Can anyone tell me what the difference is between system and component reliability target vales? I have a new project that is being launched and this is part of my planning requirements.

Can anyone tell me what the difference is between system and component reliability target vales? I have a new project that is being launched and this is part of my planning requirements. They are at different ends of the reliability spectrum. Component reliability is the reliability of each of the "pieces" of the product, subsystem, and/or system (e.g., a switch, diode, connector, clamp, etc.), and the target is generally the historic reliability (or a little better with improved designs) of past component performance.

System reliability is the composite reliability of all components and/or 'subsystems made up of many components'. The targets established for system reliability are based on calculating all of the 'pieces', whether they are in series or parallel, and other factors. These calculations require some knowledge and background into reliability engineering. I've tried to simplify it as much as possible above. Please tell us a little more about your specific situation.

Consider a car engine. Component reliability is how well each part works -- the fuels pump is 99.2% reliable, the spark plugs are 99.996% reliable, etc... System reliabilty is how well the whole system works when put together. System reliability can be improved by such approaches as including parallel components and redundancy.

Tim F

Tim,

I hate to be so ignorant on this subject but what do you mean by "System reliability can be improved by such approaches as including parallel components and redundancy" What do you mean by parallel components and redundancy? I am not a reliability engineer and not very familiar with any terms associated with reliability. Thanks for your help!

Jerry

Parallel means two (or more) items that both perform a particular task at the same time. For example, a truck might have dual tires, so that if one fails, the truck can still operate - perhaps not quite at 100%, but well enough to get by until repairs can be made.

You can also have standby components or systems. For example, an emergency generator for a hospital. This systems only kicks in when there is a failure in the primary system. The space shuttle has - I believe - 5 computers! to manage key tasks. Even if 4 of them fail, the fifth can still handle the task.

Both of these set-ups would be considered redundant, since they are not needed as long as things are working properly.

Tim


P.S. I moved & merged the posts from two other threads. I don't see any forum specifically for reliability, but this seemed like a good choice.

Tim makes good points of which I only alluded to (i.e. series vs. parallel systems). Not to be a killjoy, but like I said earlier, if you do not have some background here, it might be difficult to accomplish your task.

Again, we might be able to help if you give us some details on the specific product you are working on and what you are expected to accomplish.

The project is the launch of a new entrance door on a school bus.

The old door has premature seal failures and frame failures. The new door takes care of these issues and is cheaper as well.

The project is the launch of a new entrance door on a school bus.

The old door has premature seal failures and frame failures. The new door takes care of these issues and is cheaper as well.

Component Failure are:

Seals
Hinges
Glass
Material (metal that rusts, plastic that cracks, etc.)

System Failure would be the door fails to work.

An example of a redundant system would be if the door opened electronically, but there was still a manual door open in the event the electronic one did not work.

Does that help?

This link may be helpful.
http://www.weibull.com/systemrelwebcontents.htm

It provides a basic overview of the reliability field.

Thanks to all of you very much. This is a lot clearer now.

Excellent start.

Now, list all of the components on the old door which could independently cause the door to fail to operate (to make it easier, list only the components which were most likely to fail or, based on past history of all of the components from those style doors that did fail). Determine each of their failure rates (number of failures / time period). Roughly speaking, 1 - failure rate = reliability.

For example, if a seal failed 20 times over a 1000 hours expected usage, then the failure rate is 0.02 and the reliability (1-fr) is 0.98. If some frame component's resulting reliability (using same formula) was .985 and two other components had a .99 reliability, then .98 x .985 x .99 x .99 = 0.946 (or 54 failures in 1000 hours); the old door was 94.6% reliable

Each of these results are the respective "Component Reliabilities". Multiply them together and you get your "old door's" System Reliability. Obviously, make sure you include the failure rates for the seal and frame issues you noted.

Do the same thing for the new door. Using my example above, and assuming you solved the problem of the seal and frame component completely, only the problems with the two other components remain, that is, .99 x .99, and your new reliability is 98.01% (or 19 failures in 1000 hours).

This again is oversimplified, but hopefully makes the point.

Can anyone tell me what the difference is between system and component reliability target vales? I have a new project that is being launched and this is part of my planning requirements.

Probably a bit late for this discussion but my experience is that people start with an assumed system level failure rates and then allocate it downwards to respective failure categories. ie/ reliability budgeting.

For example : Electronic system failures may be x FIT of which 40% may be attributable to software, 40% electronic components and 20% manufacturing/assembly defects of the system. You work with your suppliers and team to ensure the budget is met.

Alternatively, you could go to your suppliers / teams and get their inputs on what the expect #'s to be. (Human nature is to pad).

If you want to reduce pro-activily, work with tools such as FMEA or DOE, to drive down probability of occurance. Other philosophies for reducing errors that could cause system failures are with software error trapping, redundancy (ie/ if 1 component fails have a backup or if a measurement has possibility of false failure, do measurement multiple times and vote on it) or if consumer safety, make it fail safe.

Be wary of component vs System performance. As a component supplier, I have seen many times an IC that works to datasheet but not in it's the application that customer is using it. Sometimes, parameters have not been specified.