PhD Defense | Moudud Hasan | In situ methods for high resolution mapping of radioactive soil contamination
In situ methods for high resolution mapping of radioactive soil contamination
Every country with older nuclear installations or NORM-related industries, including Belgium, suffers from problems related to radioactive soil contamination, usually with large volumes of contaminated soil involved. For radiological impact assessment of a site and environmental decommissioning and remediation, an accurate estimate of the three-dimensional distribution of radioactive contamination and its uncertainty (measurement and spatial) is of crucial importance. Uncertainties of such estimates in the context of remediation are sometimes very large, and further methodological developments are needed to reduce them. The main objective of this thesis is to develop new in situ measurement techniques and improve existing techniques for the determination of the distribution of radioactive contamination at affected sites and to develop and apply a methodology to estimate the associated uncertainties.
A key step in any measurement of radioactivity is the calibration of the detector efficiency. The efficiency of a detector can be obtained experimentally using a calibrated standard in a geometry identical to that of the source to be measured. In the case of in situ measurements, such a type of calibration is expensive and time consuming. Monte Carlo simulation can be used for efficiency calibration of the detector for complex geometries instead. It also facilitates sensitivity analysis of different measurement parameters. A reliable detector model is however needed for Monte Carlo efficiency calibration. In this study, a LaBr3(Ce) detector model was optimized and validated using different radioactive sources (241Am, 133Ba, 137Cs, 60Co and 152Eu) and geometries (point, extended and surface). The PENELOPE and MCNP codes were used for Monte Carlo simulations, and good agreement was observed between the simulated and experimental full energy peak efficiencies (FEPE), as their mean relative difference was 2.84% ± 1.93% and 2.79% ± 1.99% for the PENELOPE and MCNP simulations, respectively. The differences between the simulated FEPEs of the two Monte Carlo codes were negligible except for low energies (< 100 keV).
The modeled LaBr3(Ce) detector was used to develop a portable measurement setup for an in situ gamma spectrometric survey of a contaminated site. This system is also suitable for borehole gamma spectrometric measurements. The minimum detectable activity concentration (MDAC) is an important parameter of a measurement system. In this study, the MDAC of 137Cs for the measurement setup was investigated. It was observed that the MDAC varies with the position of the detector with respect to the ground surface. A 5- to 20-minute acquisition time, depending on the detector position, can be sufficient to obtain an MDAC of approximately 10% of the exemption limit of 137Cs (100 Bq/kg).
In situ gamma spectrometry above the soil surface, conducted on a predetermined grid, can be used to map the level of radioactive contamination in the soil. However, the field of view of a gamma detector can be tens of meters depending on the height of the detector above the soil surface, the source energy and the density of the soil and air. In the case of uncollimated measurements, overestimation of the contamination may result as the gamma radiation from a large area may contribute to the measurement. On the other hand, if one works with the assumption that the measurements only represent an average over certain support, local underestimation can result as well. The contribution of the adjacent areas should be disentangled to estimate the level of contamination correctly. The Tikhonov regularization-based 2D inversion methodology was used to disentangle the contribution of the adjacent area. The method performed well in assessing the 133Ba and 137Cs surface-source activity distributions. This method was also applied to calculate the 137Cs activity concentration distribution at the study site, making use of aboveground in situ gamma spectrometry measurements, which were conducted on a regular grid. The results from the inversion process agreed well with the results from the in situ borehole measurements and those from the soil samples, showing that the 2D inversion is a convenient approach to deconvolute the contribution of radioactive sources from nearby areas within a detector’s field of view and increases the resolution of spatial contamination mapping.
The depth distribution of radioactive contaminants (e.g., 137Cs) is also a critical piece of information for characterizing a contaminated site. The conventional approach to measuring the activity-depth distribution is to collect samples or cores from different depths at the site and measure them in a gamma ray spectrometry laboratory. This approach provides precise information about the activities of different radionuclides with a low level of minimum detectable activity (MDA). However, it can also be expensive and time-consuming to collect, prepare and measure individual samples. Therefore, spatial coverage can be limited, and spatial heterogeneities may not be properly accounted for through this approach. On the other hand, an in situ measurement is likely to be more representative of the measurement location, as it is less affected by the local heterogeneity of the source activity, integrating its response over a larger soil volume. Furthermore, in situ measurements are relatively low cost and quick to conduct, as statistically significant counts can be obtained in a short period because of the larger support volume. An in situ borehole gamma logging method was studied in a 137Cs-contaminated field. In this method, a gamma detector was lowered into a borehole, and gamma spectra were recorded at different depths. During borehole drilling, soil samples from different depths were also collected for low-level gamma spectrometry analysis under laboratory conditions using an HPGe detector to validate the in situ approach. The activity-depth distribution was calculated using different inversion methods, such as least squares optimization and Tikhonov regularization. The regularization parameter of the Tikhonov regularization method was estimated using three different methods, i.e., L-curve, generalized cross validation (GCV) and a prior approach. The sensitivity and uncertainty of different measurement parameters were also analyzed using the Monte Carlo method. The calculated activity-depth profiles were in good agreement with those obtained from the soil samples analysis. It was observed that the in situ method is faster and better at representing the true spatial activity variation than the conventional approach. The GCV method performed best in estimating the value of the regularization parameter.
The efficiency calculation of a detector for an in situ measurement setup in a borehole depends on many factors related to the soil properties and measurement parameters. In this study, the sensitivity of different soil characteristics and measurement parameters on simulated efficiencies for a 662 keV photon peak were investigated. In addition, a Bayesian data inversion with a Gaussian process model was used to calculate the activity concentration of 137Cs and its uncertainty considering the major sources of uncertainty identified during the sensitivity analysis, which include soil density, borehole radius, and the uncertainty in detector position in the borehole. The 95% credible interval of the calculated 137Cs activity concentrations covers those obtained from soil samples. Therefore, the vertical radioactivity distribution can be calculated using this probabilistic method for the proper estimation of uncertainty.
A dosimetric approach to measuring the activity-depth distribution using Thermoluminescence Dosimeters (TLDs) was also developed, which is a novelty in borehole measurements for radiological site characterization. The developed dosimetry-based method is less destructive and requires no inspection during measurement. For dose collection, lithium fluoride (LiF)-based thermoluminescent dosimeters (TLDs) were used to collect doses in several boreholes at the study site. The dose uptake efficiency of a TLD for several radionuclides was calculated using Monte Carlo simulation in PENELOPE. First, the natural background dose was calculated using the activity concentration of natural radionuclides and compared with the measured doses from a noncontaminated area. The natural background dose was subtracted from the measured doses to calculate the net dose from the contaminants (mainly 137Cs). From the net dose, the activity concentration-depth profiles of 137Cs were calculated using two different inversion methods, i.e., Tikhonov regularization with the generalized cross validation and Bayesian inversion. The activity concentrations of 137Cs from the TLD dosimetry agree well with the soil sample as well as with the results from in situ gamma spectrometry. However, the Tikhonov regularization produced an over-regularized activity concentration-depth profiles in a few cases. However, the Bayesian inversion method was better in calculating the activity concentration depth profile with a proper estimation of its uncertainty. The major limitation of dosimetry is that it is not feasible to identify individual radionuclides and their contribution to dosimetry data. Hence, prior information about radionuclide composition is needed to calculate the activity concentration from dosimetry data. Therefore, the TLD-based dosimetry method can be used for calculating the distribution of radioactivity concentration in soil when some prior information about the natural background and composition of radionuclides is available.
The methods studied in this thesis are particularly useful for collecting data at a contaminated site using relatively few resources. More data can help to improve the spatial resolution in contamination mapping, thus reducing the global uncertainty related to a surveying campaign, which leads to greater confidence in the estimation of the contaminated volume, a crucial parameter in a remediation plan.
Marijke Huysmans (VUB)
SCK CEN mentors:
Johan Camps (SCK CEN)
Bart Rogiers (SCK CEN)
Tim Vidmar (SCK CEN)
Click here for a list of obtained PhD degrees.