iSixSigma

Help in making decision

Six Sigma – iSixSigma Forums Old Forums General Help in making decision

Viewing 16 posts - 1 through 16 (of 16 total)
  • Author
    Posts
  • #44414

    Harsh
    Participant

    Hello to all,
    I really need some guidence from you guys. Here is my problem.
    I am having scrap rate of 14% average since last 15 days. Normally  we have the scrap of 5 to 8%. 
    Defect that causes scrap : Misrun: factors affecting the defects are die temp, furnace temp, pouring rate, tilting rate, cycle time, riser height etc. What should i use to identify which variable is affecting the defect most.?  because i am not in decision making position, i can not run DOE as it is very hard to explain to orthodox people. So is regression analysis is good to begin with>? Please advise.
    Thanks in advance,
    HARSH
     

    0
    #142230

    Adam L Bowden
    Participant

    Then your only option might just be to go and spend time with the folks working there – ask them what’s up  – I’m sure they know.
    Adam

    0
    #142231

    Stevo
    Member

    Have you asked the front line associates yet?  They might be able to give you direction.  Practical – graphical – analytical.
     
    Stevo
     

    0
    #142234

    Harsh
    Participant

    Hi,
    thanks for encourement for a young person like me..
    HARSH

    0
    #142238

    michael spearman
    Participant

    have you checked the Upper and Lower Spec limits of each step in the process, do you have inspectors or operator @ each station, either or – do they have to check the product before it leaves there station, if so you can perform a Gage R&R, to find out if they can spot the defect. A DOE is really the route to go. Check your scrap for multiple defects, each station could be adding to the problem, covering up the true nature of the defect. When one station could be the real contributor, to the cause.
    Hope this helps

    0
    #142263

    harsh sharma
    Participant

    Thanks micheal,
    Yes , we do have check points. Operator identified as many defective as they can. But i am trying to find which variable is affecting this defect. Here is what i am thinking to set an experiment.  ” 20castings/day/turntable. Measure them for misrun defect. At the same time, measure all the variables. run a Regression analysis for multiple variable.. Get the P value and hopefully, i will be able to identify which variable is affecting significantly.” sounds right to you???
    Thanks in advance.
    HARSH

    0
    #142283

    michael spearman
    Participant

    Harsh, that is a good way to look @ it, be careful – make sure your operators can spot the defect(s) – if they can’t, you will be back @ square one again. Start with your operators, next your machinery.
    Just a thought………

    0
    #142303

    JSK
    Participant

    Thanks for the advice. I am myself going to floor and take the reading and measuring the defect. I am not involving any operator in this.
    Harsh
     

    0
    #142326

    BBPT
    Participant

    Harsh,
    In my opinion, DOE is the best solution, but as you mentioned as impossible, than regression may be quiet helpful.
    regards,
    BBpt
     

    0
    #142332

    GDS
    Participant

    Harsh,
    You have received some good advice, however the key thing that caught my eye was that your scrap has doubled in the last 15 days, the first thing I would look at is what changed 15 days ago? New operator or inspector? New supervisor? New production metrics?Different raw material? What in the process has changed? Have these defects always been there and someone is just now finding or are they now producing new defects.  I do feel you can get the answers for these questions from the folks on the line.
     

    0
    #142334

    VP
    Member

    Hi Harsh
    As I look at your problem, I understand that your process defects are ‘Scraps’ and cause for the defect to occur is a ‘mis-run’ and factors or root causes for the cause or ‘mis-run’ to occur could be the variations in process parameters. Correct me if my understanding is incorrect.
    I see you’ve listed many parameters that would have influence over ‘mis-run’. When you’ve prepared this list, i’m sure you would have good knowledge over the process as well. Therefore, before jumping into statistical tools, sit down and do a logical analysis of what could be the effect of each those parameters on your process performance based on your technical knowledge. Think about what can really go wrong and to what extent in each of those parameters based on your current process controls  and instruments (system design). This would help screen those insignificant factors. If you do this exercise with real interest, most likely you would identify the root cause for your problem. If you still have not found it, then go for a statistical tool to understand better the relationship between your process parameters and output.
    regards
    VP

    0
    #142378

    Kal
    Participant

    Good, you would’ve probably finished the job by now… If in case you are still searching for solution… like someone had adviced in this tread, first look for difference in the Input Variables (be it your process parameters, people, base matl etc..etc..) , screen them for differences.. How many shots you make a day? I hope that should give you ample samples to understand the difference in actual process from your control spec…. Should be a known reason that you need to search by discussing with people or screening data… No need for DOE till your target is to reduce it further down below 6%… If you end up using Logistic for now,  I doubt if you may be able to conclude something… Honest advice, talk to people, fix the thing that has got changed.. Also update your FMEA and CONTROL PLAN documents for your followers to be pre-informed about this..

    0
    #142423

    Harsh
    Participant

    Thanks everyone,
    I really appreciate the response from everybody. But friends, here is my experience that i want to share with you. As i told you before i am not in position to run DOE, i thaught i should run Regression Analysis. When i prepared my Minitab sample expt to show my COO what is really  regression analysis is? he told me not to interfere on the scrap issue. So i backed up now…. But i am still trying to get permission from upper mgmt. 
    HARSH

    0
    #142610

    Jonathon Andell
    Participant

    It may be worth considering a multi-vari test. If you develop your sampling scheme effectively, the resulting patterns may reveal predominant sources of process variation.

    0
    #142631

    Eric Maass
    Participant

    Hey Jonathon!! Long time no see!!!(My email is [email protected] if you’d like to email me directly)
    Okay, on to the topic:
    When I’ve used Multi-Vari Charts, I also like to do Nested ANOVA (or the equivalent in GLM, General Linear Model) to get values for the variances for the sources of variation.
    Best regards,Eric Maass

    0
    #142634

    Jonathon Andell
    Participant

    Grretings Eric. I’ll shoot a “sociable” email separately.I totally agree about doing the ANOVA afterward. Once the data have been collected, there’s benefit to doing both the graphical and the statistical analysis. As you well know, the big challenge with multi-vari is in preparing an effective sampling scheme in the first place.

    0
Viewing 16 posts - 1 through 16 (of 16 total)

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