Testing of the Airdar System for Oil Leak Detection

Early detection of liquid hydrocarbon releases is crucial for the safety and efficiency of oil and gas operations, especially for pipelines and operations in remote locations. Timely mitigation of these releases minimizes their environmental and community impact; to achieve this, a sensitive and intelligent monitoring system is required. It must provide accurate monitoring of all important emission sources, detect leaks in real time, and distinguish real emission anomalies while avoiding nuisance false alarms.

By Michelle Liu, M.Eng., Dennis Prince, M.Sc., P.Eng., and Chris Apps, M.Sc., P.Eng.

In theory, the volatilization of light products from liquid hydrocarbon leaks will produce a vapor plume detectable by sensors. Using these concentration measurements and cross-referencing meteorological data, an analysis of air concentration patterns can identify spills and leaks in real time, sending the appropriate alarms to operators about identified issues. The approach is particularly useful for monitoring sites located in remote areas, or pipelines covering long distances, ensuring prompt and effective responses to leaks and spills.

This application was tested in partnership with C-FER Technologies (1999) to verify its ability to identify and locate controlled releases of liquid hydrocarbon products. Five tests were performed outdoors at a C-FER facility in Edmonton, Canada using white gas (Coleman camp fuel) or Access Western Blend (AWB), a diluted bitumen. Concentration measurements of volatile organic compounds (VOCs) were obtained at different locations around the test site. The timing and location of the simulated leaks were provided in advance for open, unblinded tests. The date, time and location of simulated spills were not made known in advance for blinded tests. A summary of the performed tests is shown in Table 1.

The hydrocarbon leak detection system consists of a central sensing unit connected to seven remote sample inlets placed around the perimeter of the site. The location of the central sensing unit and remote sample inlets are shown in Figure 1. The central sensing unit contains multiple VOC photoionization detectors (PIDs), each with a detection limit of 1 ppb, providing redundancy and enhanced reliability to the system. Air concentrations are measured continuously at each inlet location in sequence. After deployment, concentration measurements were collected by the monitoring system for two weeks during which controlled releases were not performed, to establish a site emission baseline. This baseline allows the monitoring system to better distinguish offsite sources and sources originating from within the site, minimizing false alarms.

VOC concentration measurements taken by the monitoring system during the baseline and controlled release testing period are shown in Figure 2. Highlighted regions show the start and end of each simulated spill that was performed. In Figure 2, increases in VOC concentration measurements relative to the baseline can be observed during each simulated spill. Additionally, increased VOC concentrations are observed in periods outside of these controlled releases. Potential offsite VOC sources include an oil field waste facility located approximately 1.2 km east, and an industrial area located south of the C-FER site.

An unblinded test using white gas was performed in Test 1 to verify that the system was operational, and to observe the system’s response to a hydrocarbon release. Approximately 3.78 L of white gas was poured into a pan and left for 24 hours. Increased VOC concentrations were observed after the spill was initiated, demonstrating the system was operational and that VOC vapor plumes from the simulated spill are detectable by the monitoring system.

Test 2 was a blinded release of 3.78 L of white gas poured into a pan. As shown in Figure 3, there was a noticeable increase in the sensor output six minutes after the simulated spill was initiated with several inlets intercepting vapor plumes coming from the released gas. The repeated patterns in VOC concentration data at different wind directions indicated the emission event was real and that a spill had occurred onsite. Figure 4 shows the system provided a leak location approximately 18 m from the actual location. The location accuracy was likely affected by buildings and other equipment located in the area potentially impacting the path of the wind within the yard.

Test 3 was an open test performed in two parts using AWB. The first part of the test was conducted to determine if the system would respond to volatiles coming off an AWB spill. An open 20 L drum of AWB was placed near one of the active sampling inlets and a strong response was recorded by the system, as seen in Figure 5. After confirming a signal could be detected, the full test was conducted using the open 20 L drum and three open 2 L jars of AWB. Winds were predominantly in the SSW direction at approximately 200° during the testing period. Figure 5 shows a clear increase in VOC concentration could be seen at only Inlet 4 due to the primary wind direction. This test demonstrated that volatiles from AWB can be detected by the system.

Next, a blind test using AWB was performed by placing 5 L of AWB in an open tray. After 24 hours, the system did not trigger any alarms, and an additional 5 L was added to the tray. Figure 6 shows that increases in the VOC levels can be seen in the concentration data through the testing period. However, these increases were not sustained enough to meet the alarming criteria and trigger an alarm. Additionally, at times the sensor inlet and wind direction indicated that this was an offsite source. Wind direction during this test varied between 100° and 200° at 15 to 20 km/h. Given the location of the simulated leak, these wind directions may have caused the plume to pass in between the inlets for most of the testing period. This issue can be mitigated by increasing the number of sample inlets or placing sample inlets further away from the intended monitoring area. This would enhance the chances of detection by increasing the likelihood for a sample inlet to intercept an emission plume at different wind directions.

Test 5 assessed the system’s performance in extremely cold temperatures. This test was performed over five days with ambient temperatures ranging from -28°C to -45°C. Throughout the test, the system continued to function without requiring onsite support from a technician. The simulated spill was successfully detected, with VOC levels indicating a leak four minutes after the test began. An analysis of the data provided a location estimate that was 13 m away from the actual location and 6 m away from the area of uncertainty provided by the system, as illustrated in Figure 7. Again, location error may be attributed to the effects of the building and nearby equipment in the area on wind behavior.

This third-party testing completed by C-FER successfully demonstrated that vapor plumes from liquid hydrocarbon sources are detectable and evident in air concentration patterns. Plumes from air emissions can be measured remotely and observed by this system without needing onsite support. The data collected offers valuable, actionable information that enables facility operators to enhance the management of site emissions. Continuous monitoring allows for quicker leak detection and reduces false alarms, providing greater confidence in identifying true leak events. Information on whether the sources are elevated or located offsite can also be determined. This leak detection and emissions monitoring approach has diverse applications, from odor identification and mitigation to detecting pipeline oil leaks.

REFERENCE

1. Apps, C. 2024. Applied Testing of Airdar’s Air Detection and Ranging System; C-FER Technologies.