Ant colony optimization of shielding for mixed neutron and gamma radiations

• Main Author: Azali, Muhammad Arif
• Format: Thesis
• Language: English
• Published: 2020
• Subjects: TP Chemical technology
• Online Access: http://eprints.utm.my/id/eprint/92429/1/MuhammadArifPMSChE2020.pdf.pdf; http://eprints.utm.my/id/eprint/92429/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:139300
• Institution: Universiti Teknologi Malaysia

Shielding is important in maintaining safe levels of radiation. In mixed neutrons and gamma-rays condition, more than one material are needed for shielding purposes. These materials can be arranged into a multilayer shield or they can be mixed into a composite shield. These shields have different variables and it can be increasingly complex to optimise them. In such a situation, using brute method is infeasible in terms of computation time. Recent studies have been looking into the use of metaheuristics in shield optimisation specifically the genetic algorithm (GA). This study extends the knowledge by investigating the ant colony optimisation (ACO) for shielding optimisation against mixed radiation. Three objectives were outlined: to develop an ACO-based algorithm to optimise a shield made from polyethylene (PE), boron, and tungsten, to build a GA as a comparison, and to evaluate the new shield design. MCNP5 was used for shielding calculations. There were four problem cases: a composite shield with known solution (Case 1); a composite shield with unknown solution (Case 1.2); a multilayer shield with known solution (Case 2); and a multilayer shield with unknown solution (Case 2.2). Six ACO parameters and four GA parameters were tested to observe their effects and to determine their best values for the optimisation algorithms. Four composite shields were fabricated and experimented with a 252Cf mixed neutron-gamma source. It was observed that the ACO-MCNP algorithm was significantly better than the brute method. It managed to find the exact solutions for Case 1 and Case 2 while reducing the runtime of the brute method by 81.75% and 89.01% respectively. For Case 1.2 and Case 2.2, good solutions were achieved in only 2.13 hours and 1.28 hours respectively as compared to the brute method which could take almost 70 hours to complete. The results also suggest that the ACO is a good alternative to the GA for shielding optimisation. For the experimental work, it was found that the PE composite with additives of 16 wt% boron and 16 wt% tungsten had the best mixed radiation shielding performance as compared to pure polyethylene, poly-boron (25 wt%), and poly-tungsten (25 wt%).