【转帖】forcing dimensions - how you do i

yang686526

forcing dimensions - how you do it?
so recently a plastic part we designed was measured and statistics etc about the manufactured parts were gathered.
they suggested we should change our drawings and in-turn models to reflect they're manufacturing capabilities.  we have been discussing weather our models should be updated to reflect the manufacturers inability to generate "propper" parts, or if we should just alter the drawings with specific notes indicating why the demensions have been "forced" to reflect the manufacture capabilities (ie they're tooling was probably created impropperly).
anyhow, a mistake has been made, thats not what my question relates to.  what i am asking is do we reflect the mistake in the model?  or just the drawing?  if just the drawing, how would you indicate the dimension is "forced" and varies from the design intent?
i was thinking coloring the dimension blue rather then black with a note.  this way printed drawings won't "see" it but when looked at electronically (internally) we see what is ... different.
if you are going to accept these parts as they were made, i would recommend changing the model.  if this is a temporary situation, i would note the differences on the documentation accompanying the part, but not on the drawing, which should reflect design intent.
i would not blindly overwrite dimensions on a drawing.  the drawing dimensions should match the model dimensions.
if only a few dimensions are off, perhaps add a note to the dimensions to show "as designed" state versus "as manufactured".
ideally, you should change the model to match what you are manufacturing.
more ideally, you could flog the molder until the parts are completely right.  this rarely works.
apparently we have accepted the parts.  what was mentioned in the meeting yesterday, is what happens 3 years down the road when we find a new manufacturer.  do we give them the "fumbled" model or do we give them the design intent?
hense the indecision about how to proceed.  having a design intent model and a "real world" model seems kind of too much.
are the parts acceptable, or have they simply been accepted?
if you do change suppliers down the road, do you want the new supplier to make the same part that the current supplier makes, or do you want what you really want the first supplier to make?
at any rate, never disassociate the dimensions from the model.
if the parts are acceptable, and will always be acceptable i suggest increasing the tolerance of the dimensions to cover this.  
i've seen this done on castings a number of times due to tool wear.  this way you don't have to change the model.  bad thing is you may no longer have bilateral tolerances around a nominal which is usually preferred.  
the whole point of tolerances is to accommodate acceptable manufacturing variation, i would have thought this was an example of that, it’s a variation from the nominal that has been deemed acceptable.  while you're at it look at any other tolerances which are unnecessarily tight and adjust those.  ideally tolerancing should reflect function of the part, not some vague idea that it needs to be tight in one area etc.  unfortunately i certainly don't spend as much time on this as i perhaps should or could so many of my tolerances end up a best guess (not on interferences, positional etc but general tols)
i would not force the drawing.  this can lead to all sorts of confusion in the future and even unnoticed problems from a geometry point of view.  i almost certainly wouldn’t add a note on the face of the drawing no matter what you do, though a note somewhere in the file properties can be useful.  if explanation is needed it should be on the eco or equivalent.
if you decide to actually change the dimensions rather than simply modify the tolerance i'd change the model.
i support the tick in his appeal to flog the molder.  that is the ideal situation.
depending on how much the dimensions are off, you may want to flag the off-dimensions, then create a general note on the drawing stating the "molded dims".  or, you could flag the off-dims and use a general note that applies a different tolerance range for those particular dims.
or you can sign-off on a temp waiver and hope the vendor makes them right the second time, which leads to the questions how these passed first artical inspection (if it exists at your company) in the fist place.
art without engineering is dreaming; engineering without art is calculating.
this doesn't answer your question about giving future suppliers which model, but since we are talking years, you could revise the model and drawing to reflect what you are receiving now (if they will work in your application) and when you change suppliers in the future, resurrect the original model and drawing with a new p/n or revision level.  most companies keep this type of data archived, but if yours doesn't, take steps to ensure the correct files will be available in the futureby burning them to cd or dvd and giving a copy to your drawing and data control dept.  
if the vendor didn't make per print ... i would never change a drawing to match parts made wrong. i would ship them back and the vendor eat the cost.
if the part was made per print, but is wrong ... eat the cost, fix the print, send part back to fix or remake, then go on.
chris
solidworks 06 5.1/pdmworks 06
autocad 06
there is no reason to disassociate the dimensions. what happens if you do that and in three years you send the model to a mold shop that build their molds based on the model and not the drawing? i agree with the others, if the parts are "right" and the drawing needs to be changed, change the drawing and the model or increase the tolerance to include what is acceptable. if the parts are wrong, the drawing is right and you are accepting bad parts, stop, send the back and tell them to build them to print.
altering dimensions on a drawing so that they don't reflect the model is a sure fire way to get your a$$ bit later on.
david
once again manufacturing has tried to create tools that manufacture parts that meet drawing intent.  after measuring a number of parts, it is found that the parts have a large mean shift from nominal.  but the variation about that mean is low.
so what to do?
you can accept the out of spec parts only after assembling and testing to ensure that they function within customer requirements.
if everything fits and functions with all the different out of spec parts, then document the difference.  the drawing is not the place for this unless you want to change you verified and validated design intent.  work with purchasing or supplier quality on how to document the difference.  in automotive world we do this to approve with a deviation for the out of spec parts for ppap and after.  
the bottom line is that you found one design solution and manufacturing found another.
how do you prove which is the better solution given the budget and schedule limits?
did you perform all the design analysis to verify your design intent?
did you evaluate alternate designs looking for design improvements?
are you confident that your design is right given the millions of dollars at stake?
the tools are made, and production starts very soon.
you have the choice to change tools or test the manufactured product.
tom rhodes, gdtp-s