iSixSigma

Entering the Improve Phase – Now What?

Six Sigma – iSixSigma Forums Old Forums General Entering the Improve Phase – Now What?

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #51736

    RC
    Participant

    I am working on my first project after completing 2 week Six Sigma Green Belt training. After hard work through Define, Measure, and Analyze phases, I will soon be working on the Improve phase of my project.My project involves improving a current machine group that is not holding tolerances. This is resulting in high scrap and severely impacting throughput of the cell in which they are located.I have begun a process FMEA (Failure Mode Effects Analysis) highlighting big players in the product of the two machines. I’d like to know if my following plan of action is good and if anyone has any suggestions about how to proceed in this very critical phase of the DMAIC process.I plan on taking information I have gathered from the FMEA and attacking the variables that have the highest ‘score’, that is, process variables that have the biggest impact on the features of the product. (As I write this I am getting a flood of information in my brain about critical few x’s etc.)So with this surge of knowledge and recollection (being that I am not a full time six sigma green belt, it is more an auxiliary task) I know that I have to identify the critical few inputs that are affecting my project Ys: scrap and production.I will take my critical few x’s and do somewhat of an iterative process on them? That is, I’ll examine the x’s and pretty much fall back to the measure phase as I look more closely at these variables? As I write THIS I am wondering if I am in fact ready for the Improve phase and am really still in the Analyze phase…Anyway, I suppose my real question is this: when I enter the Improve Phase, how can I improve my identified critical few x’s to achieve a better performing machine? Do I do things like install SOPs? Make machine adjustments? Basically a Design of Experiments correct? This is all new to me, I hope I’m not in over my head. My Black Belts are somewhat hard to get a hold of so I appreciate any help or insight you can provide.

    0
    #180196

    Remi
    Participant

    Hai RC,
    You are almost finished with Analyse = you have determined a list of X’s that COULD be the critical X’s. But this list was composed by OPINION (tool used = FMEA).
    First: try to check with data if those opinions were correct. Use already available data (X’s already measured compared with the Y by H0-testing; Tools are Graphs, ANOVA and Regression).
    Second: try to influence the Y by manipulating the X’s (tool = DoE). Because only then can you  find out if you have enough influence to improve the Y.
    Exception: X’s like “machine broken”, “tool worn”,'”wrong software/ setting used” etc you just repair.
    Then calculate from the DoE what the best settings for your X’s is so that the Y has its best performance and check this with a pilot run (confirmation of theoretical DoE results)
    Then you are ready for the Control phase: making sure that it stays improved.
     
    Good luck

    0
    #180199

    SiggySig
    Member

    The only way to know for sure whether you have the right X’s is to try out your improvement in a pilot. Then measure the results. Good luck.

    0
    #180210

    Ron
    Member

    Remi,
    I do not know who taught you your GB methodology but your veerbiage seems to be confused.
    In the Define Phase you have defined your specific target area and problem statement.
    In the Measure phase you have completed your detailed process map, identified the Critical and controlable variables, you have also identifed the noise varialbes that may be affecting your process. Youi have gathered data which you have fitlered down to the 80 20 rule rule states that if you fix these 20 % of problems you will improve 80% of the process and have rewritten your problem statement to address this new focused view based on data.
    In  the Analyze phase you have statistically determined which variables have a direct impact on your problem statement.
    In the improve phase you are brainstorming solutions to you verified root cause issues. You are coming up with solutions that will fix the problem you have identifed. You will also create an action plan to implement these changes. Yo8u will redo an FMEA with the implemented changed in place to show the improvement. You will verify by running pilot projects that the improvments you put inplace have indeed reduced your defects. You will create training plans to pass this knowledge off to the process owners and you will calculate the financiall impact of your actions.
     
    Hope this helps

    0
    #180235

    Remi
    Participant

    Uh Ron,
    do you combine Improve and Control into one improve phase?
    Remi

    0
    #180236

    Vallee
    Participant

    RC,First off you have just realized that each step of DMAIC is NOT a one-way process. You will find that you may be missing information or need to redefine the information you thought was right in the beginning…. as you can see from the arguments between Ron and Remi. I won’t get in the middle of that one.Caution though about scope creep, not saying that scope can not change but must be made based on valid data and management support.Your question however is how to I improve the problems I have identified…. if you are asking this at this stage you do not know the root causes and need to go back to Analyze and maybe measure.

    0
    #180257

    Mike Carnell
    Participant

    RC,
    I am probably going to get a fair amount of push back on this but I will give you my opinion.
    It doesn’t matter what phase you are in. At this point you seem to be a bit confused on what to be looking at for a response variable. You are looking at scrap and production. You said it is because you cannot hold tolerance. If that is a fact then stick with tolerance (you should be able to see that in a capability study and in a control chart). The project Y may be production and scrap but that will be a function of holding tolerance. Holding tolerance will become a Y that is a function of _________?
    What is it about the tolerance. Is it a location issue (mean) or is it a variation issue or is it both. You should be looking at which X’s are affecting the mean and which are affecting the variation. Whatever you have seen in the project that lead you to the conclusion that you couldn’t hold tolerance should tell you which catagory you are dealing with.
    I am not sure what type of machine you are dealing with or what type of material but for the most part machining is pretty well understood in terms of feeds and speeds. You should be able to sort out most of this wih hypothesis testing which is a less intrusive tool than DOE and much easier to control. Sort out as much as possible with the hypothesis testing so if you do have to run a DOE you don’t take any more variables into the DOE than necessary (extra variables will increase cost, complexity and control of a DOE). If you are going in with more than 5 you probably need to talk to someone that can help you with your data.
    This next part is pure opinion. Watch how the parts are setup and run. Talk to the people running the machines particularly if it is expensive material. Depending on what type of machine and the cost of the material they will frequently set up and sneak up on a dimension slowly so there is room for rework if something goes wrong – like the kid that advised his grandfather on staying inside the lines when you color – slow down when you get close to the line. It will skew your data to one side of the specification.
    You are supposed to fix the problem and at some level I am sure someone cares about the order you use the tools. Spend more time figuring out what information you need to know and what tool will give you that information and you will do fine. Nobody goes to statistical jail for doing something out of order – just have a rational line of thought behind what you did.
    Check your measurement systems – all of them. Understand the effects of temerature if you are working to tight tolerances. The price of a gage doesn’t assure it is either accurate or precise.
    Just my opinion.

    0
    #180260

    Vallee
    Participant

    Mike,Why should there be push back on your guidance? People often think that DMAIC steps are exclusive from each other but they are not. You and I both know that the flow of the process is the intent. If you don’t gain focus and purpose at the beginning, scope creep and resource management can get out of control. If you don’t define your end state and goal based on valid measures you are stuck in mud. If you attempt to improve symptoms and not causes you are just blowing in the wind. Finally, if you think the world is stable and control does not need to be monitored… just wait for the next mistake you created to appear. By the way, I’m not referring to you in the above statements. So the issue as stated before with RC is that he does not really understand what is wrong and if it really needs to be fixed. So he must go back and review what his scope should be and what data he actually has.

    0
    #180263

    Ron
    Member

    Uh No!

    0
    #180326

    Jonathon Andell
    Participant

    Mike:
    At some point you’ll simply have to stop making sense. Perhaps that way you can avoid push-back.
    J

    0
Viewing 10 posts - 1 through 10 (of 10 total)

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