Need to make a bi-rail loft

[pimm]

I am struggling on something that is extremely simple.  I need to match geometry built originally by customer as a bi-rail loft.  The sections are quite simple but the solution to this is eluding me.

Here is the frame I have to start with.

My 1st inclination was to attempt this as a Swept Blend.  I constructed this by choosing 2 reference curves.  As seen below the main reference curve looks great but the solid totally misses the secondary rail.

Another method that I've tried is through using a Boundary Blend.  This would work if the side surfaces didn't want to become the top surface going around the sweep.  I tried playing around with Control Points but in this specific instance it appears that they don't work properly.

I also attempted this as a Sweep but I don't even want to share those results it was so bad.

By any chance is there just a hidden bi-rail loft command in Creo?

If not would anyone have a recommendation for getting a good solid or quilt based off the curves in my top image?

[pausob]

I'd recommend for you to utilize variable section sweep.

You can add trajectories other than the origin to your sweep and use those to drive the cross-section (use CTRL key after your origin trajectory is defined)

[pimm]

Pausob:

I believe I tried this as a Variable Section Sweep.

This is what happens.  I tried different settings.

[MartinHanak]

---

@pimm wrote:

I am struggling on something that is extremely simple.  I need to match geometry built originally by customer as a bi-rail loft.  The sections are quite simple but the solution to this is eluding me.

Here is the frame I have to start with.

My 1st inclination was to attempt this as a Swept Blend.  I constructed this by choosing 2 reference curves.  As seen below the main reference curve looks great but the solid totally misses the secondary rail.

Another method that I've tried is through using a Boundary Blend.  This would work if the side surfaces didn't want to become the top surface going around the sweep.  I tried playing around with Control Points but in this specific instance it appears that they don't work properly.

I also attempted this as a Sweep but I don't even want to share those results it was so bad.

By any chance is there just a hidden bi-rail loft command in Creo?

If not would anyone have a recommendation for getting a good solid or quilt based off the curves in my top image?

---

Hi,

I think if you upload your "startup" model containg curves, somebody can experiment with it ...

[pimm]

Martin,

If I could figure out how to attach a file I could try sending a file.  I pared this out of the imported model.  No guarantees if I could get the file uploaded anyone could even open it.  I see an Insert/Edit link from the tools but it is looking for a URL. 

Since yesterday's post I was able to successfully build the boundary blend without kinks but it did not match up with the solid shape that it should represent.  The rail side and end frames were perfect but the opposing depth was considerably short of what the geometry I need to match with is.  (I was able to figure out how to use the Control Points in this application).

[MartinHanak]

---

@pimm wrote:

Martin,

If I could figure out how to attach a file I could try sending a file.  I pared this out of the imported model.  No guarantees if I could get the file uploaded anyone could even open it.  I see an Insert/Edit link from the tools but it is looking for a URL. 

Since yesterday's post I was able to successfully build the boundary blend without kinks but it did not match up with the solid shape that it should represent.  The rail side and end frames were perfect but the opposing depth was considerably short of what the geometry I need to match with is.  (I was able to figure out how to use the Control Points in this application).

---

Hi,

if you want to upload Creo files, pack them into single zip file and attach this zip file. Unfortunatelly it is not possible to upload Creo files with double extension.

[pimm]

Martin,

I finally see the place to attach.  The file still might not display properly; I've never attempted paring out data like I did.

[MartinHanak]

---

@pimm wrote:

Martin,

I finally see the place to attach.  The file still might not display properly; I've never attempted paring out data like I did.

---

Hi,

1.] 2 curves connecting 2 bottom vertices of start/end rectangles are 3D curves

2.] start/end rectangles have different dimensions

3.] are you able to get 2 curves connecting 2 top vertices of start/end rectangles ???

[pimm]

Martin:

This model was developed in Catia.  Evidently that software can do a bi-rail loft.  This was all of the curve data from their model.  Since I couldn't match up with their solid I was hoping that if I used their base construction I might be able to get an exact match.

I could develop the top curves but it would mean that I would have to use a different construction technique.

Prior to trying to use the customer's curves I built the frame as a Swept Blend.  I was seeing deviation of up to .010(.25mm) which was not good enough.

The framework for an exact match is there with the curves, but I don't know whether Creo can interpret it.

[MartinHanak]

---

@pimm wrote:

Martin,

I finally see the place to attach.  The file still might not display properly; I've never attempted paring out data like I did.

---

Hi,

I attached my test model ... inside birail-curves_2020-07-09.zip file

[pimm]

Martin:

Thank you very much for looking at my geometry and building from scratch.  I am glad that the model came across to be accessible.

Your result was similar to my own when using Boundary Blend.  It interprets it differently than Catia's bi-rail loft or the customer applied an influence that was not found in curves.

My earlier Swept Blend model was not correct but fairly close to what I need to match.  Here is how the boundary blend combination you uploaded given the limited curve data compares to the target.

I'm not sure if Creo can apply the curve structure the same way Catia could.

Thanks for your help!

[MartinHanak]

---

@pimm wrote:

Martin:

Thank you very much for looking at my geometry and building from scratch.  I am glad that the model came across to be accessible.

Your result was similar to my own when using Boundary Blend.  It interprets it differently than Catia's bi-rail loft or the customer applied an influence that was not found in curves.

My earlier Swept Blend model was not correct but fairly close to what I need to match.  Here is how the boundary blend combination you uploaded given the limited curve data compares to the target.

I'm not sure if Creo can apply the curve structure the same way Catia could.

Thanks for your help!

---

Hi,

in my test model I used section relations using trajpar function to build side surfaces. You can combine trajpar function with evalgraph function to get "better" side surfaces. See http://support.ptc.com/help/creo/creo_pma/usascii/index.html#page/fundamentals/fundamentals/fund_seven_sub/Example_Graph_Evaluation_Function.html help page.

[pimm]

Martin,

The trajpar that you used doesn't seem to show up in the model.  Is this an outside feature?  That is nice how you were able to match up on the end with both of your sweeps.

There might be times that a small influence would be helpful for matching up with a Sweep.

[MartinHanak]

---

@pimm wrote:

Martin,

The trajpar that you used doesn't seem to show up in the model.  Is this an outside feature?  That is nice how you were able to match up on the end with both of your sweeps.

There might be times that a small influence would be helpful for matching up with a Sweep.

---

[tbraxton]

It appears to me that you can realize the desired geometry with a variable section sweep and a split draft. I am commenting on the pictures only as I can not open your model in Creo 4. It looks to me that a variable section sweep with more than two trajectories would be needed to get there in a one sweep feature. Once the sweep matches then you can create a split draft to get the tapered side walls. 

Trajpar is an internal parameter available for use within the sweep feature. You can use this feature parameter to vary the section along the trajectory. You can read about it in the help files.

When reverse engineering geometry that has been imported it is often useful to break it down into multiple features created in Creo to realize the net shape of the import.

[pimm]

Tbraxton:  The very 1st picture is the curve geometry that I'm attempting to loft.  There are 2 green rails and outer curve boxes that appear black.

The challenge is to drive the outer curve boxes in a way that follows the contour of the 2 rails.

Yes, I do agree that breaking down a part into it's core shape is the way to go.  I am attempting to do this but Creo can't replicate what Catia can do given the limited amount of curves.

I would like to be able to do this with a Variable Section Sweep, but in what I am seeing and given the curves that I have it appears that the sweep will not follow the rails and also be variable.

[tbraxton]

If you post a STEP or IGES of the CATIA geometry I will look at it. The newest rev of Creo I have installed is Creo 4 and I was unable to open the model you posted above.

If I can see the imported CATIA geometry I may be able to offer more insight into a solution.

[pimm]

Tbraxton:

Sure thing.  Here is a STEP model of the curves.

[tbraxton]

Does this look right? It is a quick hack to get the shape Using 2 variable sweeps. I need to look at the imported curves some more to see why a single var sweep can not crate this shape as you experienced.

Both VSS features use trajpar to linearly control the delta (21-16) height of the rectangles at either end. Look in the sketch relations for each VSS. 

sd4=21-5*trajpar

If this basic form is correct then I will need to think about it some more to see if a single sweep using the two import curves could work.

[pimm]

Tbraxton:

The basic shape is correct; however it is apparent here as well as with what I have seen in generating this with Boundary Blend and ISDX the outer wall is not as deep as what the customer model geometry is.  For example look at the image with the Red model over laying my rough base geometry (based off my own Swept Blend feature).  (There are a number of posts here but above I explained that I could get within .010(.25mm) but that wasn't good enough.  The curves are the feature used by the customer to generate the base feature.)

Thanks for looking at my model and seeing if VSS could be incorporated!

[tbraxton]

Is the deviation you refer to as shown in my pic below? If that is the case  then the sketch of the sweep we are creating is not defined with the correct design intent. In your CATIA reference geometry is a vertex of the rectangle constrained to that rail along the entire length of the rail or is it rising above it along the rail?

Without seeing the what the final surface topology you are targeting it is difficult to say how to deal with it. You may not be able to share the final shape here but without that or some more definitive constructs describing the end goal it is hard for me to realize a solution.

 

[tbraxton]

Check this out, the topology is different from my previous and may or may not be closer to target. It was created with a single variable section sweep with normal to projection option for the section.

[pimm]

Tbraxton:  You did much better than me with the VSS.

That is good that you can get the sweep to generate as 1 feature.  You did say that you noted an issue with the topology.  Would this be a tangency or continuity issue or was the geometry just dirty?

I do see that the enclosure still doesn't want to seal as a solid.

The end result is very similar to the Boundary Blend and ISDX solutions.  The outer wall is a lot lower than Catia's solid model.

Even if Catia is using a birail loft (which Creo can't do) I am getting the opinion that they must have used some kind of influence to raise up the way they did.

The beauty of a VSS feature is that it can be built as 1 piece.

Since there are numerous methods that obtain similar results I am wondering what type of method would best allow influencing the outer mid section.  Essentially tugging the outer center upwards along the 1 sidewall while keeping all the other sidewalls at home.

[tbraxton]

Look at the enclosed model, it will require tweaking but may get you closer. You may need to change the order of the ISDX curves to get the match, I quickly did this based on visual qualitative assessment of the deviation.

You should be able to redefine the VSS from surface to solid in my last model to get the solid.

Use the curve analysis "comb" plot to inspect the CATIA import rails. They could be "smoother" but I don't think that is the issue here. Can you create 3D ISDX curves on top of the CATIA rails and then build it all in Creo?

I think I understand the deviation now. If the single VSS is close except for the "top" surface as shown in your model overlay then I would use the solid VSS as I provided and then build the top surface with ISDX and then use a tweak replace to get the "right" surface on the top. Using internal curves on a 4 side surface in ISDX should allow you to get that top surface. 

I know nothing about CATIA surface modeling but a loft through the curves you provided does not seem to define the surface that is not matching in Creo. I suspect it was not just a loft in CATIA to get that top surface either. There seems to be more involved than the two rails to define that top surface.

[pausob]

You can do a 4-curve VSS where you define curves to drive all 4-corners of the cross-section:

4-rail sweep

sweep cross-section

sweep trajectories

those curves can be tweaked (tugged around like you mention) using the "Tweak Curve" tool inside the edit-definition tool:

Curve tweaks

This is basically poor-man's ISDX - so if you have that license, that seems like the best tool for the job.