Blindly following the SPC concepts, if you would have corrected your process to below the centerline, then the 7 point test would "restart" the count. If not, it keeps on going. Remember, the ultimate point of these tests is to determine a significant change in process, and then to determine a corrective action. Adjustment of the process is not always the mandatory corrective action. The real question is why are there so many points above a mean that you are expecting to be randomly bouncing back and forth across? Has the process shifted? For example, is something wearing? That requires maintenance, not adjustment - a different reaction. Knowledge of the process is as important as the red flag of the chart.
By the way, not all processes randomly jump back and forth across the centerline. They follow a sawtooth form, which is expected to have runs - but just in one direction (either up or down). Trying to force a process with a sawtooth curve to randomly jump across the centerline will make a sane man mad. Does your process have runs in one direction (up or down) all of the time? It could quite likley be expected, and no action would be required. It has to be supported with logical process knowledge, however, such as ongoing tool wear (as in grinding). Food for thought.
Community metadata Could not be loaded. No pre-configured community forum parent found for the incoming fid: . You need to add this forum into custom_code/community.sql
Send to friend
Comments
bobdoering 1/10/2002
Blindly following the SPC concepts, if you would have corrected your process to below the centerline, then the 7 point test would "restart" the count. If not, it keeps on going. Remember, the ultimate point of these tests is to determine a significant change in process, and then to determine a corrective action. Adjustment of the process is not always the mandatory corrective action. The real question is why are there so many points above a mean that you are expecting to be randomly bouncing back and forth across? Has the process shifted? For example, is something wearing? That requires maintenance, not adjustment - a different reaction. Knowledge of the process is as important as the red flag of the chart.
By the way, not all processes randomly jump back and forth across the centerline. They follow a sawtooth form, which is expected to have runs - but just in one direction (either up or down). Trying to force a process with a sawtooth curve to randomly jump across the centerline will make a sane man mad. Does your process have runs in one direction (up or down) all of the time? It could quite likley be expected, and no action would be required. It has to be supported with logical process knowledge, however, such as ongoing tool wear (as in grinding). Food for thought.