So I'm participating in a research project.

Here's the official title: Scheduling of Jobs with Sequence-Dependent Setups on Unrelated Parallel Machines.

Sounds pretty cool, doesn't it? Let's break it down.

"Scheduling of Jobs" is pretty straightforward. Imagine that you have several machines available to do jobs that come in periodically. The machines have different capabilities, which mean that one machine may be able to complete a certain job in less time than another machine.

"Sequence-Dependent Setups." This is one of the things that makes this original research. There are certainly set-up times associated with jobs. But "Sequence-Dependent" means that the setup time on a machine will be different depending on the previous job that machine processed. So if Machine1 just processed Job1, and was going to process Job2 next, there would be a setup time of, say, 3 minutes. But if Machine1 had just processed Job3, the setup time would be 5 minutes.

"Unrelated" means they're not identical, they have different capabilities. One machine may be able to process a certain job in 10 minutes, another machine may take 25 to do the same job, and another machine may not physically have the dimensions to handle the job.

All "Parallel" means is that the machines can process jobs at the same time, as opposed to a series relationship which would require that a job be done at one machine before another.

Oh yeah, another thing. Each job has a weight and a due date. If the weight is 3, it's very important. If the weight is 1, it's not a very high priority. The due date denotes when the job is supposed to be done.

My partner and I, with guidance from my professor, need to devise a solution that minimizes the total weighted tardiness. For instance, if a job was completed 3 days late and it had a weight of 2, it's total weighted tardiness would be 3 * 2 = 6.

Trust me, it's very fun stuff.

I'm a Mennonite, and so I have this environmental influence toward the practical. But I lean the other way. I love the theoretical and the analytical, the idea that we could model stuff in the real world with math, for instance.

Do you see what I mean when I say that this is a very high-level way of looking at a problem? It's far removed from actually getting down on the plant floor, getting greasy, setting the milling machine and running the aluminum through it. Do you even run aluminum through milling machines? I assume you can, though I don't have near enough knowledge or experience with milling machines to know.

That's the type of thing you'd study as a Manufacturing Engineer, which, though similar to Industrial Engineering, has a little more of a low-level feel to it.

Tom, is this elitest to you? It might seem sort of that way, because it's a very analytical way to solve a problem. The risk you run is not connecting with the people that are actually doing the work on the floor. They know so much more about the processes themselves. Operations Research people (or IEs) just take the data and form models that hopefully will aid the floor by giving them guidance in their decision-making.

This problem was influenced by a real-world situation at NASA, by the way.