Enough Data?
Abstract
Dianna and Fred discuss a common reliability engineering dilemma: do we have enough data? Is data nirvana achievable?
Key Points
Join Dianna and Fred as they discuss a common reliability engineering dilemma: do we have enough data? Is data nirvana achievable?
Topics include:
- Hidden challenges for RE, from too little to overwhelming amounts
- Departmental silos and data quality nightmares
- The art of purposeful collection
- Buy-in for better data and how explaining the ‘why’ behind its collection can transform resistance into collaboration
Enjoy an episode of Speaking of Reliability. Where you can join friends as they discuss reliability topics. Join us as we discuss topics ranging from design for reliability techniques to field data analysis approaches.
Show Notes
In this episode of Speaking of Reliability, Fred Schenkelberg and Dianna Deeney tackle a common question for reliability engineers: Do we have enough data to do our job?
They discuss the dual challenges of both not having enough, particularly for new and unproven products that require costly testing. On the other hand, we’re sometimes overwhelmed by excessive information without clear direction on its utility.
A significant part of their discussion revolves around the accessibility and quality of it. Data often resides in departmental silos (e.g., customer service, finance, manufacturing) that don’t readily share information. They highlight the extensive time spent cleaning up inconsistent, error-prone sources from disparate systems, often finding issues like incorrect data types or format changes.
Dianna notes that while information collection falls under management responsibility and quality management systems, reliability engineering often requires advocating for specific needs that might not be captured by standard financial or customer service metrics. They emphasize the importance of foresight and cross-functional collaboration to ensure useful data is collected from the outset, ideally before a product’s market release. This includes identifying critical parameters and coordinating with customer service.
Fred shares an insightful anecdote about how poorly designed forms can lead to skewed data, illustrating the need for a “usability engineering” approach even for checklists.
They conclude that while obtaining “nirvana” with perfectly trustworthy and available data is rare, explaining the purpose and value of collection can significantly improve buy-in and collaboration across an organization.
Thanks for the delightful oboe solo from Danzi? Reicha? wind quintet last movement (Rondo?). Flute player in Vineyard quintet would kill me if played it that fast.
Thanks for the discussion of internally generated reliability data problems. I admit it is useful to compare test reliability to field reliability, especially for those who don’t learn from the past. (Sorry, generations of products have similar field reliability, because processes, customers, and environments don’t change.)
May I suggest (again) that statistically sufficient ->field<- reliability data is (usually) available from data required by GAAP. Ships (installed base, cases, sales, production, etc.) and returns (failures, warranty returns, deaths, repairs, replacements, spares sales, etc.) tell what really happens through nonparametric reliability and failure rate functions. A little work is required to convert accounting, production, and service data to field reliability, but it is interesting.