We've all pondered this question: Which came first, the chicken or the egg? While I'm fascinated by this question I won't bore you with my theories. However, I have visited many manufacturing sites and sometimes there is a little confusion on whether to collect data from certain processes using the Process or Test functions in the Data Entry Configuration. Which is best, Process or Test?
Example: A beverage manufacturer is using an automated filler (called Juice Filler01) with 10 individually adjustable nozzles to fill 10 bottles at a time. The testing requirement is to measure the weight of each bottle from each nozzle every 30 minutes. There is a tare weight collected earlier in the process and it's brought back as a database value.
Some of the manufacturers I have visited use the Test function to collect data. They have actually set up individual tests for each weight. For example, they listed "Juice Filler01" as the Process. The Tests would include a Tare Wt. (database value), ten (10) Gross Wt Tests represented as Wt Row01(Gr), Wt Row02 (Gr), and so on through Wt_Row10(Gr), and ten (10) Net Wt Tests represented as Wt Row01(Net), Wt Row02 (Net), and so on through Wt_Row10(Net). That's 21 tests in one configuration. Sure, they would be able to see the data in the view data chart, average the results, and chart the data in a graph, but is there another approach and what if there was more than one head on that filler?
Since each nozzle is individually adjustable, they should be treated as separate processes. Using this approach, there are actually ten (10) Processes on Juice Filler01 (Juice Filler01_Nozzle01, Juice Filler01_Nozzle02, and so on through Juice Filler01_10). These processes would be grouped under Juice Filler01. If there was more than one head on this filler, the Process Groups would be Juice Filler01_Head01 and Juice Filler01_Head02 and the processes would be called JF1_H1_Nozzle01 for head 1 and JF1_H2_Nozzle01 for head 2, etc. This would allow the user to calculate capability by head and nozzle. The Tests are simply Tare Weight (database value), Filled Weight (gage or keyboard), and Final Weight (calculation) for a total of three (3) in this configuration.
How would this look in the data entry configuration?
In the Required Items tab section D. Process: Juice Filler01 (Juice Filler01_Nozzle01)* would be selected. Juice Filler01 is the process group, while Juice Filler01_Nozzle01 is the Process. The * indicates that there have been reselect options identified. Within the properties for the Process, there are two Reselect Options: 1. Selectable by User + and 2. Auto sequence through all group items +. Both would be enabled in this case. Option 1 would not change from the original configuration, but Option 2 would allow all of the weights from each nozzle to be collected at one time starting with Nozzle01.
How would this change the way the data is analyzed or viewed?
In the original configuration, the data selection has the Process Juice Filler01 selected and all of the Final Weight Tests from each row selected. The View Data chart displays the Final Weight from each row from left to right across the screen, but may cause the user to have to scroll to the right in order to review all of the data. Another Test would be required to average the results from say rows 1-5 or rows 6-10. Since there are 10 individual tests, there would be 10 individual Control Charts (one for each row) and capability could be calculated for each row. A Box & Whisker Plot categorized by Test would include all Final Weight data for each row.
In the new configuration, the data selection has the process Juice Filler01_Nozzle 01 (Include All Group Members) selected. The view data chart displays the Final Weight for each process vertically in one column and can be reviewed without scrolling to the right. Since there is more than one process, but only one test, there would only be one Control Chart, but it would display all of the Processes in one graph. So if one nozzle had an issue, it would stand out. Control Charts could be created for each nozzle if desired. This would allow the user to see the true capability of each nozzle. A Box & Whisker Plot categorized by Process would include all Final Weight data for each nozzle.
When creating a new configuration using this same approach, the user only needs to select a different process and it's ready to go.
So, which is better? That's up to you. Just remember that while both methods obtained similar results, if it's adjustable, it's a Process and should be treated as such.