Using Rating Scales in the FMRA

Xfmea allows you to use qualitative ratings (e.g., "Level B - Reasonably Probable," "Level E - Extremely Unlikely," etc.) to rate causes and failure modes based on their relative probabilities of occurrence. In addition to being used in RPN calculations and qualitative criticality analysis, these ratings can also be used to assign quantitative models to cause and failure records in the FMRA (in the absence of more specific data).

By using rating scales to assign quantitative models to FMRA records, you can:

Generate a preliminary baseline reliability estimate for a system based on probabilities of failure that are assigned via qualitative occurrence ratings.

Perform a quantitative criticality analysis based on probabilities of failure that are assigned via qualitative failure probability ratings.

To use a rating scale in the FMRA, an authorized user must first configure the scale by associating a quantitative model (e.g., a constant failure probability of 5%) with each qualitative rating (e.g., "Level C - Occasional"). Then any user can choose to base an FMRA record's reliability policy on the record's assigned rating. The following sections explain how to edit the rating scales and how to base a policy on a rating scale.

Editing Rating Scales

Authorized users can edit the occurrence scale in the Rating Scale window for the current project or the library.

To edit the scale for the project, choose Project > Management > Configurable Settings > Occurrence.
To edit the scale for the library, open the Profiles/Library Manager and then click the Edit button on the Occurrence page. The window also will be displayed when you double-click one of the scales displayed on these pages.

The failure probability scale can only be edited for the current project. To do this, choose Project > Management > Configurable Settings > Probability of Failure.

In the Rating Scale window (regardless of whether it is used to edit the occurrence scale or the failure probability scale), the Quantitative Value will be treated as either a fixed probability (that is the same for all times) or a probability at a specified time (assuming an exponential distribution). You select how the value is treated in the Treat Quantitative Values As area.

Note: When you convert an existing library or project from Xfmea 5, the software will set the quantitative values of these scales to match those in the "RS DFR Focus" profile that ships with Xfmea 8. Thus, after converting data, you may have to re-assign or manually adjust the associated profiles. See Convert from Version 5.

To be able to assign fixed (i.e., time-independent) probabilities of failure, select the Fixed Probability (Q) option.
- For example, suppose you assigned a quantitative value of 0.01 to an occurrence rating of "7 - High." In this case, if a cause is assigned an occurrence rating of "7 - High" and has a reliability policy based on the occurrence rating, then it will be automatically assigned a model that represents a fixed unreliability of 1%.
To be able to assign time-dependent distributions with constant failure rates, select the Exponential Distribution at R(T) = 1-Q, and T is option. The quantitative value will then be treated as an unreliability point at the time you specify in the Time field. This unreliability point will be used to solve for the exponential distribution's parameter.

Note: If you are using the Rating Scale window to edit the occurrence scale in an enterprise library, the time specified for this option will always be entered in base units. This is because the same standard library might be used by different databases.

For example, suppose you entered 1,000 for the time and Hour for the time units. Moreover, suppose you assigned a quantitative value of 0.01 to a rating of "Level C - Occasional." In this case, if a failure has a rating of "Level C - Occasional" and a reliability policy based on the failure probability rating, then the software will solve for the exponential distribution's parameter given an unreliability of 1% at 1,000 hours. A model using this distribution will then be automatically assigned to the failure.

Note: To solve for the parameter of the exponential distribution given the unreliability point, the software uses the same calculation method used in the Quick Parameter Estimator. For example, if the given unreliability point is 1% at 1,000 hours, then the model is an exponential distribution with lambda = 0.00001. (You can choose to use mean time for exponential distributions on the Settings page of the Application Setup.)

Basing a Reliability Policy on a Rating Scale

When a user chooses to set the reliability policy of a failure or cause record based on a qualitative rating, the software will use the rating scales defined for the project to create a probability model for that record.

To base a cause's reliability policy on an occurrence rating, first make sure that the Initial Occurrence and/or Revised Occurrence properties are enabled on the FMEA > RPNs page of the interface style and that the cause has an occurrence rating assigned. Then select Based on Initial Occurrence or Based on Revised Occurrence as the policy type.
To base a failure's reliability policy on a failure probability rating, first make sure that the Failure Probability criticality analysis field is enabled for the project and that the failure has a failure probability rating assigned. Then select Based on Failure Probability as the policy type.

Tip: You can also view and select a rating scale by clicking the Select Existing icon that appears when you click inside the Initial/Revised Occurrence or Failure Probability field on the Properties tab for the selected failure/cause record.

If you later change the rating that the reliability policy is based on, or if you change the qualitative value associated with the rating in the rating scale, the software will automatically update the associated model.