Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations IFRs on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

internal precision, 11 digits. ??

Status
Not open for further replies.

skanskan

Civil/Environmental
Jul 29, 2007
278
Hello

Is it true that Catia uses 19 digit precision and NX only 11?

I understand that it could be important for CAE simulations but,
Does it really matter for surfaces and bodies?
 
Replies continue below

Recommended for you

Sorry, whoever made that statement was wrong, at least with respect to NX (I can't speak for what Dassault has done).

Computations are done in NX at 16 significant figures and then rounded off two places. You can verify this by going to...

Preferences -> User Interface -> General

...and toggled ON the option labeled 'System Precision in Information Window'.

Now create a circle of any size, go to...

Information -> Object...

...and select the circle and hit OK. When the listing window comes up, go down to the displayed value for 'Arc Length' and check the number of significant figures that you see. It should be 15 total, counting both sides of the decimal place, with the last digit always being zero (remember I stated that the last two 'computed' digits are rounded off), so you have 14 figures that you can use.

Now you allude to the fact that even if it were only 11 places that it would have been adequate for most work, and while that may be true, certainly when you consider most of the tolerance that one can realize with modern machining and fabrication technologies (and remember, that's what prescison is all about, getting products which go together how you want them to and still perform as you expected them to). Another way to think about what this really means is that this level of precision allows you to create a model which will fit inside a cube 1,000 Meters (One Kilometer) on a side with the 0,0,0 origin at the center of the cube. Any geometric computation made with respect to any object inside this 'modeling space' will be correct and will produce a result which will support whatever level of precision you've specified for an operations, be it model, analysis or manufacturing.

One last comment, the reality is that this level of accuracy may NOT play as big a role in CAE simulations as you may at first think since virtually all analysis applications, at least those which utilize finite elements methods, are actually doing their 'math' at some level of approximation much lower than the maximum precision of the system which these computations are being run on. Granted, as hardware becomes faster and memory address space has now been expanded out to unheard of capacities, more complex and accurate computations can be made, but even then one must understand that all simulations are only providing an 'idealised result' with respect to how something will actually behave in nature. Therefore, there's a point at which any attempt at additional levels of precision will only mislead you into thinking that this will actually demonstrate any better what will likily happen in the real world. Trust me, the world itself is not as precise as Newton would have liked us to believe, particularly once Einstein declared that where you observed something was as critical as to what it was you were observing and then Heisenberg warned us that even if you did managed to measure something accurately, that this act alone will have changed the result you just observed. So I think it's safe to say that 14 significant figures will probably be good enough for anything any one of us are going to need to worry about ;-)

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Industry Sector
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.
 
As I said before, I can't speak to the Dassault comments or claims, but the accuracy of the statements about NX are wrong on many issues in addition to the claim that NX can only compute to 11 significant figures. In at least two places he makes incorrect claims about the origin of the NX (it was not invented by EDS) and who Siemens acquired the company from (again, it was not EDS). NX is a product whose 'DNA' goes back nearly 40 years to a Company named United Computing located in Carson, CA, a few miles south of Los Angeles. Eventually the company was acquired by MDC (McDonnell Douglas Corp) in 1979 which was later sold to EDS in 1991. The product was originally named Unigraphics and has evolved into NX which is the designation chosen to represent the changes in the product as we began an era when the best of Ideas (SDRC was acquired by UGS, then a division of EDS in 2001) was incorporated into the Unigraphics architecture. A few years after the SDRC acquisition and the launch of NX, the company was sold by EDS to a group of venture capital firms which held onto the company until it was finally purchased by Siemens in 2007.

To give you some insite into the history of the Unigraphics product, at least up to the SDRC acquisition, please go to:


Where you can also find some background the history of SDRC.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Industry Sector
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor