Why does this sweep twist?

[samu3]

Hi All,

Can someone please take a look at the attached file. It was created using Inventor 2013.

Looking down from "Front" view, the end where the sweep starts is as it should be: the sidewalls of the rectangular sweep are straight up and down, with zero degree angle from z-axis.

The other end of the sweep though, the sidewalls have twisted slightly and have a 0.53 degree angle to z-axis, but I need them to be straight up and down.

What is causing this and how do I fix it?

Thanks in advance,

Samu

[spackle42]

Am I correct that this sweep should be purely thru the X-Y plane?  Just from a cursory inspection, the end curves in the Z axis.  I would suggest making this as a sweep using a 2D sketch to plot the path rather than 3D sketches as you can ensure the sweep stays in the given plane.

[samu3]

Thanks for your reply! That's the thing: there are elevation changes along the path - it is not parallel to the XY-plane... So I do need that 3D sketch in this case. Any other suggestions?

Samu

[JDMather]

---

@Anonymous wrote:

... there are elevation changes along the path .

Samu

---

I found several things "wrong", but before I suggest a better way -

what defines the change in elevation?  Is it a constant slope?
A helical slope? A variable slope.  Do you have any defining equation for the change in elevation?

[JDMather]

I read one of your other threads and see that you model mine tunnels.

Well, I'll come back tomorrow and suggest a completely different technique than you are using, but you will have to follow some very precise step-by-step instructions to adapt my technique to your needs.

[ic198]

Your path is made up of 3 3D sketches, each of which are made up of lots of smaller lines, which makes suggesting a fix for this much harder. How did you generate your path curves?

Normally is this case I would suggest changing the sweep type to path with guide surface, or use a loft with a sketch at both ends- however I've just tried both options and I'm struggling with those path curves

[samu3]

Thank you for your replies.

Yes, I too have tried to use a sweep with guides and loft, unfortunately to no avail.

JD, you are correct, I do work with tunnels. The file I sent is a simplified version of what I am working on, and it displays the same twisting as the original model. I have received the path file (dwg) from an outside affiliate, and there the path is one long polyline. The path does not follow an equation, but it is not arbitrary either in the sense that it is dictated by the surrounding rock, its quality and also what is above on the surface.

Currently my workflow is that I break the polyline in AutoCAD based on tunnel profiles - I use sweeps for segments with a continuous profile, and loft for segments where the profile changes linearly.

I place each polyline segment on its own layer in Acad, and then I bring those layers into their own 3D sketches in Inventor. This allows me to turn sketches containing the 3D path lines on and off for sweeping a particular profile from start to end. In my example file the 3 3D sketches are really for 2 different tunnel profiles, and the sketch in the middle is the linearly changing segment.

I realize that the root of the issue lies probably in the path lines, and have considered just bringing the vertex points from Acad and then connecting them with a spline in Inventor. Anyhow, some way or the other I need to solve the problem so that walls of the oprofile are straight up and down all the way through the whole tunnel.

JD, I am really looking forward to learning your technique! Thank you!

Samu

[JDMather]

Step 1

Drag the End of Part marker to just below the 3 3D sketches.

Right mouse button and select Delete everything below EOP.

Now drag the EOP all the way to the top of the browser hiding the 3 3D sketches.

[JDMather]

Start a new 2D sketch on the XY plane and create the 3 points shown dimensioned from the origin.

Attach the file here.

[samu3]

Ok, here we go! Thanks JD!

Samu

[JDMather]

Right click on the 2D sketch you just completed and turn off Dimension Visibility.

Click on the glass cube in upper right corner to go to the Top view.

Start a new 2D sketch on the XZ Plane.

Project Geometry the 3 sketch points you just created.

Create three construction lines from the three projected points as shown.

Attach the file here.

[samu3]

Done!

Samu

[JDMather]

Your sketched lines on the XZ plane are underconstrained - add vertical constraint to each construction line.

[JDMather]

Drag the EOP back up to just under the XY Plane sketch.

Orient in Front view.

Sketch a line from the right side point to the middle point.

Sketch an arc from the middle point to the left-most point (ignore the projected origin point) (you can do this on one smooth movement by selecting line, the right point, middle point then hold down the mouse on the middle and drag arc out to the left side point).

Exit sketch.

Drag the EOP below the XZ sketch and edit the sketch.

Sketch a 3-point arc selecting the right side line end, the left side line endpoint and then the middle line endpoint.

Exit sketch and save.

[JDMather · Accepted solution]

Rotate the view around to an isometric view by clicking on any of the corners of the glass cube.

Start a new 3D sketch and select the Intersection Curve tool and then select the 2 2D sketches.

Right click on each of the 2D sketches in the browser and turn off Visibility for the 2D sketches.

Drag the EOP down to show your original 3D sketches.
Note that we fairly closely recreated your original 3D sketches with only 3 points (the first 2D sketch defined the X,Y position and the second 2D sketch defined the Z position).  The key to using 2D sketches is that it is trivially easy to edit the point positions by editing the dimensions.  We could more closely represent your original 3D path by adding another point or two to the first 2D sketch, but don't do too many - more is not better.

After creating your Sweep profile use the Sweep with Guide surface to eliminate the twist (select the appropriate Origin Plane to keep the profile parallel to that plane along the sweep path).  Attach your file here if you can't figure it out.

[JDMather]

....another tip.

On your profile dimensions, multiply them by 100 to make the Sweep result easier to visualize.  Once you get it solved then remove the multiplication factor.  (and use equal (=) constraints rather than duplicating dimensions.

[samu3]

Brilliant JD! I really appreciate you taking the time to go over this, step by step.

Not only did I learn how to fix the twist in the sweep, I learned how to better control the 3D sketch using 2D sketches, how to simplify the geometry and the utilization of Intersection Curve. I got many new ideas that I can apply in the future. Thanks so much!

It is because of helpful people like you, JD, that this forum is such a fantastic place for learning 🙂

Samu

[ic198]

Hi JD,

I've made 2 solids, one swept along the path only, the other along the path with the XZ plane as guide, I think as in your instructions. I've split these at an arbitary point on the path and sketched the cross sections. The one using path only has maintained constant cross section (though it is twisted, as in the original question) but the other has changed dimensions. Am I doing something wrong- using the wrong plane as guide?

I often come across problems like this which is why I've been interested to follow your method.

Thanks

Ian

[samu3]

Hi Ian,

See the attached file - using the XY-plane as the guide surface fixes the twisting problem.

I don't really follow your workflow though. Why Sweeping and then splitting? In my case, I had the 3 original 3D sketches for different profiles, so I could make the 3D sketches I don't need for a particular sweep invisible.

The profiles are placed on 2D sketches on workplanes that a are normal to curve at that particular path end point.

Meaning, one feature (sweep or loft) per path (in my case, one of three). Splitting the path, not the feature.

Samu

[JDMather]

---

@Anonymous wrote:

Hi Ian,

 Why Sweeping and then splitting? .

Samu

---

I think he is demonstrating the cross-section profile did not remain a square profile.  I will have to do some experimenting, sweeping a line as surface and then Thicken might be a solution.  I'll be back in a while.