## Wednesday, October 26, 2011

### Pipe Slope Label Precision

So here’s the scenario, you have submitted plans for  a sewer line (or profile) and a comment comes back that your pipe slope labels are wrong. You go through and check them and find out that you are getting the same results as the plan checker. You think Civil 3D is stupid it can’t even calculate pipe slopes correctly. Well the real problem is located between the chair on the plans on the plan checker side and between your chair and the computer on your end.

It seems somewhere along the way some of us forgot about significant figures. The basic concept of significant figures is that the resulting number can’t be more accurate than the least accurate number used in the calculation.

The problem usually comes up when the agency requires more significant figures in the slope label then in the Grade Break labels. In this above picture if we calculate the slope based on the grade break information:

(PVIE – PVIE)/(PVIS-PVIS) = slope

The best accuracy we can get from the slope would be -0.33%. If you do the math you will find calculating the slope based on the grade breaks is -0.3284746%. This number is obviously different than the slope Civil 3D is showing.

The only way we should be confirming the accuracy of the Civil 3D label is using this formula:

((PVIS-PVIS) x slope) +PVIE = PVIE

This way we are using the available values to ensure our results meet the requirements of the significant figures provided on the plans.

Once completing this explanation of the issue with the plan checker and they don’t agree. The only accurate solution is to increase the precision of the grade break labels. Creating a label to get the slope label to be correct may cause problems later on. Especially if the contractor builds the project off the slopes provided. For long lengths the ending PVIE will be incorrect value.

forest.peterson said...

I am always upset when my accounting is short by 33 ten-thousands of a cent; I once had to explain my harsh feeling to the project business manager that it is not so much the small amount of money missing but the interest over the next hundred years that our corporation will lose, 4.8 cents... 4.8!

The rounding error results in 4 tenths of an inch, almost half an inch over 100', that seems a lot.

What is the field accuracy and precision for installed sewer line over the same distance? I searched and it looks like 4% is the specified accuracy, so your good at 12x the standard for accuracy.

I agree that plans should be exact but for the cost of sending the plans back and your time to figure out wth - all for something less than field accuracy, the accuracy only exists in the model world it is impossible in the real world. Nobody will ever instal a sanitary sewer with exactly a .3251% slope - at least not purposefully. You should backcharge them for your time.

Mike said...

I have a similar issue right now with Civil3D profile view stationing and pipe lengths.

We use 2 decimal places for stations and pipe lengths.

But pipes never have pipe lengths of exactly 2 decimals. It may display as 85.62' but the true distance between the structures could be 85.62132535 or 85.61835235235.

I'll call everything beyond the 2nd decimal the "slack". That slack doesn't show in the pipe length label, but it will show in the stations.

For example, a pipe network with 2 pipes, each having a true length of 100.123'

The pipe lengths will display 100.12’ (because 100.123 rounded to two decimals will round down)

But the stations go like this:

Str 1 ... 10+00.00
Str 2 ... 11+00.12 (rounded down from 11+00.123)
Str 3 ... 12+00.25 (because 100.123+100.123 = 200.246 ~200.25)

And the pipe length summation goes like this:

Str 1 to 2 ... 100.12
Str 2 to 3 ... 100.12 (Sum = 200.24)

----------

I wonder how others address this - nothing I've thought of has worked.

Christopher Fugitt said...

Other than purchasing a elementary math school book and highlighting the chapter on significant figures, I'd use station equations along the alignment at the structures and show them in the profile view. When they ask why they are there tell them why and maybe then they will get it. Especially since some of them will be showing the same values at two digits.

Rafael Rodriguez said...

I also had a problem like this with a reviewer who wanted exact numbers. Then after a little digging around I came across the Rounding parameter in Text Component Editor. You have three choices: Round Normal, Round Up and Truncate. But I found out that 'Normal' didn't act as I expected and 'Up' gave me the expected results most of the time. I still had to edit some of the numbers by hand so that the reviewer would accept them. I was really surprised by the whole ordeal, not what I expect from a product like this.

Anonymous said...

I totally hear you. One of my biggest complaints with reviewers is their lack of understanding significant digits. In your example, everything you are putting in has 5 significant figures, so why shouldn't the slope? Because 642.24-641.92 is 0.32, not 0.32000. Even then, sometimes there is a rounding issue. For example, what numbers are you using in your calculations? Well, C3D has 14 significant figures so you are using numbers that round to the numbers displayed. Based on that fact, your slope will actually vary between 0.318% ([642.235-641.925]/[378.605-476.035]) and 0.339% ([642.245-641.915]/[378.615-476.025]). I suppose you could create an expression for the label that calculates the slope based on the precision you want.