Intelligent Kiln Project: Genesis
July 29, 2008 by cynthia
So what if your kiln was smart? (Gosh, this sounds like a GE commercial)
But… what if your kiln really was smart? What if it sensed things happening during a firing, analyzed and decided on a course of action, then responded appropriately?
Would that make a difference in the way you fire glass? The time it takes? The way you construct molds or assemble the glass? In your firing success rates? The amount of glass you could fire at one time? Would it allow you to discover (and potentially auto-correct) a problem that occurs during firing? Most important, would it make a difference in what you can fire?
Been thinking about this for a few years, since my last trip to the Corning Museum with its incredible collection of ancient glasswork. With no electronic kilns or compatible glass or the Corning schedules or any of that stuff, these guys produced work remarkably like modern glasscraft. And while we don’t know much about their failure rates, the fact that so many glass artifacts have survived some pretty harsh conditions for thousands of years must say SOMEthing about ancient annealing schedules, right?
Kinda humbling, ain’t it? But it got me to wondering if, as glassworkers, we’ve really come all that far in the last 3,000 years…
In reproducibility, yes–the addition of “tested compatible” glass and electronic kilns dramatically improved success rates and enabled significant improvements in production cycles. Glass manufacturers’ R&D has worked miracles in making glasscrafting broadly accessible–a revolution in itself, if you think about the history of glass and its trade secrets.
Yet when it comes to the most critical part of kilncasting–the actual firing–are we really that much farther along than ancient Mesopotamians?
The typical warmglass* kiln isn’t as smart as your average microwave oven; it’s a glorified space heater with a timer. These kilns aren’t all that accurate (heck, I measured as much as a 50-degree differential from one end of mine to the other). A single stationary thermocouple with relatively low sampling rates tends to be the only feedback available, and the response to that feedback is largely binary: If the temperature is lower than the program requires, turn on the heat. If it’s at or above that temperature, turn it off.
In that respect, ancient glassmakers who left a slave watching the kiln all night were probably more effective. So why can’t we use modern slaves, i.e., computers, in the same way?
The commercial glass industry does this in spades, for gigantic telescope mirrors, vitrification of radioactive waste, etc. It’s less likely in the glass art world for a number of reasons (I think):
- Since part of the value of glass art is its “handmade” nature, there are fewer automation opportunities to begin with.
- Most glass artists don’t worry about this stuff unless the glass explodes.
- With some notable exceptions, most glass artists aren’t all that into non-glass gadgets. And even when they are, like most people they tend to buy a solution, not invent one.
- The level of investment required to create a thinking kiln hasn’t offered enough ROI for glass art, especially given typical undercapitalization (i.e., it costs too much).
Kiln casting, OTOH, is a different beast with more variables to boost failure rates:
- The glass in a casting mold stays hotter for longer, increasing the likelihood that its properties may alter.
- Cast objects are typically more 3D to begin with, i.e., the glass “thickness” may happen in any of 3 dimensions, depending on where you look.
- The mold must change its physical properties significantly during firing, i.e., stay strong on the way up and while the glass is flowing, but soften to the point of crumbling on the way down.
- The glass can react chemically (as well as mechanically and thermally) with the investment.
- Air and water entrapment can have significantly greater consequences.
- You generally invest more time, money and heart in a 6-inch casting than in a sushi plate.
Personally, I think about 75 percent of casting glass is resolving and eliminating a bewildering variety of variables so you can get back to the art. And here’s my question: If I add smarts to the kiln, i.e., automate the monitoring and response of kiln events so that I don’t have to (a) do it myself and (b) ‘way overengineer the firing schedule to compensate, will that allow me to get back to the art that much faster?
I suspect that’s the case. And the kiln casting world is standardizing a bit, paying a lot more attention to multipoint annealing (if you haven’t read Bullseye’s nicely written explanation of how to do this in a standard kiln, please do), and how (or if) glass behaves differently in casting situations than fusing.
My kilnless state makes this a good time to test this idea, so I’m talking to folks, and hooking up with a kilngod, establishing some ground rules and we’re getting ready to build me a kiln. I think this is going to be fun.
Next: Designing the kiln
———————
*now, now…no teeth-gnashing, anybody 
the reason that old glass has lasted until this time may not be because of special annealing, but, like most arts (music, paint, oral, written, etc), all the crappy stuff fell part and disappeared years ago, residing under an old roman aqueduct.
No argument there. But the difficulty of some of the work that’s survived makes me think that somebody also patiently worked out a process that included anneal soaks and slow cooldowns. The information we have says that glass was more highly prized than gold in those days–stands to reason that there would have been a strong drive to improve processes, reduce failure rates and make increasingly difficult (and profitable) work.
I’m having trouble understanding where you’re going with this. What kiln events are you thinking of monitoring? How will you discover or measure them?
Well, the idea was to create a kind of lab-in-a-kiln that would let me test different ideas, such as moisture level in a casting mold, when the glass has completely filled the mold (or if all the glass is down in the mold and it needs more), etc. Primarily, though, I was interested in eventually having a kiln that could respond to conditions inside the mold by initiating the next firing segment, shutting down, or calling for help as needed.
The ultimate goal, really, was to see if increasing kiln precision in casting, especially large castings, could reduce both the time it needed to be in the kiln and the overall failure rate.
In any case, ran into all kinds of difficulties just getting the basic kiln together, finally decided I couldn’t wait any longer (I’ve been without a working kiln since 7/08), and got a commercial kiln I can dink around with and add some of this stuff at a later time.