Hexagonal Perforation

[speckofdustin]

I've been trying off and on for the last few days to create a hexagonal
hole pattern like in the attached picture. After hours of trying, I
had to hang my head in shame and let the resident SolidWorks user model
the pattern for me in SW (which, by the way, took about 3 minutes to
figure out in that software).

So, I'm wondering how someone
would go about creating a perfect honeycomb pattern of holes in a part
in Pro/E. I got nowhere with the pattern tool and was resorting to a
lot of manual work before I gave up.

I'm guessing there's an obvious way to do this. Any suggestions?

Let's say the holes are 3mm in diameter with 1mm of material in between...





Edited by: speckofdustin

 

[CRang72]

You can use a pattern table, such as the one below. This was my first time using a pattern table, and it really didn't take me very long to figure it out.


After creating the original feature, select it, then select pattern. Change the pattern type to "Table". Select your reference dimensions (hold CTRL to select more than one). Then select the "Edit" button next to the active table name.


The first column is the pattern instance which you define, and the second and third columns are the dimensions you reference for each instance. With this you can input exact dimensions for each individual instance. May not be the quickest thing to do, but it works and doesn't take too long. I just used the "help" section on pattern tables in Pro/E for a push in the right direction.





The Result:

 

[Dell_Boy]

a table by pattern is a friggin awful way to do this.

Is this as big as you want to make it.

There are several ways to do this. The choice between is somewhat dependant on how large you want to make it in both directions and the overall final outline.


DB

 

[speckofdustin]

Interesting. I thought a table might work for this. Thanks for that suggestion, CRang72. I'll have to learn more about that. Seems like an interesting tool to play around with too.

Dell_Boy, your concerns about tiling this over a large area are appreciated. I was trying to punch a pattern like this in a part roughly 200 x 30 mm in size, so several hundred holes. I wanted the pattern to end about 2mm from the part's edge. I can't show the exact part, but it was similar to the image below.



Thanks again for the input.


Edited by: speckofdustin

 

[Dell_Boy]

The following method is probably the most suitable for what you are trying to do and please note that I have left gaps between some of the cells to assist in pattern recognition



First hex cell (bottom left) was sketched using a datum-on-the-fly as the right sketcher orientation plane.

Cell was dimension patterned 6 times in the x direction

Bottom row of 6 was dependant copied to form the top "odd" row. The only independant dimension is the single y dimension

First row is grouped

First row is group-patterned in the first direction 2 times by a combination of x and y dimensions and in the second direction 3 times by the y dimension.

A few simple relations are written to link all the increments to the AF of the hex.


DB




Edited by: Dell_Boy

 

[pjw]

Or use pattern relations?

 

[Bob_W]

Speck:


PJW was on the right track with the pattern relations. Here is the WF2 model set up with patterned axes of which you can create any other patterned feature by reference.


2008-06-05_151531_hole_pattern.prt.zip


Bob

 

[mganzer]

Can't this be done with a carefully set up fill pattern?

 

[Dell_Boy]

PJW

yes and I can think of other ways to do it as well but relation based patterns are something that I have hardly ever used so it gets hard to remember the exact syntax of memb_v lead_v and idx1 in the relations


Bob_W

the reason why I did it that way was because Dustin appeared to want an odd number of rows. Your model has 6 rows, how does it handle 7 rows.

Also your method does not lend itself well to putting in hundreds or maybe even thousands of holes. One of the reasons is you take around 3 features per hole. With a more efficient method that is more appropriate for even numbers of rows your 27 perforations could easily be modeled in around 30 features instead of 89.



DB

 

[Dell_Boy]

No. Second thoughts. Can both those ideas, I just realised a better way.

Create the first feature, pattern all the features in the x direction first and in the y direction, pattern the 1st, 3rd, 5th etc rows.

Next create a dependant copy of this pattern but make 4 dimensions independant, the x and y to the first feature and the x and y repeat values so you can interpose the even rows between the odd rows.

Write a few simple relations to link everything together

With this method you can have an odd or even number of rows and the alternate rows can have the same or differing numbers of features.

And to cap it all, a pattern of 18 individual holes can be done with a grand total 19 features.

It seems that you can have your cake and eat it too.




DB

 

[Bob_W]

DB


The third feature pattern was just a datum curve for visual effects and was not required. I could have left the axes pattern out also but thought it would make it easier for later changes of features shapes. If you want seven rows, simply turn off the 8th row. Took a couple seconds on this model. The entire model only takes a couple minutes to create using this method. There are other ways to create this as you mention so your mileage may vary.


2008-06-06_090910_hole_pattern.prt2.zip


Bob

 

[Dell_Boy]

I was mistaken in my belief that patterning to table was the method of last resort.

I had forgotten about over-patterning and manually turning off the ones that are in the way which presents a good challenge for that title


Thanks for taking the time to do the relational zig-zag pattern. i knew it could be done, I was just a little rusty in that area


DB

 

[Bob_W]

DB


I thought when I read the post he was looking for a table driven pattern, but after re-reading I think I was wrong. The table did make it more complicated than required as you had suggested. We use them occasionally but not too often since they are a little cumbersome to use and maintain. I think a regular pattern and group would work better. If I recall,the keyto getting the pattern of a pattern to work you have to include the reference datums in the group. I just did it by habit and didn't try otherwise.


Bob

 

[Bob_W]

DB


Now I probably have you confused, I reread what I had posted and think i washaving a senior moment

In another thread I had patterned a pattern using a table but this one was done with relations. So sorry for the confusion.


Bob

 

[airion]

I found an easier way. Use a fill pattern, triangle type. No math, if statements,or relations required.


2008-06-13_104940_hex_pattern_2.prt.zip


Al

 

[Dell_Boy]

Airion

While a fill pattern is very easy to construct and control it is very ss-llll-oooo-wwww to regenerate particularly when you get a high feature count.

As an example I modified the size of the hex and your model took about 35 seconds to generate in WF3 on my PC

On the other hand my semi-dependant copied pattern

2008-06-16_051017_copied_pattern.prt.zip

which has the same feature count takes about 3 seconds.

No such thing as the perfect solution. Each and every one has its own set of advantages and disadvantages. Sometimes solution A is better, sometimes it is solution B or C or D etc..


DB

 

[Asho Pulsar]

I found an easier way. Use a fill pattern, triangle type. No math, if statements,or relations required.

Fill pattern,triangle typeis the best one and easy too. Its just the matter of matching the pitch to get well structured honeywomb. Look at the picture below.