Stackup -comparison

[greenimi]

Question for the group":
(Ref: ""maximum wall thickness at one spot" vs. "maximum consistent wall thickness." from a previous discussion)

How the stackup -minimum and maximum wall distance- change (if it does ) , IF I would say that:

X maximum = maximum possible distance that can ever happen in a single cross section for cases.
x minimum = minimum possible distance that can ever happen in a single cross section for cases.

Will those two values of 2.35 and 0.65 for the maximum wall distance and minimum wall distance respectively, change with my above adjustment?
If yes, what would be the "new" values for X maximum and x minimum?

Thank you very much

 

[Ajith NJ]

Hi
You are doing the worst case analysis. Means you are finding the 2 extreme values.
2.35 and 0.65 are the max and min wall thickness at any cross section taken along the length.
In simple words worst case analysis you did, concludes that the wall thickness at any cross section will be between 0.65 and 2.35.
X maximum a X minimum assumptions you made are absolutely right. Infact this what the assumption of Worst case analysis also. This is reason why you got 2 extreme values. 2.35 and 0.65 is the fixed max and min at any cross section throughout the length. Unless and until you change the dimensions or tolerances, this value does not change.
I want to state one general statement about tolerance stack up analysis. Tolerance analysis is done at design stage where still the part is in virtual world.
In this particular case you have 3 variables namely ID, OD and TOP for OD which contributes for wall thickness. There is fixed max and min value for all 3 variables. At design stage if i want to find min wall thickness of this part, then I will take max ID, min OD and axis moved towards bottom side ,assuming we are doing stack at top side. Similarly for max wall thickness also we will be virtually calculating.

If worst case analysis is clear to you, then I request you to go through statistical tolerance analysis like RSS.

Thank you.

 

[Belanger]

In this particular case you have 3 variables

I request you to go through statistical tolerance analysis like RSS

Actually, I wouldn't conduct a statistical stack in a case like this where there are so few variables. There's no absolute rule, but usually we would want more variables in the stack (I often hear 6 or more) to make RSS reliable.
Furthermore, a "purist" will say that RSS only works when all variables are independent. That's not the case here, where the bonus tolerance (row D) is a direct function of the OD's size (row F).

 

[Ajith NJ]

Hi
No no I think you misunderstood me. I dint ask to do RSS for this particular problem. I was just mentioning about tolerance analysis methods to be aware of.
Yes I absolutely agree with you that RSS is preferred for more than 6 line items in stack.
Yes and also all the variables should be independent. Another way of specifying this is no dimension should be contributing for more than 30 to 40 percent of the overall stack result.
Sorry for bringing RSS to the discussion which would have confused initial questioner.
For this particular question, 2.35 is max possible and 0.65 is min possible wall thickness that can occur.It will not change.

Thank you.

 

[greenimi]

Gentlemen,

This is an academic discussion only and RSS and no-RSS is really irrelevant on what I am asking.
This is a case from Alex Krulikovski (Stackup book)--as you can see is quite old book (written probably in 90's per 1994 standard), but the concept and what I am trying to learn from this discussion is still valid and conceptually unchanged.

Again, my questions are:

How the stackup -minimum and maximum wall distance- change (if it does ) , IF I change a little bit the definition of what X maximum and x minimum means:

X maximum = maximum possible distance that can ever happen in a single cross section for cases.
x minimum = minimum possible distance that can ever happen in a single cross section for cases.

I see Ajith voted for no change. Ok. That is his vote.
I think I see some changes (as usually If I am wrong I will stand corrected), but I like to get more opinions before I clutter the discussions with my own calculated values.

 

[Belanger]

Since this is a simple case of OD to OD, with no other datums involved, then I don't think the answers will be different. The numbers plugged into the spreadsheet already account for form error (per Rule #1). If another datum were added, such as the end face of the part being a primary datum P, then there would be additional error between that datum P and the secondary datum A.

I recall that this was discussed years ago, and I happened to find a thread from 2013:
[URL unfurl="true"]https://www.eng-tips.com/viewthread.cfm?qid=348558[/url]

Greenimi -- is the question you're asking here (about the definition of max/min X) similar to what was going on in that other thread?

 

[Belanger]

To amplify my first paragraph in the comment above, I've included a sketch. In the original Krulikowski problem above, the max wall thickness was the distance between the OD and the UAME of the hole.
In my sketch, because of the extra datum and the perp tolerance, the max wall thickness is the distance between the OD and the RAME of the hole, but the min wall thickness will now include the tilt of the hole as an additional factor.

I haven't done this stack, and it would get messy with M and L being invoked, but hopefully it clarifies my view about "consistent thickness" vs. "thickness at one spot."

 

[greenimi]

Since this is a simple case of OD to OD, with no other datums involved, then I don't think the answers will be different.

That I was thinking too. The answers should be the same: X max .0205 no matter A is RMB or A is MMB.
Do you agree with my above statement? If you plug the values in your spreadsheet did you get the same numbers regardless if datum feature is called RMB or modified at MMB?

J-P,

Probably is the same discussion, but the configuration it is not the same.
I want to make sure that MMB or RMB is irrelevant for the type of x max or x min I am looking for in this example.

 

[greenimi]

J-P,

If you don't mind lets agree (or disagree) with my example's value (if they are going to change or not) and then I will calculate yours.

Therefore, Do you think that the values on the initial sketch (from AK book) are going to change if I've added the X maximum and x minimum "clarification" (see above)?
If yes, what are the new values?
if no, why not?

Thank you

 

[Belanger]

The answers should be the same: X max .0205 no matter A is RMB or A is MMB.

If your question is what I've quoted here, then I presume you meant to type 2.35 rather than .0205?
If so, then the max stays the same. The min would change, however, because having MMB on the datum causes a shift tolerance to appear only in the min column:

However, that's just talking about RMB versus MMB. But is that related to your question about cross-section vs. consistent wall thickness?

 

[greenimi]

If your question is what I've quoted here, then I presume you meant to type 2.35 rather than .0205?

I agree. I meant to replay in the other thread (

https://www.eng-tips.com/viewthread.cfm?qid=471498):

.0205 in IMHO the correct value regardless of MMB or RMB for datum feature A. Sorry about that. I have too many irons in the fire.

For this discussion I got the same value like yours for x minimum (0.45), but for X maximum I got 2.6.

However, that's just talking about RMB versus MMB. But is that related to your question about cross-section vs. consistent wall thickness?

Ignore RMB versus MMB (that is applicable to the other thread). But you know what, could be applicable here too. Again, even in this thread if A is RMB (as originally is shown) or you change it to MMB, IMHO the two values of X maximum and x minimum are not changing. Do you agree with that? i hope you do. If not, let me know.

Now regarding X maximum = 2.6 (my value) I got it from:
17.0/2 + (0.1+1.5)/2 -13.8/2 = 2.6

Lets reconcile my X maximum value as this looks like I am not really understanding on why yours is only 2.35.
So, J-P, in your opinion onyl X minimum will be different if the question is changed from consistent wall thickness to cross-section?
X maximum stays the same?

And yes I would confirm that my intent is "cross-section" because "consistent wall thickness" is the original answer from AK's book.

 

[Belanger]

OK -- sorry for the confusion. I wasn't really following that other thread "Concentric holes," aside than one tangential comment I made. But I'm happy to dive into this one, with the Krulikowski example above.

In this thread, let's say that Krulikowski changed the circled S to a circled M. I showed the max as staying the same (2.45) but you got 2.6.
Where does your 0.1 come from?
17.0/2 + (0.1+1.5)/2 -13.8/2 = 2.6

Since the only change is the MMB modifier, are you thinking that the datum shift is 0.1? I see the total shift as 0.4 (eventually cut in half) but that doesn't apply in the max column because datum shift is only felt with the datum feature departs away from max material (thus I plugged 0.4/2 it only into the min column).

 

[greenimi]

Where does your 0.1 come from?
17.0/2 + (0.1+1.5)/2 -13.8/2 = 2.6

I think you are correct here: I have added another 0.1 (2x 0.5) by mistake.
The middle term should have been only 1.5 which comes from 0.5 (initial positional tolerance value ) + 2 x 0.5 (bonus from departure from LMC = 1.5

So, I agree with you J-P, that the correct calculation for X max is
17.0/2 + (0.1+1.5)/2 -13.8/2 = 2.6 2.35

Therefore only x minimum will change if we change the problem from "consistent thickness" vs. "thickness at one spot."

Do you also agree that datum feature modifier on A is irrelevant for our case? I mean joggling between MMB and RMB on "A" will not make any difference for the "thickness at one spot" case? In other words X maximum and x minimum values stay unchanged.

(Those values of X maximum and x minimum will change if A is called LMB, but that in entire another case)

 

[Belanger]

I see two very different questions going on here:

1) "Are the max/min answers affected by changing datum A's reference from RMB to MMB?" We agree it's yes because the min answer changes from 0.65 to 0.45.

2) "Are the max/min answers affected by changing the question from finding the consistent thickness to finding a thickness at one spot?" That's a different question, and I see it as unrelated to the RMB/MMB question. My answer to this question #2 would be no. In order for the answers to this question to be different, we'd have to add another factor into the mix, such as the perpendicularity tolerance shown in my hand sketch above.

In other words, the max/min answers aren't different for question #2 because in the Krulikowski problem there is only a position callout and size tolerance (and size includes form error). But if another error such as tilt is added to the part, then we might get different answers because datum A transitions from a UAME to a RAME.

 

[greenimi]

1) "Are the max/min answers affected by changing datum A's reference from RMB to MMB?" We agree it's yes because the min answer changes from 0.65 to 0.45.

No. We did not. We agree that the answer (the X minimum distance) changes from .065 (as shown in the book) to .045 (as calculated both for us) because we changed the QUESTION, we changed what we were asking for and NOT because of changing the references for datum A's from RMB to MMB.
As a matter of fact changing the references from RMB to MMB is NOT GOING to change the answer for X maximum and x minimum.

Again, that is my opinion.

So again,
if the asked question is: what is the X max and X min in the consistent thickness case: I would say that the answer X max: 2.35 x min:0.65. No matter if a is MMB or a is RMB those answers would not change.

if the asked question is: what is the X max and X min in the thickness at one spot case: I would say that the answer X max: 2.35 x min:0.45. No matter if a is MMB or a is RMB those answers would not change.

Now, probably you don't agree and I am curious why not? But that is my understanding on what we talked about.
Let me know, in details, on where you do not agree with my statements above.

 

[Belanger]

We agree that the answer (the X minimum distance) changes from .065 (as shown in the book) to .045 (as calculated both for us) because we changed the QUESTION, we changed what we were asking for

No, I did not agree to that. I got a different answer solely because of the RMB-to-MMB transition. RMB vs. MMB changes the min answer because of the extra shift tolerance. I'm not sure why you disagree with that (see my spreadsheet above).

But for your particular concern, the key factor to getting a different min answer for consistent thickness vs. thickness at one spot would be whether there is a perpendicularity tolerance (or similar additional tolerance) involved; refer back to my hand sketch. With that perp tolerance, and new primary datum, the position tolerance of the OD is held with respect to the hole's RAME axis. But in the original Krulikowski picture the position tolerance of the OD was held with respect the hole's UAME axis.
Thus, with the RAME datum in my sketch, the hole could tilt and cause a thinner wall without the GD&T knowing it. But with the UAME datum that Krulikowski used, any tilt of the hole would force the OD to tilt with it, having no thinning effect.

Anyhow, that's how I'm seeing all of this.

 

[pmarc]

If we assume that the part below is produced perfectly straight, can the picture for X minimum be qualified as an illustration of consistent wall thickness?

 

[Belanger]

Thanks, pmarc -- a picture certainly helps. Yes, I would call that consistent wall thickness because that Xmin could run down the full length of the part (longitudinally). However, I stand corrected on one thing; while I arrived at that min by adding MMB, it can also occur in the RMB case when there's form error.

 

[greenimi]

However, I stand corrected on one thing; while I arrived at that min by adding MMB, it can also occur in the RMB case when there's form error.

Therefore, do we agree that MMB or RMB on "A" does not matter and the X maximum/ x minimum won't change?
I guess now we should.

Also that this statement

RMB vs. MMB changes the min answer because of the extra shift tolerance.

it is a little misleading in my opinion.

 

[Belanger]

Let me turn it back to you... Is the Krulikowski answer for Xmin flat-out wrong?
I agree about the min now, as illustrated by pmarc, but I still think there's room for a discussion about the "min at one spot" longitudinally (the one with my perp tolerance added). That's what I was thinking of when we use the "one spot" terminology.