Consistency Validations of the Distribution Definition

When you specify input parameters for a Six Sigma Analysis, DesignXplorer validates the distribution properties to ensure that there are no inconsistencies in the distribution definition. However, in some cases, invalid attribute combinations can already exist. For example, a project created with a previous version of DesignXplorer could have invalid attribute combinations. When you open the project, DesignXplorer runs checks to catch any inconsistencies in the distribution definition. The following table shows the checks performed for each distribution type:

Distribution TypeValidation Check

Uniform

Lower Bound is smaller than the Upper Bound

Triangular

Maximum Likely Value is between the Lower Bound and Upper Bound, and the Lower Bound is smaller than the Upper Bound

Normal

Standard Deviation is positive, but is not too small when compared to the Mean value:

Truncated Normal

Lower Bound is smaller than the Upper Bound, and the bounds (lower or upper) are not too far away from the mean in terms of standard deviation:

Log Normal

Standard Deviation is positive

Exponential

Decay Value is positive

Beta

Beta Shape R and Beta Shape T values are positive, their sum is not too high (Shape R + Shape T < 1000), and the Lower Bound is smaller than the Upper Bound

Weibull

Weibull Exponent is positive, and the Weibull Characteristic Value is positive and greater than the Lower Bound

If inconsistencies are found, a warning dialog box opens. You can see the Messages view for more information.

On the Project Schematic, the Design of Experiments cell has a state of Attention Required, indicating that you must edit the distribution definition to resolve any inconsistencies.

Example:

The truncated normal distribution is used in this example. It is defined by the following attributes:

In the following figure, you can see these attributes in the Properties and Chart views for the Design of Experiments cell in the Six Sigma Analysis system:

In this example, the Distribution Upper Bound (3) is far away from the Mean (1.5) when compared to the Standard Deviation (0.075). The validation check fails because (Upper Bound - Mean)/Standard Deviation = 20. To fix the inconsistency, you must either reduce the bounds, modify the mean, or modify the standard deviation.


Release 18.2 - © ANSYS, Inc. All rights reserved.