Vehicle-based Fugitive Emission Detection and Attribution within Alberta Energy Developments

Liz O’Connell and David Risk, Flux Lab Department of Earth Science, St. Francis Xavier University

July 3, 2017


Accurate and spatially-extensive methane emission data can help operators and regulators meet new reduction targets. Vehicle-based monitoring uses a truck equipped with laser-based, multi-gas analyzers that measure methane and associated thermogenic gases to a precision of ~1 ppb, as well as algorithms that detect plumes on the basis of methane and associated gases. Plumes can be attributed known upwind infrastructure using back-trajectory algorithms. Travelling at roughly 80km/hr, we can routinely detect emissions of ~10 m3/d per day from hundreds of meters away, while sampling an average of 100 (on-pad) to 400 (roadside sampling) well pads/facilities per day. The aim of this Alberta Upstream Petroleum Research Fund (AUPRF) study was to collect field air chemistry data in three Alberta Energy Developments (Medicine Hat, Lloydminster, Peace River) that would:

  1. Provide development-specific geochemical (CH4 , C2 H6 , δ13CH4 , CO2, H2 S) fingerprints for the vehicle-based gas monitoring system, to increase immediate applicability in Alberta;
  2. Quantify methane concentrations and drivers of variation, across several developments;
  3. Allow us to evaluate vented and fugitive emissions frequency and severity from several thousand pieces of infrastructure.

We collected data near thousands of wells and facilities within Lloydminster, Peace River, and Medicine Hat, Alberta during fall 2016. Over the course of six weeks, we measured CH4 (methane), C2 H6 (ethane), δ13CH4 (isotopic methane), CO2 (carbon dioxide), and H2 S (hydrogen sulfide) concentrations simultaneously at 1 Hz intervals, while following pre-planned routes along public roads. Over 6.7 million geo-located ambient gas and climate measurements were collected during this time. Using these data, we identified geochemical emission signatures of industrialsourced plumes. We also generated statistical summaries of methane concentration variability in dense areas of infrastructure, and in the background. We ran a Generalized Additive Model (GAM) to understand how different variables influence ambient background methane. Finally, using geochemical emission signatures, we enumerated plumes detected on-road and attributed them to nearby infrastructural sources within several hundred meters. If a piece of infrastructure was measured upwind and within a defined radius of an on-road anomaly (as determined by gas ratio signatures), it was tagged as a potential emitter.

Absolute raw concentrations differed between developments. We observed only mild enrichments of several tenths of a ppm CH4 in the Peace River and Medicine Hat regions, relative to the atmospheric background of methane (~1.85 ppm, NOAA 2017), whereas enrichments were markedly higher in the Lloydminster area. In Lloydminster, we occasionally measured average concentrations exceeding 6 ppm for an hour of driving (covering tens of km). We saw several hundred plumes (super-ambient air downwind of infrastructure), the relative gas ratios of which were within expected ranges for these developments. For example, we measured similar C1(CH4 ):C2(C2 H6 ) ratios of 1-2% for Peace River and Medicine Hat, whereas the Lloydminster C1:C2 ratio was just below 1%. We measured appreciably higher ratios of associate gases at Peace River, and relative to CH4 , the CO2 and H2 S values were higher than in other developments.

Outside of plumes, a statistical model showed that methane concentrations in the background air were controlled mostly by the time in which they were recorded. Other factors such as wind speed, topography, geography, and temperature were comparatively less important in predicting background variation.

Applying geochemical and geospatial filters to the data, we could attribute plumes more specifically to known upwind oil and gas infrastructure. These were the result of fugitive and vented emissions. Overall our routes passed ~1200 wells in each development, in triplicate. We found that emission frequencies varied amongst developments, but were the most common in Lloydminster, where 56% of wells were emitting methane-rich gas above the minimum detection range of 10.3 – 25.9 m3/day (dependent on atmospheric conditions each day). Active wells in Medicine Hat and Peace River followed, with 28% and 29% of wells tagged as a potential emission sources, respectively. Although active wells were the predominant source of emissions, other classes of infrastructure, including abandoned and suspended infrastructure, also contributed. Both episodic and persistent emissions were measured in each development, owing to the sporadic and unpredictable nature of oilfield related emissions. Lloydminster emissions were the most episodic.

This study demonstrates the practicality of mobile surveying as a practical, large-scale screening solution to address high-priority air quality concerns in Alberta. We hope the project outcomes will inform the development of smart policies, regulations, and best practices for the sustainable development and monitoring of Alberta’s hydrocarbon resources.

Full Report

Best Management Practices

# 16-ARPC-04