Context
At SCK CEN, we irradiate targets in our BR2 reactor to obtain radionuclides used in medical and industrial applications. These irradiations are characterized by the target type (base material), the irradiation position in the reactor (which determines the neutron intensity and energy distribution) and the duration of irradiation and cooling down (radioactive decay). The quality of the irradiation is determined by the mass of the radionuclide of interest at end-of-irradiation and the “contaminants”. Even for a 100% pure target, there will be “parasitic” interactions, producing radionuclides other than the one aimed at. Some of these will have an impact on the purity of the final material, some will have an impact on the processes between irradiation and shipping of the final product (for example, the presence of high energy gamma emitters will require much shielding). Some of these reactions will have an impact on the irradiation itself (if fission reactions can occur, the target will generate a substantial heat during irradiation and there are engineering constraints on coolability).
In its most simple form, the equations to solve are a coupled set of ODEs where the coefficients are determined by reactor physics parameters. These parameters can be calculated for the initial, unperturbed, system but in reality these will change during irradiation and hence will need to be updated (non-linear problem being linearized). In a more elaborate form, the radionuclide composition of the target is completely calculated by a reactor physics code and the objective function and constraints of the optimization problem are results of this “black box” simulation.
Of course, when the optimal irradiation conditions have been determined, the reactor physicists want to know how sensitive the optimum is to a change in these parameters.
Goal
Develop an optimization framework that couples gradient-free optimization algorithms to the simple, coupled ODEs model and the complex “black box” reactor physics simulation code to obtain an optimal irradiation condition characterized by neutron intensity, energy distribution, irradiation and cooling down time and which is constrained by presence of contaminants, heat generation, radiation emission during target handling etc. Allow for sensitivities of the optimum to input variables to be estimated. Apply this to one or two typical irradiation problems currently of interest.
Methodology
- Get acquainted with the basics of nuclear interactions, neutron activation, fission
- Implement the simple model (coupled ODEs with fixed, up-front calculated, coefficients) as generic as possible (and apply to the one or two model problems at hand)
- Embed this as objective function and constraints in an optimization framework
- Compare different optimization algorithms
- Couple the optimization framework with the full-blown reactor physics code as a black box model
- Write a report and do a decent hand-over of the code
Het vereiste minimumdiploma van de kandidaat
- Academic bachelor
De vereiste achtergrondkennis van de kandidaat
- Informatics
- Mathematics