How to calculate MTBF of a system based on subcomponents

To provide better detail, our company has bid our product for a potential government contract. Their response has requested a MTBF of the system. I have the following questions?

1. If our product is a protoype, can a MTBF been performed?

2. Without official tests can MTBF be calculated?

3. If I have the MTBF data from our suppliers for subcomponents, what is the easiest way to calculate and combine these values to represent the system?

I can do the simple math of MTBF, but how can I compile these to reflect a system MTBF?

Thanks!

MTBF is a statistical prediction and actual performance may never match the prediction. Find yourself a copy of mil-hdbk-217 to do the system prediction.

JkelleyCDS said:

To provide better detail, our company has bid our product for a potential government contract. Their response has requested a MTBF of the system. I have the following questions?

1. If our product is a protoype, can a MTBF been performed?

2. Without official tests can MTBF be calculated?

3. If I have the MTBF data from our suppliers for subcomponents, what is the easiest way to calculate and combine these values to represent the system?

I can do the simple math of MTBF, but how can I compile these to reflect a system MTBF?

Thanks!

Click to expand...

The characteristic life parameter can not be added, from components, to determine system performance...but the reciprocal (hazard rate) can. So, add the hazard rates then determine the life parameter with this resulting number.

janedoe said:

The characteristic life parameter can not be added, from components, to determine system performance...but the reciprocal (hazard rate) can. So, add the hazard rates then determine the life parameter with this resulting number.

Click to expand...

To clarify:

• Lambda = 1 / MTBF
• Lambda System = Lambda component 1 + Lambda component 2 + Lambda component n
  
• MTBF System = 1 / Lambda System
  

1. Yes
2. Yes but they would only be Predictions (which may or may not satisfy your customer)
3. Yes but ...

Change MTBF (in hrs) to Failures In Time (number) by
Failures In Time = Time Period / MTBF
for MIL-217F, Time Period is 10^6 hrs.
for SR-332, Time Period is 10^9 hrs.

Add all FITs together,
Use same formula to calculate system level MTBF.

FITs (and therefore MTBFs) are also dependent on operating and environmental conditions. So you need more than just a single MTBF from your supplier.

Regards.

All of the previous answers you got are correct, however all assume that its an electronic system for which the exponential distribution is a good assumption. If however your system includes mechanical assemblies this may no longer be correct

w_grunfeld said:

All of the previous answers you got are correct, however all assume that its an electronic system for which the exponential distribution is a good assumption. If however your system includes mechanical assemblies this may no longer be correct

Click to expand...

Excellent point that we all overlooked.

Depending on how the reliability of each component affects the others in the system as a whole may have an effect on the expected MTBF. In this case Reliability Block Diagrams (RBD) or Fault Tree Analysis may lead to a useful predictive answer to an overall MTBF. Pardon me as I cannot yet post links. Try searching for Reliability Block Diagram on Google or Wikipedia to get an idea of what this method provides.

There are also software packages that can be used for RBD simulation, the user can create one and assign failure and repair rates to the individual components and get an answer for Availability/MTBF of the system based on a simulated amount of time such as expected life or proposed deployment of a system. Relevant applications include:

ARINC Raptor
Relex OpSim
Reliasoft BlockSim

Reliability block diagrams are better to use compared to Fault tree analysis. Your works is in success line in RBD while it is in failure line in the Fault Tree. This diagrammatic method will show you that how component reliability contributes to the success or failure of the system. Looking on the success combination using RBD, you have to get a well written predictive answer.

danwsc said:

Change MTBF (in hrs) to Failures In Time (number) by
Failures In Time = Time Period / MTBF
for MIL-217F, Time Period is 10^6 hrs.
for SR-332, Time Period is 10^9 hrs.

Add all FITs together,
Use same formula to calculate system level MTBF.

FITs (and therefore MTBFs) are also dependent on operating and environmental conditions. So you need more than just a single MTBF from your supplier.

Regards.

Click to expand...

I found this link http: // www. sqconline. com/reliability/resistor1.html that is based on 217. - OBSOLETE DEAD LINK UNLINKED

How accurate is the data obtained from this online calculator??