report nrl-pt determination of surface contact area - kitchenware · 2015-09-23 · nrl-fcm...

NRL-FCM Determination of surface contact area - kitchenware

1

UNIT ‘CONSUMER SAFETY’

Kathy Van Den Houwe, Els Van Hoeck, Caroline Evrard, Fabien Bolle, Joris Van Loco

([email protected])

2013

NRL Proficiency Study Report

Determination of surface contact area -

kitchenware


2

Table of contents

1. Introduction ................................................................................................................... 33

2. Scope ............................................................................................................................. 33

3. Time frame ..................................................................................................................... 33

4. Test Materials ................................................................................................................ 33

4.1. Preparation ................................................................................................................. 33

4.2. Homogeneity assessment ......................................................................................... 44

4.3. Distribution ................................................................................................................. 44

5. Statistical evaluation of results .................................................................................... 44

5.1. Assigned values ......................................................................................................... 44

5.2. Target standard deviation ......................................................................................... 44

5.3. Evaluation criteria for laboratory performance – zu-scores .................................... 55

6. Results and conclusions .............................................................................................. 66

6.1. Participation ............................................................................................................... 66

6.2. Laboratory results and scores .................................................................................. 66

7. Conclusion .................................................................................................................... 88

8. Acknowledgements ...................................................................................................... 88

9. References ....................................................................................................................... 8

10. Annexes ....................................................................................................................... 99

10.1. Invitation letters and documents sent to the participants ..................................... 99

10.1.1. Invitation letter ........................................................................................................ 9

10.1.2. Confirmation of participation ........................................................................... 1010

10.1.3. Shipping kit information ................................................................................... 1111

10.1.4. Confirmation of the sample receipt ................................................................. 1212

10.1.5. Possible approaches to determinate the food contact surface area ............. 1313

10.1.6. Excel file for compilation of results ................................................................. 1818

10.2. Results for the homogeneity studies ...................................................................... 19

10.2.1. Results for the homogeneity studies of samp le A .............................................. 19

10.2.2. Results for the homogeneity studies of samp le B .............................................. 20

10.3. Reported results and z u-scores ............................................................................... 21

10.3.1. Reported results and z u-scores for sample A ..................................................... 21

10.3.2. Reported results and z u-scores for sample B ..................................................... 25

10.4. Tabulated z u-scores ................................................................................................. 28


3

1. Introduction

Proficiency studies are an essential element of laboratory quality assurance and allow individual laboratories to check their analytical performance while providing them objective standards to perform against.

It is one of the duties of the National Reference Laboratories to organise interlaboratory comparisons, as stated in Regulation (EC) No 882/2004 of the European Parliament and of the Council [1].

In accordance with the above requirements the National Reference Laboratory for Food Contact Materials (NRL-FCM) organised inter-laboratory comparison tests for the network of appointed national laboratories in 2013.

2. Scope

The objectives of this ILC was to assess the laboratory performance of the routine laboratories to determine the food contact surface area of 2 kitchenware objects. Each laboratory is completely free to choose the applied method to determinate the food contact surface area. If interested, laboratories can perform more methods on one sample, but this was not compulsory.

3. Time frame

The ILC was launched in November 2013. Invitation letters (10.1.1) were sent by e-mail to the routine laboratories on the 4th of November 2013. 5 Laboratories were obliged to participate and filled out a letter of confirmation of their participation (10.1.2).

Homogeneity tests were then carried out in September and October 2013.

The samples were dispatched to the participants on the 25th, 26th and 27th of November, together with shipping kit information (10.1.3), a form of confirmation of receipt of the samples (10.1.4), instructions from the EURL-FCM to determinate the food contact surface area (10.1.5) and a printed copy for the compilation of results (10.1.6).

An electronic version of these files and each laboratory’s lab code was sent individually to the participants on the 25th of November (by e-mail).

The deadline to report the results was the 10th of January 2014.

4. Test Materials 4.1. Preparation

Samples of two different kitchen utensils were purchased on the Belgian market (Table 1). All samples were labelled and then directly sent to the participants. No further sample preparation was done. Purchase and labelling were done by the NRL-FCM.

Table 1 Test Materials

Name Sample

A Silicone baking form

B Ceramic coffee cup


4

4.2. Homogeneity assessment

The samples were tested for homogeneity by the NRL-FCM in accordance with ISO 13528/2005(E) Annex B [2]. As there was no opportunity to check the homogeneity of the surface area itself, other parameters like thickness, width, length and height of the samples or specific sample parts were measured to ensure homogeneous sample dimensions for all test items. Ten randomly selected test specimens of each sample A-B were analysed. For each test specimen, 5 dimensions were measured with a calliper. The results are given in 10.2.

The standard deviations for all measured dimensions were below 0.038 mm corresponding to coefficients of variation in the range of 0.27 - 4.30%. According to these results, the sample homogeneity can be regarded as sufficient.

4.3. Distribution

The sample kits were dispatched to the participants by the NRL-FCM in November 2013. Each participant received:

a) Two samples labelled with A-B; b) The accompanying letters with instructions on sample handling, analysis and

reporting of results (10.1.3, 10.1.5, 10.1.6); c) The form to confirm the sample receipt (10.1.4).

In addition, each participant received an e-mail containing the respective laboratory code and the Excel file for reporting the results.

5. Statistical evaluation of results

5.1. Assigned values

The true values for the surface area of the samples were unknown. As there were no other reference values available, the robust mean values obtained from the reported results of the participants were used as assigned values.

The robust mean values were obtained using the Hampel estimator, as described in ISO/TS 20612 [3]. This calculation was done using the Pro Lab software [4].

The Hampel estimator is a tool of robust statistics to obtain reference values from the results of the participants of an inter-laboratory comparison test [3]. It remains viable even with more than 40% outlier laboratories [Pro Lab Manual]. It does not require replicates for the measured values and therefore could be applied in the present case. It should be noted that no tests for outliers are carried out when the Hampel estimator is used. The algorithm works in a way that values which differ from the mean value by more than 4.5 times the standard deviation do not affect the calculated results [3].

5.2. Target standard deviation

The target standard deviation (σp) determinates the limits for a satisfactory performance in an ILC test. It should be set to a value that reflects best practice for the analysis in question. The standard deviation of the reproducibility found in collaborative trials is generally considered as an appropriate indicator of the best agreement that can be obtained between laboratories. Since no comparative test data are available, the reproducibility standard deviation that was determined from the reported test results of


5

the participants was set as the target standard deviation. The reproducibility standard deviation was calculated by the help of the Q-method described in ISO/TS 20612 [3].

Again, this calculation was done using the Pro Lab software [4].

5.3. Evaluation criteria for laboratory performance – zu-scores

The individual laboratory performance was expressed in terms of zu-scores (zu) as described in ISO/TS 20612 [3].

Commonly, z-scores are used to describe the performance of laboratories. They describe the deviation of the individual laboratory result to the assigned value, standardised by the target standard deviation. A problem of z-scores is that laboratories which report values lower than the assigned value would generally obtain a better z-score than laboratories that submit values which are above the assigned value. To overcome this problem, zu-scores were developed. They represent a modified form of z-scores. [3]

The calculation of z-scores (z) and zu-scores (zu) is done as follows [3]:

(1) p

assignedlab Xxz

σ)( −

= ,

Where:

labx is the measurement result reported by a participant;

assignedX is the assigned value;

pσ is the target standard deviation for proficiency assessment.

(2)

≥

<=

0.

0.

2

1

zifzk

g

zifzk

g

zu ,

Where:

z is the z-score;

g is the quality limit, here: 2=g ;

1k and 2k are obtained by solving the following equations in an iterative procedure:

(3)

−

+−=

−

+ 211

222 2

1exp

1

2

1exp

1k

vkk

vk


6

(4) ( ) α−=Φ−Φ

−Φ−−

1)()(1

1 12

1

kkv

The uz -scores can be interpreted as follows:

2≤uz satisfactory result;

32 ≤< uz questionable result;

3>uz unsatisfactory result.

For parameters that cannot reach negative values as it is the case in the present ILC, ISO/TS 20612 generally recommends the use of zu-scores [3]. Therefore, they were chosen as criteria to describe the laboratory performance in the present study. For their calculation, again the Pro Lab software was used [4].

6. Results and conclusion 6.1. Participation

Samples were dispatched to 7 laboratories, 5 of them submitted results, corresponding to a percentage of participation of 71.4% taking also into account the participation of the IPH.

6.2. Laboratory results and scores

The results of the statistical evaluation for the surface area (SA) measurements are summarized in Table 2 and 3. The single results reported by the laboratories, the Kernel density plots and the obtained zu-scores are displayed in the annex (see 10.3 Figure 3 – 9 and 10.4 Table 8 – 9). All calculations were done with Pro Lab [4].

To measure the food contact surface area of the samples, the laboratories could apply a method of choice. Therefore all calculations were done for each method separately, but the data were also treated as a single population for statistical evaluation.


7

Table 2 Summary of the statistical evaluation for the food contact surface area determined for Sample A

Method DIN 38402 A45 "Calculation"

"Wrap in

paper"

"Wrap in

Al foil"

Single

population

Measurand SA SA SA SA

Robust Mean = Assigned value [cm] 701.2 710.3 701.6 710.4

Robust Reproducibility = Target s.d. [cm] 40.4 47.9 33.0 29.4

Rel. Reproducibility s.d. [%] 20.2 27.7 19.0 12.8

Lower limit of tolerance [cm²] ( 2−≥uz ) 622.6 617.9 637.2 652.7

Upper limit of tolerance [cm²] ( 2≤uz ) 784.4 810.1 769.2 770.6

Lower alarm limit [cm²] ( 3−≥uz ) 582.2 569.9 604.2 623.3

Upper alarm limit [cm²] ( 3≤uz ) 824.9 858.1 802.2 800.1

Number of results 4 3 3 5

Lab performance

2≤uz 4 (100%) 3 (100%) 3 (100%) 5 (100%)

32 ≤< uz 0 0 0 0

3>uz 0 0 0 0

Table 3 Summary of the statistical evaluation for the food contact surface area determined for Sample B

Method DIN 38402 A45 "Calculation"

"Wrap in

paper"

"Wrap in

Al foil"

Single

population

Measurand SA SA SA

Robust Mean = Assigned value [cm] 138.6 143.1 142.0

Robust Reproducibility = Target s.d. [cm] 5.3 1.3 4.9

Rel. Reproducibility s.d. [%] 2.7 0.8 1.6

Lower limit of tolerance [cm²] ( 2−≥uz ) 128.1 140.4 132.3

Upper limit of tolerance [cm²] ( 2≤uz ) 149.4 145.7 152.1

Lower alarm limit [cm²] ( 3−≥uz ) 122.8 139.1 127.4

Upper alarm limit [cm²] ( 3≤uz ) 154.8 147.1 157.0

Number of results 4 3 5

Lab performance

2≤uz 4 (100%) 3 (100%) 5 (100%)

32 ≤< uz 0 0 0

3>uz 0 0 0


8

The calculated zu-scores indicate a satisfying laboratory performance for all laboratories, independently on the method applied to determine the food contact surface area. Only one lab used the “wrap in aluminium foil” for the sample B. Hence, no statistical evaluation can be done for the “wrap in aluminium foil” for the sample B.

The relative reproducibility standard deviations range from 0.8 to 27.7%. Also the Kernel density plots indicated homogeneous data distributions for all samples and all measurement methods.

7. Conclusion

This ILC was an exercise to evaluate the performance for the determination of the food contact surface area of the participating laboratories.

All laboratories obtained satisfying zu scores and no difficulties were observed in the surface area determination with a method of choice.

8. Acknowledgements

The laboratories who participated in this exercise – listed below – are kindly acknowledged.

Belgisch Verpakkingsinstituut (BVI), Zellik

Federaal Laboratorium voor de Voedselveiligheid Gentbrugge, Gentbrugge

Institut Provincial Ernest Malvoz, Liège

Laboratoire Fédéral pour la Sécurité Alimentaire Liège, Wandre

Lareco S.A., Marche-en-Famenne

SGS Belgium N.V., Antwerpen

Wetenschappelijk Instituut Volksgezondheid (WIV), Elsene

9. References

[1] Regulation (EC) No 882/2004 of the European Parliament and of the Council of 29 April 2004 on official controls performed to ensure the verification of compliance with feed and food law, animal health and animal welfare rules. Last amended by Commission Implementing Regulation (EU) No 702/2013 of 22 July 2013. OJ 199, 24.7 .2013, p. 3-4

[2] ISO 13528/2005 (E). Statistical Methods for Use in Proficiency Testing by Inter-laboratory Comparisons

[3] ISO/TS 20612/2007 (E). Water quality – Inter-laboratory comparisons for proficiency testing of analytical chemistry laboratories

[4] Pro Lab Plus Version 2.14 – QuoData, Dresden (Germany) – www.quodata.de


9

10. Annexes 10.1. Invitation letters and documents sent to the participants 10.1.1. Invitation letter


10

10.1.2. Confirmation of participation


11

10.1.3. Shipping kit information


12

10.1.4. Confirmation of the sample receipt


13

10.1.5. Possible approaches to determinate the foo d contact surface area

Instructions for the EURL-FCM ILC03 2013

– part I. Determination of the Food Contact Surface Area –

I.1. Determination of the food contact surface area by calculation

Scope

This protocol explains how to determine the surface area of a kitchen utensil by calculation.

Reagents and laboratory equipment

certified ruler (accuracy 0.1 cm)

certified caliper (accuracy 0.01 mm)

Procedure

The shape of this sample part is broken down to regular geometric shapes (e.g. cylinders,

rectangular solids, truncated cones, trapezoids, ellipsoids). If necessary, an irregular shape can be

divided into different triangles or trapezoids.

The surface area of each subpart is calculated according to the corresponding mathematical

formulas (examples are listed in Table 1). For each subpart, the dimensions needed for calculation

are measured with a ruler or caliper.

The total food contact area is the sum of all subareas.


14


15

I.2. Determination of the food contact surface area by wrapping in paper and direct weighing

Scope

This protocol explains how to determine the surface area of a kitchen utensil by wrapping in paper

and weighing the paper.


paper with a constant grammage (e.g. 80 g/m2)

scissors or scalpel

balance (accuracy 0.0001 g)

Procedure

Before wrapping the sample, the paper must be checked for a sufficiently constant and homogenous

grammage. The deviation within each sheet and between different sheets of the same paper batch

shall not exceed ±1% of the average grammage.

The surface weight (grammage) of the paper is determined with paper from the same batch. It can

be done e.g. by cutting 6 pieces of paper, each with an area of 1 dm². Another approach is to

prepare a 5-step calibration (e.g. with pieces of 0.25, 0.5, 1.0, 1.5, 2.0 dm2). The intercept is set to

zero. The slope corresponds to the paper grammage.

Then the part of the sample that will foreseeably be in contact with food is wrapped in paper.

Wrapping is done as tight as possible. Excess paper is removed with a scalpel or scissors. Afterwards,

the sample is unwrapped and the wrapping material (paper) is weighed. Knowing the paper

grammage, the surface area of the sample can be calculated.

NOTE: The term wrapping shall be interpreted in a very broad sense. Important is that the paper that

is weighted at the end, represents the foreseeable contact surface area of the article.

I.3. Determination of the food contact surface area by wrapping in aluminium foil and direct

weighing

Scope

This protocol explains how to determine the surface area of a kitchen utensil by wrapping in

aluminium foil and weighing the aluminium foil.


aluminium foil with a constant surface weight (common household aluminium foil is sufficient)

scissors or scalpel



16

Procedure

Before wrapping the sample, the aluminium foil must be checked for a sufficiently constant and

homogenous grammage. The deviation within the same batch/roll shall not exceed ±1% of the

average grammage.

The surface weight of the aluminium foil is determined with foil of the same batch/roll. It can be

done e.g. by cutting 6 pieces of foil, each with an area of 1 dm². Another approach is to prepare a 5-

step calibration (e.g. with pieces of 0.25, 0.5, 1.0, 1.5, 2.0 dm2). The intercept is set to zero. The

slope corresponds to the surface weight of the aluminium foil.

Then the part of the sample that will foreseeably be in contact with food is wrapped in aluminium

foil. Wrapping is done as tight as possible. Excess aluminium foil is removed with a scalpel or

scissors. Afterwards, the sample is unwrapped and the wrapping material (aluminium foil) is

weighed. Knowing the surface weight of the aluminium foil, the surface area of the sample can be

calculated.

NOTE: The term wrapping shall be interpreted in a very broad sense. Important is that the paper that

is weighted at the end, represents the foreseeable contact surface area of the article.

I.4. Determination of the food contact surface area by drawing the sample outline on paper

Scope

This protocol explains how to determine the surface area of a kitchen utensil by drawing its outline

on paper.


paper with a constant grammage (e.g. 80 g/m2)

pencil

scissors or scalpel


Procedure

Before drawing the sample outline on paper, the paper must be checked for a sufficiently constant

and homogenous grammage. The deviation within each sheet and between different sheets of the

same paper batch shall not exceed ±1% of the average grammage.

The surface weight (grammage) of the paper is determined with paper from the same batch. It can

be done e.g. by cutting 6 pieces of paper, each with an area of 1 dm². Another approach is to

prepare a 5-step calibration (e.g. with pieces of 0.25, 0.5, 1.0, 1.5, 2.0 dm2). The intercept is set to

zero. The slope corresponds to the paper grammage.


17

Then the sample is placed on a sheet of paper and its outline is drawn on the paper. Only the part of

the sample that will foreseeably be in contact with food is outlined. To make sure that the outline is

representative for the real surface, the sample can be cut into smaller pieces and the outline of each

of the single pieces is drawn on paper. The drawings are cut and weighed. Knowing the paper

grammage, the surface area of the sample can be calculated.


18

10.1.6. Excel file for compilation of results


19

10.2. Results for the homogeneity studies 10.2.1. Results for the homogeneity studies of sam ple A

Figure 1 Dimensions measured for homogeneity testing (Sample A)

Table 4 Measured values - Homogeneity testing (Sample A)

Sample A 1 2 3 4 5 6 7 8 9 10

Thickness silicone [cm] 0.141 0.146 0.150 0.150 0.134 0.138 0.135 0.140 0.148 0.148

Height [cm] 5.834 5.885 5.891 5.847 5.834 5.789 5.884 5.894 5.868 5.838

Diameter large [cm] 1.988 1.971 1.982 1.965 1.949 1.982 1.966 1.923 1.975 1.945

Diameter medium[cm] 1.441 1.501 1.455 1.548 1.528 1.463 1.433 1.470 1.451 1.467

Diameter small [cm] 0.889 0.895 0.886 0.875 0.894 0.880 0.883 0.897 0.891 0.882

Table 5 Statistical evaluation - Homogeneity testing (Sample A)

Sample A mean [cm] s.d. [cm] CV [%]

Thickness silicone [cm] 0.143 0.006 4.30

Height [cm] 5.856 0.034 0.58

Diameter large [cm] 1.965 0.020 1.03

Diameter medium[cm] 1.476 0.038 2.57

Diameter small [cm] 0.887 0.007 0.81


20

10.2.2. Results for the homogeneity studies of sam ple B

Figure 2 Dimensions measured for homogeneity testing (Sample B)

Table 6 Measured values - Homogeneity testing (Sample B)

Sample B 1 2 3 4 5 6 7 8 9 10

Height cup [cm] 8.122 8.179 8.200 8.167 8.164 8.174 8.133 8.186 8.187 8.169

Height handle [cm] 5.036 5.047 5.012 5.054 5.094 5.008 5.004 5.044 5.062 5.081

Thickness handle [cm] 1.335 1.313 1.339 1.359 1.334 1.345 1.326 1.336 1.327 1.302

Diameter large [cm] 7.218 7.196 7.192 7.162 7.177 7.191 7.182 7.146 7.182 7.187

Diameter small [cm] 6.190 6.179 6.163 6.165 6.188 6.184 6.172 6.174 6.215 6.148

Table 7 Statistical evaluation - Homogeneity testing (Sample B)

Sample B mean [cm] s.d. [cm] CV [%]

Height cup [cm] 8.168 0.024 0.29

Height handle [cm] 5.044 0.030 0.60

Thickness handle [cm] 1.332 0.016 1.20

Diameter large [cm] 7.183 0.019 0.27

Diameter small [cm] 6.178 0.018 0.30


21

10.3. Reported results and z u-scores 10.3.1. Reported results and z u-scores for sample A

Figure 3 Summary of reported test results for the food contact surface area of sample A determined by "calculation" (a),

Kernel density estimation (b) and zu-scores (c)


22

Figure 4 Summary of reported test results for the food contact surface area of sample A determined by "wrap in paper"

(a), Kernel density estimation (b) and zu-scores (c)


23

Figure 5 Summary of reported test results for the food contact surface area of sample A determined by "wrap in Al foil"



24

Figure 6 Summary of mean test results for the food contact surface area of sample A (treated as a single population) (a),



25

10.3.2. Reported results and z u-scores for sample B

Figure 7 Summary of reported test results for the food contact surface area of sample B determined by "calculation" (a),

Kernel density estimation (b) and zu-scores (c) Figure 4 Summary of reported test results for the surface area determination of Sample A by wrapping in aluminium (a), Kernel density

estimation (b) and zu-scores (c)


26

Figure 8 Summary of reported test results for the food contact surface area of sample B determined by "wrap in paper"



27

Figure 9 Summary of mean test results for the food contact surface area of sample B (treated as a single population) (a),



28

10.4. Tabulated z u-scores

Table 8 Zu scores for sample A

Laboratory "Calculation" "Wrap in paper" "Wrap in Al foil" Single population

CS-001

0.448

0.750

CS-002 -0.412

-0.072 -0.634

CS-003 0.211 -0.011 0.526 0.092

CS-004 -0.499 -0.468 -0.479 -0.860

CS-005

CS-006 0.649

0.591

CS-007

Table 9 Zu scores for sample B

Laboratory "Calculation" "Wrap in paper" "Wrap in Al foil" Single population

CS-001

1.071

0.491

CS-002 0.833

0.681

CS-003 -1.143

-1.948

CS-004 -0.684 -0.503

0.066

CS-005

CS-006 0.925 -0.578

0.182

CS-007

report nrl-pt determination of surface contact area - kitchenware · 2015-09-23 · nrl-fcm...

Documents