Round Robin of Primary Calibration of Heat Flux meters

Petra Andersson, Ingrid Wetterlund

Fire Technology SP Arbetsrapport 2011:05

S

P

T

ec

hni

c

a

l R

es

e

ar

c

h I

ns

ti

tut

e of

S

w

ed

en

Round Robin of Primary Calibration of

Heat Flux meters

Abstract

A proficiency testing scheme, so called Round Robin, with the primary calibration methods described in ISO 14934-2 has been conducted during the period 2008-2010. Three meters were used: one 70 kW/m² Gardon meter, one 70 kW/m² Schmidt-Boelter meter and one 20 kW/m² Schmidt-Boelter meter. Five labs participated in the Round Robin. Three of the participating laboratories used the same method, i.e. method 2 of ISO 14934-2, one lab used method 3, and one lab used both method 1 of ISO 14934-2 and a national method.

Key words: Round Robin, Heat Flux meter calibration, Fire

SP Sveriges Tekniska Forskningsinstitut

SP Technical Research Institute of Sweden SP Technical Note 2011:05

ISSN 0284-5172 Borås 2012

Contents

Abstract

3

Contents

4

Acknowledgement

5

1

Introduction

6

2

Round Robin design

7

3

Results

9

3.1 Meter 1, no 15675,1 a 70 kW/m2 Gardon gauge 10

3.2 Meter 2, no 15 6921, a 70 kW/m2 Schmidt-Boelter gauge 16 3.3 Meter 3, no 158081, a 20 kW/m2 Schmidt-Boelter gauge 23

4

Discussion

31

5

Comparison with previous Round Robins

35

6

Concluding remarks

36

Acknowledgement

This work and report had not been possible to finalize without the valuable contribution from the participants in the Round Robin. Howard Yoon at the NIST physics lab, Erik Johnsson at the NIST Building & Fire lab, Jean Remy Filtz at LNE and Marc Janssens at SwRI are all thanked for their cooperation and suggestions during this work.

1

Introduction

A proficiency testing scheme, here called Round Robin, with the primary calibration methods described in ISO 14934-2 1 has been conducted during the period 2008-2010. The objective of the Round Robin was to compare the primary calibration methods listed in the ISO standard to give background data for the precision of the three methods listed in the standard.

A similar comparison was conducted in the period 2000-2004 within a working group in the FORUM for International Cooperation in Fire Research2. During that activity the ISO 14934 series of documents was still in the Committee Draft stage and had not yet been published.

Further, a comparison was previously conducted 2001-2002 within the European project “Improving heat flux meter calibration for fire testing laboratories HFCAL” SMT4-CT98-22663. The only laboratories having primary calibration methods that participated in that comparison were LNE and SP.

2

Round Robin design

The Round Robin was conducted in a star configuration with SP calibrating the meters between each of the calibrations performed with another method or at another site. Thus the meters were returned to SP between each calibration method even for those

laboratories having two methods for calibration. The star configuration ensured that the status of the heat flux meters was traceable. At the beginning and at the end of the star inter-comparison, the meters were calibrated twice at SP to determine how much the response changed between two subsequent calibrations at the beginning of the life cycle of the meters and later in their life cycle. The participating laboratories are listed in Table 1, together with the method used.

Table 1 Participating laboratories and methods Name

in report

Laboratory Method Country Contact Person

SP SP Fire technology

Spherical black-body cavity method (method 2)

Sweden Petra Andersson, Ingrid Wetterlund

Optics NIST Physics lab

Variable-temperature black-body (VTBB) method (method 3)

USA Howard Yoon

Fire NIST Building & Fire lab

Spherical black-body cavity method (method 2)

USA Erik Johnsson

LNE1 LNE Vacuum

black-body cavity (VBBC) method (method 1)

France Jean-Remy Filtz, technical staff: Jacques Hameury, Thierry Valin, Alexandre Allard

LNE2 LNE Spherical furnace* France Jean-Remy Filtz technical staff: Jacques Hameury, Thierry Valin, Alexandre Allard

SwRI Southwest Research Institute

Spherical black-body cavity method (method 2)

USA Marc Janssens

* Horizontal mounting of heat flux meter instead of as in method 2. French primary Method, which is not listed as one of the primary methods in ISO 14934-2.

Each of the labs calibration results are independent of each other, i.e. there is no traceability to another lab except for LNE2 and LNE1.

The heat flux meters listed in Table 2 were used in the Round Robin. Additional replacement meters were also purchased in case any damage should occur to any of the primary meters. None of the additional meters, however, needed to be used.

Table 2 Original set of heat flux meters

Meter no Serial no Type Measuring range, kW/m2

1 156751 Medtherm 64-7-18* 0-70

2 156921 Medtherm 64-7SB-18 0-70

3 158081 Medtherm 64-2-18** 0-20

* Gardon gauge as default type for this range ** SB gauge as default type for this range

The calibrations were conducted at ten heat flux levels which were evenly distributed over the measurement range for each of the heat flux meters. The participants were asked to perform their calibration at levels as close as possible to the levels listed in Table 3. They were also asked to set the water temperature to 25±1°C, unless the dew point was at a higher temperature.

Table 3 Calibration levels for the Round Robin

Level no Meter no 1 Meter no 2 Meter no 3

1 7 7 2 2 14 14 4 3 21 21 6 4 28 28 8 5 35 35 10 6 42 42 12 7 49 49 14 8 56 56 16 9 63 63 18 10 70 70 20

3

Results

The calibrations were conducted according to the time scheme as presented in Table 4. The initial plans were to finish the Round Robin in October 2009, but there were severe delays – as can be seen in Table 4.

Table 4 Time table for the Round Robin

Date Activity To laboratory Meters

returned to SP

2008-09-22 – 2008-11-13 Initial calibration and ageing at SP

- -

2008-11-17 Meters sent from SP NIST Optics → LNE* 2009-02-02

2009-02-03 – 2009-02-10 Calibration at SP - -

2009-02-16 Meters sent from SP LNE 2009-03-12

2009-03-16 – 2009-03-25 Calibration at SP - -

2009-03-26 Meters sent from SP NIST Optics 2009-11-23

2009-11-24 – 2009-12-07 Calibration at SP - -

2009-12-09 Meters sent from SP NIST Fire 2010-02-08

2010-02-09 – 2010-02-17 Calibration at SP - -

2010-02-18 Meters sent from SP SwRI 2010-09-13

2010-09-15 – 2010-10-06 Final calibration at SP

- -

*On 2008-12-18 NIST Optics reported a broken calibration apparatus. LNE agreed to take over. Meters were sent from NIST Optics to LNE.

Figure 1 Schematic illustration of when the meters in the RR were transferred between participating labs. Note the star arrangement with calibration by SP between all other laboratories.

Before the actual Round Robin started the meters were placed in the furnace at SP to be aged in order to avoid initial changes in the calibration results. The meters were aged at

LNE1 LNE2 NIST Optics NIST Fire SwRI SP nov -08 feb -09 mar -09 novdec -09 feb -10 sep -10

80% of the measurement range for at least 6 h. Measurements were conducted during this procedure to check that the ageing procedure had established stable meter performance. More details about the ageing can be found in Appendix 1. The first calibration set after the ageing was called SP2. The calibration SP1 was performed before ageing.

3.1

Meter 1, no 15675,1 a 70 kW/m

2

Gardon gauge

The results for Meter 1, a 70 kW/m2 Gardon gauge, are presented in Figure 2and Figure 3 for all participating laboratories.

The NIST optics lab had exchanged Meter 1 with Meter 3 and therefore only calibrated this meter up to 20 kW/m² instead of up to 70 kW/m² as was requested in the plan. As can be seen in the figures there was also some difficulties in fixing the calibration levels exactly according to Table 3. LNE indicated that they in method LNE2 only can perform calibrations above 10 kW/m² while the provided no information on why the highest calibration level was not reached.

NIST Fire and SwRI used the ISO 14934-2 standard for determining the set-point. The version of the standard they used (i.e. ISO 14934-2 2006, which was the only publicly available version at that time) was based on an old furnace in use at SP until 2005, which is different from the equipment used today. The equations that NIST Fire and SwRI used to determine the set-point will be deleted from future versions of the standard. The use of the 2006 standard to determine the set-points for the calibration resulted, however, in minor deviation between their values and the intended values.

As the NIST Fire lab and SwRI had newly bought equipment and had not had the possibility to check temperature profiles and different sizes etc. in their furnaces, as described in Appendix 2, their incident radiation results were calculated using the values of SPs furnace. The uncertainties from the method have therefore been assumed to be double the uncertainties in the SP method to account for possible differences towards the SP furnace.

Figure 2 Output signal as a function of incident radiation for meter 1. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 O ut put si gna l, m V Incident Radiation, kW/m²

156751

SP2 SP3 LNE1 SP4 LNE2 SP5 SP6 Fire SP7 SWRi SP8 SP9

Figure 3 Incident Radiation as a function of Output signal for meter 1.

In order to study the results in more detail a mean calibration curve was calculated based on all the calibration results (except for SP where only the SP5 results were included in order not to bias the mean of all results towards the SP results). SP5 was chosen as a value in the middle of the Round Robin exercise. This mean calibration curve was then deduced from the values for each of the laboratories . The results of this calculation are shown in Figure 4. As seen from the curvature this meter is slightly non-linear in its response. Figure 5 shows the difference between each lab and the mean linear regression curve expressed in percentages. As seen, the results differ between the different labs by about ± 5 % for all except NIST Optics which deviates from the others. If the lowest level is excluded which has a high deviation due to the non-linearity of the meter then the variation decreases to ± 3.2%.

Figure 4 Spread in incident radiation as a function of output signal for the different laboratories for meter 1.

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 Inc ide nt R adi at io n, k W /m ² Output signal, mV

156751

SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9

Figure 5 Deviation from mean linear regression curve in percentages.

One plot has been constructed showing the change in results in the SP calibrations for meter 1 during the course of the Round Robin, this is presented Figure 6. A small drift can perhaps be seen but the changes are all within the expanded uncertainty limits of the method.

Figure 6 Change of calibration results at SP for meter 1 during the course of the Round Robin. A slight drift can be seen but the differences are within the expanded uncertainty limits of the method (indicated as the max and min curves in the figure).

Separate plots have also been constructed for each of the labs together with the

calibration results from SP conducted before and after that labs calibration. These plots are presented in Figure 7-Figure 11. The differences between the laboratories are somewhat larger than the differences between subsequent calibrations at SP.

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 60 70 80 Re sid ual f ro m m ean , k W /m ² Incident Radiation, kW/m² Meter 1 SP2 SP3 MAX MIN SP4 SP5 SP6 SP7 SP8 SP9

Figure 7 Comparison of LNE VBBC results for meter 1 with SP results before and after.

Figure 8 Comparison of LNE SBBC results for meter 1 with SP calibrations before and after.

Figure 9 Comparison of NIST Optics lab VTBB results for meter 1 with SP calibrations before and after.

-1.0 -0.5 0.0 0.5 1.0 1.5 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 De viat io n f ro m m ean kW /m ² 156751 SP2 SP3 LNE1 SP4 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 De viat io n f ro m m ean kW /m ² 156751 SP4 LNE2 SP5

Figure 10 Comparison of NIST Fire lab (ISO 14934-2 method 2) results for meter 1 with SP calibrations before and after.

Figure 11 Comparison of SwRI (ISO 14934-2 method 2) results for meter 1 with SP calibrations before and after.

In addition, a linear regression was performed to determine a slope and intercept for the different calibration results of each calibration, these values are presented in Table 5. The standard deviation for the slope and intercept is also shown in Table 5. The standard deviation calculation includes only one of the calibrations at SP, i.e. SP5 in order not to lower the value. SP5 was chosen as an example value in the middle of the

inter-comparison.

The incident radiation for 10 mV was also calculated based on this regression and the results are presented in Table 5. The 10 mV level was chosen for comparison with the FORUM RR2 as it is the 10 mV values that are presented in the FORUM RR. The 10 mV results are however outside of the calibration range for all laboratories as the meter was only calibrated up to 70 kW/m².

As NIST Optics only calibrated the meter up to 20 kW/m² and LNE2 only started at 10 kW/m² the common calibration range is limited. Similar calculations as for the 10 mV level is therefore conducted for the 2 mV level, these values are also reported in Table 5. The 2 mV prediction is also presented in Figure 12 and the 10mV prediction in Figure 13.

-1.5 -1.0 -0.5 0.0 0.5 1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 De viat io n f ro m m ean kW /m ² 156751 SP6 Fire SP7 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 De viat io n f ro m m ean kW /m ² 156751 SP7 SWRi SP8 SP9

The Combined expanded uncertainty presented in Table 5 includes the uncertainty in the regression which is substantial in this case as the meter turned out to be slightly non-linear. The uncertainty value given in the table for the 10 mV level is taken as the uncertainty value at the highest level used in the calibration and the values is therefore under predicted as the uncertainty in kW/ m² increases with the incident radiation. Table 5 Slope, intercept and Value at 10 and 2 mV for Meter 1.

Calibration Slope Intercept Value at 10 mV (kW/m²) Value at 2 mV (kW/m²) Combined expanded uncertainty at 2 mV (kW/m²) Combined expanded uncertainty at 10 mV (kW/m²)*** SP2 7.7681 1.5104 79.2 17.0 0.98 1.3 SP3 7.7982 1.4296 79.4 17.0 0.88 1.2 LNE1 (LNE VBBC) 7.951 1.4177 80.9 17.3 0.7 2.3 SP4 7.7848 1.5128 79.4 17.1 0.9 1.2 LNE2 (LNE SBBC) 7.9287 1.8296 81.1 17.7 1.16 2.2 SP5 7.7914 1.56 79.5 17.1 0.96 1.2 NIST Optics VTBB 7.5329 0.1998 75.5 15.3 0.24 0.3 SP6 7.8073 1.515 79.6 17.1 0.95 1.2 NIST Fire 7.6676 1.9217 78.6 17.3 1.4** 3.4 SP7 7.8222 1.5544 79.8 17.2 0.95 1.2 SwRI 7.5987 1.3272 77.3 16.5 1.18** 3.3 SP8 7.854 1.5587 80.1 17.3 1 1.3 SP9 7.8552 1.5814 80.1 17.3 0.96 1.3 Standard deviation* 0.17 0.62

* Only one SP value (SP5) included in the standard deviation calculation in order not to decrease it inappropriate.

**Uncertainty based on uncertainty from method in SP furnace multiplied by two to include extra uncertainties due to unchecked temperature profile and sizes in furnaces ***Under predicted, value at highest calibration level is used

Figure 12 Predicted value at 2 mV including expanded combined uncertainty limit including uncertainty in regression for meter 1.

Figure 13 Predicted value at 10 mV including expanded combined uncertainty limit including uncertainty in regression for meter 1. Please note that these values are outside the calibration range used in the RR.

3.2

Meter 2, no 15 6921, a 70 kW/m

2

Schmidt-Boelter

gauge

The results for Meter 2, a 70 kW/m2 Schmidt-Boelter gauge, are presented in Figure 14 and Figure 15 for all participating laboratories.

14 15 16 17 18 19 20 Valu e at 2 m V, kW /m ²

156751

72 74 76 78 80 82 84 Valu e at 1 0 m V, kW /m ²

156751

Figure 14 Output signal as a function of incident radiation for meter 2.

Figure 15 Incident Radiation as a function of Output signal for meter 2.

Figure 16 shows the change in results in the SP calibrations for meter 2 during the course of the Round Robin. The results are very stable except for a small change for the final two calibrations, all differences are however within the uncertainty limits of the method (see Appendix 3).

In order to study the results in more detail a mean calibration curve was calculated based on all the calibration results except for SP where only the SP5 results were included in order not to bias the mean of all results towards the SP results. SP5 was chosen as a value in the middle of the Round Robin exercise. This mean calibration curve was then deduced from the values for each of the laboratories. The results of this calculation are shown in Figure 17. Figure 18 shows the same expressed as percentages as a function of incident radiation. The differences are large for the low levels, i.e. ±9%.

Separate diagrams have also been constructed for each of the labs together with the calibration results from SP conducted before and after that labs calibration. These diagrams are presented in Figure 19 - Figure 23. The diagrams show that the differences between the laboratories are somewhat larger than the differences between subsequent calibrations at SP. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 O ut put si gna l, m V Incident Radiation, kW/m² 156921 SP3 SP2 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Inc ide nt R adi at io n, k W /m ² Output signal, mV 156921 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9

Figure 16 Change of calibration results at SP for meter 2 during the course of the Round Robin. As seen are all differences within the combined expanded uncertainty for the calibration indicated as the max and min values in the figure.

Figure 17 Spread in incident radiation as a function of output signal for the different laboratories for meter 2.

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 60 70 80 Re sid ual f ro m m ean , k W /m ² Incident radiation, kW/m²

Meter2

SP2 SP3 MAX MIN SP4 SP5 SP6 SP7 SP8 SP9 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 156921 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9

Figure 18 Deviation from mean linear regression curve in percentages for meter 2.

Figure 19 Comparison of LNE VBBC results for meter 2 with SP results before and after.

Figure 20 Comparison of LNE spherical furnace results for meter 2 with SP calibrations before and after.

-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 156921 SP2 SP3 LNE1 SP4 -0.5 0.0 0.5 1.0 1.5 2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ²

156921

SP4 LNE2 SP5

Figure 21 Comparison of NIST Optics VTBB results for meter 2 with SP calibrations before and after.

Figure 22 Comparison of NIST Fire lab (ISO 14934-2 method 2) results for meter 2 with SP calibrations before and after.

-2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 156921 SP5 Optics SP6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 156921 SP6 Fire SP7

Figure 23 Comparison of SwRI (ISO 14934-2 method 2) results for meter 2 with SP calibrations before and after.

In addition, a linear regression was performed to determine a slope and intercept for the different calibration results of each calibration, these values are presented in Table 6. The standard deviation for the slope and intercept is also shown in Table 6. The standard deviation calculation includes only one of the calibrations at SP, i.e. SP5 in order not to lower the value. SP5 was chosen as an example value in the middle of the

inter-comparison.

The incident radiation for 10 mV was also calculated based on this regression and the results are presented in Table 6. The 10 mV level was chosen for comparison with the FORUM RR. The 10 mV value is however outside of the range NIST Optics used as they only calibrated the meter up to 48 kW/m². Similar calculations as for the 10 mV level has also been conducted for the 2 mV and 5 mV level, the 2 and 5 mV levels are within the range all labs have used for their calibrations. All the values are reported in Table 6 together with the combined expanded uncertainty for each level. The combined expanded uncertainty includes the uncertainty form the regression. In addition are the same results presented in Figure 24-Figure 26.

-3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ²

156921

SP7 SWRi SP8 SP9

Table 6 Slope, intercept and Value at 10, 5 and 2 mV for Meter 2. Calib Slope Intercep

t

Value (kW/m²) at Combined expanded uncertainty (kW/m²) at 10mV 5mV 2mV 10mV 5mV 2mV SP2 5.944 0.0143 59.5 29.7 11.9 0.7 0.4 0.2 SP3 5.9533 0.0366 59.6 29.8 11.9 0.7 0.4 0.2 LNE1 (LNE VBBC) 6.1342 0.5761 61.9 31.2 12.8 1.5 0.8 0.5 SP4 5.944 0.0566 59.5 29.8 11.9 0.7 0.4 0.32 LNE2 (LNE SBBC) 6.0053 1.2656 61.3 31.3 13.3 1.8 0.9 0.7 SP5 5.9329 0.0056 59.3 29.7 11.9 0.7 0.4 0.2 NIST Optics VTBB 5.7394 -0.0673 57.3 28.6 11.4 0.8*** 0.3 0.2 SP6 5.944 0.0462 59.5 29.8 11.9 0.7 0.4 0.2 NIST Fire 5.8743 0.3872 59.1 29.8 12.1 2.7** 1.3** 0.7** SP7 5.942 -0.0075 59.4 29.7 11.9 0.7 0.4 0.2 SwRI 5.6952 -0.0672 56.9 28.4 11.3 2.7** 1.3** 0.7** SP8 6.0018 -0.0808 59.3 29.9 11.9 0.7 0.4 0.3 SP9 5.9792 -0.0369 59.8 29.9 11.9 0.7 0.4 0.2 Standar d deviatio n* 0.16 0.52

* Only one SP value (SP5) included in the calculation in order not to decrease it inappropriate.

**Uncertainty based on uncertainty from method in SP furnace multiplied by two to include extra uncertainties due to unchecked temperature profile and sizes in furnaces ***Underpredicted, value at highest calibration level is used

Figure 24 Predicted value at 2 mV with the expanded combined uncertainty limit including uncertainty in regression for meter 2.

10 10.5 11 11.5 12 12.5 13 13.5 14 Valu e at 2 m V, kW /m ²

156921

Figure 25 Predicted value at 5 mV with the expanded combined uncertainty limit including uncertainty in regression for meter 2.

Figure 26 Predicted value at 10 mV with the expanded combined uncertainty limit including uncertainty in regression for meter 2.

3.3

Meter 3, no 158081, a 20 kW/m

2

Schmidt-Boelter

gauge

The results for Meter 3, a 20 kW/m2 Schmidt-Boelter gauge, are presented in Figure 27 and Figure 28for all participating laboratories over the entire range the used the meter at. Please note that data up to 21 kW/m² only is included in the comparison from Figure 29 and onwards despite the fact that NIST Optics had calibrated this meter up to 70 kW/m² due to exchange of meter 1 and meter 3. It should also be noted that LNE could not calibrate this meter at vacuum pressures due to the slightly smaller outer diameter of this meter, instead they conducted the calibration in their VBBC facility at normal pressure, this will however increase the convective contribution in this case.

27 28 29 30 31 32 33 Valu e at 5 m V, kW /m ²

156921

52 54 56 58 60 62 64 Valu e at 1 0 m V, kW /m ²

156921

Figure 27 Output signal as a function of Incident radiation for meter 3.

Figure 28 Incident radiation as a function of Output signal for meter 3.

Figure 29 Output signal as a function of Incident radiation for meter 3 up to 21 kW/m².

0.0 5.0 10.0 15.0 20.0 25.0 30.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 O ut put si gna l, m V Incident Radiation, kW/m² 158081 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9 0.0 10.0 20.0 30.0 40.0 50.0 60.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Inc ide nt R adi at io n, k W /m ² Output signal, mV 158081 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0.0 5.0 10.0 15.0 20.0 25.0 O ut put si gna l, m V Incident Radiation, kW/m² 158081 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9

Figure 30 Incident Radiation as a function of Output signal for meter 3 up to 21 kW/m². Figure 31 shows the change in results in the SP calibrations for meter 3 during the course of the Round Robin. A slight increase in responsivity could be seen over time, especially between SP7 and SP8, the difference is however lower than the expanded uncertainty of the method. There is however no immediate indication that the meter was damaged due to the high calibration levels conducted at NIST Optics (which was conducted between SP5 and SP6).

In order to study the results in more detail a mean calibration curve was calculated based on all the calibration results except for SP where only the SP5 results were included in order not to bias the mean of all results towards the SP results. SP5 was chosen as a value in the middle of the Round Robin exercise. This mean calibration curve was then deduced from the values for each of the laboratories. The result of this calculation is shown in Figure 32. Figure 33 shows the same expressed as percentages as a function of incident radiation.

Separate diagram has also been constructed for each of the labs together with the calibration results from SP conducted before and after that labs calibration. These diagrams are presented in Figure 34 - Figure 38.

0.0 5.0 10.0 15.0 20.0 25.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Inc ide nt R adi at io n, k W /m ² Output signal, mV 158081 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8 SP9

Figure 31 Change of calibration results at SP for meter 3 during the course of the Round Robin. A slight increase in radiation as a function of signal over time can be seen but all changes are within the uncertainty limits of the method indicated as Max and Min in the figure.

Figure 32 Spread in incident radiation as a function of output signal for the different laboratories for meter 3.

-0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0 5 10 15 20 25 Re sid ual f ro m m ean , k W /m ² Incident radiation, kW/m²

Meter 3

SP2 SP3 MAX MIN SP4 SP5 SP6 SP7 SP8 SP9 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 158081 SP2 SP3 LNE1 SP4 LNE2 SP5 Optics SP6 Fire SP7 SWRi SP8

Figure 33 Spread in incident radiation in percentages as a function of incident radiation for the different laboratories for meter 3.

Figure 34 Comparison of LNE VBBC results for meter 3 with SP results before and after.

Figure 35 Comparison of LNE spherical furnace results for meter 3 with SP calibrations before and after.

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 158081 SP2 SP3 LNE1 SP4 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 158081 SP4 LNE2 SP5

Figure 36 Comparison of NIST Optics lab VTBB results for meter 3 with SP calibration before and after.

Figure 37 Comparison of NIST Fire lab (ISO 14934-2 method 2) results for meter 3 with SP calibrations before and after.

-1.0 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 158081 SP5 Optics SP6 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 De viat io n f ro m m ean , k W /m ² 158081 SP6 Fire SP7

Figure 38 Comparison of SwRI (ISO 14934-2 method 2) results for meter 3 with SP calibrations before and after.

The differences between the laboratories are somewhat larger than the differences between subsequent calibrations at SP.

In addition, a linear regression was performed to determine a slope and intercept for the different calibration results of each calibration, these values are presented in Table 7. The standard deviation for the slope and intercept is also shown in Table 7. The standard deviation calculation includes only one of the calibrations at SP, i.e. SP5 in order not to lower the value. SP5 was chosen as an example value in the middle of the

inter-comparison.

The incident radiation for 10 mV was also calculated based on this regression and the results are presented in Table 7. The 10 mV level was chosen for comparison with the FORUM RR. Similar calculations as for the 10 mV level has also been conducted for the 5 mV level, both levels are within the range all labs have used for their calibrations. All the values are reported in Table 7 together with the combined expanded uncertainty for each level. The results are also presented in Figure 39-40.

Table 7 Slope and intercept for a linear regression, incident radiation as a function of output voltage for meter 3.

Calibration Slope Interce pt Value at 10 mV (kW/m²) Value at 5 mV (kW/m²) Combined expanded uncertainty at 5 mV (kW/m²) Combined expanded uncertainty at 10 mV (kW/m²) SP2 1.751 0.0695 17.6 8.8 0.2 0.4 SP3 1.7526 0.0386 17.6 8.8 0.2 0.4 LNE1 (LNE VBBC) 1.7879 0.8145 18.7 9.7 0.5 0.7 SP4 1.7603 0.0324 17.6 8.8 0.2 0.4 LNE2 (LNE SBBC) 1.8157 1.0503 19.2 10.1 0.6 0.8 SP5 1.7585 0.0717 17.7 8.9 0.2 0.2 NIST Optics VTBB 1.6992 -0.0047 17.0 8.5 0.1 0.4 SP6 1.7659 0.0369 17.7 8.9 0.2 0.4 NIST Fire 1.5435 0.1589 16.4 8.3 0.5** 0.9** SP7 1.7572 0.0282 17.6 8.8 0.2 0.4 SwRI 1.7665 0.0716 17.9 8.9 0.5** 1.0** SP8 1.779 0.03 17.8 8.9 0.2 0.4 SP9 1.7775 0.0395 17.8 8.9 0.2 0.4 Standard deviation* 0.072 0.45

* Only one SP value (SP5) included in the calculation in order not to decrease it inappropriate.

**Uncertainty based on uncertainty from method in SP furnace times two to include extra uncertainties due to unchecked temperature profile and sizes in furnaces

Figure 39 Predicted value at 10 mV according to linear regression obtained from the different calibrations for meter 3. The expanded combined uncertainty limit includes also the uncertainty from the regression

Figure 40 Predicted value at 5 mV according to linear regression obtained from the different calibrations for meter 3. The expanded combined uncertainty limit includes also the uncertainty from the regression

4

Discussion

Despite the intention of obtaining several values to compare between the participating laboratories only few comparable values are available. For the first meter the common calibration range only ranges from 1.2 mV to 2.6 mV. For meter 2 all laboratories

covered the agreed range i.e. 1.1 – 8.4 mV. For meter 3 the laboratories covered the range 5.0 – 10.4 mV but this meter was not calibrated in vacuum at LNE but in normal pressure which increases the convection contribution in that calibration facility. Figures 41-46 below show a comparison between the three calibration methods in ISO 14934-2 for the common calibration ranges.

Both the NIST Fire lab and SwRI used their newly bought equipment. They had not performed any calibrations/adjustments on their equipment before the RR started. The equipment used at SP during this Round Robin had replaced the old furnace that was used during the FORUM and HFCAL Round Robins. When this new equipment was taken into use two of the gauges used in one of the FORUM Round Robins were borrowed from NIST to check the new furnace. This check showed that a factory set controller adjustment had been made that needed to be removed. In addition the temperatures were checked on flanges etc. in the cooler and an approximation was made for the temperature of the cooler walls constructed as described in Appendix 2. In addition all dimensions were checked in the cooler which resulted in updated figures to be used in the incident radiation calculations.

As both NIST Fire lab and SwRI did not have their facilities fully set up at the time for the Round Robin their values are not included in the comparison, instead only the SP results are included. The SP value included is SP5, a values in the middle of the RR. For meter 1 the common calibration range is at the value of 2 mV. The plot in Figure 41 is based on the regression for the entire range for that meter for each of the laboratories while Figure 42 shows the value obtained from drawing a line through the value below 2 mV and the value above 2 mV for each calibration. This figure is included as this meter

was non-linear and therefore the outcome can be different depending on which range the labs covered. However, as seen the results are similar regardless of the comparison method. Method 1 and 2 are in agreement with each other while Method 3 gives another value. The difference is 12%.

Figure 41 Value at 2 mV for the methods in ISO 14934-2 for meter 1 based on the

regression from all 10 calibration points.

Figure 42 Value at 2 mV for the methods in in ISO 14934-2 for meter 1 based on the calibration level below 2 mV and the level above 2 mV for each of the labs.

Figure 43-44 shows the values for the three different methods at 2, 5 and10 mV for meter 3. As seen, the results are similar in difference between method 2 and 3 here while Method 1 gives a higher value for this meter. The difference between method 2 and 3 is 3-4% while the difference between method 1 and 2 is between 5 and 8 %.

14 15 16 17 18 19

Method 1 Method 2 Method3

Valu

e at

2

m

V f

ro

m

re

gr

es

sio

n

ov

er

clib

rat

io

n r

an

ge

, k

W

/m

²

156751

14 15 16 17 18 19

Method 1 Method 2 Method3

Valu

e at

2

m

V f

ro

m

p

oin

t

be

lo

w

a

nd

abo

ve

, k

W

/m

²

156751

Figure 43 Value at 2 mV for the methods in ISO 14934-2 for meter based on the regression from all 10 calibration points

Figure 44 Value at 5 mV for the methods in ISO 14934-2 for meter based on the regression from all 10 calibration points

Figure 45-Figure 46 shows the results for meter 3 at 5 and 10 mV respectively for method 2 and method 3. Method 1 is not included as it was not possible to perform this

calibration at vacuum according to method 1. As seen the results are similar to meter 2 with a difference between the two facilities of 4%.

11 11.5 12 12.5 13 13.5

Method 1 Method 2 Method 3

Valu

e at

2

m

V,

kW

/m

²

156921

28 28.5 29 29.5 30 30.5 31 31.5 32 32.5

Method 1 Method 2 Method 3

Valu

e at

5

m

V,

kW

/m

²

156921

Figure 45 Comparing the value at 10 mV for meter 3 Method 2 and Method 3 in ISO14934-2.

Figure 46 Comparing the value at 5 mV for meter 3 Method 2 and Method 3 in ISO14934-2.

If one studies all the participating laboratories results as presented in Figure 5, 18 and 33 one sees that the differences are up to 15% as for the lowest values of meter 3.

16.5 16.7 16.9 17.1 17.3 17.5 17.7 17.9 18.1 Method2 Method3 Valu e at 1 0 m V, kW /m ²

158081

8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.99 9.1 Method2 Method3 Valu e at 5 m V, kW /m ²

158081

5

Comparison with previous Round Robins

Three of the participants in this Round Robin participated in the Round Robin organized by FORUM in 2000-20042: the LNE vacuum furnace, SP and NIST Optics. At that time the SP calibration facility was not considered as a primary calibration facility and therefore no detailed comparison was conducted between SP and the other facilities. However, it is possible to extract some results from the NIST report as presented in Table 8 and Table 9.

Table 8 Slope, intercept and Value at 10 mV for Schmidt-Boelter Gauge in Forum Round Robin

Facility Slope Intercept Value at 10 mV

LNE VBBC 8.995 0.37 90.32

NIST Optics VTBB 8.640 - 86.4

SP 9.055 0.375 90.92

Table 9 Slope, intercept and Value at 10 mV for Gardon Gauge in Forum Round Robin

Facility Slope Intercept Value at 10 mV

LNE VBBC 12.083 0.62 121.45

NIST Optics VTBB 11.73 - 117.3

SP 11.912 1.26 120.38

As seen the results are in close agreement with each other for SP and LNE while the result for NIST Optics differ about 4%. Similar results extracted from Table 5, 6 and 7 are presented in Table 10. As seen the difference between SP and NIST Optics is in the same order as in the FORUM Round Robin while the difference against the LNE vacuum facility is larger in this Round Robin.

Table 10 Difference between SP facilities and the other facilities in this RR at 10 mV Difference between SP and facility, kW/m2 and (%)

Facility Meter 1 Meter 2 Meter 3

LNE VBBC 1.5 kW/m² (1.8%) 2.6 kW/m² (4.2%) 1.0 kW/m² (5.7%)** NIST Optics VTBB -3.9 kW/m² (-5%) -2 kW/m² (-3.4%) -0.7 kW/m² (-3.8%) * Calibration only performed over a small range

** Calibration not performed in vacuum

The results from the Round Robin in HFCAL3 are presented in Table 11 and Table 12. As seen the results were in close agreement for the Schmidt Boelter gauge while the results differed more for the Gardon meter in that Round Robin.

Table 11 Slope intercept and Value at 10 mV in HFCAL RR Schmidt-Boelter

Facility Slope Intercept Value at 10 mV

LNE VBBC 3.48 0.33 35.13

SP 3.44 0.3 34.7

Table 12 Slope intercept and Value at 10 mV in HFCAL RR Gardon

Facility Slope Intercept Value at 10 mV

LNE VBBC 6.08 1 61.8

6

Concluding remarks

The results indicate a somewhat larger difference between the methods in ISO 14934-2 in this RR compared to those conducted previously. The spread in results is indeed larger than the ±3% as stated in the ISO 14934-2 (about ± 5% considering mainly the results for meter 2). If one focuses on the 10 mV value as in the previous RRs then the results lies within ±4% based on the 10 mV as presented in Table 5 and Table 6 (meter 1 and meter 2).

Since there is no instrument which can give an absolutely true value of the incident radiation it is possible that the participating laboratories should add adjustments to their readings to get a better agreement between all the primary methods. An example of a adjustments method is given in Appendix 2. However, as the results of this RR gave varying results depending on the different meters the data is currently too limited in order to take any decisions on this.

The comparison between all participating labs and methods in this RR showed differences up to 15%.

7

References

1 _{ISO 14934-2:2006 Fire tests – Calibration and use of heat flux meters – Part 2 Primary}

calibration methods, First ed. 2006-02-15

2 _{W. M. Pitts, A. V. Murthy, J. L. de Ris, J. R. Filtz, K. Nygård, D. Smith, I. Wetterlund, Round}

Robin Study of Total Heat Flux Gauge Calibration at Fire Laboratories, National Institute of Standards and Technology Special Publication 1031, Gaithersburg, MD, USA, 2004

3

J-R. Filtz, M. Lievre, T. Valin, J. Hameury, I. Wetterlund, B. Persson, P. Andersson, R. Jansson, T. Lemaire, M. Öhlin, J. Myllymäki Improving heat flux meter calibration for fire testing

Appendix 1 Ageing of the meters

The ageing was conducted at 900 °C (80% of measuring range = 55 kW/m²) for Meter 1 and Meter 2 and at 585 ºC (80 % of measuring range = 16 kW/m²) for Meter 3. After an initial calibration of each meter roughly the following procedure was followed:

1. The meter was mounted in the furnace and a 2 min measurement was conducted when the temperature (900 °C or 585 °C) had stabilized, this was determined by checking the temperature and output signal reading.

2. The meter was then left in the furnace for 1 to 2 h before the next 2 min measurement was conducted.

3. The meter was removed from the furnace and left outside for 15 min. Step 1-3 was then repeated until at least 6 hours of ageing had taken place The outcome of the procedure is described in Table 13- Table 15. The measurement points are presented in Figure 47 - Figure 49.

Table 13 Meter 1

Date Comment

Oct 08 -08 First calibration SP1 Oct 09 -08

at 13:55

Meter mounted in furnace and wait for stabilisation in 75 min. First ageing measuring point taken. Meter left in furnace

Oct 09 -08 at 15:13

Second ageing point taken, i.e. close in time to first point, meter left in furnace for 30 min

15:45 Third ageing point taken. Meter taken out of furnace Oct

Oct 20 -08 at 08:05

Meter mounted in cold furnace. Wait for temperature rise and stabilisation for more than 4.5 h

Oct 20 -08 At 12:45

Fourth ageing measuring point taken

12:52 Fifth ageing measuring point taken, meter removed from furnace for ten minutes

13:02 Meter put back in furnace, sixth point taken after 1.5 h, meter taken out of furnace for 15 min

14:49 Seventh measuring point taken, meter left in furnace for 1 h 15:00 Eighth measuring point taken, meter taken out of furnace Oct 21 -09

at 06:46

Ninth measuring point taken Second calibration at SP SP2 Third calibration at SP SP3 Total approx. 8 hours

Table 14 Meter 2

Date Comment

Sep 22 -08 First calibration SP1 interrupted before last point due to customer need Sep 29 -08

at 07:48

Meter mounted in cold furnace, wait for heat up and stabilisation and first ageing measure point taken after 4.5 h, meter left in furnace

Sep 29 -08 at 12:18

Second ageing measure point taken (i.e. first and second point taken close to each other), meter left in furnace for 88 min

13:48 Third ageing measure point taken, meter taken out of furnace

14:12 Meter mounted in furnace, Fourth point taken, meter left in furnace for 89 min

14:41 Fifth point taken meter taken out of furnace Sep 30 -08

at 06:50

Meter mounted in furnace 700°C, wait for heat up and stabilisation. Sixth point taken after 86 min (08:16) Meter left in furnace

08:19 Seventh point taken after 76 min (09:35) 1st of oct Second calibration at SP SP2

6th of oct Third calibration at SP SP3 Total: about 10 hours

Table 15 Meter 3

Nov 04 -08 First calibration SP1 conducted

Nov 05 -08 Meter mounted in furnace for 1 h + 1 h (out of furnace for 10 min in between), Measuring point taken at 16:50, meter taken out of furnace Nov 06 -08

at 06:48

Meter mounted in furnace, Measuring point taken after 1 h, meter left in furnace

Nov 06 -08 at 08:40

Measuring point taken, meter left in furnace

09:30 Measuring point taken, meter taken out of furnace for 15 min 09:52 Measuring point taken, meter left in furnace for 135 min 12:07 Measuring point taken, meter taken out of furnace for 15 min 12:44 Measuring point taken, meter left in furnace for 135 min 16:00 Measuring point taken

Nov 10 -08 Second calibration at SP SP2 Nov 12 -08 Third calibration at SP SP3 Total: about 10 h

Figure 47 Ageing measuring points meter 1.

6.75 6.76 6.77 6.78 6.79 6.8 6.81 6.82 6.83 6.84 6.85 900.5 900.6 900.7 900.8 900.9 901 901.1 901.2 901.3 901.4 901.5 0 2 4 6 8 10 O ut put si gna l, m V Fur na ce te m pe ra tur e, °C Measuring point

Meter 1

Temperature Output signal

Figure 48 Ageing measuring points meter 2.

Figure 49 Ageing measuring points meter 3.

After the ageing process the meters were calibrated twice at SP before being sent out to the other laboratories. The calibration results before and after ageing are presented in Figure 50- Figure 52 together with the combined expanded uncertainty limits of the furnace at SP indicated as MAX and MIN in the figures.

9.1 9.12 9.14 9.16 9.18 9.2 900.5 900.6 900.7 900.8 900.9 901 901.1 901.2 901.3 901.4 901.5 0 1 2 3 4 5 6 7 8 O ut put si gna l, m V Fur na ce te m pe ra tur e °C Measuringpoint

Meter 2

Temperature Output signal 8.75 8.77 8.79 8.81 8.83 8.85 586 586.1 586.2 586.3 586.4 586.5 586.6 586.7 586.8 586.9 587 0 2 4 6 8 10 O ut put si gna l m V Fur na ce te m pe ra tur e °C Measuringpoint

Meter 3

Temperature Output signal

Figure 50 Calibrations before and after ageing of Meter 1

Figure 51 Calibrations before and after ageing of Meter 2

Figure 52 Calibrations before and after ageing of Meter 3

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 60 70 80 Re sid ual f ro m m ean , k W /m ² Incident Radiation, kW/m² Meter 1 SP1 SP2 SP3 MAX MIN -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 60 70 80 Re sid ual f ro m m ean , k W /m ² Incident radiation, kW/m² Meter2 SP1 SP2 SP3 MAX MIN -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0 5 10 15 20 25 Re sid ual f ro m m ean , k W /m ² Incident radiation, kW/m² Meter 3 SP1 SP2 SP3 MAX MIN

Appendix 2 Adjustments made to incident radiation calculation

and furnace control at SP

Upon arrival of the furnace bought from Mikron it turned out that the measurement thermocouple and thermocouple used for controlling the furnace showed quite different temperatures. This was due to an adjustment that Mikron had made to the controlling thermocouple. This adjustment was removed.

Temperature of cooler

The cooler is divided into two parts in the calculations, the upper and lower part. These two sections are assumed to have a uniform temperature each. The temperature of each part is the mean value of several temperatures in different locations in the cooler and it is assumed that this mean value is representative for the relevant overall cooler temperature. The mean value is calculated from a regression on temperature measurements performed on the flanges in the cooler (sight tube), on the spacer and in the holder. Figure 53 shows the measurement spots. It turned however out that it was not possible to mount any TC on spot 1 and 7 in this equipment and therefore estimates for this temperature was made based on measurement sin the old furnace and the measurements in the new furnace. The temperature readings are presented in Figure 54- Figure 56 together with the old readings.

T/C 1 T/C 2 T/C 3 T/C 4 T/C 5 T/C 6 T/C 7

Figure 53 Cross-section of the sight tube (given in Figure F.2), the spacer (given in Figure F.5), and the upper part of the holder (given in Figure F.3 of ISO 14934-2). Thermocouples were mounted at positions T/C 2, T/C 3, T/C 4, T/C 5 och T/C 6. (see also Figure 12 in SP REPORT 1991:58)

Figure 54 Temperature of cooler in new (Pink markings “N-0427”) and old (lines “–G”) furnace

Figure 55 Temperature of cooler in new (Pink markings “N-0427”) and old (lines “–G”) furnace With spacer 0 20 40 60 80 100 120 140 160 180

-5.0E+04 0.0E+00 5.0E+04 1.0E+05 1.5E+05 2.0E+05 2.5E+05

σ(Tf4-Tw4) C o o ler t em p er at u res ( d eg C ) "T/C2-N-0427" "T/C3-N-0427" T/C 1-G T/C 2-G T/C 3-G With spacer 0 20 40 60 80 100 120 140 160 180

-5.0E+04 0.0E+00 5.0E+04 1.0E+05 1.5E+05 2.0E+05 2.5E+05

σ(Tf4-Tw4) C o o ler t em p er at u res ( d eg C ) "T/C4-N-0427" "T/C5-N-0427" "T/C6-N-0427" T/C 4-G T/C 5-G T/C 6-G T/C 7-G

Figure 56 Temperature of cooler in new (Pink markings “N-0427”) and old (lines “–G”) furnace

The measurements resulted in that the equations for calculating the temperature of the cooler for the calculation of incident radiation had to be changed. The new equations were determined to :

Without spacer ring:

)

(

00099

,

0

06

,

10

4 4 2

T

w

T

f

T

w

T

=

+

+

⋅

σ

−

)

(

00078

,

0

16

,

9

4 4 3

T

w

T

f

T

w

T

=

+

+

⋅

σ

−

With spacer ring:

)

(

0010

,

0

34

,

10

4 4 2

T

w

T

f

T

w

T

=

+

+

⋅

σ

−

)

(

00052

,

0

11

,

3

4 4 3

T

w

T

f

T

w

T

=

+

+

⋅

σ

−

Reflections in cooler

We made sure that the surfaces visible from the furnace, see Figure F.1 in ISO 14934-2, were treated to give a matt black surface with low reflection. This was done using the same type of paint that is used for Gardon gauges.

Comparison with previous furnace

Two of the meters used in the FORUM RR were borrowed and the calibration of these in the new furnace was checked against the results in the FORUM RR. It proved that the measures taken above were sufficient to get similar results as in the FORUM RR.

Comparison with and without spacer ring

Calibrations were performed on several meters to make sure that the results were similar when a spacer ring were used and when no spacer ring was used. It turned out that one more adjustment was needed to achieve similar results with and without spacer ring. An example of this exercise is shown in Figure 57. This resulted in an adjustment of 4 °C for the case with spacer ring as the case without spacer ring was the case that had been compared with the FORUM RR results.

No spacer 0 20 40 60 80 100 120 140 160 180

0.0E+00 5.0E+04 1.0E+05 1.5E+05 2.0E+05 2.5E+05

σ(Tf4-Tw4) C o o ler t em p er at u res ( d eg C ) "T/C 6-N-utan" T/C 1-G-utan T/C 6-G-utan T/C 7-G-utan

Figure 57 Comparison with and without spacer ring.

Control of all distances and sizes

All distances and sizes of the cooler where checked to see if they were in compliances with the drawings in ISO 14934-2 method 2. Small adjustments were made to the calculation schemes in reflection of these measurements

y = 8.53x + 0.0498 y = 8.5512x + 0.0227 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 0.00 1.00 2.00 3.00 4.00 H e a t f lu x [k W /m ²] Output signal [mV] 701226-without-makroD 701226-with-makro090112-4gr Linjär (701226-without-makroD) Linjär (701226-with-makro090112-4gr)

Appendix 3 Reported uncertainties

LNE and SP reported their combined expanded uncertainties for each calibration level. These values are presented in Table 16- Table 18. NIST Optics uncertainty comes from their special publication 250.651. NIST Fire and SwRI did not report any uncertainty calculations, no values has been estimated for their calibrations in the tables below. Table 16 Combined expanded uncertainty for Meter 1.

Level LNE VBBC SP4 LNE SBBC NIST optics

7 0.29 0.11 0.58 0.15 14 0.38 0.20 0.66 0.29 21 0.53 0.28 0.77 0.44 28 0.75 0.35 0.878 0.59 35 0.92 0.43 1 0.73 42 1.11 0.50 1.3 0.88 49 1.3 0.57 1.4 1.03 56 1.5 0.64 1.5 1.2 63 1.9 0.70 1.8 1.3 70 2.1 0.77 2 1.5

Table 17 Combined expanded uncertainty for Meter 2.

Level LNE VBBC SP4 LNE SBBC NIST optics

7 0.26 0.11 0.57 0.15 14 0.51 0.20 0.65 0.29 21 0.58 0.28 0.76 0.44 28 0.78 0.36 0.89 0.59 35 0.93 0.43 1.1 0.73 42 1.1 0.50 1.5 0.88 49 1.3 0.57 1.5 1.03 56 1.5 0.64 1.8 1.2 63 1.5 0.71 1.8 1.3 70 1.7 0.778 2 1.5

Table 18 Combined expanded uncertainty for Meter 3.

Level LNE VBBC SP4 LNE SBBC NIST optics

2 0.10 0.05 0.042 4 0.18 0.09 0.084 6 0.24 0.13 0.13 8 0.30 0.16 0.17 10 0.32 0.20 0.58 0.21 12 0.38 0.24 0.61 0.25 14 0.44 0.27 0.65 0.29 16 0.51 0.31 0.67 0.34 18 0.59 0.34 0.71 0.38 20 0.65 0.38 0.75 0.42

1 _{Tsai, B., Gibson, C., Murthy, A., Early, E. Dewitt, D. and Saunders, R. ”Heat-Flux” Sensor}

SP Technical Research Institute of Sweden

Box 857, SE-501 15 BORÅS, SWEDEN

Telephone: +46 10 516 50 00, Telefax: +46 33 13 55 02 E-mail: info@sp.se, Internet: www.sp.se

www.sp.se

Fire Technology

SP Technical Note: 2011:05 ISSN 0284-5172

More information about publications published by SP: www.sp.se/publ

SP Technical Research Institute of Sweden

Our work is concentrated on innovation and the development of value-adding technology. Using Sweden's most extensive and advanced resources for technical evaluation, measurement technology, research and development, we make an important contribution to the competitiveness and sustainable development of industry. Research is carried out in close conjunction with universities and institutes of technology, to the benefit of a customer base of about 9000 organisations, ranging from start-up companies developing new technologies or new ideas to international groups.