THURSDAY, NOVEMBER 23, 2017
Font Size

MBBinWI

  • 0 practical defects? A CPU or CPL of around 1 isn’t great–especially if you have shifts over time in process performance for the long term.

    Reducing product cost is GREAT but I could never advocate a ppm defective rate that was appreciable to save money when the defective rate may cost more depending on many things such as customer…[Read more]

  • Well they may have designed the metric to update daily and with the technique you demonstrated, it will keep from double/over counting those that were late and not delivered (e.g. late ones that aren’t delivered aren’t in the denominator until they are shipped that day and then they are in the numerator and denominator).

  • hmmm @mbbinwi :)

  • I wish I could help but get confused. You quote 90% and 65% OEE but also quote in a later post that you have 50-65% uptimes which are in conflict with quoted OEE’s.

    Don’t forget my suggestion on modeling software to “see” how the process might perform after getting the stats clear enough.

  • If you know your typical sources of variation, you can set subgroups up to monitor the sources of those variation.

    There is no absolutely correct subgroup size. It’s purely a decision for monitoring and maintaining good process control assuming the actions plans are active and functional.

  • It means you don’t have to focus on improving this process for quality control issues. However, are there cost benefits to raising the mean much closer to the USL? Less processing, cheaper processing, fewer ….

    Just 2 cents.

  • A simulation model would be great but some confusion exists reading the blog. If the lot size is 500 parts which are created in less than 30 seconds, it seems you’d limit operation of steps A and B since step C is your bottleneck or else you’ll pile up inventory before Step C.

  • @gabriel
    I wanted to compliment you on your in depth thinking during your posts.

  • @68rs327 You asked your question on a Friday night. You’re more likely to get a faster response during the work week.

  • Monte Carlo simulations are a tool. I’m confused…average weekly demand is more than A, B, and C which total to 300 units.

    However, don’t forget about inefficiencies and losses. I don’t see why basing capacity is thought to be 1025–how many effective days do you have?

    Just a few quick thoughts.

  • Some questions:

    1. How many 2 level variables are there?
    2. Can all of these 2 level variables be considered continuous?
    3. The 4 holes – it sounds like you are treating these as categorical. Do you have to do this or can they be characterized by something like diameter which would be a continuous variable?
    4. When you say “some treatments…[Read more]

  • @mike-carnell https://www.youtube.com/watch?v=vo9AH4vG2wA

    boy, reminds me of the old days (outside of work)….

    Hope all is well with you my friend, colleague, …. Let me know when you get your visitors from RSA and maybe we’ll plan something?

  • 1. Gabriel, if you think of a statapult there are typically more than 2 settings (2 holes) for a factor but run a DOE to confirm magnitude and statistical confidence in the impact of that factor–pick 2 holes. Who knows you may find the factor has minimal impact without having to test all 4 holes. I’m just saying you may not want to use all of…[Read more]

  • I might be mistaken but I don’t see any data. You have a list of stuff that you could have sat down in a corner and produce3d yourself without doing any “brainstorming” what ever that is supposed to mean.

    It sounds like you sat in some classroom, closed your eyes and used the force to pick a problem. I don’t even see a problem. How many days are…[Read more]

  • The null hypothesis is always an equals statement i.e. one thing equals another (no difference hence “null”) You already have it in the words just put it in some jargon like X1 = X2.

    The alternate is just a difference greater than, less than or if you don’t care about direction just basic does not equal.

  • some thoughts…

    1. Remember, you’re doing a DOE to confirm cause/effect and strength of impact by each factor. I’d confirm with just using 2 of the 4 settings for the one factor and confirm impact and/or strength of interactions. You can always do another DOE after you find something.
    2. If your “gage” isn’t very good, that will be a…[Read more]

  • 2 thoughts come to mind.

    1. Get MBB training IF they have mentor/coach/teach as part of their agenda.
    2. Mentoring your GB’s are a great practice. Also consider your “clients” for the projects helping them understand what you’re doing.

  • fyi….the hypothesis testing tests “a relationship”. It doesn’t confirm “cause and effect” which can be observed with experiments such as DOE or other basic tests.

  • that’s a fairly worded statement. Be aware you’ll need a p-value (probability) to determine your line of acceptance of the hypothesis tests.

    Now you need to understand whether your two variables are continuous, discrete, etc. and pick the correct hypothesis test.

    Good luck…good work so far!

  • Did your training emphasize getting a project sponsor and being sure to attack business pain points? Do you have data on types of absenteeism? I’d say narrow down focus to …. IF you can get a project sponsor or champion.

  • Load More


5S and Lean eBooks
GAGEpack for Quality Assurance
Six Sigma Statistical and Graphical Analysis with SigmaXL
Six Sigma Online Certification: White, Yellow, Green and Black Belt

Lean Six Sigma Project Tracking

Lean and Six Sigma Project Examples
Six Sigma Online Certification: White, Yellow, Green and Black Belt

Find the Perfect Six Sigma Job

Login Form