Profile Labels for Curb - Rounding Error

[tshzi2012]

Hello.  We sre doing 3 line profiles for a road and the left and right curb should be .13' lower than the centerline elevation.  In many cases we are getting .14' or sometimes more or less.  We tried resampling the cooridor with a tighter interval but still the same results.  Any feedback would be appreciated.

[wfberry]

Last time I checked we are building roads, not watches.  You will get this quite often due to rounding, take it out to 5 places and look at it.

Bill

[Joe-Bouza]

I'm curious, maybe I misunderstand. if your leop and reop are a uniform offset from crown why 3 profiles?

[tshzi2012]

We are actually profiling the curb not the eop.  In California we always profile the curb even if its concentric.

[jmatthei]

We usually have to profile our roadway curbs in Michigan too.  It makes it easier to set and check the curb elevations, and it gives us greater peace of mind that they'll do it right when the road width changes.

Is it just a rounding error?  Do things look right when you take it to 3 decimals?

[Jeff_M]

---

@tshzi2012 wrote:

 In California we always profile the curb even if its concentric.

---

Maybe in your part of California you do, but not in any portion I've worked (Sonoma/Napa/Mendocino counties) 

Using the corridor for the profiles seems odd, though, since you get a PVI at the same frequency as the corridor. In cases where I need an eop or toc at a specific offset & slope, I just Copy the centerline profile and raise/lower by the calculated difference.

[tshzi2012]

Its definitely a rounding error.  for example on one label we have 651.3255, and when you take of .13 off f it you get .1955.  Some cases its rounding to .2, but randomly leaving other spots .19.  We are most likely going to have to spot check each elevation and do an edit label text on the ones that round incorrectly.

[klugb]

Are you sure it's not the way Civil 3D rounds?

Either way it can't be staked and constructed that accurately, but I know we are **** engineers 🙂

We have this argument all the time with pipe slopes & lengths Vs. invert elevations.

Rounding

Specifies how numbers are rounded up or down to the number of decimal places specified in Precision:

• Round Normal: If the last displayable digit is even, and the next digit is exactly 5, the program rounds down. If the last displayable digit is odd, and the next digit is exactly 5, the program rounds up. An input value of 100.12345 with precision of 4 would be displayed as 100.1234.

  If the next digit after the last displayable digit is greater than 5, the program rounds up. If the next digit after the last displayable digit is less than 5, the program rounds down.

[Joe-Bouza]

My point exactly. Now its a typical section note.

[Grimes_MG]

@ Jeff_M

Why are you questioning what he is drafting? If he wants to draft TOC, EOP, curb FL, curb batter, the subbase, mid-point of the curb face, etc, let him. The user is not asking if what he's drafting is okay with the group, what he's come here to ask help with is why AutoCAD is rounding the way it is.

Groups need to do less "my way is better, why are you doing it your way" and more "I'll help you with the glitches in AutoCAD so what you want is displayed/modeled correctly."

Also, I've seen plenty of Caltrans plans and profiles sheet include TOC's in them. In fact, Caltrans often has you provide BOC, TOC, FOC and CL's depending on the complexity of the project.

Finally, in the school I attended, if the last (precision) number ended in a 5 or greater, you rounding up; 4 or less, you rounded down. AutoCAD doesn't follow is rule.

[Jeff_M]

Grimes, first of this thread is 18 months old, why dig it up just to complain?

I really don't care what they need to show, if they need to, do it. I merely commented that not all of California has to do it as they described. If you read carefully what I posted you should see that I was questioning the methods the OP was using to get the data. They said they were getting the profile from the corridor...I suggested copying the FG profile up/down to get the desired profile as this would give them the expected difference in elevations. Using the corridor to generate the profile just adds extra calculations which may introduce rounding errors of their own. Not to mention the fact you get a PVI at every corridor frequency point (which I did mention in my initial response).

As for rounding, since AutoCAD uses a real number out to 15 significant digits, it may appear at times that it isn't rounding correctly based on the numbers you can see. But the rounding is done with the entire number in mind. So if you see 14.435 and expect it to round to 14.44 but it is displayed as 14.43 then you can bet that if you set your precision to a higher number the real number may be 14.4346823 which truly does round to 14.43.