1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

S chart Upper control limit

Discussion in 'SPC - Statistical Process Control' started by Jackson Yeo, May 14, 2024.

  1. Jackson Yeo

    Jackson Yeo New Member

    Joined:
    May 14, 2024
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    1
    Hi,
    This is Jackson Yeo, 1st time posting question her for help.

    Currently we have implemented SPC for past 2 years with more than 100 of machines in our production lines. Most of the parameters have only either upper or lower spec limit, as a result, our control limits are turn on for only on sided limit, either LCL or UCL.

    One of the problem is, we have some parameters with only lower spec limit and lower control limit, (ex two components bonding force), that mean, we expect the data point to be as high as possible so that it will be far away from the LCL trigger limit, good is, we could find some problem most of the time whenever the Xbar chart are triggered due to low Xbar points.

    As for S chart, our control limits are set at only high control limit, we ignore the low limit because we are happy if the standard deviation is at near to zero.

    However, there has been on and off trigger of S chart alarm on high limit due to few data point at too high side and widen the standard deviation. as a result, the machines were shutdown unnecessary for repair due to few data points are too good. Since our control is so strict that each machine shall be shutdown for any SPC trigger, as a result, the OEE has been impacted unnecessary by this kind of invalid high S chart trigger. Because of this, some production personnel are also requesting to remove the S chart control since it's not able to reflect the true picture.

    My question is, is there any control limit for S chart to only look into standard deviation causes by data point is drifting low instead of both high and low ? Of course, many may argue that the data at too high side shall be also trigger for improvement study, I fully agree, however, this can be studied off line instead of trigger alarm for machine shutdown unnecessary.
     
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    591
    Likes Received:
    512
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    You appear to be confused as to the purpose of control charts and their relationship with spec limits (Hint: there is no relationship). Control limits are designed to alert you to a change in the process. PERIOD. That change may be perceived as bad or as good. That may inform you as to the appropriate reaction you should take in response to the change, but the limits have fulfilled their purpose by alerting you. What you are asking for, in essence, is how you can silence half those alerts. While you may view that as getting rid of a nuisance, in reality, you are 1) increasing the variation of your process by allowing it to wander, and 2) passing up opportunities to improve the process by investigating what causes the process to move in a more desirable direction.
     
    Andy Nichols and Bev D like this.
  3. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    613
    Likes Received:
    672
    Trophy Points:
    92
    Location:
    Maine
    What Miner said. I too have had many many control charts. There are times when a signal represents something GOOD or something BAD that doesn’t mean out of spec parts. An increasing SD (beyond the properly calculated upper SD control limit) is always a cause fo concern and should be investigated.

    The other misunderstanding that is hurting you is that any OOC alert is a reason for shutting down equipment. It is not - that is an engineering decision. If the OOC condition can be investigated offline and/or it doesn’t pose an increasing risk for out of spec parts then you can continue operation.

    SPC si not intended to be a blind thing. Knowledge, experience and understanding of variation and your process must be applied for the successful utilization of Control Charts.

    One more thing: OEE is an imperfect metric that can be useful when applied in conjunction with other metrics. Meeting build plan that is tuned to demand with a quality part that has minimal variation is the over riding metric.
     
    Andy Nichols likes this.