Hi Andy,
(By the way, calling the individual items "panels" can be confusing as a collection of individual solar cells, which you are thinking of buying, wired together make up one panel.)
I answered the voltage question below, but yes, you're getting the idea of "filling the box." If you had to buy them in batches of 20 and 27 would fill your box, the other cells should either be used elsewhere, kept for future projects (my favorite - and I have the cluttered storage areas to prove it ;-), or sold to someone else. If you did put those cells into a 3x9 configuration, yes, you'd have a total of 47.25 watts of production. Keeping with your hypothetical of wanting 13.5 volts for simplicity, you could always use 9 of the extra cells to add another row and end up with a 4x9 configuration. That would up your amps and just leave 4 cells unused.
But, lets go back to the panel output watts to describe one more thing you need to be aware of. Watts out of the panel is not the same as watts into the batteries. Lets go way down to the microscope level for an instance and pretend that we're watching a 16v electron that is going into a battery that is currently at 12v. What happens? That 16v electron finds a "spot" that needs charging (using inelegant terms as the technology, and thus the particular thing being charged, varies from battery type to battery type), and gives that "spot" 12v to charge. Does the original electron end up still having 4 volts? Nope. The 4 extra volts were converted to (boo, hiss) heat. So, the actual power being absorbed by the battery at any instant in time is the number of electrons coming into it, measured by the amperage, times the present voltage of the battery. Instead of the 3x9 configuration (which would only produce 4.5 volts and thus won't charge the batteries at all), lets pretend it is an outrageous 3x90 configuration putting out 45v at 10.5A or 472.5 watts. Connect that to the batteries and what you end up with is 10.5 amps worth of electrons flowing into the batteries and each one charging the battery at only 12v. That means this hypothetical panel would have produced 472.5 watts and yet only 126 watts of power (10.5A / 12v) was being absorbed by the batteries. That is an efficiency that is somewhat less than desirable. If you want to learn more about this, search for "MPPT" on this board or on google and you'll learn about a technology in newer charge controllers that would, in this example, sense that the battery only wanted 12v and would convert the 472.5 watts into 12v at nearly 40A before applying the juice to the batteries.
Craig