modeling basic vs meadian value

[dtharrett]

Lately many of our tool shops have been cutting to the model. As a result, the tool shops have complained about modeling inconsistencies. Some of us are modeling things like metric dowels, slip fit holes and press fit holes to the basic dimension while others are modeling to the meadian value.

The problem is two fold. First the tool shops need to look at the toleranced drawings. Secondly, we need to set a standard in our department (machine design standards) with regard to modeling basic vs. median. Is there a generally accepted industry standard that we can adopt? Below is an exaple of the two senarios:

Basic method:
12mm dowel modeled at 12mm, press fit hole modeled at 12mm, slip fit hole modeled at 12mm. The drawing would call out the press fit hole as 12 M7 and the slip fit hole as 12 E7.

Median method: 12mm dowel modeled at median value of 12.0125, press fit hole modeled at 11.991, slip fit hole modeled at 12.041. The drawing would call out the press fit hole as 12 M7 and the slip fit hole as 12 E7.

Thanks in advance.

 

[ctopher]

I have also seen this problem. The software doesn't know the difference.
One way around it is to not model the holes for CNC work, only a small hole for location. Let the machinist finish them per the drawing.
Or, give them the model with the holes, and the complete drawing, then tell them to change the CNC file to make the holes per the drawing.
Some CNC programmers don't know how to manipulate the programs, they just machine per what the software spits out.

Chris
SolidWorks 10 SP5.0
ctopher's home
SolidWorks Legion

 

[kellnerp]

The shops I have worked with have always checked the geometry before cutting. Models sent to the shop need to be specifically setup for machining based on the shop's practices and probably should be dumb solids accompanied by a released drawing with tolerances and such. Letting the designer run the CNC by proxy just isn't cool. The shop is trying to cut corners. Models sent to FEA need to be setup for FEA.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

http://www.linkedin.com/in/engineeringtransport

Phenom IIx6 1100T = 8GB = FX1400 = XP64SP2 = SW2009SP3
"Node news is good news."

 

[TheTick]

What about moving on to making a mold? Then the model would need to be steel-safe, e.g. Hole diameters at large side of limits.

 

[drawoh]

dtharrett,

I model to nominal size. This seems to work in our machine and sheet metal shops. They work from DXF files. They get PDFs as well, if I have any say in the matter. We have done no machining from the model. The process seems to work. It supports design in that the nominal size is modeled separately from the tolerances.

I have done one set of rapid prototypes, and things got weird. I don't think they looked at my drawings. If I do any more rapid prototyping, I will model to median size.

An alternate approach would be to model to MMC (MMB).

Perhaps there should be a note on each drawing stating how the modeling was done.

Your inspector, probably, is working from your 2D[ ]drawings.

JHG

 

[CorBlimeyLimey]

I use nominal model sizes but supply the machine shop with parasolids and PDF's. The parasolids eliminates the donkeywork for the machinists (or programmer), but the PDF's are the "fine print" for the contract of work.

To avoid confusion, if you are going to use median sizes, then dimension limits should be used instead of type of fit, or bilateral or even symmetric tolerances.

 

[kellnerp]

It comes down to having a well documented procedure that is documented on the drawing like a note that says: Solid Model provided at nominal dimensions. Hole and Pin tolerances as called out on drawing.
OR
Interpret Solid model per Engineering Standard <give number here>.

The drawing checker will be responsible to make sure this happens before handing off to the shop.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

http://www.linkedin.com/in/engineeringtransport

Phenom IIx6 1100T = 8GB = FX1400 = XP64SP2 = SW2009SP3
"Node news is good news."

 

[dtharrett]

We are going to have a conversation with the tool shops re-iterating the need to verify against the supplied drawing. On the design side, we need to standardize (basic or median). At the moment, the majority of us are modeling basic so I am leaning toward basic for our standard. I would feel better if there was an ISO or ANSI standard that agreed with this decision. Any thoughts?

 

[ajack1]

This topic has been debated many times and I am not sure there is a one size fits all solution.

I am not sure exactly how Solidworks handles these things but in the software I use it is easy to put in a 12mm dowel and holes for example and create the fit in 2D drawings M7 for example, the downside is it is 12mm if you make it 11.991 you lose this option. So both have pluses and minuses.

Once you get away from simple shapes and holes it becomes far more complicated. The example of rapid prototyping as mentioned above being a good example. Most rapid prototyping machines require a .stl file and they are not very editable so take a great deal of time and effort and therefore cost, so in my opinion at least it is far easier to model what you want and simply let them click print, well sort of.

Would anyone dream of sending a pdf file of a flyer to a print company with notes all over it saying 5mm font needs to be 6mm etc? Surely you send them what you want them to print?

As the tick mentioned it goes to another level again when you start looking at moulds, stamping dies, forging etc with complicated surfacing on them. Do you model everything to mean or nominal? If you just go to nominal someone has to redo the whole thing again at a huge cost.

My personal opinion and I know many will disagree with this, is you should aim to set up your own standards with the one aim of making the process of the raw materials coming in to the finish product going out as quick, cheap and reliable as possible with all relevant parties involved in the decision making.

 

[drawoh]

ajack1,

I have send PDF copies of flyers to printers, and they have screwed them up. If you are printing Z-folded flyers, the PDF file must be printed at 1:1 scale. This is the exact problem we have. Our computer data must be interpreted exactly as we say it should be.

JHG

 

[ShaggyPE]

I agree with the point kellnerp brought up at the top. Make it clear that the shop is responsible for "tweaking" the model so that the finished part matches the drawing requirements.

Alternatively, if you are working closely with the shop (or internal shop), you may be able to make the process more efficient in the manner ajack1 brought up... but that puts the responsibility on good parts on you... this should come with a good cost break as well.

In general, when fits are concerned, I model to the median value. i.e. the model has clearance/interference as appropriate... not line on line.

-Dustin
Professional Engineer
Pretty good with SolidWorks

 

[ShaggyPE]

Not sure if ASME Y14.41-2003 addresses this issue... but it is worth a look.

-Dustin
Professional Engineer
Pretty good with SolidWorks

 

[dtharrett]

ShaggyPE,
I read through Y14.41 a few times. I does state "parts should be modeled at a specified dimensional state". My interpretation is that a group or designer or company should specify a standard like basic or mean modeling and stick with it. I guess what ever makes the most sense. Just not sure which way to go at the moment...

 

[rollupswx]

I have done one set of rapid prototypes, and things got weird. I don't think they looked at my drawings.

I am quite sure the DIDN'T look at your drawings. 2D is irrelevant for RP, unless you expected them to remodel your part with considerations for the tolerances of the process. RP is not a machining operation - it is not close tolerance at all.

When we make RP parts we add additional clearance to the solid model knowing that the process is not particularly precise. The orientation of the model build layers is also important.

If you don't have in-house RP capability to work out some best practices you will have to communicate in-depth with your service. Far more variability than traditional machined parts from dimensioned and toleranced drawings.

 

[RoscoeCoach]

I just read the following in ASME Y14.5-2009 ..."The general opinion is that CAD modeling should be done to the mean." Later it states "The CAD model should be drawn to the mean and BASIC dimensions. Where plus and minus tolerancing is used, the toelrance should be bilateral; not unilateral. Equal plus and minus values are perferred."

Like many of you we constantly debate this subject at our company. The to point were many drafters, designers and manufacturing engineers are mixing up the definitions of Nominal, Mean and Basic. They believe they are always one in the same. They can be at times, or not at other times all depending on the tolerance that is applied.

When doing engineering work and design I prefer to model to the nominal because it is quick and easy and I can easily change the tolerance with out haveing to deal with tiny changes to the model. However our manufacturing group always perfers the model to be at the mean. That is what they want to shoot for when machinging and it saves them time in dealing with changing the models.

I have always been for modeling to nominal values, but in practice with programing from models and 3D printers and CMM programs from models we have no choice but to stay with modeling to the mean. Until CAD systems can give us easy solutions to quickly update the models to mean reading from nominal annotations we really have no choice.

So when I am developing new layouts I design to the nominal. As I approach the final design I start to add notes to state the tolerance I would like. When it moves to production prints and models the model will be updated to be at the mean. There is some risk and effort to do this but either we do it in engineering and control the end result, or we put the burden on the manufacturing and programers to update the model so they can develop code. Either way the work is still there, so I might as well control it within engineering.

It's a pain to model to the mean but until CAD programs give us better solutions we have no choice.