Reasons Not To Max Out Roofspace.

Ok, let's say you got yourself a weird roof.  There's nothing normal about it.  It's not a square.  It's some kind of peculiar shape.  There is a way to figure out how many panels are going to fit up there, but it's not going to be as simple as getting the square area and fitting in a number of panels that match that area.
Here is the way that I've come up with that may help.

First of all, panels all have dimensions.  Also, panels require mounting systems.  Ultimately, the space on your roof has to facilitate the rails.

Start with the longest vertical measurement.  That's a good place to start, right?

Get yourself nice and comfortable along the ridge of the roof and run your tape measure down to the gutter.  Don't measure exactly from the very top to the very bottom!  In fact, if your roof overhangs beyond the structure of the house at the bottom, don't consider mounting onto that part.

Next, draw a series of horizontal lines, perhaps with a crayon onto the shingles. (the lines will wash away in the rain if you make them lightly).  Use the dimensions of the panels you intend to use.  Here we see that from the top to the bottom, there's enough room for 3 rows of portrait-oriented panels, without enough room for 2 rows to fit at the narrower part on the left.  However, if the panels are oriented landscape, three rows will fit.

Don't let someone sell you this.  It's too big for the roof.  Note the "precise" distance from gutter to the peak.  This generally means one of two things.  You're either going to see the backs of the top panels from the other side of your house, or the panels at the bottom are going to send rainfall right over your gutter.  Possibly both.  I know it looks 'great' (yeah right).  But in reality, it's impractical to fit that many panels on a roof.

Here is a way more practical approach.  There are only 6 fewer panels in this example (the one above is the 'maxed out' version).  But in this version, first of all the array will definitely fit on your roof.  And secondly, 32 panels will fit onto a single inverter as 4 strings of 8, which is great!

What would the purpose of those additional 6 panels actually equate to?  In all honesty, nothing but a bunch of issues.  First of all, 38 panels would require you to break your array into 2 separate inverters, because 38 is not divisible by a number of panels that will series into a string that is compatible with an inverter.  You'd end up with 2 strings of 10 and 2 strings of 9, probably, each on 4000w inverters.  Now you have twice as many inverters, and only 6 more panels.

I guess that's ok, there's nothing exactly wrong with 2 inverters.  But additional costs are not just that you're buying twice as many inverters (rather than a single 7kw inverter).  You also need an additional subpanel.  It might require a service upgrade.  Those 6 panels, at STC 1.2kw, are really going to only generate 900w at max power, 1.4 mW a year in an area like Connecticut.

Hmm, that's starting to sound good again.  I guess these are the questions we're forced to ask ourselves before we enter a situation like getting a PV installation though... right?