Talking about sizing and resource allocation in specific activities we need to consider time windows and estimated durations. Using linear programming techniques, an optimization model can be defined and solved by generating an optimal sequence of tasks. In practice, however, the data does not precisely represent reality, and the solution becomes non-optimal or even infeasible. In this context, it is necessary to consider the data uncertainty, dealing with the trade-off between the objective function value and robustness of the proposed plan. This study compares the robust optimization considering proposals from Soyster, Bental and Nemirovski, and Bertsimas and Sim. A case study in the area of petroleum was performed, aiming to analyze the results of each model. However, the Bental and Nemirovski model was not analyzed because of its computational size. The results suggest the advantages of considering the robust optimization model as a modeling paradigm in this case.