# Fillet Modeling for Manifolds and Volute Tongues etc

## Recommended Posts

Hi together,

Please find attached an example for parametric fillet modeling between two given surface. The parameter "OFFSET" in the scope "filletmodeling" controls the width of the fillet; you can change it e.g. in the range [0.05,0.025]. The fillet is able to change its shape along the "sweep" direction - there is an input called "factor" which can hold a function. In the attached project file, this input is constant and set to 0.5.

I use this fillet for manifolds and volute tongues etc. The following steps are defined in the feature definition "fillet":

1. Run along a main surface curve and find the corresponding position on the second surface curve
2. Calculate the difference vector between these two positions
3. Project the difference vector onto the tangential planes of the two surfaces at the two positions
4. Use the resulting vectors for setting the directions of the connecting bspline curve

The object "testing" in scope "filletmodeling" shows how this section definition works. This feature definition is then used for a meta surface. I like these kinds of definitions - they nicely show the possibilities of customized curve/surface designs by utilizing basic math stuff.

Intersection

There is another feature definition "intersection" which creates two surface curves. This intersection is based on surface curves and surface intersection points - quite nice for demonstrating loops and surface domain modeling. These surface curves are then used later on by offset curves to widen the intersection area (for squeezing in the fillet). Note that there is the assumption that the surface parameter "v" corresponds to the "sweep direction" of the surface. If your surface is swept in "u"-direction, the intersection feature definition needs to be modified slightly.

Trimming / Subsurfaces

As far as trimmed surfaces are not directly supported in CAESES/FFW, trimming can be realized by using sub surfaces. There is another feature which "cuts" out the hole by using the domain information of the intersection.

Cheers

Joerg

fillet.fdb

##### Share on other sites

Remark: This fillet style is not suited for cases where the normal vector of a tangential plane approaches the difference vector (bullet point 2 from the last post). I have seen this situation e.g. for some volute models etc.

##### Share on other sites

Here is another variant:

The vector that gets projected onto the tangential planes is now averaged so that the projection works fine for my examples, including volutes.

Cheers

Joerg

fillet_general.fdb

volutesymmetric.fdb

##### Share on other sites

And once again ... here is another version where the tangents of the bspline curve are defined via the normal and tangent vector. I think, this is my favourite generalized version :-)

Cheers

Joerg

fillet_general.fdb

##### Share on other sites

Just as a remark if performance is an issue:

You could remove all the visualization stuff from the feature definition - I usually keep it in order to have a visual possibility to check if problems occur. But, removing all the unessential objects improves the performance of this definition. Furthermore, all points could also be defined as simple vectors (FVector3) without visualization, which additionally speeds up the surface generation process.

##### Share on other sites

Please find attached the feature definition - just put it into your features folder (e.g. application data > friendship > features) for direct access in CAESES/FFW.

filletviabsplinecurve.fdf

##### Share on other sites

Hi together,

Please find attached a quick example for a volute tongue. I extracted this special modeling part from the complete volute designs for a better understanding.

Cheers

Joerg

PS: Instead of having a constant factor for the fillet, you can add a custom function for the factor of the fillet surface (and set it at the curve engine) so that the tongue surface has more degrees of freedom.

volutefillet.fdb

##### Share on other sites

Here is a version where I create a fillet between two given surfaces by using another way of finding the offseting curves (i.e. the surface curves where the fillet gets joined to). This approach is useful in case the surface curves need to have a constant offset to the original intersection curve (i.e. constant on the surface).

The presented fillet is close to the result of a radius-based fillet surface. In order to ensure constant radius, we still need to inscribe a circular arc.

Cheers

Joerg

##### Share on other sites

Just a short update:

With version > 3.1, we can now also use BReps for general fillets with a single click. Useful e.g. for blades (hub fillet), manifolds and so on where the fillet itself does not vary (or is not considered during a shape optimization).

Cheers

Joerg

##### Share on other sites

Here is an example that was made by using the breps objects in CAESES. This is one of our standard approaches to model robust fillets for variable geometries.

##### Share on other sites

Good morning Joerg,

I am a fellow researcher that is currently developing and designing an assymmetric volute for a centrifugal compressor in Caeses. I am encounting several complications when it comes to creqating a fillet between the exit duct and the volute. I have observed you have already provided a code to create a fillet within a volute nevertheless, it would be greately appreciated if you could give explicit instructions on how to model such case and how you ahcieved every step.

Kind regards,

Marta Vazquez

##### Share on other sites

Hi Marta,

We have different techniques for these fillets, depending on the shapes of the cross sections. I will get back to you directly.

Cheers

Joerg

##### Share on other sites

Hi Marta,

As discussed, we will try to find an example that we can share in this forum.

Cheers

Joerg

##### Share on other sites

Hi Marta,

attached is an example how you can model the tongue area. Basically we want to trim the volute and the outlet surface in order to have space for a fillet surface. To do this we follow these steps for each part:

1. create intersectioncurve  e.g. volute and offsetsurface of outlet

2. modify the domaincurve of the intersectioncurve. create also an offsetcurve of the new domaincurve

3. create surfacecurve from new domaincurve and a surfacecurve from the offsetcurve, which will be tangent edge for the fillet surfaces

4. create a subsurface from the new surfacecurve

5. put the subsurface into a brep

After you did thtis with both parts it should look like this:

Now you can use the edges and tangent edges to create the fillet surface with a meta surface:

The final results can look like:

best regards

Carsten

volutecompressor_Training.fdb

##### Share on other sites

Here is another variant:

The vector that gets projected onto the tangential planes is now averaged so that the projection works fine for my examples, including volutes.

Cheers

Joerg

Hi Joerg and Carsten,

I am exploring your software and its capabilities for turbomachinery application. In the moment and wanted to have a look on the 2013 volutesymmetric.fdb model as it is close to what I am looking for. Unfortunately it is not compatible with the actual CAESES version 4.3.1. any longer. I wonder if you have an updated model and could upload it into this thread?

Best regards,

Matthias

##### Share on other sites

Hi Matthias,

Please find attached the updated project.

Cheers

Ceyhan

volutesymmetric.fdb

##### Share on other sites

Hey together,

I want to create a fillet between the spiral casing and the exit cone, but when it comes to creating surface curve and subsurface, I'm getting several problem. One of them is that Caeses doesn't recognize the correct subspace, when I want to create one. Here's a simple example. Attached you can see that with the side index "1" the intersection part is captured , which has to be deleted. The complementary of this part is the correct part, which has to be kept as my subsurface. But as you can see here , there is no way to get this part by changing the side index from 1 to 4. Here's the domain curve as well . Any Idea or suggestion?

Best,

Vahid

##### Share on other sites

Hi Vahid,

Can you show a picture of a geometry that shows where you wanna go finally? Maybe there are better and easier approaches to design your geometry.

Cheers

Jörg

##### Share on other sites

Hi Jörg,

I'm working on my master thesis at MAN company, where I have to develop an automatized design process chain to design assymmetric volute casing in Caeses. Unfortunately because of company secret, I cannot provide you the original geometry, espicially here in a public Forum. I want to write a feature definition that recognize automatically the correct intersection curve between the volute an the exit cone, trim them afterwards and then create a fillet with a constant radius between the two. Since the parameter s of our geometry are very variable, the process should not be error prone or depend on individual casees. BTW, we are in contact with Mr. Mattia Brenner in this regard. But now I have a specific question. what's the best way to trim intersected volute and outlet cone?

Cheers,

Vahid

##### Share on other sites

Hi Vahid,

OK got it! Understandable.

From what I have seen so far, we either trim completely using BReps (offseting the surfaces and trim), or we use subsurfaces in combination with intersection curves, or we use both (subsurfaces & BReps) in combination. This typically depends on the volute shapes. Probably we can support you a bit more directly within this evaluation, Mattia will come back to you.

Cheers

Jörg

## Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

×   Pasted as rich text.   Paste as plain text instead

Only 75 emoji are allowed.

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.