ASTM D7720-21

Standard Guide for Statistically Evaluating Measurand Alarm Limits when Using Oil Analysis to Monitor Equipment and Oil for Fitness and Contamination

SIGNIFICANCE AND USE
5.1 Alarm limits are used extensively for condition monitoring using data from in-service lubricant sample test results. There are many bases for initially choosing values for these alarm limits. There are many questions that should be addressed. These include:  
Are those limits right or wrong?
Are there too many false positive or false negative results?
Are they practical?  
5.2 This guide teaches statistical techniques for evaluating whether alarm limits are meaningful and if they are reasonable for flagging problems requiring immediate or future action.  
5.3 This guide is intended to increase the consistency, usefulness, and dependability of condition based action recommendations by providing machinery maintenance and monitoring personnel with a meaningful and practical way to evaluate alarm limits to aid the interpretation of monitoring machinery and oil condition as well as lubricant system contamination data.
SCOPE
1.1 This guide provides specific requirements to statistically evaluate measurand alarm thresholds, which are called alarm limits, as they are applied to data collected from in-service oil analysis. These alarm limits are typically used for condition monitoring to produce severity indications relating to states of machinery wear, oil quality, and system contamination. Alarm limits distinguish or separate various levels of alarm. Four levels are common and will be used in this guide, though three levels or five levels can also be used.  
1.2 A basic statistical process control technique described herein is recommended to evaluate alarm limits when measurand data sets may be characterized as both parametric and in control. A frequency distribution for this kind of parametric data set fits a well-behaved two-tail normal distribution having a “bell” curve appearance. Statistical control limits are calculated using this technique. These control limits distinguish, at a chosen level of confidence, signal-to-noise ratio for an in-control data set from variation that has significant, assignable causes. The operator can use them to objectively create, evaluate, and adjust alarm limits.  
1.3 A statistical cumulative distribution technique described herein is also recommended to create, evaluate, and adjust alarm limits. This particular technique employs a percent cumulative distribution of sorted data set values. The technique is based on an actual data set distribution and therefore is not dependent on a presumed statistical profile. The technique may be used when the data set is either parametric or nonparametric, and it may be used if a frequency distribution appears skewed or has only a single tail. Also, this technique may be used when the data set includes special cause variation in addition to common cause variation, although the technique should be repeated when a special cause changes significantly or is eliminated. Outputs of this technique are specific measurand values corresponding to selected percentage levels in a cumulative distribution plot of the sorted data set. These percent-based measurand values are used to create, evaluate and adjust alarm limits.  
1.4 This guide may be applied to sample data from testing of in-service lubricating oil samples collected from machinery (for example, diesel, pumps, gas turbines, industrial turbines, hydraulics) whether from large fleets or individual industrial applications.  
1.5 This guide may also be applied to sample data from testing in-service oil samples collected from other equipment applications where monitoring for wear, oil condition, or system contamination are important. For example, it may be applied to data sets from oil filled transformer and circuit breaker applications.  
1.6 Alarm limit evaluating techniques, which are not statistically based are not covered by this guide. Also, the techniques of this standard may be inconsistent with the following alarm limit selection techniques: “rate-of-change,...

General Information

• Status: Published
• Publication Date: 30-Sep-2021
• ICS: 25.040.40 - Industrial process measurement and control
• Technical Committee: D02 - Petroleum Products, Liquid Fuels, and Lubricants
• Drafting Committee: D02.96.04 - Guidelines for In-Services Lubricants Analysis

Relations

• Refers: ASTM D445-24 - Standard Test Method for Kinematic Viscosity of Transparent and Opaque Liquids (and Calculation of Dynamic Viscosity)
• Effective Date: 01-Apr-2024
• Refers: ASTM D4378-24 - Standard Practice for In-Service Monitoring of Mineral Turbine Oils for Steam, Gas, and Combined Cycle Turbines
• Effective Date: 01-Mar-2024
• Refers: ASTM D7647-24 - Standard Test Method for Automatic Particle Counting of Lubricating and Hydraulic Fluids Using Dilution Techniques to Eliminate the Contribution of Water and Interfering Soft Particles by Light Extinction
• Effective Date: 01-Feb-2024
• Refers: ASTM D4175-23a - Standard Terminology Relating to Petroleum Products, Liquid Fuels, and Lubricants
• Effective Date: 15-Dec-2023
• Refers: ASTM D6786-15(2023) - Standard Test Method for Particle Count in Mineral Insulating Oil Using Automatic Optical Particle Counters
• Effective Date: 01-Dec-2023
• Refers: ASTM D6299-23a - Standard Practice for Applying Statistical Quality Assurance and Control Charting Techniques to Evaluate Analytical Measurement System Performance
• Effective Date: 01-Dec-2023
• Refers: ASTM D7484-23a - Standard Test Method for Evaluation of Automotive Engine Oils for Valve-Train Wear Performance in Cummins ISB Medium-Duty Diesel Engine
• Effective Date: 01-Nov-2023
• Refers: ASTM D6224-23 - Standard Practice for In-Service Monitoring of Lubricating Oil for Auxiliary Power Plant Equipment
• Effective Date: 01-Nov-2023
• Refers: ASTM E2412-23a - Standard Practice for Condition Monitoring of In-Service Lubricants by Trend Analysis Using Fourier Transform Infrared (FT-IR) Spectrometry
• Effective Date: 01-Nov-2023
• Refers: ASTM D445-23 - Standard Test Method for Kinematic Viscosity of Transparent and Opaque Liquids (and Calculation of Dynamic Viscosity)
• Effective Date: 01-Nov-2023
• Refers: ASTM D4175-23e1 - Standard Terminology Relating to Petroleum Products, Liquid Fuels, and Lubricants
• Effective Date: 01-Jul-2023
• Refers: ASTM D7416-09(2020) - Standard Practice for Analysis of In-Service Lubricants Using a Particular Five-Part (Dielectric Permittivity, Time-Resolved Dielectric Permittivity with Switching Magnetic Fields, Laser Particle Counter, Microscopic Debris Analysis, and Orbital Viscometer) Integrated Tester
• Effective Date: 01-May-2020
• Refers: ASTM D7484-18 - Standard Test Method for Evaluation of Automotive Engine Oils for Valve-Train Wear Performance in Cummins ISB Medium-Duty Diesel Engine
• Effective Date: 01-Oct-2018
• Refers: ASTM D5185-18 - Standard Test Method for Multielement Determination of Used and Unused Lubricating Oils and Base Oils by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES)
• Effective Date: 01-Apr-2018
• Refers: ASTM D6299-17b - Standard Practice for Applying Statistical Quality Assurance and Control Charting Techniques to Evaluate Analytical Measurement System Performance
• Effective Date: 15-Dec-2017

Overview

ASTM D7720-21: Standard Guide for Statistically Evaluating Measurand Alarm Limits when Using Oil Analysis to Monitor Equipment and Oil for Fitness and Contamination provides a comprehensive approach for statistically determining and adjusting alarm limits used in condition monitoring of industrial equipment through oil analysis. These alarm limits are critical for distinguishing machinery wear, oil degradation, and contamination, and enable maintenance teams to prioritize actions based on reliable warnings. The standard addresses consistent evaluation techniques, improving dependability and practical application in machinery upkeep and oil condition monitoring.

Key Topics

• Alarm Limits in Oil Analysis: Defines alarm limits as threshold values that trigger various levels of warning about the fitness, contamination, or wear of equipment and lubricants.
• Statistical Evaluation Methods:
  • Statistical Process Control (SPC): Recommended for parametric (normally distributed and controlled) data sets, SPC helps establish objective control limits and refine alarm thresholds on quantifiable test results.
  • Cumulative Distribution Technique: Applicable to both parametric and nonparametric data, this method uses the observed distribution of data points to set and evaluate alarm limits, making it effective even with skewed or single-tail distributions.
• Alarm Levels: Describes four typical alarm levels-WHITE (normal), GREEN (acceptable), YELLOW (warning), and RED (critical)-and their use in classifying test results from oil analysis.
• Data Set and Population Definition: Emphasizes the necessity of using statistically representative historical data from similar equipment operating under comparable conditions.

Applications

The guide is relevant for industries utilizing machinery with lubricated systems, including but not limited to diesel engines, turbines, hydraulics, pumps, and oil-filled electrical equipment such as transformers and circuit breakers. Practical applications include:

• Condition Monitoring Programs: Implementing or refining alarm limit systems to provide actionable, data-driven responses to machinery or lubricant anomalies.
• Predictive Maintenance: Enhancing maintenance strategies by minimizing false positives/negatives and focusing resources where the risk of equipment failure or lubricant contamination is highest.
• Quality Control in Oil Laboratories: Offering standardized statistical techniques for laboratories analyzing in-service oil samples, ensuring alarm limits remain meaningful as data evolves.
• Fleet and Industrial Asset Management: Supporting large-scale operations in categorizing equipment populations, tracking wear, and optimizing oil change intervals based on statistically valid alarm thresholds.

Related Standards

ASTM D7720-21 references and complements a number of other important standards within the domain of oil quality and condition monitoring. Prominent related ASTM standards include:

• D445: Test Method for Kinematic Viscosity of Transparent and Opaque Liquids
• D664: Test Method for Acid Number of Petroleum Products
• D2896: Test Method for Base Number of Petroleum Products
• D4378: Practice for In-Service Monitoring of Turbine Oils
• D6224: Practice for In-Service Monitoring of Lubricating Oil in Auxiliary Power Equipment
• D6299: Practice for Applying Statistical Quality Assurance and Control Charting Techniques
• D6595, D5185: Methods for Wear Metal and Contaminant Analysis in Lubricants
• E2412: Practice for Condition Monitoring by Trend Analysis Using FT-IR Spectrometry

Practical Value

Adopting ASTM D7720-21 enables organizations to validate and adjust alarm limits using industry-approved statistical methods, increasing the dependability of oil analysis programs. This not only supports proactive decision-making but also aligns with international best practices, reducing equipment downtime, maximizing asset life, and optimally timing maintenance interventions. By formalizing alarm threshold evaluation, maintenance teams can improve the accuracy of condition-based recommendations and support overall equipment reliability initiatives.

Standards Content (Sample)

This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
Designation: D7720 − 21
Standard Guide for
Statistically Evaluating Measurand Alarm Limits when Using
Oil Analysis to Monitor Equipment and Oil for Fitness and
Contamination
This standard is issued under the fixed designation D7720; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
1. Scope* surand values corresponding to selected percentage levels in a
cumulative distribution plot of the sorted data set. These
1.1 Thisguideprovidesspecificrequirementstostatistically
percent-based measurand values are used to create, evaluate
evaluate measurand alarm thresholds, which are called alarm
and adjust alarm limits.
limits, as they are applied to data collected from in-service oil
analysis. These alarm limits are typically used for condition
1.4 This guide may be applied to sample data from testing
monitoring to produce severity indications relating to states of
of in-service lubricating oil samples collected from machinery
machinery wear, oil quality, and system contamination. Alarm
(for example, diesel, pumps, gas turbines, industrial turbines,
limits distinguish or separate various levels of alarm. Four
hydraulics) whether from large fleets or individual industrial
levels are common and will be used in this guide, though three
applications.
levels or five levels can also be used.
1.5 This guide may also be applied to sample data from
1.2 A basic statistical process control technique described
testing in-service oil samples collected from other equipment
herein is recommended to evaluate alarm limits when mea-
applications where monitoring for wear, oil condition, or
suranddatasetsmaybecharacterizedasbothparametricandin
control. A frequency distribution for this kind of parametric system contamination are important. For example, it may be
data set fits a well-behaved two-tail normal distribution having applied to data sets from oil filled transformer and circuit
a “bell” curve appearance. Statistical control limits are calcu-
breaker applications.
lated using this technique.These control limits distinguish, at a
1.6 Alarm limit evaluating techniques, which are not statis-
chosen level of confidence, signal-to-noise ratio for an in-
ticallybasedarenotcoveredbythisguide.Also,thetechniques
control data set from variation that has significant, assignable
of this standard may be inconsistent with the following alarm
causes. The operator can use them to objectively create,
limit selection techniques: “rate-of-change,” absolute
evaluate, and adjust alarm limits.
alarming, multi-parameter alarming, and empirically derived
1.3 Astatistical cumulative distribution technique described
alarm limits.
herein is also recommended to create, evaluate, and adjust
alarm limits. This particular technique employs a percent
1.7 The techniques in this guide deliver outputs that may be
cumulativedistributionofsorteddatasetvalues.Thetechnique
compared with other alarm limit selection techniques. The
is based on an actual data set distribution and therefore is not
techniquesinthisguidedonotprecludeorsupersedelimitsthat
dependent on a presumed statistical profile.The technique may
have been established and validated by an Original Equipment
be used when the data set is either parametric or
Manufacturer (OEM) or another responsible party.
nonparametric, and it may be used if a frequency distribution
1.8 This standard does not purport to address all of the
appears skewed or has only a single tail. Also, this technique
safety concerns, if any, associated with its use. It is the
may be used when the data set includes special cause variation
responsibility of the user of this standard to establish appro-
in addition to common cause variation, although the technique
should be repeated when a special cause changes significantly priate safety, health, and environmental practices and deter-
or is eliminated. Outputs of this technique are specific mea- mine the applicability of regulatory limitations prior to use.
1.9 This international standard was developed in accor-
dance with internationally recognized principles on standard-
This guide is under the jurisdiction of ASTM Committee D02 on Petroleum
Products, Liquid Fuels, and Lubricants and is the direct responsibility of Subcom- ization established in the Decision on Principles for the
mittee D02.96.04 on Guidelines for In-Services Lubricants Analysis.
Development of International Standards, Guides and Recom-
Current edition approved Oct. 1, 2021. Published October 2021. Originally
mendations issued by the World Trade Organization Technical
approved in 2011. Last previous edition approved in 2017 as D7720 – 11 (2017).
DOI:10.1520/D7720-21. Barriers to Trade (TBT) Committee.
*A Summary of Changes section appears at the end of this standard
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
D7720 − 21
2. Referenced Documents and Derived Kinematic Viscosity of Liquids by Oscillat-
2 ing Piston Viscometer
2.1 ASTM Standards:
D7484 Test Method for Evaluation of Automotive Engine
D445 Test Method for Kinematic Viscosity of Transparent
Oils for Valve-Train Wear Performance in Cummins ISB
and Opaque Liquids (and Calculation of Dynamic Viscos-
Medium-Duty Diesel Engine
ity)
D7596 Test Method for Automatic Particle Counting and
D664 Test Method for Acid Number of Petroleum Products
Particle Shape Classification of Oils Using a Direct
by Potentiometric Titration
Imaging Integrated Tester
D974 Test Method for Acid and Base Number by Color-
D7647 Test Method for Automatic Particle Counting of
Indicator Titration
Lubricating and Hydraulic Fluids Using Dilution Tech-
D2896 TestMethodforBaseNumberofPetroleumProducts
niques to Eliminate the Contribution of Water and Inter-
by Potentiometric Perchloric Acid Titration
fering Soft Particles by Light Extinction
D4175 Terminology Relating to Petroleum Products, Liquid
D7670 Practice for Processing In-service Fluid Samples for
Fuels, and Lubricants
Particulate ContaminationAnalysis Using Membrane Fil-
D4378 Practice for In-Service Monitoring of Mineral Tur-
ters
bine Oils for Steam, Gas, and Combined Cycle Turbines
D7684 Guide for Microscopic Characterization of Particles
D4928 Test Method for Water in Crude Oils by Coulometric
from In-Service Lubricants
Karl Fischer Titration
D7685 Practice for In-Line, Full Flow, Inductive Sensor for
D5185 Test Method for Multielement Determination of
Ferromagnetic and Non-ferromagnetic Wear Debris De-
Used and Unused Lubricating Oils and Base Oils by
termination and Diagnostics for Aero-Derivative and Air-
Inductively Coupled Plasma Atomic Emission Spectrom-
craft Gas Turbine Engine Bearings
etry (ICP-AES)
D7690 Practice for Microscopic Characterization of Par-
D6224 PracticeforIn-ServiceMonitoringofLubricatingOil
ticles from In-Service Lubricants by Analytical Ferrogra-
for Auxiliary Power Plant Equipment
phy
D6299 Practice for Applying Statistical Quality Assurance
E2412 Practice for Condition Monitoring of In-Service Lu-
and Control Charting Techniques to Evaluate Analytical
bricants by Trend Analysis Using Fourier Transform
Measurement System Performance
Infrared (FT-IR) Spectrometry
D6304 Test Method for Determination of Water in Petro-
leum Products, Lubricating Oils, and Additives by Cou-
3. Terminology
lometric Karl Fischer Titration
D6439 Guide for Cleaning, Flushing, and Purification of
3.1 Definitions:
Steam, Gas, and Hydroelectric Turbine Lubrication Sys- 3.1.1 For definitions of terms used in this guide, refer to
tems
Terminology D4175.
D6595 Test Method for Determination of Wear Metals and 3.1.2 alarm, n—means of alerting the operator that a par-
ContaminantsinUsedLubricatingOilsorUsedHydraulic
ticular condition exists.
Fluids by Rotating Disc ElectrodeAtomic Emission Spec-
3.1.3 assignable cause, n—factor that contributes to varia-
trometry
tion in a process or product output that is feasible to detect and
D6786 Test Method for Particle Count in Mineral Insulating
identify; also called special cause.
Oil Using Automatic Optical Particle Counters
3.1.4 boundary lubrication, n—condition in which the fric-
D7042 Test Method for Dynamic Viscosity and Density of
tion and wear between two surfaces in relative motion are
Liquids by Stabinger Viscometer (and the Calculation of
determined by the properties of the surfaces and the properties
Kinematic Viscosity)
of the contacting fluid, other than bulk viscosity.
D7279 Test Method for Kinematic Viscosity of Transparent
3.1.4.1 Discussion—Metal to metal contact occurs and the
and Opaque Liquids by Automated Houillon Viscometer
chemistry of the system is involved. Physically adsorbed or
D7414 Test Method for Condition Monitoring of Oxidation
chemically reacted soft films (usually very thin) support
in In-Service Petroleum and Hydrocarbon Based Lubri-
contact loads. Consequently, some wear is inevitable.
cants byTrendAnalysis Using FourierTransform Infrared
3.1.5 chance cause, n—source of inherent random variation
(FT-IR) Spectrometry
in a process which is predictable within statistical limits; also
D7416 Practice for Analysis of In-Service Lubricants Using
called common cause.
a Particular Five-Part (Dielectric Permittivity, Time-
Resolved Dielectric Permittivity with Switching Magnetic
3.1.6 characteristic, n—property of items in a sample or
Fields, Laser Particle Counter, Microscopic Debris
population which, when measured, counted or otherwise
Analysis, and Orbital Viscometer) Integrated Tester
observed, helps to distinguish between the items.
D7483 TestMethodforDeterminationofDynamicViscosity
3.1.7 data set, n—logical collection of data that supports a
user function and could include one or more data tables, files,
or sources.
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
3.1.7.1 Discussion—Herein a data set is a population of
contact ASTM Customer Service at [e-mail]. For Annual Book of ASTM
values for a measurand from within a particular measurand set
Standards volume information, refer to the standard’s Document Summary page on
the ASTM website. and covering an equipment population.
D7720 − 21
3.1.8 distribution, n—as used in statistics, a set of all the 3.2 Definitions of Terms Specific to This Standard:
various values that individual observations may have and the
3.2.1 alarm limit, n—alarm condition values that delineate
frequency of their occurrence in the sample or population.
one alarm level from another within a measurand set; also
called alarm threshold.
3.1.9 measurand, n—particular quantity subject to measure-
3.2.1.1 Discussion—When several alarm levels are
ment.
designated, then a first alarm limit separates the normal level
3.1.9.1 Discussion—In industrial maintenance a measurand
fromthealertlevel,andasecondalarmlimitseparatesthealert
is sometimes called an analysis parameter.
level from action level. In other words, measurand data values
3.1.9.2 Discussion—Each measurand has a unit of measure
greater than the first alarm limit and less-than-or-equal-to the
and has a designation related to its characteristic measurement.
second alarm limit are in the state of the second level alarm.
3.1.10 nonparametric,n—termreferringtoastatisticaltech-
3.2.1.2 Discussion—An alarm limit, “X”, may be single-
nique in which the probability distribution of the constituent in
sided such as “greater than X” or “less than –X”; or it may be
the population is unknown or is not restricted to be of a
double-sided such as “greater than X and less than –X”.Alarm
specified form.
limit values may represent the same units and scale as the
3.1.11 normal distribution, n—frequency distribution char-
corresponding measurand data set, or they may be represented
acterized by a bell shaped curve and defined by two param-
as a proportion such as a percent. Alarm limit values may be
eters: mean and standard deviation.
zero-based, or they may be relative to a non-zero reference or
3.1.12 outlying observation, n—observation that appears to other baseline value.
deviate markedly in value from other members of the sample
3.2.1.3 Discussion—Statistical process control is used to
set in which it appears, also called outlier.
evaluate alarm limits comparing a control limit value with an
alarm limit value. Statistical cumulative distribution is used to
3.1.13 parametric, n—term referring to a statistical tech-
evaluate alarm limits by identifying a cumulative percent
nique that assumes the nature of the underlying frequency
values corresponding with each alarm limit value and compar-
distribution is known.
ing those results, for example, percentages of a data set in each
3.1.14 population, n—well defined set (either finite or infi-
alarm level, with expected percentages of the data set typically
nite) of elements.
associated with each alarm level.
Statistical Process Control Technique Terms
3.2.2 alarm limit set, n—collection of all the alarm limits
(alarm condition threshold values) that are needed for an
3.1.15 statistical process control (SPC), n—set of tech-
alarm-based analysis of measurands within a measurand set.
niques for improving the quality of process output by reducing
variability through the use of one or more control charts and a
3.2.3 critical equipment, n—category for important produc-
corrective action strategy used to bring the process back into a
tion assets that are not redundant or high value or highly
state of statistical control.
sensitivity or otherwise essential, also called critical assets or
critical machines.
3.1.16 state of statistical control, n—process condition
when only common causes are operating on the process.
3.2.4 equipment population, n—well defined set of like
3.1.17 center line, n—line on a control chart depicting the equipment operating under similar conditions, selected and
grouped for condition monitoring purposes; also called ma-
average level of the statistic being monitored.
chine population, asset population, and fleet.
3.1.18 control limits, n—limits on a control chart that are
3.2.4.1 Discussion—Like equipment may refer to equip-
used as criteria for signaling the need for action or judging
ment of a particular type that may include make, model,
whether a set of data does or does not indicate a state of
lubricant in use, and lubrication system. Similar conditions
statistical control based on a prescribed degree of risk.
may include environment, duty-cycle, loading conditions.
3.1.18.1 Discussion—For example, typical three-sigma lim-
its carry a risk of 0.135 % of being out of control (on one side
3.2.5 measurand set, n—meaningful assemblage of mea-
of the center line) when the process is actually in control and
surands collectively representing characteristic measurements
the statistic has a normal distribution.
that reveal modes and causes of failure within an equipment
population.
3.1.19 warning limits, n—limits on a control chart that are
3.2.5.1 Discussion—In industry, a measurand set is some-
two standard errors below and above the center line.
times called an analysis parameter set.
3.1.20 upper control limit, n—maximum value of the con-
3.2.6 noncritical equipment, n—category for production
trol chart statistic that indicates statistical control.
assets that are not critical equipment; also called balance of
3.1.21 lower control limit, n—minimum value of the control
plant.
chart statistic that indicates statistical control.
3.2.7 optimum sample interval, n—optimum (standard)
Cumulative Distribution Technique Terms
sample interval is derived from failure profile data. It is a
3.1.22 cumulative distribution, n—representation of the to- fraction of the time between initiation of a critical failure mode
tal fraction of the population, expressed as either mass-, and equipment failure. In general, sample intervals should be
volume-, area-, or number-based, that is greater than or less short enough to provide at least two samples prior to failure.
than discrete size values. Theintervalisestablishedfortheshortestcriticalfailuremode.
D7720 − 21
Alarm Level Terms (in order of severity) evaluate alarm limit values. If the data set is nonparametric or
if it includes special cause variation, then the user may apply
3.2.8 WHITE, adj—favorable level alarm designation show-
cumulative distribution technique to statistically evaluate and
ing undamaged or as-new condition having reasonable wear or
make practical adjustments to existing alarm limit values.
expected operational condition.
3.2.8.1 Discussion—Some other terms used for this level of
5. Significance and Use
alarm may include but are not limited to normal, satisfactory,
acceptable, level 1, level A, suitable for continued use and
5.1 Alarm limits are used extensively for condition moni-
good.
toring using data from in-service lubricant sample test results.
3.2.8.2 Discussion—WHITE level alarm condition is not
There are many bases for initially choosing values for these
usually accentuated by any special color indication on displays
alarm limits. There are many questions that should be ad-
or reports.
dressed. These include:
Are those limits right or wrong?
3.2.9 GREEN, adj—favorable alarm level designation
Are there too many false positive or false negative results?
showing acceptable condition and showing a measurable
Are they practical?
change in a measurand value compared with WHITE alarm
level.
5.2 This guide teaches statistical techniques for evaluating
3.2.9.1 Discussion—Some other terms used for this level of
whether alarm limits are meaningful and if they are reasonable
alarm may include but are not limited to fair, watch list,
for flagging problems requiring immediate or future action.
monitor, acceptable, level 2, level B and moderate.
5.3 This guide is intended to increase the consistency,
3.2.9.2 Discussion—GREEN level alarm condition is com-
usefulness, and dependability of condition based action recom-
monly accentuated by green letters or green highlight or green
mendations by providing machinery maintenance and monitor-
background in displays or reports.
ing personnel with a meaningful and practical way to evaluate
3.2.10 YELLOW, adj—intermediate level alarm designation
alarm limits to aid the interpretation of monitoring machinery
warning a fault condition is present and will likely need
and oil condition as well as lubricant system contamination
attention in the future.
data.
3.2.10.1 Discussion—Some other terms used for this level
of alarm may include but are not limited to amber, alert, level
6. Assumptions and Limitations
3, level C, low action priority, caution, warning, and abnormal.
6.1 The assumptions below define the ideal conditions and
3.2.10.2 Discussion—YELLOW level alarm condition is
limitations for alarm limits from a data set representing an
commonly accentuated by yellow letters or yellow highlight or
equipmentpopulation.Itisunderstoodthatidealconditionsare
yellow background in displays or reports.
not often met and that actual conditions may impact the
3.2.11 RED, adj—high level alarm designation showing
accuracy or sensitivity of the alarm limits. Assumption and
significant deterioration, review other condition information
conditions include:
and consider a possible intervention.
6.1.1 Caution should be used for data sets with too few
3.2.11.1 Discussion—Some other terms used for this level
members.
of alarm may include but are not limited to extreme, danger,
6.1.1.1 For SPC techniques using a normal distribution,
level 4, level D, unsuitable, actionable, alarm and fault.
caution should be used for data sets with fewer than 30
3.2.11.2 Discussion—RED alarm condition is commonly
members. Tentative limits can be set from as little as 10
accentuated by red letters or red highlight or red background in
samples although the quality of the limits will improve with
displays or reports.
larger populations. Larger populations (for example, in the
hundreds) can provide best alarm limits. However, the data
4. Summary of Guide
needs to be representative of the equipment population.
4.1 This guide is used to statistically evaluate and adjust
6.1.1.2 For cumulative distribution techniques regardless of
alarm limits for condition monitoring based on representative
the form of distribution, caution should be used for data sets
measurand data sets from in-service oil sample testing and
with fewer than 100 members. Tentative limits can be set from
analysis. This statistical analysis should be performed periodi-
as little as 50 samples although the quality of the limits will
cally to update alarm levels using historical data available to
improve with larger populations. Larger populations (for
the user.
example, 1000 plus) can provide best alarm limits. However
4.2 Theuserdefinesanequipmentpopulation.Theuserthen
the data needs to be representative of the equipment popula-
selectsanappropriatemeasurandsetrepresentingcharacteristic
tion.
measurements that reveal likely modes and causes of degrada-
6.1.2 The machinery process is a closed loop system
tion or failure for the lubricated machinery and for the
whereby test measurements are only affected by operations,
lubricants for that equipment population.
maintenance or the onset of a failure mode.
4.3 For each alarm based measurand the user must have a 6.1.3 An equipment population or fleet is a population of
statistically representative data set covering the equipment like machines that would be expected to be maintained
population. If the data set follows a parametric statistical accordingtothesameprotocol.Themachinesintheequipment
distribution, then the user may apply statistical process control population operated in a similar environment, under a similar
(SPC) and cumulative distribution techniques to statistically duty cycle and load conditions to include use of similar fluids
D7720 − 21
and capacities. Where machinery is maintained as such, it 6.7.2 For the case of limits based upon old data or from a
remains part of the same population, regardless of age. company that no longer produces or supports the product,
changes in lubricants or maintenance practices may have an
6.1.4 An optimum sample interval has been established
effect on the OEMs limits provided. These limits may be used
accounting for the likely or expected failure modes and at least
as a starting point for limits as discussed in 7.2.2. The
two samples will be available between failure mode initiation
techniques stated within this guide would be expected to aid
and its terminal phase.
the quality and accuracy of these limits.
6.1.5 The data set should represent historical measurements
covering at least one overhaul interval or in the case of a large
7. Procedure
fleet,shouldcoveralloperationalphasesfromnewtooverhaul.
6.1.6 Each established measurand is free from interference.
7.1 In-service lubricant sample analysis is commonly used
for condition monitoring of lubricant characteristics, lubricat-
6.2 The following comments only apply to parametric data
ing system contamination, and equipment wear. Samples are
for which the data set fits a normal distribution:
periodically and consistently collected from designated sample
6.2.1 The population satisfies a normal distribution in ac-
points on equipment and are analyzed either by an off-site
cordance with Practice D6299 Anderson-Darling (A-D) statis-
laboratory, by an on-site laboratory, by on-site test kits or by
tic which is used to objectively test for normality as described
in-line sensors.
inSubsectionA1.4ofPracticeD6299,orinaccordancewithan
7.1.1 Analysestypicallyinvolvesmultipleteststhatproduce
equivalent test for normality.
several measurands (also called analysis parameters) which
6.2.2 Most WHITE and GREEN level alarm data are
have been intentionally selected to report and measure charac-
expected to fall within two standard deviations of the mean or
teristics covering the intended range of conditions to be
represent about 94 % of all samples taken.
monitored. The group of tests (for example, test profile) is
6.2.3 Abnormalsampledataareexpectedtofalloutsidetwo
intended to target selected characteristics associated with the
standard deviations of the mean and represent about 6 % of all
asset or equipment type being monitored and produce a list of
samples taken. These data are expected to exceed a YELLOW
measurands called a measurand set (also called analysis
level alarm and unacceptable performance or an indication of
parameter set). It is common to have three alarm limits
a degrading condition is expected.
between four alarm levels associated with each alarm-based
6.3 When using cumulative distribution technique for para-
measurand. Alarm limits may be upper or lower or upper and
metric data, alarm limits may be set at points that do not lower depending on the nature of each measurand. The
coincide with standard deviations.
combination of all the alarm limits for a complete measurand
set is called an alarm limit set.
6.4 Carefulconsiderationshouldbegiventothegroupingof
7.1.2 It is not necessary for every measurand to have alarm
a population. Improved accuracy to the alarm values and limits
limits. Measurand and data values that are not alarm-based
being generated can be obtained by dividing a larger group of
have other uses such as supporting, correlating, or validity
less similar equipment/machinery into smaller more similar
checking.
ones.
7.1.3 Measurandbasedalarmlimitsserveasanintermediate
6.5 Alarm limits that are deemed to be practical must be
contributioninaprocessforconditionmonitoring.Workorders
tested at a minimum using the data set from which they were
and maintenance actions are based on a review of all data from
derived to demonstrate that the functional conclusions are
ameasurandset,onhistoricaldataandonotherinformationfor
verifiably correct.
a measurement point.
6.6 Otherstatisticalmethodsbeyondthosestatedwithinthis
7.1.4 This procedure outlines two techniques to statistically
guide may also provide reliable and useful alarm limits. This
evaluate alarm limits applied to data from in-service lubricant
guide is limited to those discussed in Section 7 as they can be
analysis condition monitoring: a statistical process control
readily applied without extensive statistical training. This
technique and a cumulative distribution technique. Both of
guide does not intend to preclude the use of other statistical
these techniques depend on statistical information from mul-
models.
tiple data sets where each data set corresponds to a measurand.
And the combination of multiple data sets covers all the
6.7 Alarm limits may be or may have been developed by
alarm-based measurands within a measurand set.
OEMs based upon experience, or in house data, or both. These
recommendations may be based upon current information or
7.2 Equipment Population—There are many types of equip-
they may have been generated by a company that no longer
ment in a condition monitoring database. A particular type of
manufactures the equipment.
equipment is selected for an equipment population that in-
6.7.1 For the case of limits based upon current data, these cludes a large number of similar equipment items having the
limits can have great value for product support and mainte- same lubricant and operating under similar conditions.Alist of
nance. This guide should be considered when variations in all measurands from a lubricant sample test profile selected for
usage and maintenance may occur. The user who wishes to an equipment population results in a measurand set represent-
depart from OEM suggested alarm limits should consider ingcharacteristicmeasurementsselectedtoreveallikelymodes
contact and discussions with the OEM when deviations from and causes of degradation or failure for the lubricated machin-
their defined limits are made. ery and for the lubricants.
D7720 − 21
TABLE 1 Generic Example
How to create a measurand set for an equipment population?
First, choose test of modes and causes. Here are examples: Then, a measurand set will list measurands specified by your
preferred methods, guides, and practices:
Particle counting D6786, D7647, D7416, D7596
Ferrous density D7416, other
Water-in-oil D4928, D6304, D6439, D7416, E2412
Lubricant chemistry D664, D974, D2896, D7414, D7416, D7484
Elemental Fe, Pb, Si, Ba, and Na D5185, D6595
Lubricant viscosity D445, D7042, D7279, D7416, D7483
Wear debris analysis D7670, D7416, D7684, D7690, D7685
7.2.1 For each measurand for which the user wishes to assets. Statistical analysis of data from an equipment popula-
evaluate alarm limits, the user produces a data set covering the tion including both critical and noncritical equipment is likely
equipment population. If the data set follows a parametric to be applied more conservatively when statistical results are
statistical distribution, then the user may apply statistical used for evaluating or adjusting alarm limits for critical
process control and cumulative distribution techniques to equipment as compared with noncritical equipment.
statistically evaluate alarm limit values. If the data set is
7.2.5 Thisguideworksbestwithasegregated,well-behaved
nonparametric or if it includes special cause variation, then the
population of identical equipment so that common cause
user may apply cumulative distribution technique to statisti-
variation in measurand data values will be small and failure
callyevaluateandmakepracticaladjustmentstoexistingalarm
modes will be readily observable in the data.
limit values.
7.2.6 It is recognized that for a laboratory with hundreds or
7.2.2 Statistical analysis suggested in this guide is most
even thousands of equipment variations, it may be time
effective using large sets of measured data values (≥30) for
consuming to create, use, and manage dozens or hundreds of
each alarm-based measurand. Historical data is necessary for
differentalarmlimitsets.Thereforepracticalapplicationofthis
statistical analysis. Thus, this guide is typically used to
guide may require compromise trade-offs when selecting
evaluate and adjust alarm limits. If sufficient historical data is
specific equipment for inclusion in a statistical equipment
not available, alarm limits for similar or related equipment can
population. The limitation in the quality of alarm limits
be used as a starting point until limits can be generated for the
generated in this fashion should be recognized.
specific equipment. Other get-started alarm limits may be
7.2.7 Itiscommonpracticeforuserstodesignateequipment
based on sources such as original equipment manufacturers,
in categories such as pumps, motors, compressors, gearboxes,
lubricantsuppliers,industryexpertconsultants.However,these
steam turbines, gas turbines, diesel engines, etc. Further
alarm limits should be migrated toward statistically based
dividing these down by make and model is desirable, particu-
alarm limits as data becomes available.
larly for fleets. Still further dividing groupings down by speed,
7.2.3 The user of this guide will need access to a historical
duty cycle, and load will yield the best alarms.
database containing the following information:
7.2.8 If a user groups equipment from too many different
7.2.3.1 Equipment information for lubricated machinery
equipment types or operational functions, then it becomes
and other equipment,
difficult to assure that resulting alarm limits are relevant and
7.2.3.2 Lubricant identity for the lubricant used in each
statistically accurate. If there are too few pieces of equipment
lubricant compartment,
in the population then variation of measured data within each
7.2.3.3 An equipment population listing a set of like equip-
population becomes too broad causing some problems not to
ment based on similarity of equipment information and lubri-
get alarmed,
...

This document is not an ASTM standard and is intended only to provide the user of an ASTM standard an indication of what changes have been made to the previous version. Because
it may not be technically possible to adequately depict all changes accurately, ASTM recommends that users consult prior editions as appropriate. In all cases only the current version
of the standard as published by ASTM is to be considered the official document.
Designation: D7720 − 11 (Reapproved 2017) D7720 − 21
Standard Guide for
Statistically Evaluating Measurand Alarm Limits when Using
Oil Analysis to Monitor Equipment and Oil for Fitness and
Contamination
This standard is issued under the fixed designation D7720; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
1. Scope Scope*
1.1 This guide provides specific requirements to statistically evaluate measurand alarm thresholds, which are called alarm limits,
as they are applied to data collected from in-service oil analysis. These alarm limits are typically used for condition monitoring
to produce severity indications relating to states of machinery wear, oil quality, and system contamination. Alarm limits distinguish
or separate various levels of alarm. Four levels are common and will be used in this guide, though three levels or five levels can
also be used.
1.2 A basic statistical process control technique described herein is recommended to evaluate alarm limits when measurand data
sets may be characterized as both parametric and in control. A frequency distribution for this kind of parametric data set fits a
well-behaved two-tail normal distribution having a “bell” curve appearance. Statistical control limits are calculated using this
technique. These control limits distinguish, at a chosen level of confidence, signal-to-noise ratio for an in-control data set from
variation that has significant, assignable causes. The operator can use them to objectively create, evaluate, and adjust alarm limits.
1.3 A statistical cumulative distribution technique described herein is also recommended to create, evaluate, and adjust alarm
limits. This particular technique employs a percent cumulative distribution of sorted data set values. The technique is based on an
actual data set distribution and therefore is not dependent on a presumed statistical profile. The technique may be used when the
data set is either parametric or nonparametric, and it may be used if a frequency distribution appears skewed or has only a single
tail. Also, this technique may be used when the data set includes special cause variation in addition to common cause variation,
although the technique should be repeated when a special cause changes significantly or is eliminated. Outputs of this technique
are specific measurand values corresponding to selected percentage levels in a cumulative distribution plot of the sorted data set.
These percent-based measurand values are used to create, evaluate and adjust alarm limits.
1.4 This guide may be applied to sample data from testing of in-service lubricating oil samples collected from machinery (for
example, diesel, pumps, gas turbines, industrial turbines, hydraulics) whether from large fleets or individual industrial applications.
1.5 This guide may also be applied to sample data from testing in-service oil samples collected from other equipment applications
where monitoring for wear, oil condition, or system contamination are important. For example, it may be applied to data sets from
oil filled transformer and circuit breaker applications.
1.6 Alarm limit evaluating techniques, which are not statistically based are not covered by this guide. Also, the techniques of this
This guide is under the jurisdiction of ASTM Committee D02 on Petroleum Products, Liquid Fuels, and Lubricants and is the direct responsibility of Subcommittee
D02.96.04 on Guidelines for In-Services Lubricants Analysis.
Current edition approved May 1, 2017Oct. 1, 2021. Published July 2017October 2021. Originally approved in 2011. Last previous edition approved in 20112017 as
D7720 – 11.D7720 – 11 (2017). DOI:10.1520DOI:10.1520/D7720-21.⁄D7720-11R17.
*A Summary of Changes section appears at the end of this standard
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
D7720 − 21
standard may be inconsistent with the following alarm limit selection techniques: “rate-of-change,” absolute alarming,
multi-parameter alarming, and empirically derived alarm limits.
1.7 The techniques in this guide deliver outputs that may be compared with other alarm limit selection techniques. The techniques
in this guide do not preclude or supersede limits that have been established and validated by an Original Equipment Manufacturer
(OEM) or another responsible party.
1.8 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility
of the user of this standard to establish appropriate safety safety, health, and healthenvironmental practices and determine the
applicability of regulatory limitations prior to use.
1.9 This international standard was developed in accordance with internationally recognized principles on standardization
established in the Decision on Principles for the Development of International Standards, Guides and Recommendations issued
by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
2. Referenced Documents
2.1 ASTM Standards:
D445 Test Method for Kinematic Viscosity of Transparent and Opaque Liquids (and Calculation of Dynamic Viscosity)
D664 Test Method for Acid Number of Petroleum Products by Potentiometric Titration
D974 Test Method for Acid and Base Number by Color-Indicator Titration
D2896 Test Method for Base Number of Petroleum Products by Potentiometric Perchloric Acid Titration
D4175 Terminology Relating to Petroleum Products, Liquid Fuels, and Lubricants
D4378 Practice for In-Service Monitoring of Mineral Turbine Oils for Steam, Gas, and Combined Cycle Turbines
D4928 Test Method for Water in Crude Oils by Coulometric Karl Fischer Titration
D5185 Test Method for Multielement Determination of Used and Unused Lubricating Oils and Base Oils by Inductively
Coupled Plasma Atomic Emission Spectrometry (ICP-AES)
D6224 Practice for In-Service Monitoring of Lubricating Oil for Auxiliary Power Plant Equipment
D6299 Practice for Applying Statistical Quality Assurance and Control Charting Techniques to Evaluate Analytical Measure-
ment System Performance
D6304 Test Method for Determination of Water in Petroleum Products, Lubricating Oils, and Additives by Coulometric Karl
Fischer Titration
D6439 Guide for Cleaning, Flushing, and Purification of Steam, Gas, and Hydroelectric Turbine Lubrication Systems
D6595 Test Method for Determination of Wear Metals and Contaminants in Used Lubricating Oils or Used Hydraulic Fluids by
Rotating Disc Electrode Atomic Emission Spectrometry
D6786 Test Method for Particle Count in Mineral Insulating Oil Using Automatic Optical Particle Counters
D7042 Test Method for Dynamic Viscosity and Density of Liquids by Stabinger Viscometer (and the Calculation of Kinematic
Viscosity)
D7279 Test Method for Kinematic Viscosity of Transparent and Opaque Liquids by Automated Houillon Viscometer
D7414 Test Method for Condition Monitoring of Oxidation in In-Service Petroleum and Hydrocarbon Based Lubricants by
Trend Analysis Using Fourier Transform Infrared (FT-IR) Spectrometry
D7416 Practice for Analysis of In-Service Lubricants Using a Particular Five-Part (Dielectric Permittivity, Time-Resolved
Dielectric Permittivity with Switching Magnetic Fields, Laser Particle Counter, Microscopic Debris Analysis, and Orbital
Viscometer) Integrated Tester
D7483 Test Method for Determination of Dynamic Viscosity and Derived Kinematic Viscosity of Liquids by Oscillating Piston
Viscometer
D7484 Test Method for Evaluation of Automotive Engine Oils for Valve-Train Wear Performance in Cummins ISB
Medium-Duty Diesel Engine
D7596 Test Method for Automatic Particle Counting and Particle Shape Classification of Oils Using a Direct Imaging Integrated
Tester
D7647 Test Method for Automatic Particle Counting of Lubricating and Hydraulic Fluids Using Dilution Techniques to
Eliminate the Contribution of Water and Interfering Soft Particles by Light Extinction
D7670 Practice for Processing In-service Fluid Samples for Particulate Contamination Analysis Using Membrane Filters
D7684 Guide for Microscopic Characterization of Particles from In-Service Lubricants
D7685 Practice for In-Line, Full Flow, Inductive Sensor for Ferromagnetic and Non-ferromagnetic Wear Debris Determination
and Diagnostics for Aero-Derivative and Aircraft Gas Turbine Engine Bearings
D7690 Practice for Microscopic Characterization of Particles from In-Service Lubricants by Analytical Ferrography
For referenced ASTM standards, visit the ASTM website, www.astm.org, or contact ASTM Customer Service at [e-mail]. For Annual Book of ASTM Standards
volume information, refer to the standard’s Document Summary page on the ASTM website.
D7720 − 21
E2412 Practice for Condition Monitoring of In-Service Lubricants by Trend Analysis Using Fourier Transform Infrared (FT-IR)
Spectrometry
3. Terminology
3.1 Definitions:
3.1.1 For definitions of terms used in this guide, refer to Terminology D4175.
3.1.2 alarm, n—means of alerting the operator that a particular condition exists.
3.1.3 assignable cause, n—factor that contributes to variation in a process or product output that is feasible to detect and identify;
also called special cause.
3.1.4 boundary lubrication, n—condition in which the friction and wear between two surfaces in relative motion are determined
by the properties of the surfaces and the properties of the contacting fluid, other than bulk viscosity.
3.1.4.1 Discussion—
Metal to metal contact occurs and the chemistry of the system is involved. Physically adsorbed or chemically reacted soft films
(usually very thin) support contact loads. Consequently, some wear is inevitable.
3.1.5 chance cause, n—source of inherent random variation in a process which is predictable within statistical limits; also called
common cause.
3.1.6 characteristic, n—property of items in a sample or population which, when measured, counted or otherwise observed, helps
to distinguish between the items.
3.1.7 data set, n—logical collection of data that supports a user function and could include one or more data tables, files, or
sources.
3.1.7.1 Discussion—
Herein a data set is a population of values for a measurand from within a particular measurand set and covering an equipment
population.
3.1.8 distribution, n—as used in statistics, a set of all the various values that individual observations may have and the frequency
of their occurrence in the sample or population.
3.1.9 measurand, n—particular quantity subject to measurement.
3.1.9.1 Discussion—
In industrial maintenance a measurand is sometimes called an analysis parameter.
3.1.9.2 Discussion—
Each measurand has a unit of measure and has a designation related to its characteristic measurement.
3.1.10 nonparametric, n—term referring to a statistical technique in which the probability distribution of the constituent in the
population is unknown or is not restricted to be of a specified form.
3.1.11 normal distribution, n—frequency distribution characterized by a bell shaped curve and defined by two parameters: mean
and standard deviation.
3.1.12 outlying observation, n—observation that appears to deviate markedly in value from other members of the sample set in
which it appears, also called outlier.
3.1.13 parametric, n—term referring to a statistical technique that assumes the nature of the underlying frequency distribution is
known.
3.1.14 population, n—well defined set (either finite or infinite) of elements.
D7720 − 21
Statistical Process Control Technique Terms
3.1.15 statistical process control (SPC), n—set of techniques for improving the quality of process output by reducing variability
through the use of one or more control charts and a corrective action strategy used to bring the process back into a state of statistical
control.
3.1.16 state of statistical control, n—process condition when only common causes are operating on the process.
3.1.17 center line, n—line on a control chart depicting the average level of the statistic being monitored.
3.1.18 control limits, n—limits on a control chart that are used as criteria for signaling the need for action or judging whether a
set of data does or does not indicate a state of statistical control based on a prescribed degree of risk.
3.1.18.1 Discussion—
For example, typical three-sigma limits carry a risk of 0.135 % of being out of control (on one side of the center line) when the
process is actually in control and the statistic has a normal distribution.
3.1.19 warning limits, n—limits on a control chart that are two standard errors below and above the center line.
3.1.20 upper control limit, n—maximum value of the control chart statistic that indicates statistical control.
3.1.21 lower control limit, n—minimum value of the control chart statistic that indicates statistical control.
Cumulative Distribution Technique Terms
3.1.22 cumulative distribution, n—representation of the total fraction of the population, expressed as either mass-, volume-, area-,
or number-based, that is greater than or less than discrete size values.
3.2 Definitions of Terms Specific to This Standard:
3.2.1 alarm limit, n—alarm condition values that delineate one alarm level from another within a measurand set; also called alarm
threshold.
3.2.1.1 Discussion—
When several alarm levels are designated, then a first alarm limit separates the normal level from the alert level, and a second alarm
limit separates the alert level from action level. In other words, measurand data values greater than the first alarm limit and
less-than-or-equal-to the second alarm limit are in the state of the second level alarm.
3.2.1.2 Discussion—
An alarm limit, “X”, may be single-sided such as “greater than X” or “less than –X”; or it may be double-sided such as “greater
than X and less than –X”. Alarm limit values may represent the same units and scale as the corresponding measurand data set, or
they may be represented as a proportion such as a percent. Alarm limit values may be zero-based, or they may be relative to a
non-zero reference or other baseline value.
3.2.1.3 Discussion—
Statistical process control is used to evaluate alarm limits comparing a control limit value with an alarm limit value. Statistical
cumulative distribution is used to evaluate alarm limits by identifying a cumulative percent values corresponding with each alarm
limit value and comparing those results, for example, percentages of a data set in each alarm level, with expected percentages of
the data set typically associated with each alarm level.
3.2.2 alarm limit set, n—collection of all the alarm limits (alarm condition threshold values) that are needed for an alarm-based
analysis of measurands within a measurand set.
3.2.3 critical equipment, n—category for important production assets that are not redundant or high value or highly sensitivity or
otherwise essential, also called critical assets or critical machines.
3.2.4 equipment population, n—well defined set of like equipment operating under similar conditions, selected and grouped for
condition monitoring purposes; also called machine population,asset population, and fleet.
D7720 − 21
3.2.4.1 Discussion—
Like equipment may refer to equipment of a particular type that may include make, model, lubricant in use, and lubrication system.
Similar conditions may include environment, duty-cycle, loading conditions.
3.2.5 measurand set, n—meaningful assemblage of measurands collectively representing characteristic measurements that reveal
modes and causes of failure within an equipment population.
3.2.5.1 Discussion—
In industry, a measurand set is sometimes called an analysis parameter set.
3.2.6 noncritical equipment, n—category for production assets that are not critical equipment; also called balance of plant.
3.2.7 optimum sample interval, n—optimum (standard) sample interval is derived from failure profile data. It is a fraction of the
time between initiation of a critical failure mode and equipment failure. In general, sample intervals should be short enough to
provide at least two samples prior to failure. The interval is established for the shortest critical failure mode.
Alarm Level Terms (in order of severity)
3.2.8 WHITE, adj—favorable level alarm designation showing undamaged or as-new condition having reasonable wear or
expected operational condition.
3.2.8.1 Discussion—
Some other terms used for this level of alarm may include but are not limited to normal, satisfactory, acceptable, level 1, level A,
suitable for continued use and good.
3.2.8.2 Discussion—
WHITE level alarm condition is not usually accentuated by any special color indication on displays or reports.
3.2.9 GREEN, adj—favorable alarm level designation showing acceptable condition and showing a measurable change in a
measurand value compared with WHITE alarm level.
3.2.9.1 Discussion—
Some other terms used for this level of alarm may include but are not limited to fair, watch list, monitor, acceptable, level 2, level
B and moderate.
3.2.9.2 Discussion—
GREEN level alarm condition is commonly accentuated by green letters or green highlight or green background in displays or
reports.
3.2.10 YELLOW, adj—intermediate level alarm designation warning a fault condition is present and will likely need attention in
the future.
3.2.10.1 Discussion—
Some other terms used for this level of alarm may include but are not limited to amber, alert, level 3, level C, low action priority,
caution, warning, and abnormal.
3.2.10.2 Discussion—
YELLOW level alarm condition is commonly accentuated by yellow letters or yellow highlight or yellow background in displays
or reports.
3.2.11 RED, adj—high level alarm designation showing significant deterioration, review other condition information and consider
a possible intervention.
3.2.11.1 Discussion—
Some other terms used for this level of alarm may include but are not limited to extreme, danger, level 4, level D, unsuitable,
actionable, alarm and fault.
3.2.11.2 Discussion—
RED alarm condition is commonly accentuated by red letters or red highlight or red background in displays or reports.
4. Summary of Guide
4.1 This guide is used to statistically evaluate and adjust alarm limits for condition monitoring based on representative measurand
data sets from in-service oil sample testing and analysis. This statistical analysis should be performed periodically to update alarm
levels using historical data available to the user.
D7720 − 21
4.2 The user defines an equipment population. The user then selects an appropriate measurand set representing characteristic
measurements that reveal likely modes and causes of degradation or failure for the lubricated machinery and for the lubricants for
that equipment population.
4.3 For each alarm based measurand the user must have a statistically representative data set covering the equipment population.
If the data set follows a parametric statistical distribution, then the user may apply statistical process control (SPC) and cumulative
distribution techniques to statistically evaluate alarm limit values. If the data set is nonparametric or if it includes special cause
variation, then the user may apply cumulative distribution technique to statistically evaluate and make practical adjustments to
existing alarm limit values.
5. Significance and Use
5.1 Alarm limits are used extensively for condition monitoring using data from in-service lubricant sample test results. There are
many bases for initially choosing values for these alarm limits. There are many questions that should be addressed. These include:
Are those limits right or wrong?
Are there too many false positive or false negative results?
Are they practical?
5.2 This guide teaches statistical techniques for evaluating whether alarm limits are meaningful and if they are reasonable for
flagging problems requiring immediate or future action.
5.3 This guide is intended to increase the consistency, usefulness, and dependability of condition based action recommendations
by providing machinery maintenance and monitoring personnel with a meaningful and practical way to evaluate alarm limits to
aid the interpretation of monitoring machinery and oil condition as well as lubricant system contamination data.
6. Assumptions and Limitations
6.1 The assumptions below define the ideal conditions and limitations for alarm limits from a data set representing an equipment
population. It is understood that ideal conditions are not often met and that actual conditions may impact the accuracy or sensitivity
of the alarm limits. Assumption and conditions include:
6.1.1 Caution should be used for data sets with too few members.
6.1.1.1 For SPC techniques using a normal distribution, caution should be used for data sets with fewer than 30 members.
Tentative limits can be set from as little as 10 samples although the quality of the limits will improve with larger populations.
Larger populations (for example, in the hundreds) can provide best alarm limits. However, the data needs to be representative of
the equipment population.
6.1.1.2 For cumulative distribution techniques regardless of the form of distribution, caution should be used for data sets with
fewer than 100 members. Tentative limits can be set from as little as 50 samples although the quality of the limits will improve
with larger populations. Larger populations (for example, 1000 plus) can provide best alarm limits. However the data needs to be
representative of the equipment population.
6.1.2 The machinery process is a closed loop system whereby test measurements are only affected by operations, maintenance or
the onset of a failure mode.
6.1.3 An equipment population or fleet is a population of like machines that would be expected to be maintained according to the
same protocol. The machines in the equipment population operated in a similar environment, under a similar duty cycle and load
conditions to include use of similar fluids and capacities. Where machinery is maintained as such, it remains part of the same
population, regardless of age.
6.1.4 An optimum sample interval has been established accounting for the likely or expected failure modes and at least two
samples will be available between failure mode initiation and its terminal phase.
6.1.5 The data set should represent historical measurements covering at least one overhaul interval or in the case of a large fleet,
should cover all operational phases from new to overhaul.
D7720 − 21
6.1.6 Each established measurand is free from interference.
6.2 The following comments only apply to parametric data for which the data set fits a normal distribution:
6.2.1 The population satisfies a normal distribution in accordance with Practice D6299 Anderson-Darling (A-D) statistic which
is used to objectively test for normality as described in Subsection A1.4 of Practice D6299, or in accordance with an equivalent
test for normality.
6.2.2 Most WHITE and GREEN level alarm data are expected to fall within two standard deviations of the mean or represent about
94 % of all samples taken.
6.2.3 Abnormal sample data are expected to fall outside two standard deviations of the mean and represent about 6 % of all
samples taken. These data are expected to exceed a YELLOW level alarm and unacceptable performance or an indication of a
degrading condition is expected.
6.3 When using cumulative distribution technique for parametric data, alarm limits may be set at points that do not coincide with
standard deviations.
6.4 Careful consideration should be given to the grouping of a population. Improved accuracy to the alarm values and limits being
generated can be obtained by dividing a larger group of less similar equipment/machinery into smaller more similar ones.
6.5 Alarm limits that are deemed to be practical must be tested at a minimum using the data set from which they were derived
to demonstrate that the functional conclusions are verifiably correct.
6.6 Other statistical methods beyond those stated within this guide may also provide reliable and useful alarm limits. This guide
is limited to those discussed in Section 7 as they can be readily applied without extensive statistical training. This guide does not
intend to preclude the use of other statistical models.
6.7 Alarm limits may be or may have been developed by OEMs based upon experience, or in house data, or both. These
recommendations may be based upon current information or they may have been generated by a company that no longer
manufactures the equipment.
6.7.1 For the case of limits based upon current data, these limits can have great value for product support and maintenance. This
guide should be considered when variations in usage and maintenance may occur. The user who wishes to depart from OEM
suggested alarm limits should consider contact and discussions with the OEM when deviations from their defined limits are made.
6.7.2 For the case of limits based upon old data or from a company that no longer produces or supports the product, changes in
lubricants or maintenance practices may have an effect on the OEMs limits provided. These limits may be used as a starting point
for limits as discussed in 7.2.2. The techniques stated within this guide would be expected to aid the quality and accuracy of these
limits.
7. Procedure
7.1 In-service lubricant sample analysis is commonly used for condition monitoring of lubricant characteristics, lubricating system
contamination, and equipment wear. Samples are periodically and consistently collected from designated sample points on
equipment and are analyzed either by an off-site laboratory, by an on-site laboratory, by on-site test kits or by in-line sensors.
7.1.1 Analyses typically involves multiple tests that produce several measurands (also called analysis parameters) which have
been intentionally selected to report and measure characteristics covering the intended range of conditions to be monitored. The
group of tests (for example, test profile) is intended to target selected characteristics associated with the asset or equipment type
being monitored and produce a list of measurands called a measurand set (also called analysis parameter set). It is common to have
three alarm limits between four alarm levels associated with each alarm-based measurand. Alarm limits may be upper or lower or
upper and lower depending on the nature of each measurand. The combination of all the alarm limits for a complete measurand
set is called an alarm limit set.
D7720 − 21
7.1.2 It is not necessary for every measurand to have alarm limits. Measurand and data values that are not alarm-based have other
uses such as supporting, correlating, or validity checking.
7.1.3 Measurand based alarm limits serve as an intermediate contribution in a process for condition monitoring. Work orders and
maintenance actions are based on a review of all data from a measurand set, on historical data and on other information for a
measurement point.
7.1.4 This procedure outlines two techniques to statistically evaluate alarm limits applied to data from in-service lubricant analysis
condition monitoring: a statistical process control technique and a cumulative distribution technique. Both of these techniques
depend on statistical information from multiple data sets where each data set corresponds to a measurand. And the combination
of multiple data sets covers all the alarm-based measurands within a measurand set.
7.2 Equipment Population—There are many types of equipment in a condition monitoring database. A particular type of
equipment is selected for an equipment population that includes a large number of similar equipment items having the same
lubricant and operating under similar conditions. A list of all measurands from a lubricant sample test profile selected for an
equipment population results in a measurand set representing characteristic measurements selected to reveal likely modes and
causes of degradation or failure for the lubricated machinery and for the lubricants.
7.2.1 For each measurand for which the user wishes to evaluate alarm limits, the user produces a data set covering the equipment
population. If the data set follows a parametric statistical distribution, then the user may apply statistical process control and
cumulative distribution techniques to statistically evaluate alarm limit values. If the data set is nonparametric or if it includes
special cause variation, then the user may apply cumulative distribution technique to statistically evaluate and make practical
adjustments to existing alarm limit values.
7.2.2 Statistical analysis suggested in this guide is most effective using large sets of measured data values (≥30) for each
alarm-based measurand. Historical data is necessary for statistical analysis. Thus, this guide is typically used to evaluate and adjust
alarm limits. If sufficient historical data is not available, alarm limits for similar or related equipment can be used as a starting point
until limits can be generated for the specific equipment. Other get-started alarm limits may be based on sources such as original
equipment manufacturers, lubricant suppliers, industry expert consultants. However, these alarm limits should be migrated toward
statistically based alarm limits as data becomes available.
7.2.3 The user of this guide will need access to a historical database containing the following information:
7.2.3.1 Equipment information for lubricated machinery and other equipment,
7.2.3.2 Lubricant identity for the lubricant used in each lubricant compartment,
7.2.3.3 An equipment population listing a set of like equipment based on similarity of equipment information and lubricant
identity,
7.2.3.4 Failure modes and causes for common problems within the equipment population,
7.2.3.5 A measurand set listing in-service sample test measurands which can identify modes and causes, and
7.2.3.6 Preferably not less than 30 historical measurand data values within the data set used with each measurand for statistical
evaluation of alarm limits.
7.2.4 As mentioned earlier, the user studies detail equipment information to designate an equipment population made up of similar
equipment, using same lubricant, and operating under similar load, operation, and environmental conditions. Within an equipment
population it is possible to have the same type of equipment that is both critical equipment, such as important production assets
that are not redundant or high value or highly sensitivity or otherwise essential, as well as noncritical equipment, such as balance
of plant equipment and redundant assets. Statistical analysis of data from an equipment population including both critical and
noncritical equipment is likely to be applied more conservatively when statistical results are used for evaluating or adjusting alarm
limits for critical equipment as compared with noncritical equipment.
7.2.5 This guide works best with a segregated, well-behaved population of identical equipment so that common cause variation
in measurand data values will be small and failure modes will be readily observable in the data.
D7720 − 21
TABLE 1 Generic Example
How to create a measurand set for an equipment population?
First, choose test of modes and causes. Here are examples: Then, a measurand set will list measurands specified by your
preferred methods, guides, and practices:
Particle counting D6786, D7647, D7416, D7596
Ferrous density D7416, other
Water-in-oil D4928, D6304, D6439, D7416, E2412
Lubricant chemistry D664, D974, D2896, D7414, D7416, D7484
Elemental Fe, Pb, Si, Ba, and Na D5185, D6595
Lubricant viscosity D445, D7042, D7279, D7416, D7483
Wear debris analysis D7670, D7416, D7684, D7690, D7685
7.2.6 It is recognized that for a laboratory with hundreds or even thousands of equipment variations, it may be time consuming
to create, use, and manage dozens or hundr
...