Atmospheric monitoring and detection of fugitive emissions for Enhanced Oil Recovery

https://doi.org/10.1016/j.ijggc.2015.11.031Get rights and content

Highlights

  • Proof of concept for a multi-gas, ratio-based, mobile fugitive gas detection technique.

  • We explain why excess ratios are preferred over raw ratios for atmospheric detection.

  • Excess ratios of CH4:H2S are most sensitive, but excess ratios of CO2:CH4 were also useful.

  • Signals from multiple sources were detected and differentiated at kilometer-scales.

  • Our results can be applied across spatial scales, and in complex environments.

Abstract

In Weyburn, Saskatchewan, carbon dioxide (CO2) is injected into the Weyburn oilfield for Enhanced Oil Recovery (EOR). Cenovus Energy Inc. operates more than 1000 active wells, processing plants, and hundreds of kilometres of pipeline infrastructure over a >100 km2 area. While vehicle-based atmospheric detection of gas leakage would be convenient for a distributed operation such as Weyburn, implementing atmospheric detection schemes, particularly those that target CO2, are challenging in that natural ecosystems and other human activities both emit CO2 and will contribute to regular false positives. Here we present field test results of a multi-gas atmospheric detection technique that uses observed trace gas ratios (CO2, CH4, and H2S) to discriminate plumes of gas originating from different sources. This work is part of a larger project focused on multi-scale fugitive emissions detection and plume discrimination. During 2013 and 2014, we undertook vehicle-based mobile surveys of CO2, CH4, H2S, and δ 13CH4, in the Weyburn oilfield, using customized Cavity Ring Down Spectroscopy (CRDS) instruments that also alternated as stationary receptors. Mobile surveys provided georeferenced observations of atmospheric gas concentrations every 20–30 m, along a route driven at roughly 70 km h−1. Data were uploaded to remote servers and processed using visualization tools that allowed us to constrain the location and timing of potential emission events. Results from one day of mobile surveying, September 24, 2013, are presented here to illustrate how industrial activities, combustion engine and flare stack source emissions can be discriminated on the basis of excess mixing gas ratios, at distances from a few hundreds metres, to kilometres, in the Weyburn oilfield.

Introduction

In the petroleum industry Enhanced Oil Recovery (EOR) through the injection of carbon dioxide (CO2) or water (H2O) has been used for decades (Emberley et al., 2005) as a method for displacing and recovering additional fractions of subsurface oil (Blunt et al., 1993). As a result of EOR, the production life of many oil fields has been extended, while at the same time shifting greenhouse gases from the atmosphere to geologic reservoirs for permanent storage. There were more than 130 active CO2 EOR operations globally (GCCSI, 2012), and because it is a mechanism of production first, and CO2 capture and storage second, EOR research has naturally tended to focus on increasing oil production rather than understanding the fate of CO2 in reservoirs and unintended atmospheric releases (Thomas, 2007). While this is still generally the case, atmospheric releases are increasingly being recognized as a diagnostic tool for production-related issues, including environmental performance. Even minor releases contribute valuable information about production factors, including the condition of infrastructure, and the care with which maintenance activities are done. In many cases, understanding small releases may help avert larger releases down the road. Detecting small fugitive EOR emissions is, however, challenging because several of the same gases involved in EOR operations are also produced naturally from a variety of sources distributed across the landscape. In this paper ‘fugitive’ emissions are defined as escaped reservoir gases with a composition (multi-gas signature) specific to the geochemistry of activities in the Weyburn oilfield, including CO2 injection (Emberley et al., 2004). Further, we note that geochemistry can vary from site to site in ways that may limit, for example, the abundance of hydrogen sulfide (H2S). Thus, a priori knowledge of a site's formation geochemistry will be vital prior to adaptation of techniques such as that presented here.

During the production, processing, and transportation of oil and natural gas, including EOR, unintended seepage can occur at a variety of points such as along well casings, pipeline, and other infrastructure. These components may include CO2, methane (CH4), and H2S and other hydrocarbon species. In the case of CH4, the US EPA estimates that over 60% of global CH4 emissions are tied to human activities, a large portion of which comes from the energy sector (US EPA, 2010). This has been supported by a number of recent studies. The teams of Miller et al. (2013) and Pétron et al. (2012) used multi-year atmospheric measurements of hydrocarbon species from oil and gas fields to suggest that the amount of CH4 leaked from the fields under study might be underestimated currently in national inventories by as much as a factor of two, although there is some disagreement on the exact magnitude of emissions (Allen et al., 2013).

As in the case for Carbon Capture and Storage (CCS) sites, CO2 comprises the bulk of injected gas in EOR. Though CO2 seems a target for monitoring, studies have found that bulk CO2 is not a conclusive indicator of leakage by itself and therefore other indicators are required, because CO2 is readily exchanged by most living organisms (Risk et al., 2013). This means that seepage events, especially those at smaller scale, may be masked by natural variations in CO2 between the biosphere, atmosphere, and also by fossil fuel burning activities in the area. Recent studies have also examined other gases such as H2S, which do occur naturally within these oil-bearing reservoirs, and the low natural abundance of such gases in the free atmosphere means that they can serve as valuable tracers in leak detection.

Table 1 lists four potential emission sources identified across the Weyburn oil field. The CO2 injected by Cenovus is sourced from the Dakota Gasification Company (DGC) and transported via a ∼300-km pipeline. At Weyburn, a water-alternating-gas (WAG) injection strategy is used, which increases oil fluidity and recoverability at production wells (Lombardi et al., 2006, Asghari et al., 2007). Following fluids separation and recycling (REC, mostly CO2), surplus fluids and gases are mixed with new DGC CO2 and re-injected back into the formation. This closed-loop, recycle system continually re-injects recovered fluids back into the reservoir at high pressure. Thus, each emission source (DGC, REC, etc.) has a unique gas composition and mixing ratio ‘fingerprint’ potentially allowing atmospheric plumes originating from these sources to be differentiated from one another, and also from natural signals within the field.

Given our knowledge about the multi-gas proportions of different sources and by measuring the concentrations of several gases simultaneously, it is in principle possible to distinguish these sources in the field based on multi-gas signatures (Romanak et al., 2012). This study field-tests a technique that associates known multi-gas ratios from different sources to anomalies recorded by mobile gas analyzers in the field. In our study, we chose to use recycled gas (REC) as an indicator of containment loss in this closed loop EOR system, whereas (Romanak et al., 2012) used a different set of ratios (CO2, CH4, vs. N2:O2). Here we present the results from a field campaign at Weyburn during late September 2013, where we test our ability to detect REC plumes in the free atmosphere, above background variation, and without prior knowledge of their presence.

Section snippets

Study methods

Our surveys focused on measuring atmospheric gas composition inside and outside the unit boundary in a grid-like search pattern, and in post-processing to identify atmospheric plumes of elevated concentrations of CO2, CH4, and H2S that originated from upwind service rig activity. Service rig activities are known to occasionally generate emissions, including both emissions from service truck combustion and flare stacks (BURN), and also potentially reservoir-type gases (such as REC). We did not

Results and discussion

This section presents three different cases in which measured eCO2:eCH4 and eCH4:H2S ratios met the required criteria for an anomaly as described above, and could be associated with service rig events. As expected, in all three cases we observed complex signals, with anomalies of both REC and BURN. All cases were observed on the same afternoon, September 24, during mobile surveying as can be seen in Fig. 1. Fig. 2 shows the survey time series. During that day, we measured peaks in concentration

Conclusions

This study presents a compositional-based atmospheric sensing strategy that can be used to discriminate between industrial, natural, and combustion emissions on the basis of site-specific gas ratios. Results from each of the observed anomalies show positive identification of atmospheric anomalies that were attributed to combustion or recycled gas sources originating from a drilling or service rig activity, and are consistent with our expectations. Additionally our results show that continuously

Acknowledgments

We would like to thank; Rae Lynn Spencer, Marc Dubord, and James Stirling of Cenovus Energy; Jocelyn Egan for the help with data sampling; and the Natural Resources Canada Eco-Energy Initiative for generous support of this work.

References (20)

There are more references available in the full text version of this article.

Cited by (17)

  • Sweet and sour: A quantitative analysis of methane emissions in contrasting Alberta, Canada, heavy oil developments

    2022, Science of the Total Environment
    Citation Excerpt :

    In the Peace River area, the GOR averages 107–175 m3/m3 (Haydu and Smith, 2015) and in the Lloydminster region, a study by New Paradigm Engineering Ltd. (2015) found that the GOR averaged 80 m3/m3 across 509 wells and varied widely from 50 to 147 m3/m3 (New Paradigm Engineering Ltd., 2015). The truck-based emissions measurement approach used in this study is featured in several past studies, including Atherton et al. (2017), Baillie et al. (2019), O'Connell et al. (2019), Williams et al. (2018), and Hurry et al. (2016). Fox et al. (2019) provide a comprehensive overview of truck-based sensing techniques, highlighting their limitations and advantages.

  • Fugitive and vented methane emissions surveying on the Weyburn CO<inf>2</inf>-EOR field in southeastern Saskatchewan, Canada

    2019, International Journal of Greenhouse Gas Control
    Citation Excerpt :

    Finally, it should be noted that there are a number of service-related emissions that do happen onsite at Weyburn, from activites such as drilling, cleanings and purging of various types, etc. These activities can often result in larger emissions than those reported in this study, but they are intermittent and short-lived (Hurry et al., 2016). The emissions measured during this study represent continuous emissions only, and are therefore not representative of total emissions from this development.

  • Atmospheric impacts of a natural gas development within the urban context of Morgantown, West Virginia

    2018, Science of the Total Environment
    Citation Excerpt :

    Using this method we sacrifice some ability to measure large regional contributions from “super-emitters” as these emissions may be masked in the background, but our sensitivity to defined CH4-rich local sources around the survey vehicle is considerably higher. As described in Atherton et al. (2017) and Hurry et al. (2016) we looked for CH4 enrichment relative to CO2 in areas of even slightly elevated (excess) concentrations. In this study we used a value of eCO2:eCH4 <50 (excess CO2 over excess CH4) as being indicative of a methane plume or local anomaly, which is 4 times the ambient CH4 relative to a natural atmospheric ratio (eCO2:eCH4 = 215).

  • Atmospheric monitoring of carbon capture and storage leakage using radiocarbon

    2017, International Journal of Greenhouse Gas Control
    Citation Excerpt :

    Even so, under many circumstances the δ13C method is likely slightly better than using CO2 mole fractions alone, and has a detection limit of about 3–5 ppm (Keeling et al., 2011; Moni and Rasse, 2014). It is also possible to inject “correlate” trace gasses into carbon reservoirs along with CO2 and monitor leakage of these gasses (Hurry et al., 2016; Susanto et al., 2016; Garcia et al., 2012; Pekney et al., 2012; Watson and Sullivan, 2012; Wells et al., 2009). The correlate gas can be intrinsically associated with a particular CO2 source used for CCS, or deliberately added.

View all citing articles on Scopus
View full text