Establishing Centerline Datum of Impeller Casting?

[Quince]

I have an impeller I'm having cast and then machined. The "as cast" drawing is where I need to establish a Center-line Datum which will be used to guide the machining of the hubs (one each at nose and tail of impeller).

Outside Diameter Measurements of the Flutes are taken at two different locations on the impeller (near nose, and near tail) to find the common axis that runs through their theoretical center. This is how I want the machinist to "Establish" the "True Center" of the impeller since the hubs as cast in the raw, cannot be relied upon to center the flutes when spun.

If it were a simple cylinder, I'd establish the Datum "-A-" by using the one outside surface (admittedly not necessarily cylindrical), and apply a tolerance to the Datum Feature. In this case, I need to show two cross sections, one each of the Flutes at the nose, and one at the tail since their respective nominal O.D.s are different, and "Land" at different positions as viewed from Nose or Tail (Flutes are spiraled).

How do I identify the two cross sections as being related to the desired Centerline Datum?

My GD&T books don't address this very well.

Thanks

B

 

[Quince]

Would it be incorrect to:

1) dimension the outside diamters of the two cross sections (one at each end of the impeller), and label each of those O.D.s "Datum A", and "Datum B" respectively.

2) Draw a centerline on the impeller which represents the line intersecting the two points described by the outside diameters. Call this centerline "Datum Axis A-B".

3) Describe surfaces of this impeller as needing to be concentric to "A-B".

 

[KENAT]

Assuming ASME Y14.5M-1994 take a look at fig4-21 & section 4.5.7.2. Single Axis of Two Coaxial Features.

I don't think it has all the answers but might help you a little. Specifically I don't think you'd explicitly call the centerline "Datum Axis A-B". However, in FCF relating to the axis you'd put [A-B] and invoke the equivalent.

Also while this may be one of the exceptions, concentric often isn't the appropriate control. Consider if runout may be more applicable.

However, this is a bit beyond me but hopefully one of the real sharp guys will see your question and give a more complete answer.

KENAT,

Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:

http://eng-tips.com/market.cfm?

What is Engineering anyway: faq1088-1484

 

[Quince]

I follow you, yes, that example is closest to my situation.

Will be interested to see if anyone might have another view on it, but I'm otherwise liking this method.

Thanks

B

 

[fsincox]

Quince, I see you do cast impellers, too. May I inquire what type of Industry?

 

[drawoh]

Quince,

Consider casting in place some datum targets. Locating centres is meaningful as per ASME Y14.5M-1994 and KENAT's suggestion, above, but I do not know how repeatable it is.

If you search on the internet, you should find some casting manuals that recommend how to do datum targets.

JHG

 

[Quince]

Hi All,

I discussed the idea of Datum Target in the casting with the machinist I'm working with. The geometry doesn't allow any obvious locations to do this, but we think the "Root" of each Flute, or the central "Hub" of the part may be the best reference points. His idea was to control the Concentricity of all elements of the part by establishing a Datum Axis, using two control diameters at specified locations from the nose of the part.

I'll continue my research based on suggestions posted here, and I thank you all.

Fsincox, these impellers are used for Mixing.

 

[SDETERS]

That is one cool Model. May I ask what Software it was modeled in. Sorry for the off topic but it is a cool model.

 

[Quince]

"SDETERS (Agricultural)
19 Jun 09 15:51
That is one cool Model. May I ask what Software it was modeled in. Sorry for the off topic but it is a cool model."

No problem, it's modeled in Solidworks, although I can't take the credit, I've "inherited" it from one of our Engineers who isn't with us anymore. It's definitely not easy to model.