How to do a Reliability analysis in JMP

[#1]

Hi everyone,

I am designing a new process and I want to calculate the likelyhood of components in the process (such as pumps, etc) to break down. This in order to determine how to manage maintenance, excess stock etc. 

How would I need to go about this? 

[#2]

Have you checked out the different Reliability and Survival Methods JMP does offer from JMP Help

https://www.jmp.com/support/help/en/17.2/#page/jmp/introduction-to-reliability-and-survival.shtml#

[#3]

Hi Jarmo,

Thank you for your reply.

There are a lot of options on this platform. Am I correct that if I want to determine the Mean time to failure I should use the Life distribution analysis?

Another question I have is that since I'm still designing my process, you think it is possible to use the profiler platform to simulate failure data?

~Veerle

[#4]

I'm not familiar with designing this type of systems or reliability statistics on this level (but I know JMP has platforms which can do something related to those), so hopefully someone else is able to comment who knows more about systems like these. I'm not exactly sure who to tag but I try @Mark_Bailey 

[#5]

It is not clear to me where you are and where you are going. Let's try a couple of questions:

1) There are "calculate" and "estimate". Why do you say "calculate"?

2) Regardless of "calculate" or "estimate", you need to do so based on some information. What is that information that you have?

[#6]

So the process as such doesn't exist yet. We have some labscale data but of course these are different components, so whilst we can use the data from a scientific point of view, in terms of reliability it is not usefull. Therefore we will have to rely either on historical data from other processes using similar equipment or on literature data.

I want to figure out how much redundancy I should have in my process, what the work in progress should be in order to keep the flow going if one of the components needs to be serviced etc. 

It's hard because it's still theoretical and we don't have data on the process itself. But we have to start somewhere...

[#7 · Accepted solution]

Your situation is no uncommon. Based on what you said, I can at least point two directions to investigate further.

1) Accelerated Life Test; see Accelerated Testing or Chap. 17-19 in Statistical Methods for Reliability Data or Chap. 8 in Applied Reliability. JMP Platform is Fit Life by X for one acceleration factor. This is a type of analysis which allows you to extrapolate the reliability metrics from your limited lab data to reliability metrics under the circumstances that you don't have data.

2) Depending on whether your system of interest is repairable or not, see Reliability Block Diagram or Repairable Systems Simulation. They require a JMP Pro license. They are tools for you to build bigger systems using smaller components which you have reliability information. And you can do "what-if" to tweak individual components' reliabilities to understand the bigger system's reliability. I see that you mentioned maintenance, so Repairable Systems Simulation is likely relevant. However, depending on your system and the details that you want to model, the complexity can quickly build up. In that case, I suggest starting with simple cases.

[#8]

Thanks! I will try that.