PRISM Forum - Message Replies
Topic: PRISM Questions and Answers
Topic Posted by: SRC
Date Posted: Wed Jan 12 8:33:33 US/Eastern 2000
Topic Description: Welcome to the PRISM forum! Please feel free to post your questions and comments about the PRISM assessment software here.
Posted by: Boutbien laurent
Organization:MBDAM (Châtillon France)
Date posted: Wed Feb 27 4:59:54 US/Eastern 2002
Subject: query on Process Grade Score
First of all, thank you for your previous answer regarding the PRISM software guide. I am now able to have a better understanding of the PRISM philosophy analysis.
However, I have now a new question concerning the relation between the Process Grade Score query and the corresponding Pi-Factor. Answering the different queries for the Process Grade Score, I have noticed that there is no corresponding between the software true figures and the expected tabulated figures summarised in tables I-4 (page 262) and H-10 (page 247). In order to illustrate this remark , please find herefafter the result of the simulation :
Type Pi (software)Grade Pi (tables)
Design 0,0419 78,3% 0,039
Manufacturing 0,0347 83,6% 0,034
Part Quality 0,0795 89,3% 0,075
Management 0,0020 89,5% 0,002
CND 0,1460 76,2% 0,134
Induced 0,0272 100% 0,009
Wearout 0,0229 100% 0,008
growth 0,927 100% not tabulated
Infant Mortality 0,271 100% not tabulated
Moreover, How can you justify constant value for the growth factor and infant mortality factor despite the existence of mathematical formulae for determining these factors (Pi function of time).
Thank you for your assistance and help.
Subject: Query on Process Grade Score
Reply Posted by: David Dylis
Organization: Reliability Analysis Center (RAC)
Date Posted: Thu Feb 28 14:36:33 US/Eastern 2002
You have identified errors with the Process Grade tables in the PRISM User Manual. The values that are being calculated in the actual PRISM software are correct. The PRISM User Manual will be updated with correct information in PRISM Version 1.4.
A premise of the Infant Mortality model is that failures are primarily a result of defects present in the system at the time it is fielded. These defects arise from anomalies in the manufacturing process. These manufacturing anomalies can be either at the part or the system level. As a result, failures resulting from these defects often manifest themselves as early-life or infant mortality failures. Since this is the case, it is important in a system level reliability model to account for the infant mortality effects.
Therefore, a model factor was developed to account for these infant mortality characteristics. The factor is applied to only the part and manufacturing failure cause categories, since these are the categories susceptible to infant mortality characteristics. The infant mortality factor was derived by analyzing component-level field failure rate data in the RAC databases that included time to failure information. Weibull distribution parameters, alpha and beta, were derived from this data. The Weibull parameters are not related to the screening strength or the precipitation efficiency, they only provide an indication of the typical observed infant mortality characteristics of various components. The model developed provides an instantaneous failure rate. If the average failure rate is desired for a given time period, this expression must be integrated and divided by the time period. Since ESS was not performed on systems from which the model was developed, the model is normalized to the situation in which ESS is not performed.
In addition to the infant mortality factor, electronic systems can also benefit from the reliability growth that can be realized if proactive actions are taken to minimize the probability of failure from specific failure causes. This improvement can be derived primarily from the effective implementation of a Failure Reporting and Corrective Action Systems (FRACAS). All electronic systems experience reliability growth to some degree during the development phase. The rate at which reliability grows is proportional to the degree to which failure causes are identified and mitigation actions are taken. A common growth model used to describe this reliability improvement with time is the Duane model.
The Duane Model was used in to develop the growth factor for PRISM. This model is typically utilized during system development, however, there is additional opportunity for reliability improvement after the system has been fielded. There is also no reason to believe that the same level of growth cannot continue if the same level of diligence is applied to the FRACAS after system deployment. However, in many situations the opportunity for reliability growth is not realized once the system is deployed. The growth factor used in the model developed is a function of the diligence in applying FRACAS on the deployed system, and adjusts values in the growth factor accordingly. This is applied because the growth process occurs during the development, and the system reliability growth that occurs in the field, if any, continues from the growth that took place previous to system deployment.