Designation:D7720−11(Reapproved2017)StandardGuideforStatisticallyEvaluatingMeasurandAlarmLimitswhenUsingOilAnalysistoMonitorEquipmentandOilforFitnessandContamination1ThisstandardisissuedunderthefixeddesignationD7720;thenumberimmediatelyfollowingthedesignationindicatestheyearoforiginaladoptionor,inthecaseofrevision,theyearoflastrevision.Anumberinparenthesesindicatestheyearoflastreapproval.Asuperscriptepsilon(´)indicatesaneditorialchangesincethelastrevisionorreapproval.1.Scope1.1Thisguideprovidesspecificrequirementstostatisticallyevaluatemeasurandalarmthresholds,whicharecalledalarmlimits,astheyareappliedtodatacollectedfromin-serviceoilanalysis.Thesealarmlimitsaretypicallyusedforconditionmonitoringtoproduceseverityindicationsrelatingtostatesofmachinerywear,oilquality,andsystemcontamination.Alarmlimitsdistinguishorseparatevariouslevelsofalarm.Fourlevelsarecommonandwillbeusedinthisguide,thoughthreelevelsorfivelevelscanalsobeused.1.2Abasicstatisticalprocesscontroltechniquedescribedhereinisrecommendedtoevaluatealarmlimitswhenmea-suranddatasetsmaybecharacterizedasbothparametricandincontrol.Afrequencydistributionforthiskindofparametricdatasetfitsawell-behavedtwo-tailnormaldistributionhavinga“bell”curveappearance.Statisticalcontrollimitsarecalcu-latedusingthistechnique.Thesecontrollimitsdistinguish,atachosenlevelofconfidence,signal-to-noiseratioforanin-controldatasetfromvariationthathassignificant,assignablecauses.Theoperatorcanusethemtoobjectivelycreate,evaluate,andadjustalarmlimits.1.3Astatisticalcumulativedistributiontechniquedescribedhereinisalsorecommendedtocreate,evaluate,andadjustalarmlimits.Thisparticulartechniqueemploysapercentcumulativedistributionofsorteddatasetvalues.Thetechniqueisbasedonanactualdatasetdistributionandthereforeisnotdependentonapresumedstatisticalprofile.Thetechniquemaybeusedwhenthedatasetiseitherparametricornonparametric,anditmaybeusedifafrequencydistributionappearsskewedorhasonlyasingletail.Also,thistechniquemaybeusedwhenthedatasetincludess...
