Incertidumbre CO

3rd Workshop Water in Food
A practical Approach towards

Measurement Uncertainty of
Tests Methods for Moisture
Content
L. Spack & C.T Reh
30/03/2004
NRC/QS - Lionel Spack
Nestl Research Center
Agenda
Introduction to measurement uncertainty

Guidelines and procedures available
Applied procedure
Comparison with reproducibility
Conclusion
30/03/2004
What are we talking about?
Measurement uncertainty expresses the degree of doubt

associated with a measurement
There is always experimental variations when we make a
measurement
A result should be reported with a tolerance interval
Result: 28 4mg
30/03/2004
Requirement for accredited laboratories
ISO/IEC 17025: 5.4.6 Estimation of uncertainty of measurement
5.4.6.2 Test laboratories should have and should apply procedures for estimating uncertainty of
measurement. In certain cases, the nature of the test method may preclude rigorous, metrologically and statistically valid calculation of uncertainty of measurement. In these cases the laboratory should at least attempt to identify all the components of uncertainty and make a reasonable
estimation, and should ensure that the form of reporting of the result does not give a wrong impression of the uncertainty. Reasonable estimation should be based on knowledge of the performance of the method and on the measurement scope and should make use of, for example,
previous experience and validation data.
30/03/2004
Use of measurement uncertainty (MU)
To enable a result to be interpreted with respect to a limit

To compare results obtained on the same material from
different laboratories
To reveal those part of a method having the greatest variability
during method validation
Uncertainty
30/03/2004
Validation
Guidelines Available
General
ISO Guide for the expression of uncertainty
Eurachem Guide for quantifying uncertainty in
analytical chemistry
VAM Protocol for uncertainty evaluation
from validation data
Specific
30/03/2004
General Approach of MU
30/03/2004
Step 1
Description of the measurement
procedure
Flow chart with the

description of all steps
of the analytical procedure
Step 2
Specification of the measurand
Relationship between the

measurand and the variables
of the analytical procedure
Step 3
Identification of uncertainty
sources
The uncertainty of each variable

is identified by means of
a cause and effect diagram
Step 4 & 5
Quantification and combination
of uncertainties
The above identified uncertainties

are quantified and combined
to give the total uncertainty on
the measurand
Case study: determination of water

content by Karl-Fischer
Titration Vessel
dried
Titration Vessel
dried
Sample
Solid
Dry Solvent
Added
Solvent
Quality
Dry Solvent
Added
Sample
Homogeneized
Water
Added + Mixed
K-F Solution
Titrated
Dry Solvent
Titred
Full Spoon
Weighed
Drop of KF
Added
Sample
Added
Water
Weighed
Empty Spoon
Weighed
Sample
Titred
30/03/2004
Main relationship of the K-F method

Vol KF t 100
bias = Mass H2O percent

mech 1000
(g/100g; %)
mech
VolKF
t
1000
Mass of sample [g]

Volume of Karl Fischer solution added for the titration of the sample [ml]
Water content of the Karl Fisher Solution [mg/ml]
Conversion of mg to g
100
To express the result in %
bias
Bias from the reference value
30/03/2004
Cause and effect diagram for K-F

method
Water content of the Karl Fisher Solution
Volume of Karl Fischer solution
Volume of KF solution for the determination of t
Temperature
Bias
Tolerance
Repeatability
Repeatability
Mass of water for the determination of t
Bias
Repeatability
Repeatability
Mass of sample
30/03/2004
Balance non-linearity
Mass percentage water
Balance Non-linearity
10
Temperature
End point
End point
Repeatability
Tolerance
Uncertainty
Bias
Validation data available for K-F

method
Precision study: repeatability conditions

analysis in duplicates
Trueness study: experiments with references
Refining the cause and effect diagram:

Repeatability branch
11
30/03/2004
Refined cause and effect diagram for

K-F method
Water content of the Karl Fisher Solution
Repeatability
Repeatability
Volume of Karl Fischer solution
End point Volume
Mass of water
Volume of KF solution for the determination of t

Temperature
Mass of sample
End point VolumeKF for sample
Bias
Tolerance
End point
Tolerance
Temperature
Mass of water for the determination of t
End point
VolumeKF for sample
Balance non-linearity
Bias
Mass percentage water
12
30/03/2004
Balance Non-linearity
Uncertainty
Mass of sample
Bias
Uncertainty budget for K-F Method:

example dietetic milk powder
Nestl Research Centre
13
Description
Value
Repeatability
(sample)
Repeatability
(KF water content)
Volume KF
(KF water content)
Mass of water
(KF water content)
Volume of KF for sample
3.19
Standard
Uncertainty u(x)
g/100g 0.057
Relative standard
uncertainty u(x)/x
0.018
5.2
mg/ml
0.029
0.006
14.3
ml
0.0084
0.0006
75
mg
0.173
0.0023
0.74
ml
0.0028
0.0037
Mass of sample
0.12
0.000173
0.0014
Intermediate standard
uncertainty
Bias
3.19
g/100g 0.064
0.020
0.0
g/100g 0.011
na
Standard uncertainty
3.19
g/100g 0.065
0.021
30/03/2004
Units
Graphical presentation of measurement

uncertainty for K-F method
uncertainty
Mass of sample
Volume KF for sample
Mass of water KF content
Volume KF water content
Repeatability KF water content
Repeatability for sample
0.005
0.01
0.015
0.02
0.025
Relative standard uncertainty
14
30/03/2004
Case study: determination of water

content by Oven method
Dish and lid
Washed
Representative
Sample
Dish and lid

Heated
Sample
Homogenized
Dish and lid

Cooled
Dish and lid

Weighed
Sample
Weighed
Sample
Dried
Sample
Cooled
15
30/03/2004
Sample
Weighed
Main relationship for the Oven method
(m2 m3 )
100 bias = Weight loss percent
(m 2 m1 )
(g/100g; %)
m1
m2
Mass of the dish, the lid and the test portion before drying [g]
m3
Mass of the dish, the lid and the test portion after drying [g]
bias
16
Mass of the dish and its lid [g]
30/03/2004
Bias from the reference value
Uncertainty budget for Oven method:

example dietetic milk powder
Description
Value
Repeatability
Units
2.77
Standard
Uncertainty u(x)
g/100g 0.04
Relative standard
uncertainty u(x)/x
0.014
M2-M3*
0.068
0.000087
0.0013
M2-M1
2.45
0.000087
0.000036
uncertainty
Bias
2.77
g/100g 0.04
0.014
-0.04
g/100g 0.014
0.35
2.81
g/100g 0.042
0.015
* with M1:83.2341, M2:85.6884, M3:85.6208
17
30/03/2004
Comparison of standard measurement

uncertainty with precision of P-tests
Example for dietetic milk powder
18
30/03/2004
Method
Karl-Fischer Oven
Result Nestl Research

Center
3.19
2.81
0.065
0.015
Number of tests
Result P-test
3.12
2.84
Reproducibility
0.095
0.083
Number of laboratories
37
129
Expression of total uncertainty as

an interval
The standard measurement uncertainty which is a standard deviation is
expressed as an interval by multiplying it by a factor k. This factor has been set
by convention as 2 for a 95% confidence level and 3 for a 99% confidence level.
By default we use k=2
Thus for water content of dietetic milk powder measured by K-F method:
3.19 ( 2 0.065) = 3.19 0.13

Thus for weight loss of dietetic milk powder measured by Oven method:
2.81 (2 0.015) = 2.81 0.03
19
30/03/2004
Conclusion
The approach is systematic and easy to

understand and to verify
It is pragmatic and based on validation data
It can be integrated in validation study
It can be transferred to other laboratories
Fulfills ISO 17025
20
30/03/2004
Thank you for your attention !!!
21
30/03/2004

Incertidumbre CO

Uploaded by

Copyright:

Available Formats

You might also like

Incertidumbre CO

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Incertidumbre CO

Uploaded by

Copyright:

Available Formats

3rd Workshop Water in Food

A practical Approach towards

NRC/QS - Lionel Spack

Nestl Research Center

Introduction to measurement uncertainty

NRC/QS - Lionel Spack

Nestl Research Center

What are we talking about?

Measurement uncertainty expresses the degree of doubt

NRC/QS - Lionel Spack

Nestl Research Center

Requirement for accredited laboratories

ISO/IEC 17025: 5.4.6 Estimation of uncertainty of measurement

NRC/QS - Lionel Spack

Nestl Research Center

Use of measurement uncertainty (MU)

To enable a result to be interpreted with respect to a limit

NRC/QS - Lionel Spack

Nestl Research Center

NRC/QS - Lionel Spack

Nestl Research Center

Flow chart with the

Relationship between the

The uncertainty of each variable

The above identified uncertainties

NRC/QS - Lionel Spack

Nestl Research Center

Case study: determination of water

NRC/QS - Lionel Spack

Nestl Research Center

Main relationship of the K-F method

bias = Mass H2O percent

Mass of sample [g]

To express the result in %

Bias from the reference value

NRC/QS - Lionel Spack

Nestl Research Center

Cause and effect diagram for K-F

Volume of Karl Fischer solution

Volume of KF solution for the determination of t

Mass of water for the determination of t

NRC/QS - Lionel Spack

Mass percentage water

Nestl Research Center

Validation data available for K-F

Precision study: repeatability conditions

Refining the cause and effect diagram:

NRC/QS - Lionel Spack

Nestl Research Center

Refined cause and effect diagram for

Volume of Karl Fischer solution

End point Volume

Volume of KF solution for the determination of t

End point VolumeKF for sample

Mass of water for the determination of t

VolumeKF for sample

Mass percentage water

NRC/QS - Lionel Spack