New Optical Gas-Imaging Technology for Quantifying Fugitive-Emission Rates
Optical gas-imaging (OGI) technology has been developed and can be used to detect leaks of volatile organic compounds (VOCs) from process equipment. Using OGI to detect leaks is more effective than using the US Environmental Protection Agency’s Method 21 because OGI is visual, making detection faster, and can survey an area instead of one component at a time. Although OGI can be very effective in detecting leaks, it does not provide a quantitative measure of leak rate (LR), hindering its adoption as a true alternative to Method 21. This paper describes development of quantitative OGI (QOGI) technology.
Method and Preliminary Results
Approaches have been proposed to establish a quantitative relationship between the pixel intensity difference with and without a plume (ΔI) and the product of concentration in ppm and path length in meters (ppm·m) for a gas column represented by a pixel in the infrared (IR) image for a given temperature differential (ΔT) between ambient air and the background. This quantitative relationship has been confirmed with a study showing that there is a monotonically increasing relationship between ΔI and concentration for uniform black background that was temperature controlled. That study’s data also showed that ΔI increases as the temperature of the background increases for a specific gas concentration.
The working principle of the QOGI can be described as briefly follows:
- IR images of a leak are analyzed for intensity on a pixel-by-pixel basis.
- Each pixel represents a column of hydrocarbon vapor between the IR camera and the background.
- Pixel contrast intensity (ΔI) is defined as the intensity difference at the pixel level between the background with and without the absorption because of hydrocarbon molecules.
- ΔI is a function of the temperature difference between the background and the plume (ΔT).
- At a given ΔT, the intensity is proportional to the number of hydrocarbon molecules in the vapor column.
- The LR is reflected in both pixel intensity and the number of pixels that have a ΔI higher than a certain threshold. Inversely, the combination of intensity and that number determines LR.
On the basis of this methodology, a computer program has been developed that captures raw IR data from an IR camera and analyzes it for LR. The IR camera must be radiometrically calibrated to make it capable of measuring temperature at the pixel level. To analyze the IR images, the user must also enter an estimate of ambient temperature and distance from the component being tested into the IR camera. All other variables required for determining LR are preprogrammed. With the captured IR images and the two user-provided input parameters, the program will calculate the mass leak rate in pounds per hour (lbm/hr).
Work to date has measured the component leak rate using QOGI technology on accurately controlled releases, with the focus on propane. Flow rate, or LR, was set using a calibrated mass-flow controller. The IR camera was positioned approximately 10 ft away from the release point. All of the tests performed to date (80 total) were conducted in an outdoor, open-air environment. The types of backgrounds tested included a uniform temperature-controlled metal board, a building wall, and gravel. These tests were conducted in sunlight and in shade, in ambient temperatures from 37 to 95°F, in relative humidity from 50 to 90%, and in various moderate wind conditions. Because the true LRs were known in these tests, the accuracy of this method can be assessed by comparing the true LR and the LR measured by QOGI (Fig. 1). Eight LRs were tested, represented by eight pairs of bars in Fig. 1. The green bars represent the true LRs, with the number indicating the rate. The purple bars represent the average LR measured by the QOGI method. The red error bars represent ±1 standard deviation from the average. Within these 80 tests, the measured LRs were between -17 and 43% from the true values.
A limited number of tests have also been performed for methane and ethylene. Leak rates were determined for these materials using IR response factors (RFs) developed on the basis of the IR spectra of methane and ethylene relative to the spectra of propane. Measured LRs used RFs developed from these known spectra (vs. direct RF measurement) and indicate good agreement between the true and measured rates for the set of tests conducted.
The accuracy of the QOGI method as discussed previously pertains to a limited range of conditions, and more-comprehensive tests are planned to characterize QOGI accuracy under a broad range of environmental conditions. The initial results are encouraging, especially when comparing measurement accuracy with inherent uncertainties in Method 21. The uncertainties associated with the current Method-21-based methodology come from three potential sources:
- Measured screening values (SVs)
- Compound and instrument-specific RFs
- Correlation equations that are applied to the SVs to estimate emissions rate
The SV is a concentration measurement that uses a probe to examine a component to determine the maximum concentration for a small set of components that potentially could leak. Concentration is not proportional to the LR but is presumed to be for estimating LR by use of Method 21. Factors such as the geometry of the leaking component, the pressure inside the equipment, wind speed, and atmospheric turbulence will affect the concentration measurement (or SV reading). For a small leak area (i.e., from a single point), the concentration measured by use of Method 21 will be much higher than that for a more-diffuse leak. This will produce significantly different SVs with Method 21, even if the leaks are controlled to the same rate.
Another source of error for Method 21 is the RF for each compound in the gas leak, which corrects detector differences for each compound in the emitted gas. The portable detector used for Method 21 surveys is calibrated with one compound, but actual material leaking may be a different compound or a mixture of compounds. To determine the true concentration of the leaked compound, an RF is applied to account for differences between calibration gas and the emitted gases. RF is a predetermined ratio between the reading of the calibration gas and the gas in question.
The EPA has compiled RFs for approximately 200 compounds. The RFs can vary by an order of magnitude for different compounds or from one instrument type to another. Per the EPA’s 1995 Protocol, if RF is less than 3, no adjustment is required. This means that the measured SV could have up to a 200% error if the SV is not applied per the protocol. In addition, if the RF does not reflect the actual mixture of compounds emitted, additional error in the estimate of LR is introduced.
Even if the SV is perfectly accurate, the potential for error exists when correlation equations are applied to the SV to estimate mass emission rate. The correlation equations were developed on the basis of field tests in which SVs were determined by Method 21 and actual mass emission rates were determined with another technique. The correlation between these paired data sets was not very strong, as indicated by low R2 values from 0.32 to 0.54. As a result, the ratio of LR predicted by these correlation equations to the measured LR ranges from approximately 0.2 to greater than 4. As such, errors in the LRs estimated with the EPA protocol for Method 21 could be in the range of -80 to 300% using Method 21 when all of the potential sources of error are propagated.
With Method 21, the concentrations are measured and emission rates are estimated. In comparison, QOGI directly measures mass LRs. In the tests reported in this paper, the errors in the QOGI results are substantially smaller than those that would be expected from application of Method 21.
It has been demonstrated, with initial but compelling data, that QOGI is technically feasible. QOGI directly measures emission rates. This is fundamentally different from Method 21, which estimates emission rates using concentration measurements, SVs, RFs, and correlation equations. With a QOGI commercial product, operators have to enter only ambient temperature and the distance from the leak site into the IR camera. The QOGI product can then capture the IR images for approximately 30 seconds and provide the operator with a measurement of the mass emission rate. Consequently, it is expected that QOGI will be able to reduce significantly the time to complete a survey while providing more-accurate measurement of emission rates.
This article, written by Special Publications Editor Adam Wilson, contains highlights of paper IPTC 18471, “New Optical Gas-Imaging Technology for Quantifying Fugitive-Emission Rates,” by Hazem Abdel-Moati, ExxonMobil Research Qatar; Jonathan Morris and Yousheng Zeng, Providence Photonics; Petroula Kangas, ExxonMobil Chemical Europe; and Duane McGregor, ExxonMobil Research and Engineering, prepared for the 2015 International Petroleum Technology Conference, Doha, Qatar, 7–9 December. The paper has not been peer reviewed. Copyright 2015 International Petroleum Technology Conference. Reproduced by permission.