Your Reliability Engineering Professional Development Site
Find all articles across all article series listed in reverse chronological order.
It seems that anytime we draw a sample, it should be taken randomly. Statistics books and papers regularly advise using a random sample. The adverse effect on results drawn from the experiment may hinge on the randomness of the selection of samples. [Read more…]
My introduction to reliability engineering was my boss asking me to sort out how long a new product will last in use.
The expectation was it would last for 20 years or more buried in Italian mountain concrete bridges.
My first thought was about living in the Dolomites for 20 years monitoring the performance of the product.
That was quickly dashed as my boss explained he wanted an answer in about 6 months.
Now this was a problem. How do you cheat time to learn about the expected lifetime of a something? Thus started my career in reliability engineering.
Life testing for reliability engineering helps us answer the question how long till failure occurs. Specifically, we find the chance of failure over some duration. [Read more…]
This nonparametric test evaluates if two continuous cumulative distributions are significantly different or not.
For example, if the assumption is two production lines producing the same product create the same resulting dimensions, comparing a set of samples from each line may reveal if that hypothesis is true or not.
There is a lot of probability, statistics and data analysis involved with reliability engineering. Why is that? Have you considered why our field of endeavor includes the use of these tools?
Let’s say there were no statistical tools. We would not be able to accurately infer a conclusion based on an observation of a few samples. We might react to everything that we observe – constantly spinning our wheels on minor issues. We might make decisions based on factors that did not include the random variation of the items. We might track failure rates, yet not really know how to determine if the few failures we observe were an indication of a major issue or not. [Read more…]
If nothing was uncertain we would not need statistics.
Since nearly everything varies in some fashion, we need a way to describe and work with that variability.
We already know this and we know about statistics as being the right set of tools. Yet we hesitate, avoid, and refuse to pick up the appropriate tool. [Read more…]
In the poem by Oliver Wendall Holmes, The One Hoss Shay, a deacon is confounded by the various parts of his carriage the fail.
And, he decides to do something about it.
But the Deacon swore (as Deacons do,
With an “I dew vum,” or an “I tell yeou,”)
He would build one shay to beat the taown
‘n’ the keounty ‘n’ all the kentry raoun’;
It should be so built that it couldn’ break daown:
“Fer,” said the Deacon, “t’s mighty plain
Thut the weakes’ place mus’ stan’ the strain;
‘n’ the way t’ fix it, uz I maintain, Is only jest T’ make that place uz strong uz the rest.”
Translating from old English, it basically means he wanted to craft a carriage using the best materials and techniques. Later, he built a very sound carriage where every part is just as strong as all the other parts. [Read more…]
You use your calculator or spreadsheet, or even a statistics software package to calculate standard deviation, which is an estimate of the population standard deviation. Yet, understanding how one could calculate standard deviation without such advanced tools may prove useful. The knowledge of basic sum of squares methods provides a foundation for ANOVA and DOE analysis techniques. [Read more…]
Is it possible for an individual to ‘do’ DFR? Is design for reliability something, like a specific technique, that is DFR?
What is DFR and how would you recognize it if it was occurring? Like meditation, nearly anyone can strike a pose that appears similar to someone in deep meditation, yet can you tell by observation if they really are mediating? Probably not. The same is true for an organization or person that declares they are doing DFR. Maybe they are or maybe not.
Environmental testing is the evaluation of a product or system in one or more stress conditions. Environmental as in that which surrounds and affects a product. Consider temperature. Is the product going to experience outdoor temperatures as found in Fargo, North Dakota or Belmopan, Belize?
The weather is one way to describe external stresses, yet it is so much more. Environmental testing may include fungus, insect, and animal exposure. The document MIL-STD-810G lists and describes testing methods for a wide range of environmental conditions. [Read more…]
Before computers and statistical software, we relied on tables to determine values for common integration problems – the normal distribution in particular. There is no closed form solution for the integral of the normal distribution probability density function, it requires advanced numerical methods to estimate the area under the curve. [Read more…]
Stop testing when the testing provides no value.
If no one is going to review the results or use the information to make a decision, those are good signs that the testing provides no value. Of course, this may be difficult to recognize.
Some time ago while working with a product development team, one of the tasks assigned was to create an ongoing reliability test plan. This was just prior to the final milestone before starting production. During development, we learned quite a bit about the product design, supply chain, and manufacturing process. Each of which included a few salient risks to reliable performance.
There are two basic ways to consider the central limit theorem. First consider a random variable, X, which has a mean, μ, and variance σ2. If we take a random sample from f(X) of size n and calculate the sample mean, X̄, then as n increases the distribution of the sample means, X̄’s approaches a normal distribution with mean, μ, and variance σ2/√n̄. The original data, X, may have any distribution and when n is suitably large the distribution of the averages will approach a normal distribution. [Read more…]
In the Monte Carlo method, one uses the idea that not all parts have the same dimensions, yet a normal distribution describing the variation of the part dimensions is not assumed.
Although the normal distribution does commonly apply, if the process includes sorting or regular adjustments or if the distribution is either clipped or skewed then the normal distribution may not be the best way to summarize the data.
As with any tolerance setting, getting it right is key for the proper functioning of a product. Monte Carlo method allows you to consider and use the appropriate models for the variations that will exist across your components. [Read more…]
A continuous distribution is useful in many statistical applications such as process capability, control charts, and confidence intervals about point estimates. On occasion time to failure, data may exhibit behavior that a normal distribution models well. [Read more…]