Adjoint-based multi-point optimization framework for mixing processes applied to a venturi mixer

Venturi mixers are used to conditionally inject fuel into an air stream when the under pressure is high enough. We consider here the injection and mixing of gaseous fuel and air in a venturi mixer, where the total mass flow rate of the mixture is given and the fuel to air ratio depends on the suction pressure in the venturi throat as well as the local friction in the fuel and air stream. This means that in practice, the fuel to air ratio in venturi mixers will vary with the mass flow rate. To prevent performance losses, a constant fuel to air ratio over a large range of mass flow rates is desired. To achieve this, we use a discrete adjoint method inside a multi-point optimization framework. The objective is to achieve a target fuel to air ratio for a range of total mass flow rates. Since the mass flow is achieved in practice using a fan that can overcome a limited pressure drop, a constraint is added for the pressure drop for a second optimization run. This approach has been implemented using the open source code SU2 and the algorithmic differentiation package CoDiPack. The optimization was encapsulated in the python framework FADO. The initial computational setup was first compared to measurements, showing a good match. After optimization, the target fuel to air ratio was achieved over the desired range of mass flow rates with excellent performance. • Extension of the SU2 framework for multi-point optimization of mixing processes. • Comparison of computations and measurements of species transport and mixing processes for a venturi mixer. • Extension of adjoint-based design of mixers to an industrial setting. • Optimization study with multiple operating conditions and a constraint.