AI program to determine best cut?

I have some aluminum and I have a design for a machine.
Is there a program that I can input what I have (different length aluminum beams) vs what I want (lengths of aluminum for machine) and it would show me the best cuts to make without waste?

I know that this sort of problem has been solved by computer in many different applications (optimizing value of lumber extracted from logs, optimizing cuts of 2-D material, etc.), but I am not aware of an off-the-shelf program for doing this. Can you provide more details of the problem?

-Predictor
http://will.dwinnell.com

I'm thinking about using a GA program to help solve a common problem.
However, I'm not sure the problem can be a proper equation for GA.
Any hints would be appreciated!

Problem:
You have a list of different size pipes you need to fulfill an order.
You also have a second list of pipes that you have on-hand.
The problem is how to cut the pipes on-hand so that you maximize the lengths of
the ones left, and still fulfill your order.

Example:
Say you have 3 pipes of 20 inches each, on-hand.
Your list says you need 2 pipes of 10 inches to fill an order.
You could just cut 10 inches off of two of the pipes on-hand and fulfill your
order. However, it is best to cut one pipe in two pieces so as to have 2 left
that are still long. Note: Either way your going to have the same number of
inches left over.

Equation Time:
Try as I might I just don't see the equation, or the goal, for GA here.
I can represent the pipes as the length in inches for each one.
And then try and determine how many in the "Need" list will fit into each of
the "Have" list.
But it gets confusing because the goal is to always end up with the longest left
over pipes you can have (that way the next order is easier to fill).
Before I reinvent the wheel, has any done anything like this? Got code?

P.S. I'm going to try to use Brint Montgomery's VBGALib (Visual Basic Genetic
Algorithm Libray) :
http://home.snu.edu/~brint.fs/vbgalib/

The problem you give is this:

Initial Situation: [20 20 20]
To Be Provided: [10 10]

Solution 1 Final Situation: [20 10 10]
Solution 2 Final Situation: [20 20]

Optimizers need to know what to optimize. We know that we would prefer Solution 2 in this case. A place to start is the average length of the remaining pipe:

Solution 1 Final Situation: [20 10 10] mean length = 13.3
Solution 2 Final Situation: [20 20] mean length = 20

This evaluation function works well for this example, though I imagine that it will get more complicated with more pieces of pipe. I wonder, too, whether there is a heuristic method of solving this problem. I would think someone has thought of this before now, perhaps in operations research.

-Predictor
http://will.dwinnell.com