13August2026
14:00 Doctoral defense Room 85 of IC2
Topic on
Optimization, Simulation, and Stochastic Approaches for Dengue Vector Control
Student
Carlos Victor Dantas Araújo
Advisor / Teacher
Fabio Luiz Usberti
Brief summary
Dengue fever is a mosquito-borne disease that represents a significant public health challenge, especially in Brazil, which recorded more than 10 million cases in 2024. A central control strategy is the application of insecticides by fogging vehicles, leaving health authorities to select which blocks to attend to and define efficient routes under limited operational resources. This thesis introduces the City Block Routing Problem (CBRP), a combinatorial optimization problem that combines characteristics of Vehicle Routing Problems and Arc Routing Problems. For the CBRP, two families of integer programming formulations are proposed, accompanied by preprocessing techniques, four variants of Lagrangian Relaxations, and a constructive greedy heuristic. To capture the spatiotemporal dynamics of dengue, a Multi-Agent Based Simulation (MABS) is calibrated and validated with historical data from Alto Santo and Limoeiro do Norte, in Ceará. The CBRP is extended to a stochastic version (SCBRP), formulated as a two-stage program, with formulations, local search, a Simulated Annealing metaheuristic, and a Simheuristic that couples a multi-start algorithm to MABS. The experiments use two benchmarks derived from real data: 39 deterministic and 36 stochastic instances with 50 scenarios each. In the deterministic study, compact formulations with preprocessing dominate in both primal and dual boundaries, Lagrangian relaxation provides competitive boundaries without a general-purpose solver, and the greedy heuristic delivers high-quality primal solutions in a few seconds. The simulator is validated using historical notifications with strong spatial and temporal correlations. In the stochastic study, Simulated Annealing substantially reduces the average gap between the two formulation families and certifies new optimal solutions, while Simheuristic, validated with the use of MABS in real epidemiological windows, demonstrates significant operational gains over the current strategy of health authorities, offering relevant contributions to public health decision-making in the control of vector-borne diseases.
Examination Board
Headlines:
| Fábio Luiz Usberti | IC / UNICAMP |
| Mariá Cristina Vasconcelos Nascimento Rosset | ITA |
| Igor Tona Peres | PUC-Rio |
| Pedro Henrique Del Bianco Hokama | IC / UNICAMP |
| Ruben Interian Kovaliova | IC / UNICAMP |
Substitutes:
| Pedro Belin Castellucci | INE / UFSC |
| Felipe da Rocha Henriques | CEFET/RJ |
| Marcos Medeiros Raimundo | IC / UNICAMP |