Absolute vs. Relative Part Accuracy

[ptc-1191590]

Can anyone provide some pro's and cons of switching from .0012 relative part accuracy to .0005 absolute accuracy? Experiences with the effects of doing so on merges, cut outs, assemblies, manufacturing models,and any other potentially troublesomeeffects would be great!

Fame and fortune awaits the most compelling response.

Thanks,

R.Swenson

This thread is inactive and closed by the PTC Community Management Team. If you would like to provide a reply and re-open this thread, please notify the moderator and reference the thread. You may also use "Start a topic" button to ask a new question. Please be sure to include what version of the PTC product you are using so another community member knowledgeable about your version may be able to assist.

[ptc-1932745]

Where do I start...



All of the tasks you list are affected. For example, we design parts
that are approx 1"x1"x1" to 8"x8"x16" for medical and aerospace
customers, and we set our accuracy to .0001 Absolute when working in
inch units in Wildfire 4 M040.



As far as I can tell, the only reason for playing with the accuracy was
to balance regen time against part 'fidelity'. As you can imagine with
such a large range in part envelopes, we were getting different results
for mold making, feature creation, even machining tool paths when the
part accuracy was set to .0012 relative. Machining 'accuracy' is
another, related issue as well, as is set to .0001 absolute.



Secondly, we do not use 'merge' and 'cutout' for making mold cavities
because these are 'solid' features and will either: 1) fail to complete
the feature, complaining about different accuracies, or 2) Try to
'match' the part accuracies when in TDO (Tool Design Option).



We use Insert | Shared Data | Copy Geometry... in part mode to do mold
cavities, etc. This is because the feature is a surface, and as AFAIK
seems to be 'immune' to the mismatched accuracy gotcha. To cutout the
cavity using the Copy Geometry... feature, select the quilt and Edit |
Solidify... in part mode. Now you have created an external
dependency when doing this, the implications with regards to Interlink,
et.al... I cannot comment on, we don't use it.



Now to do the copy geometry function, you need to make sure that you
have the proper licenses, options and extensions. We have been a long
time user of Pro-E and as such our old licenses rolled over to
'Foundation Advantage' licenses. Because of this, we need TDO, Pro
Assembly (NOT Pro-Assembly-Performance), or Pro Manufacturing floating
options enabled before we have the Copy Geometry... command available.



Again, AFAIK, 'regular' Foundation licenses do not have this limitation.



If you would like more detailed examples, and/or test cases, maybe I can
help...



Christopher Gosnell

TRIGON INC.
FPD Company
124 Hidden Valley Road
McMurray, PA 15317

[Greg_McMurray]

Here's a link to a good discussion on the subject:

http://groups.google.com/group/comp.cad.pro-engineer/browse_thread/thread/7ac7b385bb720124?hl=en&q=jeff+howard+%22accuracy%22

In Reply to Christopher Gosnell:

Where do I start...



All of the tasks you list are affected. For example, we design parts
that are approx 1"x1"x1" to 8"x8"x16" for medical and aerospace
customers, and we set our accuracy to .0001 Absolute when working in
inch units in Wildfire 4 M040.



As far as I can tell, the only reason for playing with the accuracy was
to balance regen time against part 'fidelity'. As you can imagine with
such a large range in part envelopes, we were getting different results
for mold making, feature creation, even machining tool paths when the
part accuracy was set to .0012 relative. Machining 'accuracy' is
another, related issue as well, as is set to .0001 absolute.



Secondly, we do not use 'merge' and 'cutout' for making mold cavities
because these are 'solid' features and will either: 1) fail to complete
the feature, complaining about different accuracies, or 2) Try to
'match' the part accuracies when in TDO (Tool Design Option).



We use Insert | Shared Data | Copy Geometry... in part mode to do mold
cavities, etc. This is because the feature is a surface, and as AFAIK
seems to be 'immune' to the mismatched accuracy gotcha. To cutout the
cavity using the Copy Geometry... feature, select the quilt and Edit |
Solidify... in part mode. Now you have created an external
dependency when doing this, the implications with regards to Interlink,
et.al... I cannot comment on, we don't use it.



Now to do the copy geometry function, you need to make sure that you
have the proper licenses, options and extensions. We have been a long
time user of Pro-E and as such our old licenses rolled over to
'Foundation Advantage' licenses. Because of this, we need TDO, Pro
Assembly (NOT Pro-Assembly-Performance), or Pro Manufacturing floating
options enabled before we have the Copy Geometry... command available.



Again, AFAIK, 'regular' Foundation licenses do not have this limitation.



If you would like more detailed examples, and/or test cases, maybe I can
help...



Christopher Gosnell

TRIGON INC.
FPD Company
124 Hidden Valley Road
McMurray, PA 15317

[dgschaefer]

Here's what I've understood about the difference between relative &
absolute.

Relative accuracy is the ratio between the shortest edge and longest
edge in the part. Reading the link that Greg sent, it seems that's a
simplification of the actual formula, but it get's the point across. As
the part changes, relative accuracy is recalculated at each step along
the way.

Absolute accuracy defines the absolute smallest edge allowed. It does
not change as the part changes.

Now for my experience in using them.

The default is relative and I believe that he theory was that a large
part can't be made with small details, or at least small details
wouldn't be desirable. Creating small details in a large part is costly
in manufacturing and in CPU time, so limiting detail size by part size
therefore seems reasonable and will speed model regen time based on the
part size.

In reality, we are now making more and more very large injection molded
parts which can easily and cost effectively have very small details. In
my experience, I've designed myself into a corner on parts with relative
accuracy where there has been a late design change that alters the part
size, even slightly, so that now features that had regen'ed fine with no
geom checks now simply won't work.

For this reason (and because I started my Pro|E life doing mold design),
I've used absolute accuracy exclusively, accept when driven by customer
requirements, for about 12 years. The only issue I've seen with
absolute accuracy was where it was set very tight (accidently setting a
mm part to an accuracy of 0.0001) and regen times increased
dramatically. Using a value of 0.0005 (we use 0.0001) should be fine,
and I'd absolutely make the switch.

Doug Schaefer

[ptc-1932745]

Thanks very much for the link, the discussion was very informative as it
discusses the implications of creating geometry a far distance from the
default datums. Aerospace parts, they are typically created using a
local coordinate system and then are placed in an assembly with
reference to the aircraft coordinate system. In CATIA a CATPART, is
placed in a CATPRODUCT.



The other point that most people on this forum may be already aware of,
is that the smallest edge 'accuracy' never gets finer, only coarser. So
if you start with a big block and then end up with a small final part,
features may not regen properly using relative accuracy.







Christopher Gosnell

TRIGON INC.
FPD Company
124 Hidden Valley Road
McMurray, PA 15317
PH: 724.941.5540
FX: 724.941.8322
www.fpdinc.com

[mlocascio]

This is a good forum....absolutely.

Michael P. Locascio

[ptc-1191590]

Here are the responses I received regarding relative vs. absolute part accuracy, names removed to protect the innocent. Thanks to everybody for the valuable feedback!

In general you should experience better results once your parts are maintained at a consistent absolute accuracy. Now the value of the accuracy is dependent on your units and the typical size of your products. For example, I typically run at .01mm and .004in. However, if your products are relatively small (heart defib) you may want to run it slightly tighter .005mm as an example.
I could go on and on but hopefully that helps,
---------------------------------------------------------------------------------------------
Your downstream manufacturing guys will love you in an absolute accuracy setting. Trust me. I have had great experiences with this. Call 319 444---- if so desired and I will be glad to discuss this with you.
---------------------------------------------------------------------------------------------
This discussion is much simpler than it has been in the past. The bottom line;if you have a value for absolute accuracy that meets the needs of your company, then switch your start parts to absolute.

Even PTC is going to change their template start parts to absolute accuracy, reportedly in WF5.

We were very hesitant to go with absolute accuracy because of some unanswered questions, but were forced to use it for our complex castings (due to failure of shared data between different models when using relative accuracy). Since then, absolute accuracy has proven itself as being the better choice.

Having said that, here are some possible concerns with absolute accuracy:
- if your parts vary greatly in size, it can be hard to find a single value that works for all your needs (we use 0.01 mm, or about 0.0004 inch, but it is not sufficient for some of our tiny parts/features)
- Pro/E can behave differently when calculating feature intersections/etc. (e.g., a round that might be created successfully with relative accuracy, might not be successful with absolute accuracy) - - this doesn't happen often for us though
- if somebody switches the units of a part that uses absolute accuracy, and selects the option to "Convert" units (where 1inch becomes 1mm, rather than 25.4mm), then the part accuracy value will not get changed automatically (e.g., if it was set to 0.01mm, it will convert to 0.01inch). This will require a manual change (more likely it will go uncorrected, and cause problems later).

---------------------------------------------------------------------------------------------

I assume your question applies to starting a new part with absolute accuracy vs. converting an exist relative accuracy part to absolute accuracy.

Many of our clients producing plastic parts, especially automotive, request that we have absolute accuracy set to .0003 (some) or .0004 (others). .0004 inches is about 1/100th mm. Some other solid modelers use this as their fixed default accuracy so import/export issues are reduced.
---------------------------------------------------------------------------------------------
Not to get into the actual value of the accuracy, but for me, ANY absolute value is better than the best relative one.
At least with the absolute, you know EXACTLY what your accuracy is without wondering what the largest dimension of the part is.
As to values, choose one within your manufacturing tolerances and you should be good.
---------------------------------------------------------------------------------------------
WF5 or 6 is slated for absolute accuracy by default.
---------------------------------------------------------------------------------------------
Where do I start…?
All of the tasks you list are affected. For example, we design parts that are approx 1”x1”x1” to 8”x8”x16” for medical and aerospace customers, and we set our accuracy to .0001 Absolute when working in inch units in Wildfire 4 M040.
As far as I can tell, the only reason for playing with the accuracy was to balance regen time against part ‘fidelity’. As you can imagine with such a large range in part envelopes, we were getting different results for mold making, feature creation, even machining tool paths when the part accuracy was set to .0012 relative. Machining ‘accuracy’ is another, related issue as well, as is set to .0001 absolute.
Secondly, we do not use ‘merge’ and ‘cutout’ for making mold cavities because these are ‘solid’ features and will either: 1) fail to complete the feature, complaining about different accuracies, or 2) Try to ‘match’ the part accuracies when in TDO (Tool Design Option).
We use Insert | Shared Data | Copy Geometry… in part mode to do mold cavities, etc. This is because the feature is a surface, and as AFAIK seems to be ‘immune’ to the mismatched accuracy gotcha. To cutout the cavity using the Copy Geometry… feature, select the quilt and Edit | Solidify… in part mode. Now you have created an external dependency when doing this,the implications with regards to Interlink, et.al… I cannot comment on, we don’t use it.
Now to do the copy geometry function, you need to make sure that you have the proper licenses, options and extensions. We have been a long time user of Pro-E and as such our old licenses rolled over to ‘Foundation Advantage’ licenses. Because of this, we need TDO, Pro Assembly (NOT Pro-Assembly-Performance), or Pro Manufacturing floating options enabled before we have the Copy Geometry… command available.
Again, AFAIK, ‘regular’ Foundation licenses do not have this limitation.
If you would like more detailed examples, and/or test cases, maybe I can help…
---------------------------------------------------------------------------------------------
Here's a link to a good discussion on the subject:
http://groups.google.com/group/comp.cad.pro-engineer/browse_thread/thread/7ac7b385bb720124?hl=en&q=jeff+howard+%22accuracy%22
---------------------------------------------------------------------------------------------
Thanks very much for the link; the discussion was very informative as it discusses the implications of creating geometry a far distance from the default datums. Aerospace parts, they are typically created using a local coordinate system and then are placed in an assembly with reference to the aircraft coordinate system. In CATIA a CATPART, is placed in a CATPRODUCT.
The other point that most people on this forum may be already aware of, is that the smallest edge ‘accuracy’ never gets finer, only coarser. So if you start with a big block and then end up with a small final part, features may not regen properly using relative accuracy.
---------------------------------------------------------------------------------------------
Here's what I've understood about the difference between relative & absolute.

Relative accuracy is the ratio between the shortest edge and longest edge in the part. Reading the link that Greg sent, it seems that's a simplification of the actual formula, but it get's the point across. As the part changes, relative accuracy is recalculated at each step along the way.

Absolute accuracy defines the absolute smallest edge allowed. It does not change as the part changes.

Now for my experience in using them.

The default is relative and I believe that the theory was that a large part can't be made with small details,or at least small details wouldn't be desirable. Creating small details in a large part is costly in manufacturing and in CPU time, so limiting detail size by part size therefore seems reasonable and will speed model regen time based on the part size.

In reality, we are now making more and more very large injection molded parts which can easily and cost effectively have very small details. In my experience, I've designed myself into a corner on parts with relative accuracy where there has been a late design change that alters the part size, even slightly, so that now features that had regen'ed fine with no geom checks now simply won't work.

For this reason (and because I started my Pro|E life doing mold design), I've used absolute accuracy exclusively, accept when driven by customer requirements, for about 12 years. The only issue I've seen with absolute accuracy was where it was set very tight (accidently setting a mm part to an accuracy of 0.0001)and regen times increased dramatically. Using a value of 0.0005 (we use 0.0001) should be fine, and I'd absolutely make the switch.
---------------------------------------------------------------------------------------------
If you put out a summary, could you please include the following best practice:

In the recent discussions/references, I haven't seen mention of a best practice regarding Relative Accuracy, which is: as the first created geometry, make a "sizing feature" that sets the working envelope that will contain your part. This could be a surface/curve that represents the bounding box, for example. The reason for doing this is to avoid the "effective" accuracy changing as the size of your model changes, due to use of relative accuracy.

Would still suggest using Absolute Accuracy if possible, but the above can minimize trouble for those using Relative.
---------------------------------------------------------------------------------------------
The absolute/relative issue really depends on what your goal is and what size your parts are.
SIZE: The accuracy determines what size 'anomaly' Pro/E will ignore. Relative accuracy of .0012 on a 20" part means Pro will not see an error unless it is .024" or greater. If you need better accuracy of your final part (say you need +/-.010") you need to either change the relative accuracy, or go with absolute accuracy and set it to less than .010"
GOAL: The place where absolute accuracy is really necessary is when you are interfacing between multiple parts as in Merges, Cutouts, Master Merge or Master models, etc. Anywhere that one part is going to rely on the data from another part, it is best to set both parts to the same absolute accuracy. That avoids failures due to differences in accuracy. I use this the most when I'm using a master surface model to control the outer shape of multiple plastic parts that make up an assembly.
One area where I had trouble with relative accuracy (long ago in pre-WF days) was when I reordered a feature in a model and it caused later features to fail because it changed the relative accuracy of the feature that USED to be before it and worked, now it is after and fails. I don't know if WF has fixed that.
---------------------------------------------------------------------------------------------

[CM15]

If you have very simple parts and features, and they will stay that way, relative accuracy is fine. If you have, say, complicated castings or IM plastic parts and use variable-section sweeps, swept blends, surfacing, or any of the other advanced modeling techniques, set your accuracy to the finst absolute accuracy (AA) available BEFORE you start making geometry. I've had plenty of instance where geometry has failed, I left it as an incomplete feature, set accuracy to the finest absolute setting, regenerated the feature and it worked. Also, as mentioned, with merge and cutout features the same absolute accuracy is basically required for the relevant parts. I've also seen problems with geometry created before and after setting absolute accuracy. If in doubt, setAA to finest before any geometry creation.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.

Dr. Seuss