Reading a Diagram with Multiple Radii

advice
modeling

#1

Hi SketchUp forum!
I am trying to read a diagram of a railroad profile and model its 2D equivalent in SketchUp. I rudimentarily understand how to bevel/radius a single 90deg corner, but my diagram requires a curve with multiple radii and beveling corners that aren’t exactly 90deg. Here’s the diagram I’m working with:


What are the techniques or ways used to go about doing that, and if you have any helpful tips for a total n00b, let me know! Thanks!


Best way to trace profile?
#2

Those you circled I’d probably use a Bezier Curve. I’d probably use it for most of the other curves, too. I would be tracing the drawing but only half since it is symmetrical.


#3

Not an expert on reading drawings by any means, but it looks to me as if the upper radius in your larger red oval is saying it’s 120mm radius, starts vertical, and extends down to where it meets the small 35mm radius whose centre is 19.5mm below the dashed reference line.

If you start from the top of this 120mm radius arc, you know its centre is horizontally 120m ‘left’ of the arc above it. And you know it is half of 16.5mm (i.e 8.25mm) left of the vertical centreline. So that fixes its centre as on a horizontal line 128.25mm left of the centreline. Draw part of that arc from centre, radius and angle (menu Draw/Arcs/Arc in latest two or three versions of SU).

Locate the centre of the 35mm arc from the dimensions given and draw that too, to meet the R120 arc. Erase any overlap.

Most of the rest you can draw starting from the top down, using an arc drawn roughly from known endpoint to endpoint with any small bulge, then changing the Radius in Entity info. Make the next arc tangent (shows cyan marker with default style settings).


#4

Just seen DaveR’s reply. What is the drawing to be used for? Bezier curve won’t preserve exact radii, but does that matter for your end use?


#5

I’d like as close to completely precise as possible. The purpose of this project isn’t engineering-based (heck I was planning on modelling it in Blender until I realized that straight up wouldn’t work), but I really want to strive for as much accuracy as possible.


#6

As you may already be aware, there are two features of SU that make ‘as much accuracy as possible’ a matter of careful compromise.

  1. SU draws arcs as a series of short straight-line segments. You can choose how many segments you want to use. More = more accurate, but still not precisely circular.
  2. SU has limitations on how small you can draw lines without running into problems where it considers vertices ‘too close together’ as essentially the same vertex. So if you want real ‘accuracy’ draw at an enlarged scale, then scale down to real size. See many posts (often by DaveR himself) on the way to do this.

#7

Sorry, no more tonight - 1:15am in UK here, so I’m off to bed. Maybe more tomorrow…


#8

The drawing is already done to scale. So I’d just import the image into SU and trace over the image with the line tool, using as few segments as needed.

As DaveR noted, only half needs to be drawn as it’s symmetrical. Select the half > Move + Ctrl to copy > -1 Scale of the copy to mirror.

Don’t even really need to be concerned about the overall size in the beginning. As after a profile face is created, it can be resized with the Tape Measure tool to the real size.


#9

Sorry, I was reading on my phone and couldn’t see the image that well. Since radii are given for most of the curves, I would use the 2-point Arc tool for most of that. Importing the image, still and at least using it as a guide. Tracing it might not be that effective because the image is already scaled unequally in each direction. It can be fixed reasonably well but I expect there’s enough dimensions to be able to draw the rail profile without tracing.

Like this:


#10

Benjamin;
The uic60 is a standard spec so you should find cad drawing of the profile on the net. Since you do not have the pro version it does not have the dxf import capability but those my give you a little more clarity. The drawing pic you show appears to have all the dimensions needed to lay the profile out, These give you the angle of the arc, the radius and location of the vertex. By definition at each point on a true arc the radius and curve are tangent at their intersection…
Unfortunately the native SU program does not have any true arcs because by its very nature curves are approximated by straight lines or I guess you could say cords. However, there is a plugin called true tangent you can use that should help you.
In the mean time you basic question on how to read drawing can be answered by showing you by just one example.
If correct let use know, may not be quick answer since many maybe busy with holiday.


#11

Here is case for the left side bottom, all the info is in dimensions given except some cases for the center point of the radius. Depending on what the ultimate use you have planned Su may / not be acceptable because of the errors caused by approximations used. If you measure the radius of the example there are some sub mm errors , I did not scale up to see what improvement can be done. The challenge for you is to find the location of the arc center, some of the older compass and triangle user will know how to do that, late now will show tomorrow.rail profile_2016.skp (204.3 KB)


#12

One thing you may not be aware of is that you can enter slopes when using the Protractor Tool. The drawing gives you 2 slopes and their heights. Using these, you should be able to get to this image:


#13

I’d be interested in seeing how you would do the construction of these multiple-radius arc using native. It’s always seemed harder than necessary to me - as if I’m missing something obvious.


#14

Jim, it is a little tedious but not nearly as bad with the Arc tool as it would be with the 2-Point Arc tool. I used a lot of guidelines and guide points.

I should have mentioned the part about placing the guidelines using the slope instead of an angle.

FWIW, I didn’t trace the drawing at all. I did scale it to fit the rail cross section after drawing it to the dimensions shown on the drawing.


#15

For the case you have shown the arc spec. is R= 40mm and one has to find where the arc center point is given the requirement it must be perpendicular, at tangent point, to both slopes at the same time. IMHO the arc you have shown is not the correct solution. In addition the 31.5 mm is just a Y ref. there are three other arcs per. spec. you must consider.


#16

I have been playing with the 2-point arc tool quite a lot today. The 2-point arc is able to create the 40mm radius arc without the use of any guides. Just draw the arc any size as long as you get the magenta tangent inference, then edit the radius in the Entity Info dialog. There is also another way without using Entity Info, but it’s trickier.

I did not show any arcs so I am not sure what you mean? I only showed the guides for the 40mm radius arc, and none of the compund radius arcs. I was hoping Dave would do a demo.


#17

Sorry , poor explanation. Along the slopes on the part there 4 arcs in total. Three of them use the old technique of showing when the some are so long they could go off paper ( ie the jagged lines). My post shows the model and the first arc of 40 mm and where the correct tangent points are.
BTM Trimble has included by accident I assume since it is not mentioned in the release notes a very easy way to get the correct tangent points.
It is neat I hope they do not remove since their history shows regression in very important areas they do even mentioned.


#18

it doesn’t seem correct to me…

if you offset your two guidelines each by 40mm you get the correct intersect point…

john


#19

yes =, that is the normal way one would draw on drafting table. Now that you have that part you have to do the same for the 120 R, 35R AND 7R and tie those into the rail web min thickness of 16.5 mm. I am using the first pic the op posted and have some small errors that do not make sense. GrabCad has the cad file for that profile on their site.
I need to do more checks.
Thnks for the help


#20

problem is it only gives the height for the 120r and the tangent length for the 7r so you can only guess how the 35r fits between…

if you had the sweep angles it would be easy…

john