The combustion efficiency of direct injection engines is largely dependent upon the mixing of fuel in air, thereby creating a combustible mixture. Such a process is highly dependent upon the motion of the charge in the cylinder. The shape of the intake runners and valves determines the charge motion generated within the engine. Swirl and tumble, generated along the vertical and horizontal axis respectively, govern the charge motion and hence distribution of combustible mixture. Unlike traditional parametric optimization where the parameter space has to be predetermined, adjoint optimization utilizes the gradient of objective functions obtained from a computational fluid dynamics solution to modify the shape of the original CAD geometry. During the optimization process, specific parts of the geometry can be morphed in any direction freely. The final design is a fluid volume generated as a result of such adjoint computations.
In the present study, multi-objective adjoint optimization is applied to a cylinder intake geometry wherein two symmetric runners with constant valve lift (flow-bench port geometry) are employed, with objectives of pressure loss minimization and charge motion maximization. Separate analyses have been performed for charge motion along cylindrical (swirl) and horizontal (tumble) axes. Application of the multi-objective adjoint approach on the intake geometry results in increased charge motion (mixing) and a reduction in pressure loss.