Solid sweep with guide rail

[barna.daniel]

Hello, 

I got stuck when Inventor did not have yet a solid sweep. I was happy to discover it now. However, when the toolbody is a solid and not a sketch, there is not "Guide rail" option, only "Aligned", the meaning of which is not completely clear to me. 

I would like to model the effect of a mill (i.e. toolbody is a solid cylinder) moving in 3D in a complicated way. The cylinder would move along a sweep path (calculated in a C++ code and imported via Excel), and its axis would be oriented towards a guide rail (also calculated in a C++ code and imported). The other constraint that is required to make this problem fully defined is that the axis of the cylinder is everywhere perpendicular to the tangential of the sweep path. Is there a way to model this?

Naively, a rectangle (cross section) could also be swept, giving me the possibility to use the guide rail. But if the cylinder's axis is rotating around the sweep path (which it does in my case), one can realize that this does not result in the same geometry.

Btw, what is under the hood when sweeping a 2D cross section along a path, using a guide rail? Is the cross section's plane everywhere taken to be perpendicular to the sweep path's tangential, and the guide rail is used to define the twist of the profile around the sweep path? (I would need this to approximate my true geometry - I could live with this). However, I had the impression that this is not the case, a "progress velocity" along the guide rail (different from what would result from the perpendicular constraint) tilts the cross section's plane from what I want, and this causes problems in my case. 

Thank you

Daniel 

[JDMather]

@barna.daniel 

Can you Attach an example *.ipt file (as far along as possible)?

[johnsonshiue]

Hi Daniel,

Based on your description, I suspect the exact geometry cannot be created in Inventor Solid Sweep yet. However, there should be workflows allowing the result to be very close to the exact solution. Please share an example. The forum experts and I can help take a look to see if it is doable.

Many thanks!

[barna.daniel]

Hello,

Thanks for your replies. Before I prepare a clean file (will take time, now have deadlines for several things), let me describe my case in more detail, and what has led to choosing the particular solution I am using now (which is not a solid body sweep yet). 

We design a mandrel for a high-temperature superconductor magnet. The conductor is a tape which has very strict requirements on how it can be bent. The tape is wound into grooves which are machined into a cylinder. The winding path is described by a 3D curve r(t), where t is some parameter along the curve. The Frenet-Serret frame is the curve is defined as follows. T is the tangential unit vector defined as T=r'(t)/|r'(t)|. The normal unit vector is defined as N=T'/|T'|, and the binormal vector as B = T x N. There are several tapes wound into the groove on top of each other, and the groove must be oriented along the N vector, to simplify the case (to be more precise: we have a groove direction vector S which lies in the plane spanned by B and N, i.e. perpendicular to T, but this does not matter for the description of the problem). All of these vectors T,N,B,S are changing along the path, i.e. dependent on t, but T,N,B form an orthonormal set. 

My first approach was to export discrete points along two paths from my C++ code, and import it into Inventor: the top center path of the groove, and the bottom center path. The top center path is the reference path, the bottom center path is calculated as r(t)+D*N(t) where D is the groove's depth. So by construction, the distance between the two paths is always D, *along the N direction* (which is perpendicular to the tangential of the reference path). I swept a rectangular cross section along the reference path, using the bottom center path as guide rail. I was surprised to see that neighbouring turns of this winding were intercepting. It turned out that the "scale" option of the Sweep feature was turned on, which scales the swept cross section according to the distance between the sweep path and the guide rail. Since the distance between the two paths is by construction constant *in the plane perpendicular to the tangential*, this implicitly implied that Inventor is not "progressing" along the sweep path and the guide rail in this way, i.e. the swept cross section is not perpendicular to the sweep path. Switching off the scaling option helped somewhat, but did not fully solve the problem, making me believe that my last assumption is valid. 

Finally I choose the Loft feature, because that can define cross sections at regular intervals along the sweep. From the exported points I create guide rails for the Loft by an iLogic macro, and also rectangular cross sections at regular intervals, which regularly enforce some kind of "synchronization" between the progress along the different rails. This produces nice results, but is rather complicated. And it is not perfectly exact (this is probably not a problem, differences are tiny). See the attached illustration: the red pair of lines indicates the top edges of the groove, the blue pair of lines indicates the bottom edges of the groove. The black circle represents the mill. At this instant, the mill's axis is perpendicular to the plane of the figure, but is then rotated if we progress back or forward. What is interesting is that at the top and bottom of the groove, different azimuthal positions of the mill are actually cutting (red and blue dashed lines). This would not be reproduced by a swept planar cross section, unless I miss something. The true machined shape could be modelled exactly with a solid body sweep with the following feature: (1) I can give the sweep path (center line between the two red lines) which defines the position of the top center of the mill, (2) I can give a guide rail (center line between the two blue lines), towards which the axis of the mill is oriented, and (3) the axis of the mill is also perpendicular to the tangential of the sweep path of point (1). 

A few specific questions:

- What is the mathematical apparatus behind these sweep features? How is the orientation of the swept cross section determined? 

- Is the "Spline control points" feature a Bezier curve?

The helps of these features give the impression of having quickly and easily understood what is going on, but are lacking the exact (mathematical) description.

Thank you

Daniel

[kacper.suchomski]

Hi

I think you just need to point to the surface as a reference to control the direction of the tool position.
Pointing to a control surface ensures that you maintain a constant direction (normal, if that's where you start) relative to that surface for the entire length of the sweep.

[johnsonshiue]

Hi! For the two specific questions, the regular sweep position the profile (2D or 3D volume) according to the path normal. The normal vector field isn't apparent to the users. Nor is there easy to use workflow to alter the vectors. You will need to use the Guide Surface to control that. The Guide Rail is to scale the profile (based on the point from the guide rail mapped to a point on the path). It cannot control how the profile is oriented along the path.

I believe the spline curves in Inventor are all Bezier curves.

It will be much easier to share an example that illustrates the request. Based on the discussion so far, I personally do not think the exact geometry can be created by Inventor yet. However, I would like to take a look and explore if there is any alternative workflow.

Many thanks!

[barna.daniel]

Hello,

Thank you for your answer. I think your following statement is not true:

"Nor is there easy to use workflow to alter the vectors. You will need to use the Guide Surface to control that. The Guide Rail is to scale the profile (based on the point from the guide rail mapped to a point on the path). It cannot control how the profile is oriented along the path."

I have applied a part file which demonstrates that a straight sweep path and a wavy guide rail causes the profile to twist around the sweep path.  The cross section indeed seems to be perpendicular to the sweep path - the guide rail is shorter than the sweep path, and the sweep is stopped there (before the end of the sweep path), in a perpendicular way to the sweep path. 

Thank you

Daniel

[johnsonshiue]

Hi! Many thanks for sharing the example! I am sorry my statement was false indeed. The profile orientation can be altered by the guide rail also. It seems to be consistent with the vector from a point on the path to the mapped point on the guide rail.

Unfortunately, there isn't an option to do the same for Solid Sweep at the moment.

Thanks again!