First field test of linear gas sensor net for planar detection of CO2 leakages in the unsaturated zone

https://doi.org/10.1016/j.ijggc.2013.04.014Get rights and content

Highlights

  • Linear gas sensors successfully tested for early detection of a CO2 leakage in soil.

  • Near real time determination of the evolution of the minimum CO2 plume extent.

  • Non-expected high lateral CO2 plume spreading near the surface observed.

Abstract

Gases from subsurface sources can present risks for the biosphere. To extend the lead time in managing such risks, reliable gas detection in the subsurface is required. A monitoring system was developed that is able to: (i) gather information from large areas, (ii) work efficiently within the subsurface; insensitive to changing subsurface environmental conditions while sensitive to changes in target gas concentrations, and (iii) provide a fast response. We report the first field test of linear membrane-based gas sensors: Fourteen 40 m long sensors were installed in two horizontal nets one above the other within a homogenized soil and scanned from a control station. 10 L/min of CO2 was injected into a point-like port 25 cm below the lower sensor net. Two operation modes were successfully tested to detect the CO2 plume: an active mode, which is sensitive to the concentration, and a passive mode, which is sensitive to changes in concentration. By conversion of the monitoring results the minimum contact range of seeping gas and sensor could be determined. Based on this valuable approach, an unexpectedly high level of lateral gas spread within the soil and the formation of a gas pillow below the soil surface could be observed.

Introduction

Carbon dioxide (CO2) sequestration represents a promising strategy for the mitigation of climate change even if some small fractions seep from the repository (Hepple and Benson, 2004). The early detection of gases ascending both from such repositories, but also from natural sources (e.g., volcanic gases, seam fire gases, biogenic gases such as methane and hydrogen sulfate) can extend the lead time for managing related risks. Hence, monitoring is necessary on temporal and spatial scales relevant to the process-affected area. Thus, monitoring methods are required that permit to maximize both lead time and detection probability of a leakage.

For example, risks and accidents in volcanic regions have been described as a result of the degassing of CO2 into the atmosphere (Annunziatellis et al., 2003) or, when triggered by rain events, indoors into cellars (Viveiros et al., 2009). Thereby, as observed in the area of an active tectonic fault system (Počatky-Plesná Fault Zone, Mariánské Láznĕ Fault Zone, western Eger Rift, Czech Republic), rising CO2, which is released from the upper mantle at a depth of about 30 km below surface, could degas over comparatively very small areas irrespective of its long path to the surface. Such degassing areas could contain mofettes (dry or wet) and diffuse degassing structures of up to about 100 m in length, generating variable, nearly pure CO2 fluxes depending on weather and seasonal conditions. The seeping dense CO2 influences flora and fauna depending on the local morphology of craggy landscape and causes the deaths of small animals. The individual degassing areas are arranged point-shaped alongside the fault system for a total length of about 100 km. The degassing mechanisms, geological structures and processes, the gas composition, its origin and near-surface properties of such degassing areas have been investigated thoroughly (e.g., Bräuer et al., 2011, Flechsig et al., 2008, Weinlich et al., 1999). As demonstrated by numerical simulations (Chow et al., 2009), the danger from CO2 is greatest in such craggy regions with topographic depressions where the gas can accumulate even during windy conditions.

In order to detect or observe such (potential) degassing structures in large regions, the surveying of the subsurface by geophysical methods and space-integrating gas measurements in the atmosphere has been suggested and applied. This type of atmospheric monitoring is adaptable in horizontal and uniform terrain with the aid of techniques such as open path infrared sensors in conjunction with eddy covariance (Madsen et al., 2009).

The potential for the detection of a local CO2 leakage amongst others by seismic and integrated atmospheric monitoring was studied at a regional scale within the CO2CRC (Cooperative Research Center for Greenhouse Gas Technologies) Otway demonstration project, in which carbon capture and storage (CCS) was investigated in a depleted gas field (Jenkins et al., 2012). By atmospheric measurements a point source could be detected at a distance of 700 m for favorable terms which emitted about 6 t CO2 per day (equivalent about to 2 kt/y or 2.1 m3/min). Wind, which erratically blew from the direction of the source, resulted in a rise of the concentration of about 2.5 ppm above baseline. The authors concluded a sensitivity of the applied monitoring techniques to surface leakage rates at the level of a few kt/y.

Atmospheric monitoring does not permit forewarning, which would be required for improved risk management. On the other hand corresponding integrated subsurface sensors, which could permit forewarning, are not available. For this reason, public acceptance of CCS, especially in populated regions, depends significantly on the availability of an effective forewarning system. Serious concerns regarding CCS are indeed based on the lack of early detection methods, as is shown by current public discussion.

The movement of a gas phase through the subsurface typically exhibits a complex flow behavior. In groundwater, the movement is dependent on the small-scale properties of the sediment, such as pore size and pore geometry (Geistlinger et al., 2006, Lazik et al., 2007, Selker et al., 2006), or on the aperture and orientation of fractures within the rocks (Annunziatellis et al., 2008). At larger scales, the (hierarchical) heterogeneity of the aquifer system controls the spread of the gas phase and the individual pathways control its movement (Geistlinger et al., 2009, Selker et al., 2006). However, the availability of sufficient data for a detailed description of the material and structures at the relevant scales is limited and reduces the possible resolution of model-based predictions.

Owing to the high gas–water interface tension, the movement of the nonwetting gas phase is expected to occur frequently at individual small hot spots within the groundwater. To detect fluxes from such hot spots, monitoring techniques are required that cover large areas but that remain sensitive to point-like sources. Above the groundwater level, the denser CO2 phase reaches the flexible unsaturated zone gas phase. Modeling results of CO2 flux through a 30 m-thick unsaturated zone demonstrated its potential to: (i) attenuate leaking CO2, and to decrease (ii) near-surface CO2 concentrations and (iii) the CO2 seepage. For small seepage rates, which can be assumed during the beginning of a leakage, the ratio of seepage to storage is very small (Oldenburg and Unger, 2003). The CO2 phase will occupy first an increasing part of the pore space (bulk formation) while replacing other gases and mixing with them at the interface between the growing bulk and the displaced soil gases.

Thus, we suppose: (i) the detection probability of a leakage will increase greatly in the unsaturated zone with respect to the groundwater, (ii) the CO2 concentration within the unsaturated zone will be much higher than the concentration in the atmosphere, and (iii) with respect to the attenuation phenomenon, CO2 monitoring within the unsaturated zone could enhance greatly the lead time of a developing leakage.

Therefore, the overall detection probability and lead time of a leakage could be enhanced by addition of aerial gas monitoring in the unsaturated zone to atmospheric gas monitoring in critical areas. In order to cope with the given heterogeneity of the subsurface, it is necessary to integrate a high number of point measurements (e.g., soil gas samples, flux measurement by open chambers) with limited areal support using geostatistical methods (Geistlinger et al., 2010). To enhance the areal support, passive sampling methods using gas permeable tubular membranes for the separation of the gas phase, can be combined with a pumping system and a suitable analytical device for gas measurement. (Seethapathy et al., 2008) give an excellent overview of passive sampling techniques for applications in the subsurface.

With the same aim, a new measurement method was developed based on the evolution of pressure within closed chambers, resulting from the diffusion of different gases through its gas permeable coatings - the gas selective membranes. The method was demonstrated successfully in (Lazik and Geistlinger, 2005) for the analysis of oxygen (O2) and nitrogen (N2) in a large range of 1–100%vol. This type of membrane-based gas sensor operates far from the thermodynamic equilibrium, which enables a much more rapid response compared with membrane-based gas phase separation and sampling.

Considering a single gas component mixing with a given gaseous matrix (e.g., CO2 in soil air), the theoretical framework in Lazik and Geistlinger (2005) could be simplified. The modified sensor was presented for the observation of O2 and CO2 in dry medium-grained sand in the laboratory for controlled conditions (Lazik et al., 2009). Linearity of the response behavior could be shown for a large concentration range exceeding three orders of magnitude.

The advantages of such a sensor are its simple construction and handling and its comparatively lower cost. The shape of such a membrane-based gas sensor can be adapted to the measurement problem. Furthermore, a method for in situ maintenance was demonstrated, enabling the proof of sensor quality and its effective calibration without demounting the line-sensor from the soil (Lazik and Ebert, 2012). The disadvantage is that qualitative information is required regarding the gases, the concentrations of which could vary independently of each other. A non-foreseen variability, e.g., that of the water vapor pressure, could mask a change in the CO2 concentration.

To analyze CO2 concentrations in larger areas, it is advantageous to use tubular membranes forming linear sensors (line-sensors). Line-sensors integrate over a large area like eddy flux measurements and therefore, also provide a significant average of the locally fluctuating concentrations, which is independent of the phase's composition of the supporting area.

After a brief introduction of the measurement method, the experimental study introduces a conceptually improved prototype that was optimized to observe a limited area by 40 m line-sensors, which could be arranged in a star shape around a single maintenance point. Two possible operational modes (passive, active) for leakage detection were subsequently compared in a, to that aim, homogenized soil body within the first field test performed with membrane-based line-sensors.

Section snippets

Sensor system

The operation of the gas sensor is based on the diffusive fluxes of gases through its gas-selective membrane. The individual gas fluxes correspond according to Dalton's principle with a change of the total mole number within the interior space of the sensor. With this change, the gas volume, or if this space is closed, its pressure will be modified and can be related to the concentration of the investigated gas. Therefore, the sensor has to be calibrated against different (at least two)

Preparation of the field test

The preparation of a small test field began in September 2011 at the agricultural test site in Bad Lauchstädt. The line-sensors were embedded within the top soil, which is a Chernozem. Chernozemic soils are dark brown to black in color. The humus-rich soils show a high base saturation, a stable aggregate structure and a definitive bioturbation. The 60 cm-thick Chernozem at the test site developed from loess, which now forms the basement. The Chernozem is characterized by a total silt content of

Calibration

For calibration, two line-sensors with a total length of 80 m were exposed to different gases within a closed 60 L vessel. This calibration vessel was equipped with a gas entry and a gas outlet. A reference sensor for CO2 (NDIR-sensor: CARBOCAP™ – GMP221, measurement range: 0–10%vol, response time T63 = 20 s, precision < ±(0.002 + 0.02 pCO2)%vol, manufacturer: VAISALA) was placed in this outlet. The calibration vessel was flushed from bottom to top by a permanent air flow of 0.5 L/min provided by a

Design of the sensor system

The design of the entire sensor system (control station with actuator + line-sensor network) was developed to observe a limited area situated around a single maintenance point – the control station. Using 32 line-sensors, each with a length of 40 m, an area of up to about 1000 m2 could be observed. The design of the monitoring system reduces the number of pressure sensors required. Instead of mounting gauges for each line-sensor, only one gauge is necessary for the entire network. However, the

Summary and conclusions

A prototype set-up for the areal monitoring of CO2 was successfully tested in the field. The set-up enables to scan sequentially up to 32 line-sensors with a temporal resolution of about 9 min with the aid of a controlling unit. By operating 32 such line-sensors, an area of up to about 1000 m2 around a (potential) degassing structure could be observed.

A sufficient high accuracy was demonstrated of the 40 m line-sensors during calibration. Furthermore, a nearly identical response behavior of

Acknowledgements

We thank Roland Strasburg (CEO of SMB Strasburg – Metall – Bau GmbH/Calbe) for valuable discussions and for the construction of the high-quality equipment; Benjamin Thies (student, University of Applied Sciences – Merseburg) for his help in the laboratory and during the construction of the field test; Katrin Lazik and Manfred Ebert for their help at the test field; Hans-Jörg Vogel, the editors and reviewers for valuable suggestions during the preparation and revision of the manuscript.

This work

References (33)

  • F. Viveiros et al.

    Meteorological factors controlling soil gases and indoor CO2 concentration: a permanent risk in degassing areas

    Science of The Total Environment

    (2009)
  • F.H. Weinlich et al.

    An active subcontinental mantle volatile system in the western Eger rift, Central Europe: gas flux, isotopic (He, C, and N) and compositional fingerprints

    Geochimica et Cosmochimica Acta

    (1999)
  • M. Altermann et al.

    Chernozem – Soil of the Year 2005

    Journal of Plant Nutrition and Soil Science

    (2005)
  • M. Bartholmai et al.

    Multifunctional sensor for monitoring of CO2 underground storage by comprehensive and spatially resolved measuring of gas concentrations, temperature and structural changes

    Energy Procedia

    (2013)
  • J. Brandrup et al.

    Polymer Handbook

    (1999)
  • T.K.K.C. Deepagoda et al.

    Density-corrected models for gas diffusivity and air permeability in unsaturated soil

    Vadose Zone Journal

    (2011)
  • Cited by (14)

    • Characteristic length measurement of a subsurface gas anomaly-A monitoring approach for heterogeneous flow path distributions

      2016, International Journal of Greenhouse Gas Control
      Citation Excerpt :

      During the same time, the CO2 rose only by H = 0.35–0.4 m from the injection point to a level of about 0.2 m below the soil surface. The ratio of the lateral spread to the vertical rise ROI/H = 2.5 − 2.9 fits the range (ROI/H = 2 −3) modeled by Oldenburg and Unger (2003) for a 30 m-thick unsaturated zone and improves the rough estimation of the minimum ROImin/H ≈ 3 in Lazik and Ebert (2013), which combines the vertical rise of a local CO2 leakage through a 0.3 m-thick unconsolidated Chermozem soil with the lateral CO2 spread monitored at a level of 0.2 m bgs. The general lateral spreading and the trend of higher measurement values and a higher CO2 persistence for increasing depth (which occurs irrespective of the interim change of the driving forces) confirm the influence of individual gas properties (in this case probably the comparatively high molecular mass of CO2) on the gas spreading behavior.

    • Carbon Dioxide in General

      2019, Carbon Dioxide Sensing: Fundamentals, Principles, and Applications
    View all citing articles on Scopus
    View full text