Understanding Statistical Tolerance Analysis

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 2

Understanding

Analysis

Statistical

Tolerance

Definition of Statistical Tolerance Analysis

A statistical tolerance analysis is when you take the variation of a set of inputs to calculate
the expected variation of an output of interest. In mechanical engineering, a product design is
composed of multiple features, each with tolerance values that control the variable aspects of
those features. Statistical tolerance analysis is used to understand how these tolerances
contribute
the
various
performance
characteristics
of
the
design.
1D
Tolerance
Stackup
The simplest form of tolerance analysis is the single direction, 1D Tolerance Stackup. A
1D Tolerance Stackup is created by creating a cross section of a model and adding the
tolerance values for each feature in a straight line. The variation in each contributes to the
overall
output/outcome.

Worst-Case Analysis vs RSS (Root-Sum Squared) Statistical Analysis


In a Worst-Case Analysis, each dimension will have a minimum and maximum value that
represents the range of acceptability for that dimension. Worst-Case answers the question, if I
take the maximum range on each input, what is the maximum range for the measurement of
interest or stackup? We are therefore dealing with the limits of acceptability and not
probability.
RSS (Root-Sum Squared) Statistical Analysis does not focus on the extreme values, but
focuses on the distribution of the variation for each dimension. Each dimension will have a
unique distribution of values based on the manufacturing process. Tool wear, operator
differences, changes in material and environment all contribute to variation in the dimension
value. Each dimension has its own distribution curve.
When you combine the probabilities for each dimension (each separate curve) you get the
probability for the total and therefore the distribution curve of the total. Statistical analysis
answers the question, given the distribution of variation on each dimension what is the
probability that my performance characteristic will fall within defined acceptable limits. The
limitation of RSS is that it assumes all inputs are normally distributed and all performance
characteristics have a linear relationship with the dimension. These assumptions do not
account for the breadth of conditions that exist in typical scenarios found in manufacturing.
Second Order Tolerance Analysis

Because manufacturing methods vary for different types of parts, the distribution moments or
parameters change as well. RSS only uses standard deviation and does not include the higher
moments of skewness and kurtosis that better characterize the effects tool wear, form aging
and other typical manufacturing scenarios. Second Order Tolerance Analysis incorporates all
distribution moments:

Second Order Tolerance Analysis is also needed to determine what your output is going to be
when the assembly function is not linear. In typical mechanical engineering scenarios
kinematic adjustments and other assembly behaviors result in non-linear assembly functions.
Second order calculations are much more complex so hand calculations are not advisable but
the computation accuracy is greatly improved and becomes viable within a tolerance analysis
software package.

Summary
of Statistical
Tolerance
Analysis
for
practical
useage
The choice in tolerance analysis method is based on many factors, but the can be
summarized as Which method best matches the manufacturing and inspection process of
the assembly. For simple fit problems, a 1D stack-up may be sufficient. RSS is sufficient for
the small number of scenarios where the inputs are normal and the assembly relationships
are linear. For all other scenarios, Second Order Tolerance Analysis is required to address the
real world of manufacturing.

You might also like