Creating Mesh using specific XYZ Coordinates

[Anonymous]

Long story short- If I have an Excel file with numerous XYZ coordinates, how can I create a mesh on Autodesk Inventor 2021 to hit those specific XYZ coordinates?

To explain the task in more detail, I have to build a weaving fiber and then apply a mesh pattern that hits the specific XYZ coordinate points I have on an Excel file.

So far I have the external shape of the weaving fiber (shown below):

From this weaving fiber, I need to build a mesh pattern that divides the weaving fiber above into subvolumes- specifically, a mesh that forms a consistent grid pattern across 5 different cross-sections (the two ends and the 3 orange cross-sections in the diagram below).

As mentioned, the location of the grid points created by the mesh must match exactly with the XYZ coordinates given by my partner.

Please let me know how this can be achieved on Autodesk Inventor 2021.

I am attaching the Excel file of the coordinates- please note that the excel file divides each cross-section into about 200 segments and then provides the XYZ coordinates of the corners following the 1-2-3-4 convention for each segment.

[cadman777]

Your intent is unclear to me.

When I open the Excel file in Inventor or Rhino3D I see strings of points that do not resemble your second pic.

Do you want to create a fiber along each string of points having the elliptical profile that you show in pic 1?

On second thought, it appears you have improperly formatted your excel file:

Best Practices for Importing Points 

Scroll down to bottom and see what it says.

[cadman777]

See attached example of what separating the points did.

But I'm still unclear about what you want to accomplish with these points, b/c the point groups appear to overlap.

[cadman777]

The attached model took me about 10 minutes in Rhino3D R6 (runs on Win 7, R7 runs only on Win 10).

This is only of #1, not #'s 2-4.

I'm still unsure what you want to accomplish.

Do you want to connect all the points lengthwise with a spline too?

If so, then you may need to use Python to write a script to get Rhino to do this.

Or else you need to use Grasshoppper with Rhino, which seems to be your best bet in this type of scenario.

You can visit the Rhino forum to inquire about all that (I don't know how to write Python scripts).

Rhino Forum: https://discourse.mcneel.com/

Rhino (free download works as viewer when Trial ends): https://www.rhino3d.com

 

[Anonymous]

Thank you for the assistance and apologies for the miscommunication.

I will try my best to clarify.

The following diagram is the side view of the fiber:

Note the directions of the X, Y, and Z axes and the position of the origin.

On the Excel file, each line is referring to each quadrilateral-shaped subarea of one of the five cross-sections, which is the reason why there are 12 vertical columns (x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4)- following the 1-2-3-4 convention for each quadrilateral subarea I mentioned in the initial post. Some combinations of XYZ coordinates are repeated on other lines because the left-side border of one quadrilateral subarea can be considered as the right-side border of another quadrilateral subarea.

Each of the five cross-sections across the fiber (located at Z = 0, 0.25, 0.5, 0.75, and 1 mm, as seen on the diagram above) is divided into about 200 quadrilateral-shaped subareas that form the grid pattern like the second image I attached in the original post (except in the image I drew the grid pattern only on one of the cross-sections when there should be five). In fact, lines 2-201 on the excel file refer to the cross-section at Z = 0 mm, and then so on (the rest labeled on the diagram above).

I want to create a mesh (for simulations) around and into the fiber on Autodesk Inventor 2021- from the cross-sectional view, the mesh creates the grid pattern on each of the five cross-sections by dividing each cross-section into quadrilateral subareas with specified XYZ coordinates for the corners (given by the Excel file). From the side view, the mesh follows the exact curvy path of the fiber to connect every quadrilateral subarea of one cross-section to the corresponding quadrilateral subarea of another cross-section. In other words, a mesh that divides the fiber into curvy subvolumes- kind of like 3D puzzle pieces that form the fiber when you assemble them together.

If my explanation still confuses you, then answer this question instead- if I have a bunch of XYZ coordinates on an Excel file, how do I create a mesh on Autodesk Inventor 2021 that intersects through all the specific XYZ coordinate points?

Is it possible to achieve this using Inventor only? Or does it always involve another software?

In fact, if you know other CAD programs capable of this function, then please let me know.

I have only dealt with generating random mesh on Inventor, so I appreciate your assistance.

[cadman777]

Thanx for your further explanation.

I'm clear now that all of this matrix is in ONE fiber only.

The fiber is shaped like the segment of a 'snake' moving in 3D space.

Let me think a bit more about the rest of what you said and get back to you.

I can't imagine doing any of this in Inventor automatically.

The only way I can see to accomplish this is using a programming language or doing it manually(!).

But there are others in here with more expertise with Inventor software who may be able to shed more light on how to accomplish this in Inventor. Alternatively, you can go to the Inventor Customization forum to inquire, b/c they write programs for the software and may be able to help.

Meanwhile, if I were you, I would go to the link in my previous explanation and download the Rhino3D software (R6 for Win 7 and older, R7 for Win 10), and get on their forum and ask people about how they would do it. There's some VERY sharp people in there who know the software like the back of their hand, and others who can write scripts or who use Grasshopper to accomplish such feats. GH is a visual language that enables you do to AMAZING things w/Rhino! When you download Rhino you can open my file from the previous message to see what I did as well as see the mesh commands available to you.

[cadman777]

I believe I get it now.

Please explain why your detail in pic #2 shows a rounded corner @ 4:

Also, what connects the vertices between sections of the fiber (1.1 to 2.1 to 3.1 to 4.1 to 5.1, and so on)?

Or said a different way, these tiny fibers seem like a special kind of binding aggregate in a fiber-reinforced product.

So that would mean that you have to 'grow' this fiber a certain way to give it strength (per the cross sections).

It just seems like the mesh cells define FEA nodes in this fiber.

Also, this is how it would be typically 'meshed', unless you were doing a quad-mesh:

You'll see this in my Rhino file.

I assume you want this minus the triangles lines, right?

And you want to extrude these from section to section to form a smooth fiber, right?

Also remember that this model is only one set of the 4 sets of x,y,z point coordinates.

[cadman777]

Check out his thread in the Rhino forum:

https://discourse.mcneel.com/t/quad-mesh-from-point-cloud/125941

[johnsonshiue]

Hi! If this is about FEA mesh, Inventor Stress Analysis does not have such setting. I believe you will want to use Inventor NASTRAN or Fusion Simulation.

Many thanks!

[Anonymous]

Thank you for your explanation!

I will try my best to answer your questions.

Q1. Please explain why your detail in pic #2 shows a rounded corner at 4.
My usage of "quadrilateral subarea" in the previous response is just a general term I used to refer to the shape of the subarea on a cross-section (plus, each subarea has four corners according to the Excel file). The subarea shape obviously gets more distorted from a quadrilateral the closer you get the edge (with curved boundaries), so sorry for the confusion there. Corner 4 is rounded because I was referring to the specific subarea near the middle of the first cross-section just to explain the 1-2-3-4 convention, and I thought a rectangle with one rounded corner best described that specific subarea. If you look back at the diagram, I drew an arrow pointing to a specific subarea on the cross-section shown with the grid.
In other words, please don't take "quadrilateral subarea" and the fact that Corner 4 has the rounded corner seriously- they are just general terms I used to describe the overall situation.

Q2. What connects the vertices between sections of the fiber (1.1 to 2.1 to 3.1 to 4.1 to 5.1, and so on)?
The cross-section grid pattern created by the mesh should be consistent for all five cross-sections. This means that every subarea on one cross-section has four other identical, corresponding copies on the other four cross-sections. I see that you have five cross-section grid patterns for the mesh- is it possible to connect every subarea on one cross-section to the corresponding subarea in four other cross-sections (using a specific curvy path)? I will provide the useful dimensions at the end of this paragraph, including the equation curve I used as the path for the Sweep function on Inventor.
Back to my comparison with a 3D puzzle, basically, think of the mesh as an extremely thin skeleton frame of the 3D puzzle that divides the fiber into multiple curvy subvolume pieces.

If your definition of a node is a point where mesh lines intersect, then yes- nodes should basically lie on the corners of the quadrilateral subareas, the XYZ coordinates for which are specified on the Excel file I provided at the beginning.
Yes- getting rid of the diagonal red lines to erase the triangular mesh pattern would be perfect!

The following diagrams specify the dimensions of the fiber.

The cross-section stays consistent throughout and all units are in mm.

Cross-Section:

Sweep Path Equation:

Quick question about Rhino- once I complete the frame of the desired mesh on Rhino, will I be able to import and apply the mesh to my fiber already built on Inventor? Or will I need to rebuild the fiber on Rhino?

Once again, I appreciate your tips and willingness to assist me.

[cadman777]

Q1 & Q2: OK, sounds good.

Along those lines...

Here's what Pascal did 'manually' in the Rhino forum (quad-mesh in Rhino 7):

Far as I know Rhino does NOT has an equation curve.

But you can do much more using GH inside Rhino:

https://discourse.mcneel.com/t/equation-based-curves/41338/9

I think you would build your model in Rhino (like I did) and export it out as a generic file format (like STEP) and then import it into Inventor as a generic file. Here's a PRICEY add-in for going from native Rhino to native Inventor features:

https://www.bimdex.com/rhino-to-inventor-exporter.php

Check out the thread I linked you to above to see how this discussion is progressing in the Rhino forum...

[Anonymous]

johnsonshiue,

Thank you for offering an alternative approach.

To clarify- I have a bunch of XYZ coordinates of points in an Excel file and I want to generate a mesh on Inventor 2021 where the XYZ positions of the nodes lie exactly on the XYZ coordinate points from the Excel.

Can you please describe in more detail how to go about achieving this task on Inventor Nastran or Fusion 360 Simulation?

I appreciate your assistance.

[cadman777]

Incidentally, usually when I create models in 3D CAD I place the sections perpendicular to the curve's tangency, and not all parallel like you show (unless that's what you intend to accomplish). That way the cross sectional boundary at each cross section is properly represented in the extrusion. Otherwise, unless the point cloud makes provision for each skewed section (the middle 3), the extrusion will be skewed and inaccurate.

It would be like having a circle at position 2-201, and a circle at position 202-401, etc., and then extruding the 2-201 true circular section through 202-401 and expecting it to be truly circular, when it will actually be elliptical. So typically I would, if possible, make an accurate curve at the center of all sections by making each section perpendicular to the curve's tangency at the point a sketch work plane crosses the curve. And that curve I typically would draw through the center of all sections, if possible. That way you setup your model sketches accurately so the geometry doesn't become unmanagable down-stream.

vs.

[johnsonshiue]

Hi! I thought you were talking about Stress Analysis (FEA). Meshing in FEA is the key pre-process step to perform simulation. Inventor Stress Analysis offers a few options to generate mesh but specifying coordinate isn't an option. Inventor NASTRAN and Fusion Simulation may have the ability.

But, I think I misunderstood your request. You want to create the mesh on the Brep body without doing FEA.

Many thanks!

[cadman777]

Forgot to answer this one:

"In fact, if you know other CAD programs capable of this function, then please let me know."

I don't know of any b/c I don't know much about meshing programs (except a handful of scan/RE programs).

But if you go into the Fusion forum you may run into people who have used other software besides Fusion or Maya or MeshMixer. They will be able to tell you which software can do this kind of thing, if possible.

Also, if you go to this Grasshopper and watch the first few videos you may find a way to do this automatically:

https://vimeopro.com/rhino/grasshopper-getting-started-by-david-rutten/video/79844298

Maybe you can start with your cross section profile and then divide it up and draw lines between the vertical and horizontal points. then connect them all from section to section. Just a thought. But you have to spend a few hours learning the basics and doing trial and error. I really like the visual programming over the coding method of doing things. You can accomplish a lot of higher level modeling with the graphic programming interface that you can't do with the basic commands in Rhino. Check it out, you may get some ideas on how to accomplish this challenge.

[Anonymous]

Thank you for the additional insights.

I am writing this response to clarify parallel vs perpendicular extrusion and then to ask a few questions about the mesh feature on Inventor.

The reason why I drew the three orange grid mesh cross-sections in the middle parallel to the two black grid mesh cross-sections on the edges is simply because it's one of the requirements.

If you review the original Excel file, you will notice that lines 202-401 have a common Z value of 0.25 mm, lines 402-601 have Z = 0.5 mm, and lines 602-801 have Z = 0.75 mm.

I understand your statement that this requirement makes the task extra challenging, but the fact that about 200 lines have the same Z value implies that all five cross-sections must stay vertical regardless of the extrusion path.

I am currently an undergraduate student, so I was recently introduced to the concept of mesh.

I feel like understanding how the mesh feature exactly works on Inventor is the key to developing approaches, so please answer the following questions to the best of your knowledge.

Q1. When I generate and apply a random mesh on Inventor, does the mesh divide only the surface of the object into subareas, or does it divide the whole object into subvolumes (i.e. does the mesh cut into the object)?

Q2. I am sure you are well aware by now that I have to create three internal cross-sectional grid mesh patterns (three vertical orange lines in the side view diagram).

I am not sure how this will go, but on Inventor 2021, can I pick an internal surface of the fiber (maybe by creating 3 vertical work planes, located at Z = 0.25 mm, 0.5 mm, and 0.75 mm, or by splitting the fiber about the three vertical work planes) and apply separate grid mesh pattern for those internal surfaces?

I was thinking that if your answer to Q1 is that Inventor mesh only cuts the surface into subareas, then maybe I can apply the mesh on the surface first and then apply three additional cross-sectional grid mesh separately on the three internal cross-sections?

Q3. Once I generate and apply a mesh on a part on Inventor 2021, will I be able to generate an Excel file of the XYZ coordinate locations of the mesh nodes?

As always, I truly appreciate your assistance.

[cadman777]

Thanx for the clarifications about the coordinates.

When answering your questions I can only speak about the general concepts of the features.

But others from Autodesk will have better answers than me, b/c they are the developers of the product and know it's design functions.

Q1: Only the surface. Meshes are surface features. They create 'shells' or 'skins' using either triangulation or quadrilaterals (depending on the software's functions).

Q2: That's how I would do it if starting from scratch. But since you already have all the coordinates in x,y,z, you already have them when you import the CSV file into an Inventor sketch. But if you want to start from scratch, then you can do all of it manually. But that would be WAY too much work placing the points individually (1000 points!).

     So, I would say the easy part is importing the point clouds. But the rest of it is difficult b/c it's manual work drawing all the lines 'connecting the dots' in the vertical and horizontal directions. Maybe someone else in here knows a better way to make that grid for you from lines/curves/splines?

     Another thing to consider is that the mesh may not be the best way to solve this challenge. Correct me if I'm mistaken, but it seems you want to make a single fiber that consists of a bundle of fibers. If so, I'm not sure using a mesh is the best way to approach this, due to how Inventor creates extrusions. Maybe sweeping a closed area along a path is more suitable a method to accomplish this? Is the pic in one of my previous posts showing the snake-like extrusion representative of what you want to accomplish?

     Another question: Are all the grids the same on each cross-section, except for position on the plane? In other words, can you make one grid on the first work plane and then copy that sketch to each successive plane, except position the whole thing according to the main equation curve? If so, then this may be a key to getting your project accomplished. Whether or not Inventor 2021 can do it automatically is above my pay grade at this point, b/c I don't have that new a version of it. 

Q3: I don't know about 2021, but 2010 (my version) is impossible to do that 'out of the box'. Maybe someone can do it writing an iLogic or VBA rule, or maybe one of the add-ons for Inventor can do it like NasTran FEA, but I can't tell you b/c I don't know. Maybe someone else in here can say?

     Also, I don't know if Inventor 2021 can create a quad-mesh like you want. But if it can, then you probably would have to place a point at every node and then run an iLogic or VBA rule to find and i.d. each point's x,y,z coordinate and compile it in a data list and output it to a text file. I'm not a coder, so I can't tell you how to do that. Or maybe you can export the sketch into DXF format and bring it into AutoCAD and select all point, then invoke the List command to compile that coordinate data list?

[swalton]

I've been watching this thread and I have one key question:

Why do you need to model this mesh? 

Maybe there is a different way to get to your end goal.

You may get more detailed information about a FEA mesh if you use Inventor Nastran to generate it.  This tool creates standard Nastran input files that you may be able to parse with other software tools to get the XYZ coordinates of the mesh.  

[Anonymous]

cadman777,

Sorry for the delayed response- I have a wifi problem yesterday.

To answer your questions:

Q1. Is the pic in one of my previous posts showing the snake-like extrusion representative of what you want to accomplish?

If you are referring to the picture included under "Message 11" (on 6/15, 7 AM), then that seems correct.

I assume each subarea of the grid mesh pattern is extruded to the corresponding subarea on the next grid mesh pattern?

Q2. Are all the grids the same on each cross-section, except for position on the plane?

Exactly. Another way to explain it is you have one cross-sectional mesh grid, and then four additional copies are made at different locations of the curve equation I provided earlier.

I still have a few approaches that I want to attempt on Inventor and Fusion 360, but I have a few questions about Rhino in case my attempts are unsuccessful.

Q1. I know Inventor and a bit of Fusion 360 and Solidworks.

How difficult is Rhino to learn, especially in a short amount of time?

Q2. I don't think Rhino is a program purchased by my University unlike Inventor and Fusion 360.

If that's the case, then will I be able to access Rhino only for the duration of the free trial?

Q3. I remember you mentioned that Rhino doesn't have an equation function, so creating the exact fiber on Rhino will be less accurate compared to Inventor.

Will I be able to import the fiber part that I already have on Inventor onto Rhino and work on the mesh from there?

As always, I appreciate your assistance.

[Anonymous]

swalton,

Thank you for following along.

The initial Excel file is something my partner generated using MATLAB.

Eventually, I will hand over my Inventor file of the fiber + mesh to my partner, who will use MATLAB to come up with a mathematical model of the fiber's behavior.

Thank you for the advice.