datum system on complicated part

[bxbzq]

I've realized the most difficult part of applying GD&T is datum system establishment.

I'm working on a part that is huge in size and has thousands of features on it. It is an air flow inlet part. The part has outer housing and inner housing. The air flows between outer housing and inner housing. The part assembles with a cylinder to its right, a bearing inside the inner housing. There are two Supports to support the part. A rotor shaft goes in bearing and on the part there are 3 grooves for 3 rings which are coaxial with rotor shaft. Detailed description please see attached sketch.

Currently in our drawing there are just 2 datum features are specified which are horizontal and vertical flange. I think there is room to improve the existing datum system of the part.
Comments, suggestions, ideas are appreciated.

 

[bxbzq]

hmm, is there anything wrong with my description? I thought my question would be interesting. Looks not true.

 

[CheckerHater]

You are the boss.
You are the only one who knows how the part works.
You are the one who (possibly) knows how the part will be manufactured.
You are the one to decide how many datums to have. Sometimes it may be possible to produce good part based on general tolerances all alone.
Sorry. Still feel like Monday.

 

[pmarc]

bxbzq,
It is not that your question is not interesting - it is simply too general. I am pretty sure that most of us here would really like to help you, as this is real-world application, however without knowing more it is really difficult, if possible, to provide any useful info.

I personally think your assembly needs a datum axis established from one of cylindrical features, but that is all I am able to help at the moment.

 

[fsincox]

I agree it is interesting to see more complicated parts but to do GD&T properly you must be well aquainted with the function of the part in the assembly, you are the only one, here, who knows it.
Frank

 

[bxbzq]

Ok. Now I feel better.
For this part, I don't know how to bring very detailed description up to you guys. I thought it may not be necessary for datum system. Maybe later during the weekend I can work on the sketch to explain more. But let me try to start with this: there are two things that puzzle me the most right now.

1. I totally agree with pmac that a datum axis is needed. But, when using a diameter to establish a datum axis, a rule of thumb is its length should be sufficient to establish a repeatable relationship with gage. In my case there are 4 cylinder features are used to mate with or hold another part. I put the diameters of these 4 features in view A in the sketch of my original post. The problem is all of them are very short in axial direction (compared with diameters). So I'm not sure if the rule is valid in this case.

2. The part assembles with following parts:
-- A cylinder
-- 3 rings
-- A bearing
-- 2 Supports symmetrically
I was wondering if an individual datum reference frame is needed for each assembly process. If it is, then it would end up with 1+3+1+1=6 datum reference frames. CheckerHater says, it's all up to me. Is this why people say applying gd&t is art instead of engineering? I still think there is one approach works best for function, maybe another one for measurement, another for manufacturing.

Sorry for providing you very less details and asking for valuable help.

 

[CheckerHater]

There are many different approaches to datums and GD&T.
Some say your GD&T should represent the function of the part.
Some say it should represent machining and related fixturing.
Some say it should represent quality control procedure.
It looks like in reality it should be meaningful compromise between all three; and that where it becomes art.
Say, your part may have five different interfaces for assembling with other components, and every interface may have each own datum(s); it is not technically wrong.
On the other hand, if all five can be machined from one single fixturing; it makes sense to have only one datum framework to represent that fixturing and dimension everything in relation to that framework.
This is why we are so curious about details.

 

[bxbzq]

Ok, I'm going to change a bit the too-general question to more specific ones. I drafted a simplified sketch as attached. This is not actual design but I try to simulate in a simple way, also to avoid information security issue. Proe really drives me nuts when adding gd&t stuff to the sketch. Anyway, here comes the first one.

The part function requires the 2 X ø18 holes have 0 offset to the BRG CENTERPLANE established by the 8 X ø6 holes. See section A. I'm not sure how to reflect this design requirement by any callout. Any ideas?

Thanks.

 

[MechNorth]

I think we're getting closer to having enough info to be helpful, but still not quite there yet.
1) Re the surface that you show as Datum-A in the first drawing and Datum-B in the second drawing. Is this the surface that intimately mates with the mating part, or is it the recessed surface? I suspect it should be the recessed face because it would prevent turbulence caused by air hitting the edge of the gap that would otherwise be there.
2) I suspect that the recessed face that I mention in (1), above, should be the primary datum feature because it is more critical to the alignment of the compressor shaft than the mating faces of the two cast intake-bell halves.
3) The bearing; conical thrust or regular cylindrical mount? It is drawn as a regular cylindrical mount with a conical seat, hence the question. The difference is important because if that conical taper is functional and therefore adequately controlled, it could work effectively with other datum features to establish a datum axis via multiple datum features.
Even if it is a semi-cylindrical seat (per cast half), it can still be used with other semi-cylindrical grooves to establish a datum axis via multiple datum features.
4) The axis of the compressor shaft should establish the relationship of the mating split-faces of the bell halves. This would make the axis the second datum, based on functionality.
5) The mating split-faces of the bell halves would effectively eliminate the final dof; rotation about the datum axis (compressor shaft axis).


Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[MechNorth]

The position tolerance of 0.2(M) already allows some shift, and the datum shift afforded by the (M) on the Datum-C reference allows things to shift as well, wrt the datum. In other words, you're never going to have 0-shift. The Dia-6 holes appear to be tapped holes (i.e. M6 callout instead of Dia-6), and therefore are not practical as the basis of critical alignment. How do they affect/establish the bearing centerplane?


Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[bxbzq]

Jim,

It is the datum B surface makes the main contact with a mating part. A gap exists between the recessed face and boss face of mating part. There is no turbulence concern in this area. From functionality standpoint, the whole shaft needs to be horizontal, so the split-face is selected as primary datum. Using split-face as primary datum is also process oriented datum selection, I believe. Then comes the mating function and coaxiality.

The bearing has a trunnion on each lateral side. Each trunnion fits into a bore on a pedestal. The pedestal then sits on my part on the surface that has 4 M6 tapped holes(thanks for pointing this out), and is bolted down by four M6 bolts. So axially and vertically the bearing is controlled by the bore on the pedestal. On my part in question, I can say basically the bearing is controlled axially by the eight M6 holes, vertically by datum D, inner split-face. The 8mm width (datum E) tab controls its lateral movement. So the bearing centerplane (taking bearing axial length as a FOS) is established by the 8 tapped holes.

I don't have sketches for the bearing and pedestal right now. I wish I could be more clear.

On the drawing I attached on 7/6, I should've added 151 basic to the four holes to the left, and 89 basic to the four holes to the right to datum B.
I think I want the 0 offset to be directly controlled while this is not.

 

[pmarc]

bxbzq,
Assign 8 hole pattern as a datum feature, let's say F, by attaching datum feature symbol to FCF controlling its location wrt |D|B|E| and control location of 2 holes by positional callout |pos.|dia.0(M)|D|F(M)|.

You may also want to re-locate datum feature symbols for A and D to other views, so that it is clear which features are selected as datum features A and D. I would say that current representation is at least confusing, not mentioning that showing GD&T symbols within part's contour should be avoided whenever possible (even if it is a rough sketch).

 

[bxbzq]

pmarc,
Agree on the datum D symbol location.
If I work to ISO, would the hole pattern datum feature approach still be valid?

 

[pmarc]

Yes, it would.

 

[MechNorth]

bxbzq; just going by what I've seen for compressor inlets before. I still have philosophical issues with using the pattern of tapped holes to control location with the expection of zero shift. It seems that you are using the tapped holes to adjust the location of the bearings ... thus the issue ... the bearing location is not repeatable by the machined casting alone. Thus, it is relatively irrelevant from the perspective of controlling things. Pls provide some clarification. Tks.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[bxbzq]

Jim,
Hope attached pic explains what I meant. I just modeled them quickly to show the concept.

 

[MechNorth]

Sorry for the delay; prep then vacation.
I can understand what you want the 8 holes to do, but tapped holes automatically have clearance (play) to begin with, and the holes within the pattern will have positional error wrt each other. To me it is not an effective datum feature for this application. Add that you are using shims ... the value as a pattern of datum features drops further.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[bxbzq]

I found out couple of weeks ago that axial movement of pedestal is constrained by the side walls - the side walls are straight cut, and there will be shims used between pedestal and the side walls. So definitely the datum feature of size is the width of side walls. Sorry for the confusion. This is a process of learning gdt and my part.

By the way, I remember there was a thread discussing threaded hole as datum. I could not find it.