iSixSigma

New to Reliability Testing

Six Sigma – iSixSigma Forums Old Forums General New to Reliability Testing

  • This topic has 3 replies, 2 voices, and was last updated 18 years ago by A.B.
Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #31561

    Aidan
    Participant

    I am very new to reliability testing, unfortunately my manager is expecting me to be an expert myself. Basically I need to perform a reliability testing on a disposible medical device (one time use only). There will not be time involved in this case, but the # of cycle involved here. The physical will use it for one time or one cycle only, and then the device is scrapped. However, to establish a certain level of confident, I need to show that the device is good at least X cycles. My idea of performing the testing is to 1. Determine the sample size 2. Determine a fixed cycle T to run the units to see if they survive or fail – Nonrepairable system 3. Analyze the data.
    I am kind of doing this from scratch. It’s much appreciated if you can help me with it. Thanks!

    0
    #83279

    A.B
    Participant

    What is the item under study ? How does it fail ? Do you have prior reliability data ?
    A.B

    0
    #83293

    Aidan
    Participant

    The item under study is flowrate (cc/sec), tested using a flow tester. The acceptance criteria now is 10cc/sec. However, based on previous experience, the result is going to be either mostly 0 cc/sec (good) or 0.5-1.5 cc/sec. How shall I design the reliability testing?

    0
    #83319

    A.B
    Participant

    As far as I understand your disposable ( non repairable ) device is not concerned by infant mortality nor by wearout types of failures.
    Therefore it is recommended to use the exponential distribution as the lifetime distribution model to do the testing.
    You are using a Type II Censoring of Data. Number of failures is  known in advance and constant ( = 1 )
    At this stage it’s good to use “Time Intervals” to have a good or ( approximate ) figure of the TTF ( Time To Failure ).
    Start reliability testing of 5 to 10 samples ( might be less or more depending on the resources available for the test ).
    Record Times To Failure ( TTF ).
    Calculate MTTF – ( Sum of TTFs / number of samples tested ). Note at this stage you can analyze the TTF figures and compare them to  MTTF in order to define the “Cycle” unit as a function of time.
    Calculate the failure rate Lambda – 1/MTTF or 1/Cycle.
    You can then use the Exponential Distribution Formula and other associated equations to calculate reliability figures.
    A.B

    0
Viewing 4 posts - 1 through 4 (of 4 total)

The forum ‘General’ is closed to new topics and replies.