chains of tolerances

[simlaz]

Hello everyone

I have a doubt that it assails me: how to handle tolerances in an assembly if the components have both tolls. dimensional shape and position? will the axieme have as tolerance the sum of those of each component? are added by type (dimensional, shape and position)

please... give me the right
Thank you.

 

[meccanicamg]

the overall result will be the overlap of the effects of dimensional tolerance with the geometric one

 

[simlaz]

Can you please give me some examples?

 

[meccanicamg]

if you have a series of parallelepipeds (10 pieces) with all tolls +/- 0.1 you will have the total length will be 10*0,1/10*(-0,1) then total length +/- 1 mm.
if you add an error of parallelism of the faces of 1° that for the measure will pass x, you will have 10*x/10*(-x).
evaluating everything you will have that the minimum is the sum of the minimum and the maximum sum of the maximum

 

[simlaz]

Suppose we have:

component to
thickness 10 +/- 0.1,
Parallelism tolerance on 0.02 sup inf and sup
0.05 sup surface parallel tolerance compared to inf

component b
thickness 5 +/- 0.2
Parallelism tolerance on 0.03 sup inf and sup
0.06 sup surface parallel tolerance compared to inf

how do I share the piece given by a+b?

...thank you

 

[meccanicamg]

Suppose we have:

component to
thickness 10 +/- 0.1,
Parallelism tolerance on 0.02 sup inf and sup
0.05 sup surface parallel tolerance compared to inf

component b
thickness 5 +/- 0.2
Parallelism tolerance on 0.03 sup inf and sup
0.06 sup surface parallel tolerance compared to inf

how do I share the piece given by a+b?

...thank you

If you make a drawing, we understand what you're talking about. The theory is always the same.

 

[simlaz]

Sorry... I made some past copy errors in the previous message.
I attach a "crash" as a reference to what I ask

Thank you.

 

[meccanicamg]

Sorry... I made some past copy errors in the previous message.
I attach a "crash" as a reference to what I ask

Thank you.

body to:
minimum = 10-0,1-0,02-0,05-0,02
= 10+0,1+0,02+0,05+0,02

body b:
minimum = 10-0,2-0,03-0,06-0,03
= 10+0,2+0,03+0,06+0,03

However it is quite absurd to have all those tolerances of form, not even a jonson block.

do the accounts and a+b becomes

maximum = maximum to +max b
minimum = minimum a + minimum b

I would say it's elementary

 

[simlaz]

but we are sure that the effects are cumulative... at the bottom we are talking about slightly different concepts: some tolerance ranges are non-consecutive overlaps.
I have doubts

 

[Er Presidente]

It's a matter of luck, like horoscope, gabala or weather forecast, if you say dog, you have no hope.
building the "cubests" must think that if the "sort will warn you" your suppliers, respecting the law of murphy (if anything can go wrong, it will do it), they will give you or all the absolutely identical blocks but at the lower limit of the single tolerance or, all absolutely identical to the upper limit.

What will you do in that case? will you curse the malasort?

Perhaps our school training system should introduce a course for mechanics, "building work construction with the use of full bricks, other known as bricks".
By raising a wall built with very precise bricks you could organize challenges where our young students could try in the noble art of building construction, you could also reward those who could respect the size of the project.

p.s.: there will be a reason if the "muratories" work in centimeters!
:smile:

 

[Fulvio Romano]

In any case, do you approach the problem as a statistic?
no numbers are added, but confidence intervals.
If you want to generalize without being too much to reason, you can describe each tolerance as a normal probability distribution. then subdue distributions in terms of media and variance.

...I never thought about it...what if it works...:smile:

 

[Er Presidente]

In any case, do you approach the problem as a statistic?
no numbers are added, but confidence intervals.
If you want to generalize without being too much to reason, you can describe each tolerance as a normal probability distribution. then subdue distributions in terms of media and variance.

...I never thought about it...what if it works...:smile:

Of course it works, we'd miss it, the important thing is that you know before you're paying a chance that the machine will jump, not the certainty.
:smile:

 

[Fulvio Romano]

Of course it works, we'd miss it, the important thing is that you know before you're paying a chance that the machine will jump, not the certainty.
:smile:

No, wait, I didn't know if you were joking or I didn't explain.

tolerances are a probability that a piece has a certain size. But of course the integral of probability throughout the tolerance range must be one, so you are sure.

And then you remember the signature of d-prom? :

remember well: when you design something that are more than three pieces counting the screws nuts and washers, never bet on the balls until you saw it mounted and working, otherwise it will pass very little time before you find yourself reduced to the eunuch condition.

 

[Ing Italy]

of course the integral of probability
":

probability density if distribution is continuous:smile:

 

[Er Presidente]

...
tolerances are a probability that a piece has a certain size. . .

I wouldn't agree, that would be the working class.
tolerance imposes the maximum and minimum oscillation around a "target", nothing says on the probability that you have on each single combination.

 

[Fulvio Romano]

I wouldn't agree, that would be the working class.
tolerance imposes the maximum and minimum oscillation around a "target", nothing says on the probability that you have on each single combination.

In fact, this was my doubt.
tolerance is not a normal distribution, but a range of eligible values, all equivalent. therefore there is no average (there is no more likely value than others) and obviously no variance.

sin. . .

 

[Er Presidente]

In fact, this was my doubt.
tolerance is not a normal distribution, but a range of eligible values, all equivalent. therefore there is no average (there is no more likely value than others) and obviously no variance.

sin. . .

according to murphy law the probability is directly proportional to the gravity of the consequences!

:biggrin:

 

[Er Presidente]

da wiki:
the first postulator of the law was the engineer of the American aeronautics edward murphy, but the current formulation is due to the military doctor john paul stapp.
edward murphy was one of the engineers of the experiments with rocket-su-rotaia performed by the us air force in 1949 to verify the tolerance of the human body to violent accelerations (usaf project mx981). an experiment provided a group of 16 accelerometers mounted on different parts of the subject's body. two ways were possible in which each sensor could be hooked to its support, and methodically the technicians mounted all 16 in the wrong way.
murphy pronounced his historical sentence,
« if there are two or more ways to do one thing,
and one of these ways can lead to a catastrophe,
Then someone will do it like that. »
which was reported by the subject of the test (the major doctor John Paul Stapp) to a press conference a few days later.

 

[Fulvio Romano]

In fact, when you have an important decision to make, better to launch the coin.. .at least the probability of wrong is "only" of 50%

:tongue:

 

[simlaz]

Thanks for the involvement... I think we're digging.

Could you pf go back to my hexanium and explain how you would quote a+b?

thanks in advance