Reliability for a New Product - How to measure?

[csasun]

We have a lot of information and practical applications on various 'testing based' statistical analysis for obtaining reliability metrics - in this forum & website.

For more practical purposes, where we have a 'product', let us say a two-wheeler, there are many components and many failure modes possible. Some of them are strength degradation type (eg. fatigue), some are wear-and-tear type, some are material properties degeneration type (eg rubber ageing), some are performance deterioration type(eg reduced mileage) and so on.

In such complex situations, it is often suggested that 'a basket of reliability tools' like: Design Analysis, Design FMEA, Simulation modelling, Proto testing, Weibull analysis, FTA, Accelerated Life Testing, DOE, QFD etc. are to be used.

At the same time, we are often confronted with the 'management question' as to 'what all these analysis means in terms of 'Reliability, Durability and Life'! - One easily understood 'metric' that integrates all these results.......

In such complex scenario, where do we start and how do we prepare mind map and ensure we pluck the right strings through out the NPD plan.

[Pantelis]

Not sure what you mean by one metric … However there is (and you should have) a map on how you would follow all these steps at each stage to get to achieve the desired goals. This would be along the lines of following a DFR (Design for Reliability) process.
I teach a course and present the metrics and processes. See course outline at http://www.reliasoft.com/seminars/gencourses/rs560.htm (http://www.reliasoft.com/seminars/gencourses/rs560.htm) as this may help.

[Pantelis]

As a follow up , I guess I did not realize where this question was posted ... “suggest topics” and I assumed it was a general question. If so ignore my last reply as this assumed that it was a question and not a suggestion for a topic.

[csasun]

Not sure what you mean by one metric … However there is (and you should have) a map on how you would follow all these steps at each stage to get to achieve the desired goals. This would be along the lines of following a DFR (Design for Reliability) process.
I teach a course and present the metrics and processes. See course outline at http://www.reliasoft.com/seminars/gencourses/rs560.htm (http://www.reliasoft.com/seminars/gencourses/rs560.htm) as this may help.
Similar to your course outline of DFR, there is a lot written as how to plan, assess and ensure reliability - different tools & techniques for different contexts.

But, in a business scenario, for management appraisal, it is necessary to have one `overall measure' that will be representing the progress of that dimension. For example, to monitor sales, we have 'Sales in USD, Sales in numbers and Market share' as three key parameters that people look at constantly.

Similarly, for reliability, do we have one unified measure that will talk about the overall progress during product design? This becomes important because, FMEA has one measure: RPN, Design Analysis has its own parameters like: resonance frequency analysis, stress plots deflection etc. during testing, it is number of cycles or hours (and there are so many different types of tests conducted - component level, system lvel, complete product level etc), during actual customer trials - it is number of troubles experienced etc.

When we need to appraise management, what should be the system of metrics we need to have that will integrate all the above indicated measures as one 'dash board' ?

Lastly, I do not know if this is part of general discussion or a suggestion of key topic of interest - in either case, I thought it is very important aspect of Reliability management - not well addressed in literature

[csasun]

We have a lot of information and practical applications on various 'testing based' statistical analysis for obtaining reliability metrics - in this forum & website.

For more practical purposes, where we have a 'product', let us say a two-wheeler, there are many components and many failure modes possible. Some of them are strength degradation type (eg. fatigue), some are wear-and-tear type, some are material properties degeneration type (eg rubber ageing), some are performance deterioration type(eg reduced mileage) and so on.

In such complex situations, it is often suggested that 'a basket of reliability tools' like: Design Analysis, Design FMEA, Simulation modelling, Proto testing, Weibull analysis, FTA, Accelerated Life Testing, DOE, QFD etc. are to be used.

At the same time, we are often confronted with the 'management question' as to 'what all these analysis means in terms of 'Reliability, Durability and Life'! - One easily understood 'metric' that integrates all these results.......

In such complex scenario, where do we start and how do we prepare mind map and ensure we pluck the right strings through out the NPD plan.
---------
Dear Pantelis,
I haven't heard from you for long. Any new thoughs? In either case, will it not be good idea to introduce this as a separate topic of discussion - Reliability Metrics for NPD?

Just to clarify or support on ONE METRIC idea, consider GDP as a metric of Nation's productivity, in similar lines for reliability?

[avanti]

..................In such complex scenario, where do we start and how do we prepare mind map and ensure we pluck the right strings through out the NPD plan.

To present progress to management or customers seems like an easy task.

In my case I usually present a monthly chart that plots reliability vs. time (or any other parameter vs. time that is relevant to your industry).
Initially I make a math model and run it several times and each time assuming a fixed reliability for each component (say from .99 to 1.0). This does two things; (i) it shows if any changes to the design might be necessary such as adding redundancy, and (ii) it provides a reliability goal for supplier components.

I subsequently refine component reliabilities by using most of the techniques you mention in post #1. I also include 'mil-spec' predictions and even guesses occasionally. As development proceeds, I update the analysis and send the updated rel vs. time chart to management.

Hope this helps.