Study On Optimal Design Methods For The Radiation Field Of Gamma Irradiation Facilities

The thesis does study surround the optimal design of radiation field of gamma irradiation facilities,and it includes the content of three following respects concretely:the new methods of study on the optimal deployment of source pencils in the irradiators,the precise calculation methods of study and the building of high performance distributed computing platform for the radiation dose distribution around the irradiators.The main achievements of this thesis are as follows:The history,industrial scale and industrial output of radiation processing technologies at domestic and abroad are introduced firstly,and then the equipment structures,machining principle,processing advantages,industry standards,especially several dosimetry problems existing in the?radiation processing technology are summarized and analyzed in details in the chapter1.The deployment result of source pencils in the irradiator determines the uniformity of radiation distribution surround the irradiator,and it then has a further impact on the machining quality and efficiency of goods.Based on the basic idea of plant growth simulation algorithm?PGSA?,a new algorithm for the optimal deployment of source pencils was proposed in the chapter 2.The code of new algorithm was written in C++language and tested by three examples,the test results showed that the new method could give a number of deployment plans in high level for users to choose,and the related dose uniformity ratio?DUR?values?1.21¹.39?of radiation distribution reached the domestic leading level.In order to further improve the computational efficiency of new algorithm according to the performance analysis result by GNU“gprof”,OpenMP and GPU parallel technologies were used to achieve two new parallel algorithms and their codes can be run on the CPU and GPU separately.Both of them were tested by using seven large-scale benchmarks,and the test results showed that the performance of algorithm achieved by OpenMP was obviously improved but still not be satisfied,by contrast,the efficiency of parallel algorithm code running on the GPU increased by tens or hundreds of times as compared with the serial version on the CPU.The new GPU-based algorithm is able to complete the source pencils'deployment work of most large gamma irradiators in which the installed capacity of Co-60 is not more than 1.48×10¹⁷ Bq?4 million Ci?.If more computation resources are supplied,the GPU-based parallel algorithm has the ability to meet the requirements of any scale gamma irradiators.For the convenience of users'on-site operation,deployment plans can be converted from data lists to graphics by the code written in MATLAB.On the basis of optimal deployment of source pencils in the gamma irradiators,the details of radiation field surround the source rack and especially of dose distribution in the irradiated goods are needed to be obtained.In the chapter 3,an Monte Carlo program named MCNP for particle transport simulation was used firstly to calculate the dose distribution in the small water spheres?stand for the irradiated goods?which are irradiated by single C-188 bare Cobalt-60 pencil,and the results were in good agreement with the theoretical ones,most of the difference were less than 5%,individual difference did not exceed 5.21%.Such work verified the feasibility and effectiveness of the application of MCNP.On this basis,partly combined with peer experiences,the thesis put forward a general method and principle of establishing simulation model and developed a common set of template code by which the radiation dose distribution of any scale single rack gamma irradiators at home and abroad could be getting easily and precisely.The code was tested by using the parameters of a small irradiator,and the computation time was 283.8 minutes,the results were validated by the theoretical ones with the difference in less than 2.2%.In addition,the work is still not perfect which needs to be further compared and fixed with the real data.Because Monte Carlo codes usually takes a lot of computing resources of processor,the thesis in the chapter 4 gave a short introduction about the most commonly used methods of distributed parallel computing such as PVM,MPI,OpenMP and GPU.Combined with our own practice,the deployment methods for a small scale parallel computing cluster which consists of four nodes was detailed from two aspects of PVM,and MPI,and the daily management and application packages were achieved by Windows OS comes with batch processing.The results of test examples and user experiences indicated that the cluster had the characteristics of inexpensive,flexible deployment,low difficulty to learn and use,stable and reliable and high competitiveness.The feasibility and effectiveness of the methods proposed in the thesis have been confirmed,which can be used for the optimal design of gamma irradiators'radiation field and the detailed calculation of absorbed dose in the irradiated goods.In addition,some aspects such as the revision for the Monte Carlo codes by the real data,and total system integration solutions,will require further research and improvement.