This dissertation presents a mixed integer program model to solve a ship routing and scheduling with stock control in ports, also known as maritime inventory routing. This model considers a heterogeneous fleet, carrying multiples products. It also has the ability to transform one product into another inside ships. This aspect allows the model to choose which product it wishes to deliver to a client with a less restrict quality specification in his demand. No model presented in other studies has this capability. Another possibility covered by this model is to combine mandatory demands with optional ones. The model is applied to a real case of maritime transportation of dirty oil products in a Brazilian oil company, whose current model has a series of small problems that hinders its use. Due to the long time it takes to get the optimal solution, the relax-and-fix and fix-andoptimize heuristics are used to get good solutions in a reduced time. With the use of computational experiments in a series of scenarios, it has proved the quality of the solutions found by the proposed methods, testing different configurations and discretizations of time. The results presented prove the superiority of the methods in comparison to the pure mathematical model. The proposed model has shown great potential to replace the current one and to achieve the improvement for the ship routing intended by the company.