Siemens PLC Reliability Numbers - MIL STD 217

Anyone familiar with the MIL STD 217 and 217F as to how the two are related ?

I have been looking for Siemens 555-1105, and i have not been able to find the MTTF or MTBF for it. I have tried to contact Siemens and they only provide the numbers as per MIL STD 217F, which makes no sense to me. Maybe someone with more exposure to this help me out.

555-1101 CPU tested under MIL STD 217
Hours 109,795
Years 12.53

555-1101 CPU tested under MIL STD 217F
Hours 8762
Years 1.00

555-1105 CPU tested under MIL STD 217F
Hours 14179
Years 1.62

555-1105 CPU tested under MIL STD 217
Hours ?
Years ?

rajy2r said:

Anyone familiar with the MIL STD 217 and 217F as to how the two are related ?

Click to expand...

They are one and the same. 'F' is just the latest revision, it came out in 1991, although there has been a couple of newer notices that affect the document.

This method for determining MTBF is a simple parts count method. You add up all the failure rates of all the components and that is the failure rate of the unit.

It looks like Siemens has done this twice (for some modules). With the updates or changes to the operating assumptions they have two different calculated values.

I'm not sure which is the latest value.

The estimated MTBF of the 555-1105 is 14179 hours or about 1.62 years of operation. So, if you have one of these beasties running 9 to 5, five days a week, you can expect to replace it once every seven years. Run it 24/7/365 expect to replace it in about 1-1/2 years.

With this method they don't use real data of field returns, so it isn't all that accurate. It helps compare two systems in regard to complexity but that is about it.

As with all statistics, it is really only useful when you have an entire fleet and are trying to figure out how many of a particular type of unit will be down or how many spares you will need to buy over their lifetime.

If I had a thousand of these I would want to budget and schedule for replacing 50 of them every month. If I had one or two, it's a total **** shoot. They may fail tomorrow or next year, no way of telling.

Sorry for replying back so late.

If they are the same standard, then why does the MTBF goes from 12.53 to 1 year ? I am not sure why the numbers are so different based on the two. Does this mean they had it wrong the first time ?

Well i am working on a project with a directive to make sure that the PLC has a minimum mean time between failure of 250,000 hours. (28.53 Years).

I looked at AB 1756-L61 and the MTBF is 1,000,000 (114 Years).

I wouldn't expect to replace a PLC every year ? From the 555-1101 CPU tested under MIL STD 217 a figure of 12.53 looks more realistic rather then 1 year ?

Maybe it has something to do with how old Siemens (TI) 555 is. I doubt there is many new ones anymore.

AB 1756-L61 is much newer CPU so i cant find those comparable in any means.

rajy2r said:

If they are the same standard, then why does the MTBF goes from 12.53 to 1 year ? I am not sure why the numbers are so different based on the two. Does this mean they had it wrong the first time ?

Click to expand...

I wish I could say why the difference. Differences can appear based on operating assumptions (temperatures, voltages, etc). I wouldn't necessarily expect a twelve-fold change.

rajy2r said:

Well i am working on a project with a directive to make sure that the PLC has a minimum mean time between failure of 250,000 hours. (28.53 Years).

I looked at AB 1756-L61 and the MTBF is 1,000,000 (114 Years).

I wouldn't expect to replace a PLC every year ? From the 555-1101 CPU tested under MIL STD 217 a figure of 12.53 looks more realistic rather then 1 year ?

Click to expand...

MIL-STD_217 does tend to overestimate the failure rate. Nothing is actually 'tested.' You simply count all the parts. Add up the failure rates of the those parts as listed in the standard and you get the estimated MTBF of the assembly.

I don't think it serves as a model for realistic performance. Rather it serves to compare two similar items on an even footing.

AB on the other hand uses data from field returns, which generally tends to underestimate the failure rate.