Figure 1: The GasClam is made of stainless steel and has the ability to continuously measure a number of parameters in a borehole simultaneously.

Soil-gas monitoring is an important aspect of contaminated land site investigation and landfill management, as the data is critical to the risk assessment process. The objectives of gas monitoring programs are to determine the true subsurface gas regime and predict how this may change in the future. This is currently achieved by discrete periodic static measurements of gas concentrations from which the gas regime is inferred.

Flaws in the current approach to quantifying and predicting risk arising from soil-gas are explicitly identified in industry and regulatory agency literature,^[1] and are implicit in the continuing evolution of guidance notes. The underlying cause of flaws is that whilst accurate quantification of risk should require accurate measurement of soil-gas concentration and of soil-gas fluxes, neither is directly measured, and both are likely to be temporally variable.

Measurement is indirect because soil-gas concentration is inferred from periodic sampling of gases that accumulate within a borehole; the flux is then inferred from these readings. The unit of flux is volume/time, therefore it can not be directly measured without time series data.

With the ability to collect time series data, an improved measurement of flux can be achieved, and temporal variability can be quantified and explained. This will improve understanding of processes, thereby reducing uncertainty inherent in using indirect measurements, or those lacking in temporal resolution.

Overview of the technology

Figure 2: The monitor can be placed into a standard borehole and flush-mounted for security.

Contaminated land and landfill industry regulators recognize the need for more representative data,[1] but cost has prevented collection of continuous soil-gas measurements. Recently, engineers have been able to apply reliable, miniature, infrared sensors to continuous monitoring instruments. The technology could allow time-series data to become the industry's new standard.

The GasClam, pictured in Figure 1, is a good example of the technology. The sensor, manufactured from stainless steel, is intrinsically safe, with ingress protection rated IP-68 as defined by the international standard, IEC 60529. Fit into a 50-mm borehole, the device can measure methane, CO₂, oxygen and H₂S concentrations, as well as atmospheric pressure, borehole pressure and temperature. Water levels can be measured with an optional pressure transducer. The device fits securely within a borehole (see Figure 2), while still allowing the hole to be vented. Sampling frequency is variable from 2 minutes to once per day. The data can be downloaded through a notebook PC using an RS 232 communication cable. It can be powered for 90 days by two alkaline D-cells based on hourly sampling.

Changing risk assessment

Enlarge this picture Figure 3: Continuous gas concentration data from the Christmas Borehole, a perimeter borehole at a landfill thought to indicate gas migration problems only at Christmas. Continuous data readings demonstrated the fallacy of that conclusion and provided an explanation.

Currently, the most common approach relies on discrete gas concentration measurements, from which representative ground gas concentrations and gas migration potential are inferred. As system data is resolved, uncertainties in these inferences remain. For example, the frequency of variation in gas concentration may be higher than the sampling frequency, in which case measurement will not be representative. This sampling frequency and variability in the gas concentration have produced some dramatic anecdotes between landfill industry professionals, such as when analysis of periodic sampling was thought to show high gas concentration only at Christmas.

That particular story is worth a closer look. Figure 3 shows continuous gas concentration data from what operators called "The Christmas Borehole," a landfill perimeter borehole thought to indicate gas migration problems only at yuletide. A period of continuous data collection overcame the artifact arising from the monthly sampling frequency mismatching with the variability of concentration. The continuous data clearly showed that though the CH4 concentration was variable, it was not only high at Christmas. Collecting spot samples on days with the green dots compared to days with the black dots would result in an extremely different perception of risk.

Importantly, time series data also revealed that the frequency of variation in gas concentration was highly variable. In the above example the rate of change was on a daily/weekly timescale, however data collected at another site indicated gas concentrations changing by up to 40 percent in minutes. From November 19 to 22 the sampling frequency was 10 minutes; after this the frequency was reduced to 1 hour and the variability was not observed.

Concentration duration curves

Enlarge this picture Figure 4: Concentration duration curves for CH4. The red line is constructed from the high frequency continuous data and the black lines compiled from random samples. The data from Set A indicate that the methane concentration is always near the lower explosive limit of methane and would indicate a very high risk. The data from Set B indicate a much lower concentration and therefore lower risk. In reality the real gas regime is somewhere in between the two.

Collection of more highly time-resolved data allows the construction of meaningful concentration duration curves. Analogous to hydrological flow duration curves, these provide a more direct interpretation of risk than typically available from conventional monitoring. The value of continuous measurement is best shown by comparing concentration duration curves from the system with data representing conventional periodic weekly sampling taken from the continuous data set.

Two such sets are shown in Figure 4. The red line is constructed from the high frequency continuous data and the black lines are compiled from random samples taken from the continuous data set to represent conventional periodic weekly sampling (see spot sampling A and B). The data from Set A indicate that the methane concentration was always near 5 percent. This is near the lower explosive limit of methane, and would indicate a very high risk. The data from Set B indicate a much lower concentration and therefore lower risk. In reality, the real gas regime was somewhere between the two: lower than the lower explosive limit for methane, but close enough to be of concern.

Correlations

Enlarge this picture Figure 5: The expected relationship between atmospheric pressure and gas concentration is clear, when the pressure falls, concentration increases and vice-versa.

Higher temporal resolution of not only gas concentration but other environmental variables allows their interrelationships to be more clearly defined. This in turn allows dominant controls on gas concentration to be recognized, and better prediction of gas concentration as other parameters change.

Atmospheric pressure is considered to be a strong driving force for gas migration.^[1] In general, it is assumed that concentrations are higher when pressure is low, and vice versa. Because of this, current guidance (e.g. CIRIA Report 665) recommends collecting at least one spot sample below 1000 mbar in falling pressure.

Continuous monitoring data in Figure 5 shows the expected relationship between pressure and concentration. Current guidance states that a spot sample should be taken when the atmospheric pressure is 1,000 mbar and falling to represent worst-case scenario. If a sample was taken at point (a) compared to point (b), both of which satisfy this condition, a very different risk would be perceived, indicating the arbitrary nature of this value. However, the arbitrary nature of the 1,000-mbar limit is clear as concentration continues to vary depending on changes in atmospheric pressure, rather than displaying a clear dependency on the absolute atmospheric pressure.

The widely reported relationship between pressure and concentration does not always exist; the inverse relationship was observed at a neighboring borehole. In Figure 6, the expected relationship between atmospheric pressure and concentration did not exist. From July 2 to 22 the inverse relationship was observed, i.e. as pressure increased, the concentration increased.

Collecting time series data

Enlarge this picture Figure 6: In this continuous data set, the expected relationship between atmospheric pressure and concentration does not exist.

With the ability to collect continuous data, it is possible to purge a borehole and collect information on how the concentration recovers over time. This information is important because the rate at which the concentration recovers is directly related to the migration/generation potential.

Continuous gas-monitoring data has revealed several potential flaws in the existing monitoring methodologies. The identification of soil-gas regimes that vary on a site-specific basis indicates the potential for a mismatch between the frequency of sampling and the variability of gas concentration, demonstrating the importance of selecting an appropriate sampling frequency to avoid missing valuable information. This can clearly be seen when comparing the concentration duration curves from high-frequency data and spot-sample measurements.

The ability to simultaneously monitor environmental parameters and concentration will provide an understanding of the processes contributing to soil-gas production and migration. Initial results suggest that the relationship between environmental parameters and concentration are complex and currently poorly understood. The potential for further understanding will allow a more representative conceptual model. This has a further impact on risk assessment, which is currently based on inferences of worst-case conditions determined by limited periodic measurements of gas concentration.

Results of pump tests indicate that absolute concentration may not be the most important factor to consider when performing a risk assessment. Currently it is assumed that all boreholes behave similarly, but, it is now clear this is not the case. It is likely that pump tests will become standard practice in the future. PE