An Elusive Product Life Time Definition
The following note and question appear in my email the other day. I had given the definition of reliability quite a bit of thought, yet have not really thought too much about a definition of ‘product life time’.
So after answering Najib’s question I thought it may make a good conversation starter here. Give it a quite read, and add how you would answer the questions Najib poses.
Najib’s note
Hi Fred,
Many thank for your very interesting articles concerning reliability topics.
I was wondering if there was a common definition of the product life time.
I know that there are several standard which specify a definition of the MTBF.
What about the product life time?
Do you know a standard which defines the product life time?
I look forward to hearing your feedback.
Najib
Fred’s reply
Hi Najib,
Thanks for the note and question. I do not have a clear definition of product life time, and maybe should write one.
MTBF, as you know, is not product lifetime, it is a measure of the inverse of the chance of failure per unit time.
Customers define or set expectations concerning product lifetime. If I expect my desktop computer to last 5 years, that is life time that I want and expect. Product/system development teams have to understand what the customer wants/expects in order to deliver a product that lasts long enough to meet the customer’s expectation.
Here are some related definitions from my upcoming glossary on reliability terms.
Asset life: The period of time from asset creation to end of life or retirement.
End of life: The instant an item fails to perform its intended function within desired specifications with no recourse for corrective maintenance. May also refer to the retirement or decommissioning of an item even if still in working order.
Field lifetime: The expected duration an item is expected to function or provide value for the customer.
Life: See conditional probability of failure, or probability of survival. Often stated as a duration missing the probability element. Together duration and probability define a statement of life.
Useful life: The number of life units from manufacture to when the item has an unrepairable failure or unacceptable failure rate.
Generally, the term product lifetime is from the customer or end user perspective, and is often defined as the time between installation, commissioning or startup, till removed, decommissioned, or failure.
In other words, how long will the item work as expected for the customer?
hope that helps,
Fred
Additional Notes
Often when I hear someone talking about a product’s life or the life of a system I think about duration. The part that is missing is the probability of successful operation over that duration. To say something has a 5 year lifetime really is pretty meaningless unless you also state the chance an item will survive that same 5 years.
For repairable systems, such as my car, when one says it will last 10 years, that implied that includes the regular maintenance and maybe a trip to the body shop or a new windshield or two. We are often more interested in availability and cost of ownership, not just reliability. Again the life time statement on its own is not all that helpful.
As you know, how we define the start and end of product life is often rather fuzzy. Is it on first power up in the factory or by the customer? Is the end of life when shut down or disconnected, or when sent out with the trash? Many times we do not know when a product ends its period of use by a customer unless we take deliberate actions to gather that information.
Product life time in some circles bounds the duration from concept to the retirement of the last product. Others define it as the warranty period or the duration we produce and support a product.
I do not have a single definition that all of us may agree upon. Still looking for that better answer for Najib. Do you have a better answer?
Use the comment section below to provide your answer, comments, or additional questions concerning a definition of product life time.
Hi Fred. We see this all the time. The nomination of a life time is simply an arbitrary figure that more often than not has no relation to failure rate or probability of mission success. I suspect that this is a hangover from airframe philosophy where an airframe has a limited number of hours life based on the vibration profile and the structural design.
Where it causes issues is when an item OEM claims a life that doesnt align with the OEM’s plans or te limitations of technology. We often see items within systems that claim a 10 year life but are hit by obsolescence issues within the first 2 years of fielded use. We hen find that the OEM has not invested in spares or planned updates for that duration so the 10 year figure is just a complete ass pluck and should not be used for any planning purposes.
Well said Tim,
Much like MTBF, product life is a meaningless gesture. We really want to know the chance of failure over a specific duration. That really isn’t so difficult, one would think.
Cheers,
Fred
Fred,
A couple of thoughts may apply since it seems to me that product life isn’t just one thing.
First, marketing decisions often affect the notion of how long something is supposed to last. Cell phones have very short marketing lives, and new features tend to drive that. As a result, consumers tend replace phones over a (relatively) short time. Refrigerators tend to be in place a long time (a couple of decades?), in contrast. This has implications for design and maintenance, and it seems that, as you have advocated here, a probabilistic way of describing behavior and performance is in order.
The other thought that occurs to me is that some products will tend to be in place until they stop working (that is, meeting expectations). There’s a bit of subjectivity in describing wear out, but it’s not meaningless. We replaced a computer recently since it started shutting down after being on for a few minutes. The engineering question is one of understanding how long we want something to last before wear out becomes a problem.
In both cases, and they’re not exclusive, we have to describe what we mean by product life, and that’s a specification problem. The obvious problem is that it’s very hard to enumerate everything that goes into a specification like this.
One path forward is to think about the economics of being in business. A company may sell a product for some time (a year, five years, etc.), and support it for some time after that. Warranty considerations may also influence this as well. At the same time, if we think of reliability as the persistence of quality over time, then we should try to determine when loss of quality (via changes in key performance indicators) begins to negatively affect customer experience.
And if a supplier says that they will support a 10 year life, and then they don’t, that’s not a technology problem. It’s an honesty problem. Tim’s issue can be managed by insisting that suppliers discuss their product roadmap. If market-driven obsolescence is significant, it needs to be part of the supply chain discussion.
Hi Paul,
Nice expansion of the discussion and you bring up a few points worth considering. I like the connection to the economics, yet production and support durations are not the same as product lifetime in my mind. I still use a few hand tools first used by my grandfather – the manufacturer is long gone and production/support for those tools expired long ago, yet the tool’s lifetime continues.
Cheers,
Fred
You’re right. It’s awfully hard to buy a 1949 Buick, especially a new one, but there are a few from the original production that are still running.
For me, one important distinction is that survival of individual units at long times isn’t indicative of the population. At the same time, technology does drive changes in what is available and what can be maintained. Your 1983 IBM PC may work just fine, but it’s hard to do any computing with it, even if it can run Windows 3.1.
Effectively, this seems to meant that there is equipment that functions, but doesn’t meet business needs today. If that same equipment is being used by an individual, none of that may matter, but if your small business is based on having a functioning 1983 IBM PC, that’s a risk that has to be managed.
What’s clear is this: product lifetime has to be defined, and life cycle cost modeled. As you correctly imply, this is something we need to think clearly about. Thanks for “provoking my thoughts.” 🙂
A great discussion and I would also be very interested in getting other people‘s views.
To me, the product lifetime is the cut-off after which the product owner/designer cannot demonstrate with sufficient confidence that the product can fulfil its functions when used and maintained according to its specifications. It is driven by the product’s constituent part shortest lifetime. For a physical component it is easily understood as the time after which degradation is unavoidable (fatigue, erosion, chemical degradation…).
Following on to Mark’s point, this definition is quite limiting when you take obsolescence into consideration.
I have encountered clients who believed that product life was equal to MTBF for a repairable product!
Hi Marie,
Good comment, although I do not know if the lifetime is limited to the design team’s ability to demonstrate or prove. Often products last and are expected to last longer than anticipated by the design team. Other times cost and other constraints limit the team’s ability to conduct life testing beyond the absolute minimum.
No, I would not link life testing or estimation/modeling to the definition of product lifetime.
Cheers,
Fred