【作者】 刘诚；

【导师】 王德忠；

【作者基本信息】 上海交通大学 ， 核能科学与工程， 2013， 博士

【摘要】 随着核能工业发展和放射性同位素日趋广泛应用，各种放射性废物大量产生和堆积。目前，我国核电运行以来产生的放射性废物大量贮存在核电厂的废物暂存库内，若不能得到及时处置，将可能成为影响核电安全运行的隐患。除此之外，国防产生的放射性废物、核设施退役产生的放射性废物如何安全处置也是一个迫切需要解决的问题。在对放射性废物最终处置之前，必须对放射性废物中的核素组成与活度进行准确鉴别与测量，为其暂存、运输和最终处置提供科学依据。由于中低放废物的特殊性，目前普遍采用无损分析方法（Non-DestructiveAssay，NDA）对废物桶进行测量。NDA方法又包括：无源和有源的γ射线分析法、无源和有源的中子分析法、量热分析法。其中γ射线分析法利用样品本身发射的γ射线来对样品进行定量分析，且不会产生二次放射性废物，是应用最为广泛的NDA方法之一。废物桶的γ射线扫描技术经历了分段γ扫描技术（Segmented GammaScanning，SGS）和层析γ扫描技术（Tomographic Gamma Scanning，TGS）两个阶段。对比SGS技术与TGS技术，SGS测量过程简单、速度快，但测量非均匀样品时具有较大的误差；TGS可以比较精确地获得桶内物质与放射性核素的分布，但复杂的测量过程难以满足大规模探测要求。针对上述问题，本文以研究更加快速准确的中低放射性废物活度探测技术为目的，对改进型SGS技术和TGS技术开展了研究，主要研究内容和成果包括：（1）双探测器改进型SGS技术理论及测量系统的建立传统的SGS技术采用桶内每层介质与放射性核素都均匀分布的假设，因此在对非均匀分布的样品进行测量时会出现较大的误差。本文提出的双探测器改进型SGS技术采用桶内每层介质均匀分布的假设，但是认为放射性核素以等效的环形线源的形式存在，通过两个探测器在不同位置测到的计数率分析得到该层内核素的等效半径。通过公式推导详细说明了该改进型SGS技术的理论基础，并详细介绍了其活度重建算法及测量系统布置。（2）双探测器改进型SGS技术的模拟及实验验证采用MCNP（Monte-Carlo N Particle）程序对不同密度的均匀介质单点源以及多点源、非均匀介质单点源的情况进行了模拟测量，并对模拟数据进行了重建分析。对于均匀介质，在最极端的单点源情况下，双探测器改进型SGS对137Cs的活度重建结果的相对误差为：密度为0.3g/cm3时在±10%范围内，0.6g/cm3时在±15%范围内，1.0g/cm3时在±20%范围内，1.5g/cm3时在±25%范围内。同时，对一非均匀单点源样品也进行了实验测量。模拟测量数据和实验测量数据的重建结果均表明，双探测器改进型SGS技术比SGS技术具有更高的测量精度。（3）基于动网格的TGS图像重建研究针对目前TGS技术中网格划分较粗，分辨率较低的问题，提出采用动网格进行TGS图像重建，并进一步提出了一种适合于TGS重建的自适应动网格加密算法，在点源附近区域采用小尺寸网格对“热点”进行准确定位，对不存在核素或者核素活度较低的区域采用大尺寸网格。采用MCNP程序对一非均匀介质的点源样品进行了TGS模拟扫描，对模拟数据的重建结果表明：采用动网格进行TGS发射重建能够在对“热点”核素进行准确定位和减少网格数量的同时减小发射重建活度误差，提高TGS的测量精度。（4）基于动网格发射重建的TGS测量优化研究针对目前TGS测量时间过长而尚未大规模工业应用的问题，提出了减少扫描次数并利用动网格进行发射重建的解决方法。为验证该方法的可行性并确定测量优化方案，针对旋转24次平移4次的TGS原始扫描方案，设置了四种减少转动或（和）平动次数的简化测量方案，采用MCNP程序对均匀介质和非均匀介质两种样品按照各种扫描方案进行了模拟扫描，并采用固定网格进行透射重建，采用固定网格和动网格分别进行发射重建，进而研究各种简化方案的重建误差变化情况。对各测量方案下的216个单点源以及100组多点源工况的重建误差进行统计分析后对比发现：由旋转24次平移4次的标准扫描方案简化到旋转12次平移4次后，再采用动网格进行发射重建，测量精度不会明显改变，但简化后测量次数减少一半，能够明显缩短TGS的测量时间。更多还原

【Abstract】 Large amounts of low and intermediate level radioactive waste (LILW) will beproduced and piled up with the nuclear power industry development and the increasingwidespread application of radioisotopes. Currently, a large number of radioactive wastegenerated in the operation of nuclear power plant are stored in the temporary repository,and it may affect the safe operation of nuclear power if the nuclear waste can not gettimely disposal. In addition, the safe disposal of radioactive waste from national defenseindustry and decommissioning of nuclear facilities is also an urgent problem. Thedistribution of the radionuclide composition and activity must be accurately measuredbefore the final disposal of LILW to provide a scientific basis for its storage,transportation and final disposal.Due to the particularity of LILW, non-destructive assay (NDA) method iscommonly used to measure the waste drums. The γ-ray analysis method is based on theγ-ray emitted by the sample itself and does not produce the secondary radioactive waste,so it is one of the most widely-used NDA method. The γ-ray scanning technique forwaste drums has undergone two stages: Segmented Gamma Scanning (SGS) andTomographic Gamma Scanning (TGS). The SGS measurement process in simple and fast,but with a large measurement error; TGS can obtain the distribution of the material andradionuclides in the waste drum accurately, but it is difficult to meet the requirements oflarge-scale detection because of its complexity and long time of the measurement process.The research of improved SGS (ISGS) and TGS method is carried out in this paper,aiming at the more rapid and accurate detection technique for LILW. The main contentsand results are as follows:(1) The establishment of the theory and measurement system of ISGSThe traditional SGS assume that the matrix and radioactive sources are distributeduniformly in each vertical segment, resulting in a large error in measuring thenon-uniform sample. The ISGS proposed in this paper assume the uniform matrix in eachsegment, but the radioactive sources exist in the form of an equivalent ring source, andthe equivalent radius is analyzed by the count rates of two detectors at different positions.The theoretical basis of the ISGS is described in detail by formula derivation, and theactivity reconstruction algorithm and measurement system layout are also presented.(2) Simulation and experimental verification of ISGSThe simulation measurements of the cases of single source in homogeneous matrix,multiple sources in homogeneous matrix and single source in heterogeneous matrix arecarried out by using MCNP code. For the homogeneous matrix, in the most extreme caseof a single point source, the relative errors of reconstruction activity results of ISGS arein the range of±10%at the density of0.3g/cm3,±15%at the density of0.6g/cm3,±20%at the density of1.0g/cm3, and±25%at the density of1.5g/cm3. Meanwhile, theexperimental measurement of a heterogeneous sample is also carried out. Thereconstruction results of simulation measurement data and experimental data both showthat the ISGS has a higher accuracy than SGS.(3) Study of TGS image reconstruction using dynamic gridsIn order to solve the problem of coarse grids and low resolution in TGS, usingdynamic grids in TGS image reconstruction is proposed. A adaptive grid refinementstrategy which is suitable for TGS reconstruction is developed in order to locate smallsize grids in the vicinity of the point source for accurate positioning of the ‘hot spots’ and locate big size grids in the area of no or low radioactivity. The simulation measurementsof a heterogeneous matrix with point sources are carried out by using MCNP code. Thereconstruction results of simulation measurement data demonstrate that dynamic grids inemission reconstruction outperform the fixed grids in terms of the accuracy and ’hotspots’ positioning with fewer grids in most cases.(4) Study on the optimization of TGS detection by using dynamic grids in theemission reconstructionReducing the scan times and using dynamic grids in emission reconstruction isproposed to solve the problem of the long measurement time of TGS. To verify thefeasibility of the solution and find the optimal scan scheme, we set five differentsimplified scan schemes which reduce the rotation or (and) translation number based onthe original scan scheme of24rotating times and4translation times. The simulationmeasurements of a homogeneous sample and a heterogeneous sample with point sourcesare carried out by using MCNP code, and the transmission reconstruction is implementedby fixed grids, the emission reconstruction is implemented by fixed grids and dynamicgrids respectively to compare the activity reconstruction error of all the scan schemes.The statistical results of the reconstruction error of216single point sources and100multiple point sources indicate that simplification of TGS scan process from24rotatingtimes and4translation times to12rotating times and4translation times will lead to littlechange of the emission reconstruction error if dynamic grids are applied, but thissimplification will lead to a obvious reduction of TGS scan time.更多还原