Theoretical Datum for a Complex-Contoured Surface... 2

[NutZach]

Good morning folks!

Long-time reader... first-time poster... ASME GDTP-2009 Senior...

Here's the question: Y14.5-2009 figure 4.3, bottom row... We see an example of how a "complex" datum feature MIGHT (in this one particular example from that figure) be able to produce a combination of theoretical datums consisting of an axis, point, and center plane... That's all well and good for a tapered / elongated symmetric feature like the one we see in that image, but... Suppose I've called out as a primary datum feature an irregular lofted complex-contoured surface? A surface that isn't tapered and therefore doesn't converge to any particular point... a surface that doesn't have a median plane... a surface that doesn't have any logical placement for a theoretical axis... I'm talking about a complex-contoured lofted surface.

What is the theoretical "datum" for this lofted surface? Is it just an infinite number of datum points? Does it even exist at all? Why wouldn't we just refer to that datum as a "mathematically-defined surface"? Or perhaps we should just skip this consideration altogether and just go straight to what the true geometric counterpart (simulator) looks like? 2009/2018 both define a datum to be "a theoretically exact point, axis, line, plane, or combination thereof derived from the theoretical datum feature simulator." It certainly seems like we really ought to be including "mathematically-defined complex-contoured surface" in that list as well.

Thoughts?

 

[chez311]

I'm not sure I see the value in additional complexity of this type, especially for an uncommon use case in most fields.

The section on theoretical datum definition is pretty arbitrary in a lot of ways anyway, I've never thought of it as very important as to what exact theoretical datums (point/line/plane) a particular datum feature (or combination of datum features) produces, the most useful aspect might be for a lot of people a different way to visualize how DOFs are constrained. People spend a lot of time worrying about it because the standard spends a decent amount of time on it and most of us are very used to theoretical points/lines/planes from geometry and linear algebra so its natural to gravitate to these concepts, but in my experience the more important part in regards to part definition is DOF constraint and how datum features interact with datum feature simulators/TGC especially as it relates to the specified Material Boundary condition.

For a common use case of what I mean and a question asked here relatively often, consider a non-symmetric pattern of holes used as a datum feature. Is the datum axis produced at one of the holes, at the centroid, or somewhere else? It really doesn't matter, it is arbitrary as long as its consistent when taking measurements on a CMM or other device.

2009/2018 both define a datum to be "a theoretically exact point, axis, line, plane, or combination thereof derived from the theoretical datum feature simulator." It certainly seems like we really ought to be including "mathematically-defined complex-contoured surface" in that list as well.

As the standard says, a theoretical datum can only be a point/line/plane or combination thereof - there is no reason to expand it to any other theoretical geometry. As I noted above, even if we can't pinpoint "where" they are in space they can be defined arbitrarily (as long as they are done so consistently) and constrained in the applicable DOF.

 

[NutZach]

I do agree that this whole exercise of considering theoretical datums is somewhat irrelevant. The only reason it's bothering me? We have an answer to the question "what's the datum for this datum feature?" for literally everything else... So by saying "a mathematically-defined complex-contoured surface can also be a theoretical datum" - we can close the loop on that last piece of the puzzle.

And while it might seem uncommon for some? We do a LOT of contoured surface primary datum callouts in our organization. Part of why I'm asking the question today is because of how frequently I see something like this being done incorrectly (without material boundary consideration, without customized-DRF controls, etc). So when it comes time to explain to teammates what the contoured datum actually means? I get tripped up over what the actual theoretical datum is supposed to be...

Regarding your non-symmetric pattern of holes used as a datum feature? I would say the "theoretical datum" for that pattern of holes is actually a pattern of axes, perfectly located to one another; not a single axis.

 

[chez311]

The only reason it's bothering me? We have an answer to the question "what's the datum for this datum feature?" for literally everything else... So by saying "a mathematically-defined complex-contoured surface can also be a theoretical datum" - we can close the loop on that last piece of the puzzle.

You have an answer to that already. Even though your complex feature is not symmetric, I'd say it still falls under Y14.5-2009 para 4.3(g) "complex."

So when it comes time to explain to teammates what the contoured datum actually means? I get tripped up over what the actual theoretical datum is supposed to be...

I don't think discussing theoretical datum geometry helps that much - why not discuss what your simulator/TGC looks like and how it constrains DOF? Regardless, creating a theoretical complex datum thats not a point/line/plane is at odds with the current definition of a DRF, and requires a total rework of how DRF's are defined. A Datum Reference Frame is three mutually orthogonal intersecting planes - point/line/plane datums are essentially the basis of establishing this DRF, they don't exist in a vacuum by themselves. Some simple cases would be a plane being coincident (or parallel) with one of the planes of a DRF, a line where two of these planes intersect (or coincident with at least one of the planes), a point where all three intersect. At least this is how its "supposed" to work in the nice neat examples we're shown in the standard, this can get ambiguous pretty quickly when we introduce some more complex examples.

Regarding your non-symmetric pattern of holes used as a datum feature? I would say the "theoretical datum" for that pattern of holes is actually a pattern of axes, perfectly located to one another; not a single axis.

That doesn't help much does it? You still need to locate and orient a single DRF consisting of three mutually orthogonal planes. Hence why I say it pretty much doesn't matter, as long as its consistent during measurements as long as your simulators/TGC are setup correctly and the correct DOF are constrained.

If you were really concerned about it, you could specify an origin for your DRF on your drawing similar to Y14.5-2009 Fig 4-28.

 

[jassco]

Hi, NutZach:

Well, it does not matter how complicated your loft surface is. Your loft feature should be defined with BASIC dimensions directly or indirectly to origin and 3 mutual perpendicular planes that you choose for your part. Your inspector will probe as many points as he needs to establish the datum.

Best regards,

Alex

 

[NutZach]

You have an answer to that already. Even though your complex feature is not symmetric, I'd say it still falls under Y14.5-2009 para 4.3(g) "complex."

Only with regard to the fact that this "complex" datum controls all six DOF. Beyond that? The component elements of this particular datum (point, axis, plane) will not match the component elements of an otherwise asymmetric irregularly-shaped lofted surface... I think it would be silly if not counterintuitive to suggest that the datum for any and all irregular lofted surfaces shall consist of a point, axis, and plane... That was only the case for 4-3g because it was tapered (converged to a 'point' along a particular 'axis'), and was oblong / symmetric about a particular 'plane'. This is almost never the case for a typical lofted surface.

A Datum Reference Frame is three mutually orthogonal intersecting planes - point/line/plane datums are essentially the basis of establishing this DRF, they don't exist in a vacuum by themselves.

Yes - agreed. Theoretical datums really just serve as the "middleman" between physical simulators on your inspection tool and the actual DRF coordinate system... but...

why not discuss what your simulator/TGC looks like and how it constrains DOF?

...I certainly will explain how the TGC / datum simulator will work... But when I teach the Y14.5 standard across the enterprise? I need to be able to explain the difference between a datum feature and a datum. Students are going to ask that question. So when a fuselage-skin designer comes up to me and asks "what about when the engineer calls out the OML loft as a primary datum feature? What does that 'datum' look like?" - I need to have an answer for that. Perhaps when the Y14.5 committee decides to do away with the concept of theoretical datums and maybe skip straight to the topic of TGCs? I won't need an answer to this question anymore... But until then? The logical progression is supposed to be: 1) datum feature/target called out on drawing --> 2) corresponding simulator based on TGC is designed into inspection tool --> 3) simulator serves as a means of "mapping" a theoretical datum relative to the produced part --> 4) after all datum simulation is complete (in the correct order of precedence), these theoretical datums serve as "scaffolding" for locating / orienting a 3D cartesian coordinate system (DRF) relative to the physical part. So basically? I'm trying to rationalize step 3 above for lofted surfaces.

(regarding pattern of axes as a datum) That doesn't help much does it? You still need to locate and orient a single DRF consisting of three mutually orthogonal planes.

As a general rule... If you can't identify the theoretical datum for which you've defined a datum feature? Then there's an inherent risk (however minimal) that you may not have a complete understanding of all of those degrees of freedom that are being controlled... This might not be a big deal for a pattern of holes - It's pretty obvious that you'll usually lock down at LEAST 2x translations and all 3x rotations with that... but perhaps for a VERY gradually contoured lofted surface? The designer might logically conclude that "hey... it's ALMOST planar... so perhaps I'll just treat it like a plane, let the part slide around on the inspection tool, and proceed to define a secondary and tertiary datum feature..." - this is what we're trying to avoid. If they determined instead "well, this theoretical datum is technically NOT a plane... and is in fact a complex-contoured surface... and complex contoured surfaces - no matter how gradual the contour - shall control all six DOFs" - then they'll think about implementing custom DRF controls to permit the inspector to allow that part to "slide around" on the inspection tool so they can pin it at opposite corners with a secondary / tertiary datum feature. When contoured datum features are being used? We need to be super deliberate in what we will and will not allow for the inspector. Does that make sense?

If you were really concerned about it, you could specify an origin for your DRF on your drawing similar to Y14.5-2009 Fig 4-28.

Thankfully, the majority of my work over the past 5+ years has been fully model-based per Y14.41-2012. And per that standard, anytime a DRF is identified? A dedicated DRF coordinate system MUST be included within that model; that's not optional. So no matter how I call out my datum features in any of these parts... There shall be a DRF coordinate system, complete with visual response (highlighted planes of the coordinate system to signify translational DOF control; highlighted axes of the coordinate system to signify rotational DOF control). Placement of this coordinate system is entirely at the discretion of the designer, but it HAS to be placed somewhere (ideally, somewhere that makes sense). We'll typically use this method of visual response in conjunction with [z,u,v] / [x,y] / [w] custom DRF controls to make it crystal clear which datums are doing what.

 

[3DDave]

Because the majority of the committee is on the inspection side, they don't consider that the original definition for a hole is a feature that is given it's orientation and location relative to an axis; the concept for that axis exists prior to the existence of the hole. Similarly a complex surface is defined from a particular coordinate system that exists before the surface does and that coordinate system is the theoretical datum (or whatever they renamed it to in the latest version.)

The inverse problem - how to discover that same datum from a portion of an imperfect surface that might closely resemble some neighboring areas and is therefore not entirely unique - that's a different problem. If you understand the nature of that problem then you understand how to deal with it.

I'm guessing there aren't tolerance analysts who generate potential variations and the likely outcomes based on those variations. "The designer might logically conclude that "hey... it's ALMOST planar." is not something that happens when there is a tolerance analyst involved.

 

[ewh]

Welcome to the forum, NutZach!

"Know the rules well, so you can break them effectively."
-Dalai Lama XIV

 

[Burunduk]

NutZach, welcome.
When you reference those complex lofted surfaces as primary datum features in your company, what true geometric counterpart / datum feature simulator is implemented down the process? Do you actually construct physically or virtually the inverse shape of the datum feature? I would expect datum targets to be applied for such complex features.

 

[axym]

NutZach and Chez311,

Very interesting thread. You've touched on several ideas that I've had myself. Here are a few initial comments:

The datum for a complex surface is defined as an axis/point/centerplane combination. I agree that this definition adds little or no value, especially for a lofted surface. An axis/point/centerplane combination could be defined, but it would be completely arbitrary.

I completely agree that DOF constraint is accomplished via the contact between the datum features and the TGC's. Datums and a DRF are non-essential concepts that are somewhat useful for visualizing constraint, but only for very simple datum feature configurations (single planar surface, single orthogonal cylinder, single orthogonal slot, etc.). For more complex datum feature configurations, datums and the DRF become arbitrary and confusing.

Y14.5's treatment of the datum and DRF resulting from a pattern of parallel holes is especially misleading. Figure 4-26 in 2009 and Figure 7-18 in 2018 show the datum for a 4-hole pattern as an axis - this is wrong. If we follow the table at the beginning of Y14.5's DRF section, a pattern of parallel holes corresponds to the Linear Extruded Shape (f) and the datum would be an axis and centerplane. This datum would still be non-unique and arbitrary, but at least it would show the DOF constraint correctly.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[NutZach]

what true geometric counterpart / datum feature simulator is implemented down the process? Do you actually construct physically or virtually the inverse shape of the datum feature? I would expect datum targets to be applied for such complex features.

Regarding how the simulator might be implemented... First and foremost - I would never reference that lofted datum RMB... A physical RMB simulator for a large lofted surface is just not feasible. For the case of a large composite part? Assuming no modifications have been made to the original "true profile" of the lofted datum surface for springback or other factors within the layup tool, the surface that was machined INTO that layup tool (or perhaps a trim/drill fixture... or for big $$$, a separate inspection fixture), can serve as a fixed material boundary datum feature simulator. If the form of that lofted surface was toleranced using a unilateral-subtractive (circle-U-zero) profile tolerance? Then it would hold true that the max material boundary for that datum feature would be the as-modeled "basic" surface from the 3D model. So in that scenario, referencing this datum at MMB would be identical to referencing it at [BASIC]. Provided the layup tool (or perhaps a separate trim/drill or inspection fixture) can be produced to a reasonably precise tooling tolerance? This would be how one might simulate a lofted datum surface on a physical inspection tool.

By using a customized DRF to unlock those translational and rotational degrees of freedom that would be "tangent" to that lofted surface (treating the loft as if it were a plane), we are permitting this part to "slide around" on this lofted inspection tool until secondary and perhaps tertiary datum simulators can be engaged to constrain those remaining degrees of freedom more accurately.

 

[NutZach]

Datums and a DRF are non-essential concepts that are somewhat useful for visualizing constraint, but only for very simple datum feature configurations (single planar surface, single orthogonal cylinder, single orthogonal slot, etc.). For more complex datum feature configurations, datums and the DRF become arbitrary and confusing.

Have to disagree a little bit here... I agree the theoretical datum doesn't mean a whole lot for a lofted datum surface (other than for the purposes of teaching the class, I guess)... But the presence of a DRF coordinate system? That's going to be absolutely essential. Especially when we start using customized DRF controls (2009 para. 4.19)... If I assign [z,u,v] as the degrees of freedom controlled by a lofted primary datum surface... those custom DRF controls mean absolutely nothing unless I have a coordinate system defined in that model to show specifically how x, y, and z (and therefore u, v, and w) are configured relative to the part.

 

[axym]

NutZach,

I agree that a coordinate system would be essential if a customized DRF was used.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[3DDave]

Theoretical datums are only useful if you intend to understand the contribution of measurement errors on the final product. If there isn't any interest in that error contribution then ignoring the theoretical datum concept is perfectly fine. Most work isn't precise enough to bother with that contribution.

 

[Burunduk]

NutZach,
What you are describing reminds me of figure 4-28 in the '09 standard. With datum A referenced as BASIC to avoid implying that the simulator should be of some adjustable curvature. It is also a datum target application, which definitely makes sense. And the datum of datum feature A - well, the figure does its best to guide you not to mind it; it communicates - "here's your datum reference frame, and that's how you relate it by basic dimensions to the datum feature simulator". Doesn't this example and the text that refers to it answer your questions?

 

[Burunduk]

I disagree with some opinions expressed here, suggesting that datums, let alone DRFs, are insignificant concepts.
In many cases the datum derived from a datum feature plays a main role.
Take for example a simple stepped shaft when one diameter is designated as datum feature A and another diameter has a position tolerance applied to it referencing only datum feature A. Sure you could go on and on about how that datum feature A and its simulator, or "t.g.c" constrains 2 translational and 2 rotational degrees of freedom, but that doesn't tell the drawing user anything about where the tolerance zone is. In practice, a collet may be used to simulate the center axis used as the datum, and the tolerance zone has to be centered to that axis. There you have the requirement covered without even having to get into the datum reference frame, although it is defined too, and becomes important when a secondary datum for clocking might be needed for controlling a differently located feature. When you generalize these concepts to all possible types of datum features, yes there are times when it makes sense to skip on the datums directly to a datum reference frame, representing the measurement coordinate system, and care only about the basic relationship of the DRF with the datum or datum targets simulators. I think that every case where datum targets are involved is like that. But does it mean generally that datums are not necessary? I doubt it.

 

[NutZach]

Just getting back to the main point of the post... I'm simply suggesting that theoretical datums - however important or unimportant they might be for any particular situation - really ought to include mathematically-defined complex-contoured surfaces. More specifically - the "definition" for a datum can be split into "primitive geometric elements" like points, axes, and planes, as well as "mathematically-defined complex contoured surfaces" - which cannot be easily characterized using primitive geometric elements. This closes the loop on that "what is the datum for this datum feature" question I had before... EVEN if that question is somewhat unimportant. All I'm trying to do here is add consistency across the board with regard to how we define theoretical datums.

 

[NutZach]

And the datum of datum feature A - well, the figure does its best to guide you not to mind it; it communicates - "here's your datum reference frame, and that's how you relate it by basic dimensions to the datum feature simulator". Doesn't this example and the text that refers to it answer your questions?

2009 figure 4-28 is actually showing a linear-extruded datum feature per figure 4.3f; otherwise the presence of a secondary datum feature (12x pattern of holes) would result in an overconstrained DRF. Also, you'll notice the side-view within figure 4-28 appears to be showing a constant cross-section in that particular head-on direction... But that's somewhat besides the point here. The standard really shouldn't be "talking around" this particular issue... If every single other type of datum feature has a corresponding theoretical datum? It's my opinion that a complex contoured surface should have one as well.

 

[chez311]

NutZach,

Whether or not I think it's important, let's say we added "mathematically defined surface" to the current point/line/plane list of allowed datums. What then? I don't really see how that even helps you without totally reworking how DRFs are defined. You still have to locate and orient three mutually perpendicular planes requiring either arbitrary or specifically defined origin - it's not like it makes that process any less ambiguous.

Besides literally having a picture to point at in the standard, I'm really not clear on what value it adds.

 

[Burunduk]

NutZach,
So what would be your suggestion to the committee? What do you imagine the datum for a "mathematically-defined complex-contoured surface" could look like?

By the same token...

More specifically - the "definition" for a datum can be split into "primitive geometric elements" like points, axes, and planes, as well as "mathematically-defined complex contoured surfaces" - which cannot be easily characterized using primitive geometric elements

I'm not sure that I follow this. Is the bolded portion a suggestion for another type of datum, which differs from elements like planes, points and axes, or do you mention it as a type of datum feature, which can't be characterized by the mentioned elements, but could be characterized by... what?