What is reliability data and how can it be gathered and used to help you make better products and/or reduce your production costs? Let’s find out…
Prefer listening to reading?
What kind of reliability data can we get?
Reliability data can come from product returns after it’s been on sale and in customers’ hands in the field, from reliability testing (especially such as from failed tests and qualifications) and also the issue tracking database during design and development, etc.
Even if this is a new product that has not been on sale before, you will still have gathered useful reliability data during the design and development process.
If you have been following a proper reliability testing program you’ll certainly have data from the test results during EVT, DVY, and PVT where you will have done engineering-level testing and multiple tests on prototypes. This data then becomes very helpful for the next product you make, especially if it’s a similar product or an iteration of the one in question that uses, say, some of the same components and designs, because you know that it can be tested in a similar way and that the issues you originally found can help you reduce risks, costs, and engineering time, etc, because you already know what to avoid. (00:43)
Keeping a database of past issues or ‘lessons learnt.’
Most companies that sell the same product to the same customers for some time will collect data on returned products in a database during the triage process.
For example, you might highlight that products that suffered from issues used a component from a specific vendor, whereas other batches that were fine used the same component, but sourced from a different vendor. Your learning from this reliability data, then, is not to source from the first vendor for future products. This is just one lesson learnt, imagine how many there would be for a product that uses hundreds of components..!
If an issue is fixed, it’s also important to note it in the database, then the next person who is checking that particular problem can understand what the problem was, how it was caused, and what was done to fix it.
When you start the development process for a new product it is worth having a meeting to go over the information in the database and decide which lessons learnt can help in this case in order to reduce costs, time, risks, etc. (04:43)
Where do issues that occur during manufacturing come from?
If development was done correctly and the reliability testing revealed all of the issues that the product and components had by creating proper test cases that mimicked worst-case scenarios in the field by the end users into a reliability test plan, you usually shouldn’t see any of these problems once a product goes into manufacturing. So any problems occurring at this stage should be simply manufacturing-related.
It takes a lot of work to make processes repeatable, products are made within tolerances and they can cope with the stresses they will encounter in the field, etc, so there is still a lot to do at the manufacturing stage, but at least you won’t have reliability issues stemming from development if you’ve gone through comprehensive reliability testing beforehand. (10:29)
How can reliability data be used to make better products?
You have the data from feedback on how and where products failed, so how do you use this to influence V2.0 of that product, or others in the future?
Your reliability team can track the different ‘failure modes’ and create a Pareto analysis that will outline the 20% that cause 80% of the issues. These are the problems that you focus on fixing in your new product, then you have a far more reliable product already. Within the 20% of key issues, you can go deeper. Are they design, supplier, manufacturer, or component-related? Once the issue has been pinpointed, you can focus on that. For example, if component problems are common, you know that you will need to source and use different components next time. (12:50)
How user experience can influence a new testing plan.
Even if a product passed reliability testing the first time round it may have been returned because users abused it in a totally unexpected way. This information can then be used to create an all-new reliability testing plan for the next product because it is a lesson learnt. The engineers will create new user test cases so the new product will not suffer from the same problems even if it is ‘abused’ in that way in future. (17:16)
Over-engineering and reliability: Mercedes Cars in the 1980s and NOKIA phones that seemed to be indestructible.
Products were engineered in the past that probably offered more value in terms of reliability than customers actually required. Mercedes cars in the 80s were well-known as having engines so reliable that they’d do 500,000km or more. Few non-commercial vehicles today have that level of reliability built-in. The same can be said for NOKIA phones in the early 2000s. Everyone knows the stories about the 3210 which had a battery that lasted forever and was so tough it could fall a few stories without breaking.
You have to walk the line between making products reliable enough to satisfy your target customers’ needs and being so clunky, over-reliable and over-engineered that they become expensive and unattractive. NOKIA’s phones were so reliable that customers didn’t need to buy a new phone; this somewhat shot the company in the foot as it was hard to get new sales for this reason.
The level of reliability and engineering done on a product depends on the industry. For example, airplanes and medical products need to be as close to 100% reliable as possible in order to avoid accidents that can lead to a loss of life. So redundancies are added to guarantee safety that just wouldn’t be practical in consumer products due to the additional costs involved. (19:46)
How can reliability data be used to cut production costs?
We’ve already seen that some products have been over-engineered and too reliable, increasing costs for customers. You can use reliability data to avoid this scenario. You may encounter these findings:
- You’re using expensive custom parts that could be replaced by better value off-the-shelf standard parts instead, for instance.
- There are too many parts being used to make the product overall, so these could perhaps be replaced with just one or two reliable parts instead.
- You can reduce the number of suppliers you’re currently using and buy more components from fewer where possible to enjoy lower prices than if buying small amounts from many.
- Redesign old parts using DFR principles to be more reliable and then you may be able to eliminate some completely, so you have a better, more up-to-date product design. (25:11)
An example from the early automobile industry: Henry Ford’s Model-T Kingpin
Henry Ford sent people to check on scrapped Model-T’s to discover which parts had commonly failed and which were more reliable. They found that the kingpin (which connected the front axel to the body of the car) never seemed to fail, so, because it was so reliable, Ford is said to have demanded that it was remade to be inferior in order to reduce costs while still being reliable enough.
Making appropriate changes to parts based on reliability data could help reduce your costs a lot, but you need to pick and choose carefully. In this case, the kingpin is a critical part that will lead to accidents if it fails, therefore it may actually be a poor candidate for redesigning to save on costs as it could have led to increased repair costs and even legal costs if people were injured. Also, you need to take into account the costs to make the new part, retooling, testing, etc, are not cheap, so would it actually be a false economy?
An example here is that of smartphones vs. feature phones. The former are not as reliable as the latter, yet they’re preferred by users as they offer many other functions and benefits, so reduced reliability is an acceptable trade-off. In the old days products were often built-to-last, but that’s not the case today and in fact, has the full consent of users in many cases. (27:56)
Takata airbag safety scandal.
Around 80 million faulty airbags used by numerous car brands were found to be at risk of launching metal shards into users’ faces when the airbags were deployed in accidents, so this is considered to be the largest product recall of all time. There have been 26 deaths and 400 injuries to date due to these airbags. So, in this case, we can see that this is not a product that you want to be cutting costs on, as safety and reliability are critical here.
This also teaches us about the importance of having second-source suppliers for critical components, because the different automakers relied only on Takata without using any other suppliers which resulted in the problem being so widespread and serious. (33:52)
- Get help from Sofeast to do Reliability Engineering & Testing (in a China lab) – we provide reliability engineering services including reliability testing on component and PCB levels as well as the product level on various types of consumer products. Some of these tests include drop, vibration, temperature and humidity and package testing. We can customize the right tests for your product.
- The story of the NOKIA 3210
- How To Do Product Reliability Testing?
- Product Quality and Reliability Issues: Typical Classification
- DFR Design For Reliability Guide & Implementation Secrets
- Henry Ford kingpin story
- Takata airbag recall