Neutron Probe Technique for Moisture Monitoring in Landfills

NEUTRON PROBE TECHNIQUE FOR MOISTURE MONITORING IN LANDFILLS

S.T.S. Yuen, Q.J. Wang, J.R. Styles and T.A. McMahon

Department of Civil & Environmental Engineering, University of Melbourne, Parkville, Victoria 3052, Australia

SUMMARY : The paper begins by highlighting the need for indirect/non-destructive measurements of moisture in MSW landfills. It then presents the findings of a study comprising a feasibility assessment, a laboratory investigation and a field application to substantiate that a neutron probe can practically assess moisture content of MSW despite some limitations. A field application procedure is also recommended.

1.INTRODUCTION

With the recent shift in the design of municipal solid waste (MSW) landfill moving from a permanent containment storage (dry cell) concept to a bio-reactor (wet cell) approach, there is a growing need for in-situ monitoring of MSW moisture. For a successful bio-reactor landfill, it is essential to maintain the MSW at an optimum moisture state favourable for bio-degradation. The measurement of MSW moisture is needed in landfill related research such as:

• leachate recirculation studies to assess the performance of the recirculation systems,
• predicating methane formation potential of an existing landfill,
• landfill cell water balance to predict leachate generation, and
• studies of saturated/unsaturated flow in a MSW medium.

While the measurement of moisture content of a landfill can be achieved by gravimetric determination of samples directly collected from the landfill, the direct method has certain operational constraints. The sampling of waste, often involving either drilling or excavation in the landfill, can be destructive to the containment system. It can also be expensive if substantial and repetitive sampling is required. It also involves the drying of samples which takes time. To avoid the above limitations, the availability of a rapid, non-destructive, and easily repeatable indirect method would be desirable, particularly in situations where a large spatial moisture measurement of a repetitive nature is required.

Although a number of standard techniques of in-situ soil moisture measurement are available in agricultural and engineering applications, very little information is available in the literature regarding the measurement of moisture in a MSW medium.

This MSW moisture measurement study was conducted as part of a full-scale leachate recirculation bio-reactor landfill research project which has been described by Yuen et al. (1995). The study comprised three components, namely a feasibility assessment of all standard moisture measurement methods, a laboratory testing program to confirm viability, and a field application in a real landfill situation.

2.FEASIBILITY ASSESSMENT

A wide range of soil moisture measurement devices commonly employed in agricultural and engineering applications were assessed. They can be broadly divided into the following categories based on their principles of operation :

• electromagnetic techniques,
• electrical or thermal conductivity,
• tensiometric techniques, and
• neutron scattering.

The principles of operation, advantages and disadvantages of each of the above methods used in a soil medium have been discussed in detail by other studies (e.g. Schmugge et al., 1980; Gardner, 1986; Tyndale-Biscoe and Malano, 1993). However, due to the complexity and highly heterogeneous nature of MSW, many difficulties and limitations exist when they are used in a landfill. A detailed discussion regarding their suitability in MSW is given by Yuen et al. (1996). The selection of a suitable method was based on the following criteria :

• reliability,
• ease of measurement,
• non-destructiveness,
• repeatability,
• acceptable accuracy, and
• large effective sampling volume.

While all of the above criteria are obvious, it is important to emphasise the importance of a large effective sampling volume to achieve “macroscopic” scale sampling. The heterogeneity of MSW is much higher than that of soil. Both the composition and size of individual component can vary in a wide range and each type of material may exhibit a different moisture/suction characteristic. For example, a pocket of food waste would carry a much higher moisture content than a piece of plastic even when they are subject to the same hydrological condition. Hence, to study the moisture state of a landfill, it is only appropriate to look at the moisture regime at a larger scale and it is important that moisture contents of macroscopic samples are measured.

The high level of salinity and heterogeneity of a MSW medium prevent the use of most of the above techniques. The feasibility assessment (Yuen et al., 1996) concluded that there is no ideal indirect/non-invasive method for MSW. Nevertheless, neutron scattering technique combined with the use of in-situ access tube was identified as having the potential to produce acceptable results within certain limitations.

The operating principles of a neutron moisture meter are well documented (e.g. Goodspeed, 1981 and Stone, 1990). Neutrons with high energy are emitted by a radioactive source into the soil and are slowed down (thermalised) by elastic collisions with nuclei of atoms. As hydrogen has a very low atomic weight, it can slow neutrons more effectively than other elements. The density of the resultant cloud of slowed neutrons (which can be detected by a counter) is taken to be proportional to the total number of hydrogen atoms per unit volume of soil. Assuming these hydrogen atoms have a direct correlation with soil moisture, the volumetric moisture content can then be determined from an established calibration curve.

There are certain distinctive advantages of using a neutron probe in a MSW medium. It has a relatively large radius of influence (generally between 150 mm in wet soil to 700 mm in dry soil (Gardener, 1986) )and can measure a continuous full depth profile through a access tube. Hence it provides a spatial coverage superior to any point sampling methods. This macroscopic sampling feature helps to better represent moisture content of a heterogeneous MSW medium. Once the access tubes are in place, no maintenance is required. To record temporal changes, measurement can be repeated at the same location as often as required by simply taking readings through the same tube.

However, the neutron scattering method has its limitations particularly when it is applied in a MSW medium.

3.POTENTIAL LIMITATIONS WITH NEUTRON SCATTERING

3.1Bound Hydrogen Effect

Hydrogen atoms may exist in a soil as free water or as bound hydrogen (e.g. in the form of organic matter). Although this bound hydrogen cannot be removed by oven-drying, it affects neutron response in the same way as free water (Holmes, 1966). This effect in soil has been discussed by Greacen et al. (1981), Hauser (1984) and Dickey (1990) and can be shown graphically as in Figure 1. Any presence of bound hydrogen would cause a parallel up-shift in the calibration line but without a change of slope. The magnitude of this vertical shift depends directly on the quantity of bound hydrogen.

As MSW by its nature consists of a significant proportion of material containing bound hydrogen (e.g. plastic and wood), this bound hydrogen bias could be substantial in MSW.

3.2Neutron Capture Effect

As neutrons are slowed down (thermalised) they are subject to capture by various elements that have an affinity for neutrons. This absorption capacity varies from element to element. Greacen et al. (1981), Hauser (1984) and Dickey (1990) suggested that this neutron capture effect can be reflected graphically as shown in Figure 2. The neutron absorption elements present in the soil essentially decrease the number of thermalised neutrons and this reduction is proportional to moisture content. This effectively reduces the gradient of the calibration curve.

Table 1 (Dickey 1990) shows the absorption capacities of some common elements. Boron although having a relatively high absorption capacity, is considered not important as it is scarce in soil and MSW. Generally, the elements of concern which are common in MSW are iron, potassium and chlorine. With their presence, one would expect a certain degree of neutron absorption bias.

3.3Density Effect

The effects of soil density on neutron probe calibration have been investigated by many researchers (e.g. Holmes, 1966; Greacen and Schrale, 1976). Basically, a change in density affects the neutron count because the density change may result in change in both bound hydrogen and neutron capture effects. For example, an increase in the density of a soil containing bound hydrogen would increase the count rate due to more hydrogen per unit volume of soil. On the contrary, increasing the density of a soil containing neutron absorption elements would decrease the count rate due to more neutron capture per unit volume.

4. LABORATORY INVESTIGATION

4.1Methodology

In order to quantify the above neutron probe limitations associated with its use in MSW, a laboratory program comprising a series of tests was conducted to determine the calibration curves for :

• Test (a) - Sand,
• Test (b) - Sand with ferrous metal (2% by dry mass),
• Test (c) - Sand with plastic ( 5% by dry mass),
• Test (d) - Sand with wood (10% by dry mass), and
• Test (e) - Sand wetted with leachate (instead of water).

Test (a) would provide a basic calibration curve to be used as the standard for comparison. The material used was a very fine laboratory grade silica sand which contained no bound hydrogen. The constitutional elements of the sand, i.e. silicon and oxygen, should exhibit negligible neutron capture (refer to Table 1). Hence the resultant calibration curve could be considered to have no bias due to either bound hydrogen or neutron capture effects.

The materials in Tests (b), (c) and (d) were chosen to represent some of the most common MSW components that may exhibit bias in neutron counts. The percentages used were based on their average proportions commonly found in landfills. The mixtures were used to test sensitivity on neutron count rather than to represent their true proportion in a landfill.

Test (b) was conducted with a mixture of sand and ferrous metal filings. The result would reflect any neutron capture effect due to the presence of the strong neutron absorption iron element (refer to Table 1). Ferrous metal is a common component in landfills although it is only present in a very small proportion.

Test (c) used a mixture of plastic granules and sand. Polyethylene (PE) was used in the test. It is the most common form of plastic (generally used for sheetings and carrier bags) in the plastic component of MSW. The result would provide some indication of the bound hydrogen effect coming from the hydrogen atoms attached to the polyethylene polymers. The use of polyvinyl chloride (PVC) was avoided as the attached chlorine atoms would produce a secondary effect of neutron capture which may complicate the interpretation of the results.

Test (d) was conducted with wood chips which would reflect the bound hydrogen effect due to the presence of natural organic matter such as cellulose which is also common in paper, garden waste, timber and kitchen waste that occupy a substantial portion of MSW.

Test (e) with the presence of leachate was expected to show both bound hydrogen and neutron capture bias in the calibration curve, as leachate commonly contains a considerable amount of organic matter (bound hydrogen effect) as well as a high level of chloride (neutron capture effect). The leachate used in the test was collected from a young landfill cell and was tested to exhibit both contents in typical concentrations.

All six tests were conducted in a 120 litre, 700mm diameter drum with a vertical aluminium access tube (40mm internal diameter 2mm thick) installed in the centre. In selecting the size of the tube, it is desirable to minimise the air gap between the probe (of 38mm diameter) and the tube as the gap constitutes a discontinuity in the system being measured. This is also important from the point of view of reproducibility of probe location and avoidance of possible asymmetry effects. Aluminium tubing was used as it is virtually transparent to neutrons. Steel and PVC tubings were avoided because of their high absorption of slow neutrons. The neutron counts were measured by using a CPN Corporation model 503 DR neutron depth moisture probe. Seven neutron counts were taken for each moisture measurement, averaged and then converted to count ratio (i.e. average count to standard count).

In order to minimise the effect due to density change, every attempt was made to maintain a consistent density for each batch during compaction. Large samples were taken from various locations of the drum for gravimetric moisture determination. For the conversion of gravimetric moisture to volumetric measurement, the bulk density of each batch was required. In this case an average bulk density was used and obtained by weighing all the material placed in the test drum of a known volume.

4.2Results and Discussions

Results of each test on dry densities suggest that the compaction of the mix was maintained reasonably consistent through out. However, to minimise any density effect, count ratios were corrected to account for density variation as proposed by Greacen and Schrale (1976) which uses an empirical relationship that assumes count rate at constant total volumetric water content is proportional to the square root of density.

The calibration curves (with corrected count ratios) as fitted by the least squares method are plotted in Figure 3. The slopes, m and intercepts, c of the six lines are listed in Table 2.

To assess bound hydrogen effect, we can take the intercept of Test (a) (c= -0.21) as the datum for comparison since the sand mix contained no bound hydrogen. Referring to Table 2, the variation in intercepts c agrees very well with the bound hydrogen theory and follows the trend as discussed in Section 3.1. While the increase in the leachate mix (c = +0.07) was relatively mild, the increases in the 5% plastic mix (c = +0.43) and in the 10% wood mix (c = +0.38) were very obvious. The 2% Fe mix gave an intercept value very close to the sand line (c = -0.01) which reflected the absence of bound hydrogen.

For neutron capture effect, as indicated in Table 2, the changes in gradient followed very closely the pattern as predicted by the theory described in Section 3.2. The gradient of Test (a) (m=6.24) was taken as the datum since the sand exhibited no neutron capture. For the 5% plastic mix with no neutron capture, the gradient remained almost the same (m= +0.1 or +1.6%). For the leachate mix, the gradient dropped only slightly (m = -0.3 or -4.8%). For the 2% Fe mix, the drop was higher (m= -0.6 or -9.6%) but the change was still small. The above reflects that neutron capture effect caused by the presence of chloride in the leachate and the presence of the iron filings in the Fe mix was relatively mild. However, for the 10% wood mix line, there was no logical explanation for the gradient drop (m = -0.6 or -9.6%). As this drop in gradient was small, it might simply be due to experimental error.

The above results thus indicate that the intercept of calibration curve can change substantially in a MSW medium from sample to sample due to the presence of material containing a high amount of bound hydrogen. This means a successful application to measure absolute moisture content would require an individual calibration curve appropriate to each in-situ sample and that would be practically very difficult to achieve.

Nevertheless, the results also suggest that the gradient of calibration curve may drop only slightly from sample to sample even with the presence of neutron capture elements provided their presence is not excessive. Based on this assumption, one can use the gradient of the standard sand curve to interpret moisture change from the change in neutron count and the error involved would be small.

An example can be taken to quantify this error. Considering a change in count ratio of 0.5 is detected by the neutron probe, using the gradient of the standard curve would interpret a corresponding change of moisture of 0.080. If the sample actually contains a neutron capture element as in the case of the 5% Fe mix, using the gradient of the correct curve, the actual moisture change should then be 0.088. In this case the moisture change is under-estimated by 9 %. Of course this error grows as the presence of neutron capture element increases. In general the use of the standard curve would tend to slightly under-estimate moisture change.

Working with MSW of high heterogeneity, it would be very difficult to quantity this error associated with each sample. Using the standard curve to measure all samples would inevitably produce results of unacceptable errors in occasional cases of extreme composition. However, the macroscopic sampling feature as discussed previously would help to smooth out any of these extreme measurements and to maintain a reasonable “average” moisture change profile.

If a knowledge of the composition of the MSW sample is available, the error can be minimised by selecting a closer and more representative calibration curve. However, this is difficult to achieve in practice. For practicality, the standard curve can be used in most applications provided the above limitations are observed.

In practice, when a hole is being drilled to install the access tube, in-situ samples along the depth profile can be collected to determine the gravimetric moisture content and to estimate dry density. Once the initial volumetric moisture profile is established, any subsequent moisture change detected by change in neutron count can then be used to calculate the new volumetric moisture content.

5.FIELD APPLICATION

5.1Description of Full-Scale Trial

The above neutron probe technique was applied in a real landfill situation to measure MSW moisture. In-situ vertical aluminium access tubes each 12 m long were installed in the Lyndhurst bio-reactor landfill test cell (located 35 km south-east of Melbourne, Australia) to monitor seasonal moisture change as well as to monitor moisture change due to leachate recirculation (Yuen et al. 1997).

The same neutron probe used in the laboratory tests was employed except in this case a longer connection cable was used to allow the probe to reach the full depth of the access tubes. The material and size of the access tubes were identical to that used in the laboratory.

Considerable difficulties were encountered in the installation of the access tube. Each tube had to penetrate 12m down into the landfill and the longest available tubing length was 6m, therefore at least one joint was required. As the internal diameter of the tube (40mm) is only marginally larger than the diameter of the neutron probe (38mm), welding the tubes inevitably reduces the effective internal diameter at the joint and would not allow the probe through. After some trials, a “pin and socket” joint was successfully employed and glued with Araldite (a two-pack epoxy glue). The joint was pre-made in the workshop by machine turning to remove half of the thickness of the tubes, one end externally and the other end internally.