iSixSigma

MSA for online meters in continuous proces

Six Sigma – iSixSigma › Forums › Old Forums › General › MSA for online meters in continuous proces

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • April 7, 2005 at 7:03 am #38948

    SVD
    ★ 10 Years ★
    Member

    Ours is continuous process (24 hr, 365 days), we have chosen metric to improve efficiency of process. The measurement is from flow meters (transmitter) installed on the process lines.
    When it comes to MSA of these meters, we can’t stop the process to check flow meters.
    If anybody has done such study , pl enlighten me on this subject.

    0
    April 7, 2005 at 12:47 pm #117354

    Schuette
    ★ 10 Years ★
    Participant

    You don’t do an MSA on flow meters – you calibrate them to a known standard.  They should have come with calibration documentation when they were purchased/installed, along with recommended recalibration intervals.  You should have replacements, so that you can pull one and replace it with a fresh calibrated unit while the pulled one is recalibrated.  If you don’t have the equipment, find a contract service that will do it.

    0
    April 7, 2005 at 12:59 pm #117355

    NS
    ★ 10 Years ★
    Participant

    You are mixing up “accuracy” with “precision”.  Calibration will solve the first issue, and should be done prior to the Gage.
    Of course you can do a Gage R&R on continuous measurements.  You have to make the assumption that for “short” spurts or periods of time that the reading is stable (ie. the unit is at the same setting, or gives the same results).  This works best with on-line measurements systems.  You then repeat for different levels or settings.

    0
    April 7, 2005 at 3:48 pm #117373

    Schuette
    ★ 10 Years ★
    Participant

    NS,
    I disagree – you calibrate for a range of flows and pressures, and that’s it.  How are you going to R&R an in-line flowmeter and discount pump flucuations, viscosity changes, etc.  You would have no idea what the actual velocity and pressure is vs the flow meter error.  You can an idea of the total variation of pressure and flow rate as you set the pump to different flow rates and pressures, but you can’t isolate the meter results. You are dependent on how consistent the pump is.  If you do it as a destructive R&R, you are assuming the flow is constant, which I don’t believe would be a good assumption.  Which is pump fluctuation, and which is meter flux?
    Walk me through an example of how you would do a R&R on an inline flow meter and get any meaningful data on the meter.  I’m interested.

    0
    April 8, 2005 at 8:17 am #117413

    Anonymous
    ★ 20 Years ★
    Guest

    I agree with your position Jim … it is also worth noting that the accuracy of a rotometer or flow meter is usually specified as a percentage of ‘full scale deflection.’
    In the early days of Six Sigma, we found three significant sources of variation in furnaces:
    – Flow meters working at 95% of fsd – instead of 50%
    – variation in the position of spike thermo-couples due to cantilever vibaration and post preventative maintenance
    – gas flows to low for the volume of the tube – this was the fault of management because they always stated no one should change the process – instead of everyone should understand the process!
    Regards,
    Andy
     

    0
    April 8, 2005 at 4:48 pm #117437

    NS
    ★ 10 Years ★
    Participant

    Say you had a continuous in-line flowmeter for natural gas.
    Set the reading at 1000.  Measure a “burst (20?)” of short-term readings.  The time-frame would be very short, and depends on the instrumentation in your plant (how fast  you measure and record).  Do the same for 1500, 2000, 2500, etc.  You go to Minitab, parts=levels that you chose, 1 operator (unless you are lucky enough to have 2 of these units in the same pipe – unlikely), and 20 or so replicates per level.
    This has nothing to do with calibration.  This was already performed before the Gage or MSA.  We are doing a Repeatability Gage, not a Bias study.
    You assume that the readings were constant over the “burst”.  For most in-line flowmeters this is reasonable over a 10 second period.  If you pass the Gage R&R, you can assume this to be a “worst case” scenario.  As you mentioned, there will be some “actual” variation due to the “gas properties or flows” so the real Gage values will be better than you calculated in Minitab.  If you fail (unlikely with most instruments), then you don’t know if it was your instrument or if your gas was actually changing its properties.
    Hope this clarifies.

    0
    April 9, 2005 at 10:01 am #117464

    Anonymous
    ★ 20 Years ★
    Guest

    NS,
    I don’t understand your advice. Can you please provide an example, even if you just make up the data. Perhaps you could use Minitab to generate it?
    From my point of view, flow meters are calibrated off-line. When the calibration is performed a number of measurements are taken at various set-points and compared against a master. We can then estimate the tolerance (variation) at the particular set point. While it is true that it can be done at a single set point, it is preferable to do it over the entire range since processes are often adjusted to compensate for variation in raw materials.
    In your scheme, you appear to repeat measurements at the same set point, but how do you know what the actual flow is if the flow meter is still connected? Do you infer it from a process response, or are there other means of measuring flow?
    Andy

    0
    April 11, 2005 at 5:32 am #117499

    SVD
    ★ 10 Years ★
    Member

    Thanks Andy, Jim and NS
    We have calibration schedule in place , it is followed as per vendor recommendations. People have doubt whether indicated flow is correct? – i.e. accuracy which depends on installations, sensing lines, inernal orifice bore intactness etc. We don’t have other measures to check accuracy in readings- like pump flow setting , diecrence in tank level rise etc. Basically it is used for steam, air , nitrogen etc and hence accuracy is always questioned. I feel many more BB’s may be facing such problem in there measure phase and wanted to know the solution for this.
    Does any body ommitted these from metric selection and went for better metric selection to capture the efficicncy of process? How?
    SVD

    0
    April 19, 2005 at 12:52 pm #118019

    Manfred
    ★ 10 Years ★
    Participant

    The easiest way is, to install an second one Flow meter in an bypass. So you have even an redundant system.

    0
    May 5, 2005 at 4:06 am #118908

    SVD
    ★ 10 Years ★
    Member

    Thanks for the suggestion. Where it is possible we are providing bypass with provision for new meter. But it is not possible for all lines as we can not afford closing down continuos operation (rarely we get opportunity of plant shutdown).
    Can we go ahead with only callibration of transmitters only, which is not a MSA (R&R)as per Six SIgma methodology.
    SVD

    0
  • Author
    Posts
Viewing 10 posts - 1 through 10 (of 10 total)

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