Thursday, 27 September 2012

Revit Needs to Go Back to School

Here's a simple maths problem. Take a circle of radius 10. This radius could be any unit of measurement;  millimetres,inches,light years,baby steps. To find the area of this circle, you apply the formula:

πr²

Therefore the area of our circle is 3.14159 x (10²) = 314.159

Seems simple enough?

Not according to Revit

Draw a filled region using a circle of radius 10.
Firstly, do this using metric units (mm & mm²) and you get:












Now do this in imperial units (decimal in & in²) and you get:












The metric figure is 99.98% accurate, the imperial a little closer. Big deal you may say, and in most cases you'd be correct, certainly in a small example like this. But what if I take this circle and make it bigger, then I extrude it to make some 3D geometry, then I copy this around many times, then I want to know the volume of these elements?  Small errors start to compound into more significant overall errors

pi is a constant. Revit knows this is a circle. So why the discrepancy?









5 comments:

Steve said...

Looks like it only affects filled region area. Rooms and spaces seem fine. Fwiw, I personally don't rely on or use filled regions for area values that really matter.

Darren Snook said...

Hi Steve, thanks for your thoughts. It's actually affecting volume calculations, which is how I first came across this. I used filled regions here as an example to demonstrate the point. I first thought it might be because circles are essentially faceted, but I've checked that and it's not the case. If you take a cylinder and do an Area*Length calculation and compare it to Revit's volume calculation, the figures are different that in some cases, even rounding doesn't bring out the same figure

Anonymous said...

I personally use Filled Regions in calculations for sloped roof calculations in MEP design due to the fact that the Roof slopes are not recognized Rooms/Spaces. Thank you for the information.

DaveP said...

If you took that circle, made it bigger, and copied it around a bunch of times, it's still going to be 99.98% accurate. Sure, you're going to be off by a larger absolute number, but its still a 0.02% error, and it will always be a 0.02% error, no matter how large your sample size.
On a hundred dollar item, that's two cents. Now say that error happens on a $100,000 house. That's about 20 bucks. I thinks It's safe to say, you'll find a more than a few 20 dollar errors.

Miranda Stahl said...

So is that another glitch of Revit? Or another Revit LT System Requirements flaw?