Consistency check for large parts

Lesley M · 2010-11-26 10:50:28

Hello,

we're making large parts and have problems with the stock and design consistency checks.

Let me first list some important information:
-Units: inch
-Part size: ~200x100x15
-Cutting tolerance: 0.02
-We use *.stl models for the design and stock
-64 bit, 8GB RAM
-VERICUT Version: 6.2.1

Both consistency checks detect errors inside the *.stl models (stock and design) and tries to repair it, without succes.
We tried the PolyFix, also without succes.
We opened the models in a stl viewer software and indeed it reported some errors in the *.stl models.

But when the cutting tolerance is increased to 0.03, the stock consistency check is ablo to repair the model.
Why is that?
What is the influence of the cutting tolerance on the stock consistency check?
We then saved the corrected stock model as a *.vct and used this as stock instead of the .stl model, set back the cutting tolerance to 0.02 ==> cannot repair model!

During AUTO-DIFF, we experienced folowing:
case 1: cutting tolerance set back to 0.02. Run simulation. Perform AUTO-DIFF with design consistency check and use solid comparison method, by region is off ==> cannot repair model.
case 2: by region is on, over the entire stock ==> cannot repair model.
case 3: by region is on, over only half of the stock ==> model repaired! (same for the other half of the part)
case 4: might be memory problem, so we added extra 4 GB RAM to the computer (now 12 GB RAM) and repeated case 1 and 2 ==> no difference: cannot repair model.
Whay can the design model only be repaired if the AUTO-DIFF is performed over a smaller region?

Best regards,
Lesley M

paehv · 2010-11-29 06:49:54

Cutting tolerance represents the size of a cutâ€™s volume measured in any direction. Cuts that are larger than the Cutting Tolerance in any one direction will be detected.
The cutting tolerance is probably used to check misalignments in the stl with model consistency check.

Your value is this in mm or inch?
If it's mm than it's better is to set it to a larger value (0.5 to 1.0) and with autodiff, use "improve cut stock tolerance" an compare by region. Based on your gauge/excess tolerance vericut will regenerate each region and improve tolerance. (if you notice that your model fails on 0.02, just don't use this as tolerance.)
An extra benefit of a larger base cutting tolerance is that it can speed up your simulation dramatically
We always use a base cutting tolerance of 1.0 (mm) and catch all errors.

Lesley M · 2010-11-29 07:38:41

Hi Patrick,
Thank you for your reply.

Units are in inch.

Before, we tried a cutting tolerance of 0.05 (= 1.27 for mm units).
This was indeed a lot faster for simulation but AUTO-DIFF (simple solid method) was not able to find small gouges, for example: gouges of 0.02" (= 0.5 mm) were not detected.
The option "improve cut stock tolerance" makes the computers compute for hours and finally crahes ... even, for small and simple parts.
Do you have better experience with this function?

It surprises me that you are able to catch all erros with a cutting tolerance of 1 (mm).
Don't you get problems with fastfeed errors and holder-stock collisions?

I did some more tests to understand our problem:

case1:
I devided the stock in 4 parts, in 4 different *.stl models and added them to the stock component.
The for models are positioned in line, representing the complete stock model.
The cutting tolerance is set to 0.02.
Stock consistency check ==> cannot repair model.

case2:
Next, I hided the two inner models. So only the two outer models are visible. With a big gap between them.
Stock consistency check ==> cannot repair model.

case3:
Same as case2, but I moved the two visible parts closer to each other, but not tuching.
Stock consistency check ==> model repaired!
The two *stl models were checked and repaired individually.

case4:
I created two diffrent stock models with each two of the four *stl models.
Stock consistency check ==> model repaired!
The four *stl models were checked and repaired individually.
Disadvantage: you need to do the AUTO-DIFF twice: with stock1 and stock2.

Can anyone explain me why VERICUT's stock consistency check acts this way?
Is the volume taken by the cut stock models outlines too big?

Thanks a lot,
Lesley M

paehv · 2010-11-29 11:54:12

Lesley,
Cannot explain how Vericut deals with consistency cheking, perhaps some CGTech wizard can.

Why does a cutting tolerance of 1(mm) catches (almost) everything?..
The value is not a liniar dimension but a volume. In my case 1 cubic mm
Let's say you have a 10mm endmill (footprint of this tool = pi x 5 x 5 = 78.5 square mm)
When you rapid into the model 0.02mm this would lead to a volume of 78.5 x 0.02 = 1.57 (caught)
gauges smaller than 1 cubic mm will be detected with autodiff > improve... > use regions.
Due to system memory issues we hardly ever use a gouge tolerance smaller than 0.1 mm (We also have large freeform parts)

NathanG · 2010-11-29 23:54:22

PolyFix will attempt to make a consistent surface by flipping triangles. Consistency Check will try to fix gaps and overlaps for AUTO-DIFF solid comparisons. It fails when it cannot resolve overlapped edges and close gaps. The larger the tolerance the more room Consistency Check has to make patches. Tighten the tolerances and Consistency Check has to conform more to the original model, where sometimes a satisfactory solution is not feasible. Start with water-tight models and things go smoothly almost without exception. Consistency Check can help to a point but when a model is bad enough it may require third-party software designed for fixing bad models, and extensive manual manipulation within that package to boot.

From the documentation:
---
Design Consistency Checkâ€”When selected during a "Solid" comparison, checks consistency of solid design models, including: check for "watertight" solid- repair improperly trimmed surfaces (overlaps and gaps), and reconstruct insignificant missing surfaces. This option is highly recommended for design models creating from importing IGES data, and other model files which produce unsatisfactory results from the comparison.

When selected during a "Profile" comparison, Design Consistency Check is applicable when building a profile from an STL, VERICUT Polygon, or VERICUT Solid design model. While checking design model consistency can take a while, it provides the best insurance for reliable AUTO-DIFF results. If you've used the model in VERICUT before and know the quality is good enough, you can save time by clearing this checkbox. AUTO-DIFF will take less time, but results may not be reliable if the solid is not water-tight.
---

Why does a cutting tolerance of 1mm catch almost everything? Because of your particular parts and their cut feature sizes. If you were making parts the size of a skyscrapper then that tolerance would take up to much memory. If you were making delicate and miniature medical parts you might not catch even half the features being cut. You mentioned pi for example. We can compute pi forever, so at some point we need to say that we have met our needs for precision. We could get progressively more accurate in calculating pi and feel really good about it, but have our system slow to a crawl in the process. That is why we have cutting tolerance!

paehv · 2010-11-30 08:25:36

Nathan,
Your right about part size and memory. It all depends on your parts and the system you use for verification.
We use Vericut now for about 12 years and for a long time we used a realy small cutting tolerance (mostly based on tool because memery wasn't availible for based on stock)
Simulations took fore-ever, sometimes the simulation took longer than the actual cutting on the machine.
Our CGTech support (UK) has convinced us to switch to a larger value by explaining that cutting tolerance is NOT equal to part tolerance. The introduction of autodiff over regions with improved tolerance was a big step foreward in vericut. Also the option "Constant gauge check" made live for us, programmers, much easier.

VERICUT Users' Forum

#1 2010-11-26 10:50:28

Consistency check for large parts

#2 2010-11-29 06:49:54

Re: Consistency check for large parts

#3 2010-11-29 07:38:41

Re: Consistency check for large parts

#4 2010-11-29 11:54:12

Re: Consistency check for large parts

#5 2010-11-29 23:54:22

Re: Consistency check for large parts

#6 2010-11-30 08:25:36

Re: Consistency check for large parts

Board footer