In the pursuit of reducing fuel consumption and meeting stringent emission regulations,controlling the air charge in heavy-duty vehicle combustion engines is critical. Instead ofthe traditional way of controlling an engine, which means using separate feed-forward andfeedback, this study explores the usage of a Nonlinear Model Predictive Control (NMPC). Anengine model, describing a physical engine, has been constructed and parameterized withdata from the engine manufacturer. Sub models were parameterized using the least squaremethods. With complete model parametrization, this report shows results of the minimizingof a cost-function that optimizes parameter values according to model-faults. Two solvers were used and compared for a simplified example, controlling a throttle.Both leading to similar results. One of the solvers were further used to control the completeengine model. The objective was to achieve a desired combustion, minimize losses andthereby reducing emissions. The NMPC was simulated in two different drive cycles. The results showed that a satisfactory engine model has to be used in the controller toachieve optimal control. Several results show that this method of controlling an engine canbe beneficial. / <p>B-huset, Plan 3</p>