Volume of a solid body

[NXsupport]

NX4..

I have been observing when the geometry of the solid body is built from a complex splines and shapes, the volume value changes..

for example, if the volume for body when it is one single solid is xxx.xxxmm^3, when you start splitting this body into multiple pieces, the consolidated volume starts to vary..

Is this a bug in volume calculation in NX?

 

[JohnRBaker]

Can you provide an example part which exhibits this behavior?

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[hudson888]

Does it really vary by much? We have always found left and right handed linked mirrored parts would wary slightly in volume. I believe that the system basically divides the body into numbers of cubes or like prisms in order to calculate how many fit into the swept volume. The stated accuracy is I think 99% but that is per solid so if you split a single body four ways worst case scenario you might get 5% variation if the original was 1% on the high side and the others all fluked 1% each on the low side for example. More than that seems like another kind of problem may be at work.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[JohnRBaker]

That's correct, the volume is computed using integral methods, calculating the volume under a surface so it's an approximation for anything that's more complicated than a block.

After the initial post above, I decided to test this in NX 6.0.4.3 using a very large and complex model of a casting (413 features, 901 faces, 2012 edges and 3.8 meters long x 2.4 meters wide x 0.7 meters high). I first measured the volume of the original model, which I saved. I then performed a 'Split Body' operation, cutting the model basically in half, and measured the volume of each half and then added these two values together and compared it to the original volume, and the difference was LESS than 0.01%. Close enough for government work ;-)

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[NXsupport]

Yes Hudson, it varies quite significantly approx 13% in one case.. This happens when the body is constructed thru a mesh of complex B surfaces..We checked for geometric irregularities , but the model passed all the examine geometry checks..

The volume behavior for a body built from a complex B-surface mesh in NX6 is also same.. there is significant variation in the volume value

 

[JohnRBaker]

Again, can you provide an example which exhibits this behavior so that we could have someone look at it?

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[hudson888]

Worth sending a copy of the model to GTAC so we can all rest assured that volume calculations are bullet proof.

Personally I had cause rather recently to design a product of quite irregular shape which itself needs to be suspended hanging freely from a cable. Using measure body tools I optimised the design for balance and sited a hole on the centre of gravity. We then made accurate prototypes that hung from that string with the top face so flat an level that it would hold a bead of water steady. We all trust the system to be accurate and intellectually accept it, but is sure is fun to have a working example.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[NXJockey]

If you are getting different values between machines or users, check the tolerance which the weight calculation is being performed to.

I have had this happen to me on several occasions, with the hindsight of experience, we now have the measurement accuracy locked at a site level.

0.999 seems to give a good trade off against accuracy and performance for our components which contain free form surfaces.

On a side note, we have seen the weight calculation become more accurate as we have migrated from v16 > v18 > NX2 > NX4 (all though some users oddly believe that because the new value is different, it must be wrong!!! Users, who’d have them…)