Vertical Parabolic Arc

[I-Sokolov]

Hi Rick

I confused your topic Parabolic Arc

First, why do you use "A = (end gradient - start gradient)/(2 * horizontal length)"?
IFC documentation says that RadiusOfCurvature is "minimum parabola radius" and I think it should define A.

Second, it seems you assume the parabola axis is vertical, but I do not see this fact from IFC doc.
My understanding of the segment definition is the following... dit I miss something?

@peterrdf

[RickBrice]

@I-Sokolov excellent questions.

First, why do you use "A = (end gradient - start gradient)/(2 * horizontal length)"?
IFC documentation says that RadiusOfCurvature is "minimum parabola radius" and I think it should define A.

The RadiusOfCurvature attribute is for IfcAlignmentSegmentVertical. I am defining the geometric representation of the parabola for IfcCurveSegment.ParentCurve = IfcPolynomialCurve. My terms A, B, C are the polynomial coefficients.

Second, it seems you assume the parabola axis is vertical, but I do not see this fact from IFC doc.

In general, the axis of an IfcPolynomialCurve could be in any direction. However, in the context of an alignment, the IfcGradientCurve is defined in a "Distance Along Horizontal", "Elevation" coordinate system. As such, the parabola has a vertical axis in this special case.

The entire section I've written for vertical needs a lot of attention. It should present the semantic definition and provide a mapping to the geometric definition. I haven't written that yet. I only captured the key parts of geometry.

[I-Sokolov]

Yes, the RadiusOfCurvature attribute is for IfcAlignmentSegmentVertical.
But EndGradient, StartGradient, HorizontalLength are also atributes of IfcAlignmentSegmentVertical.

Why do you use EndGradient, StartGradient, HorizontalLength to calculate polynomial coefficient A and not using RadiusOfCurvature?

A = 1 / 2* RadiusOfCurvature.

Am I wrong?

[RickBrice]

StartGradient, EndGradient, and HorizontalLength are required attributes. RadiusOfCurvature is an optional attribute that might not be available.

You are correct that A = 1 / 2 * RadiusOfCurvature where
K = (EndGradient - StartGradient)/HorizontalLength and RadiusOfCurvature = 1 / K

I think the bSI validation service checks the RadiusOfCurvature against the other parameters using this relationship.

[I-Sokolov]

Ok, it can make sense even I expected optional is to avoid it for constant gradient, it misses WR to specify exactly.

Next, why do you think parabola must have vertical axis in (Distance,Elevation)?
Why my picture is invalid? (S-distance, H-elevation)

[I-Sokolov]

[RickBrice]

In my 30+ years of bridge engineering experience in the US, I have never seen parabolic vertical curves defined that way.

With that said, your use of the parabola is entirely possible using the IfcCurveSegment.Placement, SegmentStart, and SegmentLength parameters.

The IfcPolynomialCurve doesn't have parameters that would rotate the axes.

[I-Sokolov]

ok, I accept this is not practical case but would like the clear statement in IFC doc
Thanks for clarifications!