferably after some cooling. At some point a furnace door will have to be opened to assess the wreck and how easily it can be cleared.
If time was not a factor and I was the production manager I would let the furnace cool to below 1000F and then analyze the wreck. However this is the auto business and I'm just the metallurgist
, so things don't always work the way I like it. This means the wreck sometimes must be assessed AND cleared while hot, and parts are cherry-red.The rule that I learned from the old-timers years ago was to work at the wreck no more than five minutes, then close the door to allow the N2 to recover. I'm being asked to publish a maximum time that a door can be opened which I'm reluctant to do on just my opinion. Five minutes always worked and even if some decarb formed it was easy to rework.
You may have figured out by now that the door was opened for far longer than five minutes, creating a lot of rework and loss of product due to excessive scale...What I'm looking for is any published evidence of an empirical relationship between duration of air entry and how much decarb or scale forms in the process. The material is 8822 or 8620 steel. Thanks for your help!





RE: Furnace door open 'x' minutes = 'y' scale
I think, however, you have the best source of information at your disposal; the old-timers. While they may not fully understand what is going on with the steel, they do have a very good idea of what works (or better, what doesn't work). As an old boss of mine once told me, "we've screwed things up just about every way possible", meaning they've done everything the right way, the wrong way, and the "God-knows-how" way and seen the results. Often, this information is much more valuable than all the scientific analysis you can perform.
My suggestion, use the 5 minute limit in your procedure citing the old-timer's experience as your justification. If you later gain emperical evidence (the amount of rework time or scrap, for example) that 5 minute limit is too long (or too short), you can certianly change it. Based on the best information you have, though, it seems to be the best choice.
rp
RE: Furnace door open 'x' minutes = 'y' scale
rmw
RE: Furnace door open 'x' minutes = 'y' scale
But your temps and your atmosphere are completely different from mine, so the issue may not exist.