|
Drooling: and Other Phenomena that Know no Equation
Our printheads are doing very strange things. We don't know why, we don't know how to mitigate it, we don't know how to solve it. Basically, we know nothing except that the problem began two weeks ago and that the consequences are dire. Welcome to proto manufacturing, where anything can and will go wrong with no advance warning and no clear cause in sight.
In an earlier post, I discussed the difficulty of executing a design in the incredibly complex physical world. The context then was around readying a set of photographs for exhibit, but I think the same observations apply to inkjet manufacturing. When building a printhead, there are so many places where accidents can happen that it's impossible to design them out from the get-go. Think of the intractability of quantitatively predicting the dynamic interactions between electronics, silicon, ink, plastics, and adhesives. Or all the unknowns involved with outside suppliers. To avoid sitting on a design for 20 years, your best bet is just to propose something based on past experience, start making it, and wait for things to go wrong.
This build-break-analyze-fix cycle brings a product to market much faster than, say, the analyze-analyze-analyze-build approach. The more you fail, the more you learn. That’s the key to successful manufacturing: to fall flat on your face early and often, learning from each mistake. Design and manufacturing methods begin with educated guesses derived from previous programs and require years of trial and error adjustment to arrive at something that can be relied on. Once stable, the manufacturing line becomes a recipe-following, high-volume, low-cost affair.
It's this final stage of the factory life cycle that most people think of when someone says "factory." You imagine a hall full of drones operating by rote, machines humming, with lots of identical product flowing off conveyor belts. Seems boring, doesn't it? Well, it is, and it's exactly what goes on in "old technology" plants like refineries or paper mills. Luckily, we never see that side of things here. The three years between when we build a new printhead design and when we reach stability are, quite frankly, managed chaos. Each new product is a bundle of mysteries. Once we exhaust those puzzles, we move things offshore and move on to the next product.
Things can be pretty rough in the first year or so of a factory's life. There are few automated tools, the bulk of the work being manual. For example, our solder traces are currently done using a stencil and solder squeegee. Yikes. More than 50% of the time is "thrash," wherein the factory is involved in a problem-solving loop. And generally, everything is changing as we learn from our mistakes: the design, the performance specs, the manufacturing tools and processes.
In a way, it's sad, but we don't use the Navier-Stokes equations, Vapor-Liquid equilibrium predictions, or block-copolymer self-assembly models to assist in this task of design and process refinement. It'd be nice for personal edification, but it's too inexact and inefficient when it comes to making the right decisions. A PhD physicist can predict that ink will flow correctly if you build to a certain specification, but if the cartridge doesn't print when you test it, you're back to square one.
This is a far cry from what engineers are taught in college. Our homework assignments and class projects pivoted about the mighty equation: deterministic rules derived from the first principles of physics. As stated in the unwritten Bible of the Academy, in the beginning God made the laws of fluid mechanics and thermodynamics; then, on the second day, reality was made in their image and likeness. Consequently, a good engineering student lives and dies by these arcane formulae, viewing everything through the filter of symbolic arithmetic.
Years ago, I used to fancy myself an engineering theorist. I studied long and hard, learning applications of group theory to chemistry and how to solve strangely entangled chaotic PDE’s. When I graduated from UT, I went to a very fine graduate program to apprentice under the masters of Chemical Engineering Theory. It was a decision predicated on the notion that “real-world” problems were too simple, that the most challenging ones lay in the cerebral domain of the mind.
But eventually, it occurred to me that I was viewing the theory-reality dichotomy upside-down. The academic problems merely appeared more complex because they were trivial enough to be adequately represented in the tedious window-dressing of mathematics, while industrial solutions were so complex as to be approachable only via straightforward empiricism. As I eventually realized, the nuances of reality were much more beautiful and thought provoking than those found in the most sophisticated think tanks. The cooked-up academic research suddenly appeared phony and inanely simplistic in comparison to what happens in the realm outside of man’s control. And, I began to doubt the Genesis chapter of academic Scripture. It seemed to make more sense that reality came first, followed by the equations that described them.
I don’t mean to completely dismiss the value of theory. It has its place in an R&D program, but more in the sense of logic games, not traditional mathematical modeling. It's a theory of troubleshooting, not of universal understanding; of experimental correlation, not first-principles deduction.
It has taken me a long time to realize that the value of an engineering education is not in what you learn, but in how you learn. Although a student takes classes in a specific aspect of physical reality - chemical, electrical, mechanical, or biological - these topics are merely the means for instilling analytical skills that are more universally applicable. The subject matter is the medium, the implicit problem solving lesson the message.
So while equations don't have much use per se, the study of them does, each one acting as a virtual thought experiment capable of breeding intellectual clarity and honesty. Mathematics is the perfect language for such inculcation, avoiding the vagaries of the English language and revealing specious arguments or stupid conjecture ruthlessly. As masters of the equation, engineering students develop an intuition for analytical exactness, for cleanly separating variables and learning how to make sense of complex problems.
And that's the most highly prized trait of an engineer. It's been two years since I last invoked any chemical engineering design equations, but I rely on my analytical instincts daily. And I think that loss of specificity is a good thing because I'd quickly tire of re-visiting the same old problems ad nauseum. Distillation. Reactor Design. Heat Load Analysis. Nice thought experiments, all of them, but not interesting careers. While the chaotic environment of tech manufacturing often leads to a certain kind of insanity, it's certainly more interesting than polishing the Oil Refinery for the next forty years. And I'd rather be slightly nuts than thoroughly bored.
Occasionally in my current incarnation as a manufacturing development engineer, there are moments when I miss the late nights hunched over strange theoretical scribblings and caffeinated discussions on phase equilibrium in the graduate student lounge. But the thing about theory is that you can do it anytime, anywhere. So while I’ve long departed the ivied walls, it hasn’t precluded my playing with the odd equation when the mood strikes. A few times a month, I’ll check out from the factory floor, open a textbook, pull out a green computation pad, load the Pentel P205 with lead, and stroll the grounds of the beautiful and simplistic for a few hours.
I imagine a perfect gaseous sphere gradually growing within a fluidic continuum, derive the pressure profile, and calculate the stress tensor inside and outside the bubble. When done, I neatly draw a box around the equations in thick, dark lead and stand back in admiration. It’s a nice break, but gradually the call of reality beckons. I close the books, take out the garbage, and go into work the next day to solve the infinitely more complex problem of keeping our ink from spilling on the floor.
|
