iSixSigma

Piece Count Measurement System

Six Sigma – iSixSigma Forums Old Forums General Piece Count Measurement System

Viewing 14 posts - 1 through 14 (of 14 total)
  • Author
    Posts
  • #29848

    billybob
    Participant

    Hello Possum Lovers,
    My project involves daily production output.  How do I verify the measurement system when the piece count is taken from a computer screen? 
    I know I am concerned that the equipment counter is counting correctly, but how about the person who records the data from the computer screen to a sheet of paper, and then the accuracy of the next person doing data entry into the production software, and then my entry into minitab….all possible sources of data error?  
    How do i verify “people skills” that their combined inputs of the same data into the measuring system that I am taking data from is correct when their is more than one possible source of data error of the very same data?
     
     

    0
    #77148

    Edwards
    Participant

    Billybob,
    I know from recent postings that you have just completed GB training.  Not sure what you’ve been taught…but sounds like you’re feeling that you need to verify the measurement system.  Are you sure that it’s causing you a problem?  When I was newly graduated I thought that checking the measurement system was a must do in every project.  But now only visit them if they are causing problems…If you get bogged down in this you could end up checking how every operator uses a micrometer, caliper etc.
    Sounds like there are a few non value add steps in the process…operator takesreading from computer and writes it down, passes it on etc.  Try to eliminate these NVA steps, mistake proof the process.
    David
     

    0
    #77151

    James A
    Participant

    OK Billybob, let me suggest the following:-
    1) Long term, you need to simplify the system – so that the data gets entered once into a data ‘pool’ and can be accessed by all interested parties direct from there – so reducing errors.  Your systems/IT people should be able to come up with a few good(?) ways of doing this – a relational/SQL type database seems like a good place to start
    2) You may have to resort to manually checking all the data yourself initially – you know, if you want a job doing right, DIYFS.  Once you have a level of confidence in the data collection and distribution, then you can at least prove you’ve considered the problem, and acted accordingly. If you find it’s a problem, then you need to come up with a solution to that issue first, and then go back to the real ‘project’ work.
    Don’t worry about seemingly going round in circles – iteration will always occur in projects, and does help (tho’ it is a bit frustrating).
    3) If the person immediately behind you in this long chain of repetitive data entry currently puts the data into anything vaguely Microsofty, then you can just copy and paste the numbers direct into Minitab (saving hours of typing, and allowing you to go home on time to feed the pining possums).
    Does this help?
    James A

    0
    #77159

    melvin
    Participant

    David:
    You wrote: “When I was newly graduated I thought that checking the measurement system was a must do in every project.  But now only visit them if they are causing problems…”
    How on earth are you going to know if a measurement system is giving you problems if you do not take steps to statistically validate it?
    This type of mentality is dead wrong – old school – and not in the spirit of any data driven process improvement initiative.
    No one is suggesting that a mockery is made of things by fine-tooth combing every last number that is crunched.  There are ways to validate every measurement system and, if done correctly, the results can be as insightful and rewarding as the process improvements efforts that typically follow.

    0
    #77164

    Cannizzo
    Participant

    I agree fully with Bob’s reply.
    Billybob, I have some thoughts on your MSA but don’t have time this moment. I promise to write more later.
    –Carol

    0
    #77169

    billybob
    Participant

    Cool; and thanks!
    Later,
    Billybob

    0
    #77178

    Mike Carnell
    Participant

    Billybob,
    I’m coming in with David and James. If your process map shows the piece of paper going to some filing cabinet or some other NVA type activity then the easiest way to eliminate the transcription (that’s copying in Billybob speak) errors is to eliminate the operation. If you take out all the NVA first then you will only have to worry about the things in the process that are affecting customers (the value chain). Those are technically the only things you get paid for so nobody really care how well you do the NVA stuff.
    If you had raccoons you wouldn’t need more time to feed them. They find their own food. The down side is they wash it in the swimming pool (cement pond in Billybob speak).
    Good luck.

    0
    #77191

    melvin
    Participant

    ….considering what Mike and James have said and reiterating what I have said:
    I agree that it is always best to simplify a system – measurement or not – as much as possible.  This is almost always a fail safe way to decrease variability.  With that said, once a measurement system is simplified and free from non value-added steps, it must be validated.  Failing to validate the measurement system is like lying to yourself – similar to assuming there is actually a market for possums.

    0
    #77195

    Cannizzo
    Participant

    Billybob,
    Sorry for the delay in response. You wrote: “How do I verify the measurement system when the piece count is taken from a computer screen?” So I believe you want to ensure data consistency and stability. A few others are talking about simplifying your measurement system, and that good but I’m going to talk about how to verify it.
    First, you should determine the factors that could cause the measurement of piece counts to vary. You mentioned four factors in your message (which is great that you’ve identified them already):

    equipment counter is counting correctly
    person who records the data from the computer screen to a sheet of paper
    the accuracy of the next person doing data entry into the production software
    Billybob’s entry into Minitab
    Data is consistent if any two people who measure the same things arrive at essentially the same answer. Data is stable if the measurement remains consistent over time.
    Gage R&R studies are used to validate measurement systems used to collect continuous data. Since your values are discrete (e.g. 1, 2, 3, 4…), we need to find another validation technique.
    For discrete data, you must have a master set of materials to examine for defects or a validated checker to correctly determine whether or not the opportunities (what the person recording the data from the screen sees) are a defect or not.
    Here’s an example. I once ran a project where I was looking at cycle time for a repair process. Items arrived via FedEx in the morning and entered the process. I needed to verify when they arrived and was faced with a written time sheet that would be attached to the item; the time sheet would list the time the item entered the shipping/receiving area. Now I could have sat there every day for a week to verify that the different clerks didn’t delay in putting time sheets on the items and that the time/date was correct, but that would have caused the “halo effect” (when watched, operators suddenly perform without error). What I ended up doing was calling up FedEx and asking them to fax me the month’s worth of delivery sheets so I could see by item, the time/date they were signed for in the shipping/receiving area.
    Once you have your master set of answers, you verify against the data. You can then determine your accuracy: the number of opportunities and the number of piece parts that the operator wrote down incorrectly (defects). I’ve always used 90% as the cut-off for acceptability of discrete MSA. I wouldn’t use lower than that, but you can increase it if you feel it’s necessary for the situation.
    So how do you do it in your specific problem? “Give a man a fish and he’ll eat for a day. Teach a man to fish and he’ll eat for life.” Unfortunately, I don’t have an answer for you without more details on your exact situation. I hope the above information and example did help to some degree. Let me know if my reply is off base with what you were asking.
    –Carol

    0
    #77197

    billybob
    Participant

    Carol,
    Thanks, its a running start for me to consider!  I see I can verify data entered into the “system” because we hold the form the data is taken from for a short period of time.  But my biggest concern is the data taken at the machine from the computer screen, that the digets aren’t transposed or the wrong value written down.
    I appreciate the fact that some PLCs can be fed directly into the system to remove operator data error.  That is not a cheap process for a factory filled with equipment, nor is data entry really the problem from the higher level or lower level.
    Thats my first “dig” with 6S, I need to spend a lot of time verifying something I know is not wrong, so then I can move on to fixing the real problems that occur within the actual process.  I’ll tone that down a bit, my confidence that while the operator packs the box with product we count the full box and compare it to the PLC; they are the same, so I am confident without doing  MSA that my counter is correct, but I must still jump through that hoop to make my BB happy.
    Jeepers; I’m getting to serious, the project must be getting into my head.  At least it’s Friday!
    Later,
    Billybob
     

    0
    #77203

    Cannizzo
    Participant

    Billybob,
    I think you hit an an all-too-often-forgotten point that the tools are here for our benefit, not to create extra hoop for us to jump through. If you don’t believe that you have a measurement system issue, then no MSA is necessary. I (as an MBB) would ask you, “how do you know unless you measure it once?” It’s a vicious cycle.
    It’s also tough for your MBB in another way. S/he need to help you pick the right tools to use without wasting extra energy, but also needs to balance that with the fact that you are “training” and should use many of the tools to ensure that you understand them (even if they don’t find anything relevant or significant). Your next project may need it and by that time you’ll be a Black Belt without a coach, or a colleague will ask you for help and it would be terrific if you knew the answer because you’ve used the tool in the past.
    Direct feeds of data (a potential solution to an issue) is a terrific idea if defects are occuring due to operators. We haven’t necessarily determined that to be the case yet. I look forward to hearing more about your project as you progress.
    Have a good weekend,
    –Carol

    0
    #77347

    Andrew M. Brody
    Participant

    After reading David’s and Carol’s responses to you, Billy Bob, I tend to agree with David that there are too many “touches” in your system, and, ideally, direct data input into minitab would be ideal.  However, I can also relate to the dilemma that the expense is prohibitive.
    Unfortunately, I don’t know what kind of software you are using to pick up your initial data before it is “touched by human hands”.  One option that I have used under different circumstances is to find a way to print out a snapshot of your data screen (even if it is by camera), and then make a direct comparison to the data entered into Minitab after it is entered (obviously).
    To avoid the “halo” effect, you may have to get creative (sneaky, deceptive, etc.) to get your picture .
    Good Luck,
    Andy
     

    0
    #77352

    john beaudoin
    Participant

    You could be wasting a lot of time as the others said…
    However, you should plot the daily numbers that are being produced now to see if they are in control.  Look for any points outside the 3 Sigma limits (Special Causes).  If you have none, you just saved time.  If there are some, you can look at them to see if there was an error in the process you mentioned that caused it.  If so, you can implement things to eliminate these.  Then, once in control, move onto something that will save you money and save your customers grief.

    0
    #77357

    Swaggerty
    Participant

    BillyBob,
                  You may want to contact your IT/IM person and see if there is a FTP function you can use. I run tons of reports in our WMS system and FTP straight to EXCEL. I’ve been with with my current company 2 years and when I first got here people were running reports on WMS, printing them out and then typing them into excel. FTP has been around forever, but often it is forgotten about and not used.

    0
Viewing 14 posts - 1 through 14 (of 14 total)

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