Designation: D3864 − 12

Standard Guide for

On-Line Monitoring Systems for Water Analysis1
This standard is issued under the fixed designation D3864; 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 2.2 ASTM Special Technical Publication:

STP 442 Manual on Water4
1.1 This guide covers the selection, establishment,
application, and validation and verification of monitoring 3. Terminology
systems for determining water characteristics by continual 3.1 Definitions—For definitions of terms used in this guide
sampling, automatic analysis, and recording or otherwise refer to Terminology D1129.
signaling of output data. The system chosen will depend on the
purpose for which it is intended: whether it is for regulatory 3.2 Definitions of Terms Specific to This Standard:
compliance, process monitoring, or to alert the user of adverse 3.2.1 Calibrations:
3.2.1.1 laboratory calibration curve for flow-through
trends. If it is to be used for regulatory compliance, the method
systems—calibration curve calculated from withdrawn samples
published or referenced in the regulations should be used in
or additional standards that may be spiked or diluted and
conjunction with this guide and other ASTM methods.
analyzed using the appropriate laboratory analyzer.
1.2 This standard does not purport to address all of the 3.2.1.2 laboratory calibration curve for flow-through
safety concerns, if any, associated with its use. It is the systems—type of sample used to generate a laboratory calibra-
responsibility of the user of this standard to establish appro- tion curve for flow-through systems.
priate safety and health practices and determine the applica- 3.2.1.3 line sample calibration—coincidental comparison
bility of regulatory limitations prior to use. Specific hazard of a line sample and adjustment of a continuous analyzer to the
statements are given in Section 7. compared laboratory analyzer or a second continuous analyzer.
3.2.1.4 multiple standard calibration —where the calibra-
2. Referenced Documents tion curve is calculated from a series of calibration standards
covering the range of the measurements of the sample being
2.1 ASTM Standards:2 analyzed.
D1129 Terminology Relating to Water 3.2.1.5 probe calibration—where the probe is removed
D1193 Specification for Reagent Water from the sample stream and exposed to a calibration solution
D3370 Practices for Sampling Water from Closed Conduits and the analyzer is adjusted to indicate the appropriate value.
D4210 Practice for Intralaboratory Quality Control Proce- Alternately, two probes are exposed to the same solution and
dures and a Discussion on Reporting Low-Level Data the on-line analyzer is adjusted to coincide with the pre-
(Withdrawn 2002)3 calibrated laboratory instrument.
D5540 Practice for Flow Control and Temperature Control 3.2.1.6 reference sample calibration —coincidental com-
for On-Line Water Sampling and Analysis parison of a reference sample and adjustment of a continuous
E178 Practice for Dealing With Outlying Observations analyzer to the compared laboratory analyzer results.
3.2.2 cycle time—the interval between repetitive sample
introductions in a monitoring system with discrete sampling.
1
This guide is under the jurisdiction of ASTM Committee D19 on Water and is 3.2.3 drift—the change in system output, with constant input
the direct responsibility of Subcommittee D19.03 on Sampling Water and Water- over a stated time period of unadjusted, continuous operation;
Formed Deposits, Analysis of Water for Power Generation and Process Use, usually expressed as percentage of full scale over a 24-h
On-Line Water Analysis, and Surveillance of Water.
Current edition approved June 1, 2012. Published October 2012. Originally
period.
approved in 1979. Last previous edition approved in 2006 as D3864 – 06. DOI: 3.2.3.1 span drift—drift when the input is at a constant,
10.1520/D3864-12. stated upscale value.
2
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
contact ASTM Customer Service at [email protected]. For Annual Book of ASTM 3.2.3.2 zero drift—drift when the input is at zero.
Standards volume information, refer to the standard’s Document Summary page on
the ASTM website.
3 4
The last approved version of this historical standard is referenced on Available from ASTM Headquarters. Contact Customer Service, 100 Barr
www.astm.org. Harbor Drive, West Conshohocken, PA 19428-2959.

Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States

1
 D3864 − 12
3.2.4 full scale—the maximum measuring limit of the sys- 3.2.17.1 line sample—a process sample withdrawn from the
tem for a given range. sample port (3.2.16) during a period when the process stream
3.2.5 input—the value of the parameter being measured at flowing through the continuous analyzer is of uniform quality
the inlet to the analyzer. and the analyzer result displayed is essentially constant.
Laboratory tests or results from a second continuous analyzer
3.2.6 interference—an undesired output caused by a sub-
are obtained from each sample and compared with the con-
stance or substances other than the one being measured.
tinuous analyzer results obtained at the time of sampling.
3.2.6.1 Discussion—The effect of interfering substance(s)
on the measured parameter of interest should be expressed as 3.2.17.2 reference sample—can be a primary standard or a
a percentage change (6) in the measured component as the dilution of a primary standard of known reference value. The
interference varies from 0 to 100 % of the measuring scale. If reference value must be established through multiple testing
the interference is nonlinear, an algebraic expression should be using an appropriate ASTM or other standard laboratory test
developed (or curve plotted) to show the varying effect. method. Bulk quantities of the reference sample must be stored
and handled to avoid contamination or degradation. One or
3.2.7 laboratory analyzer—a device that measures the more reference samples encompassing the range of the ana-
chemical composition or a specific physical, chemical, or lyzer may be required.
biological property of a sample.
NOTE 1—It is essential that the laboratory analyzer be checked carefully
3.2.8 limit of detection—a concentration of twice the crite- before these tests are performed to ensure compliance with the require-
rion of detection when it has been decided that the risk of ments of the standard test procedure. To further ensure proper operation it
making a Type II error is equal to a Type I error as described is recommended that a previously calibrated reference sample or an
in Practice D4210. in-house control standard of known concentration be tested to validate the
operations of the laboratory analyzer.
3.2.9 linearity—the extent to which an actual analyzer
reading agrees with the reading predicted by a straight line 3.2.18 validations—a one-time comprehensive examination
drawn between upper and lower calibration points—generally of analytical results.
zero and full-scale. (The maximum deviation from linearity is 3.2.18.1 reference sample validations—a reference sample
frequently expressed as a percentage of full-scale.) is analyzed a minimum of seven times by an appropriate
3.2.10 monitoring system—the integrated equipment pack- continuous analyzer and by an appropriate laboratory analyzer.
age comprising sampling system, analyzer, and data output A comparison is made between the average continuous ana-
equipment, required to perform water quality analysis auto- lyzer results and the average laboratory results using the
matically. Student’s t test at 95 % confidence coefficient, two-tailed test
as described in 14.1. Passing the Student’s t test signifies the
3.2.10.1 analyzer—a device that continually measures the
continuous analyzer’s average analysis of the reference sample
specific physical, chemical, or biological property of a sample.
is not statistically significantly different from the laboratory
3.2.10.2 data acquisition equipment—analog or digital de- analyzer’s average analysis of the same reference sample
vices for acquiring, processing, or recording, or a combination (validation test acceptable). Failing the ``t” test signifies a
thereof, the output signals from the analyzer. statistically significant difference exists (validation test not
3.2.10.3 sampling system—equipment necessary to deliver a acceptable).
continual representative sample to the analyzer. 3.2.18.2 line sample validations—a line sample is analyzed
3.2.11 output—a signal, usually electrical, that is related to coincidentally a minimum of seven times by an appropriate
the parametric measurement and is the intended input to data continuous analyzer and an appropriate laboratory analyzer or
acquisition equipment. a second continuous analyzer. A comparison is made on the
3.2.12 range—the region defined by the minimum and differences between the coincidental results using the Student’s
maximum measurable limits. t test at 95 % confidence coefficient, two-tailed test, to evaluate
3.2.13 repeatability—a measure of the precision of one whether the average difference is statistically significantly
analyzer to repeat its results on independent introduction of the different from zero difference as described in 14.2.
same sample at different time intervals. 3.2.19 verification—a periodic or routine procedure to en-
3.2.14 reproducibility—a measure of the precision of differ- sure reliability of analytical results.
ent analyzers to repeat results on the same sample. 3.2.19.1 line sample verification—a line sample is analyzed
3.2.15 response time—the time interval from a step change as described in 3.2.18.2, and the results of the difference
in the input or output reading to 90 % of the ultimate reading. between the continuous analyzer and the laboratory analyzer or
a second continuous analyzer is plotted on a control chart. If
3.2.15.1 lag time—the time interval from a step change in
the calculated difference between the continuous analyzer and
input to the first corresponding change in output.
the laboratory analyzer or a second continuous analyzer is
3.2.15.2 total time—the time interval from a step change in within 63 Sd, the continuous analyzer is considered verified. If
the input to a constant analyzer signal output. the calculated difference is outside 63 Sd the continuous
3.2.16 sample port—that point in the sample-conditioning analyzer is considered out of control (not verified).
system where samples for laboratory analysis are taken. 3.2.19.2 reference sample verification—a reference sample
3.2.17 samples: is analyzed as described in 3.2.18.1 and the results of the

2
 D3864 − 12
differences between the continuous analyzer and the laboratory 7. Hazards
analyzer are plotted on a control chart. If the calculated 7.1 Each analyzer installation shall be given a thorough
difference between the continuous analyzer and the laboratory safety engineering study. 6
analyzer is within 63 Sd the continuous analyzer is considered
verified. 7.2 Electrically, the monitoring system as well as the
Discussion— If the calculated difference is outside 63 Sd individual components, shall meet all code requirements for
the continuous analyzer is considered out of control (not the particular area classification.
verified). 7.2.1 All analyzers using 120 V, alternating current, 60 Hz,
3-wire systems shall observe polarity and shall not use me-
3.3 Symbols:—Sd = standard deviation chanical adapters for 2-wire outlets.
7.2.2 Check the neutral side of the power supply at the
4. Summary of Guide
analyzer to see that it is at ground potential.
4.1 This guide provides a unified approach to the use of 7.2.3 Connect the analyzer’s ground connection to earth
on-line monitoring systems for water quality analysis. It ground and check for proper continuity.
presents definitions of terms, safety precautions, system design 7.2.4 The metallic framework of the analyzer shall be at
and installation considerations, calibration techniques, general ground potential.
operating procedures, and comments relating to validation and 7.2.5 Consider additional protection in the form of properly
verification procedures. sized ground fault interrupters for each individual application.
7.2.6 Analyzers containing electrically heated sections shall
5. Significance and Use have a temperature-limit device.
5.1 Many of the manual and automated laboratory methods 7.2.7 The analyzer, and any related electrical equipment (the
for measurement of physical, chemical, and biological param- system), shall have a properly sized power cutoff switch and a
eters in water and waste water are adaptable to on-line fuse or breaker on the “hot” side of the line(s) of each device.
sampling and analysis. The resulting real-time data output can 7.3 Give full consideration to safe disposal of the analyzer’s
have a variety of uses, including confirming regulatory spent samples and reagents.
compliance, controlling process operations, or detecting leaks
or spills. 7.4 Provide pressure relief valves, if applicable, to protect
both the analyzer and monitoring system.
5.2 This guide is intended to be a common reference that
can be applied to all water quality monitoring systems. 7.5 Take precautions when using cylinders containing gases
However, calibration, validation, and verification sections may or liquids under pressure. Helpful guidance may be obtained
be inappropriate for certain tests since the act of removing a from Refs (1–4). 7
sample from a flowing stream may change the sample. 7.5.1 Gas cylinders must be handled by trained personnel
only.
5.3 Technical details of the specific methodology are con- 7.5.2 Fasten gas cylinders to a rigid structure.
tained in the pertinent ASTM standard test methods, which will 7.5.3 Take special safety precautions when using or storing
reference this practice for guidance in selection of systems and combustible or toxic gases to ensure that the system is safe and
their proper implementation. free from leaks.
5.4 This guide complements descriptive information on this 7.6 Gas piping, where possible, shall be metallic, especially
subject found in the ASTM Manual on Water. 4 inside the analyzer housing.
6. Reagents
8. Measurement Objectives
6.1 Purity of Reagents—Reagent grade chemicals shall be
8.1 Carefully define the measurement objective for the
used in all tests. Unless otherwise indicated, it is intended that
monitoring system before selecting components of the system
all reagents shall conform to the specifications of the Commit-
and set specifications realistically, to meet the objective. Terms
tee on Analytical Reagents of the American Chemical Soci-
used as specifications shall be consistent with the terminology
ety. 5 Other grades may be used, provided it is first ascertained
in Section 3.
that the reagent is of sufficiently high purity to permit its use
without lessening the accuracy of the determination. 8.2 If the monitoring system is intended primarily to deter-
mine compliance with regulatory standards, the accuracy,
6.2 Purity of Water— Unless otherwise indicated, the refer-
precision, frequency of sampling, and response time may be
ence to water shall be understood to mean reagent water that
dictated by the requirements of the regulations. A high degree
meets the purity specification of Specification D1193 Type I or
of stability and on-line reliability is generally required. The
Type II water.

5 6
Reagent Chemicals, American Chemical Society Specifications, American The user, equipment, supplier, and installer should be familiar with require-
Chemical Society, Washington, DC. For suggestions on the testing of reagents not ments of the National Electrical Code, any local applicable electrical code, U.L.
listed by the American Chemical Society, see Analar Standards for Laboratory Safety Codes, and the Occupational Safety and Health Standards (Federal Register,
Chemicals, BDH Ltd., Poole, Dorset, U.K., and the United States Pharmacopeia Vol 36, No. 105, Part II, May 29, 1971).
7
and National Formulary, U.S. Pharmaceutical Convention, Inc. (USPC), Rockville, The boldface numbers in parentheses refer to the list of references at the end of
MD. this standard.

3
 D3864 − 12
analyzer response for a specific parameter must be referenced 9.7 Provide necessary sample conditioning equipment (for
to a recognized or specified laboratory method approved by the example, filters, diluters, homogenizers, stream splitters), that
regulatory agency. is consistent with the defined measurement objective.
8.3 Monitoring systems intended to detect leaks and uncon- 9.8 Provide a connection, when necessary, for introducing
trolled discharges, that is, spills, to protect treatment plants or standard samples or withdrawing check samples immediately
receiving waters, require short sampling cycles and rapid upstream of the analyzer.
response. Typically, these will activate alarms to alert operating 9.9 Keep single- or multiple-sample streams that interface a
personnel. They then may cause flow to be diverted from single analyzer flowing all the time. Keep the manifold close to
normal channels until the upset has passed or has been the analyzer to minimize cross-contamination.
corrected. Frequently, the monitoring system is used in some
9.10 Always keep sample lines as short as possible.
way to locate and identify the source of the spill.
9.11 Provide appropriate protection of sample lines from
8.4 Systems that monitor the performance of process opera- extremely hot or freezing temperatures.
tions such as waste treatment, may have varying degrees of
sophistication and complexity, depending on the specific nature
10. Considerations for Analyzer Selection
of the application.
8.4.1 Simple, inexpensive, and low-precision analyzers with 10.1 The analyzer selected must meet the measurement
indicating or recording devices and alarms are acceptable for objective of the system over the complete range of application.
monitoring trends in operating parameters and for alerting 10.1.1 Precision and accuracy of measurement and response
operating personnel to off-standard performance. time for the parameter of interest shall coincide with system
specifications at all levels of measurement.
8.4.2 Monitoring systems that provide data to be used to
10.1.2 Interference shall be insignificant relative to the
manually control process operations or to manually set auto-
measured component or shall be controllable. When used for
matic controllers are generally more complex and frequently
regulatory compliance, known interferences shall not affect the
require that outputs be transmitted long distances.
reading more than 5 % from the true value.
8.4.3 Monitoring systems intended to process data for 10.1.3 If required for compliance, the analyzer shall be
operating guidance or management presentation and to provide capable of validation by calibration with approved and certified
varying degrees of automatic process control must be compat- standard reference materials using standard ASTM (or equiva-
ible with digital computers or telemetering systems. The lent) tests.
reliability and stability of such systems, particularly the data
output equipment, shall be high. 10.2 In choosing a specific analyzer for a specific
application, on line reliability of the instrument is of prime
9. Sample System Design Considerations concern.
10.2.1 Downtime for maintenance because of component
9.1 Carefully examine the measurement objectives of the failures or other malfunction shall be minimal. Ease,
monitoring system and select a sampling system that matches promptness, minimal cost of repair or replacement are essen-
these requirements. tial.
9.2 Review all sample requirements with the equipment 10.2.2 The analyzer shall be stable. Drift and changes in
supplier. Be sure to define accurately all conditions of intended response with changes in conditions such as flow and tempera-
operation, the components in the sample and expected varia- ture shall be insignificant or means for compensation shall be
tions in the measured parameters. provided. Sample flow variations may have a significant effect
on measured analyte concentrations. Flow rate control shall be
9.3 Choose materials of construction for the parts that will established as specified in Practice D5540. Sample flow rate
be in contact with the sample, that do not react with the sample shall be maintained within limits to maintain the necessary
to cause subsequent contamination, corrosion, or other damage precision of the continuous on line monitor.
to critical parts or sorption of measurable components and 10.2.3 The analyzer shall be relatively simple and easy to
maintain sample integrity. operate and maintain at a satisfactory level of performance.
9.4 Select the sampling point(s) so as to provide a repre-
sentative and measurable sample as close as possible to the 11. Data Output Equipment Considerations
sample system and analyzer, and as outlined in Practices 11.1 Equipment for the acquisition of output data from the
D3370. analyzer shall meet the requirements of the measuring objec-
9.5 Design the sample probe to be consistent with the tives for the monitoring system.
measurement objective and to require a minimum of mainte- 11.2 Visual or audible alarms and simple output meters are
nance. acceptable and desirable in many applications.
9.6 Select the sample transfer system, including pumps and 11.3 The analyzer output can be recorded locally at the field
transfer lines, so that the integrity of the sample is maintained location. The digital or analog signal is frequently transmitted
from sampling point to analyzer, especially with respect to to a centralized location, such as a control room, often by a data
suspension of solids and biological growth. line shared with other instruments.

4
 D3864 − 12
11.4 Records or real-time data can be transferred to com- constant, the effect of flow rate variation on measured analyte concentra-
puters for storage, process control, or report generation. tion shall be evaluated. Limits for flow rate variation shall be established
to maintain the necessary precision of the continuous on line monitor.
11.5 Process equipment such as valves and pumps can be
13.2 Reference Sample Calibration :
actuated by output generated by analyzers in a number of ways:
11.5.1 Recorded and output meters can have set points as 13.2.1 With the reference sample flowing uniformly through
integral parts of their design which actuate the equipment the analyzer sampling line, allow the continuous analyzer
directly for either on-off or proportional control. readout to equilibrate.
11.5.2 Controllers can be manually adjusted in response to 13.2.2 Record time, sample number, date, and the corre-
analyzer signals read from a recorder or from output presented sponding continuous analyzer readout, and immediately ana-
in a data report, typed or displayed on a cathode ray tube. lyze the reference sample using the appropriate laboratory
11.5.3 Direct digital process control is possible in more analysis test method.
complicated and sophisticated systems, where real-time ana- 13.2.3 Determine the continuous analyzer calibration ad-
lyzer output is integrated with other process data and used to justment required so that results of laboratory analysis and the
maintain desirable process conditions. continuous analyzer readout coincide. Adjust the analyzer
controls accordingly.
12. Installation of Monitoring System 13.2.4 Repeat this procedure until no further change is
12.1 Obtain information required for installation and opera- needed, consistent with the quality of data required.
tion of the monitoring system from the supplier. 13.3 Line Sample Calibration:
12.2 Study operational data and design parameters fur- 13.3.1 With the sample flowing through the continuous
nished by the supplier before installation. analyzer sampling line uniformly and the continuous analyzer
readout as close as possible to an equilibrium value, connect a
12.3 Choose materials of construction and components of second on line analyzer either downstream or on a parallel
the monitoring system to withstand the environment in which sample line, or withdraw a sample from the inlet stream as
it is installed. described in Practices D3370.
12.4 Select a location for the analyzer that is as close as
NOTE 3—The connection should be made in such a way so as not to
possible to the sample intake and which provides adequate contaminate the flowing sample.
protection from extremes of temperature and humidity, where
this is essential for proper performance. 13.3.2 Record time, date, continuous analyzer results and
the second on line analyzer results, or immediately analyze the
12.5 Provide a convenient access to the entire monitoring withdrawn sample using the appropriate laboratory analysis
system. test method.
12.6 Provide proper outlets for the analyzer’s exit streams 13.3.3 Determine the continuous analyzer calibration ad-
so that no liquid or gas pressure buildup occurs (see 7.4). justment required so that the results of the on line continuous
12.7 After the installation has been completed, allow the analyzers agree with the second on line analyzer or the
analyzer to stabilize and calibrate before testing performance laboratory analysis.
specifications. NOTE 4—It is essential that the second on line continuous analyzer be
checked carefully before this calibration is performed to ensure compli-
13. Calibration ance with the requirements of the standard test procedure. To further
13.1 Establish a written calibration procedure and frequency ensure proper operation it is recommended that a reference sample or
in-house control standard of known quality be tested to validate the
consistent with the parameter being measured and the accuracy operation of the second on line continuous analyzer.
and reliability demanded by the measurement or control
objectives based on the following: 13.3.4 Adjust the continuous analyzer with the analyzer
13.1.1 Consult the analyzer supplier to determine the best controls accordingly.
calibration procedure to use with the specific analyzer in a 13.4 Multiple Standard Calibration :
particular application. 13.4.1 Prepare a series of calibration standards covering the
13.1.2 When required for regulatory compliance, use cali- range of measurements for the sample being analyzed, follow-
bration procedures specified by the appropriate agency. ing instructions in the test method or in the analyzer supplier’s
13.1.3 Refer to ASTM standards, where applicable, to instructions.
determine appropriate calibration standards. 13.4.2 Check all operating conditions of the system in
13.1.4 Provide calibration standards at concentrations and accordance with the analyzer specifications, and allow suffi-
compositions as close as possible to those of the sample stream cient time for instrument equilibrium.
being analyzed. 13.4.3 Introduce a calibration standard of a concentration
13.1.5 Before calibration, ensure that the sampling system level recommended by the instrument supplier into the ana-
and output instrumentation are functioning properly and that all lyzer using the recommended instrument operating procedure.
preliminary adjustments to the analyzer required by the proce- Activate the readout equipment.
dure have been made. 13.4.4 After sufficient sample has been allowed to flow
NOTE 2—Flow rate changes may affect continuous on line analyzer through the analyzer, adjust the readout to conform to the
measured analyte concentration. If flow rates cannot be maintained desired value.

5
 D3864 − 12
13.4.5 Repeat 13.3.3 for the remaining standards from the 13.7 After initial calibration with standard solutions or
calibration series, recording the equilibrium readout value each actual samples, as in 13.2 through 13.5, analyzer calibrations
time. can be rechecked with secondary standards.
13.4.6 Plot a calibration curve of standard value versus 13.7.1 An electrical signal may be imposed to produce an
readout response from the above data. analyzer output corresponding to a specific value produced by
13.4.7 Discard any standard when any change of composi- the parameters being analyzed.
tion is detected. 13.7.2 A solution containing material other than the com-
ponent of interest, but producing the same analyzer output as
13.5 Laboratory Calibration Sample for Flow-Through Sys-
that component, may be used in place of the standard solution.
tem:
13.7.3 An optical filter may be placed in the beam of a
13.5.1 Withdraw from the spot sampling line or otherwise photometric analyzer to produce an output equivalent to that
obtain directly from the sample stream sufficient sample for produced by the component of interest.
calibration, representative of one concentration within the
range of measurement of the analyzer (see Practices D3370).
14. Validation Procedures
13.5.2 Analyze the sample for the parameter of interest
using the appropriate laboratory analysis test method. 14.1 Reference Sample Validation Procedure:
13.5.3 If necessary, prepare additional standards to cover 14.1.1 Obtain the reference sample and determine the ref-
the range of interest by dilution with reagent water or by erence value in accordance with 3.2.18.2.
“spiking” with known amounts of an appropriate standard. 14.1.2 Store the reference sample under conditions that will
13.5.4 Serially, introduce the standards into the continuous not cause contamination or degradation of the reference sample
analyzer, using the recommended instrument operating proce- concentration. Because storage conditions and factors that
dures. Allow the continuous analyzer readout to reach affect sample stability change with time, confirm the reference
equilibrium, and record the equilibrium readout value each value at periodic intervals. The frequency of confirmation can
time. best be determined by the user of the analyzer.
13.5.5 Plot a calibration curve of concentration of parameter 14.1.3 Obtain a minimum of seven coincidental laboratory
being determined versus readout response from the readout and continuous analyzer results of the reference sample, by
data. introducing the reference sample into the continuous analyzer
or laboratory analyzer and recording the results. Preferably use
13.6 Probe Calibration: different qualified operators to make the multiple determina-
13.6.1 Provide special calibration procedure for continuous tions over a period of time, with routine testing in the interim,
analyzers for which the instrumental measuring technique until sufficient data have been obtained for analysis.
utilizes a sensor that is inserted directly into the sample, for 14.1.4 More than seven test results on the reference sample
example, pH, dissolved oxygen, conductivity. are often necessary to attain an average value with acceptable
13.6.2 Prepare two calibration solutions in accordance with confidence limits. This will vary significantly for different
the appropriate test method, selecting them to bracket the laboratory procedures and reference sample concentrations.
anticipated value of measurement. This applies for both laboratory and continuous analysis.
13.6.3 Remove the probe from the sample stream, clean if 14.1.5 Tabulate the laboratory and continuous analyzer
appropriate and perform any necessary maintenance. results and their differences. Check for outliers using the
13.6.4 Fill a test container with the first calibration solution. Grubbs test criterion in Annex A1.
The container shall have the means for monitoring temperature 14.1.6 Calculate the laboratory analyzer variance from the
and, where appropriate, provide and maintain an adequate flow individual test results, excluding any outliers found in 14.1.5,
of sample past the sensor. as follows:
13.6.5 Insert the probe in the container containing the
calibration solution and, using the procedure provided by the
suppliers, adjust controls so that the analyzer output coincides SL 2 5
F(
n

i51
XL 2 2
~ ( X L!
nL
2

G (1)
~n L 2 1!
with the accepted value of the standard. Make necessary
adjustments for temperature compensation. where:
13.6.6 Rinse the probe thoroughly, place it in a second SL2 = variance of the laboratory test results,
container containing the other calibration solution and readjust XL = individual laboratory analyzer test results on the
the controls, if necessary, so that the output agrees with the reference sample,
value of this guide. nL = number of laboratory analyzer test results, and
13.6.7 Recheck with both solutions at least once. If either X̄L = ( X L
n
= arithmetic average of the laboratory analyzer
point differs from the true value by a significant amount, as L

determined by the quality of measurement required, perform test results.

necessary maintenance, and recalibrate. 14.1.7 Determine whether the precision of the laboratory
13.6.8 Alternatively, insert a second probe, with indepen- test results on the reference sample is statistically significantly
dent readout equipment and previously calibrated, into the different from the historical precision of the laboratory test
sample alongside the probe and calibrate in situ, by adjusting method. The statistical criterion for this purpose is the F test as
its controls until the outputs of the two probes coincide. follows:

6
 D3864 − 12
SB 2 n
F5
Ss 2
(2) ( ~X
i51
c 2 X̄ c ! 2
nc 2 1
(4)

where: where:
SB 2 = larger variance, either SL 2 or Sh 2, Xc = individual continuous analyzer results on the reference
Ss 2 = smaller variance, either SL 2 or Sh 2, sample,
SL 2 = variance of the laboratory test results on the refer- nc = number of continuous analyzer test results, and
ence sample as determined in 14.1.6, X̄c = ( X c
vL = degrees of freedom for laboratory analysis of refer- 5 arithmetic average of the continuous analyzer
n c
ence sample (nL − 1), test results.
Sh 2 = historical variance for the laboratory analysis with nh
determinations, and 14.1.10 Apply the F test as follows to determine whether the
vh = degrees of freedom for historical laboratory analyzer variance of the laboratory analyzer (SL 2) and the variance of
tests (nL − 1). the continuous analyzer (Sc 2) are statistically significantly
14.1.8 Compare the calculated F value with the critical F different:
value given in Table 1 for the appropriate degrees of freedom SB 2
F5 (5)
in the numerator (vL or vh) and appropriate degrees of freedom Ss 2
in the denominator ( vh or vL).
where:
14.1.8.1 If the calculated F value exceeds the critical F
value obtained from Table 1, there is at least a 95 % probability SB 2 = larger variance, either SL 2 or Sc 2,
that the reference sample laboratory analyzer data precision is Ss 2 = smaller variance, either SL 2 or Sc 2,
SL 2 = variance of the laboratory test results on the refer-
statistically significantly different from the historical precision
ence sample as determined in 14.1.6,
for that laboratory analyzer. In this event, the reasons for the
vL = degrees of freedom for laboratory analysis of refer-
substandard test precision should be determined, appropriate
ence sample (nL − 1),
corrective actions to the procedure or laboratory analyzer, or Sc 2 = variance of the continuous analyzer test results on
both, and a minimum of seven new tests on the reference the reference sample in 14.1.9, and
sample repeated in accordance with 14.1.3 through 14.1.8 until vc = degrees of freedom for the continuous analyzer,
acceptable laboratory test precision is obtained. (nc − 1).
14.1.9 Calculate the variance of the continuous analyzer
excluding outliers rejected in 14.1.5 as follows: 14.1.11 Compare the calculated F value with the critical F
value given in Table 1 for the appropriate degrees of freedom

Sc 2 5
F(n

i51
Xc 2 2
~ ( X c!
nc
2

G (3)
in the numerator (vL or vh) and the appropriate degrees of
freedom in the denominator (vh or vL).
~nc 2 1!
or

TABLE 1 F-Distribution: Degrees of Freedom for Numerator

d/n A 1 2 3 4 5 6 7 8 9 10 12 15 20 B
1 161 200 216 225 230 234 237 239 241 242 244 246 248
2 18.5 19.0 19.2 19.2 19.3 19.3 19.4 19.4 19.4 19.4 19.4 19.4 19.4
3 10.1 9.55 9.28 9.12 9.01 8.94 8.87 8.85 8.81 8.79 8.74 8.70 8.66
4 7.71 6.94 6.59 6.39 6.26 6.16 6.09 6.04 6.00 5.96 5.91 5.86 5.80
5 6.61 5.79 5.41 5.19 5.05 4.95 4.88 4.81 4.77 4.74 4.68 4.62 4.56
6 5.99 5.14 4.76 4.53 4.39 4.28 4.21 4.15 4.10 4.06 4.00 3.94 3.87
7 5.59 4.74 4.35 4.12 3.97 3.87 3.79 3.73 3.68 3.64 3.57 3.61 3.44
8 5.32 4.46 4.07 3.84 3.69 3.58 3.50 3.44 3.39 3.35 3.28 3.22 3.15
9 5.12 4.26 3.86 3.63 3.48 3.37 3.29 3.23 3.18 3.14 3.07 3.01 2.94
10 4.96 4.10 3.70 3.48 3.33 3.22 3.14 3.07 3.02 2.98 2.91 2.85 2.77
11 4.84 3.98 3.59 3.36 3.20 3.09 3.01 2.95 2.90 2.85 2.79 2.72 2.65
12 4.75 3.89 3.49 3.26 3.11 3.00 2.91 2.85 2.80 2.75 2.69 2.62 2.54
13 4.67 3.81 3.41 3.18 3.03 2.92 2.83 2.77 2.71 2.67 2.60 2.53 2.46
14 4.60 3.74 3.34 3.11 2.96 2.85 2.76 2.70 2.65 2.60 2.53 2.46 2.39
15 4.54 3.68 3.29 3.06 2.90 2.79 2.71 2.64 2.59 2.54 2.48 2.40 2.33
16 4.49 3.63 3.24 3.01 2.85 2.74 2.66 2.59 2.54 2.49 2.42 2.35 2.28
17 4.45 3.59 3.20 2.96 2.81 2.70 2.61 2.55 2.49 2.45 2.38 2.31 2.33
18 4.41 3.55 3.16 2.93 2.77 2.66 2.58 2.51 2.46 2.41 2.34 2.27 2.19
19 4.38 3.52 3.13 2.90 2.74 2.63 2.54 2.48 2.42 2.38 2.31 2.23 2.16
20 4.35 3.49 3.10 2.87 2.71 2.60 2.51 2.45 2.39 2.35 2.28 2.20 2.12
` 3.84 3.00 2.60 2.37 2.21 2.10 2.01 1.94 1.88 1.83 1.75 1.67 1.57
A
Where: n = degrees of freedom in the numerator and d = degrees of freedom in the denominator (for example, if n = 6 and d = 15 the critical F value is 2.79).
B
Expanded tables may be found in statistical reference books, see also “Standard Probability and Statistics,” CRC Press, 1991.

7
 D3864 − 12
14.1.12 If the calculated F value is equal to or less than the 14.1.16.1 Compute the degrees of freedom for the t test as
critical F value obtained from Table 1, apply the Student’s t follows:
test to determine if there is a statistically significant difference
between the average continuous analyzer results and the FS DG
SL 2 SC 2
nL
1
nC
2

average laboratory analyzer result. If the computed F is greater degrees of freedom 5 22 (9)
than the critical F value proceed to 14.1.16. S D S D
SL 2 2
nL
1
SC 2 2
nC

t5
? X̄ 2 X̄ ?
c L
(6)
n L 11 n C 11
S p round the computed value to the nearest whole number.

Sp 5 Π2
~ v L! S L 1 ~ v c! S c
n L 1n c 2 2
2

3 S 1
1
1
nL nc D (7)
14.1.17 Compare the calculated t value from above with the
critical t value from Table 2 for the degrees of freedom
computed above.
where: 14.1.18 Refer to 14.1.14 and 14.1.15 for the proper inter-
Sp = pooled standard deviation for the difference between pretation of the comparison in 14.1.17.
X̄L and X̄c, 14.1.19 Calculate the t value for the differences as follows:
X̄L = arithmetic average laboratory analyzer results,
X̄c = arithmetic average continuous analyzer results, d̄ =n d
t5 (10)
vL = degrees of freedom for laboratory analysis (nL − 1), Sd
vc = degrees of freedom for the continuous analyzer re- where:
sults (nc − 1),
d̄ = average difference,
SL 2 = variance of laboratory analyzer results, and
nd = number of differences, and
Sc 2 = variance of the continuous analyzer results.
Sd = standard deviation of the differences.
14.1.13 Compare the calculated t value from Eq 6 with the
14.1.20 Compare the calculated t value from above with the
critical t value from Table 2 for the (nL + nc − 2).
critical t value from Table 2 for nd − 1 degrees of freedom.
14.1.14 If the calculated t value is equal to or less than the
14.1.21 Refer to 14.1.14 and 14.1.15 for the proper inter-
critical t value, the continuous analyzer can be expected to give
pretation of the comparison in 14.1.20.
essentially the same average results as the laboratory analyzer.
14.1.15 If the calculated t value exceeds the critical t value 14.2 Line Sample Validation Procedure:
there is at least a 95 % probability that the continuous analyzer 14.2.1 The line sample method is used primarily for the
and the laboratory analyzer are not giving the same average test validation of the continuous analyzer operation where the
results. Continuous analyzer validity is therefore suspect. process stream is in service and available. The line sample
Further investigation of the continuous analyzer function and method therefore is not applicable for predelivery validation of
operation should be made to correct the probable bias. the continuous analyzer or for calibration before start-up and is
14.1.16 Calculate the t value as follows: not a viable alternative to the reference sample method under
these conditions.
t5
?X C 2 XL ? (8) 14.2.2 For continuous analyzer applications or process
ΠSL 2 SC 2
nL
1
nC
stream conditions, or both, that negate the practical use of the
reference sample method for predelivery or initial validation,
the line sample method is used for the analyzer validation and
is applied at appropriate times in the process when the process
TABLE 2 Table of t at 5 % Probability Level stream corresponds to low, midscale, and high concentrations
Degrees of Freedom
t
within the range of continuous analyzer operation.
(N − 1) 14.2.3 Obtain a minimum of seven line samples, preferably
1 12.706 during times of stable continuous analyzer results, using
2 4.303
3 3.182 different qualified operators, over a period of time, with routine
4 2.776 testing in the interim, until sufficient samples have been
5 2.571 obtained.
6 2.447
7 2.365 14.2.4 Record the continuous analyzer results at the time
8 2.306 each sample is withdrawn.
9 2.262
10 2.228
14.2.5 Determine the laboratory analyzer results using an
11 2.201 appropriate ASTM or standard test method.
12 2.179 14.2.6 Tabulate the difference between each continuous
13 2.160
14 2.145 analyzer result and its corresponding laboratory analyzer result
15 2.131 as follows:
16 2.120
17 2.110 di 5 Xc 2 X L (11)
18 2.101
19 2.093 where:
20 2.086
di = individual difference,

8
 D3864 − 12

Xc = continuous analyzer result, and n

~ ( d i! 2
! (d 2
XL = laboratory analyzer result. i 2
i51 nd
Sd 5 (17)
14.2.7 Check the set of differences for outliers by the ~nd 2 1!
Grubbs test as described in Annex A1. where:
14.2.8 Compute the average difference and the standard d̄ = average of differences,
deviation of the individual differences excluding outliers re- di = individual differences,
jected in 14.2.6 as follows: nd = number of differences, and
Sd = standard deviation of the differences.
d̄ 5
(d i
(12)
nd 15.1.4 Apply the Student’s t test to determine if a statistical
significant bias exists between the average difference and zero
n
~ ( d i! 2
!
difference as follows:
(d
i51
i
2
2
nd
Sd 5 (13) d̄ =n d
~nd 2 1! t5 (18)
Sd
where:
d̄ = average differences, 15.1.5 Compare the calculated t value to the critical t value
di = individual differences, from Table 2 for (n − 1) degrees freedom.
nd = number of differences, and 15.1.6 If the calculated t value is equal to or less than the
Sd = standard deviation of the differences. critical t value, center the control chart around zero difference.
14.2.9 Apply the t test as follows to check for a possible 15.1.7 If the calculated t value is greater than the critical t
systematic difference (bias) between the continuous analyzer value, center the control chart around the average difference
results and the laboratory analyzer results: (d̄). Make further investigations of the continuous analyzer
function and operation to resolve the probable bias indicated.
d̄ =n d 15.1.8 Calculate the upper and lower control chart limits as
t5 (14)
Sd follows:
14.2.10 Compare the calculated t value to the critical t value UCL 5 0 or d̄13 3 S d (19)
from Table 2 for (nd − 1) degrees of freedom.
14.2.11 If the calculated t value is equal to or less than the LCL 5 0 or d̄13 3 S d (20)
critical t value, the continuous analyzer can be expected to give where:
essentially the same average results as the laboratory analyzer.
UCL = upper control limit,
14.2.12 If the t value is greater than the critical t value, there LCL = lower control limit,
is at least a 95 % probability that the continuous analyzer and 0 = center line if pass t test,
the laboratory analyzer are not giving the same average results. d̄ = center line if fail t test, and
Therefore, the continuous analyzer validity is suspect. Make 3 × Sd = 99 % confidence interval, where Sd is the standard
further investigations of the continuous analyzer function and deviation of the differences.
operation to resolve the probable bias indicated.
15.1.9 Periodically, by introducing the reference sample,
compare the calculated differences between the continuous
15. Verification Procedures
analyzer and the coincidental analysis results from the labora-
15.1 Reference Sample Verification Procedure: tory analyzer. The frequency for the periodic check will depend
15.1.1 Tabulate the differences between the continuous on the stability of the continuous analyzer. The frequency
analyzer results and the laboratory analyzer results to the should be short enough to detect instrument drift or malfunc-
reference sample in 14.1 as follows: tion but long enough so as to not be a nuisance to the operator
of the continuous analyzer.
d i 5 Xc 2 XL (15)
15.1.10 If the calculated difference is within the UCL and
where: LCL, the continuous analyzer is considered verified.
di = individual difference, 15.1.11 If the calculated difference is outside the UCL or
Xc = continuous analyzer result, and LCL, the continuous analyzer is considered out of control.
XL = laboratory analyzer result. Further investigation of the continuous analyzer function and
15.1.2 Apply the Grubbs test for outliers to the tabulated operation should be made to correct the problem.
differences in accordance with Annex A1. 15.2 Line Sample Verification Procedure:
15.1.3 Calculate the average difference and the standard 15.2.1 Calculate the upper and lower control chart limits as
deviation of the individual differences excluding outliers re- follows:
jected in 15.1.2 as follows:
UCL 5 0 or d̄13 3 S d (21)
d̄ 5
(d i
(16)
nd LCL 5 0 or d̄13 3 S d (22)

9
 D3864 − 12
where: form on the output signal. Most analyses are recorded as direct
UCL = upper control limit, readouts based on instrument calibration.
LCL = lower control limit,
0 = center line if pass t test, 17. Precision
d̄ = center line if fail t test, and
3 × Sd = 99 % confidence interval, where Sd is the standard 17.1 Preferably, each laboratory standard test method that is
deviation of the differences. applied to on line monitoring shall include its own precision
section based on cooperative test program results.
15.2.2 Periodically, compare the calculated differences be-
tween the continuous analyzer and a coincidental laboratory 17.2 If it is desirable to validate the monitoring system
analysis or a second continuous analyzer. The frequency for the results relative to the laboratory method, statistical equivalency
periodic check depends on the stability of the continuous of the data generated by the two techniques shall be demon-
analyzer. The frequency should be short enough to detect strated by applying the“ t” test for mean differences of the
instrument drift or malfunction but not long enough to be a paired observations at the 95 % (P ≤ 0.05) confidence level
nuisance to the operator of the continuous analyzer. (t0.025 − two-tailed). This shows whether any significant differ-
15.2.3 If the calculated difference is within the UCL and ence exists between the mean values of the differences and zero
LCL, the continuous analyzer is considered verified. for paired observations having different values. The procedure
15.2.4 If the calculated difference is outside the UCL or for this test is given in the annex.
LCL, the continuous analyzer is considered out of control.
Further investigation of the continuous analyzer function and
18. Keywords
operation should be made to correct the problem.
18.1 automatic analysis; continuous sampling; monitoring
16. Calculation systems; on line; validation; verification; water analysis
16.1 Each individual monitoring system and ASTM test
method chosen determines the calculations necessary to per-

ANNEX

(Mandatory Information)

A1. REJECTION OF INDIVIDUAL OUTLIERS

A1.1 Rejection of Individual Outliers—Absolutely no data T n 5 ~ x̄ 2 x l ! /s t

should be discarded unless valid statistical criteria show them where: x̄ and st are the current estimates of the mean and
clearly to be erroneous or aberrant. Control charts and variance overall standard deviation for all retained data at the concen-
analyses may be misleading in some cases. Statistical refer- tration of material associated with xn (the highest data) and xl
ences provide valid criteria for excluding data from precision (the lowest data).
evaluations. When the experimenter is aware that a gross A1.1.1.1 Because the direction of the value being tested is
deviation from prescribed experimental procedure has not known beforehand and because the experimenter is inter-
occurred, the resultant observation should be discarded. Oth- ested in detecting values that could be either high or low, a
erwise the most extreme value among the data at each two-sided test is being performed (see Grubbs (5) for detailed
concentration of material may be tested by calculating its T explanation).
value. A1.1.2 If the suspected outlier fails the test in 10.5.1,
A1.1.1 If the T value is greater than the critical value recalculate the x̄ and st from the remaining data at this
recorded in Table A1.1 at the selected significance level (see concentration of material.
Practice E178), the outlier may be rejected (the selection of the A1.1.3 If there is another suspected value among the re-
significance level is left up to the collaborative study chair- maining data for a specific concentration level, a second
man). The test criterion for the suspected outlier, xl or xn, is as iteration of 10.5.1 may be justified but is generally not
follows: recommended.
T n 5 ~ x n 2 x̄ ! /s t A1.1.4 A completed example is shown in X1.3.

10
 D3864 − 12
TABLE A1.1 Grubbs Distribution
Critical Values for T (Two-Sided Test) When Standard Deviation is Calculated
from the Same Samples (for Outliers) A
10 % 5% 2% 1%
Number of
Significance Significance Significance Significance
Observations, n
Level Level Level Level
3 1.15 1.15 1.15 1.15
4 1.46 1.48 1.49 1.50
5 1.67 1.71 1.75 1.76
6 1.82 1.89 1.94 1.97
7 1.94 2.02 2.10 2.14
8 2.03 2.13 2.22 2.27
9 2.11 2.21 2.32 2.39
10 2.18 2.29 2.41 2.48
11 2.23 2.36 2.48 2.56
12 2.29 2.41 2.55 2.64
13 2.33 2.46 2.61 2.70
14 2.37 2.51 2.66 2.75
15 2.41 2.55 2.70 2.81
16 2.44 2.58 2.75 2.85
17 2.47 2.62 2.78 2.89
18 2.50 2.65 2.82 2.93
19 2.53 2.68 2.85 2.97
20 2.56 2.71 2.88 3.00
21 2.58 2.73 2.91 3.03
22 2.60 2.76 2.94 3.06
23 2.62 2.78 2.96 3.09
24 2.64 2.80 2.99 3.11
25 2.66 2.82 3.01 3.13
30 2.75 2.91 3.10 3.24
35 2.82 2.98 3.18 3.32
40 2.87 3.04 3.24 3.38
45 2.92 3.08 3.29 3.43
50 2.96 3.13 3.34 3.48
60 3.03 3.20 3.41 3.56
70 3.09 3.26 3.47 3.62
80 3.14 3.30 3.52 3.67
90 3.18 3.35 3.56 3.72
100 3.21 3.38 3.60 3.75
A
Values of T for N # 25 are based on those given in Ref (5). For n > 25, the values
of T are approximated. All values have been adjusted for division by n − 1 instead
of n in calculating s . Tabulated values come from Practice E178 and may also be
found in Grubbs (5). Levels of significance shown in Practice E178 were doubled,
since this is a two-sided test for significance instead of a one-sided test.

APPENDIXES

(Nonmandatory Information)

X1. REFERENCE SAMPLE VALIDATION DETERMINATION

X1.1 The following is a sample calculation for determining TABLE X1.1 Reference Sample Tabulated On Line and Laboratory
the validity of continuous analyzer system results relative to a Results
reference sample analyzed on both a continuous and laboratory On line Laboratory
Difference
Response Response
analyzer or a second continuous analyzer in accordance with 26.8 27 −0.2
14.1. 20 17 3
26 31 −5
25 26 −1
X1.2 Tabulate continuous and laboratory analyzer or a 21.3 20 1.3
second continuous analyzer results for at least seven matched 21 20 1
20.9 18 2.9
paired analysis of the reference sample, as in Table X1.1. 20.8 20 0.8
20.5 19 1.5
X1.3 Check the extreme outliers using the Grubbs rejection 15 15 0
21.3 19 2.3
criterion in A1.1 for continuous, laboratory, and difference X̄ = 21.69 21.09 0.60
values. S = 3.269 4.826 2.244

X1.3.1 Laboratory:

11
 D3864 − 12

X 2 X̄ r 31 2 21.09 20.10 2 15
TN 5
nr
5 5 2.053 Tl 5 5 1.369
Sr 4.826 3.725
T ~ α/250.05,10! 5 2.29, conclude data are not outliers
X̄ r 2 X lr 21.09 2 15
Tl 5 5 5 1.262 Continuous:
Sr 4.826
T ~ α/250.05,11! 5 2.36, conclude data are not outliers 26.8 2 21.26
TN 5 5 1.788
3.099
where:
21.26 2 15
Xnr = highest laboratory data, Tl 5 5 2.020
3.099
X̄r = average of the laboratory data,
Sr = standard deviation of the laboratory data, and T ~ α/250.05,10! 5 2.29, conclude data are not outliers
X lr = lowest laboratory data.
Difference:
Continuous:
3 2 1.160
TN 5 5 1.386
X nc 2 X̄ c 26.8 2 21.69 1.327
TN 5 5 5 1.563
Sc 3.269 1.160 2 ~ 21 !
Tl 5 5 1.628
X̄ c 2 X 1.327
lc 21.69 2 15
TN 5 5 5 2.046 T ~ α/250.05,10! 5 2.29, conclude data are not outliers
Sc 3.269
T ~ α/250.05,11! 5 2.36, conclude data are not outliers X1.4 Determine whether the precision of the laboratory test
where: results on the reference sample are statistically significantly
different from the historical precision of the laboratory of the
Xnc = highest continuous data,
X̄c = average of the continuous data, analysis as follows:
Sc = standard deviation of the continuous data, and S r2
X lc = lowest continuous data. F5
σ r2
Difference: ~ 4.826! 2
F5 5 1.822
X nd 2 X̄ d 2.9 2 0.60 ~ 3.575! 2
TN 5 5 5 1.025
Sd 2.244 F ~ α50.05,9,9 ! 5 3.18, conclude variances are not
statistically significantly different
X̄ d 2 X ld 0.60 2 ~ 25 !
Tl 5 5 5 2.495 where:
Sd 2.244
T ~ α50.05,11! 5 2.36, conclude data are not outliers Sr 2 = variance of laboratory results from the reference
sample validation (Sr ) 2, and
where: 2
σr = historical variance for this laboratory method, from
Xnd = highest difference data, previous quality control data.
X̄d = average of the difference data,
If the F test is failed, evaluate, investigate, and eliminate the
Sd = standard deviation of the difference data, and
X ld = lowest difference data. reason for failure.
Reject this data pair, and recalculate Grubbs values from X1.5 Determine whether the precision of the laboratory
data shown in Table X1.2. analysis of the reference sample are statistically significantly
Laboratory: different from the precision of the continuous analysis as
27 2 20.10 follows:
TN 5 5 1.852
3.725 SL 2
F5
SS 2
TABLE X1.2 Reference Sample Tabulated On Line and Laboratory
Results without Outliers where:
On line Laboratory SL 2 = larger variance, and
Response Response
Difference
SS 2 = smaller variance.
26.8 27 −0.2
20 17 3
~ 3.725! 2
F5 5 1.445
25 26 −1 ~ 3.099! 2
21.3 20 1.3
F ~ α50.05,9,9 ! 5 3.18, conclude variances are not
21 20 1
20.9 18 2.9 statistically significantly different
20.8 20 0.8
20.5 19 1.5 X1.6 Calculate the Student’s t test for the averages as
15 15 0 follows:
21.3 19 2.3
X̄ = 21.26 20.10 1.160
S = 3.099 3.725 1.327 X̄ c 2 X̄ l
t5
Sp

12
 D3864 − 12

Sp 5 Πn l 1n c 2 2 S
v l ~ S l 2 ! 1v c ~ S c 2 !
3
1 1
1
nl nc D X1.7 Calculate the Student’s t test for the difference as
follows:
v 5 n l 1n c 2 2 d̄ =n d 1.160 =10
t5 5 5 2.762
21.26 2 20.10 Sd 1.327
t5 5 0.757
1.532 t ~ α/250.05,9 ! 5 2.262, conclude there is a statistical

Sp 5 Π9 ~ 3.099! 2 19 ~ 3.725! 2
10110 2 2
3
1
1S1
10 10
5 1.532 D significant bias in the difference values.
Instrument pair can not be used for validation.
NOTE X1.1—If the Student’s t test fails in either X1.6 or X1.7, the
t ~ α/250.05,9 ! 5 2.262, conclude no statistically
instrument pair can not be validated.
significantly difference. Instrument pairs
are not statistically significantly different.

X2. LINE SAMPLE VALIDATION DETERMINATION

X2.1 The following is a sample calculation for determining X2.3 Check for extreme outliers using the Grubbs rejection
the validity of continuous analyzer system results relative to a criterion in A1.1 for difference values.
laboratory or second continuous analyzer in accordance with Differences:
14.2.
X nd 2 X̄ d 0.70 2 0.155
Tn 5 5 5 1.822
X2.2 Tabulate continuous and laboratory analyzer or a Sd 0.299
second continuous analyzer results for at least seven matched
X̄ d 2 X 0.155 2 ~ 20.19!
paired analysis of the process stream, as in Table X2.1. Tl 5
ld
5 5 1.154
Sd 0.299
TABLE X2.1 Line Sample Tabulated On Line and Laboratory T ~ α50.05,7 ! 5 2.02, conclude data are not outliers
Results
Continuous Analyzer Second Continuous
X2.4 Calculate the Student’s t test for the difference as
Difference follows:
Response Analyzer Response
8.37 8.33 0.04
4.61 4.52 0.09 d̄ =n d 0.155 =7
5.87 6.06 −0.19 t5 5 5 1.372
Sd 0.299
5.80 5.81 −0.01
6.86 6.81 0.05 t ~ α/250.05,6 ! 5 2.447, conclude no statistical significant
6.03 5.33 0.70
difference. Instrument pair can be used
5.45 5.04 0.41
X̄ = ... ... 0.155 for validation.
S = ... ... 0.299 NOTE X2.1—If the t test fails, the instrument pair cannot be used for
validation.

X3. REFERENCE SAMPLE VERIFICATION

X3.1 Develop a control chart centered at zero difference 63 X3.3 If the difference value is within 63 Sd , the continuous
Sd analyzer is considered verified. If not, the continuous analyzer
where: Sd = standard deviation of the differences. should be considered out-of-service until the difference is
resolved.
X3.2 Periodically compare, calculate, and plot the differ-
ence between the continuous analyzer and the coincidental
laboratory analysis or a second continuous analyzer to the
reference sample.

13
 D3864 − 12

X4. LINE SAMPLE VERIFICATION

X4.1 Develop a control chart centered at zero differences laboratory or second continuous to the process stream.
63 Sd where:
Sd = standard deviation of the differences X4.3 If the difference value is within 63 Sd, the continuous
analyzer is considered verified. If not, the continuous analyzer
X4.2 Periodically compare, calculate, and plot the differ- should be considered out-of-service until the difference is
ence between the continuous analyzer and the coincident resolved.

REFERENCES

(1) “Safe Handling of Compressed Gases,” Pamphlet P-1, Compressed (4) Sax, N. I., Dangerous Properties of Industrial Materials, 3rd ed.,
Gas Association, Inc., New York, NY. 1968, Reinhold Book Corp., New York, NY.
(2) “Compressed Gases, Safe Practices,” Pamphlet No. 95, National (5) Grubbs, F. E., and Beck, G., “Extension of Sample Sizes and
Safety Council, Chicago, IL. Percentage Points for Significance Tests of Outlying Observations,”
(3) “Chemical Safety Data Sheets,” Manufacturing Chemists Technometrics, Vol 14, No. 4, November 1972, pp. 847–854.
Association, 1825 Connecticut Ave., N.W., Washington, DC 20009.

ASTM International takes no position respecting the validity of any patent rights asserted in connection with any item mentioned
in this standard. Users of this standard are expressly advised that determination of the validity of any such patent rights, and the risk
of infringement of such rights, are entirely their own responsibility.

This standard is subject to revision at any time by the responsible technical committee and must be reviewed every five years and
if not revised, either reapproved or withdrawn. Your comments are invited either for revision of this standard or for additional standards
and should be addressed to ASTM International Headquarters. Your comments will receive careful consideration at a meeting of the
responsible technical committee, which you may attend. If you feel that your comments have not received a fair hearing you should
make your views known to the ASTM Committee on Standards, at the address shown below.

This standard is copyrighted by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959,
United States. Individual reprints (single or multiple copies) of this standard may be obtained by contacting ASTM at the above
address or at 610-832-9585 (phone), 610-832-9555 (fax), or [email protected] (e-mail); or through the ASTM website
(www.astm.org). Permission rights to photocopy the standard may also be secured from the ASTM website (www.astm.org/
COPYRIGHT/).

14