I agree with Dan, i.e. It's better to have too much canvas than not enough.... especially when each test costs a dollar or two.
Something else to consider... I was shocked to discover that canvas shrinks! My first 24 x 20" canvas print was almost 3/8" short on the 20" side. I'm new to canvas printing, but I"ve been printing on paper for decades. It never ocurred to me that canvas would be any different. Certainly I had screwed it up, so I tested the same file with paper and that print was accurate to a gnat's eyelash. Coincidentally, when I logged-on to this forum to ask WTF!!!, someone else with the same experience had already posted the question, and there were several good answers.
The best solution, for me, was to adjust the Paper Feed Adjustment to the plus side. My first test was at +30, and the difference was barely measurable. The second test was at +70 (maximum), and that got me almost 1/16" over. I backed it off to +65 which put me on the button. I then made three identical 24 x 20 prints.
Since then, I've discovered that different canvases shrink at different rates... varying by manufacturer and by different canvases made by the same manufacturer. It appears to me that, in general, the higher the cotton content, the more it shrinks... and it seems reasonable to assume that shrinkage increases in proportion to the weight of the canvas. And, it also stands to reason that shrinkage is a percentage of the measurement, so the longer the print, the more you'll see it. (I hate to even think this... but... there probably is some correlation between ink density being laid down and shrinkage, too, but to try and figure that would drive me almost sane. Wouldn't it be nice if Epson's software developers would include automatic compensation for all those factors?)