report no.

2/17**

An evaluation of an
optical gas imaging
system for the
quantification of fugitive
hydrocarbon emissions
 report no. 2/17

An evaluation of an optical
gas imaging system for the
quantification of fugitive
hydrocarbon emissions
Prepared by the Concawe Air Quality OGI Ad-Hoc Group (Optical Gas Imaging):

C. Caico
L. Fragu
L. Gonzalez
C. Juéry
P. Kangas
C. Lawson
J. Negroni
P. Roberts
B. Smithers
P. Tupper
K. Vaskinen

Reproduction permitted with due acknowledgement

 Concawe
Brussels
January 2017

I
 report no. 2/17

ABSTRACT

This report provides the results of a preliminary study undertaken to test the
performance of a quantitative optical gas imaging (QOGI) system. This has been
developed to measure the mass emission rate of hydrocarbon leaks from industrial
process equipment.

Testing was performed using controlled releases in the range 1.7 g/h to 1000 g/h. For
comparison purposes a conventional sniffing technique was also used for leak
detection with emission estimates determined using the correlation equations from
US EPA Method 21.

Sixty one leak tests were performed. The QOGI system detected all of these leaks
but it was found that quantification required a differential temperature of > 5°C
between the released gas and the background.

Where leak rate quantification was achieved with the QOGI system the differences
between the values determined and the known release rates were within a range of
-23% to 69%, with an average difference of 6%. By comparison, where the US EPA
Method 21 correlations were used to estimate leak rates these differed from the
known release rates within a range of -92% to 667%, with an average difference of
31%.

KEYWORDS

Fugitive emissions, quantification, LDAR, optical gas imaging, OGI, quantitative

optical gas imaging, QOGI, refinery.

INTERNET

This report is available as an Adobe pdf file on the Concawe website

([Link]).

NOTE
Considerable efforts have been made to assure the accuracy and reliability of the information
contained in this publication. However, neither Concawe nor any company participating in
Concawe can accept liability for any loss, damage or injury whatsoever resulting from the use of
this information.

This report does not necessarily represent the views of any company participating in Concawe.

II
 report no. 2/17

CONTENTS Page
1. INTRODUCTION 1
1.1. CURRENT USE OF OPTICAL GAS IMAGING (OGI) 1
1.2. OGI OPERATING PRINCIPLES 1
1.3. QOGI METHOD: FROM QUALITATIVE TO QUANTITATIVE 1
1.4. STUDY OBJECTIVE AND OVERVIEW 2

2. LEAK GENERATION EQUIPMENT & METHODS 5

2.1. VITO LDAR INSTALLATION 5
2.2. GENERATION OF KNOWN LEAK RATES 5
2.2.1. MFC Calibration 6

3. LEAK DETECTION METHODS 7

3.1. OPTICAL GAS IMAGING (OGI) CAMERA 7
3.1.1. OGI camera sensitivity check 7
3.2. TVA 1000B AND DILUTION KIT 7

4. LEAK QUANTIFICATION METHODS 8

4.1. QUANTITATIVE OPTICAL GAS IMAGING DEVICE (QOGI -
MODEL QL100) 8
4.1.1. QL100 synchronisation with OGI camera 8
4.1.2. Response factors 9
4.1.3. Temperature and distance parameters 9
4.1.4. Use of enhanced backgrounds 9
4.2. EPA CORRELATION APPROACH (METHOD 21) 9

5. TEST RESULTS 11
5.1. PRE-TEST SUMMARY 11
5.2. FULL TEST RESULTS 15
5.3. CALCULATED EMISSION RATES 19
5.3.1. Method 21 vs. Generated leak rate 19
5.3.2. QOGI vs. Generated leak rate 19
5.3.3. QOGI vs. Method 21 20

6. CONCLUSIONS 22

7. REFERENCES 23

ANNEX A – MASS FLOW CONTROLLER (MSC) CALIBRATION DATA 24

III
 report no. 2/17

SUMMARY

This report presents the results from the preliminary testing of a quantitative optical
gas imaging (QOGI) system, consisting of an infrared optical gas imaging (OGI)
camera and a quantification module. The QOGI system allows an operator to detect
and quantify mass emission rates of hydrocarbon gases.

The objective of this study was to assess the ability of the QOGI system to detect
fugitive emission sources and to quantify their mass release rates using a number of
controlled but realistic releases. The study was conducted at VITO’s leak detection
and repair (LDAR) training facility in Mol, Belgium in Summer 2015.

OGI cameras allow leaks of hydrocarbon substances to be visualised, which can be

very useful when conducting LDAR programs. The operating principle is
straightforward and uses the fact that hydrocarbon gases absorb infra-red light. If gas
passes between the camera and a reference surface, light is absorbed by the gas
and an image will be formed. OGI cameras are very good visual tools, but currently
their use in LDAR programs is limited to detection as they are considered to be
applicable only for qualitative monitoring.

The QOGI technology used for this study [7] is a patented technology designed to
work with the FLIR GF 300/320 series of infra-red (IR) cameras [5]. The QL100
quantification module uses the image information from a OGI camera and
supplementary data (distance from camera to leak source, environmental conditions
and gas optical properties) to calculate the hydrocarbon mass emission rate. This
QOGI system is designed to fulfil the needs of an LDAR program to detect
hydrocarbon leak sources and determine their mass rates of release.

The LDAR installation was used to generate controlled hydrocarbon releases

simulating leaks of varying sizes from different process plant components (i.e. valves,
open-ended pipes and flanges). The release rates ranged from 1.7 to 1000 g/h. All
the gas releases took place at ambient temperature. Because the QOGI technique
uses naturally occurring infra-red radiation and not a reference source of radiation (an
‘illuminant’), a wide range of camera positions, weather conditions and background
scenes were explored. For comparison purposes, a conventional sniffing technique
(EN 15446:2008) [1] was used for leak detection and the US EPA Method 21
(Correlation Approach) [9] was used to estimate the emissions of the leaking
components.

A total of 61 scenarios were explored in two test periods:

The first series of tests (hereafter called the ‘pre-test’) used the normal backgrounds
available at the test facility with the aim of identifying conditions suitable for detection
and quantification. A total of 37 releases were carried out.

The second series of tests (hereafter called the ‘full test’) introduced some artificial
backgrounds where necessary to enhance the difference in temperature between the
background surface and the ambient air. A total of 24 releases were carried out.

IV
 report no. 2/17

The results showed that all 61 releases could be detected using the camera even
though the environmental conditions were such that there was little difference
between the temperatures of the gas and the background facility structure which
included various built materials. This is fully consistent with experience with the FLIR
GF 300/320 camera series that gas can be detected if there is temperature difference
as small as 1 to 2 degrees between the gas and the contrasting background.

The QOGI system provided quantified leak rates for 31 of the 61 detected releases.
This is because quantification requires a greater temperature difference, T, between
the gas and the background than for detection. From these tests it was found that the
current technology required a T > 5 C in order to successfully quantify emissions.

In the pre-test period, 7 out of the 37 release scenarios were quantified by the QOGI
system and for each of these scenarios the T was in excess of 5C.

In the full test period all 24 release scenarios were quantified by the QOGI system
with the backgrounds being adjusted when necessary, using a temperature
enhancement device, to ensure T > 5C.

It is foreseen that T will be an important criterion or Data Quality Indicator (DQI) for
any QOGI method. Furthermore, when a specific test method is developed for QOGI,
a refined criterion for T must be considered.

Where leak rate quantification was achieved with the QOGI system, the differences
between the values determined and the known release rates were within a range of -
23% to 69%, with an average difference of 6%. By comparison, where the US EPA
Method 21 correlations were used to estimate leak rates these differed from the
known release rates within a range of -92% to 667%, with an average difference of
31%.

Overall the QOGI results from this evaluation were very promising. This was a
demanding test of the technique because there was no operational heat release on
the test facility and hence both the released gas and all surfaces were close to
ambient temperature. Work needs to be done to refine the criterion that T has to be
> 5C to achieve quantification and also to identify quick and effective means of
providing an enhanced background to increase T when needed. On an operational
site with heat release and fluids at process temperatures the likelihood of needing
such action is less than during the controlled release conditions. This should be
considered as a factor in future investigations.

The QOGI technology evaluated in this study was still in the development stage at the
time of these tests in June of 2015. It is recommended that further testing using the
latest version of the technology is undertaken in field conditions. .

V
 report no. 2/17

1. INTRODUCTION

1.1. CURRENT USE OF OPTICAL GAS IMAGING (OGI)

Infrared cameras have gained increasing application in recent years in the oil and gas
(O&G) industry for their ability to rapidly detect leaking equipment. As well as their
use in LDAR programmes, OGI inspections are typically performed after maintenance
activities, to investigate odour sources, follow up on alarms from other sensors, etc.
OGI therefore now plays an important role in the safety and environmental programs
in both upstream and downstream oil refining and also petrochemical facilities.

In the petrochemical sector, the US EPA has promulgated an Alternative Work

Practice (AWP) [10] which allows operators to use OGI as part of an LDAR program.
However, the productivity advantage of using OGI in US LDAR programs is negated
by the requirement to also perform Method 21 at least once annually. Hence, there
are no facilities in the US currently utilizing OGI in LDAR programs. In Europe, OGI is
considered as a Best Available Technique (BAT) for LDAR (in the refining BREF [4]
and in the common waste water and gas treatment/management BREF [3]). There
are some European countries that accept OGI as a stand-alone method for LDAR.

1.2. OGI OPERATING PRINCIPLES

The operating principle of the OGI camera is to detect the absorption of infra-red light
by a gas plume. The base infra-red signal originates from a background, shown
uniform in Figure 1 below. Infra-red (IR) radiation flux is related to temperature. If the
background is warmer than the released gas (IB > IG) then the gas appears dark; if it
is cooler the gas appears white.

1
 report no. 2/17

Figure 1 Impact of gas plume on infra-red signal from background (IB =

background Intensity, IG= Gas Intensity) [12] (Courtesy of
Providence Photonics)

The OGI camera differs from other instruments that use light absorption to detect VOC
emissions in that it uses background radiation as the source of radiance. Therefore,
for an image to be recorded there must be a temperature difference (T) between the
gas and the background. Other techniques, e.g. differential infra-red absorption
(DIRA), use a controlled radiant source and at least two wavelengths, one absorbed
and one not absorbed, to make an absolute measurement of the amount of light
absorbed by the molecules in the light path between the camera and background.
Such systems either use a fixed radiation source and sensor or, for more easy
alignment, a retro-reflector and co-aligned sensor and source.

Another feature of the OGI camera is its narrow bandpass filter. A given gas is only
detected if its infra-red absorption spectrum overlaps the absorption band of the
camera. The minimum detection limit for a given gas depends on the number of
molecules between the camera and the background.

For more details of the OGI camera used for these tests, see reference [5].

1.3. QOGI METHOD: FROM QUALITATIVE TO QUANTITATIVE

In a step beyond using OGI for detection purposes only, a new methodology for
deriving an emission rate from an OGI camera image has been developed [7]. This
Quantitative Optical Gas Imaging (QOGI) system uses a conventional OGI camera
coupled to an external device that runs analysis software. Prior to QOGI, OGI
techniques were effective visual tools but had limited application for LDAR compliance
since they were only qualitative. Consequently the American Petroleum Institute (API)
developed OGI leak/no-leak factors [6] to permit the quantification of VOC emissions
rates from leaks detected by an OGI camera.

1
 report no. 2/17

Two parameters affecting the performance of the QOGI system to detect and quantify
releases (see Figure 2) are the temperature difference between the released gas and
the background (T) and the total number of molecules in a line of sight through the
plume to the background, referred to as concentration-path length (CL). This path-
integrated concentration is measured using the brightness of each pixel in the image.
The system will have detection/quantification limits represented by a minimum value
of CL and this minimum value can be expected to decrease as T increases.

A further parameter, considered fixed in this work, is the sensitivity of the camera to
the wavelength that is absorbed. This is set using a specific wavelength filter on the
camera matched to the target gas. Because many hydrocarbons have similar
absorption spectra, response factors can be used to account for different target gas
compositions.

Figure 2. The parameters that can affect the performance of the QOGI
system [12] (Courtesy of Providence Photonics)

1.4. STUDY OBJECTIVE AND OVERVIEW

This Concawe report describes the first evaluation trials of QOGI in a European
context. A series of controlled releases, comprising realistic simulations of emissions
from process plant leaks, were carried out at the VITO LDAR training facility. A QOGI
system was used to detect and quantify the releases found and the results compared
to the known release rates. Additionally, a traditional sniffing method was used to
detect the emissions and the Method 21 correlations were used to provide calculation-
based leak estimates for comparison purposes.

The parameters in Table 1 were explored to determine how the technology performs
under a variety of conditions. Similar conditions were used in earlier controlled tests
performed by Concawe [2] and are believed to be representative of field LDAR.

2
 report no. 2/17

Table 1 Key parameters and variables explored during testing

Key Parameters Types / Ranges

Background scene brick wall, concrete, metal, sky
Leaking component flange, valve, open-ended pipe
Volatile organic gas methane, propane, propylene, a mixture of the
three (~33% each)
Leak rate 1.7 – 1000 g/h
Camera distance from leak source (meters) 2, 3, 5 and 10

Test matrices were developed to encompass a combination of the aforementioned

parameters resulting in a total of 61 testing scenarios. Backgrounds similar to those
found in refineries, such as concrete and pipe/metalwork, were used. The range of
leak rates generated reflected the range of emissions that might be observed in the
field. Propane, propylene, methane and a mixture of these 3 gases were used to
generate the leaks.

For each test release the OGI camera was used to scan for and detect the source of
emission. When a leak was detected, images of the leak were recorded by the OGI
camera (attached to a tripod to steady the image) and the QL100 device was
connected via a USB cable to the OGI camera to analyse the IR camera images and
to calculate the emission rate of the leaking gas in g/h (see Figure 3). For each leak,
three scans of 30 seconds each were recorded and the average leak rate determined.

Meteorological conditions, i.e. wind direction and speed, ambient air temperature,
relative humidity and the barometric pressure, were measured using an All-in-One
Compact Weather Station. Using a common date/time reference the data were logged
every 10 seconds to allow for synchronisation with the recorded plume images.

3
 report no. 2/17

Figure 3 Set-up of QOGI system used during the study (Courtesy of

Providence Photonics)

4
 report no. 2/17

2. LEAK GENERATION EQUIPMENT & METHODS

2.1. VITO LDAR INSTALLATION

The LDAR installation at the VITO facility is designed to generate releases simulating
leaks from various equipment pieces (e.g. flanges, valve stems, valve bonnets, and
open-ended pipes). All of the controlled leak tests using one of the three pure gases
(propane, methane, propylene) or an approximately equal mixture of the three were
performed using this installation. A schematic overview of the LDAR installation at the
VITO site in Mol is presented in Figure 4.

Figure 4 Schematic overview of the LDAR installation

2.2. GENERATION OF KNOWN LEAK RATES

A mass flow controller (MFC) was used to generate the mass flow rate. A MFC is a
self-regulating device that can generate a stable flow rate by comparing the input
signal provided by the operator (set-point) to the value from the mass flow sensor and
adjusting the proportional valve accordingly to achieve the required flow. Any back
pressure that might be caused by the LDAR installation does not influence the flow
rate of the MFC.
MFCs with the following ranges were used to generate known leak rates:

 Propane:
o 0 - 10 g/h
o 0 - 300 g/h
o 0 - 1200 g/h
 Propylene:
o 0 - 5 g/h
o 0 - 200 g/h
 Methane:
o 0 - 200 g/h.

5
 report no. 2/17

2.2.1. MFC Calibration

VITO calibrated the MFCs using a mercury sealed piston prover. This is subjected to
VITO’s ISO 17025 accreditation and allows them to calibrate MFCs with an
uncertainty of 0.3%. The results of these calibrations are reported in Annex A.

The calibration of the MFCs was performed before the first test day and at the end of
the final test day. Each set point that was used was calibrated with propane. The
stability of the mass flow controller was monitored during each test. The flow was
required to be constant within 1% within the elapsed time of the test or it had to be
repeated. In practice, during the tests all releases met the stability condition. The
system was calibrated to deliver fixed flow rates of propane, methane, propylene or a
mixture of these three gases in approximately equal proportions. Figure 5 shows a
picture of the MFCs used during the study.

Figure 5 Mass flow controllers used during the study (Courtesy of

Providence Photonics)

6
 report no. 2/17

3. LEAK DETECTION METHODS

3.1. OPTICAL GAS IMAGING (OGI) CAMERA

The FLIR GF320 camera was used to survey the controlled leaks. The camera utilised
a 38 mm optical lens and was equipped with the latest firmware version. Prior to the
study, the camera was synchronized to the quantification module. This accounts for
any differences between individual OGI camera responses and is a required step for
the QOGI method. The process takes place the first time a camera is used with a
module. The synchronization took a couple of hours and was conducted at the VITO
test facility.

3.1.1. OGI camera sensitivity check

Prior to each day of testing, the camera’s ability to detect a 5 g/h leak of propylene
was verified in the field. Propylene was selected for verification purpose because it is
one of the gases having the highest detection limit for the camera used in the study.
This approach is consistent with the suggested method in the US EPA LDAR
Alternate Work Practice (AWP) [10].

3.2. TVA 1000B AND DILUTION KIT

The Toxic Vapour Analyser (TVA 1000B) applied in the sniffing method was used in
flame ionisation detector (FID) mode to record the screening value (gas concentration
in ppmv) of the leaking components before and after testing, consistent with EN
15446:2008 [1]. This screening value, corrected for the response factor of the actual
gas, was used to estimate the leak rate with the EPA Method 21 correlations (see
Section 4.2).

The working principle of the flame ionization detector (FID) is to ionize the
hydrocarbons in a combustion chamber utilising a flame produced by the combustion
of hydrogen and air. The ions are subsequently attracted to a collector electrode, and
quantified based on the current produced. When there is not enough air in the vicinity
of the leak (e.g. at elevated hydrocarbon concentration), the combustion cannot take
place and the FID provides no response. This is referred to as ‘flame-out’.

A dilution probe can be used to enrich oxygen deficient samples by adding ambient
air to the combustion chamber. The use of a dilution probe allows flame-out to be
avoided and this enables the recording of screening value for the larger leaks, which
permits their mass estimation with the Method 21 correlations. A dilution probe
(Century Dilutor Kit; part No. CR010MR) was applied during some of the tests (i.e.
those with the highest leak rates).

The analyser (TVA 1000B) was calibrated each day prior to use and drift checked
throughout the day to evaluate the bias and accuracy of the screening measurements
used for the EPA Correlation Approach (Method 21) - see Section 4.2. Zero air and
500 and 10,000 ppmv methane-in-air were used for daily analyser calibration. The
500 ppm standard was used for analyser drift checks throughout the day and at the
end of testing each day.

7
 report no. 2/17

4. LEAK QUANTIFICATION METHODS

4.1. QUANTIFICATION MODULE

The quantification module utilizes proprietary algorithms to automatically calculate

mass leak rates in g/hr andanalyses IR images of a leak to determine the intensity on
a pixel-by-pixel basis. Each pixel represents a column of hydrocarbon vapour
between the camera and the background. Pixel contrast intensity is a function of
temperature difference (∆T) between the background and the plume. At a given ∆T,
the contrast intensity is proportional to the number of hydrocarbon molecules in the
vapour column. The leak rate drives the pixel intensity and the number of pixels
covering the plume. US Patent 9225915 B2 [11] provides a more detailed description
of the method.

4.1.1. OGI camera and quantification module synchronisation

The quantification module must be synchronised to the specific OGI camera being
used. This is needed to account for variations between OGI cameras and must be
performed for the temperature range setting to be used and the lens.

All of the data captured during the study utilised the 38mm lens and the 10 - 60C
temperature range. It is important to note that temperature range is the only camera
setting which has an effect on the QOGI method used in the study. The user is free
to change the polarity of the camera or switch to manual or high sensitivity modes if
desired, but the temperature range must match the synchronisation. It is possible to
synchronise a single quantification module to multiple lenses or temperature ranges.
For this study, only one combination of lens and temperature range was selected.
Figure 6 shows a FLIR GF320 and a black body set up for synchronisation.

Figure 6 Synchronisation of the OGI camera with the quantification

module used for the study (Courtesy of Providence Photonics)

8
 report no. 2/17

4.1.2. Response factors

The OGI camera used for the study is a single spectrum camera. This means that it
is not able to distinguish between different compounds; instead it measures the
response to different compounds in the same mid-wave IR spectral window (3.3µm
to 3.4µm). The QOGI system has been calibrated to propane, but if the operator
knows the specific compound (or composition) a response factor (RF) can be applied
to adjust the result. The response factor takes into account the relative sensitivity of
the compound (or compounds) within the spectral window of the OGI camera, as well
as the molecular weight of the compound, for mass emission calculations. The QOGI
system has built in RF values for many common compounds. In this study, releases
of propane, propylene, methane, and a mixture of all three compounds were
quantified. The propane results were directly measured and the rest used compound
specific response factors.

4.1.3. Temperature and distance parameters

The QOGI system requires the user to provide the ambient temperature and the
distance from the OGI camera to the leak. The ambient temperature was measured
with a National Institute of Standards and Technology (NIST) traceable version of the
ThermoWorks Thermapen. The ambient temperature is generally taken at the location
of the OGI camera but should be representative of the conditions at the site of the
leak. Distance measurements were obtained with a tape measure.

4.1.4. Use of enhanced backgrounds

In cases where ∆T is not sufficient to apply the QL100 method, the background can
be enhanced to generate the needed temperature differential. Enhancing the
background can be typically accomplished by applying a heated or cooled surface
behind the leak. During this study cold towels were used to lower the background
temperature when ∆T was insufficient for quantification. In practice, enhanced
backgrounds will probably not be necessary depending on the environmental
conditions.

4.2. EPA CORRELATION APPROACH (METHOD 21)

The EPA correlation approach, commonly referred to as Method 21, was applied to
derive mass emissions for the TVA screening values (SV). The methodology for
Method 21 is described in US EPA Report 453/R95-017 [8]. The Method 21
correlations were originally derived by vacuum bagging leaks of several equipment
pieces (e.g. valves) under different service conditions such as light or heavy liquid.
The resulting correlations allow mass leak rate to be determined from a screening
value for the equipment types and services in the database. A large degree of
variation was found during the derivation of the correlations for Method 21 including
the fact that several screening values of the same magnitude could have represented
mass emission rates with several orders of magnitude difference. The accuracy of the
correlation for predicting the emission from a single source is therefore poor. In LDAR
surveys this can result in so called ‘false negatives’ (large leak with low screening
value) and ‘false positives’ (very small leak with high screening value). However, as
the correlations are applied to all leaks marked for repair the individual uncertainty is
less important and averaging means that the total mass emissions determined from
an LDAR campaign should be in reasonable agreement with the real emissions.
Therefore, the Method 21 correlations are only statistically meaningful if applied to a
very large number of leaks [2].

9
 report no. 2/17

Examples of the equations applied, in this case for a valve, using the TVA screening
values (SV) and the Method 21 correlation and pegged factors are:

For SV between 1 - 100,000 ppmv: Leak rate (kg/h) = 2.29 × 10-6 × Screening
Value0.746

For SV above 100,000 ppmv: Leak rate (kg/h) = pegged value emission factor (kg/h)

When the leak concentration was high and caused the FID instrument to flame-out, a
dilution probe with a dilution factor (DF) = 10 was used to obtain a screening value.
The screening values measured with a dilution probe were used in the same
equations above to get a “diluted” leak rate, which was then multiplied by the dilution
factor to get the real leak rate. If the TVA screening gave a ‘flame-out’ reading even
with the dilution probe being used, the pegged value emission factor was used. When
the generated gas was not methane, the TVA 1000B response factors for the gas
were used to correct the methane equivalent screening value recorded by the
instrument.

10
 report no. 2/17

5. TEST RESULTS

5.1. PRE-TEST SUMMARY

The purpose of the pre-test was to learn more about the application and limitations of
the QOGI system to permit the finalisation of the test protocol for the full testing. The
environmental conditions during the pre-test were challenging with cold temperatures
and rain. In practice, LDAR surveys are not generally performed during periods of
high wind speed or rainy weather, despite the fact that no normative guidelines on
recommended environmental conditions have been developed to-date.

37 tests were undertaken during the pre-test and all of the generated leaks were
detected by the OGI camera, regardless of background, leak rate, gas composition or
weather conditions.

When there was sufficient IR radiance between the background and a leaked plume,
the QOGI system provided measured leak rates. However, for most tests the
background was a brick wall and there was insufficient T to apply the QL100 module.
During the second test day, no data were recorded that met the T threshold of >
5C. No enhanced backgrounds were attempted during the pre-test, but it is
reasonable to expect that results could have been obtained for most of the leak
scenarios with the use of enhanced backgrounds.

The results for propylene were obtained through the use of a response factor based
on its IR spectrum and molecular weight, as described in Section 4.1.2, and provided
accuracy similar to the propane results. The manufacturers advise that the Response
Factor approach can be used for many other compounds, with similar measurement
accuracy expected.

The pre-test could not evaluate the effects of distance because all of the quantifiable
scenarios coincidentally occurred at 3 meters. Testing was undertaken in the full tests
to evaluate changes in distance (see Section 5.2).

The sniffing method was applied to 14 leaks and quantified using Method 21
correlations and pegged factors. For the other 23 scenarios, the sniffing method was
not applied because flame-out was expected, given the high release rate.

The pre-test parameters and results for QOGI and Method 21 are summarized below
in Table 2 and Table 3.

11
 report no.2/17

Table 2. Pre-test results for May 4, 2015

Test / Ambient conditions Leak generation Method 21 QOGI system
No. Time Temp Humidity Wind Wind Leak Mass Gas used TVA- Calculated Diff. Backgroun Distance ∆T Calculated Diff.
(°C) (%) speed direction component release 1000B emission (2) d from °C emission (2)
(m/s) type/ID rate screening rate with % scene camera rate from %
(g/h) value Method 21 to leak QL100
(ppm CH4) using RF source (g/h)
(Note 1) (g/h) (m)
° Direction
1 10:45 16.4 - 1.15 189 SSW Open end: 1 10.3 Propane F.O. 79 667% Brick wall 3m <5 No result
2 10:57 16.4 - 1.11 189 SSW Open end: 1 50.6 Propane (Note 3) 79 56% Brick wall 3m <5 No result
3 11:14 16.7 - 1.09 189 SSW Open end: 1 16.9 Propane (Note 3) 79 367% Brick wall 3m <5 No result
4 11:19 16.3 - 0.95 196 SSW Open end: 1 198.6 Propane (Note 3) 79 -60% Brick wall 3m <5 No result
5 11:27 16.5 - 1.00 145 SSE Open end: 1 198.6 Propane (Note 3) 79 -60% Brick wall 5m <5 No result
6 11:35 16.6 - 0.84 145 SSE Open end: 1 198.6 Propane (Note 3) 79 -60% Brick wall 10m <5 No result
7 11:40 16.6 - 1.01 164 SSE Open end: 1 49.5 Propane (Note 3) 79 60% Brick wall 10m <5 No result
8 11:46 16.8 - 1.19 156 SSE Open end: 1 49.5 Propane (Note 3) 79 60% Brick wall 5m <5 No result
9 13:40 19.9 52.0 1.73 138 SE Flange: 5 49.5 Propane F.O. 84 70% Concrete 3m 9.4 54.4 10%
10 13:55 19.4 52.0 1.58 209 SSW Flange: 5 16.6 Propane 50 000 25.3 53% Concrete 3m 10.9 14.1 -15%
11 14:00 19.6 52.0 1.53 212 SSW Flange: 5 10.1 Propane 30 000 17.7 75% Concrete 3m 7.7 14.5 44%
12 14:15 20.6 51.4 1.50 198 SSW Flange: 5 10.1 Propane (Note 4) Concrete 2m <5 No result
13 14:25 20.2 51.5 1.60 198 SSW Flange: 5 198.6 Propane F.O. 84 -58% Concrete 3m 7.3 335.9 69%
14 14:45 20.1 51.7 1.43 178 S Valve: 4 49.5 Propane No data Metal door 3m 6.0 55.3 12%
15 14:50 20.3 51.2 1.35 160 SSE Valve: 4 16.6 Propane 35 000 16.3 -2% Metal door 3m <5 No result
16 15:05 20.3 51.2 1.61 159 SSE Valve: 4 49.6 Propylene 110 000 38.0 -23% Concrete 3m <5 No result
17 15:15 20.3 51.5 1.92 267 W Valve: 4 16.7 Propylene 30 000 14.4 -14% Concrete 3m 11.5 20.4 22%
18 15:25 20.7 50.7 1.39 238 WSW Valve: 4 10 Propylene 25 000 12.6 26% Concrete 3m 5.1 16.5 65%
19 15:40 20.5 51.0 1.27 105 ESE Valve: 4 175.4 Propylene F.O. 140 -20% Concrete 3m <5 No result

12
 report no.2/17

Table notes:
1. Unless indicated otherwise, all Method 21 screening values were measured with a dilution probe (dilution factor 1:10)
2. Difference = (determined leak rate – release rate) / release rate (%)
3. For this component and type of gas a flame-out reading was recorded at a lower leak rate. Method 21 was not performed. A flame-out reading was assumed for
all similar scenarios with higher leak rates.
4. In tests 11 and 12 the same gas, leak rate and leaking component were used. Only one Method 21 screening was performed.

13
 report no.2/17

Table 3 Pre-test results for May 5, 2015

Test / Ambient conditions Leak generation Method 21 QOGI system
No Time Temp Humidit Wind Wind Leak Mass Gas TVA- Calculate Diff. Backgroun Distance ∆T Calculate Diff.
. (°C) y speed direction component releas used 1000B d (2) d from °C d (2)
(%) (m/s) type/ID e rate emission % scene camera emission %
(g/h) screening rate with to leak rate from
value Method source QL100
(ppm 21 (m) (g/h)
Directio
° CH4) using RF
n
(Note 1) (g/h)
20 15:15 15.4 - 2.61 234 SW Open end: 2 1000 Propane F.O. 79 -92% Brick wall 3 <5 No result
21 15:35 16.3 - 3.93 231 SW Flange 3 49.5 Propane No data Brick wall 3 <5 No result
22 15:39 16.3 - 2.93 223 SW Flange 3 198.6 Propane No data Brick wall 3 <5 No result
23 15:44 16.9 - 2.65 221 SW Flange 3 1.7 Propane 7300 (3) 1.4 -17% Brick wall 2 <5 No result
24 15:57 17.1 48.6 4.08 225 SW Valve 4 198.6 Propane No data Brick wall 3 <5 No result
25 16:07 17.3 49.2 2.71 236 WSW Valve 4 1.7 Propane 22 000 (3) 2.1 24% Brick wall 3 <5 No result
26 16:15 17.3 48.9 2.47 227 SW Open end: 1 1.7 Propane 55 000 (3) 2.6 52% Brick wall 3 <5 No result
27 16:25 17.3 50.2 2.26 231 SW Open end: 1 49.5 Propane No data Metal door 3 <5 No result
28 16:32 17.2 50.1 2.67 231 SW Flange 3 49.5 Propane No data Metal door 3 <5 No result
29 16:39 17.2 50.0 2.30 239 WSW Valve 4 49.5 Propane No data Metal door 3 <5 No result
30 16:45 17.4 49.9 2.67 240 WSW Valve 4 49.5 Propane No data Metal door 2 <5 No result
31 16:51 17.6 48.8 2.83 237 WSW Valve 4 49.5 Propane No data Metal door 5 <5 No result
32 16:55 17.8 48.2 2.59 232 SW Valve 4 49.5 Propane No data Metal door 9 <5 No result
33 17:10 17.6 44.7 3.35 242 WSW Valve 4 167.2 Methane No data Metal door 3 <5 No result
34 17:24 18.1 44.2 1.73 212 SW Valve 4 49.8 Methane No data Metal door 3 <5 No result
35 17:30 18.1 43.3 2.20 222 SW Valve 4 148.9 Mixture No data Metal door 3 <5 No result
36 17:37 17.7 41.9 2.50 224 SW Valve 4 49.8 Mixture No data Metal door 3 <5 No result
37 17:42 17.9 41.5 2.32 234 SW Valve 4 30.1 Mixture No data Metal door 3 <5 No result
14
 report no.2/17

Table Notes
1: Unless indicated otherwise, all Method 21 screening values were measured with the dilution probe (dilution factor 1:10)
2: Difference = (calculated emission rate – release rate) / release rate (%)
3: Measurements were performed without dilution probe

16
 report no. 2/17

5.2. FULL TEST RESULTS

24 tests were undertaken and, as with the pre-testing, leaks were detected by the
OGI camera for all of these tests regardless of the background, leak rate, gas
composition or weather conditions.

For all 24 test conditions, the background (either naturally occurring or enhanced as
required) provided sufficient IR radiance, making quantification possible. On the
second day, the naturally occurring background - a brick wall with north orientation -
did not provide sufficient T to allow leak quantification with the QOGI system.
Hence, a towel which has previously been placed in a bucket of ice water was used
as an ‘enhanced background’. It was positioned in front of the brick wall and provided
a sufficient temperature difference (negative T in this case) between the
background and the plume to make leak quantification with the QL100 possible for
each of these leaks. The accuracy of the QOGI results using the enhanced
background scenarios was comparable to those using natural backgrounds.

The results for propylene, methane and the mixture (a blend of propylene, propane
and methane in equal volumes) were obtained through the use of a response factor
based on IR spectra of the gases using QOGI propane calibration curves. For the
other gases and the blend, the accuracy of leak quantification by QOGI was similar
to the propane results. The Response Factor approach can be used for many other
compounds, with similar measurement accuracy expected.

Changes in distances were explored during the full testing for different gases and
leak rates. The QOGI system was set in distances of 2 - 8 meters from the leak
source. The results showed that the accuracy of QOGI leak quantification was not
significantly affected by changes in distance.

The sniffing method was applied to 6 leaks and quantification was provided using
Method 21 correlations and pegged factors. For the other 18 scenarios, the sniffing
method was not applied because flame-out was expected, given the high release
rate. The full testing parameters and results for the QOGI system and Method 21 are
summarized in Table 4 and Table 5.

15
 report no. 2/17

Table 4 Test Results for June 15, 2015

Test / Ambient conditions Leak generation Method 21 QOGI system

No. Time Tem Humidity Wind Wind Leak Mass Gas used TVA-1000B Calculated Diff. Background Distance ∆T Calculated Diff.
p (%) spee direction component release screening emission (2) scene from °C emission (2)
(°C) d type/ID rate value rate with % camera rate from %
(m/s) (g/h) (ppm CH4) Method 21 to leak QL100 (g/h)
(Note 1) using RF source
(g/h) (m)
° Direction
1 14:03 20.5 36.0 0.80 81 E Lower Flange 5 16.5 Propane 30 000 17.7 7% Concrete pad 2.74 16.1 21.3 29%
2 14:15 20.8 35.7 0.68 91 E Lower Flange 5 49.7 Propane 40 000 21.7 -56% Concrete pad 2.74 15.1 43.5 -12%
3 14:31 20.9 36.0 0.78 100 E Lower Flange 5 200.5 Propane No data Concrete pad 2.74 19.7 224.9 12%
4 14:45 20.6 36.2 1.11 76 ENE Lower Flange 5 998.70 Propane No data Concrete pad 2.74 19.3 1118.6 12%
5 15:01 21.0 34.3 0.70 213 SSW Lower Flange 5 49.8 Methane F.O. 84.0 69% Concrete pad 2.74 13.1 45.3 -9%
6 15:06 21.2 33.3 0.94 93 ENE Lower Flange 5 169.7 Methane (Note 3) 84.0 69% Concrete pad 2.74 19.4 168.6 -1%
7 15:14 21.0 33.9 0.56 149 SSE Lower Flange 5 49.7 Mix No data Concrete pad 2.74 20.3 43.7 -12%
8 15:20 21.4 33.3 1.10 75 ENE Lower Flange 5 149.4 Mix No data Concrete pad 2.74 21.5 148.8 0%
9 15:26 20.8 33.7 1.15 76 ENE Lower Flange 5 16.7 Propylene 13 000 9.7 -42% Concrete pad 2.74 15.3 14.5 -13%
10 15:31 20.4 35.1 0.63 79 ENE Lower Flange 5 49.9 Propylene 44 000 23.0 -54% Concrete pad 2.74 20.2 42.5 -15%
11 15:35 20.3 35.4 0.30 174 S Lower Flange 5 204.7 Propylene No data Concrete pad 2.74 20.6 237.7 16%
12 16:08 20.4 34.8 1.03 73 ENE Open end 1 49.7 Propane No data Sky 2.74 -11.0 58.5 18%
13 16:42 20.8 34.7 0.57 159 SSE Open end 1 49.7 Propane No data Brick Wall 2.74 9.3 54.5 10%
14 16:51 20.2 35.7 0.51 113 ESE Open end 1 49.7 Propane No data Brick Wall 2 2.74 30.7 46.4 -7%

16
 report no. 2/17

Test / Ambient conditions Leak generation Method 21 QOGI system

No. Time Tem Humidity Wind Wind Leak Mass Gas used TVA-1000B Calculated Diff. Background Distance ∆T Calculated Diff.
p (%) speed direction component release screening emission (2) scene from °C emission (2)
(°C) (m/s) type/ID rate value rate with % camera rate from %
(g/h) (ppm CH4) Method 21 to leak QL100 (g/h)
(Note 1) using RF source
(g/h) (m)
° Direction
15 17:00 20.1 34.4 0.56 189 S Open end 1 200.5 Propane No data Brick Wall 2 2.74 33.0 238.5 19%
16 17:12 20.0 35.5 0.71 158 SSE Valve 4 49.7 Propane No data Brick Wall 2 2.74 34.0 53.3 7%
17 17:17 19.7 36.1 0.67 128 SE Valve 4 200.5 Propane No data Brick Wall 2 2.74 33.9 207.1 3%
18 17:24 19.6 35.4 0.58 111 ESE Valve 4 200.5 Propane No data Brick Wall 2 4.57 23.7 237.3 18%

Table Notes:
1: Unless indicated otherwise, all Method 21 screening values were measured with the dilution probe (dilution factor 1:10
2: Difference = (calculated emission rate – release rate) / release rate (%)
3: For this component, a flame-out reading was recorded at a lower leak rate. Method 21 was not performed for this scenario. A flame-out reading was assumed.

17
 report no. 2/17

Table 5 Test Results for June 16, 2015

Test / Ambient conditions Leak generation Method 21 Results QOGI system

No. Time Temp Humidity Wind Wind Leak Mass Gas used TVA- Calculated Diff. Backgroun Distance ∆T Calculated Diff.
(°C) (%) speed direction component release 1000B emission (2) d from °C emission (2)
(m/s) type/ID rate rate with % scene camera rate from %
(g/h) screening Method 21 to leak QL100 (g/h)
value using RF source average
(ppm (g/h) (m)
° Direction CH4)
(Note 1)
20 10:17 15.3 44.5 0.53 109 ESE Valve 4 49.7 Propane (Note 3) 19.7 -60% Cool Towel 3.05 -9.8 41.2 -17%
21 10:33 14.9 45.9 0.38 171 S Valve 4 49.7 Propane 45,000 19.7 -60% Cool Towel 1.83 -7.8 43.8 -12%
22 10:43 14.9 45.9 0.38 171 S Valve 4 49.7 Propane (Note 3) 19.7 -60% Cool Towel 4.57 -6.8 38.5 -23%
23 10:52 15.5 47.4 0.37 145 SE Valve 4 49.7 Propane (Note 3) 19.7 -60% Cool Towel 8.23 -8.0 45.0 -9%
Middle
24 11:19 16.2 45.2 0.51 247 WSW 49.7 Propylene No data Cool Towel 3.05 -8.3 46.6 -6%
Flange

Table Notes:
1: Unless indicated otherwise, all Method 21 screening values were measured with the dilution probe (dilution factor 1:10).
2: Difference = (calculated emission rate – release rate) / release rate (%)
3: These tests were all the same as test number 21. The same gas, leak rate and leaking component were used. Only one Method 21 screening was performed.

18
 report no. 2/17

5.3. CALCULATED EMISSION RATES

5.3.1. Method 21 vs. Generated leak rate

For 31 out of 61 leak scenarios, a Method 21 screening value was recorded (or
assumed similar to an earlier recorded value) - see Tables 2 to 5. Due to the high
leak rates generated during the tests, many scenarios gave flame-out, even after
using the dilution probe. If a flame-out reading occurred at a given flow rate for a
given component and gas, no Method 21 measurement was performed for the higher
flow rates. It was assumed that a flame-out reading would have resulted. Pegged
values for the flame-out scenarios are included in the data analysis results in Table
6, as these values are used for leak mass estimation when applying Method 21.

For scenarios with identical leak rates, component and gas, only one Method 21
screening was performed. These scenarios are marked in Tables 2 to 5 (See table
footnotes).

Table 6 Method 21 - Comparison between calculated emission data and

known release rates
Method 21 - Difference between calculated emissions and known release
rates 1
Minimum -92%
Average 31%
Standard deviation 155%
Median -4%
Maximum 667%
Table note 1: Difference = (calculated emission rate – release rate) /
release rate (%)

The Method 21 results obtained in this study were better than observed in previous
Concawe field operations [2]. One explanation is that a TVA specific response factor
was used in the calculation of emission rates. In addition, the dilution probe was used
to the maximum extent so that higher leak rates could be better quantified. If the
dilution probe had not been used, a pegged value would have been assigned to
more leak rates, resulting in a more conservative estimation.

5.3.2. QOGI vs. Generated leak rate

For the leak scenarios where quantification was possible (31 out of 61), a
comparison of the differences between the calculated emissions and the known
release rates for both the pre-test and full test is provided in Table 7.

19
 report no. 2/17

Table 7 QOGI system - Comparison between calculated emission data

and known release rates

QOGI - Difference between calculated emissions and

known release rates 1
Minimum -23%
Average 6%
Standard deviation 22%
Median 2%
Maximum 69%
Table note 1: Difference = (calculated emission rate – release rate) /
release rate (%)

5.3.3. QOGI vs. Method 21

To permit a direct comparison of the accuracy of the emission estimates derived

using Method 21 and the quantified leak rates provided by the QOGI system, Tables
6 and 7 have been combined in Table 8. This shows that despite the good
quantitative results obtained by Method 21 in this study compared to other Concawe
studies [2], the QOGI system out-performed Method 21.

Table 8 Comparison of QOGI vs. Method 21

QOGI vs. Method 21 – Comparison of differences between calculated
emissions and known release rates
Difference 1 QOGI Method 21
Minimum -23% -92%
Average 6% 31%
Standard deviation 22% 155%
Median 2% -4%
Maximum 69% 667%
Table note 1: Difference = (calculated emission rate – release rate) /
release rate (%)

However, the number of quantifiable scenarios used to derive the results in Table 8
are not the same for QOGI and Method 21. In order to allow for a better comparison,
the results of both methods for the most commonly generated leak rate (i.e. 50 g/h)
are illustrated in Figure 7. This analysis also shows that QOGI out-performed
Method 21.

Seven release rates were used during the pre-test and full test campaigns; 1.7, 10,
16.5, 50, 175, 200 and 1000 g/h. Figure 8 provides the average values for the M21
and QOGI calculated emission rates for the tests at each of the flow rates. This
shows that the QOGI system provided good agreement with the set release rates
over the entire range of mass flows whereas the M21 methodology significantly
under-estimated the emissions at the higher flow rates. This is because pegged

20
 report no. 2/17

values had to be used to estimate the mass flows at the higher hydrocarbon
concentrations due to flame-out of the TOC monitor used for sniffing (see Section
4.2).

Figure 7 Comparison box whisker plot for Method 21 and QOGI at a

generated leak rate of approximately 50 g/h

90

80
Max = 84 g/h
Determined emission rate (g/h)

70

60
Max = 58.5 g/h
50

40 Min = 38.5 g/h

30

20
Min = 19.7 g/h
10

0
Method 21 QOGI

Figure 8 Comparison of emission rates determined by M21 and QOGI for

each of the set release rates (1.7, 10, 16.5, 50, 175, 200 and
1000 g/h)

1200

1000
Emission rate (g/h)

800 Release rate

600
M21
400 estimation
200
QOGI
0 quantification

21
 report no. 2/17

6. CONCLUSIONS

The aim of this study was to evaluate the emerging field of Quantitative Optical Gas
Imaging and assess its applications in the oil and gas (O&G) industry and in
particular the refining sector. Leak Detection and Repair (LDAR) is one application
that seems promising for QOGI. As the US EPA Method 21 (M21) is the most
commonly used methodology for LDAR monitoring, it was included in the study for a
side-by-side comparison.

The test data confirm that the M21 correlations cannot be used to accurately
estimate individual leak rates because of the statistical method used in their
development. The QOGI system, on the other hand, has been shown to be able to
provide accurate quantification for individual leaks over the range 1.7 to 1000 g/h.
Even where estimation is made for the total emissions from a number of leak sources
(for example as would occur in an LDAR survey), the QOGI system provided a more
accurate total flux value than M21. This suggests that QOGI could indeed be used
for numerous applications in the O&G industry, including demonstration of LDAR
compliance.

One important finding of the study was the application of response factors (RFs) for
QOGI. The study demonstrated that RFs can be used to calibrate a QOGI module
to one gas and then use that module to measure another gas (similar to the use of
the TOC monitor in Method 21). This makes the application of QOGI very practical
and eliminates the need to calibrate the module to each compound. It also allows
the quantification of fugitive emissions from process streams composed of multiple
compounds.

Another important finding was the importance of T as a data quality indicator for
QOGI methods. Though there was only one QOGI technology evaluated in this
study, the T requirement is fundamental to the infrared imaging technology and
would apply to any optical imaging based quantitative method. If QOGI is to be used
as a standard and prescribed method in the O&G industry, T must be measured
and validated as part of the method. The QOGI system used in this study, for
example, provides the operator with a screening tool which allows for immediate
assessment of T.

The study used a T threshold of 5C to define the applicability of the QOGI system.
Further study completed by the technology supplier indicates that this limit could be
reduced to as low as 2 to 3C, even though accuracy and precision may be lower.
As previously mentioned, this needs to be confirmed through additional testing.
Future test studies would be best performed in field conditions rather than with
controlled leak scenarios to ensure that the method is exposed to a variety of
environments and to evaluate the practical application of enhanced backgrounds in
the field. Another area for further study is the effects of wind speed and temperature
as data quality indicators for QOGI methods.

The QOGI technology evaluated in this study was still in the development stage at
the time of these tests in June of 2015. It is recommended that testing of the latest
version of the technology be untaken in field conditions.

22
 report no. 2/17

7. REFERENCES

1. CEN (2008) Fugitive and diffuse emissions of common concern to industry sectors
– measurement of fugitive emission of vapours generating from equipment and
piping leaks. EN 15446. Brussels: Comité Européen de Normalisation

2. Concawe (2015) Techniques for detecting and quantifying fugitive emissions –

results of comparative field studies. Report No. 6/15. Brussels: Concawe

3. EU (2014) Best available techniques (BAT) reference document for common waste
water and waste gas treatment/management systems in the chemical sector. Report
EUR 28112 EN. Seville: European Commission Joint Research Centre

4. EU (2015) Best available techniques (BAT) reference document for the refining of
mineral oil and gas. Report EUR 27140 EN. Seville: European Commission Joint
Research Centre

5. FLIR webpage:
[Link]

6. Lev-On, M. et al (2007) Derivation of new emission factors for quantification of mass

emissions when using optical gas imaging for detecting leaks. J Air Waste Manag
Assoc 57, 9, 1061-1070

7. Providence photonics webpage:

[Link]

8. US EPA (1995) Protocol for equipment leak emission estimates. Report No. EPA-
453/R-95-017. Research Triangle Park NC: US Environmental Protection Agency

9. US EPA (1999) Code of federal regulations standard test method 21: determination
of volatile organic compound leaks. 40CFR Part 60, appendix A. Research Triangle
Park, NC: US Environmental Protection Agency

10. US EPA (2008) Alternative work practice to detect leaks from equipment. 40CFR
parts 60, 63 and 65. Federal register 73, 246, 78199-78219. Research Triangle Park,
NC: US Environmental Protection Agency

11. Yousheng, Z. and Morris, J. (2015). Calibration and quantification method for gas
imaging camera. US 9225915 B2

12. Yousheng, Z, et al (2016). Rediscovering infrared cameras. 4C HSE Conference,

February 22-25, 2016, Austin,Texas

23
 report no. 2/17

ANNEX A – MASS FLOW CONTROLLER (MSC) CALIBRATION DATA

Table 9 MFC calibration data – tests on May 4th and 5th

Gas MFC flow Setpoint Temp Barometric Volume Mass

range % °C pressure flow flow
g/h mbar g/h
ml/min Nml/min
Propane 0-5 10.5 22.5 1021.1 9.12 8.49 1.0
17.0 22.4 1021.1 15.2 14.08 1.7
0 - 300 2.8 22.3 1021.1 91.55 85.30 10.1
5.0 22.2 1021.2 150.80 140.56 16.6
16.1 22.2 1021.2 450.60 420.00 49.5
33.3 22.0 1021.2 903.40 842.62 99.3
67.7 21.8 1021.2 1805.00 1684.71 198.6
0 - 1200 51.0 13.2 1003.9 9008.00 8513.46 1003.4
Propylene 0-5 3.3 22.0 1021.3 9.47 8.83 1.0
3.6 21.8 1021.4 15.36 14.34 1.6
5.0 22.3 1003.9 48.68 44.59 5.0
0 - 200 7.9 21.5 1021.4 94.70 88.50 10.0
11.4 21.3 1021.4 158.40 148.12 16.7
29.4 21.9 1021.3 472.50 440.90 49.6
57.3 21.8 1021.3 942.90 880.15 99.0
100.0 21.7 1021.3 1670.00 1559.38 175.4
Methane 0 - 200 1.4 21.2 1021.3 25.35 23.71 1.0
1.8 21.0 1021.3 41.50 38.84 1.7
6.9 20.8 1021.3 248.10 232.38 10.0
11.0 21.1 1021.3 412.70 386.15 16.5
31.0 21.2 1021.3 1243.00 1162.64 49.8
60.0 21.1 1021.3 2463.00 2304.55 98.8
100.0 20.8 1021.3 4165.00 3901.03 167.2

24
 report no. 2/17

Table 10 MFC calibration data – tests on June 15th and 16th

Gas MFC Setpoint Temp Barometric Volume Mass

flow % °C pressure flow flow
range mbar g/h
g/h ml/min Nml/min
Propane 0 - 10 10.5 23.3 1020.1 9.17 8.51 1.0
17.0 23.3 1020.1 15.08 13.99 1.6
0 - 300 3.3 24.4 1020.7 91.48 84.60 10.0
5.5 24.3 1020.7 151.80 140.42 16.5
16.7 24.2 1020.7 456.00 421.97 49.7
34.0 24.2 1020.7 915.50 847.17 99.8
69.0 24.1 1020.7 1838.00 1701.40 200.5
0 - 1200 52.5 23.6 1020.7 9139.00 8474.04 998.7
Propylene 0-5 24.2 22.0 1025.0 9.52 8.91 1.0
39.3 22.0 1025.0 15.88 14.87 1.7
0 - 200 2.2 21.4 1025.2 48.21 45.23 5.1
4.4 22.1 1025.0 95.50 89.38 10.1
7.5 22.0 1025.0 158.30 148.20 16.7
23.5 22.0 1025.0 474.10 443.85 49.9
48.0 22.0 1025.0 949.40 888.82 100.0
100.0 21.9 1025.0 1943.00 1819.64 204.7
Methane 0 - 200 1.5 24.1 1020.6 23.09 21.37 0.9
1.9 24.0 1020.6 40.65 37.64 1.6
7.0 24.0 1020.6 251.60 232.96 10.0
11.0 24.0 1020.5 416.10 385.23 16.5
31.0 24.1 1020.5 1256.00 1162.43 49.8
60.0 24.0 1020.5 2501.00 2315.45 99.2
100.0 24.0 1020.5 4278.00 3960.62 169.7

25
 report no. 2/17

Concawe
Boulevard du Souverain 165
B-1160 Brussels
Belgium

Tel: +32-2-566 91 60
Fax: +32-2-566 91 81
e-mail: info@[Link]
website: [Link]